Ultraviolet

Spectroscopy
and UV lasers

edited by
Prabhakar Misra
Howard University
Washington, D.C.
Mark A. Dubinskii
Magnon, lnc.
Reisterstown, Maryland

M A R C E L

MARCEL INC.
DEKKER, NEWYORK BASEL
D E K K E R

Copyright © 2002 by Taylor & Francis

ISBN: 0-8247-0668-4

This book is printed on acid-free paper.

Headquarters
Marcel Dekker, Inc.
270 Madison Avenue, New York, NY 10016
tel: 212-696-9000; fax: 212-685-4540

Eastern Hemisphere Distribution

Marcel Dekker AG
Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland
tel: 41-61-261-8482; fax: 41-61-261-8896

World Wide Web

http:/ /www.taylorandfrancis.com

The publisher offers discounts on this book when ordered in bulk quantities. For more
information, write to Special Sales/Professional Marketing at the headquarters address
above.

Copyright  2002 by Marcel Dekker, Inc. All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or by any
means, electronic or mechanical, including photocopying, microfilming, and recording,
or by any information storage and retrieval system, without permission in writing from
the publisher.

Current printing (last digit):

10 9 8 7 6 5 4 3 2 1

PRINTED IN THE UNITED STATES OF AMERICA

Copyright © 2002 by Taylor & Francis

Preface

Ultraviolet Spectroscopy and UV Lasers covers a range of subjects, from ul-

traviolet (UV) and vacuum ultraviolet (VUV) sources to the latest advances in
instrumentation and techniques for absorption, emission, and fluorescence spec-
troscopy. The book will prove useful to scientists pursuing spectroscopy-related
research in fields as varied and diverse as optical physics and engineering, analyti-
cal chemistry, biology, and laser technology. The book can also serve as a refer-
ence text for a two-semester special topics course for upper-level undergraduate
and graduate students in these disciplines.
In the opening chapter, ‘‘Ultraviolet and Vacuum Ultraviolet Sources and
Materials for Lithography,’’ Liberman and Rothschild review radiation sources in
the UV and VUV, followed by an appraisal of the key factors governing efficient
detection of UV radiation. Their overview of sources and detectors, although
presented in the context of lithographic applications, will prove useful for other
UV spectroscopic applications as well. UV optical metrology is covered under
two subheadings: measurements of bulk material properties and surfaces and
measurement of thin film properties. Candidate optical materials for lithographic
systems are reviewed in light of their index homogeneity, birefringent properties,
and durability. An important consideration in the use of optical components for
metrological applications at UV wavelengths is the issue of large absorption coef-
ficients exhibited by vapors and surface adsorbates, which is the subject of discus-
sion in the concluding segment of this chapter.
The second chapter, ‘‘Laser Optogalvanic Spectroscopy of Discharge Plas-
mas in the Ultraviolet Region’’ is coauthored by one of the editors (PM) of this
book and provides an overview of the optogalvanic effect. Owing to its sensitivity
and selectivity, the optogalvanic technique has been utilized for a wide variety
of spectroscopic applications, including analytical flame spectrometry, atomic

Copyright © 2002 by Taylor & Francis

and molecular spectroscopy, laser stabilization and calibration, and plasma diag-
nostics. Haridass, Major, Misra, and Han provide both an experimental back-
ground and a theoretical framework for reporting an extensive array of optogal-
vanic spectral transitions and selective waveforms in the UV for both neon and
argon. The mechanism associated with the generation of the optogalvanic signal
is covered in depth, and a theoretical model is developed and used to analyze
and interpret time-resolved waveforms of specific optogalvanic transitions of
neon and argon in the context of electron collisional ionization and electron colli-
sional transfer rates in a discharge medium.
In Chapter 3, ‘‘Spectra of the Isotopomers of CO⫹, N2⫹, and NO in the
Ultraviolet,’’ Reddy and Haridass describe the experimental recording of data
with a hollow cathode discharge tube, followed by a subsequent detailed analysis
of the electronic spectra of these species. The ultraviolet spectroscopy of these
molecules is important for the understanding and elucidation of significant astro-
physical processes, as well as for improved understanding of associated terrestrial
environmental and combustion phenomena. A review of the relevant electronic
configurations and the general spectroscopic equations needed for a detailed spec-
tral analysis and interpretation is provided. A comprehensive historical overview
of past and present work is provided, in turn, for CO⫹, N2⫹, and NO, followed by
presentation and analysis of UV spectra for the isotopomers of all three molecular
species. Schematic energy level diagrams illustrate the allowed spectral transi-
tions and the associated branches in detail. Precise molecular parameters charac-
terizing the various electronic states of the different species have been determined
using rigorous nonlinear least-squares fitting routines and are tabulated for easy
reference.
Photoabsorption cross-section measurements in the ultraviolet and vacuum
ultraviolet are the subject of Chapter 4. Yoshino outlines the earlier cross-section
measurements—prior to 1980—and discusses the current instrumentation avail-
able for precise cross-section measurements in the UV and VUV. A major uncer-
tainty in cross-section measurements with a single path spectrometer is in estimat-
ing the background intensity level, since it may vary with wavelength and time
during the course of the photoabsorption scan. It is also often difficult to maintain
the gas pressure constant within the absorption cell, particularly for reactive and/
or absorbent species. Yoshino summarizes these difficulties in the context of
cross-section measurements of the photoabsorption continuum under lower reso-
lution, citing the specific example of the H2O molecule. He then reviews measure-
ments of cross-sections for molecular bands possessing associated fine rotational
structure under superior resolution in relation to the Doppler widths of relatively
simpler molecules. Line-by-line cross-section measurements of the Schumann-
Runge bands of O2 and their temperature dependence are then discussed. Line
and band oscillator strengths are subsequently related via the Hönl-London factor,

Copyright © 2002 by Taylor & Francis

whereby once the band oscillator strengths are established, cross-sections for any
temperature can in principle be determined.
In Chapter 5, ‘‘Ultraviolet and Vacuum Ultraviolet Laser Spectroscopy Us-
ing Fluorescence and Time-of-Flight Mass Detection,’’ Lipson and Shi review
the principles and techniques of ultraviolet lasers produced by frequency doubling
in anisotropic crystals, nonresonant third harmonic generation, and two-photon
resonant four-wave sum- and difference-frequency generation in isotropic gases.
Their application focus is on single-photon excitations of supersonically jet-
cooled gas-phase molecules that are detected either via laser-induced fluorescence
or by ion detection via time-of-flight mass spectrometry. The authors provide a
detailed discussion of supersonic jet expansions, which have now been estab-
lished as a powerful tool for gas-phase spectroscopy, because a jet environment
is highly conducive to the production and probing of rovibrationally cold mole-
cules, free radicals, and clusters. This chapter provides practical examples of
UV and VUV laser-induced fluorescence and time-of-flight excitation spectra
in supersonic jets. The spectroscopy of several metal monohalide radicals and
diatomic halogens is presented in some detail. A comprehensive catalog of atomic
and molecular systems studied with VUV lasers is provided in tabular form.
The table includes a summary of the electronic states investigated by specific
spectroscopic methods for various atomic and molecular species, along with the
pertinent literature references. The use of VUV lasers in tandem with mass spec-
trometry provides unique opportunities for detection and characterization of or-
ganic molecules. The practical use of these techniques is illustrated in the context
of organic molecules (with several examples), and a detailed referenced tabula-
tion of organic molecular systems detected by nonresonant VUV ionization has
been included. The authors also provide a comprehensive bibliography of the
literature associated with the twin techniques of fluorescence spectroscopy and
time-of-flight mass detection.
In Chapter 6, Cefalas and Sarantopoulou review spectroscopy and applica-
tions of diatomic and triatomic molecules assisted by laser light at 157.6 nm.
The authors begin with a description of the useful characteristics of the 157.6
nm radiation source—a molecular fluorine laser—and of its utility for photo-
lithography, and then lead up to a summary of the laser’s usage for elucidating
the spectroscopy of such varied molecules as OH, O2, HCl, DCl, CH3Cl, H2, OCS,
H2O, HDO, ICN, AsF2, O3, CO2, and SiH. Spectroscopic investigations relating
to the properties of neutral or charged clusters can be accomplished in the VUV
via optical and mass spectroscopic techniques. Such approaches to study clusters
in the VUV possess two distinct advantages, namely, that systems with a large
band gap can be studied and that excitation of inner shell electrons allows ele-
ment-specific information to be collected. In this context, the VUV spectroscopy
of triatomic clusters is reviewed. It is followed by a discussion of the laser-

Copyright © 2002 by Taylor & Francis

induced fluorescence and mass spectroscopy of diatomic and triatomic clusters
of mercury and a description of the interesting quantum beat phenomenon in
mercury triatomic clusters, following de-excitation from the excited state to the
ground state.
In ‘‘Tunable Solid-State Ultraviolet and Vacuum Ultraviolet Lasers’’
(Chapter 7), Ilev and Waynant comprehensively review the all-solid-state concept
for producing widely tunable short-wavelength laser generation using nonlinear
optical techniques and specific solid-state active materials (e.g., the rare-earth-
doped wide band gap crystals). The authors provide an analytical outline of the
general principles concerning basic nonlinear effects and then consider the pri-
mary all-solid-state nonlinear techniques and optical materials utilized for gener-
ating tunable UV/VUV radiation. Significant progress achieved by nonlinear fre-
quency upconversion techniques is also presented. Optical fibers provide a good
solid-state medium for broad-band optical frequency conversions, and when laser
radiation is propagated in such fibers suitable conditions are realized for efficient
nonlinear processes, which include, among others, stimulated Raman scattering,
four-photon mixing, and self-phase modulation. Nonlinear effects in optical fibers
are presented in considerable detail, especially those involving stimulated Raman
scattering and its application for the design and development of tunable fiber
Raman lasers in the UV. In addition to the all-solid-state nonlinear techniques
for tunable UV/VUV laser generation, the use of rare-earth-activated wide band
gap fluoride dielectric crystals for the development of solid-state UV/VUV lasers
is reviewed, and leads naturally to the subject matter of the next chapter.
In Chapter 8, ‘‘VUV Laser Spectroscopy of Trivalent Rare-Earth Ions in
Wide Band Gap Fluoride Crystals,’’ methods and techniques used in vacuum
ultraviolet spectroscopy are reviewed by Sarantopoulou and Cefalas, with special
focus on the molecular fluorine laser and VUV absorption spectroscopy. Optical
and electronic properties of wide band gap fluoride dielectric crystals doped with
trivalent rare-earth ions are surveyed, followed by a section on the VUV absorp-
tion spectroscopy of these ions. Tables provide the energy positions of the 4f n⫺15d
electronic configuration level onsets of the trivalent rare-earth ions in various
crystal hosts, together with crystal field splitting of the main 4f n electronic
configuration levels in the presence of octahedral and tetragonal symmetry
crystal fields. The laser-induced fluorescence spectroscopy of various rare-earth-
activating ions excited at 157.6 nm is presented and discussed in relation to the
active center formation with different site symmetries, direct emission and/or
weak electron–phonon interactions, emission due to repopulation via phonon
trapping and reabsorption of lattice vibrations, and spin-forbidden transitions.
In Chapter 9, ‘‘Spectroscopy of Broad-Band UV-Emitting Materials Based
on Trivalent Rare-Earth Ions’’ by Moncorgé, the latest developments in broad-
band UV laser systems are discussed and the pros and cons of using specific
rare-earth ions for different applications are evaluated. Data summarizing the

Copyright © 2002 by Taylor & Francis

positions and fluorescence lifetimes of metastable levels in different hosts doped
with rare-earth ions are presented, along with the related excited-state absorption
spectra. Color center formation, also known as solarization, following excitation
by the UV light, can be a critical problem in UV-emitting materials with direct
excitation. Such solarization effects are discussed with respect to polarization of
the optical transitions and in relation to the purity and dopant substitution pro-
cesses associated with the host crystals used.
Oldenburg and Eden, in ‘‘Generation of Coherent Ultraviolet and Vacuum
Ultraviolet Radiation by Nonlinear Processes in Intense Optical Fields’’ (Chapter
10) focus on three important processes: continuum generation, four-wave mixing,
and high-order harmonic generation. Researchers in the past decade and a half
have observed a variety of nonlinear effects that are a direct result of intense
optical fields available to them. The generation of odd harmonics when an intense
optical field interacts with an atom or molecule in the gas phase offers the poten-
tial for the development of a table-top source emitting soft X-ray radiation or
coherent XUV. In addition, harmonic generation is also an exceptional tool for
investigating the interaction between an atomic system and a strong optical field.
The authors explore all of the above issues and their ramifications.
In Chapter 11, Liu, Sarukura, and Dubinskii (a coeditor of this volume)
comprehensively review the all-solid-state, short-pulse, tunable, ultraviolet laser
sources based on Ce3⫹-activated fluoride crystals. Rare earth ions incorporated
in appropriate host crystals exhibit vibronic lasing and can be implemented for
the development of compact and widely tunable ultrashort-pulse laser sources in
the UV. Prior to detailed discussions of newer solid-state laser materials, the
authors review the existing UV laser systems that use efficient frequency conver-
sion and also those that allow direct generation of coherent UV radiation. The
spectroscopic and laser properties of Ce :LLF and Ce :LiCAF crystals are re-
viewed in depth, followed by a discussion on the generation of subnanosecond
pulses from Ce3⫹-doped fluoride lasers. Short-cavity subnanosecond pulse gener-
ation, combined with regenerative amplification, has been used in the design of
the novel so-called self-injection-seeded pulse train lasers. The authors introduce
a generalized passive version of this device, followed by detailed descriptions
of the self-injection-seeded pulse train Ce: LLF and Ce: LiCAF lasers. Tunability
data for the short-cavity Ce: LLF and Ce: LiCAF lasers are presented in the near-
UV, followed by a demonstration of tunable subnanosecond pulse generation
around 230 nm by sum-frequency generation.
Fluorescence lifetime spectroscopy requires UV pulses with pulse durations
shorter than the fluorescence lifetime of a large number of molecules. ‘‘Diode-
Pumped Picosecond UV Lasers by Nonlinear Frequency Conversions’’ (Chapter
12) addresses the critical issues related to the development of compact, efficient,
and reliable diode-pumped short-pulsed solid-state lasers that can be effectively
utilized for such fluorescence lifetime measurements. Balembois et al. present

Copyright © 2002 by Taylor & Francis

several diode-pumped sources that emit picosecond pulses in the UV. Three kinds
of techniques have been used to generate the picosecond pulses (followed by
appropriate multipass or regenerative amplification): active mode locking of a
laser cavity, gain-switching of a laser diode, and passive Q-switching of an ultra-
short-cavity microchip laser. All the sources developed use diode-pumped laser
crystals that emit in the near infrared, and the emission is subsequently converted
to UV wavelengths utilizing nonlinear crystals for harmonic generation. A pair
of laser sources has been developed, based on a Ti:sapphire regenerative ampli-
fier seeded by either an actively mode-locked diode-pumped Cr3⫹ :LiSrAlF6 pico-
second oscillator or a gain-switched picosecond laser diode. These lasers pro-
duced an average power of 1 mW in the wavelength region 270–285 nm. Another
efficient laser source described here is based on a passive Q-switched Nd :YAG
laser that seeds a multipass (specifically 2 or 4) Nd: YVO4 amplifier. Such poten-
tially portable systems hold promise for optical sampling and for the semiconduc-
tor industry, besides their obvious utility for fluorescence spectroscopy.
Ultraviolet spectroscopy can provide a wealth of information about the
Earth’s atmosphere via remote sensing techniques. In ‘‘Atmospheric Ultraviolet
Spectroscopy,’’ Chance and Rothman survey the field of UV remote sensing
measurements, with a special focus on satellite-based measurements. They pro-
vide an overview of the underlying physics, followed by a description of the data
analysis techniques used to process raw data and extract meaningful atmospheric
parameters. The HITRAN spectroscopic database has become the standard refer-
ence for absorption cross-section information related to atmospheric measure-
ments. The authors report on the progress and status of the extension of the
HITRAN database into the ultraviolet regime, and point out that the UV data are
stored in a separate directory in the HITRAN compilation and further subdivided
into cross-sectional and line parameter data. The concluding segment of Chapter
13 presents an overview of satellite UV spectrometers that have been used for
atmospheric investigations followed by a survey of current and planned instru-
mentation for atmospheric ultraviolet spectroscopy.
In the final chapter of the book, ‘‘Ultraviolet Spectroscopy in Astronomy,’’
Carruthers provides an account of the early measurements and the difficulties
encountered in recording UV spectra of astronomical objects. He then describes
the characteristics required of materials for transmission optics and those of coat-
ings for reflective optics used in designing efficient UV spectroscopic instrumen-
tation. Typical spectrograph configurations and designs are reviewed in the con-
text of space measurements and stellar observations. The evolution of ultraviolet
detectors is reviewed, covering photographic films, photomultipliers, channel
electron multipliers, and electronic imaging devices. Examples of current UV
spectrographic instruments and their applications to UV space astronomy are
presented, which include measurements with the Hubble Space Telescope, the
Hopkins UV Telescope, the Interstellar Medium-Absorption Profile Spectrome-

Copyright © 2002 by Taylor & Francis

ter, and the Far-Ultraviolet Spectroscopic Explorer, as well as Extreme-UV (be-
low about 90 nm) Spectroscopy Missions. A summary of the scientific results
from various UV space astronomy missions is provided. Different types of mea-
surements are summarized: those related to the solar atmosphere and solar activ-
ity, those involving temperatures and luminosities of hot stars, as well as measure-
ments of the composition and properties of the interstellar medium and studies
relating to extragalactic objects and cosmology.
The editors wish to acknowledge all of the contributors in putting the book
together, despite their busy schedules. We would also like to acknowledge Ms.
Barbara Mathieu’s tireless efforts in this endeavor and appreciate Mr. Russell
Dekker’s patience and support.
Professor Misra wishes to dedicate this book in fond and cherished memory
of his father, Shri Prem Krishna Misra, who always valued a good book. Dr.
Dubinskii would like to dedicate his efforts on this book with gratefulness to the
memory of his teacher, Professor S. A. Altshuller, one of the true pioneers of
electronic paramagnetic resonance and nuclear magnetic resonance.

Prabhakar Misra
Mark A. Dubinskii

Copyright © 2002 by Taylor & Francis

Contents

Preface
Contributors

Ultraviolet Spectroscopy in Chemical Physics, Combustion,

Plasma Science, and Photolithography
1. Ultraviolet and Vacuum Ultraviolet Sources and Materials for
Lithography
Vladimir Liberman and Mordechai Rothschild
2. Laser Optogalvanic Spectroscopy of Discharge Plasmas in the
Ultraviolet Region
C. Haridass, H. Major, Prabhakar Misra, and Xianming L. Han
3. Spectra of the Isotopomers of CO⫹, N2⫹, and NO in the
Ultraviolet
S. Paddi Reddy and C. Haridass
4. Photoabsorption Cross-Section Measurements in the Ultraviolet
and Vacuum Ultraviolet
Kouichi Yoshino
5. Ultraviolet and Vacuum Ultraviolet Laser Spectroscopy Using
Fluorescence and Time-of-Flight Mass Detection
R. H. Lipson and Y. J. Shi
6. Spectroscopy and Applications of Diatomic and Triatomic
Molecules Assisted by Laser Light at 157.6 nm
Alciviadis-Constantinos Cefalas and Evangelia Sarantopoulou

Copyright © 2002 by Taylor & Francis

 Spectroscopy Associated with the Search for New Ultraviolet and
Vacuum Ultraviolet Solid-State Materials for Laser Technology

7. Tunable Solid-State Ultraviolet and Vacuum Ultraviolet Lasers

Ilko K. Ilev and Ronald W. Waynant
8. VUV Laser Spectroscopy of Trivalent Rare-Earth Ions in Wide
Band Gap Fluoride Crystals
Evangelia Sarantopoulou and Alciviadis-Constantinos Cefalas
9. Spectroscopy of Broad-Band UV-Emitting Materials Based on
Trivalent Rare-Earth Ions
Richard Moncorgé

Solid-State Tunable Lasers for Ultraviolet Applications

10. Generation of Coherent Ultraviolet and Vacuum Ultraviolet

Radiation by Nonlinear Processes in Intense Optical Fields
A. L. Oldenburg and J. G. Eden
11. All-Solid-State, Short-Pulse, Tunable, Ultraviolet Laser Sources
Based on Ce3⫹-Activated Fluoride Crystals
Zhenlin Liu, Nobuhiko Sarukura, and Mark A. Dubinskii
12. Diode-Pumped Picosecond UV Lasers by Nonlinear Frequency
Conversions
François Balembois, Frédéric Druon, Patrick Georges, and
Alain Brun

Ultraviolet Spectroscopy in Atmospheric Science and Astronomy

13. Atmospheric Ultraviolet Spectroscopy

Kelly Chance and Laurence S. Rothman
14. Ultraviolet Spectroscopy in Astronomy
George R. Carruthers

Copyright © 2002 by Taylor & Francis

Contributors

François Balembois, Ph.D. Non Linear Optics Group, Laboratoire Charles

Fabry de l’Institut d’Optique, Orsay, France

Alain Brun, Ph.D. Non Linear Optics Group, Laboratoire Charles Fabry de
l’Institut d’Optique, Orsay, France

George R. Carruthers, Ph.D. Space Science Division, Naval Research Labo-

ratory, Washington, D.C.

Alciviadis-Constantinos Cefalas, Ph.D. National Hellenic Research Founda-

tion, Theoretical and Physical Chemistry Institute, Athens, Greece

Kelly Chance, Ph.D. Atomic and Molecular Physics Division, Harvard–

Smithsonian Center for Astrophysics, Cambridge, Massachusetts

Frédéric Druon, Ph.D. Non Linear Optics Group, Laboratoire Charles Fabry
de l’Institut d’Optique, Orsay, France

Mark A. Dubinskii, Ph.D. Magnon, Inc., Reisterstown, Maryland

J. G. Eden, Ph.D. Department of Electrical and Computer Engineering, and

Office of Vice-Chancellor for Research, University of Illinois, Urbana, Illinois

Patrick Georges, Ph.D. Non Linear Optics Group, Laboratoire Charles Fabry
de l’Institut d’Optique, Orsay, France

Copyright © 2002 by Taylor & Francis

Xianming L. Han, Ph.D. Department of Physics and Astronomy, Butler Uni-
versity, Indianapolis, Indiana

C. Haridass, Ph.D.* Department of Physics and Astronomy, Howard Univer-

sity, Washington, D.C.

Ilko K. Ilev, Ph.D. U.S. Food and Drug Administration, Rockville, Maryland

Vladimir Liberman, Ph.D. Lincoln Laboratory, Massachusetts Institute of

Technology, Lexington, Massachusetts

R. H. Lipson, Ph.D. Department of Chemistry, University of Western Ontario,

London, Ontario, Canada

Zhenlin Liu, Ph.D. Hosono Transparent Electro-Active Materials Project, Ex-

ploratory Research for Advanced Technology, Japan Science and Technology
Corporation, Kawasaki, Kanagawa, Japan

H. Major, Ph.D. Department of Physics and Astronomy, Howard University,

Washington, D.C.

Prabhakar Misra, Ph.D. Department of Physics and Astronomy and the Cen-
ter for the Study of Terrestrial and Extraterrestrial Atmospheres, Howard Univer-
sity, Washington, D.C.

Richard Moncorgé, Ph.D. Centre Interdisciplinaire de Recherche Ions-Lasers,

UMR 6637 CNRS-CEA-ISMRA, Université de Caen, Caen, France

A. L. Oldenburg, Ph.D. Department of Electrical and Computer Engineering,

University of Illinois, Urbana, Illinois

S. Paddi Reddy, D.Sc. Department of Physics and Physical Oceanography,

Memorial University of Newfoundland, St. John’s, Newfoundland, Canada

Laurence S. Rothman, Ph.D. Atomic and Molecular Physics Division, Har-

vard–Smithsonian Center for Astrophysics, Cambridge, Massachusetts

Mordechai Rothschild, Ph.D. Submicrometer Technology Group, Lincoln

Laboratory, Massachusetts Institute of Technology, Lexington, Massachusetts

* Current affiliation: Department of Physical Sciences, Belfry School, Belfry, Kentucky.

Copyright © 2002 by Taylor & Francis

 Evangelia Sarantopoulou, Ph.D. National Hellenic Research Foundation,
Theoretical and Physical Chemistry Institute, Athens, Greece

Nobuhiko Sarukura, Ph.D. Institute for Molecular Science, Okazaki National

Research Institutes, Myodaiji, Okazaki, Japan

Y. J. Shi Department of Chemistry, University of Western Ontario, London,

Ontario, Canada

Ronald W. Waynant, Ph.D. Electro-Optics Branch, U.S. Food and Drug Ad-
ministration, Rockville, Maryland

Kouichi Yoshino, Ph.D. Atomic and Molecular Physics Division, Harvard–

Smithsonian Center for Astrophysics, Cambridge, Massachusetts

Copyright © 2002 by Taylor & Francis

1
Ultraviolet and Vacuum
Ultraviolet Sources and Materials
for Lithography

Vladimir Liberman and Mordechai Rothschild

Massachusetts Institute of Technology, Lexington, Massachusetts

I. MOTIVATION FOR ULTRAVIOLET LITHOGRAPHY

The relentless drive of the semiconductor industry to smaller device dimensions

continues to present new challenges for all the enabling fabrication technologies,
and in particular to microlithography. At present, the most advanced microelec-
tronic devices have dimensions of 0.18 µm, and it is expected that these will be
reduced further within the next few years to 100 nm and below. The mainstay
of mass production lithography has been projection optical lithography, in which
the pattern on a photomask is imaged in reduction onto the silicon wafer. From
fundamental principles we know that the resolution in such a configuration is
directly proportional to the wavelength. Thus, the most straightforward way of
reducing printed dimensions is to use radiation sources at shorter wavelengths.
Several years ago the industry made the transition from a continuous-wave mer-
cury discharge lamp at 365 nm to a pulsed excimer laser operating at 248 nm
(KrF lasers). It is widely expected that in the near future a similar transition will
take place to 193 nm (ArF lasers). Beyond that, 157 nm (F2 lasers) is being
explored for sub-100-nm lithography. This shift in lithographic wavelengths from
the near-ultraviolet (UV) to deep-UV and eventually to vacuum-UV (VUV) poses
new challenges related to optical materials, optical coatings, detectors, and ambi-
ent control. This chapter summarizes the key issues encountered at the shorter
lithographic wavelengths, and the present state of the art in addressing them.

Copyright © 2002 by Taylor & Francis

 The definition of the terms ‘‘deep-UV’’ and ‘‘vacuum-UV’’ is not always
unambiguous. In lithography ‘‘deep-UV’’ has been applied to 248 and 193 nm,
and ‘‘VUV’’ to 157 nm. This chapter will follow this convention: wavelengths
in the region 250–190 nm are designated as ‘‘deep-UV,’’ and those between 190
and 100 nm as ‘‘VUV.’’ The name ‘‘vacuum-UV’’ for wavelengths below 190
nm is a misnomer, since the radiation can propagate in purged ambients as well,
such as nitrogen, argon, or helium. The main restriction is that molecular oxygen
be absent, since it absorbs strongly in this spectral region (see discussions in Sec.
VII). However, even at 193 nm one should avoid the presence of significant
amounts of oxygen, because discrete absorption bands do exist there and the
atmospheric attenuation can vary from wavelength to wavelength.

II. RADIATION SOURCES

In the deep-UV and VUV spectral regions there exist several radiation sources
that can be adapted to specific applications. The main use of such sources in
lithography is, of course, for the actual lithographic process (i.e., exposure of
resist-coated wafers). However, there are other needs as well: for interferometry,
absorption measurements, and at-wavelength inspection of photomasks, among
others. The lithographic radiation sources must have high power in order to facili-
tate a high throughput of the projection systems, and they must be robust, with
a low cost of ownership.
The sources that have become the de facto industry standard are excimer
lasers. These gas lasers rely on electric discharges to generate the requisite laser
population inversion. At 248 nm the radiating species are krypton fluoride mole-
cules. These are formed in their electronically excited state by gas-phase recombi-
nation of krypton and fluorine atoms in the presence of a buffer gas (helium or
neon). At 193 nm the lasing species are ArF molecules, and at 157 nm they are
F2 molecules. The ground states of the KrF and ArF molecules are dissociative,
and therefore a population inversion (and amplified emission) is established as
soon as the upper state is populated via collisions. Although this feature makes
excimer lasers efficient, it also causes the natural linewidth to be quite broad (a
few hundred picometers). In lithography the upper limit on acceptable laser
linewidth is determined by the dispersion of the lens materials and by the allow-
able chromatic aberration of the optical system. For high-numerical aperture
optics, the laser linewidth must be narrowed to less than 1 pm [1]. This task is
achieved with intracavity optical components such as prisms and gratings. Line-
narrowed excimer lasers at 248 and 193 nm are available commercially from
several vendors. Their pulse repetition rates exceed 1 kHz and are expected to ap-
proach 5 kHz in the near future. Their average power is ⬃10 W [2,3]. Exci-
mer lasers have a high gain and, therefore, the pulse duration is quite short,

Copyright © 2002 by Taylor & Francis

typically ⬃10–20 ns full-width at half maximum. Only few cavity round trips
can be generated, and the spatial coherence is consequently low (less than 100
µm correlation length) [4].
The 157 nm laser is slightly different. The lower electronic state of the
lasing transition is nondissociative, and, therefore, the natural output spectrum
has a discrete structure. It consists of two peaks, each ⬃1 pm wide, separated
by ⬃100 pm [5]. It is relatively easy to select one of the lines, using intracavity
optics, but it is quite difficult to narrow the width of the remaining line to less
than ⬃0.5 pm. In addition, the 157 nm emission of molecular fluorine is accompa-
nied by a red emission (⬃700 nm) of fluorine atoms generated in the discharge
[6]. This red emission is especially noticeable when the buffer gas is helium, and
amounts to 3–5% of the total laser power.
Solid-state coherent radiation sources in this spectral range lag significantly
behind excimer lasers with regard to power, cost, and reliability. There are no
lasers operating in the deep-UV and VUV, but there are schemes for frequency
upconversion using Nd: YAG or Ti:sapphire lasers [7]. For instance, the fifth
harmonic of a 1.06 µm Nd: YAG laser emits at 213 nm. The nonlinear media
required for upconversion are usually crystals such as barium beta-borate (BBO)
or lithium borate (LBO). They are transparent to ⬃190 nm, but their exact trans-
mission at 193 nm, as well as their resistance to laser damage at this wavelength,
are strongly determined by crystalline impurities and defects. The efficiency of
the frequency upconversion process into these short wavelengths is less than 10%.
Other upconversion schemes do not rely on nonlinear crystals but on the genera-
tion of anti-Stokes lines in a gaseous Raman-active medium, or on near-resonant
nonlinear processes in certain vapors [8]. At present none of these techniques has
been turned into commercially available systems. All of the approaches relying
on nonlinear optical processes require high peak powers of a pump laser, and
are therefore inherently pulsed sources. They also hold the promise of narrow-
linewidth, highly coherent radiation, in contrast with the excimer lasers.
Incoherent sources in the deep-UV and VUV are based on electrical dis-
charges of gases, whether continuous wave (cw) or pulsed. Weak lamps, but
covering a broad spectral range, employ low-pressure hydrogen or deuterium
molecules. They emit mostly at ⬃140–170 nm in several rotational–vibrational
bands, and they also have weak continua at wavelengths below and above this
range [9]. Deuterium lamps are frequently used in spectrophotometers. More in-
tense lamps have limited spectral range. For example, mercury lamps radiate at
254 and 185 nm, corresponding to atomic transitions in mercury. The relative
intensity of the 185 nm line varies, since it may be strongly attenuated by absorp-
tion in the window material used for the discharge envelope. Other commercially
available lamps rely on discharges in xenon (172 nm), krypton (145 nm), or argon
gas (126 nm) [10]. The spectra of these are broad, 5–10 nm, and are due to
radiative transitions in the respective dimer molecules. The lamp efficiency and

Copyright © 2002 by Taylor & Francis

its reliability decrease rapidly with wavelength. Therefore, xenon lamps are much
easier to obtain and maintain than argon lamps. Their power is nevertheless
low compared to that of excimer lasers: it is measured in milliwatts per square
centimeter, as opposed to the watts one encounters with lasers. One more wave-
length should be mentioned: 121.6 nm. This wavelength of an atomic transition
in hydrogen atoms, the so-called Lyman-alpha line, can be obtained quite effi-
ciently in discharges containing hydrogen. Several schemes have been proposed
for higher-power 121.6 nm lamps, but they have not yet been demonstrated exper-
imentally.

III. DETECTION OF UV RADIATION

The choice of the detector for the UV region is dictated by several factors related
to the incident radiation:
1. Temporal shape of incident radiation and required temporal resolution
2. Incident level of radiation and required dynamic range of the detector
3. Incident wavelength of radiation and required wavelength sensitivity
of the detector
4. Durability expectations of the detector
Four types of detectors are commonly used: pyroelectric detectors, thermopiles,
photodiodes, and photomultiplier tubes.
The operation of pyroelectric devices is based on a temperature dependence
of the electric polarization in certain classes of materials, such as ferroelectrics.
Thus, a change in temperature resulting from an incident laser pulse leads to a
voltage signal [11]. Pyroelectric detectors are well suited for direct energy and
power measurements of excimer laser radiation. For pulsed energy detection,
pyroelectric detectors can accommodate fluences as high as hundreds of mJ/cm2 /
pulse and as low as µJ/cm2 /pulse with fast amplifier circuitry. Within this range,
three to four orders of magnitude of dynamic range for pulse energy detection
can be commonly obtained. For a flat spectral response and efficient energy cou-
pling, the front detector surface may be coated with an absorbing organic paint
or a baked-on black ceramic coating. As an alternative, uncoated pyroelectric
detectors can be used, whose front surface electrode is a thin layer of chromium.
The uncoated design offers the advantage of a much faster detector response for
resolving individual pulses at repetition rates above a few hundred hertz; how-
ever, the wavelength response is determined by the spectral reflectivity of chro-
mium, and thus can vary by 15–20% across the UV spectrum. When operating
pyroelectric detectors in fluence regimens below the material damage threshold,
the detector durability is determined by the lifetime of the front coating or a front
surface electrode. For instance, in recent tests performed at the Massachusetts

Copyright © 2002 by Taylor & Francis

Institute of Technology (MIT) Lincoln Laboratory at 157 nm, a commercial pyro-
electric detector was observed to fail after 3 ⫻ 109 pulses at an incident fluence
of 0.5 mJ/cm2 /pulse. The mode of failure appeared to be delamination of the
front surface electrode [12].
The pyroelectric response of a detector may be a function of many factors,
including crystal size and type, the value of the pyroelectric coefficient, and its
dependence on temperature. Therefore, the absolute calibration of pyroelectric
detectors is not always straightforward. At present, the National Institute of Stan-
dards and Technology (NIST) offers calibration services for pyroelectric detec-
tors at 248 and 193 nm with an accuracy of ⫾ 1% [13].
Thermopile devices are useful for measuring continuous wave (CW) radia-
tion sources or average power of pulsed UV radiation. A thermopile consists of
an array of thermocouple or thermistor devices that convert incident power into
a temperature rise. Because of their simplicity, thermopile probes are consider-
ably less expensive than pyroelectric devices. They can measure intensities from
several kilowatts to milliwatts per cm2. The front surface of a thermopile is typi-
cally a black, highly absorptive coating, with a spectral response constant within
a few percent over the UV spectral region [14]. The main utility of thermopile
probes is the direct determination of incident laser power and laser stability. With
NIST calibration, the measurement accuracy can be better than 2% [13]. The
most common damage mechanism involves ablation of the black coating. The
coating can be reapplied by a detector manufacturer at a nominal cost.
Photodiodes are an excellent choice for detecting highly attenuated pulsed
excimer laser radiation when high temporal resolution (better than 1 ns) and wide
dynamic range are required. Photodiodes operate on the principle of converting
incident photons into electron–hole pairs, which are swept to the opposing ends
of a p–n junction, generating current at the device terminals. Diodes for UV
applications can be made from a variety of semiconducting materials, including
Si, PtSi, and GaP [15]. Photodiodes exhibit excellent linearity over several orders
of magnitude: pulsed UV laser energies from under 1 nJ/cm2 to hundreds of µJ/
cm2 can be measured. The external wavelength sensitivity of the photodiode is
dependent on the surface reflectance, which is a function of the substrate’s optical
constants. In addition, the internal quantum efficiency of the device increases
with photon energy. The combined effect is a varying spectral responsivity. For
instance, for a typical Si-based p–n junction device, a 50% increase in overall
sensitivity may be expected as the wavelength is reduced from 248 to 157 nm.
Furthermore, as the device ages under UV laser exposure, its wavelength respon-
sivity may change [16].
For metrology applications requiring extremely low-level signal detection,
such as transmittance, reflectance, and ellipsometry measurements in the ultravio-
let, photomultiplier tubes (PMTs) offer the best solution. In a popular configura-
tion the PMT window is coated with a fluorescing material, such as sodium salyci-

Copyright © 2002 by Taylor & Francis

late. Ultraviolet light striking the sodium salycilate will cause it to fluoresce at
a wavelength of approximately 400 nm, which can in turn be detected by a PMT
designed for the visible spectral region [17]. For ultimate detection of low signals,
UV-sensitive solar-blind PMTs are used. These have photocathodes insensitive
to visible radiation, such as CsI or CsTe. PMTs have constant responsivities over
thousands of hours when used with low-level CW radiation. They are not gener-
ally used for measuring pulsed UV laser radiation since both sodium salycilate
and photocathode materials can be easily damaged with high peak powers.

IV. UV OPTICAL METROLOGY

A. Bulk Materials
In discussing UV optical metrology relevant to microlithography, it is convenient
to classify the topic into two parts: measurement of bulk material properties and
surfaces, and measurement of thin film properties. Bulk materials relevant to UV
microlithography would include lens materials, such as fused silica, single crystal
fluorides (calcium fluoride, barium fluoride, and others), and photomask materi-
als, such as fused silica. For lens materials, important optical properties include
transmission and loss at the lithographic wavelength, and highly accurate refrac-
tive index values. The transmission and optical losses of the lens materials have
immediate impact on the overall system transmission and throughput. Further-
more, residual absorption causes lens heating, which, through the temperature
dependence of the index of refraction, can cause unacceptable lens aberrations.
An accurate knowledge of the index of refraction, its temperature dependence,
and its dispersion is essential to the design and tolerancing of near-diffraction
limited optics such as those used in lithographic systems.
The transmission and maximum loss requirements are a sensitive function
of a particular lens design. For 193 nm and 157 nm applications, a consensus
[12] has been developed that acceptable losses in a bulk material should not
exceed 0.5%/cm at the wavelength of use, which is expressed as a maximum
bulk absorption coefficient of 0.002/cm, base 10.
When prescreening optical materials, one must separate bulk losses, such
as bulk absorption and scatter, which are intrinsic to the material, from surface
losses, which are a function of surface polishing and ambient conditions. For a
sample of length l the internal transmission is:

Tint ⫽ 10⫺[2β⫹(α⫹γ)l] (1)

where β refers to the loss of a single surface, including absorption and scatter;
α is the internal absorption coefficient per unit length; and γ is the bulk scatter

Copyright © 2002 by Taylor & Francis

coefficient per unit length [18]. Experimentally, one measures the external trans-
mission, which includes the internal transmission as well as surface reflections
due to index mismatch (Fresnel losses). The internal transmission can then be
approximated by:

Tint ⫽ Text(n ⫹ 1)4/(16n2) (2)

where n is the refractive index of the material. The procedure for obtaining bulk
losses involves measuring the transmission of several samples of different
lengths, all with the same nominal surface finish. Then, applying Eqs. (1) and
(2) above, one derives α ⫹ γ. In order to obtain the bulk absorption coefficient,
separate bulk scatter measurements must be performed. Additional metrology
complications may arise because the material characteristics can change in the
course of laser irradiation: surface losses may change due to laser cleaning or
contamination and bulk losses may vary due to laser induced bleaching or absorp-
tion. Thus, when long-term irradiation experiments requiring a high degree of
accuracy are needed, it is desirable to irradiate several samples of the same grade
and different lengths simultaneously.
Because of the expected high initial transmission of lens materials, samples
at least several centimeters in length need to be measured to obtain appreciable
loss signals. Commercial spectrophotometers are not well suited for measuring
longer samples. This is due in part to the limited length of the sample compart-
ment, and in part is a result of the high divergence of the spectrometer beam.
From our experience at MIT Lincoln Laboratory, the above issues can cause
transmission measurement errors ⬎2% for an 8 cm-long sample at wavelengths
below 200 nm. This measurement discrepancy of ⬎0.25%/cm is a significant
fraction of the specification of initial material transmission.
An alternative, more accurate, and precise technique of transmission mea-
surements of long bulk samples utilizes an excimer laser light source. This tech-
nique is especially useful in conjunction with long-term materials durability stud-
ies, since transmission measurements can be performed in situ [18]. A typical
measurement setup (Fig. 1) involves using a beamsplitter before the sample to
reflect part of the incident beam onto a reference detector. The main part of the
laser beam is incident on the sample detector, which is positioned directly behind
the sample station. Measurements are obtained ratiometrically. For the incident
laser fluences of interest (from 0.1 to 10 mJ/cm2 /pulse), pyroelectric detectors
are well suited to the task of transmission measurements. Most material evalua-
tion studies for UV lithographic lens applications are performed using the laser
ratiometry method described above, whether they are carried out by the material
suppliers, by independent laboratories, or by lithographic lens manufacturers. In
using pulsed laser ratiometry for in situ single-wavelength transmission measure-

Copyright © 2002 by Taylor & Francis

Figure 1 Schematic representation of an in-situ laser-based transmission measurement
scheme. A. Measurement configuration for obtaining ‘‘100%’’ transmission. B. Measure-
ment configuration for obtaining sample transmission.

ments, one must be careful to separate potential two-photon absorption from lin-
ear absorption. The total absorption coefficient is
α tot ⫽ α 0 ⫹ βI (3)
where α0 is the linear absorption coefficient, β is the two-photon absorption coef-
ficient, and I is the peak intensity of the laser pulse. The two-photon absorption
coefficient can be obtained by plotting the total absorption coefficient as a func-
tion of incident peak intensity and obtaining the slope of the linear regression.
Typical two-photon absorption measurements for either fused silica or calcium
fluoride at 193 nm yield β ⫽ 0.5 ⫺ 1 cm/GW (base 10) [19–22]. For a pulse
width on the order of 10 ns and a fluence of 10 mJ/cm2 /pulse, βI ⫽ 0.0005–
0.001/cm (base 10), compared to the specification of αtot less than 0.002/cm (base
10). Thus, in order to keep the effect of two-photon absorption below ⬃25% of
the total measured value, fluences below 10 mJ/cm2 /pulse should be used in laser
ratiometry.
Microlithographic projection systems are designed to have near-diffraction
limited performance, with very small wavefront aberrations. Their design is sensi-
tive to the exact value of the lens material’s index of refraction, which must be
known to within 10 ppm. The most accurate measurements to date of the index
of refraction of materials transparent in the UV region have been obtained by
the minimum deviation method [23].

Copyright © 2002 by Taylor & Francis

B. Thin Films
Accurate knowledge of the optical properties of thin films is important to many
areas of microlithography. For instance, a lithographic lens manufacturer may
want to develop a variety of multilayer thin film coatings applied to lens surfaces
that will either eliminate surface reflections or allow the element to reflect a
prescribed amount, depending on the particular lens design. Although the princi-
ples of designing such thin film coatings are well understood, the optical proper-
ties of the constituent layers often are not. This lack of information has hampered
the development of high-performance coatings in the UV wavelength range.
Many lithographic applications require knowledge of the refractive index
and absorption coefficients of thin films, in addition to reflection and transmission
data. For instance, knowledge of the refractive index of thin films for lens coat-
ings is needed for precise coating designs. Another application is the tailoring
and modeling of photoresist performance, which require knowledge of the optical
constants of resists and their constituents (such as photoacid generators). Similar
data are needed for the optimization of bottom antireflectance coatings (BARCs),
which are applied to the wafer surface prior to the deposition of photoresists to
suppress back reflections into the resist. The extraction of optical constants of
thin films is best performed with ellipsometry, as discussed in Sec. IV.B.2 below.

1. Transmittance and Reflectance Measurements

For the lifetime testing of optical coatings, accurate spectrophotometric reflec-
tance and transmittance measurements need to be performed periodically. The
stringent demands of the material’s durability impose strict requirements on the
measurement reproducibility. For instance, for durability studies of antireflec-
tance coatings and high reflector coatings, both transmission and reflectance over
the UV spectral range may need to be recorded over periods of months with
repeatability around 0.1%. In addition, it is not sufficient to record single-wave-
length data that are available with a laser-based measurement (as is the case
with bulk materials), since useful information about degradation mechanisms of
coatings can be learned only from their full spectral behavior.
Transmission and reflection measurements in the near-UV spectrum can
be performed to high accuracy and precision, but special challenges arise for
metrology below 200 nm. Both water vapor and oxygen absorb strongly in this
wavelength region and therefore the instruments must either be purged with a
transparent gas or operated under vacuum. The ambient constraints complicate
the instrument design, and significantly increase the times needed for sample
transfer and measurement. All the transmitting windows, bulb envelopes, and
photomultiplier windows must be made either of UV-grade synthetic fused silica
or, for operation below 180 nm, of magnesium or calcium fluoride. If polarizing

Copyright © 2002 by Taylor & Francis

elements are needed in the VUV, the most common option is Rochon prisms
made of optically contacted magnesium fluoride elements. They are designed to
deflect the extraordinary ray by several degrees while letting the ordinary ray
pass straight through.
A simple VUV spectrometer design utilizes a ratiometric method to normal-
ize bulb intensity fluctuations during a wavelength scan. In that configuration,
light from a monochromator is incident on a beamsplitter, where a part of the
beam is deflected onto a reference detector and the rest of the beam is used for
sample transmission measurement. By contrast, in a dual beam–single detector
system the two beams are formed using a chopper wheel. During a wavelength
scan, for every chopper cycle both a reference and a signal scan are obtained.
The technique produces a superior signal-to-noise ratio. The dual beam–single
detector designs are quite common for near-UV and visible spectrometers, but
they have only recently been incorporated into VUV spectrometers.
In addition to the challenges inherent in obtaining precise readings in the
deep-UV, accuracy is difficult to achieve since there are no primary transmission
or reflection standards in this wavelength range. The difficulty in developing such
standards in the deep-UV stems from the fact that the optical properties of materi-
als are not known with sufficient accuracy, and surface losses due to polishing
or surface contamination can significantly influence measurements below 200
nm (see Sec. VII) [24]. Therefore, standards for sample preparation and surface
treatment need to be established as part of every measurement procedure. The
impact of surface cleaning on the transmission and reflection measurements has
only recently been assessed, and a considerable amount of work must be per-
formed before reliable transferable standards are developed.
Several specular reflectance configurations for high accuracy and precision
measurements have been successfully implemented in the deep-UV and VUV.
Integrating spheres, while popular at longer wavelengths, are not practical below
200 nm because of the lack of a suitable coating reflector material. In the absence
of reflectance standards for the VUV range, one needs to use self-referencing
schemes, including goniometer-based measurements and use of absolute reflec-
tance accessories, such as V–W and V–N methods. In the latter, the nomenclature
describes the beam path traced by a light beam in the baseline and reflectance
measurement positions, respectively. The advantage of goniometer over fixed-
angle setups is an ability to measure a wide range of specular reflectance signals.
However, this arrangement is not compatible with so-called straight-through
single-detector spectrometers. For the single-detector schemes, V–W and V–N
methods can be used (Fig. 2). Both methods utilize movable mirrors to ensure
that, for both baseline and reflectance measurements, the same number of mirrors
direct the light beam and the path length is identical for both baseline and sample
measurements.

Copyright © 2002 by Taylor & Francis

 (A) (B)

Figure 2 Geometric representations of V-N (A) and V-W reflectance (B) methods. MF,
fixed mirror; MR, rotatable mirror; MM, movable mirror.

2. Ellipsometry
Although accurate measurements of reflection and transmission of thin films, as
described above, can be used to extract the material optical constants [25], vari-
able angle spectroscopic ellipsometry (VASE) is better suited for such purposes.
Ellipsometric parameters relate to a ratio of two reflectivities of different polariza-
tions; thus, the measurements are self-normalizing. Additionally, the phase
change upon reflection, obtained in ellipsometric measurements, is sensitive to
the presence of very thin films and to the interface structure and therefore can
yield valuable information in these areas [26,27]. The extension of VASE to
VUV encounters certain practical challenges, similar to those discussed above
in the context of spectrophotometry: sources, coating and window materials, po-
larizers and detectors need to be engineered or optimized for this wavelength
range. Ellipsometers must work in a purged or vacuum ambient; thus, instrument
access is limited, and automation of the alignment procedure must be engineered.
Significant progress has been made recently towards the design of purged UV

Copyright © 2002 by Taylor & Francis

VASE, and commercial systems are now available with comparable performance
to that obtained in the near-UV [28,29].

V. OPTICAL MATERIALS

There is a limited choice of optical materials in the deep-UV and VUV that would
both meet the initial requirements of the excimer-laser based lithographic systems
and exhibit the required lifetime durability. For instance, some materials with
excellent damage resistance, such as crystalline quartz, magnesium fluoride, and
sapphire, are intrinsically birefringent and are thus unsuitable for a lithographic
system that is polarization-sensitive. Coupled with the initial requirements of low
birefringence, candidate materials must exhibit excellent index homogeneity and
must be available in sufficient quantity to meet the demand of the semiconductor
industry. The only practical materials that could meet the preceding requirements
are amorphous fused silica and crystalline fluorides, primarily calcium and bar-
ium fluorides.

A. Fused Silica
The fused silica used in UV lithography applications is of very high purity. All
the trace element impurities are in the sub-100 parts per billion (ppb) level. Hy-
droxyl groups and hydrogen are present in various amounts in 193 nm and 248
nm grade fused silica, because they apparently improve the resistance of fused
silica to laser induced damage. Fused silica for 157 nm applications is somewhat
different: since OH absorbs at this wavelength, the material must be ‘‘dried.’’
Instead of the hydroxyl groups that are removed, fluorine atoms are added to the
network. This process extends the transmission edge of the fused silica to below
157 nm, since fluorine titrates absorbing defects and dangling bonds.
The high-purity fused silica represents an atomic structure very close to
the ideal continuous random network, comprised of perfect Si(O1/2)4 tetrahedra
(Fig. 3A) joined at the corners with statistically determined Si–O–Si and dihedral
angles [30,31]. In an amorphous fused silica the tetrahedra form closed rings
with varying numbers of members, from three tetrahedra and up (Fig. 3B).
Modeling the absorption band-edge of fused silica by an Urbach rule for
near band-edge absorption by excitons [32] predicts a lower absorption limit of
⬃10⫺5 /cm at 193 nm. However, the most transparent fused silica that has been
measured exhibits much larger absorption coefficients, in the range of 10⫺3 /cm.
This discrepancy between expectation and experiment indicates that the transpar-
ency of fused silica in the deep UV is limited by defects, whether intrinsic or
radiation-induced, by OH impurities, and by bulk scattering (Fig. 4) [30].

Copyright © 2002 by Taylor & Francis

 (A) (B)

Figure 3 A. Basic structural unit of silicon dioxide. B. Two-dimensional representation

of the amorphous fused silica. (From Refs. 30, 31.)

Several types of lattice defects contribute to the deep-UV absorption of

fused silica. One is the nonbridging oxygen hole center, the spectral signature
of which is an absorption band at ⬃4.8 eV (260 nm). This defect can be repre-
sented by the notation DSi–O⋅, where the three parallel lines represent three
separate bonds to oxygen atoms, and the dot represents an unpaired electron. The
width of the band is broad enough to overlap the 248 nm excimer laser line.
However, this defect does not contribute to absorption at 193 nm and below.
The second defect, the E′ center, can be represented by the notation DSi⋅,
and refers to an unpaired electron on a silicon atom. The absorption band caused
by this defect is centered at 5.8 eV (215 nm) and is broad enough that it overlaps
the 193 nm laser line. The laser-induced formation of E′ centers is believed to
be the dominant cause of the observed degradation in deep-UV transmission
when fused silica is irradiated with 193 nm lasers.
Scattering-induced transmission losses in fused silica arise from statistical
density fluctuations. The general form of this scattering loss, γ, can be expressed
by [33]
γ ⬀ (n8/λ4)p2Tf β T (4)
where n is the refractive index, λ is the wavelength, p is the photoelastic constant,
Tf is the fictive temperature, and βT is the isothermal compressibility at Tf. The
contribution of the bulk scattering losses in the deep-UV can be extrapolated
from total integrated scatter measurements performed in the near-UV and visible,
using the expression above to correct for wavelength and dispersion effects. From
those measurements, a bulk scatter loss of γ ⬇ 10⫺3 /cm at 193 nm has been
derived [18]. Thus, for the high-purity fused silica, bulk scatter is comparable to
absorption from intrinsic defects.
The most critical optical elements in a lithographic system are those form-
ing the projection optics. It is expected that in a manufacturing environment they

Copyright © 2002 by Taylor & Francis

Figure 4 Attenuation coefficients of amorphous fused silica as a function of wavelength
and energy. (From Ref. 30.)

will be exposed to fluences up to ⬃0.1 mJ/cm2 /pulse for 4–10 ⫻ 109 pulses per
year. The fluences are low enough that the electric fields are orders of magnitude
below the threshold for dielectric breakdown in fused silica. They are also low
enough that, coupled with the low absorption coefficient, the induced temperature
rise is only a few degrees even at high pulse repetition rates. Therefore, any laser-
induced changes are not caused by either breakdown or thermal effects. Instead,
there may be small photo-induced processes whose cumulative effect is notice-
able only after large pulse counts. Several laser-induced phenomena in fused
silica have been reported, but their mechanistic details and even their scaling

Copyright © 2002 by Taylor & Francis

laws are not yet fully understood. It has been previously observed that 248 nm
irradiation at higher fluences of initially transparent fused silica causes a sudden
drop in transmission after a few million pulses, a phenomenon termed strong
absorption transition (SAT) [34]. This effect has not been documented with lasers
below 248 nm, nor has its fluence scaling been reported. Other changes that occur
in fused silica after prolonged irradiation are a gradual loss in transmission and
a laser-induced structural rearrangement of the fused silica network, which can
manifest itself as densification. All the above phenomena adversely affect the
performance of optical systems by changing either their transmission or their
effective optical path length.
Our description of irradiation-induced damage of fused silica will be struc-
tured in two parts: high-dose damage studies of fused silica for 193 nm lens
materials applications, and lower-dose irradiation studies of modified (fluorine-
doped) fused silica for 157 nm photomasks. The behavior of fused silica at 248
nm is qualitatively similar to that at 193 nm but the damage rates are about one
order of magnitude smaller.

1. Fused Silica for 193 nm and 248 nm Applications

Most earlier reports of UV laser-induced damage to bulk fused silica are based
on studies performed at fairly high fluences (10 2 mJ/cm2 /pulse) pulse counts (on
the order of 10 ⫻ 106 pulses) [31,35–46]. More recent damage studies of fused
silica are specifically aimed at understanding degradation effects relevant to mi-
crolithography (i.e., those observed in the lower-fluence, higher pulse count re-
gimes) [18,47–51]. At MIT Lincoln Laboratory we surveyed 10 grades of fused
silica from six commercial suppliers designed for 193 nm [50]. All were desig-
nated as lithographic grade. To increase the pulse accumulation rate the irradia-
tion test bed used 193 nm radiation from two free-running 400 Hz excimer lasers,
achieving an effective 800 Hz pulse repetition rate. The incident laser beams
were temporally interleaved and spatially overlapped by a 50/50 beam splitter
and subsequently distributed along 12 beamlines to deliver an incident laser flu-
ence in the range 0.25–4 mJ/cm2 /pulse. More than 100 different samples were
tested. For the material with the highest transmission, lifetime durability tests
were extended to pulse counts on the order of 109.
The details of material growth were not made available, in order to protect
the suppliers’ confidentiality. The study aimed instead to assess the state of fused
silica as it applies to serving the needs of the lithographic community by deriving
general trends in degradation behavior and analyzing the spread among various
grades.
Induced Absorption. For the samples tested, the initial 193 nm bulk ab-
sorption coefficient was found to vary from 0.001 to 0.01/cm (base 10), de-
pending on the grade tested. The majority of the samples had initial absorption

Copyright © 2002 by Taylor & Francis

coefficients below 0.002/cm (base 10), which is a target value for maximum
acceptable absorption derived in consultation with lithographic lens manufactur-
ers. Upon irradiation, several different trends were observed (Fig. 5). Most sam-
ples displayed varying degrees of transmission recovery at the onset of irradia-
tion. Depending on the amount of recovery, this phenomenon could be either
laser-induced surface cleaning (1–2% total per sample, length-independent) or
bulk bleaching (up to 2% per cm, scaling with sample length). Thereafter, the
193 nm absorption coefficients were found either to reach a plateau or exhibit
slow continuous growth. No SAT phenomenon was observed for any sample for
the largest doses exposed (5 ⫻ 109 pulses at 1 mJ/cm2 /pulse).
In general, the details of the degradation mechanism of fused silica under
low-level irradiation can be a fairly complex function of small amounts of impuri-
ties in the glass, such as OH and H2. A recent study has correlated the degree
of damage under similar irradiation condition to the amount of hydrogen dis-
solved in the glass network [48]. The time-dependent absorption behavior has
been modeled as an interplay of two competing processes: activation of pre-
existing defects into E′ centers by incident photons and quenching of these defects
by dissolved hydrogen to form nonabsorbing SiH bonds [48].

Figure 5 193 nm absorption coefficients of several grades of fused silica as a function

of a number of laser pulses (in millions). The incident fluence is 1 mJ/cm2 /pulse. (From
Ref. 50.)

Copyright © 2002 by Taylor & Francis

 The absorption measurements shown in Figure 5 represent a permanently
induced absorption. However, fused silica also exhibits a temporary reduction in
transmission following initial exposure to laser irradiation, and a partial recovery
in transmission once the laser is turned off [40,48,52]. Typical time resolved
absorption data for a fused silica sample are shown in Figure 6. [53]
Laser-Induced Structural Changes: Densification and Rarefaction. It has
been reported for several decades that, when exposed to prolonged low levels
of ionizing and nonionizing radiation, fused silica undergoes densification, as
manifested by an increase in its refractive index in the irradiated area. This phe-
nomenon has been observed to occur also with the subbandgap irradiation of
excimer lasers. The somewhat indirect observations of densification are small,
typically a few parts per million (ppm). Because of their magnitude it has not
been possible to obtain direct evidence of the structural rearrangement of the
lattice or the density increase. Interferometric measurements of wavefront retar-
dation or measurements of stress-induced birefringence through the photoelastic
effect are used instead to measure the amount of densification. Nevertheless, the
impact of material densification, even in the ppm range, on the performance of
a high-quality lithographic system can be devastating, causing unacceptable
wavefront aberrations. Over a lens lifetime, it is estimated that a maximum of 1
ppm densification may be tolerated.

Figure 6 Temporal behavior of 193 nm fused silica absorption as the incident laser is
turned on and off. Data shown are obtained for an 8-cm long sample at an incident fluence
of 0.25 mJ/cm2 /pulse. (From Ref. 53.)

Copyright © 2002 by Taylor & Francis

 It has been proposed that UV laser-induced densification follows a power
law. If N is the pulse count and the fluence is Φ, then the fractional densification
δ is
δ ⫽ k(NΦ 2/τ)x (5)

where k and x are empirically derived constants, with x typically between 0.5
and 0.8, and τ is a measure of effective laser pulsewidth, assuming that a two-
photon effect is operative [54]. This power law has been confirmed experimen-
tally in long-term studies performed at MIT Lincoln Laboratory and elsewhere
[19,39,50,51,55,56]. The irradiation was typically with unpolarized light, at flu-
ences of ⬍0.5–100 mJ/cm2 /pulse, and for pulse counts up to 5 ⫻ 109. Some of
these studies reported that k and x as defined in Eq. (5) varied somewhat between
grades and with incident fluence levels.
To extend the applicability of the densification scaling to the lifetime doses
of lithographic tools (40–100 ⫻ 109 pulses at 0.1 mJ/cm2 /pulse), an ultramara-
thon irradiation study has been recently performed [57]. It used time-delay pulse
multiplexing of a 2 kHz lithography-grade 193 nm laser beam for an effective
operating rate of 8 kHz. The study used polarized light and periodically assessed
both interferometric wavefront distortion and stress-induced birefringence of the
irradiated fused silica samples. Rather than showing conventional densification
through increase in the optical pathlength, the study revealed the opposite effect:
rarefaction. The magnitude of this effect was about 0.5 ppm/cm after 40 ⫻ 109
pulses. Early studies of fused silica did detect expansion under ionizing irradia-
tion [58–60], but this is the first demonstration of such a phenomenon with UV
laser light. The accompanying birefringence signal did not suggest a development
of a symmetrical stress field that would be induced by a round laser spot, but
exhibited anisotropy that seemed to be induced by the linear polarization of the
laser. While at present the laser-induced rarefaction phenomenon is not well un-
derstood, it is clearly relevant to attempts to predict the practical lifetime of litho-
graphic lens optics.

2. Modified Fused Silica

Even the highest-quality 193 nm fused silica is not transmissive enough at 157
nm. However, in the last year significant progress has been made to improve its
transmission substantially in the 155–175 wavelength region by modifying its
composition [61,62]. The general approach consists of two steps: reduce the hy-
droxyl content of fused silica, since the OH moieties absorb at wavelengths above
about 175 nm; and introduce fluorine into the glassy network, in order to titrate
all the unpaired electrons on the silicon atoms. Figure 7 shows the transmission
of conventional UV-grade fused silica and modified fused silica [63]. The trans-
mission of a 1 cm thick modified fused silica is about 82% at 157 nm, as measured

Copyright © 2002 by Taylor & Francis

Figure 7 Transmission comparison of standard UV grade fused silica (dot-dash) and
a modified fused silica (solid line) samples.

in a vacuum-UV spectrometer. This value should be compared to the theoretical

maximum of 88%, obtained from the index of refraction and by allowing for only
Fresnel reflection losses. It thus appears that there still is a residual absorption of
⬃6% in the sample measured. In fact, 3–4% losses can be attributed to surface
losses that are removed under laser radiation. The intrinsic absorption coefficient
of the modified fused silica can be as low as 0.02/cm (base 10) at 157 nm.
The primary application of modified fused silica is as a transmissive sub-
strate for photomasks at 157 nm. The requirements for good mask material are
more relaxed than those for lens materials. High initial transmission needs to be
ensured only for a 6 mm thickness (the standard mask thickness), and lifetime
requirements must be met only for ⬃60 ⫻ 106 pulses at 0.1 mJ/cm2 /pulse. Several
modified fused silica samples were evaluated at MIT Lincoln Laboratory, and
some met the criteria for low initial absorption as well as for stable transmission
at the pulse counts and fluences listed above [12].

B. Calcium Fluoride
The lattice structure of crystalline calcium fluoride is shown in Figure 8. Each
fluoride ion is tetrahedrally bound to four calcium ions. Calcium fluoride has
excellent transmission down to 130 nm. It is inert, has very low water solubility,
and can be grown in large-diameter boules. Because of its excellent resistance
to high-power laser irradiation, the material has found extensive use in illuminator

Copyright © 2002 by Taylor & Francis

Figure 8 A unit cell of calcium fluoride. Light balls, F⫺ ions; dark balls, Ca2⫹ ions.

designs for 193 nm lithographic systems. The growth in material processing and
polishing infrastructure has positioned CaF2 as the primary candidate for lens
materials for 157 nm lithography as well.
Early studies assessing transmission and radiation-induced color centers
in undoped calcium fluoride had difficulties separating impurity-induced from
intrinsic defects [64–66]. Nevertheless, the most often seen intrinsic color centers
are the F center, a single electron trapped in a fluorine vacancy; and the M center,
an aggregate of two nearest F centers. The absorption peak of the F band is at
⬃380 nm, while the M center is dichroic with the two main peaks at 360 nm
and ⬃550–600 nm (see Fig. 9). Since the intrinsic color centers are sufficiently
far from 193 nm, the transmission properties at that wavelength and at 157 nm
are limited by impurities in the material. In fact, long-term low-fluence 193 nm
durability studies of calcium fluoride showed that the initial absorption first in-
creases and then saturates [18]. Similar long-term studies under 157 nm irradia-
tion revealed no added 157 nm absorption for pulse counts in excess of 1 ⫻ 109
[12,63]. Although considerable variations exist among calcium fluoride suppliers,
total absorption coefficients (initial plus laser induced) of less than 0.003 cm⫺1
(base 10) can be readily obtained for a number of materials at both 193 and 157
nm wavelengths.
Due to its single-crystal nature, calcium fluoride does not undergo radia-
tion-induced density changes such as those observed in fused silica. Furthermore,
bulk scattering in the material is at least an order of magnitude less than that of
fused silica. The main challenges in producing CaF2 for lithographic lens applica-
tions include the ability to grow large-diameter crystals (⬎200 mm) with no
dislocations, perfect crystallinity over the full open aperture, low stress-induced

Copyright © 2002 by Taylor & Francis

Figure 9 Color centers in CaF2 induced by 193 nm irradiation.

birefringence (⬍1 nm/cm at 633 nm), and good index homogeneity (better than
1 ppm at 663 nm).

C. Other Fluorides
For projection lens systems utilizing an all-refractive optical design, a second
lens material would be required to provide achromatization across the spectral
bandwidth of the 157 nm excimer laser (⬇1 pm). The second material option is
being pursued in addition to engineering the lasers to bandwidths below ⬇0.5
pm. Such second refractive lens materials must meet the same stringent require-
ments as CaF2: transmission, durability, index homogeneity, low stress birefrin-
gence, among others. The material should also have good mechanical, thermal,
and environmental stability.
Table 1 summarizes key properties of several potential material candidates.

Copyright © 2002 by Taylor & Francis

Table 1 Relevant Properties of Potential Lens Materials for 157-nm Lithography

Index
Refractive dispersion, Thermal
index at dn/dλ dn/dT in N2 OPD Conductivity
Material 20°C (/nm) (ppm per °C) (⫻ 10⫺6 /deg) (W/m/K) Comments

CaF2 1.559 ⫺0.0026 8 12 9.7 Primary choice for lens material.

Modified SiO2 1.654 ⫺0.0050 39 24 1.4 Marginal absorption (⬎0.02/cm),
long-term radiation durability
appears satisfactory.
LiF 1.486 NA 0–4 12–15 11.7 Optical quality, radiation durabil-
ity, polishing may pose chal-
lenges.
SrF2 1.576 ⫺0.0031 9 12 8.5 Index dispersion is marginally dif-
ferent from that of CaF2.
BaF2 1.657 ⫺0.0044 10.6 13.6 7 Index and dispersion complement
CaF2. Transmission and laser
damage must be further tested.
Prone to thermal shock.

All wavelength-dependent data are at 157.6 nm. All material properties are from Ref. 67, with the exception of LiF, which is from Ref. 69.

Copyright © 2002 by Taylor & Francis

All the materials listed are nonbirefringent and are expected to have good trans-
mission properties at 157 nm. The optical pathlength difference (OPD) shown
in Table 1 is defined as
OPD ⫽ 1/n[(n ⫺ 1)α ⫹ dn/dT] (6)
where α is the coefficient of thermal expansion. The highest-performing modified
fused silica has transmission somewhat inferior to that of single crystal fluorides.
It also has a thermal conductivity about an order of magnitude lower than that
of fluorides, resulting in a higher temperature rise per unit incident laser flux. Its
OPD per degree is the highest of the materials listed because of its high thermo-
optic coefficient dn/dT at 157 nm. These disadvantages, in addition to marginal
initial transmission, would limit the use of modified fused silica to very few
elements in an optical train. Of the remaining materials listed, barium fluoride
is the best candidate. It has sufficiently different dispersion from that of calcium
fluoride to enable achromatization over ⬃1 pm [67], high thermal conductivity,
and good radiation durability when exposed to gamma rays [68], and in initial
157 nm durability testing [12,63]. However, an infrastructure for growing large
sizes of this material still needs to be developed. Techniques for polishing and
figuring of this material to the required tolerances have not yet reached maturity,
nor have low-stress birefringence and index inhomogeneity been demonstrated.
Strontium fluoride has optical properties intermediate between those of cal-
cium fluoride and barium fluoride. In recent 157 nm tests it showed promising
initial transmission and laser durability results [12]. However, refractive index
and dispersion measurements performed at NIST [67] suggest that the index dis-
persion of SrF2 may not be sufficiently different from that of CaF2 to warrant its
use as a second refractive material. Thus, there is not likely to be a strong demand
for strontium fluoride for UV lithographic applications.

VI. OPTICAL COATINGS

Dielectric coatings applied to lithographic elements can be classified into three

main categories: antireflective designs for throughput enhancement and ghost
image suppression, high reflectors for beam steering purposes; and partial
reflectors/beamsplitters employed primarily in catadioptric designs. The coating
stacks will consist of layers of alternating high- and low-index materials, as com-
pared to the substrate index. The layers must be transparent enough at the wave-
length of use to enable an efficient design and to ensure good damage resistance.
For 157 nm lithography the choice of coating materials will be primarily limited
to fluorides, since most oxide compounds are too absorptive. For instance, low-
index materials may include magnesium and aluminium fluorides, while high-
index materials may include lanthanum and gadolinium fluorides. For 193 nm

Copyright © 2002 by Taylor & Francis

the choice of materials is more relaxed; oxides may also be used, in addition to
fluorides. Therefore, the oxides of aluminum, hafnium and zirconium may also
be considered as high-index candidates, while for low-index materials silicon
dioxide is a possible choice.
In addition to the optical properties of the materials used in thin film coat-
ings, the microscopic structure of the films plays an important role. Thin films
tend to be more porous than bulk materials, and airborne molecules may adsorb
onto the surfaces of the micropores. It was previously observed that water can
incorporate into thin films, thus changing their effective index of refraction and
thereby shifting the optical response of multilayers. This effect occurs even at
wavelengths at which water is transparent (400–700 nm) [70], and it would be
even more pronounced below 180 nm, where water also absorbs radiation. This
effect may be mitigated by increasing the film deposition temperature, since the
packing density changes from ⬃70% at room temperature to over 90% at 300°C
[71]. However, higher deposition temperatures are not always practical because
of thermally induced stresses at the substrate–coating interface. For UV applica-
tions, both water and hydrocarbon contamination need to be considered in their
effect on coating layers. Although condensed water is nearly transparent above
200 nm, it does absorb significantly in the VUV range. The absorption coefficient
of condensed water is 7/µm (base 10) at 157 nm. Hydrocarbons have strong
absorption throughout the UV range, depending on their molecular structure.
Even small amounts of water adsorption can have significant effects on the
performance of multilayer coatings in the VUV range. By way of example, we
have modeled the effect of 1% moisture incorporation into coating layers for a
LaF3 /MgF2 system designed for 157 nm [63]. The simulations (Fig. 10) show
that even such a modest amount of water can cause a drop of 2% in the transmit-
tance of an antireflective design and in the reflectance of a high-reflectance sys-
tem. By comparison, for a design centered at 193 nm, where the absorption coef-
ficient of water is ⬍1 ⫻ 10⫺5 /µm, a change of less than 0.2% is observed for
either an antireflective or a full reflector stack.

A. COATINGS DURABILITY STUDIES

Laser-induced coating damage has received a large amount of attention in recent

years because of high-power laser optics applications [42,72–75]. Most studies
involved the determination of laser-induced damage thresholds (LIDTs) using
so-called one-on-one or S-on-one measurement techniques. The former method
is a measurement of energy required to produce damage to a coating in a single
laser pulse. The latter involves a determination of a maximum laser pulse energy
that a coating can withstand for a fixed number of pulses, usually a few tens to
a few hundreds. Typical one-on-one LIDT thresholds for dielectric coatings in

Copyright © 2002 by Taylor & Francis

(A)

(B)

Figure 10 Effects of 1% moisture incorporation into thin film dielectric layers on (A)
transmission of a three-layer antireflectance coating and (B) reflection of a 20-layer full
reflector. (From Ref. 63.)

the UV are several J/cm2 and S-on-one thresholds are two to four times lower
[42].
A recent investigation of 248 nm laser damage of multilayer dielectric
stacks employed photothermal displacement and mirage techniques to assess 1-
on-1 LIDTs in both fluoride and oxide systems [72]. The study found that LIDT
values were dominated by defect absorption within the dielectric layers for a
LaF3 /MgF2 system, and by interface absorption for Al2O3 /SiO2 layers. In the case
of interface absorption, thermal transport of the energy away from the absorbing
region was found to be a critical factor in determining the damage threshold.

Copyright © 2002 by Taylor & Francis

Mann et al. [74] also found that the substrate’s thermal conductivity was a critical
parameter in determining LIDT values of dielectric 248 nm antireflective coat-
ings. The work was extended to 193 nm studies of dielectric-based reflectors. It
found significant effects due to hydrocarbon contamination and diffusion into the
dielectric layers. Laser conditioning was found to drive out the contaminants and
to increase the 193 nm LIDT values.
Few studies have addressed long-term, low-fluence degradation of optical
coatings. Two previous studies analyzed UV degradation of dielectric optics at
248 and 308 nm, for exposures up to ⬇108 pulses and moderately high laser
fluences (up to 0.5 J/cm2 /pulse) [74,75]. Both studies reported no visible damage
to the coatings, only a slight degradation in optical properties and a possible
increase in the absorption coefficient of the dielectric layers.
Within the lifetime of a lithographic tool an average antireflective coating
will be subjected to up to 1011 pulses at laser repetition rates of 1 kHz and fluences
ranging from 0.1 mJ/cm2 /pulse to a few mJ/cm2 /pulse. Since these fluence values
are three orders of magnitude below typical single-pulse damage thresholds, one
would not expect the coating performance to be determined by LIDT values.
Instead, long-term cumulative exposure effects may result either in subtle degra-
dation of optical properties or, perhaps, the eventual failure of the coating. Thus,
systematic experiments designed to identify the extent and mechanisms of laser-
induced degradation of optical coatings under realistic exposure conditions are
needed.
Long-term damage assessment of antireflective coatings for 193 nm lithog-
raphy was performed at MIT Lincoln Laboratory using incident fluences of ⬍15
mJ/cm2 /pulse and pulse counts of up to 10 9 pulses [50]. Antireflective coatings
from nine different commercial suppliers were evaluated under identical condi-
tions. Upon initial irradiation all the coatings underwent a certain amount of
surface cleaning, which resulted in higher transmission. The magnitude of this
recovery was up to 2%, and its magnitude was found to vary across coatings.
This so-called laser cleaning is observed also with bulk materials, but it is more
pronounced in coatings, presumably because it is related to surface roughness
and/or porosity.
A variety of catastrophic failures of coatings were also observed, although
these tended to occur early in the course of the irradiation studies, at pulse counts
under 100 ⫻ 10 6. The failure phenomena included high losses, either in the layers
or at the interfaces, peeling of the coating, or an apparent thinning of the layers.
It was also noted in these studies that antireflective coatings on fused silica
had ⬃1% higher 193 nm transmission than coatings deposited on calcium fluoride
substrates. In the course of 193 nm laser irradiation these losses decreased to the
level of the losses on the SiO2 substrates. This cleaning was very slow, and could
take place over 10 9 pulses. No correlation was found between surface roughness
and the amount of optical losses. The losses are believed to be an interface phe-

Copyright © 2002 by Taylor & Francis

nomenon, and the ‘‘laser cleaning’’ perhaps a slow, laser-induced out-diffusion
of contaminants through the coating layers.
Thin-film coatings at 157 nm are only now beginning to emerge with the
performance and durability comparable to that observed in 193 nm lithography.
The main difficulties are qualitatively the same as at longer wavelengths, but
quantitatively more pronounced: a limited choice of materials, larger impact of
contaminants, as well as the lack of a robust metrology infrastructure.

VII. ATMOSPHERIC ABSORPTION

AND SURFACE CONTAMINATION

A critical issue for the operation of optical components and for metrology at
short wavelengths is the high absorption coefficient of vapors and surface ad-
sorbates. As can be seen from Figure 11, propagation of UV radiation becomes
progressively difficult below 200 nm due to gas-phase absorption by O2 and H2O
molecules [76]. Even at 193 nm, purging the laser beamline with nitrogen is
recommended to suppress ozone formation. By comparison, 157 nm radiation
can propagate in an air atmosphere only over distances on the order of a millime-
ter before being completely absorbed.
Beam delivery systems in the VUV spectral region may be purged with an
inert gas or with nitrogen, which is transparent to wavelengths below 120 nm.
As an alternative, experiments may be performed under vacuum. However, great

Figure 11 Absorption coefficients of water vapor and oxygen throughout the ultraviolet
range. (From Ref. 76.)

Copyright © 2002 by Taylor & Francis

care must be taken that the vacuum is established by an oil-free pump and that
ultrahigh vacuum techniques are practiced to prevent surface contamination of
samples from gas-phase adsorbates. Figure 12 shows that 30 min exposure of a
precleaned CaF2 sample to a 5 ⫻ 10⫺5 torr of vacuum resulted in a transmission
drop of several percent throughout the VUV range [24]. In that example, the
vacuum was maintained by an oil-free turbomolecular pump backed by an oil-
based roughing pump. The contamination can be explained by backstreaming of
hydrocarbon-based oils into the vacuum chamber. Note that the original sample
transmission could be completely restored by in situ 157 nm laser cleaning.
Ambient contamination and sample packaging can drastically affect mea-
surements in the VUV region. To put the absorptivity of contaminants in perspec-
tive, Figure 13 shows the absorption of liquid water and solid polyethylene, a
representative material for modeling common organic contaminants such as al-
kanes [77]. Even a very thin film of 1 nm organic material can cause a 1% trans-
mission drop at 157 nm. For a 1 nm layer of water, an almost 2% drop in transmis-
sion is expected. In fact, in one set of experiments, in the course of sample
exchange with a supplier of optical coatings, several antireflectance coatings inad-
vertently came in contact with a clean polyethylene packaging membrane. All

Figure 12 Ambient effects on transmission of a 2 mm thick piece of CaF2 in the VUV

range. (From Ref. 24.)

Copyright © 2002 by Taylor & Francis

Figure 13 Absorbance per nanometer of a thin layer of polyethylene and condensed
water. (From Ref. 77.)

these samples consistently exhibited transmission around 70% when measured

in the VUV spectrometer prior to laser exposure. However, upon brief exposure
to 157 nm irradiation, dramatic transmission recovery was observed (⬎25% in-
crease in transmission). By comparison, coating samples shipped in such a way
that only an edge contact was made inside the shipping container exhibited only
a modest transmission increase (about 1–2%). The effect of surface contamina-
tion on samples may be expected to be more dramatic than on uncoated ones
because of the possibility of detuning of the dielectric stack and because of the
inherent porosity of some dielectric layers, as has been discussed in Sec. VI.

ACKNOWLEDGMENTS

The authors thank all the people and organizations that over the last few years
have provided us with samples to be tested and have shared with us their vast

Copyright © 2002 by Taylor & Francis

amounts of technical insight. Special thanks to our colleagues at MIT Lincoln
Laboratory, without whom much of the data reported here would have been im-
possible to obtain. This work was sponsored in part by the Advanced Lithography
Program of the Defense Advanced Research Projects Agency, under Air Force
contract F19628-95-C-0002 and in part under a Collaborative Research and De-
velopment Agreement between MIT Lincoln Laboratory and SEMATECH. Opin-
ions, interpretations, conclusions, and recommendations are those of the authors
and are not necessarily endorsed by the United States Government.

REFERENCES

1. AI Ershov, T Hofmann, WN Partlo, IV Fomenkov, G Everage, PP Das, D Myers.

Proc SPIE 3334: 1021–1030, 1998.
2. U Stamm, R Patzel, J Kleinschmidt, K Vogler, W Zschocke, I Bragin, and D Basting.
Proc SPIE 3334: 1010–1013, 1998.
3. TP Duffey, T Embree, T Ishihara, R Morton, WN Partlo, T Watson, R Sandstrom.
Proc SPIE 3334: 1014–1020, 1998.
4. R Sandstrom. Proc SPIE 1264: 505–519, 1990.
5. K Vogler, U Stamm, I Bragin, F Voss, SV Govorkov, G Hua, J Kleinschmidt, R
Paetzel. Proc SPIE 4000: 1515–1528, 2000.
6. CB Collins, FW Lee, JM Carrol. Appl Phys Lett 37: 857–859, 1980.
7. U Stamm, W Zschocke, B Nikolaus, P Genter, D Basting. Proc SPIE 2991: 194–
200, 1997.
8. T Uchimura, T Onoda, C-H Lin, T Imasaka. Rev. Sci. Instrum. 70: 3254–3258,
1999.
9. Oriel Company. http://www.oriel.com/homepage/down/pdfla.htm
10. Ushio Semiconductor Company http://www.ushiosemi.com/excimer.htm
11. W Tiffany. http://www.molectron.com/AppNote005a.html
12. V Liberman, M Rothschild, JHC Sedlacek, RS Uttaro, AK Bates, K Orvek. Proc
SPIE 4000: 488–495, 2000.
13. ML Dowell, CL Cromer, RW Leonhardt, TR Scott. AIP Proc 449, 539–541, 1998.
14. JD Buck. http://www.molectron.com/AppNote002a.html
15. Hamamatsu Company. http://usa.hamamatsu.com/opto-semi/photodiodes/default.
htm
16. R Gupta, KR Lykke, P-S Shaw, JL Dehmer. Proc SPIE 3818: 27–33, 1999.
17. Mc Pherson Inc. http://www.mcphersoninc.com/detectors/model650detectorassem-
bly.htm
18. V Liberman, M Rothschild, JHC Sedlacek, RS Uttaro, A Grenville, AK Bates, C
Van Peski Opt Lett 24: 58–60, 1999.
19. NV Morozov. Proc SPIE 2428: 153–169, 1995.
20. KR Mann, E Eva. Proc SPIE 3334: 1055–1061, 1998.
21. R DeSalvo, AA Said, DJ Hagan, EW Van Stryland, M Sheik-Bahae. IEEE J Quan-
tum Electron. 32: 1324–33, 1996.
22. E Eva, K Mann. Appl Surf Sci 109–110, 52–57, 1997.
23. R Gupta, JH Burnett, U Griesmann, M Walhout. Appl Opt 37: 5964–5968, 1998.

Copyright © 2002 by Taylor & Francis

24. TM Bloomstein, V Liberman, M Rothschild, DE Hardy, RB Goodman. Proc SPIE
3676: 342–349, 1999.
25. V Liberman, TM Bloomstein, M Rothschild. Proc SPIE 3998: 480–491, 2000.
26. JA Woollam, B Johs, CM Herzinger, J Hilfiker, R Synowicki, CL Bungay. Proc
SPIE CR72, 3–28, 1999.
27. B Johs, JA Woollam, CM Herzinger, J Hilfiker, R Synowicki, CL Bungay. Proc
SPIE CR72, 29–57, 1999.
28. JN Hilfiker, B Sing, RA Synowicki, CL Bungay. Proc SPIE 3998: 390–398, 2000.
29. P Boher, JP Piel, P Evrard, C Defranoux, M Espinosa, J Stehle. Proc SPIE 3998:
379–389, 2000.
30. DL Griscom. J Ceram Soc Jpn 99: 923–942, 1991.
31. DL Griscom. Proc SPIE 541: 38–59, 1985.
32. IT Godmanis, AN Trukhin, K Hubner. Physica Status Solidi B 116: 279–287, 1983.
33. S Sakaguchi, S Todoroki, S Shibata. J Am Ceram Soc 79: 2821–2824, 1996.
34. DJ Krajnovich, IK Pour, AC Tam, WP Leung, MV Kulkarni. Opt Lett 18: 453–
455, 1993.
35. DR Sempolinski, TP Seward, C Smith, N Borrelli, C Rosplock. J Non Crystal Solids
203: 69–77, 1995.
36. P Schermerhorn. Proc SPIE 1835: 70–79, 1992.
37. R Schenker, L Eichner, H Vaidya, S Vaidya, WG Oldham. Proc SPIE 2440: 118–
25, 1995.
38. S Thomas, B Kuhn. Proc SPIE 2966: 56–64, 1996.
39. RE Schenker, WG Oldham. J Appl Phys 82: 1065–1071, 1997.
40. C Pfleiderer, N Leclerc, KO Greulich. J Non Crystal Solids 159: 145–153, 1993.
41. E Eva, K Mann, S Thomas. Proc SPIE 2966: 72–99, 1997.
42. SS Wiseall, DC Emmony. Proc SPIE 369: 521–526, 1983.
43. E Eva, K Mann. Appl Phys A 62: 143–149, 1996.
44. JH Stahis, MA Kastner. Philos Mag 49: 357–362, 1984.
45. V Uhl, KO Greulich, S Thomas. Appl Phys A 65: 457–462, 1997.
46. TE Tsai, DL Griscom. Phys Rev Lett 67: 2517–20, 1991.
47. C Smith, NF Borrelli, DC Allan, TP Seward. Proc SPIE 3051: 116–21, 1997.
48. RJ Araujo, NF Borrelli, C Smith. Proc SPIE 3424: 2–9, 1998.
49. V Liberman, M Rothschild, JHC Sedlacek, RS Uttaro, A Grenville, AK Bates, CK
Van Peski. Proc SPIE 3427: 411–419, 1998.
50. V Liberman, M Rothschild, JH Sedlacek, RS Uttaro, AK Bates, CK Van Peski. Proc
SPIE 3578: 2–15, 1999.
51. V Liberman, M Rothschild, JHC Sedlacek, RS Uttaro, A Grenville. J Non Cryst
Solids 244: 159–171, 1999.
52. N Leclerc, C Pfleiderer, H Hitzler, J Wolfrum, KO Greulich, S Thomas, W English.
J Non Crystal Solids 149: 115–121, 1992.
53. V Liberman, M Rothschild, JHC Sedlacek, RS Uttaro, AK Bates, C Van Peski. Proc
SPIE 3679: 1137–1142, 1999.
54. TP Duffey, T Embree, T Ishihara, R Morton, WN Partlo, T Watson, R Sandstrom.
Proc SPIE 3334: 1014–1020, 1998.
55. DC Allan, C Smith, NF Borrelli, TP Seward III. Opt Lett 21: 1960–1962, 1996.
56. NF Borrelli, C Smith, DC Allan, TP Seward, III. J Opt Soc Am B 14: 1606–15,
1996.

Copyright © 2002 by Taylor & Francis

57. RG Morton, RL Sandstrom, GM Blumenstock, Z Bor, CK Van Peski. Proc SPIE
4000: 496–510, 2000.
58. CB Norris, EP EerNisse. J Appl Phys 45: 3876–3882, 1974.
59. JE Shelby. J Appl Phys 50: 3702–3706, 1979.
60. JA Ruller, EJ Friebele. J Non Crystal Solids 136: 163–72, 1991.
61. CM Smith, LA Moore. Proc SPIE 3676: 834–841, 1999.
62. Y Ikuta, S Kikugawa, T Kawahara, H Mishiro, N Shimodaira, A Masui, S Yoshi-
zawa. Proc SPIE 4000: 1510–1514, 2000.
63. V Liberman, TM Bloomstein, M Rothschild, JHC Sedlacek, RS Uttaro, AK Bates,
CV Peski, K Orvek. J Vac Sci Technol B 17: 3273–3279, 1999.
64. BG Ravi, S Ramasamy. Int J Mod Phys B 6: 2809–2836, 1992.
65. DR Rao, HN Bose. J Phys Soc Jpn 28: 152–7, 1970.
66. F Abba, Y Marqueton, EA Decamps. Physica Status Solidi A 35: 129–132, 1969.
67. JH Burnett, R Gupta, U Griesmann. Proc SPIE 4000: 1503–1509, 2000.
68. S Majewski, MK Bentley. Nucl Instrum Methods Phys Res A, Accel 260: 373–376,
1987.
69. P Laporte, JL Subtil, M Courbon, M Bon, L Vincent. J Opt Soc Am 73: 1062–
1069, 1983.
70. S Ogura, N Sugawara, R Hiraga. Thin Solid Films 30: 3–10, 1975.
71. U Kaiser, N Kaiser, P Weissbrodt, U Mademann, E Hacker, H Muller. Thin Solid
Films 217: 7–16, 1992.
72. E Welsch, K Ettrich, H Blaschke, P Thomsen-Schmidt, D Schafer, N Kaiser. Opt
Eng 36: 504–14, 1997.
73. F Rainer, WH Lowdermilk, D Milam, CK Carniglia, TT Hart, TL Lichtenstein. Appl
Opt 24: 496–500, 1985.
74. K Mann, B Granitza, E Eva. Proc SPIE 2966: 496–504, 1997.
75. DJ Krajnovich, M Kulkarni, W Leung, AC Tam, A Spool, B York. Appl Opt 31:
6062–75, 1992.
76. H Okabe, Photochemistry of Small Molecules. New York: Wiley Interscience, 1978.
77. ED Palik, ed. Handbook of Optical Constants of Solids. vol. 2. San Diego, CA:
Academic Press, 1991.

Copyright © 2002 by Taylor & Francis

2
Laser Optogalvanic Spectroscopy
of Discharge Plasmas
in the Ultraviolet Region

C. Haridass,* H. Major, and Prabhakar Misra

Howard University, Washington, D.C.

Xianming L. Han
Butler University, Indianapolis, Indiana

I. INTRODUCTION

The optogalvanic effect (OGE) is due to a change (increase or decrease) in the

electrical properties (conductivity) within a self-sustained gaseous discharge
when illuminated by radiation that is resonant with an atomic or molecular transi-
tion of the element within the discharge. The main principle of this effect is that
the absorption of radiation causes a redistribution of populations in the atomic
or molecular energy levels. Under steady-state discharge conditions, there exists
a dynamic equilibrium (as a result of various radiative and/or collisional and/or
collective processes) between the various plasma species causing a well-defined
impedance to the flow of current. At dynamic equilibrium, changes in the
electron/ion densities and/or mobilities, with a concomitant change in the electri-
cal impedance of the plasma, are caused by optical perturbations. Such a change
in impedance alters the current in the plasma and can be either real for direct
current (DC) discharges or complex for alternating current (AC) discharges.
The optogalvanic effect has been used for many years as a sensitive and
reliable method for recording calibration spectra when tunable lasers are em-
ployed for monitoring species (e.g., the hydroxyl radical, OH) that are of primary

* Current affiliation: Belfry School, Belfry, Kentucky.

Copyright © 2002 by Taylor & Francis

importance to stratospheric photochemistry or tropospheric air quality. Optogal-
vanic spectroscopic/detection methods are well-established techniques, although
the optogalvanic effect in discharges is not completely well understood. Optogal-
vanic spectroscopy in a gas discharge plasma is differentiated from other charged
particle detection techniques, such as space–charge-limited diodes [1] and gas-
filled proportional counters [2], by the presence of a sustained discharge involving
an electron gas at a relatively high temperature (0.5–10 eV) [3]. Laser optogal-
vanic (LOG) spectroscopy is performed by directing a tunable dye laser beam
into a plasma (such as that generated in a hollow cathode discharge lamp). When
the wavelength of the laser beam coincides with the absorption of a species in
the plasma, the rate of ionization of that species changes momentarily because
of laser-perturbed collisional ionization. The associated impedance change can
be detected as a voltage drop across a ballast resistor in the lamp feeding circuit.
If the laser photon flux is high enough, ionization can occur due to the
laser photoionization of excited-state species produced by stepwise resonant or
multiphoton laser excitation (whereby the collisional step is replaced by
photoionization). The term resonance ionization spectroscopy (RIS) or multipho-
ton ionization (MPI) is then used. MPI may also be referred to as resonance
multiphoton ionization or (REMPI). RIS and MPI differ from the optogalvanic
effect in that while they both occur in the plasma/flame environment, they do
not necessarily require such a medium [4].

II. OVERVIEW OF PAST AND PRESENT WORK

The earliest observation related to the OGE was made by Foote and Mohler [5].
They used a tungsten lamp dispersed by a monochromator while studying the
photoionization of Cs vapor in a thermionic diode tube, in an effort to detect
small levels of ionization. These authors also observed ionization of the Cs oc-
curring when the sample was irradiated by wavelengths longer than 318 nm,
which corresponded to wavelengths of the atomic Cs spectrum. They postulated
that radiation at these wavelengths produced excited-state atoms, which upon
collision with other atoms acquired sufficient additional energy to become ion-
ized. Penning [6] observed the OGE for atoms when he detected a variation in
the impedance of a neon discharge irradiated by emission from an adjacent neon
discharge. Terenin [7] observed the phenomena in molecules when he studied
photoionization of salt vapors. Garscadden et al. [8] observed the OGE brought
about by a change in the discharge current of a gas-discharge laser when the
laser came over threshold.
It was not until the development of tunable dye lasers, however, that the
OGE became widely used as a spectroscopic tool. Green et al. [9] demonstrated
that high-sensitivity spectra of the species in a discharge could be obtained with

Copyright © 2002 by Taylor & Francis

a tunable dye laser. The electronic excitation of the atoms in the discharge al-
lowed for the observation of transitions starting from the metastable or excited
states, while the use of a hollow-cathode discharge made it possible to perform
spectroscopy on a gas-phase sample of refractory elements produced by cathodic
sputtering. Optogalvanic transitions were observed for various species sputtered
from hollow cathodes [10–13] and from gas-discharge flashlamps filled with
neon and argon [9,14–20]. Optogalvanic signals from commercial neon lamps,
extensively used as spectral light sources, have been reported in the 540–750
nm [9,14,16,21] and 2440–2780 nm [15] regions. Gusev and Kompanets [22]
pointed out the difficulty of obtaining OG resonances with neon at wavelengths
shorter than 580 nm, owing to low oscillator strengths of the resonances. Zhu et
al., [23] however, observed laser-assisted OG signals with neon in the 337–598
nm region using a commercial iron-neon hollow cathode lamp. These authors
identified 223 OG transitions associated with neon energy levels and found that
there were more neon OG lines in the near UV region than in the yellow and
red region and that these UV transitions possessed fairly strong intensities. The
identified lines in the study by Zhu et al. [23] were calibrated with interference
fringes from the two surfaces of an etalon recorded simultaneously with the OG
signal. OG transitions of argon have also been observed in the following wave-
length regions: 367–422 nm [20], 415–670 nm [17], 425–700 nm [22], 420–
740 nm [18], 555–575 nm [24], 727–772 nm [16], 360–740 nm [19], and 2440–
2780 nm [15].
Most of the work cited above, for both neon and argon, describes the gen-
eral nature of the optogalvanic signals in various wavelength regimens. However,
very few studies [25,26,27] reported in the literature identify and quantitatively
characterize the dominant physical processes contributing to the production of
the optogalvanic signals in a discharge plasma. Han et al., [25] using a simple
mathematical rate equation model, with only a few parameters characterizing
the atomic state population, produced a good simulation for the observed time-
dependent OG waveform for a specific neon transition. In contrast, a model given
by Stewart et al. [27] used nearly 150 parameters for the same transition of the
neon atom and still failed to describe the observed waveform satisfactorily.
Owing to its sensitivity and selectivity, the optogalvanic technique has been
used successfully for a wide variety of laser spectroscopic experiments, including
analytical flame spectroscopy/spectrometry [28–30], laser calibration [11,14,19],
laser stabilization [31,32], plasma diagnostics [33–34], atomic spectroscopy
[4,21], molecular spectroscopy [35–38], and dynamics [39–41].
Turk et al. [28] introduced analytical flame spectrometry using laser-
enhanced ionization (the optogalvanic effect in flames). They compared their
detection limits to other methods of the time and found competitive sensitivity
for a number of elements with a much simpler system. While aspirating a dilute
sodium solution into the flame, a measurable increase in the current through the

Copyright © 2002 by Taylor & Francis

flame was observed when the laser wavelength was tuned to the sodium resonance
lines at 589.0 nm and 589.6 nm. The signal responded linearly to Na concentra-
tions in the range 2–1000 ng/ml [29].
Skolnick [31] used the OGE as a method to stabilize the frequency of a
single J value CO2 laser to the center of its output power. Skolnick’s method
of frequency stabilization was based on modulation of the laser frequency and
simultaneous sensing of the resulting perturbation of the laser plasma tube’s im-
pedance. Such stabilization resulted in a simple frequency-control system without
using a 10 µm light detector or any ancillary optical component [31].
Ausschnitt et al. [33] demonstrated the application of multiphoton optogal-
vanic spectroscopy to hydrogen discharge plasma diagnostics. The ground and
excited-state densities and the translational temperatures of hydrogen and deute-
rium atoms within the plasmas were probed [33]. The use of two unequal energy
photons resonant with an intermediate state enabled three-dimensional probing
of the ground state and discrimination between hydrogen and deuterium.
The first LOG detection of molecular ions (N2⫹ and CO⫹) was reported by
Walkup et al. [38] Their results indicated an optogalvanic mechanism that in-
volves direct alteration of ion mobility by laser excitation due to a difference in
charge-exchange collision rates for excited versus ground-state ions. This tech-
nique proved to be a sensitive probe of ions in the cathode dark space [38]. The
LOG spectroscopic technique may be regarded as a complement to the traditional
absorption or fluorescence methods rather than a replacement.
Intermodulated optogalvanic spectroscopy (IMOGS), a newer method of
Doppler-free saturation spectroscopy, was described by Lawler et al. [42] Closely
related to intermodulated fluorescence spectroscopy, IMOGS, uses the detection
scheme of optogalvanic spectroscopy. The authors compared the sensitivity of
IMOGS to that of Doppler-free saturated absorption spectroscopy. Using appro-
priate parameters in their experiment, they estimated that (S/N)IMOGS /(S/N)SAS ⫽
100. Since other spectroscopic techniques have been devised that can improve
the sensitivity of saturated interference spectroscopy, they concluded that IMOGS
would be most useful in the region of the spectrum where low-noise detectors,
interferometric-quality optics, or high-quality polarizers were unavailable.
In order to comply with environmentally mandated regulations, the US
Department of Energy is attempting to implement online, real-time monitors
(with limits of detection of 1 pCi l⫺1) to measure the concentration of radionu-
clides in the off-gases of mixed hazardous waste and radioactive waste thermal
treatment systems [43–44]. A detection limit of 1 pCi l⫺1 corresponds to 465 µg
m⫺3 for U-235 and 2980 µg m⫺3 for U-238. Abhilasha et al. [43] developed a
LOGS system to measure the concentration of uranium present in the off-gases
of mixed hazardous waste thermal treatment systems. Their technique yielded
limits several orders of magnitude below that attainable by current real-time on-
line radioactive disintegration counting methods and that sought by US Depart-

Copyright © 2002 by Taylor & Francis

ment of Energy (4.6 µg m⫺3 corresponding to a detection limit of 0.0099 pCi l⫺1
for U-235 or 0.0016 pCi l⫺1 for U-238). It was noted, however, that the bandwidth
of their current laser system (nominally 0.4 cm⫺1 in the UV) was insufficient to
resolve the uranium isotopically.
A field-deployable LOGS system was developed by Monts et al. [44] more
recently to measure the concentration of metal species present in thermal treat-
ment systems. LOGS is an ultrasensitive diagnostic tool for online real-time mea-
surements of the concentration of volatile toxic trace metals, radionuclides, and
other gas-phase pollutants present in off-gases of mixed-waste thermal treatment
systems, where metal species exist primarily as airborne metal oxides and/or
metallic particles rather than as free single atoms. Real-time monitoring of air-
borne contaminants plays a vital role in the protection of personnel and in safe-
guarding the environment [44]. Since the amplitude of the LOGS signal can be
related to species concentration, LOGS has the capability of monitoring, in real
time, the concentration of a pollutant of interest.
Miron et al. [39] studied the dynamics of the OGE in hollow cathode lamps
by recording time-dependent OG signals and treating the hollow cathode lamp
(HCL) as a circuit element. They found that a stable operation of the HCL does
not require a high-impedance source, like that needed for a longitudinal discharge
tube (for pressure ranges of 1–10 torr). The investigators reported that the dura-
tion of the OG signal (a few microseconds) is independent of the lifetime (10–
500 ns) of excited levels populated by the laser. They concluded that it was in
fact the impulse response of the electronic circuit (which includes the HCL). The
variation of the response time with current implied that the discharge recovered
more quickly with increased electron density.
This chapter will fill a gap that exists in the optogalvanic spectral map in
the near ultraviolet (UV) region, and study the time-dependent OG waveforms
of neon and argon, in order to determine the dominant process or processes affect-
ing the OGE and thereby quantify important associated rate parameters. Argon
and neon are excellent choices for especially accomplishing the first objective,
because they are commonly used buffer gases within hollow cathode lamps and
provide an acceptable density of lines in the visible and near (UV) regions. The
near UV region is of importance because in this wavelength domain fall the
electronic spectra of important atmospheric and combustion species, which in-
clude chemical intermediates such as the hydroxyl and alkoxy radicals. The sec-
ond and more ambitious objective was complicated because the environment of
a discharge is too complex to be described by a single general theoretical model;
and a quantitative understanding of the OGE is possible only if a reasonable
model of the discharge can be made and the cross sections of the involved pro-
cesses are well known. An effective use of the rate equation formalism to describe
the different discharge processes can lead to a very large number of equations
for which cross-section values are unavailable. Thus, any knowledge of the prop-

Copyright © 2002 by Taylor & Francis

erties of a particular discharge is usually largely empirical. An extended and
refined version of a mathematical rate equation model for the population distribu-
tion of the atomic species in a commercial hollow cathode discharge lamp, set
forth by Han et al. [25] in a relatively recent paper on collisional ionization in
a discharge plasma, was used in this investigation. The time-resolved profiles of
LOG waveforms of argon and neon were analyzed using commercial hollow
cathode discharge lamps, with an emphasis on the low-discharge current regimen
(0.2–3.0 mA), since it had not been covered in previous studies and promised
to provide new insight into multilevel atomic systems in a gas discharge plasma.
The physics of time-resolved OG waveforms provided quantitative information
on the rates of excited-state collisional processes in a gas discharge plasma.
Collisional ionization is a primary chemical process occurring in the upper
atmospheres and in ionospheres of planets. The flux of energetic photons in the
upper atmosphere can induce chemical reactions that are extremely difficult to
reproduce under similar conditions in the laboratory. An understanding of the
ionization processes and the subsequent chemistry is essential for answering
questions such as the phenomenon responsible for the existence of sharp ioniza-
tion layers (so-called ledges), or to the ionic contributions of various physical
phenomena, such as thermospheric heating and auroral excitations. The study of
the optogalvanic effect may provide some critical insights into more effective
parameterization in models of ionospheric excitation. Neon and argon are good
candidates for prototype studies because they do not posses rotational or vibra-
tional degrees of freedom. However, neon is isoelectronic with HF, CH4, and
other trace gases, while argon is isoelectronic with HCl, F2, and other species.
By studying the collisional excitation of these noble gases under controlled condi-
tions, suitable approximations can be made to explain the behavior of important
trace gases in the upper atmosphere. By fitting the OG waveforms, parameters
associated with rates of decay and those proportional to the cross sections of
electron collisional ionization are obtained. Allowances have to be made for rota-
tional and vibrational degrees of freedom, and decay rates obtained would proba-
bly not be very accurate, but there should be many similarities between the neon/
argon prototype and a so-called unified atom framework of the trace gases.

III. EXPERIMENTAL WORK

The experimental arrangement used for laser optogalvanic spectroscopy is shown

in Figure 1. In this figure, a dye laser (DL) is pumped by the second harmonic
of a Nd :YAG-laser (YL) running at 10 Hz. The output beam had a pulse duration
of about 20 ns and a nominal line width of 0.07 cm⫺1 (without any intracavity
etalon). The tuning range in the near UV (and visible) was covered by several
laser dyes. One of the beams passed through a 1 mm aperture and entered either

Copyright © 2002 by Taylor & Francis

Figure 1 Schematic experimental arrangement for laser optogalvanic spectroscopy YL,
Nd :YAG laser; DL, dye laser; QW, quartz wedge; PD, photodiode; AP, aperture; NL,
negative lens; ET, etalon; HCL, hollow cathode discharge lamp; PS, power supply; R,
resistor; C, capacitance; BC, boxcar; OSC, oscilloscope; PC, personal computer.

a commercial (Perkin-Elmer) iron–neon HCL or a Laser Galvatron (L2783-

26ANE-FE, Hamamastu Co.), which is also an HCL containing argon (2 Torr),
neon (3 Torr), and trace amounts of iron vapor. The second split beam can be
directed to record simultaneously the laser-induced fluorescence (LIF) spectrum
of a free radical (e.g., OH) in a propane–air flame. An interference fringe pattern
was also simultaneously recorded using the etalon for calibration purposes. A
high-voltage power supply (PS) and a ballast resistor (R) of 20 KΩ for the iron–
neon lamp (and 30KΩ for the Ar-containing commercial Laser Galvatron) were
used to record the OG spectra. When the laser pulse was resonantly absorbed by
the discharge medium, the voltage across the lamp varied, and these variations
were suitably coupled via a capacitor (C) to a boxcar (BC) integrator. Temporal
evolution of the signal was recorded using a digital oscilloscope (OSC). Outputs
of the boxcar and the photodiode were recorded with a computer-aided (PC) data
acquisition system.

IV. THEORY
A. Mechanism
The mechanism of the optogalvanic effect involves laser enhancement or suppres-
sion of the ionization rates of a particular species present in the plasma. Several

Copyright © 2002 by Taylor & Francis

models have been developed, each appropriate for a given type of plasma, but
they are essentially phenomenological and do not account for many of the com-
plex processes occurring in ionized gases that could, in principle, be involved in
a laser-induced impedance change. Yet the fact that a number of different plasma
sources exhibit the OGE (flames, positive column DC discharges, hollow cathode
DC discharges, radiofrequency (RF) discharges) attests to its generality [37].
Illumination of a gas discharge with radiation at a wavelength correspond-
ing to an atomic transition of a species in the discharge causes perturbations to
the steady-state population of two or more levels. This, in turn, causes a change
in the electrical properties of the discharge. The optogalvanic effect can corre-
spond to an increase or decrease in discharge current depending on the kinetics
of the levels whose populations are perturbed by the laser. When the laser fre-
quency is tuned to the transitions EL1 → EL2 between two levels of atoms or ions
in the discharge, the laser initially transfers some population in EL1 to EL2, which
subsequently relaxes to EL3 by radiative and/or collision processes in a time frame
of tens of nanoseconds. Owing to the different ionization probabilities from EL1
and EL3, such a population change will result in a change in the discharge current,
∆I, which is then detected as a voltage change ∆V ⫽ R ∆I across the ballast
resistor R. In general, positive and negative signals are observed, depending on
the characteristics of levels EL1 and EL3 involved in the transition EL1 → EL3. If
IPL1 is the total ionization probability of an atom in level EL1, the voltage change
∆V produced by the laser-induced change ∆nL1 ⫽ nfL1 ⫺ niL1 is given by: [45]
∆V ⫽ R ∆I ⫽ a[∆nL1IPL1 ⫹ ∆nL3 IPL3] (1)
where ∆nL1 is the laser-induced change in the population of level 1 (final ⫺
initial) and ∆ nL3 is the change in the population of level 3. If the laser excites
atoms from a level EL1, with a small probability of ionization, to a level EL3,
which has a larger probability of ionization, the discharge current will increase.
If on the other hand, the net effect of the laser takes atoms from a level with a
large probability of ionization to a level with a small probability of ionization,
the discharge current will decrease. The latter often arises when the atom is ex-
cited out of a long-lived metastable state into a higher-lying state that is not
metastable, which results in rapid decay to a state lower than the original metasta-
ble state. Some of the lowest excited levels in noble gases are metastable, having
values of J ⫽ 0 or 2. Metastable atoms exist in high concentrations in glow
discharges and thus can have a great impact on the optogalvanic effect.
The convention used throughout the present study is based on an external
measuring circuit that takes the OG signal off the cathode end of the discharge
lamp. The traditional positive OG signal is one in which there is a net increase
in ionization or increase in current (∆I ⬎ 0). It is the result of a decrease in
impedance, which appears as a decrease in voltage ∆V across the HCL (or an
increase in voltage across the ballast resistor) that causes the waveform to appear

Copyright © 2002 by Taylor & Francis

with an initial negative peak. A negative OG signal results from the depletion
of the metastable population. The net effect is an increase in impedance and
therefore an increase in ∆V, whereby ∆I ⬍ 0. (Note: When signals are measured
from the anode, all signs are reversed.)
Several different ionization reactions can contribute to the discharge cur-
rent. Electron impact ionization, collisions among the discharge species, and Pen-
ning ionization are perhaps the most important and likely candidates. We use the
notation: B ⫽ buffer gas atom, S ⫽ sputtered atom and B* ⫽ an excited state
of B, so that various ionization processes may be represented as follows.
Electron impact ionization occurs from excited levels as well as from the
ground state, and although it is not always the dominant ionizing reaction, it
usually contributes to the effect to some degree. It may result from a one-step
ionization defined as
B ⫹ e⫺ → B⫹ ⫹ 2 e⫺ (2)
Such direct ionization dominates at low pressures, since the electrons can acquire
enough energy between collisions to ionize an atom. A two-step, or multistep
ionization, is also possible and defined by
B ⫹ e⫺ → B* ⫹ e⫺ (3)
B* ⫹ e⫺ → B⫹ ⫹ 2 e⫺
This kind of process is important in noble gas discharges and has a significant
impact on the optogalvanic effect.
Collisions among the excited atoms of the discharge species can be repre-
sented by
B* ⫹ B* → B⫹ ⫹ B ⫹ e⫺ (4)
These types of collisions are mostly seen in the case of noble gas atoms and are
very effective because of the high energies (on the order of 16.6 eV for neon
and 11.6 eV for argon) of the metastable levels involved. The buffer gas ions
liberate atoms from the surface of the cathode by colliding with the cathode
material. Thus, the liberated atoms form a minor constituent of the discharge
species.
Penning ionization can be represented by
B* ⫹ S → S⫹ ⫹ B ⫹ e⫺ (5)
In Penning ionization, the excited buffer gas atoms may ionize the sputtered
atoms. This kind of process is important for discharges in noble gases and influ-
ences the characteristics of the observed discharge current. In general, at higher
discharge currents, the mechanism of electron impact ionization dominates over
Penning ionization, leaving the latter hardly noticeable [46,47].

Copyright © 2002 by Taylor & Francis

 Broglia et al. [48] demonstrated that photoionization can influence the dis-
charge current in hollow cathode DC lamps. The researchers showed that the
dark space of hollow-cathode discharges exhibits signals much more rapidly than
the conventional OG signals and related them to direct laser photoionization.
Since our work was based on results in which the laser beam is collimated to
fill the region between the electrodes, it was first assumed that we might be able
to see this feature; however, the energy flux was so weak in the low-current
region used in this study that we could not do so. Broglia et al. [49] also admitted
that experimental evidence showed that in a low-power regimen only the OG
signal characterized by the µs time scale arises.
It was still evident, however, that the induced current resulting from the
OGE could, in principle, be due to many factors. One of the main objectives of
the present work was to study extensive sets of OG transitions of neon and argon.
As expected, the largest identifiable contribution of the OGE waveform turned
out to be the electron collisional ionization of the buffer gas, and contributions
such as Penning ionization and others only appeared as either slight additions or
perturbations.
A nonlinear least-squares fitting program written by Han [50] of Butler
University in Fortran code was used and was capable of fitting multiple exponen-
tial functions to experimental data of the form:

W ⫽ ae⫺bt ⫹ ce⫺d t ⫹ ee⫺f t ⫹ ⋅ ⋅ ⋅

The fit could run up to 200 iterations or until a precision of 10⫺6 was reached.
W is the waveform intensity; a, c, e, . . . , etc., are coefficients related to effective
collisional cross sections; and b, d, f, . . . , etc., are exponential parameters that
allow quantification of the decay rates and thus the effective lifetime of the states
involved in the OG transition.

B. Theoretical Model
A model involving only buffer gas (neon or argon) transitions, developed by Han
et al., [25] was modified and used in this study to account for the results associated
with time-resolved waveforms of LOG transitions. Neon or argon atoms are de-
tected by the optogalvanic effect (via increased or decreased discharge current)
only as they are ionized. When the laser is tuned to a transition of the buffer gas,
the number of atoms in each state is determined mainly by three processes elec-
tron collisional excitation, radiative depopulation, and electron collisional ioniza-
tion. Certain other nonnegligible processes can contribute to the OG effect: in
particular, the redistribution of excited buffer gas atoms due to atomic and elec-
tronic collisions. Subsequent relaxation of these atoms to the corresponding lower
states may be subject to further ionization. In terms of the Paschen notation, for

Copyright © 2002 by Taylor & Francis

Figure 2 A simplified two-level energy diagram illustrates the dynamics involved in
the generation of the optogalvanic signal (* indicates laser excitation and radiative decays).
In the case of neon 1s5 –2p9 transition, |L1〉 represents |1s5〉, |L2〉 represents |2p9〉, and |L′2〉
represents other |2p〉 energy states. For other optogalvanic transitions, |L2〉 and |L′2〉 could
represent the same energy states.

the neon 1s–2p transition, there are four 1s states and ten 2p states, and therefore
a total of 30 allowed radiative transitions [51,52].
Prior to laser excitation, there exists a state of dynamic equilibrium in the
atomic distribution in each energy level. Upon laser excitation, a fraction of the
excited buffer gas atoms is collisionally transferred from |L2〉 to other |L′2〉 states
and then radiatively decays back to the |L3〉 state, which should happen within
the radiative lifetime of the |L2〉 state. To a first-degree approximation, let us treat
the other three 1s states as a collective |L3〉 state for the sake of simplicity of
mathematical derivation. We also make the approximation that, after one colli-
sion, the atoms transferred out of states |L1〉 and |L3〉 reach the steady-state distri-
bution, and therefore we neglect the possibility of collisional transfer between
|L1〉 and |L3〉. The preceding above approximations were validated by the observed
experimental data. If these approximations were invalid, all four 1s states would
have to be treated separately. The solution for the time-dependent OG signal
would consist of four exponential functions instead of the two exponential func-
tions considered by Han et al. [25] in treating the visible neon transition 1s5 –
2p9.
The population change over time (as illustrated in Fig. 2) can be expressed
as
dNL1
⫽ RL1 ⫺ NL1(ΓL1 ⫹ IPL1 ⫹ IPL1s) (6)
dt

Copyright © 2002 by Taylor & Francis

 dNL3
⫽ RL3 ⫺ NL3(ΓL3 ⫹ IPL3 ⫹ IPL3s) (7)
dt

where RL1 and RL3 are the rate constants of collisional excitation to the states |L1〉
and |L3〉, respectively, from all lower states. NLi and ΓLi are, respectively, the neon
population and effective decay rates of the state |Li〉. IPLi is the total probability
of ionization of the state |Li〉, I is the current, and PLi is related to the collisional
ionization cross section of the state |Li〉. PL1s and PL3s are the electron collisional
transfer rate parameters to 1s steady-state distribution from |L1〉 and |L3〉 states,
respectively. At steady state, dNL1 /dt ⫽ dNL3 /dt ⫽ 0. Therefore, the steady-state
population of each state can be expressed as

RL1
nL1 ⫽ (8)
ΓL1 ⫹ IPL1 ⫹ IPL1s

and

RL3
nL3 ⫽ (9)
ΓL3 ⫹ IPL3 ⫹ IPL3s

Solving Eqs. (6) and (7) gives the change in population of both |L1〉 and |L3〉
states as a function of time:

NL1(t) ⫽ nL1 ⫹ cL1 exp[⫺(ΓL1 ⫹ IPL1 ⫹ IPL1s)t] (10)

NL3(t) ⫽ nL3 ⫹ cL3 exp[⫺(ΓL3 ⫹ IPL3 ⫹ IPL3s)t] (11)

where cL1 and cL3 coefficients that will be determined by the initial conditions.
Thus, the population changes caused by the laser excitation are

∆NL1(t) ⫽ NL1(t) ⫺ nL1 ⫽ cL1 exp[⫺(ΓL1 ⫹ IPL1 ⫹ IPL1s)t] (12)

∆NL3(t) ⫽ NL3(t) ⫺ nL3 ⫽ cL3 exp[⫺(ΓL3 ⫹ IPL3 ⫹ IPL3s)t] (13)

for |L1〉 and |L3〉 states, respectively. After laser excitation, the neon population
redistribution in the 1s states gives the initial conditions as

∆NL1(0) ⫽ ⫺∆N0 and ∆NL3(0) ⫽ ⫹∆ N0 (14)

Using the initial conditions given in Eq. (14), the change in populations given
in Eq. (12) and (13) can be shown to be

∆NL1(t) ⫽ ⫺∆N0 exp[⫺(ΓL1 ⫹ IPL1 ⫹ IPL1s)t] (15)

∆NL3(t) ⫽ ∆N0 exp[⫺(ΓL3 ⫹ IPL3 ⫹ IPL3s)t] (16)

Copyright © 2002 by Taylor & Francis

Therefore, the OGE signal due to ionization of |L1〉 and |L3〉 states can be ex-
pressed as
∆w(t) ⫽ ∆NL1IPL1 ⫹ ∆NL3IPL3
⫽ ∆N0{IPL3 exp[⫺(ΓL3 ⫹ IPL3 ⫹ IPL3s)t] (17)
⫺ IPL1 exp[⫺(ΓL1 ⫹ IPL1 ⫹ IPL1s)t]}

Equation (17) is recognized as a form composed of two exponential decays asso-

ciated with |L1〉 and |L3〉 states and gives the experimental time-resolved OGE
signal as
∆w(t)
w(t) ⫽ ⫽ ae⫺b t ⫺ ce⫺dt (18)
∆N0
where
a ⫽ IPL3 (19)
b ⫽ ΓL3 ⫹ I(PL3 ⫹ PL3s) (20)
c ⫽ IPL1 (21)
d ⫽ ΓL1 ⫹ I(PL1 ⫹ PL1s) (22)
By varying the discharge current, and therefore changing the evolution rate,
a set of time-resolved OGE waveforms can be compiled to reveal the effects of
varying discharge currents. Based on the theoretical model described above, each
spectrum can then be fitted by a nonlinear least-squares method to obtain the
parameters associated with the optical energy states. The effect of the instrumen-
tal time constant τ is crucial in determining the fast time region of the OGE
waveform and can only be determined experimentally [53,54]. The instrumental
response becomes less significant as the OGE signal evolves in the time domain.
When τ is included, Eq. (18) can be rewritten as:
a c
S(t) ⫽ (e⫺bt ⫺ e(t/τ)) ⫺ (e⫺dt ⫺ e(t/τ)) (23)
1 ⫺ bτ 1 ⫺ dτ
where a and c are the positive and negative amplitudes, b and d are proportional to
the decay rates of the 1s states, and τ is the instrumental time constant. Collisional
ionization parameters are proportional to the electron collisional ionization cross
sections associated with the |L1〉, |L3〉 states. Han et al. [25] were successful in
fitting the waveforms resulting from the neon transition at 640.22 nm in the 4–
14 mA current range. The present investigation extended the limits of this same
visible transition down to 0.2 mA. Other neon transitions and several argon transi-

Copyright © 2002 by Taylor & Francis

tions (both in the visible and UV range) were subsequently analyzed to expand
the model further and test the limits of the basic theory.

V. IDENTIFICATION AND ASSIGNMENT

OF OPTOGALVANIC TRANSITIONS USING
THE J–L COUPLING SCHEME

Precise wavelength calibration of tunable lasers is critical when performing LIF

spectroscopy of atoms and molecules. Although some OG spectral wavelength
tables do exist, very few of the compilations are either readily available or com-
plete enough for calibrating tunable laser output in the near-UV region. The laser-
assisted optogalvanic effect provides a good solution for existing wavelength
calibration inadequacies in these regions, which proves vital for combustion and
atmospheric monitoring purposes.
OG spectroscopy can occasionally provide high levels of intensity for tran-
sitions of some species that produce very weak signals or those that are undetect-
able by more traditional methods. The differences in OG and fluorescence spectra
can be particularly dramatic for transitions arising from metastable levels in noble
gases [26], which is not surprising since the intensity of the OG signal is not
an absolute measure of the oscillator strength of the transition. The mechanism
responsible for the OG signal is complex but it is clearly dependent upon the
discharge current.

A. Wavelength Calculations and Edlen’s Formula

In order to accurately identify and assign the new lines of the optogalvanic spectra
of neon and argon, the positions of the rotational lines of the OH radical were
also simultaneously measured and their bin numbers d recorded. The air wave-
lengths λair of the reference OH lines were taken from Dieke and Crosswhite [55],
and their corresponding bin numbers, were fitted to the least-square polynomial:
N

λair ⫽ 冱 a (d ⫺ d )
i⫽0
i 0
i
(24)

where d 0 is the bin number reading of the first reference line and the ai is the
least-square coefficient associated with the polynomial fit of order N. The wave-
lengths in air (in Å) were then converted to vacuum wavenumbers ν (in cm⫺1)
νair ⫽ 108/(η ⫻ λ vac) in cm⫺1 (25)
where λ vac is in 10⫺8 cm and the index of refraction in air η is given by Edlen’s
formula [56]:

Copyright © 2002 by Taylor & Francis

 2,942,810 25,540
η ⫽ 1 ⫹ 6432.8 ⫻ 10⫺8 ⫹ ⫹ (26)
146 ⫻ 108 ⫺ v 2 41 ⫻ 108 ⫺ v 2

Vacuum wavenumbers of the energy levels for neon and argon were taken from
Bashkin and Stoner [57], who have compiled the atomic energy levels and the
pertinent Grotrian diagrams. The source for neon and argon air wavelengths was
a set of tables of spectral lines for neutral and ionized atoms by Striganov and
Sventitskii [58], which we have referred to as the atlas. Additional argon energy-
level tables used were supplied by Sugar [59].

B. Spectroscopy of the Noble Gases and Racah Notation

Once the neon and argon spectra were recorded and precisely cross-calibrated
using the rotationally resolved LIF spectrum of the OH radical, the OG transitions
were assigned using the J–L coupling scheme represented in the Racah notation
[60]. Although cumbersome, Racah’s notation provides the best model of the
physical situation of the excited states of noble gases. Pashen notation, most
commonly used, especially for neon, is simply a system of shorthand symbols
that must be treated as arbitrary names given to the energy levels [60]. Even
though the familiar letters s, p, and d are used, we cannot assume that a Paschen
symbol with the letter s, for example, refers to an outer electron in an s orbit.
To find out which quantum-mechanical model represents a specific level, we
must consult a list of the symbols that relates them to the model, such as Moore’s
atomic-energy-level tables [61]. Figure 3 and 4 illustrate these ideas for the lowest
excited states of neon and argon atoms, respectively.
The p-shell of noble gases, being completely filled, ends with the electron
configurations 2p6, 3p6, 4p6, and so on. The total angular momentum J, the orbital
angular momentum L, and the total spin of such closed shells are all zero. When
excited, one of the p-electrons moves out of its closed shell leaving a five-electron
core behind. In the case of neon, for example, the excited configuration is of the
form 2p53s, 2p53p, and so on (omitting the preceding closed shells, 1s2, 2s2, etc.).
J–L coupling (using the Racah notation) is the best approximation of the
coupling mechanism in an excited noble gas atom. In J–L coupling, the orbital
angular momentum of the outer electron l is coupled to the total angular momen-
tum of the core Jc, and a resultant vector K is obtained by K ⫽ Jc ⫹ l. Total
angular momentum J of the atom is obtained by adding (vectorially) the spin of
the outer electron to K. Racah notation consists of the outer electron configuration
followed by [K], that is, 3s[3/2] or 3s′[1/2], where the prime is an indication of
the antiparallel combination of core orbital momentum and spin. Thus, for the
lowest excited states of neon, two possibilities are obtained from 3s[3/2], where
J ⫽ 2 or 1, and can be written 3s[3/2]2, 3s[3/2]1; while 3s′[1/2]0, 3s′[1/2]1 are
the other two possibilities. Two of these levels are metastable with long radiative

Copyright © 2002 by Taylor & Francis

Figure 3 Plot of the lowest excited states of the neon atom displaying relative energy
levels, the number of terms, and both Racah (R) and Paschen (P) notations. *, Some terms
in this group have special Paschen symbols.

lifetimes, since they have J ⫽ 0 and J ⫽ 2, and cannot undergo an electric dipole
transition to the ground state with J ⫽ 0, because of the selection rule (∆J ⫽ J′
⫺ J ⫽ 0, ⫾ 1; excluding 0-0). The 3p[K] configurations (Jc ⫽ 3/2, 1/2) lead to
10 possible states.
The change in the angular momentum of the core from Jc ⫽ 1/2 to 3/2
affects the energy levels of neon by relatively small amounts, and thus the levels
for different Jc are not shown separately on the neon chart (Fig. 3). The primes
of the Racah symbols are also omitted. The separation of the energy levels with
different Jc increases progressively with increasing atomic number and, therefore,
is already noticeable in the case of argon. Note the grouping of the levels ac-
cording to Jc values as well as the primes are shown in the argon chart (Fig. 4).
The columns in the charts contain infinitely many energy levels with the Racah
configuration numbers ns, ns′, np, np′, and so on. All unprimed sequences tend
to the same limit; the energy of ionization for a singly ionized noble gas atom
in the 2P3/20 state. All primed sequences tend to a different limit, which represents
the ionization energy required to produce an ion in the 2P1/20 state. The latter
energy is greater, according to Hund’s rules, and the energy difference increases
with increasing atomic number.

Copyright © 2002 by Taylor & Francis

Figure 4 Plot of the lowest excited states of the argon atom displaying relative energy
levels, the number of terms and both Racah (R) and Paschen (P) notations. *, Miscellane-
ous or improvised symbols.

Table 1 provides a guideline for the values of the angular momentum of the
core, Jc; the resultant vector, K; the total angular momentum, J; and the number of
terms in the J–L coupling scheme.

C. Assigned Neon Transitions

A series of measurements has previously been carried out to observe the laser-
assisted OG signals of neon in an Fe–Ne hollow cathode lamp in the 337–598
nm wavelength interval [62]. In the present work, a total of 220 OG transitions
were observed. Some of the observed neon transitions were revisited and looked
at from a theoretical perspective. The approach was to confirm the earlier optogal-
vanic assignments by comparing the experimental wavenumbers with calculated
wavenumbers and to extend the earlier analysis by identifying and assigning new
transitions. The transitions that were experimentally observed were analyzed us-
ing the approach described below.
The vacuum wavenumbers of the energy levels for the neon atom were
taken from Bashkin and Stoner [57] for ns, n′s′, np, n′p′, nd and n′d′; and the
vacuum wavenumbers for the allowed transitions (∆ J ⫽ 0, ⫾ 1) were determined

Copyright © 2002 by Taylor & Francis

Table 1 Calculation of Quantum Numbers and the Numbers of Terms in a J–L Coupling Scheme for the Noble Gases
Jc 1 value Allowed K
Excited state (lc ⫽ 1, sc ⫽ 1/2) (outer) values J ⫽ K ⫾ 1/2 Number of terms

2p53s Jc ⫽ 3/2, 1/2 0 3/2, 1/2 2,1; 1,0 4

2p53p Jc ⫽ 3/2, 1/2 0, 1 5/2, 3/2, 1/2 3,2; 2,1; 1,0 6 ⫹ 4 ⫽ 10
3/2, 1/2 2,1; 1,0
2p53d Jc ⫽ 3/2, 1/2 0, 1, 2 7/2, 5/2, 3/2, 1/2 4,3; 3,2; 2,1; 1,0 8 ⫹ 4 ⫽ 12
3/2, 1/2 2,1; 1,0

Copyright © 2002 by Taylor & Francis

by subtracting the energies of the respective levels, s⫺p, s′ ⫺ p, . . . and so
on. These vacuum wavenumbers were then converted into air wavelengths using
Edlen’s formula (Eq. 26) for the refractive index, η. The Striganov and Sventitskii
[58] atlas was used as reference for the neon air wavelengths. The calculated
vacuum wavenumbers, the air refractive indices, the calculated air wavelengths,
together with the air wavelengths reported in the atlas, and the explicitly assigned
optogalvanic transitions have been summarized in Table 2.
Based on Table 2, it is evident that the calculated values agree well with
those reported in the literature, thereby confirming the accuracy of the present
theoretical treatment. All but the last three transitions have been assigned in our

Table 2 Calculated and Observed Neon Optogalvanic Spectral Lines

and Their Assignments

Calculated Calculateda Atlasb Observedc

(νvac in cm⫺1) η (λair Å) (λair Å) (λair Å) Transition

29665.8806 1.00028726 3369.9078 3369.9069 — 3s[1.5]02-4p′[0.5]1

29615.4275 1.88828720 3375.6490 3375.6489 — 3s[1.5]02-4p′[1.5]1
29249.3118 1.00028681 3417.9035 3417.9031 — 3s[1.5]01-4p′[1.5]2
29248.4335 1.00028681 3418.0062 3418.007 — 3s[1.4]01-4p′[0.5]1
29197.9804 1.00028676 3423.9126 3423.9120 — 3s[1.5]01-4p′[1.5]1
28996.5111 1.00028655 3447.7028 3447.7022 — 3s[1.5]02-4p [1.5]2
28970.7809 1.00028652 3450.7650 3450.7641 — 3s[1.5]02-4p [1.5]1
28942.0143 1.00028649 3454.1949 3454.1942 — 3s[1.5]02-4p [0.5]0
28889.0801 1.00028644 3460.5243 3460.5235 — 3s′[0.5]00-4p′[0.5]1
28857.2727 1.00028640 3464.3387 3464.3385 — 3s[1.5]02-4p [2/5]2
28838.6270 1.00028638 3466.5787 3466.5781 — 3s′[0.5]00-4p′[1.5]1
28788.8632 1.00028633 3472.5711 3472.5706 — 3s[1.5]02-4p [2.5]3
28579.0640 1.00028612 3498.0640 3498.0632 — 3s[1.5]01-4p [1.5]2
28553.3338 1.00028609 3501.2163 3501.2154 — 3s[1.5]01-4p [1.5]1
28476.0316 1.00028601 3510.7212 3510.7207 — 3s[1.5]02-4p [0.5]1
28439.8256 1.00028597 3515.1907 3515.1900 — 3s[1.5]01-4p [2.5]2
28397.1683 1.00028593 3520.4712 3520.4714 — 3s′[0.5]01-4p′[0.5]0
27819.8852 1.00028535 3593.5257 3593.5263 — 3s′[0.5]01-4p′[1.5]2
27768.5538 1.00028530 3600.1687 3600.1694 — 3s′[0.5]01-4p′[1.5]1
27699.2311 1.00028524 3609.1790 3609.1787 — 3s′[0.5]00-4p[0.5]1
23469.0079 1.00028148 4259.7395 — 4259.77 3p[2.5]3-13s[1.5]01
23435.2085 1.00028146 4265.8832 — 4265.89 3p[2.5]2-10s′[0.5]01
20288.6972 1.00027915 4927.4772 — 4927.48 3p[0.5]0-10s′[0.5]1
a
Calculated using the energy levels in Ref. 57 and Edlen’s formula [56].
b
From Ref. 58.
c
Observed air wavelength in our laboratory not reported in the atlas [58].

Copyright © 2002 by Taylor & Francis

Table 3 Optogalvanic Spectral Lines of Ar Atoms in a Laser Galvatron (301.800–
320.800 nm)

Observed Atlas λ Calculated

λ (nm) (nm) λ (nm)a ∆λ (nm)b Transition

301.907 — 301.902 — 4s[3/2]02-13p[3/2]2

301.944 — 301.940 — 4s[3/2]02-13p[5/2]2
303.486 — 303.483 — 4s[3/2]02-12p[3/2]1
303.495 — 303.501 — 4s[3/2]02-12p[5/2]3
303.536 — 303.534 — 4s[3/2]02-12p[5/2]2
303.592 — 303.597 — 4s[3/2]02-12p[1/2]1
303.938 — 303.938 — 4s[3/2]01-17p[5/2]2
304.564 — 304.563 — 4s[3/2]01-16p[5/2]2
305.292 — 305.292 — 4s[3/2]01-15p[3/2]2
305.327 — 305.327 — 4s[3/2]01-15p[5/2]2
305.627 305.628 — 0.001 4s[3/2]02-11p[3/2]1
305.696 — 305.696 — 4s[3/2]02-11p[5/2]2
305.717 — 305.716 — 4s[3/2]02-11p[1/2]1
306.207 306.206 — ⫺0.001 4s[3/2]02-8p′[3/2]2
306.285 306.282 — ⫺0.003 4s[3/2]02-8p′[3/2]1
306.347 306.344 — ⫺0.003 4s[3/2]02-8p′[1/2]1
307.538 — 307.538 — 4s[3/2]01-13p[3/2]2
308.647 308.647 — 0.0 4s[3/2]02-10p[3/2]1
308.781 308.781 — 0.0 4s[3/2]02-10p[5/2]3
308.917 308.917 — 0.0 4s[3/2]02-10p[1/2]1
309.018 309.018 — 0.0 4s[3/2]01-12p[1/2]0
309.130 309.132 — 0.002 4s[3/2]02-6f′[5/2]3
309.175 — 309.177 — 4s[3/2]01-12p[3/2]2
309.181 — 309.178 — 4s[3/2]01-12p[3/2]1
310.012 310.010 — ⫺0.002* 4s[3/2]01-8f[5/2]2,3
311.066 311.066 — 0.0 4s[3/2]01-11p[1/2]0
311.406 — 311.405 — 4s[3/2]01-11p[3/2]1
311.497 311.497 — 0.0 4s[3/2]01-11p[1/2]1
311.620 311.622 — 0.002 4s[3/2]01-8p′[1/2]0
312.005 312.006 — 0.001 4s[3/2]01-8p′[3/2]2
313.080 313.080 — 0.0 4s[3/2]02-9p[3/2]2
313.106 313.104 — ⫺0.002 4s[3/2]02-9p[3/2]1
313.230 313.231 — 0.001 4s[3/2]02-9p[5/2]2
313.286 313.287 — 0.001 4s[3/2]02-9p[5/2]3
313.429 313.427 — ⫺0.002 4s[3/2]02-9p[1/2]1
314.264 314.260 — ⫺0.004* 4s[3/2]01-10p[1/2]0
314.540 314.542 — 0.002 4s[3/2]01-10p[3/2]2
314.560 314.563 — 0.001 4s[3/2]01-10p[3/2]1
314.823 314.820 — ⫺0.003 4s[3/2]01-10p[1/2]1
315.040 315.042 — 0.002 4s[3/2]01-6f′[5/2]2

Copyright © 2002 by Taylor & Francis

Table 3 Continued

Observed Atlas λ Calculated

λ (nm) (nm) λ (nm)a ∆λ (nm)b Transition

315.152 315.152 — 0.0 4s[3/2]02-7f[5/2]3,2

315.229 315.229 — 0.0 4s[3/2]02-7f[3/2]1,2
315.957 315.955 — ⫺0.002 4s[3/2]01-8f[5/2]2
316.005 316.006 — 0.001 4s[3/2]01-8f[5/2]1,2
317.222 317.218 — ⫺0.004 4s[3/2]02-7p′[3/2]2
317.296 317.296 — 0.0 4s[3/2]02-7p′[1/2]1
317.371 317.371 — 0.0 4s[3/2]02-7p′[3/2]1
318.663 318.663 — 0.0 4s[3/2]01-9p[1/2]0
319.145 319.150 — 0.005 4s[3/2]01-9p[3/2]2
319.173 319.172 — ⫺0.001 4s[3/2]01-9p[3/2]1
319.492 319.493 — 0.001 4s′[1/2]00-11p[1/2]1
319.510 319.512 — 0.002c 4s[3/2]01-9p[1/2]1
320.036 320.039 — 0.003 4s[3/2]02-8p[3/2]2
320.086 320.084 — ⫺0.002 4s[3/2]02-8p[3/2]1
320.112 320.112 — 0.0 4s′[1/2]00-8p′[3/2]1
320.285 320.285 — 0.0 4s[3/2]02-8p[5/2]2
320.368 320.366 — ⫺0.002 4s[3/2]02-8p[5/2]3
320.749 320.750 — 0.001 4s[3/2]02-8p[1/2]1
a
New OG spectral lines of Ar-atom assigned based on calculations using Eq. (26).
b
∆λ ⫽ (Atlas-observed) value.
c
MIT atlas [63].

laboratory earlier, but the wavelengths determined were either atlas or observed
values. We have treated these same transitions theoretically using the method
described above to calculate their wavelengths. During this process, to the best
of our knowledge, three new neon OG transitions were identified and assigned
and are presented in Table 2.

D. Assigned Argon Transitions

The OG spectrum of argon, recorded using a laser galvatron, was recorded simul-
taneously with the LIF spectrum of the OH radical in a propane–air flame. The
spectral region 290.886–320.750 nm was covered by using several dyes. Equa-
tion (24) was used to calculate the wavelengths of the argon lines in air. These
values were then converted to vacuum wavenumbers ν (cm⫺1) using Eqs. (25)
and (26). Table 3 contains a summary of all the argon transitions observed with
the laser galvatron in the 301.800–320.800 nm wavelength region. The transitions
were calculated based on the energy levels of the ns, n′s′, np . . . atomic levels

Copyright © 2002 by Taylor & Francis

taken from the Grotrian diagram compliation [57] and subsequently compared
with the values of the transitions given in the atlas [58] and the NIST database
compilation [59].

VI. MODELING OF OPTOGALVANIC WAVEFORMS

A. Optogalvanic Waveforms Associated with the Atomic
Neon Transitions in a Hollow Cathode Discharge
For the neon 1s–2p transition, using the Paschen notation, there are four 1s states
and ten 2p states, which yield a total of 30 allowed radiative transitions. These
allowed transitions are shown in Figure 5. Of the four 1s states, two (1s3 and
1s5) are metastable with radiative lifetimes on the order of 1s, while the other
two (1s2 and 1s4) decay to the ground state within 1 ns. Lawler [64] has shown
that the optogalvanic effect in the positive column could be modeled in terms of
the perturbed rate equations, including the dynamic impedance. His theoretical
model showed excellent agreement with the experimental results for helium.
Doughty and Lawler [65] extended the model containing terms of the perturbed-
rate equations, including the dynamic impedance proposed by Lawler, [64] to
neon in the visible region and obtained good agreement with a simplified descrip-
tion of its fairly complex energy level system. Kane [66] has modeled the satu-
rated laser optogalvanic effect in neon and explained the variation in the optogal-

Figure 5 Energy level diagram shows all allowed neon transitions from the first to the
second excited state. 1S–2P, Paschen notation.

Copyright © 2002 by Taylor & Francis

vanic line profile with laser power. The improved theoretical understanding of
the optogalvanic effect in a variety of discharges and under different conditions
has made it an important quantitative diagnostic technique. However, as models
have progressed from a simple basic structure to a more comprehensive inclusion
of important discharge processes, it is apparent that some of the descriptions did
not agree well with the experimental observations [67]. Such discrepancies led
Stewart et al. [68] to develop a model that included and identified specifically and
quantitatively the dominant effects of electron collisional transfer and resonance
radiation wall losses. The authors included separate rate equations (with about
150 parameters) for neon atoms, in each of the four 1s states, and discussed the
agreement of the model with their measured optogalvanic signals for a number
of transitions in the visible region involving 1s4 and 1s5 states, and also over a
wide range of discharge gas pressures and tube currents. Han et al. [25] subse-
quently used a theoretical model involving a handful of parameters (four) to study
the optogalvanic effect associated with the signal of neon that Stewart et al. [68]
had analyzed earlier. They clearly identified and quantitatively characterized the
dominant physical processes that contributed to the production of the OG signal.
Literature survey revealed, to the best of our knowledge, that so far this type of
modeling has not been performed systematically for the atomic neon transitions
that fall in the ultraviolet spectral region. This prompted us to carry out a compre-
hensive study of the time-resolved waveforms of neon transitions in the ultravio-
let region and thereby attempt to identify and qualitatively characterize the domi-
nant physical processes contributing to the production of the OG signal.
To illustrate our ideas, we have chosen the neon 3s[3/2]01 –5p[5/2]2 (1s4 –
4p8 in Paschen notation) transition at 301.735 nm recorded at various discharge
currents: 0.6, 0.8, 1.0, 1.2, 1.4, and 1.6 mA. In case of the 1s4 →4p8 transition
for neon, there are a total of 30 optically allowed transitions, one of which in-
volves the 1s4 and 4p8 states and is presented in the energy level diagram shown
in Figure 6. On laser excitation from the 1s4 to the 4p8 state, the excited neon
atoms relax to optically allowed 1s2, 1s3, and 1s5 states. The hollow cathode lamp
we have used for recording the neon OG spectra presented here is a commercial
Fe–Ne hollow cathode lamp. The wavelengths of the OG-spectral lines of neon
were calibrated using an interference pattern recorded with an etalon. Our main
aim has been to fit the signal independently and to obtain parameters that yield
information about the decay rates, the amplitudes associated with the energy
states involved in the transition, and also to determine the instrumental time con-
stant.
To perform a fit of the time-resolved OG waveforms, we have to consider
(based on Fig. 6) the involvement of four (1s4, 1s2, 1s3, and 1s5) optically decaying
states. According to this scheme, the observed waveform can be fitted to the
following expression

Copyright © 2002 by Taylor & Francis

Figure 6 Schematic representation of the energy level diagram shows the neon (1s4 –
4p8) transition at 301.735 nm and the associated optically allowed transitions.

a c
s(t) ⫽ (e⫺bt ⫺ e⫺t/τ) ⫹ (e⫺dt ⫺ e⫺t/τ)
1 ⫺ bτ 1 ⫺ dτ
(27)
e g
⫹ (e⫺ft ⫺ e⫺t/τ ) ⫹ (e⫺ht ⫺ e⫺t/τ )
1 ⫺ fτ 1 ⫺ hτ
where a, c, e, and g are the positive, negative, positive, and negative amplitudes,
respectively; and b, d, f, and h are the decay rates for the states involved in
the transition, respectively. τ is the instrumental time constant of the waveform
representing the OG signal. Although, in principle, one can include all the four
terms to fit the observed optogalvanic signal, in the present work we have used
only the first two terms in Eq. (27) to fit the observed neon OG waveforms satis-
factorily for three (0.6, 0.8, and 1.0 mA) currents. A nonlinear least-squares pro-
gram was used to fit Eq. (27). The observed and fitted time-resolved optogalvanic
waveforms for 0.6 mA, 0.8 mA, and 1.0 mA currents, respectively, are shown
in Figure 7. The values of the fitted constants along with their standard deviations
for the three currents are given in Table 4. Owing to a high correlation between
the amplitudes a and c, we have considered the value of c to be a constant in
the least-squares fit for the three currents.
The observed time-resolved OG waveform for 1.2 mA current was likewise
fitted to the first two terms in Eq. (27), while the remaining two OG waveforms
recorded at 1.4 and 1.6 mA currents, respectively, were satisfactorily fitted to the
first three terms in Eq. (27). The observed and fitted time-resolved optogalvanic
waveforms for 1.2, 1.4, and 1.6 mA currents, respectively, are shown in Figure
8. The values of the fitted constants, along with their standard deviations, are
given in Table 5.

Copyright © 2002 by Taylor & Francis

Figure 7 Fitted and observed optogalvanic waveforms of the Ne-transition (1s4 –4p8)
at 301.735 nm for three (0.6, 0.8, and 1.0 mA) different currents.

Owing to a high correlation between the amplitudes a and c for the wave-
form recorded at 1.2 mA, and that among the amplitudes a, c, and e for the
waveforms recorded at currents of 1.4 and 1.6 mA, in the present work the values
of c and e were fixed in the least-squares fit. Based on Tables 4 and 5, it is evident
that the parameters b and d (representing the decay rates) show increasing and
decreasing trends, respectively, as the discharge current is increased. However,
the decay rate parameter f remains constant as the current is increased.

Table 4 Fitted Parameters Obtained from a Nonlinear Least-Squares Fit of the

Observed Optogalvanic Waveforms of the Neon Transition at 301.735 nm for 0.6, 0.8,
and 1.0 mA Currents

Parameter Value (for 0.6 mA) Value (for 0.8 mA) Value (for 1.0 mA)

τ 5.47(11) (µs) 5.11(11) (µs) 5.13(14) (µs)

a 8.855(49) ⫻ 10⫺1 (V) 6.705(63) ⫻ 10⫺1 (V) 8.196(91) ⫻ 10⫺1 (V)
b 4.640(44) ⫻ 10⫺1 (µs⫺1) 5.133(63) ⫻ 10⫺1 (µs⫺1) 5.318(81) ⫻ 10⫺1 (µs⫺1)
c ⫺6.78099 ⫻ 10⫺1 (V)a ⫺4.11148 ⫻ 10⫺1 (V)a ⫺5.35445 ⫻ 10⫺1 (V)a
d 3.698(16) ⫻ 10⫺1 (µs⫺1) 3.361(12) ⫻ 10⫺1 (µs⫺1) 3.654(18) ⫻ 10⫺1 (µs⫺1)
a
Fixed in the least-squares fit.

Copyright © 2002 by Taylor & Francis

Figure 8 Fitted and observed optogalvanic waveforms of the Ne-transition (1s4 –4p8)
at 301.735 nm for three (1.2, 1.4, and 1.6 mA) different currents.

The average value of τ obtained from the data set associated with the neon
301.735 nm transition is 5.15 µs ⫾ 5%. A good agreement between different
waveform fits is not always possible because the discharge lamp is part of the
electrical circuit and has a variable nature. The slightest change in conditions,
such as temperature, current setting, laser power setting, or wavelength tuning,

Table 5 Fitted Parameters Obtained from a Nonlinear Least-Squares Fit

of the Observed Optogalvanic Waveforms of the Neon Transition at 301.735 nm
for 1.2, 1.4, and 1.6 mA Currents

Parameter Value (for 1.2 mA) Value (for 1.4 mA) Value (for 1.6 mA)

τ 4.87(32) (µs) 5.87(15) (µs) 5.40(14) (µs)

a 8.60(25) ⫻ 10⫺1 (V) 6.63(17) ⫻ 10⫺1 (V) 7.22(21) ⫻ 10⫺1 (V)
b 5.69(23) ⫻ 10⫺1 (µs⫺1) 9.20(22) ⫻ 10⫺1 (µs⫺1) 1.008(26) (µs⫺1)
c ⫺5.29626 ⫻ 10⫺1 (V)a ⫺4.35617 ⫻ 10⫺1 (V)a ⫺4.36331 ⫻ 10⫺1 (V)a
d 3.699(43) ⫻ 10⫺1 (µs⫺1) 1.791(27) ⫻ 10⫺1 (µs⫺1) 1.799(29) ⫻ 10⫺1 (µs⫺1)
e 2.68471 ⫻ 10⫺1 (V)a 2.68447 ⫻ 10⫺1 (V)a
f 1.487(29) ⫻ 10 (µs ) 1.488(32) ⫻ 10⫺1 (µs⫺1)
⫺1 ⫺1

a
Fixed in the least-squares fit.

Copyright © 2002 by Taylor & Francis

can cause a large change in the plasma response, and thus experimental time
constants can be far from constant within data sets.
Figure 9 represents a plot of the parameter b versus current (I) in mA. It
indicates the linear relationship between relaxation rate and discharge current.
Recalling Eq. (20), b ⫽ ΓL3 ⫹ I(PL3 ⫹ PL3s), where ΓL3 is the decay rate of the
other s states, I is the current, and P is the rate parameter related to the total
electron collisional ionization cross section. The intercept (ΓL3) of the line is the
effective decay rate (0.37/µs), implying an effective lifetime of 2.8 µs for the
upper states (1s2, 1s3, or 1s5). The parameter d associated with the lower state
(1s4) has a relatively constant value of 0.36/µs ⫾ 5% and implies an intercept
(ΓL1) of 0.36/µs corresponding to a lifetime of 2.8 µs. Thus, within our experi-
mental uncertainty, the effective decay rates of both the lower and upper states
have the same value: 2.8 µs. The effective lifetime derived from the s states is
much longer than the radiative lifetime of the 4p8 state (⬃2 ns).
The values of the electron ionization rate parameters PL1 and PL3 are related
to electron ionization cross sections in complicated ways. These depend upon
electron collisional ionization cross sections, the electron energy distributions in

Figure 9 A plot of the parameter b for the Ne transition at 301.735 nm as a function

of discharge current I exhibits direct proportionality between the relaxation rate and the
current.

Copyright © 2002 by Taylor & Francis

the discharge lamp, and the discharge geometry. In cases where redistribution of
the 1S block occurs, they are the combination electron ionization rate parameters
plus electronic collisional transfer rate parameters. The slope of the line in Figure
9, therefore, cannot be directly converted into the electron collisional ionization
cross section associated with the upper state. However, the slope does contain
information about the electron collisional ionization and electronic collisional
transfer rates that can aid theoretical calculations in discharge physics.

B. Optogalvanic Waveforms Associated with the Atomic

Argon Transitions in a Hollow Cathode Discharge
The optogalvanic transitions of atomic argon prove very useful, especially in the
UV, in calibrating tunable dye laser wavelengths because often weak emission
lines could be excited via the optogalvanic effect. Argon and neon are both attrac-
tive choices for use in wavelength calibration of spectra because gaseous or vola-
tile elements are present in a sample discharge volume at higher concentrations
and hence the associated OG signals are easier to measure than those of less
volatile elements. Furthermore, commercial hollow cathode lamps are generally
filled with argon or neon as buffer gas and therefore these atoms are readily
available. The optogalvanic spectrum of atomic argon in the near-ultraviolet re-
gion plays an important role in our understanding of combustion processes be-
cause it overlaps with the electronic spectrum of the hydroxyl radical. OG signals
from argon have been reported in a fragmented manner in the literature. Yet so
far (to the best of our knowledge) there is no systematic study of the time-resolved
waveforms of argon in the UV using a model that includes and identifies specifi-
cally and quantitatively the dominant effects of electron collisional transfer and
resonance radiation wall losses.

C. Argon UV Transition at 297.905 nm

In the present work, we have chosen the argon II 4s2P ⫺ 4p′2Po (J ⫽ 1/2–1/2)
transition at 297.905 nm and recorded its time-resolved OGE waveform at two
(0.25 and 0.5 mA) different currents in order to understand the collisional ioniza-
tion of the excited state of argon atoms in a gas discharge plasma. A typical time-
resolved experimental OG waveform extending from 0 to 250 µs for 0.25 mA
and from 0 to 400 µs for 0.5 mA, together with the corresponding fitted curves,
are shown in Figures 10 and 11. Unlike the optogalvanic signals of neon transi-
tions produced in the hollow cathode tube, which have less noise, the optogal-
vanic signals of argon have oscillations associated with it. It is believed to be
due to the electrical resistance of the lamp and its associated circuit, which, when
combined with the negative resistance, creates characteristic oscillations. Such
oscillations have been found by Lee et al. [69] while studying the optogalvanic
signals of tin with commercial hollow-cathode lamps. The observed waveforms

Copyright © 2002 by Taylor & Francis

Figure 10 Fitted and observed waveforms of the argon II 4s2P–4p′ 2P° transition at
297.905 nm for 0.25 mA.

Figure 11 Fitted and observed waveforms of the argon II 4s2P–4p′2p° transition at

297.905 nm for 0.50 mA.

Copyright © 2002 by Taylor & Francis

were fitted to the expression given in Eq. (27). For the optogalvanic signal re-
corded at a discharge current of 0.25 mA, only the first term in Eq. (27) was
used in the nonlinear least-squares fit to obtain the parameters τ (instrumental
time constant), a (negative amplitude), and b (the decay rate). Curve fitting with
these three parameters gave excellent results, as shown in Figure 10. The fitted
parameters, along with their standard deviations, are given in Table 6. For the
OG signal recorded at a discharge current of 0.5 mA, the two terms in Eq. (27)
were likewise used to obtain a satisfactory fit between the observed and fitted
waveforms. Owing to high correlation between the amplitudes (a and c) and the
decay rates (b and d), only the values of three parameters (τ, a and b), along
with their standard deviations, are given in Table 6. A plot of the values of the
parameter b vs. current would not be convincing evidence for linearity, since
there are only two points (which would result in an automatic straight line), but
the increase in amplitude with increasing current is obvious by nothing the a
parameter values in Table 6.

D. Argon UV Transition at 320.366 nm

One more OG waveform study of the argon 4s[3/2]02 –8p[5/2]3 transition at
320.366 nm has been chosen in the present chapter to allow us to understand the
collisional ionization of a different excited state in a gas discharge plasma. Time-
resolved OG waveforms of three different current (0.2, 0.4, and 0.6 mA) are
shown in Figure 12. The observed waveforms were fitted to the expression given
in Eq. (27). The first two terms were used in the nonlinear least-squares fit to
obtain the relevant parameters, namely, τ (instrumental time constant), a (positive

Table 6 Fitted Parameters Obtained from a Least-

Squares Fit of the Observed Optogalvanic Waveforms
of the Argon Transition at 297.905 nm for Currents
0.25 and 0.5 mA

Current

Parameter 0.25 mA 0.50 mA

τ(µs) 3.44(23) 4.39(59)

a(V) 0.03322(97) 0.091(8)
b(µs⫺1) 6.11(23) ⫻ 10⫺2 1.28(18) ⫻ 10⫺1
c(V)a ⫺0.004(1)
d(µs⫺1) 1.81(54) ⫻ 10⫺2
a
Fixed in the least-square fit.

Copyright © 2002 by Taylor & Francis

Figure 12 Fitted and observed OG waveforms of the argon 4s[3/2]02 –8p[5/2]3 transition
at 320.266 nm for three (0.2, 0.4, and 0.6 mA) different currents.

Table 7 Fitted Parameters Obtained from a Nonlinear Least-

Squares Fit of the Observed Optogalvanic Waveforms of the Argon
Transition at 320.366 nm for Currents 0.2, 0.4, and 0.6 mA

Current

Parameter 0.2 mA 0.4 mA 0.6 mA

τ(µs) 19(2) 8.33 (75) 4.97 (65)

a (V) 0.326(45) 0.294 (27) 0.207 (30)
b (µs⫺1) 0.419(44) 0.416 (30) 0.406 (53)
c (V)a ⫺0.039 ⫺0.053 ⫺0.021
d (µs⫺1) 0.078(64) 0.1336 (74) 0.1052 (83)
a
Fixed in the least-square fit.

Copyright © 2002 by Taylor & Francis

amplitude), and b and d (the decay rate constants). The fitted parameters, along
with their standard deviations, are given in Table 7.

VII. SUMMARY AND CONCLUSIONS

The optogalvanic effect gives rise to efficient, sensitive, and economical spectro-
scopic and detection methods. The OGE complements other optical techniques,
such as traditional absorption and fluorescence spectroscopy and optoacoustic
spectroscopy. It is intrinsically more sensitive than absorption spectroscopy, since
the OGE is based upon a signal whose background is zero, while absorption
spectroscopy records a small variation superimposed on a large signal. When
compared to LIF, the OGE has the advantage of not being affected either by the
background signal due to the discharge luminosity or by the scattering of the
excitation signal. This becomes particularly important when the fluorescence is
detected at the same wavelength as the absorbed radiation.
Of the various laser-assisted spectroscopic and detection methods, the opto-
galvanic effect is economical, since commercially available lamps can be used.
A glow discharge is an inexpensive way to obtain large densities of excited states
in volatile elements, especially those involving metastable states. Gaseous states
of refractory elements are easily produced in hollow cathode lamps via sputtering.
Rydberg states can also be studied by the OG technique, since these can be
reached by exciting radiative transitions starting from metastable levels that are
well populated in discharges, particularly in the noble gases. In general, it has
been demonstrated that OG techniques perform favorably compared to traditional
optical techniques, with the extra advantage of inherent simplicity. However, they
cannot provide the maximum signal-to-noise ratio in detecting spectra.
The present investigation began by recording extensive LOG spectra of
neon and argon in the near-ultraviolet region, where the known and well-
documented neon lines were used to calibrate the rotationally resolved LIF spec-
trum of the hydroxyl radical, which in turn was used to calibrate unknown argon
OG spectral transitions in the near UV. By adopting this method, several new
argon OG transitions were identified and assigned using the J–L coupling
scheme. A dependable set of OG transitions suitable for calibration of tunable
lasers in the visible and near-UV regions are now available, which is so vital for
combustion and atmospheric monitoring purposes. To the best of our knowledge,
all the experimentally observed argon OG transitions in the present study, span-
ning the wavelength region 301.800–320.800 nm (a total of 58 lines), have been
documented optogalvanically in our laboratory.

Copyright © 2002 by Taylor & Francis

 The detailed investigation presented here was centered around the optogal-
vanic effect associated with a DC hollow cathode lamp, which is a special form
of DC discharge characterized by two main regions: the cathode dark region
(where 90% of the voltage drop takes place within a region ⬍1 mm) and an
expanded negative glow region (showing ⬃5% voltage drop). It was also conve-
nient to use commercial hollow cathode lamps because they are relatively inex-
pensive, and can be found containing many elements in the periodic table, and
also possess very low noise levels (on the order of shot noise). By irradiating
the entire cathode cavity, the signals produced were in fact a combination of
many of the different processes described in the theory section. As expected, the
prominent process determining the OGE signal, in the current regimen of the
study, was consistent with the electron collisional ionization of the buffer gas
(either neon or argon). As previously stated, either superposition or perturbation
was used to account for other processes evident in the recorded waveforms.
We found the following general trends:
The amplitude of the OG signal increased with increasing discharge current
until saturation due to an increase in electron population was reached.
The amplitude of the OG signal increased with increasing laser power (for
a fixed discharge current) until optical saturation was reached.
The relaxation rates depended linearly on the current.
Saturation due to increased laser power for a fixed current can be explained
as optical saturation. For each neon and argon transition studied, whenever there
were enough data recorded for a detailed analysis, the linear relationship between
relaxation rates and the discharge current was confirmed.
The results of this investigation support the theory that collisional ioniza-
tion is the dominant factor in producing the OGE signal in the low-current regi-
men (0.2–3.0 mA). This conclusion is based on the fact that the ionization stage
of the OG process is consistent with the collisional ionization model and the fit
between the data and the mathematical model was impressively close.
In addition, the analysis performed to understand time-dependent OG
waveforms generated at different currents yielded quantitative information about
relaxation rates and also determined parameters related to the electron collisional
cross sections of the states involved in the transitions.
Perhaps one of the greatest disadvantages of the OGE is that the useful
regimen of a discharge is limited to a narrow range of parameters. The condition
becomes particularly delicate when the current has to be continually increased
(as is necessary for sputtering in hollow cathode lamps). When performing OG
experiments in commercial lamps, each lamp has a small current and voltage
range, with only minimal changes in pressure possible.
There are times, however, when the OG technique is the best detection

Copyright © 2002 by Taylor & Francis

method available. One example would be when absorption spectroscopy is not
efficient, because the sample is optically thin and there is no fluorescence signal
available, as is often the case in molecular infrared spectroscopy.
The greatest difficulty encountered in the present investigation was the in-
terference from the electrical resonance oscillations while performing waveform
studies in the low-current regimen (0.2–3.0 mA). Working around the inconve-
nience caused by these oscillations led to new insights into the electrical proper-
ties of the hollow cathode lamp. Our study of the oscillations themselves provided
a method of determining the current limits within which dependable OG data
could be taken. It is envisioned that a more specific study of the oscillations
themselves will not only produce more information about the OG phenomenon
but also yield new ways in which the effect may be turned into an advantage to
allow for a better understanding of discharge phenomena. This specific goal was
outside the scope of the present investigations, however, and is recommended as
a subject for further study. It is hoped that the new quantitative information per-
taining to the optogalvanic effect provided here can be used to gain better under-
standing of plasma and gaseous media.

ACKNOWLEDGMENT

We acknowledge the many helpful discussions and critique of the manuscript

provided by Dr. J.C. Travis of the National Institute of Standards and Technology
(Gaithersburg, MD).

REFERENCES

1. D Popescu, ML Pascu, CB Collins, GW Johnson, I Popescu. Phys Rev A 8: 1666,

1973.
2. GS Hurst, MG Payne, SD Kramer, JP Young. Rev Mod Phys 51: 767, 1979.
3. JEM Goldsmith, JE Lawler. Optogalvanic spectroscopy. Contemp phys 11(2): 235–
248, 1981.
4. JC Travis, GC Turk, JR DeVoe, PK Schenck, and CA van Dijk. Prog Analyt Atom
Spectrosc 7: 199, 1984.
5. PD Foote, FL Mohler. Photo-electric ionization of caesium vapor. Phys Rev 26:
195, 1925.
6. FM Penning, Physica 8: 137, 1928.
7. A Terenin, Photoionization of salt vapors. Phys Rev 36: 147–148, 1930.
8. A Garscadden, P Bletzinger, EM Friar. J Appl Phys 35: 3432, 1964.
9. RB Green, RA Keller, GG Luther, PK Schenck, JC Travis. Galvanic detection of
optical absorptions in a gas discharge. Appl Phys Lett 29: 727–729, 1976.
10. RA Keller, R Engleman, Jr., BA Palmer. Appl Opt 19: 836, 1980.

Copyright © 2002 by Taylor & Francis

11. NJ Dovichi, DS Moore, RA Keller, Appl Opt 21: 1468, 1982.
12. BM Suri, R Kapoor, GD Saksena, PRK Rao. Opt Commun 52: 315, 1985.
13. F Babin, P Camus, JM Gagne, P Pillet, J Boulmer. Opt Lett 12: 468, 1987.
14. DS King, PK Schenck, KC Smyth, JC Travis. Appl. Opt 16: 2617, 1977.
15. MH Begemann, RJ Saykally. Opt Commun 40: 277, 1982.
16. JR Nestor, Appl Opt 21: 4154, 1982.
17. BR Reddy, P Venkteswarlu, MC George. Opt Commun 75: 267, 1990.
18. M Duncan, Devonshire. Inst Phys Conf Ser 113(6): 207, 1990.
19. Y Oki, T Izuha, M Maeda, C Honda, Hasegawa, H Futani, J Izumi, K Masuda. Jpn
J Appl Phys 30: L1744, 1991.
20. M Hippler, J Pfab. Opt Commun 97: 347, 1993.
21. KC Smyth, PK Schenck. Chem Phys Lett 55: 466, 1978.
22. VM Gusev, ON Kompanets. Sov J Quantum Electron 17: 1515, 1978.
23. X Zhu, AH Nur, P Misra. J Quant Spectrosc. Radiat Transfer 52: 167, 1994.
24. M-C Su, RR Ortiz, DL Monts. Opt Commun 61: 257, 1987.
25. XL Han, W Wisehart, SE Conner, M-C Su, DL Monts. Collisional ionization of
excited state neon in a gas discharge plasma. Contrib Plasma Phys 34: 439–452,
1995.
26. B Barbieri, N Beverini, A Sasso. Optogalvanic spectroscopy. Rev Mod Phys 62(3):
612, 1990.
27. RS Stewart, KI Hamad, KW McKnight. In: RS Stewart, JE Lawler, eds. Optogal-
vanic Spectroscopy. Institute of Physics Conference Series no 113. Bristol, UK:
Institute of Physics, p 89, 1990.
28. GC Turk, JC Travis, JR DeVoe. Anayltical flame spectrometry with laser enhanced
ionization. Anal Chem 50(6): 817–820, 1978.
29. GC Turk. The application of laser enhanced ionization to analytical flame spectrome-
try, Ph.D. dissertation, University of Maryland, College Park, MD, 1978.
30. PK Schenck, JC Travis, GC Turk, TC O’Haver. Laser-enhanced ionization flame
velocimeter. Appl Spectrosc 36(2): 168, 1982.
31. ML Skolnick. Use of plasma tube impedance variations to frequency stabilize a CO2
laser. IEEE J Quant Electron QE(6): 139–140, 1970
32. RB Green, RA Keller, GG Luther, PK Schenck, JC Travis. Use of an opto-galvanic
effect to frequency-lock a continuous wave dye laser, IEEE J Quant Electron 13:
63, 1977.
33. CP Ausschnitt, GC Bjorklund, RR Freeman. Hydrogen plasma diagnostics by reso-
nant multiphoton optogalvanic spectroscopy. Appl Phys Lett 33: 851–853,
1978.
34. RW Shaw, CM Barshick, LW Jennings, JP Young, JM Ramsey. Mass Spectrum.
10: 316, 1996.
35. CR Webster, CT Rettner. Laser optogalvanic spectroscopy of molecules. Laser Fo-
cus 19: 41–51, 1983.
36. CR Webster, RT Menzies. Infrared laser optogalvanic spectroscopy of molecules.
J Chem Phys 78(5): 2121, 1983.
37. HH Telle. Optogalvanic spectroscopy of molecules and complexes. In: RS Stewart,
JE Lawler, eds. Proceedings of the Second International Meeting on Optogalvanic
Spectroscopy and Allied Topics. Glasgow: Strathclyde University, 1990.

Copyright © 2002 by Taylor & Francis

38. R Walkup, RW Dreyfus, P Avouris. Laser optogalvanic detection of molecular ions.
Phys Rev Let 50(23): 1846, 1983.
39. E Miron, I Smilanski, J Liran, S Lave, G Erez. Dynamic optogalvanic effect in rare
gases and uranium. IEEE J Quan Electron QE-15, 194–196, 1979.
40. N Matsuno. Time-resolved spectroscopy of collisional ionization dynamics in neon
discharge plasma,’’ Master’s Thesis, Butler University, 1997.
41. R Shuker, M Hakham-Itzhaq. Optogalvanic spectroscopy of plasma processes and
autoinization levels. In: RS Stewart, JE Lawler, eds. Optogalvanic Spectroscopy.
Institute of Physics Conference Series 113. New York: Institute of Physics, 263,
1990.
42. JE Lawler, AI Lerence Ferguson, JEM Goldsmith, DJ Jackson, SL Schawlow. Dopp-
ler-free intermodulated optogalvanic spectroscopy. Phys Rev Let 42(16): 1046–
1049, 1979.
43. Abhilasha, S Qian, DL Monts. Laser optogalvanic detection of airborne uranium
particles. Appl Phys B 65: 625–632, 1997.
44. DL Monts, Abhilasha, S Qian, D Kumar, X Yao, SP McGlynn. Comparison of atom-
ization sources for a field-deployable laser optogalvanic spectrometry system.
J Thermophys. Heat Transf 12(1): 66, 1998.
45. Demtroder, W. Laser Spectroscopy, 2nd ed. New York: Springer-Verlag, 405–407,
1996.
46. C Haridass, Y-B She, H Major, P Misra. Proceeding of the International Conference
on Laser ’97, New Orleans, LA. McLean, VA: STS Press, 514–521, 1998.
47. P Misra, C Haridass, H Major. Laser optogalvanic spectroscopy of neon and argon
in a discharge plasma and its significance for microgravity combustion. Proceedings
of the 5th International Microgravity Combustion Workshop, Cleveland, OH, 205,
1999.
48. MF Broglia, F Catoni, A Montone, P Zampetti. Galvanic detection of laser photoion-
ization in hollow-cathode discharges: experimental and theoretical study. Phy Rev
A 36(2): 705–714, 1987.
49. M Brogia, F Catoni, P Zampetti. Temporal behavior of the optogalvanic signal in
a hollow cathode lamp. J Phys (Paris), C7–44, 479, 1983.
50. XL Han, Private communication. Butler University, 1999.
51. JJ Brehm, WJ Mullin. Introduction to the Structure of Matter. New York: John Wi-
ley & Sons, 528, 1989.
52. HE White. Introduction to Atomic Spectra. New York: McGraw-Hill, 203–205,
274–277, 1934.
53. R Shuker, A Ben-Amar, G Erez. Theoretical and experimental study of the resonant
optogalvanic effect in neon discharge. J Phys (Paris), C7–44: 35, 1983.
54. A Ben-Amar, G Erez, RJ Shuker. J Appl Phys 54: 3688, 1983.
55. DH Dieke, HM Crosswhite. J Quant Spectrosc. Radiat. Transfer. 2: 97–199, 1961.
56. BJ Edlen. J Opt Soc Am 43: 339–344, 1953.
57. S Bashkin, JO Stoner, Jr. Atomic Energy Levels and Grotrian Diagrams, vols. 1, 2.
New York: North-Holland, 1978.
58. AR Striganov, NS Sventitskii. Tables of Spectral Lines of Neutralized and Ionized
Atoms. New York: Plenum, 1968.
59. Sugar, J. Private communication, NIST, 1998.

Copyright © 2002 by Taylor & Francis

60. BA Lengyel. Introduction to Laser Physics. New York: John Wiley & Sons, 177,
1967
61. CE Moore. Atomic energy levels. National Standards Reference Data Service, Na-
tional Bureau of Standards, vol 1: 1971.
62. AH Nur. Laser optogalvanic spectroscopy and laser-induced chemical kinetic studies
pertaining to the methoxyl radical. Ph.D. dissertation, Howard University, Washing-
ton, DC, 1994.
63. FM Phelps, III. MIT Wavelengths Tables, vol. 2, Cambridge, UA: MIT Press, 1982.
64. JE Lawler. Phys Rev A, 22: 1025, 1980.
65. DK Doughty, JE Lawler. Phys Rev A, 28: 773, 1983.
66. DM Kane. J Appl Phys 56: 1267, 1984.
67. A Sasso, M Ciocca, E Arimondo, J Opt Soc Am B 5: 1984, 1988.
68. Stewart, RS, RW McKnight, KI Hamad. J Phys D, 23: 832, 1990.
69. SP Lee, EW Rothe, GP Rothe. J Appl Phys 61(1): 109, 1987.

Copyright © 2002 by Taylor & Francis

3
Spectra of the Isotopomers of CO⫹,
N2⫹, and NO in the Ultraviolet

S. Paddi Reddy
Memorial University of Newfoundland, St. John’s,
Newfoundland, Canada

C. Haridass
Howard University, Washington, D.C.*

I. INTRODUCTION

This chapter describes the experimental results obtained from the ultraviolet spec-
tra of the isotopomers of CO⫹, N2⫹, and NO excited in a hollow-cathode discharge
tube of special design, developed in our laboratory. Spectra of these molecules are
of considerable importance in the understanding of several aspects of molecular
spectroscopy and their applications to astrophysical phenomena.
A study of the electronic spectra of the molecular ion CO⫹ enables us to
understand the chemical and physical processes that take place in the solar and
stellar atmospheres, comet tails, and the interstellar space. The molecular ion has
also been used as a monitoring probe for chemical dynamics in the investigation
of the environmental research and combustion processes. The electronic spectrum
of CO⫹ consists of four band systems: the comet tail (A 2 Π i –X 2 ∑⫹), the First
Negative (B 2 ∑⫹ –X 2 ∑⫹), the Baldet-Johnson (B 2 ∑⫹ –A 2 Π i), and the Marchand-
D’Incan-Janin (C 2∆ r –A 2 Π i) systems occurring in the region 1800–8500 Å. Pluv-
inel and Baldet [1] first observed the comet-tail system in the tail of the comet
Morehouse 1908c. Dixon and Woods [2] observed the first terrestrial molecular
ion CO⫹ by microwave spectroscopy. Strong evidence of the presence of CO⫹
in Orion molecular cloud 1 was provided by Erickson and coresearchers [3]. The
comet-tail system of CO⫹ that occurs in the spectral region 3080–8500 Å has

* Current affiliation: Belfry School, Belfry, Kentucky.

Copyright © 2002 by Taylor & Francis

applications in the study of radiative heating of hypersonic spacecrafts at escape
velocity and in the atmospheric fringe of the planet Venus, which contains consid-
erable amounts of CO2. Spectroscopic observations of these objects by rockets
and satellites have increased their importance in astrophysics. The previous work
done on the spectra of the 12C16O⫹ ion by numerous researchers has been reviewed
by Krupenie [4], Huber and Herzberg [5], Jakubek and coresearchers [6], and
Haridass and colleagues [7,8]. Prasad and Reddy [9], Reddy and Prasad [10],
and Prasad and Reddy [11] have reviewed the work done on the A–X, B–A, and
B–X systems of 13C18O⫹, respectively.
Spectra of neutral molecular nitrogen N2 and its ion N2⫹ play a significant
role in the atmospheric and astrophysical phenomena. For example, the Meinel
(A 2 Π i,u –X 2 ∑g⫹) and the First Negative (B 2∑u⫹ –X 2∑g⫹) band systems of 14 N2⫹
were observed in the auroral emission; the First Positive (B 3Πg –A 3∑u⫹), Second
Positive (C 3Πu –B 3Πg), Vegard-Kaplan (A 3∑u⫹ –X 1∑g⫹) and Lyman-Birge-Hope-
field (A 1Πg –X 1∑g⫹) band systems of 14N2 were also observed in the auroral emis-
sions; the First Negative system of 14 N2⫹ was also found in the spectra of comet
tails (see the review article by Lofthus and Krupenie [12]). From radio astronomi-
cal measurements the abundance ratios of isotopes such as 15 N/ 14 N were estimated
for lunar samples, solar wind, and meteoritic samples (see Clayton [13]). The
study of molecular spectra of celestial sources provides the abundance ratios of
isotopes such as 15 N/ 14 N, which, in turn, gives a better understanding of the nu-
clear processes through which the energy is generated in the celestial objects.
The laboratory data on the spectra of 14 N2 , 14 N2⫹, 15 N2 , and 15 N2⫹ would
be very useful in the study of their spectra resulting from the atmospheric and
astrophysical phenomena mentioned above. The spectra of 14 N2 and 14 N2⫹ have
been thoroughly investigated in the laboratory. However, the spectroscopic
data on 15 N2 and 15 N2⫹ are either fragmentary or nonexistent. In an attempt to
obtain comprehensive laboratory data on the spectra of 15 N2⫹, its First Negative
(B 2∑u⫹ –X 2∑g⫹) system has been investigated in considerable detail for the first
time by Reddy and Prasad [14] in our laboratory.
Nitric oxide (NO) is produced in the earth’s atmosphere mainly by the
oxidation of N2O (which is a by-product of microbial metabolism) by the excited
atomic O(1D) and to some extent by high-altitude aircrafts, nuclear blasts, volca-
noes, lightning, and other phenomena [15]. It also plays a role in the destruction
of ozone (O3) but is involved in the photochemical production of O3 in the tropo-
sphere by reaction with smog; it is a major product of internal combustion engines
and combustion power plants [16]. The gamma (γ) (A 2 ∑⫹ –X 2 Πr) system of NO
has important applications in atmospheric science. For example, the bands of this
system are used to measure the NO column densities in the mesosphere in order
to interpret emission in aurora [17], and are also found to be potentially attractive
for an optically pumped ultraviolet laser [18].

Copyright © 2002 by Taylor & Francis

II. THE HOLLOW-CATHODE DISCHARGE TUBE

The details of the design of the hollow-cathode discharge tube and the associated
gas-handling system are shown schematically in Figure 1. The hollow cathode
(F) made from a copper cylinder is 75 mm long, 17 mm in outer diameter, and
1.5 mm in wall thickness. It is sliver-soldered (H) to the lower section of a 19
mm inner diameter Kovar-Pyrex seal (G). The upper end of this seal is joined
to the main Pyrex glass body (D), 140 mm long and 46 mm in outer diameter,
of the discharge tube. The side arm (E), 17 mm in outer diameter, branches out
from the main body. A tungsten anode (A) is fused into the branch of the side
arm. Two quartz (SI-UV) windows (B and C) (ESCO Products Inc., Oak Ridge,
NJ), 1.5 mm thick, are attached to the ground end-surfaces of the anode and
cathode branches of the discharge tube with low vapor pressure Torr Seal (Varian
Associated Inc., Lexington, MA). Ball and socket arrangements (not shown in
Fig. 1) are provided, one for each of the reservoir sections (R1 and R2) and the
discharge tube to facilitate their easy connection to and disconnection from the
pumping system. The discharge tube and the secondary reservoir R2 are thor-
oughly evacuated first. A small quantity of the experimental gas N2 or CO is then
admitted from the primary reservoir R1 into R2 through stopcock S1. By closing
the stopcock S3 , a small amount of the gas from R2 is admitted into the discharge
tube through stopcock S2. Under normal operating conditions, an applied direct
current (DC) voltage of 1100 V from a power supply unit rated at 2000 V and
250 mA gives a discharge current of 65 mA. The discharge is operated when the
gas is at rest. In the best operating conditions the pressure inside the discharge

Figure 1 Schematic diagram of the hollow-cathode discharge tube and the associated
gas-handling system. A. tungsten anode; B, C. quartz windows; D. Pyrex glass body; E.
anode branch; F. copper hollow-cathode; G. Kovar-Pyrex seal; H. silver-soldering; R1 , R2 ,
primary and secondary gas reservoirs; S1 , S2 , S3: stopcocks. Arrow shows the connection to
the vacuum system.

Copyright © 2002 by Taylor & Francis

tube is ⬃100 Pa as read on a thermocouple gauge. The hollow-cathode discharge
tube described here is similar in some respects to the one used by Herzberg and
coresearchers [19], who excited the emission spectra of triatomic H3 and D3 in
the cathode column of a hollow-cathode discharge tube cooled with liquid nitro-
gen and operated in a vertical configuration with flowing H2 and D2 , respectively.
These spectra correspond to the transitions between their Rydberg states.
The emissions from the anode column and cathode glow with the carbon
monoxide gas were photographed on a 2 m Bausch and Lomb dual grating spec-
trograph in the first order of a 1200 grooves/mm grating blazed at 1.0 µm. Those
with the molecular nitrogen gas were photographed on the same spectrograph in
the first order of a 600 grooves/mm grating blazed at 2.5 µm. Kodak Spectrum
Analysis No 1 and 103-F plates were used to photograph the spectra of N2 and
N2⫹ from 334 to 523 nm. Kodak SWR plates were used to record the spectra of
CO, CO⫹, and CO2 from 210 to 300 nm. Exposure times varied from 15 to 30
min. A DC Fe arc was used as a source for the reference spectra. Details of the
hollow-cathode discharge tube are given by Reddy and Prasad [20].
As an example to illustrate the characteristics of the hollow-cathode dis-
charge tube, the spectra of the anode column and cathode glow obtained with
the N2 gas from 334 to 523 nm are shown in juxtaposition in Figure 2. The

Figure 2 Spectra excited in the hollow-cathode discharge tube with the N2 gas from
3340 to 5230 Å. Top. Spectrum of the anode column shows the Second Positive (C 3Πu –
B 3Πg) system of the neutral 14 N2 molecule. Bottom. Spectrum of the cathode glow shows
the First Negative System (B 2 ∑u⫹ –X 2∑g⫹) of 14 N2⫹. Excitation conditions for both spectra
are identical.

Copyright © 2002 by Taylor & Francis

spectrum of the anode column (indicated by (A)) contains almost exclusively the
Second Positive (C 3Πu –B 3Πg) system of the neutral N2 molecule. The only ex-
ception is the appearance of the prominent 0-0 band of the First Negative system
of N2⫹ with low intensity. The spectrum of the cathode glow (indicated by (B)
in Fig. 2) consists of the First Negative (B 2∑u⫹ –X 2∑g⫹) system of N2⫹. Also seen
in Figure 2B is the occurrence of some normally very intense bands of the Second
Positive system of N2 in the cathode glow with comparatively lower intensity.
The vibrational assignments of the bands of N2 and N2⫹ are identified in Figure
2 and their wavelength data can be obtained from Lofthus and Krupenie [12].

III. ELECTRONIC CONFIGURATIONS

The type of molecular binding and the nature of the electronic states of a molecule
are determined by electrons in the outermost shells of the constituent atoms of
the molecule. The electronic configurations (Herzberg [21]) of the ground states
of the C, N, and O atoms are:
C: K 2s22p2; N: K 2s2 2p3; O: K 2s2 2p4.
The electronic configurations of the ground (X 2 ∑⫹) and the first two excited
states (A 2 Π i and B 2 ∑⫹) of CO⫹ are:
K K (zσ)2 (yσ)2 (wπ)4 (xσ): X 2 ∑⫹
... ... ... (yσ)2 (wπ)3 (xσ)2: A 2Πi
... ... ... (yσ) (wπ)4 (xσ)2: B 2 ∑⫹
The electronic configuration of the ground X 2∑g⫹ state and the first excited state
A 2 Π i of homonuclear N2⫹ are given by
K K (zσg )2 (yσu )2 (wπu)4 (xσg): X 2 ∑⫹
... ... ... ... (wπu)3 (xσg)2: A 2 Π i,u
The next excited states B 2∑g⫹ and C 2∑u⫹ of N2⫹ arise from the mixture of the
following configurations:
K K (zσg)2 (yσu) (wπu)4 (xσg)2
and
. . . . . . . . . (yσu)2 (wπu)3 (xσg) (vπ)
The configurations of the ground state X 2 Πr and the first excited doublet state
A 2 ∑⫹ of NO are:
K K (zσ)2 (yσ)2 (wπ)4 (xσ) 2 vπ: X 2 Πr
... ... ... ... ... ... uσ: A 2 ∑⫹

Copyright © 2002 by Taylor & Francis

In these configurations the electron from the antibonding vπ orbital is excited to
the uσ Rydberg orbital.

IV. GENERAL SPECTROSCOPIC EQUATIONS

The total energy E (in ergs) of a diatomic molecule (excluding translational and
nuclear spin energies) is expressed by [21]:

E ⫽ Ee ⫹ Ev ⫹ Er (1)
and the term value T (in cm⫺1) in terms of the vibrational and rotational quantum
numbers v and J respectively is written as
T ⫽ T e ⫹ G(v) ⫹ F v (J) (2)
where Te , G (v) and Fv (J) are the electronic, vibrational, and rotational terms,
respectively. The electronic terms of different multiple components of an elec-
tronic state (such as 2 Π, etc.) are
T e ⫽ T 0 ⫹ AΛ ∑ s (3)
where T0 is the electronic term, Α and ∑s are the quantized projections along the
internuclear axis of the electron orbital angular momentum L and its spin angular
momentum S. A is the spin orbit coupling constant and is positive or negative
depending on whether the electronic state is regular or inverted. For ∑ and Π
states Α ⫽ 0 and 1 respectively. The electronic state ∑ is designated as ∑⫹ or
∑⫺ depending on whether the wave function Ψe remains unchanged or changes
sign upon reflection at the plane passing along the internuclear axis. The multi-
plicity of an electronic state is given by 2S ⫹ 1, which is the number of compo-
nents along the internuclear axis. The vibrational term G(v) is expressed as
G(v) ⫽ ω e (v ⫹ 1/2) ⫺ ω e x e (v ⫹ 1/2) 2 ⫹ ω e y e (v ⫹ 1/2) 3 ⫹ ⋅ ⋅ ⋅ (4)

where ωe is the vibrational frequency and ωe xe , ωe ye , and so on are the anharmonic

terms of the vibrational motion. The wave number νo of the band origin is given
by
ν o ⫽ ν e ⫹ G′ (v′) ⫺ G″ (v″) (5)

The isotopic shift ∆ν of a vibrational band is given by the expression

∆ν ⫽ (1 ⫺ ρ) [ω′e (v′ ⫹ 1/2) ⫺ ω″e (v″ ⫹ 1/2)]
⫺(1 ⫺ ρ 2) [ω′e x′e (v′ ⫹ 1/2) 2 ⫺ ω″e x″e (v″ ⫹ 1/2) 2] (6)
⫹(1 ⫺ ρ 3) [ω′e y′e (v′ ⫹ 1/2) 3 ⫺ ω″e y″e (v″ ⫹ 1/2) 3]

Copyright © 2002 by Taylor & Francis

where ρ ⫽ (µ1 /µ2)1/2, µ1 and µ2 being the reduced masses of the isotopomers.
The rotational and stretching constants Bv and Dv are expressed as
B v ⫽ B e ⫺ α e (v ⫹ 1/2) ⫹ γ e (v ⫹ 1/2) 2 ⫹ δ e (v ⫹ 1/2) 3 ⫹ ⋅ ⋅ ⋅ (7)
and
D v ⫽ D e ⫹ β e (v ⫹ 1/2) ε e (v ⫹ 1/2) 2 ⫹ ⋅ ⋅ ⋅ (8)

V. C16Oⴙ
12

A. Earlier Work on 12C16Oⴙ

1. The Comet-Tail System
After the earlier laboratory work of Fowler [22,23] on 12C16O⫹, several researchers
observed the bands of the comet-tail (A–X) system (3080 Å–8500 Å) of this
molecular ion [4,5]. Baldet [24] observed most of the known bands of this system
and Birge [25] identified the transition as A 2 Πr – 2 ∑⫹. Coster and colleagues [26],
Schmid and Gerö [27], and Bulthuis [28] performed the rotational analysis of
several of these bands (v′ ⫽ 2 to 11 and v″ ⫽ O–2). However, Rao [29] did the
rotational analysis of the 0–2, 0–3, and 0–4 bands of this system and identified
the upper state A as an inverted 2 Π (i.e., 2 Π i) state. Rao revised the vibrational
assignments by lowering the previous assignments by three units. These new
assignments were later confirmed by Asundi and colleagues [30] and Dhumwad
et al. from the observed isotope shifts of the bands of the A–X system and the
B–A system of 13C16O⫹ and 12C18O⫹, respectively, from the corresponding bands
of 12C16O⫹. Gagnaire and Goure [32] reanalyzed the 2–0 band, but their data were
found to be erroneous by Coxon and Foster [33]. Katayama and Welsh [34] and
Brown et al. [35] reinvestigated the 0–0 band observed in laser-induced fluores-
cence. Coxon and Foster [33] carried out the deperturbation analysis of the experi-
mental data of four bands of earlier researchers: 5-0 (i.e., 8–0 band) [26], 10–
1 and 10–2 (i.e., 13–1 and 13–2) bands [28], and 0-2 band [29] and concluded
that the A, v ⫽ 0, 5, and 10, vibrational levels are perturbed. They found that
the perturbation shifts up to ⬃14.5 cm⫺1 in the A 2 Π1/2 ⫺ X 2 ∑⫹ subband at J ⫽
14.5 and 22.5 cm⫺1 in the two doublet components of A 2 Π1/2 , v ⫽ 0, 2.5 ⱕ J ⱕ
7.5 in the R21 and Q32 branches. They suggested that this perturbation was caused
by the mixing of the A 2 Π1/2 , v ⫽ 0 level, with X 2 ∑⫹, v ⫽ 10 level, which is
associated with the unusual double crossing between the e levels of X 2 ∑⫹. The
perturbations in the A 2 Π, v ⫽ 0 level, relative to those in the A 2 Π, v ⫽ 5 and
10 levels, are weak. As the off-diagonal spin-orbit matrix elements of the A, v ⫽
0–X, v ⫽ 0, interaction were obtained incorrectly in [34]. Coxon and Foster [33]
refitted the data [34] for the 0–0 band of the A–X system. Vujisic and Pesic [36]
reinvestigated the rotational structure of the 4–0, 3–0, and 0–2 bands.

Copyright © 2002 by Taylor & Francis

2. The First Negative (B 2 ∑⫹ –X 2 ∑⫹) System
The vibrational assignments of this system were given by Blackburn [37] and
Biskamp [38]. Coster et al. [26] and Schmid [39] performed the rotational analy-
sis of a few bands of this system prior to Biskamp [38] without providing reliable
rotational constants. Schmid and colleagues [27] did the rotational analysis of
the 0–0 band of this system and established that the ground state X is the common
lower state of the comet-tail and first negative systems. Woods [40] studied the
splitting of B 2 ∑⫹ and X 2 ∑⫹ states. Rao [29] analyzed 15 bands of this system:
the | γ ′–γ ″ | value reported by him is one-half of that reported by Woods. Recently
Misra and co-researchers [41] re-examined the 0–0, 0–1, and 0–2 bands of the
B–X system and reported the spin-splitting constants of several vibrational levels
of the X and B states by combining the earlier experimental data of Rao [29].
They supported the latter’s conclusions regarding the existing controversy about
the magnitude of the | γ ′ ⫺ γ″| differences.

3. The Baldet-Johnson (B 2 ∑⫹ –A 2 Π i) System

After the initial observations of this system by Baldet [42], Johnson [43] and
Bulthuis [44] performed the rotational analysis of the 0–0 and 0–1 bands. Rao
and Sarma [45] summarized the results of the A–X, B–X, and B–A systems of
CO⫹. Conkić and coresearchers [46] reinvestigated the rotational structure of the
0–1 band, but others [35] who attempted to reanalyze the former’s wave-number
data claimed that their results were questionable because of possible errors in
calibration and measurements. Jakubek and colleagues [6] performed the rota-
tional analysis of the 1–0, 0–0, and 0–1 bands of this system under higher resolu-
tion and obtained molecular constants of the B 2 ∑⫹, v ⫽ 0 and 1 and A 2 Π i , v ⫽
0 and 1 states.

4. The Marchand-D’Incan-Janin (C 2∆r – 2 Π i) System

This system, which is inherently weak, was first observed at lower resolution by
Marchand et al. [47], who assigned the vibrational quantum numbers for the
observed bands. Cossart and Cossart-Magos [48] observed several bands of this
system in the presence of Ne and did the rotational analysis of the 0–2, 1–1, and
2–2 bands, photographed under higher resolution. These authors confirmed the
vibrational assignments of Marchand et al. [47] by observation of isotopic shifts
of the 13C16O⫹ bands from those of 12 C16O⫹. They obtained molecular constants
of C 2∆r by holding the constants for the A 2 Π, v ⫽ 2, level, determined from
Gagnaire and Goure [32], fixed in the fit, and those of the A, v ⫽ 1 by interpola-

Copyright © 2002 by Taylor & Francis

tion. However, data from Gagnaire and Goure [32] were found erroneous by
Coxon and Foster [33].

5. Microwave and Infrared Studies

The 12C16O⫹ ion has a strong electric dipole moment and hence is expected to
have strong microwave and infrared spectra. Its pure rotational spectrum is the
X 2 ∑⫹, v ⫽ 0 level, which was observed in the laboratory by Sastry and core-
searchers [49] and Piltch and coresearchers [50] in the millimeter region, and by
van den Heuvel and colleagues [51] in the submillimeter region. Bogey and co-
workers [52,53] reported observing rotational transitions in the X, v ⫽ 0–4 levels.
Davies and Rothwell [54] measured the fundamental band in the X state of CO⫹
in emission using velocity modulation technique. Haridass et al. [7] reviewed the
earlier work done on the four electronic band systems of 12 C16O⫹ and its infrared
and microwave spectra. They have also analyzed the rotational structure of the
individual bands of the A–X system and reanalyzed the data of the B–X [55,41]
and B–A [6] systems, and the infrared [54] and microwave data of the v ⫽ 0,
1, and 2 levels [49,53].
The molecular constants and band origins obtained from these analyses
were combined and all multiple estimates were reduced to a single set of values
using a grand merge procedure. It was believed that the nonsmooth variations
of molecular constants in the A 2 Π i state, especially the higher-order constants
such as Dv , ADv , pv , and qv , are attributed to the neglect of perturbations. Bembe-
nek and coresearchers [56] recently recorded new bands in the Baldet-Johnson
(B 2 ∑⫹ –A 2 Π i) system of 12C16O⫹ and performed rotational analysis. They have
excluded the perturbed lines of the bands with a common lower v″ ⫽ 0 level
and claimed that the molecular constants of the B 2 ∑⫹ and A 2 Π i states thus ob-
tained show a smooth variation with the vibrational quantum number (v ⫹ 1/2).
They pointed out that the Λ-doubling parameters pv and qv of the A 2 Π i state
obtained by Haridass and coresearchers [7] do not obey traditionally recognized
polynomial dependence on the vibrational quantum number (v ⫹ 1/2). Weak
perturbations in the A 2 Π i , v ⫽ 0, state at 2.5 ⱕ J ⱕ 6.5 due to X 2 ∑⫹, v ⫽ 10
state result in extra lines of the subband of A 2 Π1/2 –X 2 ∑⫹, which have been
observed by Katayama and Welsh [34] in the laser-induced spectra of 12 C16O⫹.
Deperturbation analysis for the A 2 Π i state of CO⫹ from the available spectro-
scopic data [26,28,34,55] of the perturbed levels A, v ⫽ 0, 5, and 10 using a
modern direct fitting approach was carried out by Coxon and Foster [33]. These
authors have also calculated an accurate Rydberg, Klein, Rees (RKR) potential
curve for the X 2 ∑⫹ state and tabulated the term values and the RKR turning
points for high-lying vibrational levels. Remeasurement and subsequent analysis
of the 0–0 band of the A–X system of CO⫹ using laser-induced fluorescence

Copyright © 2002 by Taylor & Francis

excitation were also carried out by Brown and coresearchers [35]. From the time-
resolved laser-induced fluorescence of CO⫹, radiative decay rates and the veloc-
ity-averaged cross sections for the A 2 Π i (v ⫽ 0–3) levels have been obtained
by Miller and colleagues [57,58].

B. Experimental Details
The molecular ion CO⫹ was excited in the cathode column of the hollow-cathode
discharge tube. The high-resolution spectra of seven bands were photographed
on the 3.4 m Jarrell-Ash spectrograph equipped with a 1200 grooves/mm grating
blazed at 1.4 µm: five bands (4–0, 3–0, 2–0, 1–0, and 2–1) in the third order and
two bands (1–1 and 0–2) in the second order. Three bands were photographed on
the 2.0-m Bausch and Lomb spectrograph, two bands (0–1 and 0–3) in the second
order of a 1200 grooves/mm grating blazed at 1.0 µm, and one band (0–4) in
the third order of a 600 grooves/mm grating blazed at 2.5 µm. The slit width
was maintained at 30 µm for the former and 20 µm for the latter. The reciprocal
dispersions of the spectra photographed varied from 0.60 Å/mm at 3820 Å in
the third order to 1.05 Å/mm at 7200 Å in the second order. Overlapping orders
of the spectra were eliminated using Corning and Hoya glass filters. The exposure
times on the photographic plates (103 a-o, 103-F, and 1-N) varied from 45 min
to 12 h. An Fe–Ne hollow-cathode lamp was used as source for the reference
spectra whose wavelengths were taken from work by Crosswhite [59].

C. Analysis of the Spectra

The vacuum wavenumbers of the band heads of 17 bands of the comet-tail
(A 2 Π i –X 2 ∑⫹) system of CO⫹, their relative intensities, and the vibrational as-
signments are given in Table 1. In this system, for the bands 8–1, 7–1, 5–0, 3–
3, and 1–2, for which the rotational analysis was not carried out, the estimated
band origins (T′π –T″∑) using the derived molecular constants (see below) are also
included in Table 1. Each of the bands of this system gives rise to four heads
formed by the R21ee , Q21fe /R22ff , R11ee , and Q11fe /R12ff branches in the order of de-
creasing wavenumber. The two heads formed by the first two branches are from
the 2 Π1/2 – 2 ∑⫹ component and the other two heads formed by the latter two
branches are from the 2 Π3/2 – 2 ∑⫹ component. The head due to the R11ee , branch
of the 8–1 band and the head due to the Q21fe /R22ff branches of the 7–1 band
could not be measured. The upper A 2 Π i state of the comet-tail system belongs
to Hund’s case (a) and its lower X 2 ∑⫹ state belongs to Hund’s case (b). The
rotational structure of a band arising from a 2 Π i – 2 ∑⫹ transition gives rise to 12
branches designated R21ee , R22ff , Q21ef , Q22ef , P21ee , and P22ff from 2 Π1/2 – 2 ∑⫹, and
R11ee , R12ff , Q11fe , Q12ef , P11ee , and P12ff from 2 Π3/2 – 2 ∑⫹, as shown schematically in
Figure 3. The parity of the levels shown in the e/f notation is according to Brown

Copyright © 2002 by Taylor & Francis

Table 1 Comet-Tail (A 2 Π i –X 2 ∑ ⫹) Band System of 12 C 16 O ⫹ Molecule

Band head a Band origin b Band origin c Relative Assignment

(cm ⫺1) T′π –T″∑ (cm ⫺1) T′–T″ (cm ⫺1) intensity d v′–v″

29826.11 e 29750.58 f vw 8–1

29817.05 e
29692.28 e
29288.41 29212.897(2) 29212.896(2) w 6–0
29276.55
29163.94
29153.11
28484.47 e 28403.09 f w 7–1
28367.51 e
28347.42 e
27891.72 27812.16 f m 5–0
27878.69
27763.28
27751.66
26462.51 26384.722(1) 26384.720(1) s 4–0
26449.35
26336.33
26324.26
25009.13 24930.349(1) 24930.349(2) vs 3–0
24994.99
24882.58
24870.02
23529.01 23449.235(1) 23449.253(1) vs 2–0
23514.03
23402.28
23388.91
22022.07 21941.223(1) 21941.215(1) m 1–0
22006.17
21895.08
21880.93
21345.58 21265.241(1) 21265.239(1) m 2–1
21330.09
21218.80
21204.94
19839.33 19757.564(2) 19757.565(1) m 1–1
19822.62
19712.07
19697.34

Copyright © 2002 by Taylor & Francis

Table 1 Continued
a
Band head Band origin b Band origin c Relative Assignment
(cm ⫺1) T′π –T″∑ (cm ⫺1) T′–T″ (cm ⫺1) intensity d v′–v″

18550.12 18469.59 f vw 3–3

18535.64 2
18424.15 e
18409.14 e
18305.32 18222.331(2) 18222.289(1) s 0–1
18287.64
18177.66
18162.05
17685.77 e 17603.70 f vw 1–2
17674.91 e
17559.40 e
17542.94 e
17070.47 16988.472(2) 16988.480(1) vw 2–3
17053.51
16943.22
16928.20
16152.55 16068.569(2) 16068.578(1) s 0–2
16133.97
16024.72
16008.42
14030.53 13945.476(1) 13945.473(1) m 0–3
14011.01
13902.46
13885.49
11938.78 11852.560(1) 11852.557(1) w 0–4
11918.22
11810.37
11792.68
a
The four heads identified for each band are formed by the R 21ee , Q 21fe /R 22ff , R 11ee , and Q 11fe /R 12ff
branches in the order of decreasing wavenumber.
b
The number in parentheses indicates the uncertainty in the last digit and corresponds to one standard
deviation.
c
Band origins obtained from the global fit.
d
Abbreviations for the relative intensities vs, s, m, w, and vw represent very strong, strong, medium,
weak, and very weak, respectively.
e
Measured from the medium dispersion spectrum.
f
Estimated value from the derived molecular constants.

Copyright © 2002 by Taylor & Francis

Figure 3 A schematic energy level diagram shows the first few rotational transitions
for all 12 branches of a band of a 2 Π i – 2 ∑⫹ system.

Copyright © 2002 by Taylor & Francis

Table 2 Matrix Elements of the Hamiltonian
for States 2 ∑ ⫹ and 2 Π

Molecular Matrix
constant Labeling a element b,c

Tv 1, 1 1
2, 2 1
3, 3 1
Bv 1, 1 x(x ⫿ 1)
2, 2 x2 ⫺ 1
3, 3 x2 ⫹ 1
2, 3 ⫺(x 2 ⫺ 1) 1/2
Dv 1, 1 x 2 (x ⫿ 1) 2
2, 2 ⫺x 2 (x 2 ⫺ 1)
3, 3 ⫺x 2 (x 2 ⫹ 3)
2, 3 2x 2 (x 2 ⫺ 1) 1/2
γv 1, 1 0.5(x ⫿ 1)
Av 2, 2 0.5
3, 3 ⫺0.5
A Dv 2, 2 0.5 (x 2 ⫺ 1)
3, 3 ⫺0.5 (x 2 ⫹ 1)
pv 3, 3 ⫿0.5x
qv 3, 3 ⫿x
2, 3 ⫾0.5x(x 2 ⫺ 1) 1/2
a
Labels 1, 2, and 3 refer to states 2 ∑ ⫹, 2 Π 3/2, and 2
Π 1/2 , respectively.
b
x ⫽ J ⫹ 0.5.
c
In the notation ⫾ and ⫿ the upper and lower signs refer to
the e and f levels, respectively.
Source: Ref. 60.

and coresearchers [60]. As the bands of this system are degraded to longer wave-
lengths, the R21ee , Q21fe /R22ff , R11ee , and Q11fe /R12ff branches form four different
heads in the rotational structure of a band. The characteristic band origin
T π′ –T″∑ for a band is estimated from the analysis (see below). The band origins
of the 2 Π1/2 – 2 ∑⫹ and 2 Π3/2 – 2 ∑⫹ subbands of a band are obtained from (T′π –T ″∑)
⫾ (1/2)Av , Av being the spin–orbit coupling constant of the 2 Π state. The effective
Hamiltonian for the 2 Π and 2 ∑⫹ states of diatomic molecules, discussed in detail
by Brown and colleagues [61], was used to analyze the rotational structure of
the comet-tail bands of 12C16O⫹. A complete list of the matrix elements of this
Hamiltonian for the 2 Π and 2 ∑⫹ states used by Haridass et al. [7,8] are taken
from Amiot and coresearchers [62] and Douay and co-workers [63] and are listed
in Table 2. The relevant matrix elements of the Hamiltonian are:

Copyright © 2002 by Taylor & Francis

 ∑ e,f : T v ⫹ B v x (x ⫿ 1) ⫺ D v x 2 (x ⫿ 1) 2 ⫹ 0.5 γv (x ⫿ 1)
2

2
Π e,f
1/2 : T v ⫺ 0.5A v ⫺ 0.5A Dv [x ⫹ 1] ⫹ B v [x ⫹ 1]
2 2

⫺ D v x 2 [x 2 ⫹ 3] ⫿ 0.5p v x ⫿ q v x
2
Π e,f
3/2 : T v ⫺ 0.5A v ⫺ 0.5A Dv [x ⫺ 1] ⫹ B v [x ⫺1]
2 2
(9)
⫺ D v x 2 [x 2 ⫺ 1]
{ Π 1/2 , Π 3/2} : ⫺ B v [x 2 ⫺ 1] 1/2 ⫹ 2.0D v x 2 [x 2 ⫺ 1] 1/2
2 2 e,f

⫾ 0.5q v x[x 2 ⫺ 1] 1/2 ,

where Bv and Dv are the rotational constants, γv is the spin-rotation constant, Av
and ADv are the spin-orbit constants, p and q are the Λ-doubling constants, and
x ⫽ (J ⫹ 0.5). The upper and lower state signs ⫾ or ⫿ in the above matrix
elements refer to the e/f levels, respectively. The rotational structure of a section
of the 0–2 band of the comet-tail system photographed on the 3.4 m Jarrell-Ash
spectrograph in the second order of the 1200 grooves/mm grating is shown in
Figure 4. The four heads and the 10 branches (out of the expected 12) of their
rotational structure are clearly identified. Each of the branches, Q22ef /P21ee in the
2
Π1/2 – 2 ∑⫹ subband and R12ff /Q11fe in the 2 Π3/2 – 2 ∑⫹ subband, forms two close pairs
and appears as one branch. The intensities of the lines of the resolved pairs (R22ff ,
Q21fe) and (Q12ef , P11ee) are found to be approximately equal. These pairs of
branches give directly the spin-splitting in the ground 2 ∑⫹ state. To verify the
observed intensity pattern of a band in the 2 Π i – 2 ∑⫹ transition of CO⫹, intensities
of the branches were calculated using expressions given by Earls [64]. The calcu-
lated intensities are displayed in Figures 5 and 6. In these figures, it is clearly
seen that the ratio of the intensities of the unresolved pairs (Q22ef /P21ee and Q11fe /
R12ff) is approximately 3:1 and the ratio of the resolved pairs (Q22ef and P21ee and
Q12ef and P11ee) is approximately 1:1. In the case of the 1–0 and 1–1 bands (i.e.,
bands with a common upper level), it was possible to observe 12 branches since
the bands were strong and the rotational lines were resolved. To rule out the
possibility of systematic shifts in the observed spectra due to various experimen-
tal procedures, the spectra can in principle be calibrated with the atomic lines of
carbon and oxygen if they appear in the CO⫹ spectrum itself. However, these
atomic lines were not observed [7,8]. The emission spectrum has the advantage
of displaying several bands involving the same upper vibrational levels and there-
fore provides a critical check of the assignments and systematic shifts. Since the
ground (2 ∑⫹) state rotational levels of CO⫹ are known accurately, it is possible
to calculate the shifts for the bands arising from the same v′ and different v″, by
combining the observed wavenumbers of the rotational levels with the corre-
sponding term values of the ground state. The ground state term values for v ⫽
0, 1, 2, 3, and 4 for both e and f levels are calculated from the Dunham parameters
and the spin-rotation constant γv given by Bogey et al. [53] and the G(v) values
calculated by Coxon and Foster [33]. For example, the rovibrational term values

Copyright © 2002 by Taylor & Francis

Figure 4 Rotational structure of a section of the 0–2 band of the comet-tail (A 2 Π i –
X 2 ∑⫹) system of 12C16O⫹, photographed on the 3.4 m Jarrell-Ash spectrograph in the
second order of a 1200 grooves/mm grating.

of A 2 Π i , v ⫽ 0, e and f levels are obtained by combining the observed wavenum-

bers of the 0–1, 0–2, 0–3, and 0–4 bands of the A–X system to the corresponding
term values of the v ⫽ 1, 2, 3, and 4 levels of the X 2 ∑⫹ state. The term values
of the v′ ⫽ 0 level calculated from the 0–2 band are found to be consistently
lower by 0.197 cm⫺1 than those calculated from the 0–1, 0–3, and 0–4 bands.
Using a similar procedure, the rovibronic term values of the A 2 Π i , v ⫽ 1 and
2 levels are calculated by combining the wavenumbers of the 1–0 and 1–1, 2–
0, and 2–1 bands, respectively, of the A–X system to the corresponding term
values of the X 2 ∑⫹, v″ ⫽ 0 and 1 as mentioned above. The calculated term
values of the v′ ⫽ 1 level from the 1–0 band are found to be consistently lower
by 0.265 cm⫺1 than those calculated from the 1–1 band. A systematic shift was
observed for the term values of the v′ ⫽ 2 level calculated from the 2–1 band
with those calculated from the 2–0 band and is found to be ⫺0.113 cm⫺1. The
corrected vacuum wavenumbers and the rotational quantum numbers of the spec-

Copyright © 2002 by Taylor & Francis

Figure 5 Intensity distribution of the six branches in the 0–2 band of the comet-tail
(2 Π1/2 – 2 ∑⫹) subsystem of 12C 16O⫹.

Figure 6 Intensity distribution of the six branches in the 0–2 band of the comet-tail
(2 Π3/2 – 2 ∑⫹) subsystem of 12C16O⫹.

Copyright © 2002 by Taylor & Francis

tral lines of all the 10 bands (4–0, 3–0, 2–0, 1–0, 2–1, 1–1, 0–1, 0–2, 0–3, and
0–4) are deposited in the Electronic Depository of the Supplementary Material
of the Journal of Molecular Spectroscopy [8].

D. RESULTS AND DISCUSSION

The wavenumber data of the 10 bands, the infrared data of the rovibronic spectra
[54], and the microwave data of the v ⫽ 0, 1, 2, 3, and 4 levels for the X 2 ∑⫹
state [49,53]—a total of 2513 transitions—were used simultaneously in a nonlin-
ear least-squares fit [8]. The weights were calculated from the expression w ⫽
1/σ2, where σ is the uncertainty for the spectral positions of the lines. The values
of σ are in the range 0.01–0.1 cm⫺1 for the optical data, 0.001–0.01 cm⫺1 for
the infrared data, and 0.2–0.6 ⫻ 10⫺5 cm⫺1 for the microwave data. The residuals
between the calculated and observed wavenumber data of the optical spectra for
all 10 bands are also given in the Electronic Depository of Supplementary Mate-
rial [8]. The dimensionless variance in the fit is 2.5 and the root-mean-square
deviation is ⬃0.016 cm⫺1. A set of 55 molecular parameters for the v ⫽ 0 to 4
levels for both states A 2 Π i and X 2 ∑⫹ thus obtained is provided in Table 3.
From this table the value of the Λ-doubling parameters pv and qv do not show
irregularities in their v-dependence. The B0 value of the A 2 Π i state obtained by
Haridass et al. [8] is 1.5796211(84) cm⫺1. The high precision of the B0 value up
to the sixth decimal place is due to the inclusion of the microwave data. This
value is in agreement in the fourth decimal place within the error limits to the
values 1.579629(102) and 1.579710(104) cm⫺1 [6] reported from analysis of the
individual bands 0–0 and 1–0 of the B 2 ∑⫹ –A 2 Π i system. Reanalyzed data of
the 1–0 band [6] show that the value for B0 of the A state is 1.579606(74) cm⫺1.
Using the value of 1.579496(26) cm⫺1 and the values of the other molecular
parameters reported by Bembenek and co-workers [56], the wavenumbers of the
0–v″, bands with v″ ⫽ 1 to 4 of the A–X system can be reproduced up to J ⫽
22.5 with an accuracy of 0.01 cm⫺1. Beyond this J value there is a deviation in
the (νobs⫺νcal) values from ⫺0.028 to ⫺0.195 cm⫺1 for J ⫽ 24.5 to J ⫽ 33.5.
Thus we conclude that the molecular constants given in Table 3 can be used to
reproduce the wavenumbers up to the J levels [8]. Since least-squares technique
is only an interpolation method, extrapolation of the data to higher rotational
quantum numbers will not give accurate values. From Table 3 we claim that the
values of the Λ-doubling parameters pv and qv do not show irregularities in their
v-dependence. The Bv values listed in Table 3 were fitted to the standard spectro-
scopic relation given by Eq. (7), and the resulting equilibrium molecular constants
Be , αe , γe , and δe along with their standard deviations are presented in Table 4.
The Dv values were fitted to Eq. (8) and the resulting values of De and βe for the
A state, and De , βe , and εe for the X state are also listed in Table 4. The value

Copyright © 2002 by Taylor & Francis

Table 3 Molecular Constants a (in cm ⫺1) of the A 2 Π i and X 2 ∑ ⫹ States of 12 C 16 O ⫹

Molecular
constant v⫽0 v⫽1 v⫽2 v⫽3 v⫽4
A 2Πi
Tv 20406.2166(18) 21941.4793(12) 23449.2278(14) 24930.3444(18) 26384.7229(17)
Bv 1.5796211(84) 1.5600719(74) 1.5407277(96) 1.521349(11) 1.5019686(98)
D v ⫻ 10 6 6.735(11) 6.4641(96) 6.661(12) 6.690(12) 6.600(10)
⫺A v 122.0513(18) 121.9829(20) 121.8874(23) 121.8126(30) 121.8720(29)
⫺A Dv ⫻ 104 2.179(55) 1.357(63) 1.467(70) 1.546(80) 0.804(81)
⫺q v ⫻ 10 4 2.619(75) 2.439(93) 2.61(11) 2.21(10) 3.16(12)
p v ⫻ 10 2 1.549(14) 1.293(17) 1.259(18) 1.054(21) 0.943(23)

X 2∑ ⫹
Tv 0 2183.92064(36) 4337.4571(19) 6460.7397(27) 8553.6771(28)
Bv 1.967462265(60) 1.94843818(64) 1.92935252(90) 1.910180(16) 1.890985(16)
D v ⫻ 10 6 6.3170(18) 6.3674(47) 6.4278(81) 6.399(22) 6.475(20)
γ v ⫻ 10 3 9.10556(50) 9.0495(34) 8.9729(38) 8.972(65) 8.766(63)
a
Number in parentheses is the uncertainty in the last digit and corresponds to one standard deviation.

Copyright © 2002 by Taylor & Francis

Table 4 Equilibrium Molecular Constants a, b
(in cm ⫺1)
for the A 2 Π i and X 2 ∑ ⫹ States of 12 C 16 O ⫹

Molecular
constants X 2 ∑⫹ A 2Πi

Te 19628.298(69)
G (0) 1103.23(21)
ωe 2214.15(15) 1562.79(18)
ωeXe 15.150(25) 13.926(89)
ωe ye 0.65(12)
Be 1.97695082(80) 1.589482(20)
α e ⫻ 10 2 1.89615(19) 1.9806(36)
γ e ⫻ 10 4 ⫺3.018(70) 1.66(18)
δ e ⫻ 10 4 ⫺1.92(24)
D e ⫻ 10 6 6.2856(44) 6.752(12)
β e ⫻ 10 8 6.52(81) ⫺3.01(36)
ε e ⫻ 10 9 ⫺5.5(2.2)
p e ⫻ 10 2 1.631(22)
α p ⫻ 10 3 1.99(23)
β p ⫻ 10 4 1.08(49)
a
Weighted least-squares fits were used to obtain the molecular constants
here.
b
Number in parentheses is the uncertainty in the last digit and corresponds
to one standard deviation.

of D1 ⫽ 6.4641(96) ⫻ 10⫺6 cm⫺1 of the A state is not included in determining

the parameters De and βe. The equilibrium molecular constants pe , αp , and βp are
determined by fitting the values of the Λ-doubling parameter pv given in Table
3 to the expression

p v ⫽ p e ⫺ α p (v ⫹ 1/2) ⫹ β p (v ⫹ 1/2) 2 (10)

and their values are also given in Table 4. The term values Tv for both A and X
states given in Table 3 were fitted to the expression:
T v ⫽ T e ⫹ ω e (v ⫹ 1/2) ⫺ ω e x e (v ⫹ 1/2) 2 ⫹ ω e y e (v ⫹ 1/2) 3 (11)
and the vibrational constants ωe , ωexe , and ωe ye and Te thus obtained along with
their standard deviations are also listed in Table 4. In fitting the Eqs. (7), (8),
(10), and (11), a weighted least-squares fit was used in each case. The 2 Π i state
of the comet-tail system of CO⫹ belongs neither to Hund’s case (a) nor to Hund’s
case (b), but to intermediate case (case (a) for small rotation and case (b) for
large rotation). To confirm this a Fortrat diagram for the 0–2 band of the comet-

Copyright © 2002 by Taylor & Francis

Figure 7 Fortrat diagram of the 0–2 band of the A 2 Π i –X 2 ∑⫹ system of 12 C16O⫹.

tail system of 12 C16O⫹ is constructed and shown in Figure 7. In this figure, the
branches R22ff and R11ee , Q22ef and Q11fe , and P22ff and P11ee draw closer together
with increasing J, as expected for the intermediate case [see, for example, Herz-
berg [21]. For the 2 Π i state, the value of A/B is found to be ⬇⫺77. Jakubek and
coresearchers [6] did the rotational analysis of the 0–0, 1–0, and 0–1 bands of
the Baldet-Johnson (B 2 ∑⫹ –A 2 Π i) system of 12C16O⫹ and obtained the merged
molecular constants of the B and A states. Their merged molecular constants of
the B state are given in Table 5. The controversy over the | γ v′ –γ″v | difference of
the bands of the First Negative (B 2 ∑⫹ –X 2 ∑⫹) system existing in the literature
was reexamined by Haridass and coresearchers [7]. The values of |γ′v –γ″v | ob-
tained by Haridass et al. [7] together with those of Woods [40] and Rao [55] are

Table 5 Merged Molecular Constants of the B 2 ∑ ⫹

State of 12 C 16 O ⫹

v⫽0 v⫽1

Bv 1.784564(54) 1.754520(56)
D v ⫻ 10 6 7.941(78) 8.153(78)
γ v ⫻ 10 2 2.116(11) 2.065(15)
Tv 25227.6135(24) 26906.9837(29)

Copyright © 2002 by Taylor & Francis

Table 6 Comparison of the Values of | γ′v –γ ″v | (in
cm ⫺1)

|γ ′v –γ″v |
Haridass and
v′–v″ Woods (40) Rao (55) co-researchers (7)

0–1 0.0180 0.009 0.0092

0–2 0.0159 0.008 0.0092
0–3 0.0153 0.010 0.0093
1–2 0.0172 0.009 0.0092
1–3 0.0158 0.008 0.0092
1–4 0.0150 0.010 0.0067
1–5 0.0159 0.009 0.0079
2–4 0.0149 0.009 0.0085
2–5 0.0153 0.009 0.0096
2–6 0.0150 0.008 0.0069
3–5 0.0153 0.007 0.0071

presented in Table 6. The values obtained by Haridass and colleagues [7] are
closer to those of Rao [55], except for the 1–4 band, but differ a great deal from
those of Woods [40].

VI. C18Oⴙ
13

A. Earlier Work on the Spectra of Other Isotopomers

of CO
From a study of the comet-tail system (A 2 Π–X 2 ∑⫹) of 12C16O, Rao [29] identi-
fied the state A to be an inverted 2 Π i (Sec. V). The new vibrational assignments
in [29] were confirmed from the observed isotope shifts of the band heads of the
C O [30] and those of the Baldet-Johnson system (B 2 ∑⫹ –A 2 Π i) of 12C18O⫹
13 16 ⫹

[31]. The 2–0 and 0–0 bands, respectively, of the A–X system of the 13C16O⫹
molecule have been investigated [65,66]. The rotational analysis of five bands
(4–0, 3–0, 2–0, 1–0, and 2–1) of this system of the 14C16O⫹ molecule has been
performed [67]. The rotational structure of four bands (4–0, 3–0, 2–0, and 0–
2) of this system of the 12C18O⫹ molecule has been analyzed [68]. In our laboratory
several investigations [69–72] were made on the electronic spectra of the neutral
13 18
C O molecule. Prasad and Reddy [9] have reported the first observation of nine
bands (5–0, 4–0, 3–0, 2–0, 1–0, 2–1, 1–1, 0–1, and 0–2) of the comet-tail
system of the 13C18O⫹ molecule and the results obtained from the rotational analy-
sis of eight of them (except the very weak 1–0 band).

Copyright © 2002 by Taylor & Francis

 Regarding the work done on the Baldet-Johnson system of the isotopically
substituted carbon monoxide ions, Conkić and coresearchers [73] analyzed the
1–0 and 0–0 bands of 12C18O⫹, the 1–0, 0–0, and 0–1 bands of 13C16O⫹, and the
1–0 band of 12C16O⫹. Brown and colleagues [74], who attempted to reanalyze
the wave number data of Conkić et al. [73], claimed that the data are questionable
because of possible errors in calibration and measurements. Jakubek et al. [75]
performed the rotational analysis of the 1–0, 0–0, and 0–1 bands of this system
in the 14C16O⫹ ion. Reddy and Prasad [10] observed the 1–0, 0–0, and 0–1 bands
of this system of 13C18O⫹ and analyzed this rotational structure for the first time.
Misra and coresearchers [41] performed a partial analysis of the 0–0, 0–
1, and 0–2 bands of the First Negative system of 13C16O⫹, and reported a few
rotational constants and | γ ′ ⫺ γ ′| differences for the 0–1 and 0–2 bands. Shvang-
iradze and colleagues [76] and Pešić and coresearchers [77] studied the isotope
shifts of the bandheads of this system of 13C16O⫹ relative to those of 12C16O⫹.
Regarding the work done on this system of 12C18O⫹, the isotope shifts of the
bandheads have been studied by Pešić et al. [77]; Janjić and Pešić [78] recorded
the spectrum at moderate dispersion and performed the rotational analysis of 10
bands. The spin splitting of the spectral lines was not resolved. Twelve bandheads
of the First Negative system and reported isotope shifts of these bands relative
to those of 13C16O⫹ have been reported [77]. Pešić and coresearchers [79] recorded
six bands of this system at a moderate dispersion. The spin splitting was not
observed in any of these bands, and their assignment of rotational quantum num-
bers was incorrect, particularly for the 0–1 band. Prasad and Reddy [11] observed
a total of 22 bands with v′ ⫽ 0–4 and v″ ⫽ 0–7 of the First Negative system
and analyzed the rotational structure of 13 of these bands with v′ ⫽ 0–2 and
v″ ⫽ 0–5, photographed under high resolution.

B. Experimental Details
The A–X, B–A, and B–X systems of 13C18O⫹ were excited in the cathode column
of the cathode of the hollow-cathode discharge tube (Sec. III). Carbon-13 and
oxygen-18 gas with a rated purity of 99% 13C and 95% 18O (Merck Sharpe and
Dohme) was used. A DC voltage of 1100 V applied between the electrodes main-
tained the discharge at a current of ⬃65 mA when the pressure inside the dis-
charge tube was ⬃0.8 torr. An Fe–Ne hollow-cathode lamp was used as a source
of reference spectra [59] for the three band systems. Nine bands of the comet-
tail system of 13C18O⫹ occurring in the spectral region 3165–6165 Å were first
photographed under medium dispersion on a 2 m Bausch and Lomb dual grating
spectrograph, in the first order of a 600 grooves/mm grating, blazed at 2.5 µm.
Two of these bands (0–1 and 0–2) were photographed under high resolution in
the second order of a 1200 grooves/mm grating blazed at 1.0 µm on the Bausch
and Lomb spectrograph. Of the remaining seven bands (5–0, 4–0, 3–0, 2–0, 1–

Copyright © 2002 by Taylor & Francis

0, 2–1, and 1–1), which were photographed under high resolution on a 3.4 m
Jarrell-Ash spectrograph, the 1–1 band was recorded in the second order and the
rest were in the third order of a 1200 grooves/mm grating blazed at 1.4 µm. The
exposure times for the high-resolution spectra on the Bausch and Lomb spectro-
graph were about 1 h, whereas on the Jarrell-Ash spectrograph the times varied
from 45 min (for the 3–0 band) to 8.5 h (for the 1–1 band). Kodak Spectrum
Analysis No. 1, 103 a-O, and 103-F plates were used to photograph the spectra.
The slit widths were maintained at 20 and 30 µm on the Bausch and Lomb and
the Jarrell-Ash spectrographs, respectively. The reciprocal dispersions of the
spectra are ⬃0.56 Å/mm at 3950 Å in the third order on the Jarrell-Ash spectro-
graph and about 1.42 Å/mm at 6200 Å in the second order on the Bausch and
Lomb spectrograph.
The 1–0, 0–0, and 0–1 bands of the Baldet-Johnson system of 13C18O⫹
occurring in the spectral region 3700–4225 Å were photographed on Kodak Spec-
trum Analysis No. 1 plates under high resolution on a 3.4 m Jarrell-Ash Ebert
grating spectrograph in the third order of a 1200 grooves/mm grating blazed at
µm. The exposure time was 1 h for the stronger 1–0 and 0–0 bands and about
2 h for the 0–1 band. A slit width of 30 µm gave a reciprocal dispersion of 0.58
Å/mm at 3960 Å. The accuracy of the measurements was ⫾0.003 Å.
A total of 22 bands of the B–X system of 13C18O⫹, degraded to longer
wavelengths and occurring in the spectral region 2110–2755 Å, were initially
photographed in the first order of a 1200 grooves/mm grating blazed at 1.0 µm,
on a 2 m Bausch and Lomb dual-grating spectrograph. Thirteen of these bands
were then photographed under high resolution in the fifth and sixth orders of a
1200 grooves/mm grating blazed at 1.4 µm on a 3.4 m Jarrell-Ash Ebert grating
spectrograph using Kodak SWR plates. The 0–3, 1–4, and 2–5 bands were re-
corded in the fifth order and the remaining 10 bands (1–0, 2–1, 0–0, 1–1, 0–1,
1–2, 2–3, 0–2, 1–3, and 2–4) were recorded in the sixth order. The slit width
was maintained at 20 µm for the strong bands and at 30 µm for the weak bands
to photograph the high-resolution spectra. An exposure time of 30 min was re-
quired to photograph the weak bands under medium dispersion on the Bausch
and Lomb spectrograph. The exposure times for the high-resolution spectra var-
ied from 5 min for the 0–0 band to 2 h for the 2–4 band. The measured reciprocal
dispersions are about 0.2 Å/mm at 2450 Å for high-resolution spectra on the
Jarrell-Ash spectrograph and 4.1 Å/mm for the medium-resolution spectrum on
the Bausch and Lomb spectrograph. The accuracy of the measurements was
⫾0.002 Å for the high-resolution spectra and ⫾0.02 Å for the medium-resolution
spectrum.

C. Results and Discussion

The rotational quantum numbers and the vacuum wavenumbers of the spectral
lines of eight bands (5–0, 4–0, 3–0, 2–0, 2–1, 1–1, 0–1 and 0–2) of the comet-

Copyright © 2002 by Taylor & Francis

tail system of 13C18O⫹ are listed as a supplementary publication of the Journal
of Chemical Physics [9]. The data for the spectral lines of 1–0, 0–0 and 0–1 bands
of the Baldet-Johnson system of 13C18O⫹ are listed similarly [10]. The rotational
quantum numbers and the vacuum wavenumbers of 13 bands out of total of 22
of the First Negative system, which are rotationally analyzed, are deposited in
the Editor’s Office of the Journal of Molecular Spectroscopy, [11]. The authors
[11] may also be contacted for the spectral data. In the rest of this subsection
we describe the analysis of the First Negative system of 13C18O⫹.

1. Spectrum of CO⫹ in the Region 2113 Å–2656 Å

A spectrogram of the First Negative (B 2 ∑⫹ –X 2 ∑⫹) system of 13C18O⫹ photo-
graphed under medium dispersion is shown in Figure 8. Of a total of 22 bands
observed, 18 are shown in this spectrogram. Because the commercial sample of
12 18
C O used in the present work contained some traces of 12C and 16O, a few
bandheads of 12C18O⫹ and 13C16O⫹ were also observed on the photographic plates.
In addition, a few bands of the fourth positive (A 1Π–X 1∑⫹) system of 13C18O
can also be seen in Figure 8. The vacuum wavenumbers of the bandheads of
13 18 ⫹
C O , their relative intensities, and the vibrational quantum numbers are given
by Prasad et al. [11].

Figure 8 The First Negative System (B 2 ∑⫹ –X 2 ∑⫹) of the 13C18O⫹, photographed on

a 2 m Bausch and Lomb dual grating spectrograph in the first order of a 1200 grooves/
mm grating in the region 2110–2660 Å. (a) Fourth Positive System (A 1Π–X 1∑⫹) of
C O; (b), (c), (d) First Negative System of 12C18O⫹, 13C18O⫹, and 13C 16O⫹, respectively.
13 18

Copyright © 2002 by Taylor & Francis

2. Rotational Analysis of the First Negative System
The First Negative (B 2 ∑⫹ –X 2 ∑⫹) system of 13C18O⫹ represents a transition be-
tween two states that belong to Hund’s case (b). The rotational structure of a
band arising from such a transition consists of four main branches (R11ee , R22ff ,
P11ee , and P22ff) and two satellite branches, (RQ21fe and PQ12fe) as shown schemati-
cally in Figure 9. Here the subscripts 1 and 2 refer to the F1 and F2 components,
respectively, and e and f refer to the parities of the rotational levels. Because the
two satellite branches are normally very weak, they may not even be observed
in many cases. If the spin splitting of the spectral lines is not fully resolved, one
would see only one R branch and one P branch, instead of two for each branch.
As in the B–X system of 13C18O⫹, if the bands are degraded to longer wavelengths,

Figure 9 A schematic energy level diagram showing the first few rotational transitions
in a band of a 2 ∑u⫹ – 2 ∑g⫹ system.

Copyright © 2002 by Taylor & Francis

Table 7 Equilibrium Molecular Constants a (in cm ⫺1, unless stated otherwise)
of the X 2 ∑ ⫹, A 2 Π i , and B 2 ∑ ⫹ States of 13 C 18 O ⫹

Molecular
constant X 2∑⫹ A 2Πi B 2∑⫹

Te 0.0 20731.04(1) 45877.04(1)

ωe 2110.039(4) 1488.940(4) 1650.665(4)
ωe xe 13.738(2) 12.411(2) 23.983(1)
ωe ye ⫺0.0032(2) 0.0228(2)
Be 1.79468(2) 1.44258(2) 1.63374(2)
αe 0.016440(3) 0.016791(4) 0.026003(7)
I e (g cm2) 15.5979(3) ⫻ 10 ⫺40 19.4050(3) ⫻ 10 ⫺40 17.1344(2) ⫻ 10 ⫺40
r e (Å) 1.11546(1) 1.24416(1) 1.16911(1)
a
The number in parentheses indicates the uncertainty in the last digit and corresponds to one standard
deviation.

R branch forms the head. In the present study the spin splitting of the rotational
lines was observed in several bands but the satellite branches were not observed.
The |γ′–γ ″ | difference for the bands of this system is very small; the extent of
the splitting in the spectral lines is also very small.
A method of ‘‘merging’’ procedure was used [11] to obtain various molecu-
lar parameters for the X 2 ∑⫹, A 2 Π i , and B 2 ∑⫹ states of 13C18O⫹. The reader is
referred to that work [11] for further details of the ‘‘merging’’ procedure. The

Table 8 Isotope Shifts (in cm ⫺1) in the First Negative System of 13 C 18 O ⫹

Band Observed Calculated Band Observed Calculated

1–0 ⫹65.6 ⫹65.0 3–5 a ⫺272.7 ⫺274.5
2–1 ⫹36.1 ⫹35.5 4–6 a ⫺296.2 ⫺298.6
0–0 ⫺11.0 ⫺11.6 0–3 ⫺306.8 ⫺307.2
1–1 ⫺35.6 ⫺36.3 1–4 ⫺323.1 ⫺323.6
2–2 a ⫺58.9 ⫺63.1 2–5 ⫺341.5 ⫺341.9
0–1 ⫺112.7 ⫺112.9 3–6 a ⫺360.4 ⫺361.9
1–2 ⫺134.6 ⫺134.9 4–7 a ⫺382.4 ⫺283.2
2–3 ⫺158.5 ⫺158.8 0–4 a ⫺397.8 ⫺400.1
0–2 ⫺211.0 ⫺211.4 1–5 a ⫺412.8 ⫺413.8
1–3 ⫺230.2 ⫺230.6 2–6 a ⫺428.0 ⫺429.3
2–4 ⫺251.3 ⫺251.8 3–7 a ⫺445.8 ⫺446.4
a
The observed isotope shift is obtained as the difference in the vacuum wavenumbers of the band
heads. For the remaining heads, it is the difference in the band origins.

Copyright © 2002 by Taylor & Francis

equilibrium molecular constants of the X, A, and B states of 13C18O⫹ resulting
from the analysis are listed in Table 7.

3. Isotope Shifts
The observed isotope shifts of all the 22 bands of the First Negative system of
13
C18O⫹ relative to the corresponding bands of 12C16O⫹ (i.e., ∆ν (12C16O⫹ – 13C18O⫹))
are listed in Table 8. The isotope shifts calculated from Eq. (6) are also listed
in the same table. In Eq. (6) the vibrational constants for the X and B states
were taken from Haridass et al. [7,8]. In Eq. (6) the parameter ρ ⫽ [µ(12C16O)/
µ(13C18O)]1/2 ⫽ 0.9530. The agreement between the observed and the calculated
isotope shifts is good in the case of bands for which the band origin data were
used and is satisfactory for those for which the band head data were used [11].

VII. 15
N2ⴙ
A. Earlier Work
Even though the First Negative (B 2∑u⫹ –X 2∑g⫹) system of 14N2⫹ was first observed
more than a century ago, Fassbender [80] was the first to perform a partial rota-
tional analysis of a few bands and list the band head positions for 36 bands. The
previous work done on this system has been reviewed by Lofthus [81], Tyte and
Nicholls [82], and Lofthus and Krupenie [12]. Dick and colleagues [83], Gottscho
and co-workers [84], Chevaleyre and Perott [85], and Klynning and Pages [86]
reanalyzed several bands of this system. Gottscho and coresearchers [84] per-
formed a complete deperturbation analysis of the bands of this system and pre-
sented precise constants for the X, A, and B states. They concluded that there is
no evidence of 4∑u⫹ –B 2∑u⫹ perturbations and that A 2 Π i,u is the only state causing
perturbations in the B 2∑u⫹ state. Recently, Michaud et al. [87,88] did the rota-
tional analysis of the B–X system of 14 N2 recorded under high-resolution Fourier
spectrometry and observed new perturbations in the B 2 ∑⫹ (v ⫽ 0) state up to N
⫽ 84 by A 2 Π (v ⫽ 10, 11, 12, 13).
Wood and Dieke investigated the B–X system of (14 N 15 N)⫹ [89] and 15 N2⫹
[90] and confirmed the vibrational numbering. In the latter paper, they have esti-
mated the nuclear spin of 15 N from the intensity alternations in the rotational
structure of the 1–0, 0–0, 1–2, 0–1, and 0–2 bands of 15 N2⫹ but did not obtain
any molecular constants from the analysis of the spectra.
Colbourn and Douglas [91] also reported the band origin of the 0–0 band
of this system of 15 N2⫹. In our laboratory, the B–X system of 15 N2⫹ was excited
in a hollow-cathode discharge tube of special design [22] with a very pure sample
of 15 N2 by Reddy and Prasad [92].

Copyright © 2002 by Taylor & Francis

B. Experimental Details
The design of the hollow-cathode discharge tube in which the B–X system of
15
N2⫹ was excited is described in Sec. II and was recorded by photographing the
cathode glow of the discharge. Nitrogen-15 gas rated at a purity of 99.9% of 15 N
(Merck Sharpe and Dohme Canada Limited) was used. A DC voltage of 1100
V applied between the electrodes maintained the discharge at current of ⫺65
mA. The emission spectrum was photographed under medium dispersion in the
spectral region 3750–5170 Å on a 2 m Bausch and Lomb dual-grating spectro-
graph in the first order of a 600 grooves/mm grating blazed at 2.5 µm. The high-
resolution spectra of the bands of this system were photographed on the Bausch
and Lomb instrument in the third order of a 1200 grooves/mm grating blazed at
1.0 µm and also on a 3.4 m Jarrell-Ash spectrograph in the second and third
orders of a 1200 grooves/mm grating blazed at 1.4 µm. The exposure times for
the high-resolution spectra varied from 5 s (for the strong lines of the 0–0 band)
to 5 hr (for the 0–3, 1–4, and 2–5 bands). Kodak Spectrum Analysis No. 1, 103
a-O, and 103-F plates were used to photograph the spectra. The slit width was
maintained at 20 µm on the Bausch and Lomb spectrograph but varied from 15
µm to 30 µm on the Jarrell-Ash spectrograph, depending on the intensity of the
band. The reciprocal dispersions of the spectra vary from 0.51 Å/mm at 4200 Å
in third order to 0.95 Å in the second order. An Fe-Ne hollow-cathode lamp was
used as the source for the reference spectra whose wavelengths were taken from
work by Crosswhite [59]. The accuracy of the measurements was ⫾ 0.002 Å.

C. Analysis of the Spectra

Fifteen bands of the First Negative system of 15 N2⫹ degraded to shorter wave-
lengths have been identified. In addition, some bands of the Second Positive
(C 3Πu –B 3Πg) system of 15 N2 can also be seen. In most of the excitation condi-
tions, the bands of this system are generally overlapped by those of the second
positive system. Under favorable conditions, the intensity of the Second Positive
system may be reduced but it is very difficult to suppress it completely. The weak
3–5 and 3–6 bands of the First Negative system of 15 N2⫹ could not be photo-
graphed under high resolution. The vacuum wavenumbers of the band heads of
all the remaining 13 bands measured on plates photographed under high resolu-
tion, their relative intensities, and the vibrational quantum numbers are given in
Table 9.
The rotational structure of the 1–3 band of the First Negative system of
15
N2⫹ photographed under high resolution on the Bausch and Lomb spectrograph
is shown in Figure 10. It is clearly shown in this figure that the rotational lines
with the odd N values are stronger than those with the even N values, giving

Copyright © 2002 by Taylor & Francis

Table 9 Band Heads and Band Origins of the First Negative
(B 2 ∑ u ⫹ ⫺ X 2 ∑ g ⫹) System of 15 N 2 ⫹

Band head Band origin a Relative Assignment

(cm ⫺1) (cm ⫺1) intensity b v′–v″

27968.4 27997.49c m 2–1

27827.6 27855.985(6) m 1–0
25729.0 25753.656(6) m 1–1
25538.4 25562.783(6) vs 0–0
23862.4 23884.296(6) w 2–3
23659.9 23681.709(7) m 1–2
23438.7 23460.493(5) s 0–1
21854.2 21873.709(6) w 2–4
21620.9 21640.266(6) m 1–3
21369.1 21388.491(6) s 0–2
19876.8 19894.179(14) vw 2–5
19612.5 19629.869(7) w 1–4
19329.8 19347.246(6) m 0–3
a
The number in the parentheses indicates the uncertainty in the last digit
and corresponds to one standard deviation.
b
Abbreviations for relative intensities vs, s, m, w, and vw represent very
strong, strong, medium, weak, and very weak, respectively.
c
Calculated from the term values of v′ ⫽ 2 and v″ ⫽ 1 levels.

Figure 10 Rotational structure of the 1–3 band of the First Negative System (B 2∑u⫹ –
X 2∑g⫹) of 15 N2⫹ photographed on the Bausch and Lomb spectrograph in the third order
of a 1200 grooves/mm grating.

Copyright © 2002 by Taylor & Francis

alternation of intensity in the band. Similar alternation of intensity is observed
for all the bands of this system. The electron spin splitting of the rotational lines
of the 1–3 band can also be clearly seen for the spectral lines with N ⱖ 5. Similar
splitting of the rotational lines is observed in the other bands with v′ ⫽ 1 for
N ⱖ 5 and in those with v′ ⫽ 0 for N ⱖ 19. For the bands with v′ ⫽ 2, for
which the rotational lines are identified only up to N ⫽ 19, the spin splitting is
not observed. The satellite branches PQ12 and RQ21 are not observed in any of the
bands of this system. The rotational quantum numbers and the vacuum wavenum-
bers of the spectral lines of all the twelve bands analyzed are listed by Albritton
and colleagues [93]. The vacuum wavenumbers of the spectral lines of all the
branches of a band were simultaneously fitted to the expressions of the R1 , R2 ,
P1 , and P2 branches (see Fig. 9), which are given by

R 1 (N) ⫽ υ 0 ⫹ F 1′ (N ⫹ 1) ⫺ F ″1 (N) (12)

R 2 (N) ⫽ υ 0 ⫹ F 2′ (N ⫹ 1) ⫺ F ″2 (N) (13)

P 1 (N) ⫽ υ 0 ⫹ F 1′ (N ⫹ 1) ⫺ F ″1 (N) (14)

P 2 (N) ⫽ υ 0 ⫹ F 2′ (N ⫹ 1) ⫺ F ″2 (N) (15)

where υ0 is the band origin

F 1 (N) ⫽ B v N(N ⫹ 1) ⫺ D v N 2 (N ⫹ 1) 2 ⫹ 1/2 γ v N (16)

and

F 2 (N) ⫽ B v N(N ⫹ 1) ⫺ D v N 2 (N ⫹ 1) 2 ⫹ 1/2 γ v (N ⫹ 1) (17)

and the molecular constants were estimated by the method of least-squares. The
(νobs⫺νcalc) values, obtained from the least-squares fits of the vacuum wavenum-
bers of individual bands, are given in parentheses in Table 2 of Ref. [7]. However,
these are not given for the perturbed lines and also for a very few unperturbed
lines that are excluded from the analysis. In general, the standard deviation of
such a least-squares fit is ⫺0.03 cm⫺1.
Using the method of merging proposed by Albritton and colleagues [93]
and Coxon [94] in which the molecular constants obtained from the analysis
of the individual bands by least-squares fits are combined together, giving due
consideration to their uncertainties and the correlations existing among them, a
unique set of molecular constants for the B and X states of the First Negative
system of 15 N2⫹ was obtained. The band origins thus obtained from the merged
least-squares fit are included in Table 9. The uncertainty in the origin of the 2–
5 band appears to be high because the number of spectral lines used in the analysis
is limited. The Bv , Dv , and γv values of states X and B, also obtained from the

Copyright © 2002 by Taylor & Francis

Table 10 Rotational Constants a (in cm ⫺1) of the X 2 ∑ g ⫹ and B 2 ∑ u ⫹ States of 15 N 2 ⫹

X 2∑g⫹ B 2∑u⫹
Vibrational
level Bv D v ⫻ 10 6 ⫺γ v ⫻ 10 3 Bv D v ⫻ 10 6 γ v ⫻ 10 3
0 1.79520(8) 5.5(1) 18.3(4) 1.93731(9) 6.6(2) 1.1(9)
1 1.77817(8) 5.6(1) 18.8(4) 1.91709(7) 6.4(1) 1.2(4)
2 1.76114(8) 5.8(1) 17.2(5) 1.89548(12) 16.0(3)
3 1.74342(8) 5.5(1) 19.1(5)
4 1.72639(10) 6.2(2) 19.1(6)
5 1.70783(23) 4.0(6)
a
The number in the parentheses indicates the uncertainty in the last digit and corresponds to one
standard deviation.

same merged least-squares fit, are listed in Table 10. The γ0 value of the B state
was estimated exclusively from the wavenumber data of N ⫽ 33–43 in the 0–
0 band and was not included in the merged fit. The value of γ2 of state B could
not be estimated because the spin splitting was not observed in any of the bands
with v′ ⫽ 2. Similarly, γ5 of state X could not be obtained because the v″ ⫽ 5
level was observed only in the 2–5 band. The Be and αe values of states X and
B were obtained from the expression
B v ⫽ B e ⫺ α e (v ⫹ 1/2) (7-a)
using the Bv values listed in Table 10. The Be and αe values thus obtained are
listed in Table 11. The equilibrium internuclear distance re and the corresponding

Table 11 Equilibrium Molecular Constants a (in cm ⫺1, or

otherwise specified) of the X 2 ∑ g ⫹ and B 2 ∑ g ⫹ states of 15 N ⫹ 2

Molecular
constant X 2∑g⫹ B 2∑g⫹

Te 0.0 25460.258(9)
ωe 2132.584(9) 2342.811(9)
ωe xe 15.063(4) 24.656(3)
ωe ye ⫺0.0324(4)
Be 1.80392(8) 1.94775(10)
αe 0.01720(1) 0.02046(4)
r e (Å) 1.1162 1.0742
I g (g cm2) 1.5518 ⫻ 10 ⫺39 1.4372 ⫻ 10 ⫺39
a
The number in the parentheses indicates the uncertainty in the last digit
and corresponds to one standard deviation.

Copyright © 2002 by Taylor & Francis

moment of inertia Ie of states X and B, obtained from their respective Be values,
are also presented in the same table.

D. Vibrational Analysis and Isotope Shifts

All the origins of the bands of the B–X system of 15 N2⫹, except that of the 2–
1 band (see Table 9), were fitted to the expression
υ 0 ⫽ υ e ⫹ ω′e (v′ ⫹ 1/2) ⫺ ω′e x′e (v′ ⫹ 1/2) 2 ⫹ ω′e y′e (v′ ⫹ 1/2) 3
(18)
⫺ω″e (v″ ⫹ 1/2) ⫹ ω″e x″e (v″ ⫹ 1/2) ⫺ ω′e y″e (v″ ⫹ 1/2) 3
The system origin νe (⫽Te ) and the vibrational constants were obtained simulta-
neously from this fit and are listed in Table 11. Here the term value Te of state B
is the same as νe of the B–X system. Because only three vibrational levels (v ⫽ 0,
1, and 2) are observed for the B 2 ∑u⫹ state, ωe ye could not be estimated for it.
The isotope shifts of the bands of the First Negative system of 15 N2⫹ are
obtained as the differences of the origins of the 14 N2⫹ bands of this system and
those of the corresponding bands of 15 N2⫹ [i.e., ν0(14 N2⫹) ⫺ ν0(15 N2⫹)]. The values
of ν0(14 N2⫹) for the bands of this system were either directly taken from Jakubek
et al. [6] or calculated from the vibrational constants reported by them. The values
of ν0(15 N2⫹) are the experimental values obtained by Albritton and colleagues
[93]. The observed isotope shifts thus obtained are listed in Table 12. The isotope

Table 12 Isotope Shifts in the First Negative

System of 15 N 2 ⫹

Isotope shift
∆ν(14 N 2⫹ – 15 N 2 ⫹)

Band Observed Calculated

2–1 84.37 84.33
1–0 81.70 81.99
1–1 9.26 9.53
0–0 3.24 3.48
2–3 ⫺54.16 ⫺54.05
1–2 ⫺61.01 ⫺60.76
0–1 ⫺69.21 ⫺68.98
2–4 ⫺120.06 ⫺119.93
1–3 ⫺129.03 ⫺128.85
0–2 ⫺139.44 ⫺139.27
2–5 ⫺183.79 ⫺183.56
1–4 ⫺195.14 ⫺194.72
0–3 ⫺207.68 ⫺207.36

Copyright © 2002 by Taylor & Francis

shifts were also calculated from Eq. (6) using the vibrational constants of the X
and B states of 14 N2⫹ taken from Chevaleyre et al. [85] and the value of ρ ⫽
[µ(14 N2⫹ /µ(15 N2⫹)]1/2 ⫽ 0.9662 and are listed in the same table. The agreement
between the observed and calculated isotope shifts is very good.

E. Perturbations in the B 2 ∑uⴙ State of 15N2ⴙ

Among the 12 bands of the first negative system of 15 N2⫹, for which the rotational
structure is analyzed, some irregularities are observed only in the 0–0, 0–1, 0–
2, and 0–3 bands. This indicates that the v ⫽ 0 level of the B 2 ∑u⫹ state of 15 N2⫹
is perturbed. In the 0–1 and 0–3 bands, the rotational lines are identified only
up to N ⫽ 21 and it is difficult to make any conclusions from their structure
regarding the perturbations. In the 0–0 and 0–2 bands the spectral lines are identi-
fied up to N ⫽ 43 and 26, respectively, and the perturbations in the v ⫽ 0 level
are confirmed with the help of similar irregularities in the structure of these two
bands. From the vacuum wavenumbers of the rotational lines in the 0–0 band it
has been noticed that the spectral lines belonging to R1(N) and R2(N) branches
are completely identified, whereas the lines of P1(N) and P2(N) branches with
N ⫽ 22–32 are not. Hence, only the lines of the R1(N) and R2(N) branches were
used in the analysis of the perturbation. Using the appropriate molecular con-
stants, the wavenumbers of the R1(N) and R2(N) lines of the 0–0 band were calcu-
lated. The deviations (υobs⫺υcalc) for these lines are plotted against N′ in Figure
11. It can be seen in this figure that the maximum perturbations are observed at
N′ ⫽ 24 and 29 of the F1 levels and at N′ ⫽ 21 and 26 of the F2 levels. Two
extra lines are observed at the maximum perturbations for R1 (23) (N′ ⫽ 24) and
R2(20) (N′ ⫽ 21) corresponding to F1 and F2 levels respectively. The plots here
indicate that this is a heterogeneous perturbation for which ∆Λ ≠ 0. In such a
situation, a ∑u⫹ state can be perturbed only by a Πu state. Since the maximum
perturbations are occurring at four different N values and the perturbed state is
the 2 ∑u⫹ state, the perturbing Πu state is obviously a 2 Πu state, which is identified
as A 2 Π i,u on the basis of its proximity to B 2 ∑u⫹.
The molecular constants provided by Jakubek et al. [6] for A 2 Π i,u of 14 N2⫹
and the value of ρ(⫽ 0.9662) were used to obtain the corresponding constants
of 15 N2⫹. Using these constants of state A of 15 N2⫹, the (Te ⫹ G(v)) and Bv values
of the vibrational levels of state A were calculated. Because the Bv values of
state A are smaller than the B0 value of state B, only the vibrational levels of
state A, which are slightly above the v ⫽ 0 level of state B, can perturb the latter.
The v ⫽ 0 level of state B is at 26625.33 cm⫺1 with respect to the minimum of
the potential energy curve of state X. Only the levels with v ⱖ 10 lie above the
v ⫽ 0 level of state B. Hence, any vibrational level with v ⱖ 10 of state A might
be perturbing the B 2 ∑u⫹ state. The actual perturbing vibrational level is found
by the method of trial and error. According to Kronig [95], at the points of maxi-

Copyright © 2002 by Taylor & Francis

Figure 11 Plot of deviations (νobs⫺νcalc) of the rotational lines of the R1 and R2 branches
versus the rotational quantum number N′ of the upper state in the 0–0 band. The lines
joining the dots represent the R1 branch, and those joining the crosses represent the R2
branch.

mum perturbation, both the perturbed and the perturbing levels have equal energy
and the same J value. Using this principle, the Bv value of the perturbing level was
calculated from the corresponding F(J) value, which is the difference between the
total energy of the perturbed rotational level (Te ⫹ G(v) ⫹ F(J) of state B) and
the sum of the electronic and vibrational terms of the perturbing level (Te ⫹ G(v)
of state A). The approximate value of Bv for state A, v ⫽ 10, thus obtained is
1.333 cm⫺1 and the corresponding value calculated from the known values of Be
and α e is 1.445 cm⫺1. Levels A, v ⱖ 11, are too high to perturb the observed
rotational levels of the B, v ⫽ 0, state. The calculated position of the A, v ⫽ 11,

Copyright © 2002 by Taylor & Francis

is 28466.1 cm⫺1 above the X 2 ∑g⫹ state. It is thus concluded that the level, v ⫽
10 is perturbing the B, v ⫽ 0, level. No perturbations are observed in the v ⫽
1 and 2 levels of state B within the observed rotational structure (up to N ⫽ 34
and 18, respectively).

VIII. NO
A. Introduction
Strutt (96) was the first to observe the γ (A 2 ∑⫹ –X 2 Πr) system of NO. Over the
years numerous investigations have been carried out on different types of spectra
of NO: electronic spectra, vibration–rotation spectra, pure rotation spectra, rota-
tional Raman spectra, magnetic rotation spectra, among others. For the previous
work on the spectra of NO, the reader is referred to work by Miescher and Huber
[97] and Huber and Herzberg [5]. Cisak and coresearchers [98] measured a part
of the γ system of 14 N 16O and 15 N 16O recorded under medium resolution of a
quartz prism spectrograph and obtained the vibrational constants and isotopic
displacements. Engleman and colleagues [99] studied the 1–0 band of the γ sys-
tem of 15 N 16O and 14 N 18O (in addition to some bands of 14 N 16O). Engleman and
Rouse [100] studied 17 bands of the γ system of 14 N 16O in the spectral region
2700–3100 Å under high resolution and determined the molecular constants of
v ⫽ 0 to 16 of the X 2 Π state and v ⫽ 0 to 5 of the A 2 ∑⫹ state. Freedman and
Nicholls [101] reanalyzed the high-resolution data of the 1–0 and 0–0 bands of
the γ system of 14 N 16O and the 1–2 band of 14 N 18O and 15 N 16O, obtained by
Engleman and co-workers [99], using the perturbation technique described by
Zare and coresearchers [102].
Griggs et al. [103] recorded the infrared 1–0 and 2–0 bands of 15 N 18O
in the X 2 Π state under high resolution and obtained rotational and vibrational
constants. Amiot and colleagues [104] recorded the high-resolution Fourier spec-
tra of the infrared 1–0 band of 14 N 16O, 14 N 17O, 14 N 18O, and 15 N 16O and the 2–1
band of 14 N 16O, all in the X 2 Π state and obtained accurate molecular constants
for v ⫽ 0, 1, and 2 of 14 N 16O and v ⫽ 0 and 1 of the other three isotopomers.
Amiot and Guelachvili [105] investigated the infrared 1–0 and 2–1 bands of
14 16
N O and the infrared 1–0 band of 15 N 17O and 15 N 18O with a Fourier spectrome-
ter. Molecular constants were determined by the use of the ‘‘direct approach’’
method. Valentin and coresearchers [106] and Henry et al. [107] reported the
analysis of the infrared 1–0, 2–0, and 3–0 bands of 14 N 16O in the X2 Π state.
Teffo and colleagues [108] recorded the fundamental band (1–0), the high-energy
satellite X 2 Π3/2 –X 2 Π1/2 (1–0) subband, and the overtone bands 2–0 and 3–0,
of 15 N 16O and 15 N 18O, and obtained the vibrational constants and the vibrational
dependence of the rotational constants. Precise Λ doubling and hyperfine struc-

Copyright © 2002 by Taylor & Francis

ture constants of 14 N 16O were obtained by Meerts and Dymanus [109] using the
molecular beam technique. Patel and Kerl [110] used optoacoustic spectroscopy
for the study of 14 N 16O and obtained accurate values of the Λ doubling for the
X 2 Π1/2 v ⫽ 0 and 1 levels. Dale and coresearchers [111] studied the high-resolu-
tion laser magnetic resonance and infrared–radiofrequency double-resonance
spectra of the 1–0 and 2–1 bands of the ground X 2 Π state of 14 N 16O, 15 N 16O,
14 17
N O, and 14 N 18O.
The γ (A 2 ∑⫹ –X 2 Πr) system of the NO isotopomers was recently investi-
gated by Wang and colleagues in our laboratory [112]. In the present section we
report the band head measurements of the bands 1–0 and 0–v″ with v″ ⫽ 0–4
of the γ system of the nitric oxide isotopomers 14 N 16O, 15 N 16O, 14 N 18O, and 15 N 18O
in the spectral region 2140–2730 Å, excited in the anode glow of a two-column
hollow-cathode discharge tube and photographed under medium resolution. The
rotational structure of the 0–1, 0–2, and 0–3 bands of 15 N 18O was also recorded
under high resolution. The experimental data of these three bands were combined
with those of the infrared 1–0, 2–0, and 3–0 (9, 11, 14) absorption bands of the
X 2 Πr state of 15 N 18O in a global fit, using an effective Hamiltonian for the 2 Π
state given by Brown and colleagues [61] and the matrix elements for the 2 Π
and 2 ∑⫹ states given by Amiot et al. [113], and molecular constants were derived
for the A 2 ∑⫹ and X 2 Π states.

B. Experimental Details
The γ system of the nitric oxide isotopomers was excited in the anode column
of a hollow-cathode discharge tube [20]. Gases nitrogen-15 and oxygen-18 with
specified purities of 98% and 99.9%, respectively, were used (Matheson Gas
Products). A DC voltage of 1100 V applied between the electrodes of the dis-
charge tube maintained the discharge at a current of ⬃75 mA when the pressure
inside the discharge tube was ⬃ 0.8 torr. Under optimum experimental conditions
the discharge was steady and bright. The spectrum of 14 N 16O was obtained with
natural presence of nitrogen-14, oxygen-16 of low-pressure air in the discharge
tube. When nitrogen-15 alone was admitted into the discharge tube, 15 N 16O was
excited and when oxygen-18 alone was admitted, 14 N 18O was excited. When both
nitrogen-15 and oxygen-18 were admitted into the discharge tube at an approxi-
mate ratio of 1:2, 15 N 18O was excited.
In our laboratory the γ system of the NO isotopomers was photographed
in the spectral region 2140–2730 Å under medium dispersion on a 2.0 m Bausch
and Lomb dual-grating spectrograph, both in the first order of a 600 grooves/
mm grating blazed at 2.5 µm (with a reciprocal dispersion of 4.1 Å/mm at 2157Å)
and in the first order of a 1200 grooves/mm grating blazed at 1.0 µm (with a
dispersion of 2.0 Å/mm at 2367 Å). The spectra of the γ system of 15 N 18O in the

Copyright © 2002 by Taylor & Francis

same spectral region were also photographed under high resolution on a 3.4 m
Jarrell-Ash spectrograph equipped with a 1200 grooves/mm grating blazed at 1.4
µm in the fifth order (with a dispersion of 0.34 Å at 2455 Å). The slit width was
set at 20 µm for the Bausch and Lomb spectrograph and at 25 µm for the Jarrell-
Ash instrument. Iron arc and copper arc, both of DC type, were used as the
sources of the standard spectra [114,115] recorded on the former spectrograph
and an iron–neon hollow-cathode lamp was used as the standard source [59] for
the high-resolution spectra recorded on the latter spectrograph in the fourth order
of the grating. Kodak SWR photographic plates were used to record the spectra.
Overlapping orders of the spectra were eliminated with a Hoya U-340 optical
filter and a chlorine gas filter. The exposure times varied from 3 s to 4.5 hr
depending on the spectrograph, the intensity of the band, and the type of filter
used. The accuracies of measurement were approximately ⫾0.012 Å for the
medium-resolution spectra and ⫾0.003 Å for the high-resolution spectra.

C. Analysis of the Spectra

1. Vibrational Structure
The γ (A 2 ∑⫹ –X 2 Πr) system of the nitric oxide isotopomers 14 N 16O, 15 N 16O,
14
N 18O, and 15 N 18O were photographed under medium resolution as stated above.
The resulting spectra of 14 N 16O and 15 N 18O are shown in Figure 12. In this figure
the isotopic shifts of the bands can be seen very clearly. As the bands are degraded
to shorter wavelengths, P12ee and P22ff /Q12ef branches of the 2 ∑⫹ – 2 Π3/2 subsystem
and P11ee and Q11ef /P21ff branches of the 2 ∑⫹ – 2 Π1/2 subsystem form four distinct
heads of a band in the order of the increasing wavenumber. The vacuum wave-
numbers for all the 24 band heads of the six bands of each of the four isotopomers
of NO, and their vibrational assignments and relative intensities are listed in Table
13. In this table, the data of the 0–1, 0–2, and 0–3 bands of 15 N 18O were obtained
from the spectra photographed under high resolution and the rest of the data is
from the medium-resolution spectra. The γ system of 15 N 18O and several bands
of 14 N 18O of this system were observed for the first time in our laboratory [112].

2. Rotational Analysis of the 0–1, 0–2, and 0–3 Bands

of 15 N 18O
The rotational structure of a band degraded to shorter wavelengths of the A 2 ∑⫹ –
X 2 Πr system gives rise to 12 branches designated as 2 ∑⫹ – 2 Π3/2: P12ee , P22ff , Q12fe ,
Q22fe , R12ee , R22ff , and 2 ∑⫹ – 2 Π1/2: P11ee , P21ff , Q11ef , Q21fe , R11ee , and R21ff and is shown
schematically in Figure 13. In this figure the parity of the rotational levels in the
e/f notation described by Brown et al. [60] is used and the assignments of ⫹

Copyright © 2002 by Taylor & Francis

Figure 12 The gamma system of 14 N 16O and 15 N 18O excited in the anode column of a
hollow-cathode discharge tube and photographed in the spectral region 2100–2730 Å un-
der medium resolution.

and ⫺ signs proposed by Geuzebroek and coresearchers [116] on the basis of

experimental evidence are adopted. In the rotational analysis carried out in the
present work we used an effective Hamiltonian for the 2 Π state, which includes
the terms representing the vibration and rotation of the nuclei, centrifugal distor-
tion, fine structure, and the hyperfine structure discussed in detail by Brown and
coresearchers [61], which is given by

H ⫽ H 0 ⫹ H rot ⫹ H cd ⫹ H fs ⫹ H hfs (19)

See Amiot et al. [113] for details of the terms in Eq. (19). The corresponding
matrix elements were also calculated by these authors. The matrix elements of
the Hamiltonian for the upper 2 Π state and the lower 2 ∑⫹ state of the γ system
of NO are listed in Ref. 113. Douay and co-workers [117] made some corrections
for the matrix elements γ, γD , and γH given by Gatterer and Junkes [114]. The
matrix elements of these Hamiltonians relevant to the present work are:

Copyright © 2002 by Taylor & Francis

Table 13 Band Heads a,b (in cm ⫺1) of the A 2 ∑ ⫹ –X 2 Π r System of 14 N 16O, 15 N 16O,
14 18
N O, and 15 N 18O

Assignment
14
N 16 O 15
N 16 O 14
N 18 O 15
N 18 O v′–v″

46394.1 vw 2 46347.6 w 46329.3 vw 46280.6 w 1–0

46420.3 46370.9 46354.9 46305.5
46520.3 46470.7 46453.9 46404.3
46541.0 46488.4 46470.6 46426.6
44051.7 w 44047.5 m 44049.6 m 44044.8 m 0–0
44076.4 44071.2 44072.8 44068.4
44174.2 44172.1 44171.1 44171.2
44194.4 44193.0 44187.9 44189.5
42177.8 s 42207.8 s 42221.4 s 42252.31 s 0–1
42201.3 42230.0 42243.8 42273.92
42300.8 42330.4 42344.6 42375.99
42320.9 42350.9 42360.4 42394.27
40332.5 s 40393.5 s 40422.3 s 40486.36 s 0–2
40354.3 40414.8 40442.9 40506.52
40454.9 40517.7 40544.3 40609.25
40473.1 40534.6 40560.8 40626.65
38513.6 vs 38606.2 vs 38649.5 vs 38746.01 vs 0–3
38534.1 38625.9 38669.3 38764.86
38636.2 38726.9 38770.9 38868.29
38653.1 38745.2 38787.2 38884.54
36723.8 s 36846.0 s 36903.3 s 37030.3 s 0–4
36742.8 36864.7 36920.4 37048.0
36845.4 36964.9 37024.9 37152.0
36864.1 36984.2 37039.3 37167.2
a
The four heads identified for each band are formed by the P 12ee , P 22ff /Q 12ef , P 11ee and P 21ff /Q 11ef branches
in order of increasing wavenumber.
b
The 1–0, 0–0, and 0–4 band heads are measured under the medium dispersion and the 0–1, 0–2,
and 0–3 band heads of 15 N 18O are measured under the high dispersion.
c
The abbreviations for the relative intensities vs, s, m, w and vw represent very strong, strong, me-
dium, weak and very weak, respectively.

2
∑ e,f : T v ⫹ B v (x ⫹ 1) (x ⫹ 1 ⫿ 1) ⫺ D v (x ⫹ 1) 2 (x ⫹ 1 ⫿ 1) 2
2
Π 3/2e,f : T v ⫹ 0.5A v ⫺ 0.5A Dv [(x ⫹ 1]) 2 ⫺ 1] ⫹ B v [(x ⫹ 1) 2 ⫹ 1]
⫺ D v (x ⫹ 1) 2 [(x ⫹ 1) 2 ⫺ 1]
2
Π 1/2e,f : T v ⫺ 0.5A v ⫺ 0.5A Dv [(x ⫹ 1) 2 ⫹ 1] ⫹ B v [(x ⫹ 1) 2 ⫹ 1] (20)
⫺ D v (x ⫹ 1) [(x ⫹ 1) ⫹ 3] ⫿ 0.5 p v (x ⫹ 1) ⫿ q v (x ⫹ 1)
2 2

{2 Π 3/2 , 2 Π 1/2} e,f : ⫺B v [(x ⫹ 1) 2 ⫺ 1] 1/2 ⫹ 2.0D v (x ⫹ 1) 2 [(x ⫹ 1) 2 ⫺ 1] 1/2

⫾ 0.5q v (x ⫹ 1) [(x ⫹ 1) 2 ⫺ 1] 1/2

Copyright © 2002 by Taylor & Francis

Figure 13 Schematic energy level diagram shows the first few rotational transitions of
all 12 branches of a band of a 2 ∑⫹ – 2 Πr transition. Dashed lines are satellite branches.

Copyright © 2002 by Taylor & Francis

where Bv and Dv are the rotational constants, Av and Adv are the spin-orbit con-
stants, p and q are the Λ-doubling constants of the Π state, and x ⫽ (J ⫹ 0.5) ⫺
1. The upper and lower state signs (⫾ or ⫿) in the preceding matrix elements
refer to the e/f levels, respectively.
Of the three bands of the γ system of 15 N 18O photographed under high
resolution on the 3.4 Jarrell-Ash spectrograph, the rotational structure of the 0–
2 band is shown in Figure 14. In this figure the four heads and the eight branches
expected in a 2 ∑⫹ – 2 Πr transition are clearly identified. The pairs of branches P22ff
and Q12ef , and R12ee and Q22fe of the 2 ∑⫹ – 2 Π3/2 subsystem and Q11ef and P21ff , and
Q21fe and R11ee of the 2 ∑⫹ – 2 Π1/2 subsystem could not be resolved into two separate
branches in the present work. The rotational quantum numbers and the vacuum
wavenumbers of the spectral lines of the 0–1, 0–2 and 0–3 bands are listed by
Brown and colleagues [61].
The wavenumber data of the lines of the 0–1, 0–2, and 0–3 bands of the
γ system of 15 N 18O [112] can in principle be analyzed by one of the following
methods. The first involves fixing the molecular constants of the X 2 Π state,

Figure 14 Rotational structure of the 0–2 band of the γ (A 2 ∑⫹ –X 2 Π) system of 15 N 18O,

photographed on the 3.4 m Jarrell-Ash spectrograph in the fifth order of a 1200 grooves/
mm grating.

Copyright © 2002 by Taylor & Francis

which have been determined exclusively from the infrared vibration-rotation
measurements [11,14], and then obtaining the molecular constants of the A 2 ∑⫹
state from our data of the γ system. The second entails analyzing the wavenumber
data of the infrared vibration–rotation bands together with the γ system in a global
fit. In our work [112] we have adopted the second method because the molecular
constants of the ground 2 Π state of the γ system of 15 N 18O [103,105,109] were
determined by different methods stated briefly here.
Griggs and coresearchers [103] used separate expressions for the T(v,J) of
the infrared 1–0 and 2–0 bands in the individual 2 Π1/2 and 2 Π3/2 components and
did not take into account the interaction between them. Amiot and Guelachvili
[105] adopted a ‘‘direct approach method’’ in the analysis of the 1–0 infrared
band. In this method, the matrix elements of the Hamiltonian were derived con-
sidering that the main contributor to the Λ doubling is a 2 ∑⫺ state. Teffo and
colleagues [108] analyzed the 2–0, 3–0 infrared bands and the 2 Π3/2 – 2 Π1/2 (1–
0) subband using the matrix elements given by Zare et al. [102].
The wavenumber data of the 0–1, 0–2, and 0–3 bands of the γ system of
15 18
N O and the infrared data of the vibration-rotation spectra of 15 N 18O arising
from the ground 2 Π state [103,105,108] and estimated rotational constants of the
A and X states are included. A simultaneous, weighted, nonlinear least-squares
fit of all the lines of these data was carried out. The weights are calculated from
the expression w ⫽ 1/σ2, where σ is the uncertainty for the spectral positions of
the lines. The values of σ are 0.008 cm⫺1 for the data of Griggs et al. [103],
0.0002 cm⫺1 for those of Valentin and colleagues [106], 0.0005 cm⫺1 for those
of Meerts and Dymanus [109], and in the range 0.01–0.1 cm⫺1 for our data.
Residuals between the calculated and observed frequencies of the optical data
for all three bands are given by Meerts and Dymanus [109] (Table 2). For a total
of 1308 wavenumbers used in the fit, the variance was 1.46 and the root mean
square deviation was only ⬃0.065 cm⫺1. The molecular constants of the v ⫽ 0
level of the A state and those of the v ⫽ 0, 1, 2, and 3 of the X state thus obtained
are listed in Table 14. From this table it is clear that the molecular parameters
for the 2 Πr state show a smooth variation with (v ⫹ 1/2). It should be mentioned
that the Λ-doubling parameter q for v ⫽ 2 and 3 is fixed in the present analysis.
The signs of the Λ-doubling parameters convey the information about the charac-
ter of the perturber. Vivie and Peyerimhoff [118] concluded, based on theoretical
considerations, that the 2 ∑⫹ and the G 2 ∑⫺ state are dominant perturbers and
determine the magnitude of Λ splitting of the X 2 Π state. The sign of the parame-
ter p depends not only on the symmetry and location of the perturber, but also
on the sign of the effective spin-orbit coupling parameter between the ∑ and the Π
states. The sign of q is determined by the symmetry and location of the dominant
perturber. Thus, if the value of the Λ-doubling parameter q is ⫹ve, the dominant
perturber is either ∑⫹ above the Π state or ∑⫺ below the Π state, whereas if the
value of q is ⫺ve the dominant perturber is ∑⫺ above the Π state or ∑⫹ below

Copyright © 2002 by Taylor & Francis

Table 14 Rotational Constants a (in cm ⫺1) of the X 2 Π r and A 2 ∑ ⫺ States of 15 N 18 O

Molecular
constant v⫽0 v⫽1 v⫽2 v⫽3

A ∑ 2 ⫹

T 44130.244(2)
B 1.812980(7)
D ⫻ 10 6 4.694(6)
X 2Πr
T (2 Π 1/2) 0 1793.30725(4) 3560.977(1) 5303.050(2)
B 1.548215(2) 1.532899(2) 1.517569(5) 1.50218(1)
D ⫻ 10 6 4.551(5) 4.568(4) 4.595(6) 4.62(1)
A 123.1429(1) 122.9109(1) 122.665(2) 122.41(3)
A D ⫻ 10 4 1.518(7) 1.442(7) 1.37(4) 1.36(9)
p ⫻ 10 2 1.067(1) 1.066(1) 1.08(1) 1.09(2)
q ⫻ 10 5 7.6(2) 7.6(1) 1.0 b 1.0 b
a
Number in parentheses is the uncertainty in the last digit and corresponds to one standard deviation.
b
Fixed in the least-squares fit.

the Π state. The value of both p and q of the X 2 Π state are ⫹ve in [112] (see
also 119, 120). The Bv values of the 2 Πr state thus obtained were fitted to the
relation.
B v ⫽ B e ⫺ α e (v ⫹ 1/2) ⫹ γ e (v ⫹ 1/2) 2 (7-b)
and the equilibrium rotational constants Be , αe , and γe thus obtained are given in
Table 15. The Dv values were fitted to the relation
D v ⫽ D e ⫹ β e (v ⫹ 1/2) (8-a)
and the resulting values of De and βe are listed in Table 15. The equilibrium
molecular constants Ae , αA , and βA are determined by fitting the values of the
spin-orbit constant Av given in Table 14 to the expression
A v ⫽ A e ⫺ α A (v ⫹ 1/2) ⫹ β A (v ⫹ 1/2) 2 (21)
and their values are given in Table 15. From the T values given in Table 14, we
have determined the vibrational constants ωe , ωe xe , and G(0) of 2 Πr; these are
listed in Table 15. The 2 Π state neither strictly belongs to Hund’s case (a) (large
| A |/B) nor to Hund’s case (b) (small | A|/B) in most of the actual cases, but be-
longs to mixed case [case (a) for the small rotation and case (b) for large rotation].
A Fortrat diagram for the 0–2 band of the γ system of 15N 18O is shown in Figure
15. In this figure, with increasing J, the branches P11ee and P22ff , Q11ef and Q22fe ,
and R11ee and R22ff , draw closer together. This indicates the transition of X 2 Π

Copyright © 2002 by Taylor & Francis

Table 15 Equilibrium
molecular constants a (in cm ⫺1)
of the X 2 Π r State of 15 N 18 O
Constant Value

G(0) 906.25(2)
ωe 1818.92(2)
ωe xe 12.809(5)
Be 1.55585(2)
α e ⫻ 10 2 1.527(2)
γ e ⫻ 10 5 ⫺1.8(5)
D e ⫻ 10 6 4.537(3)
β e ⫻ 10 8 2.3(2)
Ae 123.258(3)
αA 0.226(4)
β A ⫻ 10 3 ⫺4.5(9)
a
Number in parentheses is the uncer-
tainty in the last digit and corresponds
to one standard deviation.

state from Hund’s case (a) at lower J values to case (b) at higher J values [see,
for example, (21)]. The value of A/B ⬇ 79 for the ground 2 Π state.

D. Vibrational Analysis
The vibrational quanta ∆G(v ⫹ 1/2) between the levels v and v ⫹ 1 of a compo-
nent of an electronic state are represented by [32]:

∆G (v ⫹ 1/2) ⫽ G(v ⫹ 1) ⫺ G(v) (22)

⫽ (ω e ⫺ ω e x e) ⫺ 2ω e x e (v ⫹ 1/2)

The observed vibrational quanta ∆G(v ⫹ 1/2) for the P21ff /Q11ef band heads in the
2
Π1/2 component and for the P22ff /Q12ef band heads in the 2 Π3/2 component, of
14 16
N O, 15N 16O, and 14N 18O (see Table 13) are separately fitted to a linear least-
squares program and the values of (ωe ⫺ ωexe) and 2ωexe as per Eq. (22) were
obtained. The values of the vibrational constants ωe and ωexe obtained from the
observed band head data for the two components of the X 2 Π state for the three
isotopomers are listed in Table 16. In the same table we have listed the value of
∆G′(1/2) for 14N 16O, 15N 16O, and 14N 18O obtained from the average value of ∆G′(1/2)
from the P21ff /Q11ef and P22ff /Q12ef band heads, since only two levels v′ ⫽ 0 and
1 are observed for A2 ∑⫹ in [112].

Copyright © 2002 by Taylor & Francis

Figure 15 Fortrat diagram of the 0–2 band of the gamma system of 15 N 18O.

Table 16 Vibrational Constants a,b (in cm ⫺1) of the X 2 Π r and A 2 ∑ ⫹ states of 14 N 16O,
15 16
N O, and 14 N 18O
15
N 16 O 14
N 18 O
14
N 16 O
Observed Observed Calculated Observed Calculated
State Constant value value value value value

A 2∑⫹ ∆G′(1/2) 2345.3 2297.6 2282.4

X 2 Π 3/2 ωe 1902.95(64) 1868.06(85) 1868.92(63) 1855.0(1.3) 1852.83(62)
ωe xe 13.91(27) 13.29(18) 13.42(26) 13.38(28) 13.19(26)
X 2 Π 1/2 ωe 1902.9(1.9) 1869.75(96) 1868.9(1.9) 1853.35(86) 1852.8(1.8)
ωe xe 14.07(42) 13.51(21) 13.57(41) 13.24(18) 13.34(40)
a
Number in parentheses is the uncertainty in the last digit and corresponds to one standard deviation.
b
Obtained from the bandhead data given in Table 13.

Copyright © 2002 by Taylor & Francis

ACKNOWLEDGMENTS

This work was supported in part by Research Grant No. A-2440 awarded to
S.P.R. by the Natural Sciences and Engineering Research Council of Canada.
The assistance received from George Varghese during the preparation of the
manuscript is also acknowledged.

REFERENCES

1. AB Pluvinel, F Baldet. Compt Rend 148: 759–762, 1909.

2. TA Dixon, RC Woods. Phys Rev Lett 34: 61–63, 1975.
3. NR Erickson, RL Snell, RB Loren, L Mundy, RL Plambeck. Astrophys J 245: L83,
1981.
4. PH Krupenie. Natl Bur Stand 5: 1966.
5. HP Huber, G Herzberg. Molecular Spectra and Molecular Structure. Vol. IV.
Princeton, NJ: Van Nostrand-Reinhold, 1979.
6. Z Jakubek, R Kepa, A Para, M Rytel. Can J Phys 65: 94–100, 1987.
7. C Haridass, CVV Prasad, SP Reddy. Astrophys J 388: 669–677, 1992.
8. C Haridass, CVV Prasad, SP Reddy. J Mol Spectrosc 199:180–187, 2000.
9. CVV Prasad, SP Reddy. J Chem Phys 90: 3010–3014, 1989.
10. SP Reddy, CVV Prasad. J Chem Phys 91: 1972–1977, 1989.
11. CVV Prasad, SP Reddy. J Mol Spectrosc 144: 323–333, 1990.
12. A Lofthus, PH Krupenie. J Phys Chem Ref Data 6: 113–307, 1977.
13. RN Clayton. In: J Audouze, ed. CNO Isotopes in Astrophysics. Boston: Reidel,
1977.
14. SP Reddy, CVV Prasad. Astrophys J 331: 572–582, 1988.
15. IS McDermid, JB Laudenslager. J Quant Spectrosc Radiat Transfer 27: 483–492,
1982.
16. B Kivel, H Mayer, H Bethe. Ann Phys 2:57–80, 1957.
17. SP Langhoff, CW Bauschlicher Jr, H Patridge. J Chem Phys 89: 4909–4917, 1988.
18. MD Burrows, SL Banghcum, RC Oldernorg. Appl Phys Lett 46: 22–24, 1985.
19. G Herzberg, H Lew, JL Sloan, JKG Watson. Can J Phys 59:425–440, 1981.
20. SP Reddy, CVV Prasad. J Phys E Sci Instrum 22: 306–308, 1989.
21. G Herzberg. Molecular Spectra and Molecular Structure I. Spectra of Diatomic
Molecules. Malabar, FL: Kreger, 1991, pp. 149, 261, 315.
22. A Fowler. Monthly Notices R Astron Soc 70: 176–182, 1909.
23. A Fowler. Monthly Notices R Astron Soc 70: 484–496, 1910.
24. F Baldet. Compt Rend 180: 271–273; 820–822, 1925.
25. RT Binge. Nature 116: 170–171L, 1925.
26. D Coster, HH Brons, H Bulthuis. Z Physik 79: 787–822, 1932.
27. R Schmid, L. Gerö. Z Physik 86: 297–313, 1933.
28. H. Bulthuis. Proc Acad Sci Amsterdam 38: 604–617, 1935.
29. KN Rao. Astrophys J 111: 306–313, 1950.
30. RK Asundi, RK Dhumwad, AB Patwardhan. J Mol Spectrosc 34: 528–532, 1970.

Copyright © 2002 by Taylor & Francis

 31. RK Dhumwad, AB Patwardhan, VT Kulkarni. J Mol Spectrosc 78: 341–343,
1979.
32. H Gagnaire, JP Goure. Can J Phys 54: 2111–2117, 1976.
33. JA Coxon, SC Foster. J Mol Spectrosc 93: 117–130, 1982.
34. DH Katayama, JA Welsh. J Chem Phys 75: 4224–4230, 1981.
35. RD Brown, RG Dittman, DC McGilvery. J Mol Spectrosc 104: 337–342, 1984.
36. BR Vujisić, DS Pesić. J Mol Spectrosc 128: 334–339, 1988.
37. CM Blackburn. Proc Natl Acad Sci 11: 28–34, 1925.
38. H Biskamp. Z Physik 86: 33–41, 1933.
39. RH Schmid. Phys Rev 42: 182–188, 1932.
40. LH Woods. Phys Rev 63: 431–432, 1943.
41. P Misra, DW Ferguson, KN Rao, E Williams Jr, CW Mathews. J Mol Spectrosc
125: 54–65, 1982.
42. F. Baldet. Compt Rend 178: 1525–1527, 1924.
43. RC Johnson. Proc R Soc (London) Ser A 108: 343–355, 1925.
44. H Bulthuis. Physica 1: 873–880, 1934.
45. KN Rao, KS Sarma. Mem Soc R Sci Liege Collect 13: 181–1986, 1953.
46. L Ćonkić, JD Janjić, DS Pesić, D Rakotoarijimy, BR Vujisić, S Weniger. Astrophys
J 226: 1162–1170, 1978.
47. J Marchand, JD Incan, J Janin. Spectrochim Acta 25A: 605–609, 1969.
48. D Cossart, C Cossart-Magos. J Mol Spectrosc 141: 59–78, 1990.
49. KVLN Sastry, P Helminger, E Herbst, FC DeLucia. Astrophys J 250: L91–L92,
1981.
50. ND Piltch, PG Szanto, TG Anderson, CS Gudeman, TA Dixon, RC Woods. J Chem
Phys 76: 3385–3388, 1982.
51. FC van den Heuvel, WL Meerts, A Dymanus. Chem Phys Lett 92: 215, 1982.
52. M Bogey, C Demuynck, JL Destombes. Mol Phys 46: 679–681, 1982.
53. M Bogey, C Demuynck, JL Destombes. J Chem Phys 79: 4704–4707, 1983.
54. PB Davies, WJ Rothwell. J Chem Phys 83: 5450–5452, 1985.
55. KN Rao. Astrophys J 111: 50–59, 1950.
56. Z Bembenek, U Domin, R Kepa, K Porada, M Rytel, M Zachwieja, Z Jakubek,
JD Janjić. J Mol Spectrosc 165: 205–218, 1994.
57. TA Miller, VE Bondybey. Chem Phys Lett 50: 275–277, 1977.
58. VE Bondybey, TA Miller. J Chem Phys 69: 3597–3602, 1978.
59. HM Crosswhite. J Res Natl Bur Stand A Phys Chem A79: 17–68, 1975.
60. JM Brown, JT Hougen, KP Huber, JWC Johns, I Kopp, H Lefebvre-Brion, AJ
Merer, DA Ramsay, J Rostas, RN Zare. J Mol Spectrosc 55: 500–503, 1975.
61. JM Brown, EA Colbourn, JKG Watson, FD Wayne. J Mol Spectrosc 74: 294–318,
1979.
62. C Amiot, JP Maillard, J Chauville. J Mol Spectrosc 87: 196–218, 1981.
63. M Douay, SA Rogers, PF Bernath. Mol Phys 64:425–436, 1988.
64. LT Earls. Phys Rev 48: 423–424, 1935.
65. BR Vujisić, DS Pešić. J Mol Spectrosc 128: 334, 1988.
66. RD Brown, RG Dittman, DC McGilvery, PD Godfrey. J Mol Spectrosc 101: 61,
1983.
67. Z Jakubek, R Kepa, JD Janjić. J Mol Spectrosc 125: 43, 1987.

Copyright © 2002 by Taylor & Francis

 68. BR Vujisić, DS Pešić, S Weniger, D Rakotoarijimy. Indian J Pure Appl Phys 18:
370, 1980.
69. CVV Prasad, GL Bhale, SP Reddy. J Mol Spectrosc 104: 165, 1984.
70. CVV Prasad, SP Reddy, M Sandys-Wunsch. J Mol Spectrosc 114: 436, 1985.
71. CVV Prasad, GL Bhale, SP Reddy. J Mol Spectrosc 121: 261, 1987.
72. CVV Prasad, SP Reddy. J Mol Spectrosc 121: 261, 1987.
73. L. Conkić, JD Janjić, DS Pešić, D Rakotoarijimy, VR Vujisić, S Weniger.
Astrophys J 226: 1162, 1978.
74. RD Brown, RG Dittman, DG McGilvery, PD Godfrey. J Mol Spectrosc 101: 61,
1983.
75. Z Jakubek, R Kepa, Z. Rzeszut, M Rytel. J Mol Spectrosc 119: 280, 1986.
76. RR Shvangiradze, KA Oganezov, B Ya Chikhladze. Opt Spectrosc 8: 239–242,
1960.
77. D Pešić, J Janić, D Marković, M Rytel, T. Sewiec. Bull Chem Soc Belgrade 39:
249–255, 1974.
78. JD Janjić, DS Pešić. Astrophys J 209: 642–647, 1976.
79. DS Pešić, D Ž Marković. DS Janković. Fiz 7: 83–89, 1975.
80. M Fassbender. J Physik 30: 73, 1924.
81. A Lofthus. Spectroscopic Report No. 2. University of Oslo, Norway: Department
of Physics, 1960.
82. DC Tyte, RW Nicholls. Identification Atlas of Molecular Spectra No. 3. London,
Ontario, Canada: Department of Physics, The University of Western Ontario, 1965.
83. A Dick, W Benesch, HM Crosswhite, SG Tilford, RA Gottscho, RW Field. J Mol
Spectrosc 69: 95–108, 1978.
84. RA Gottscho, RW Field, KA Dick, W Benesch. J Mol Spectrosc 74: 435–455,
1979.
85. J Chevaleyre, JP Perott. J Mol Spectrosc 85: 85–96, 1981.
86. L Klynning, P Pages. Phys Scripta 25: 543–560, 1982.
87. F Michaud, F Roux, S David, An-Dien Nguyen. Appl Opt 35: 2867–2873,
1996.
88. F Michaud, F Roux, SP Davis, An-Dien Nguyen, CO Laux. J Mol Spectrosc 203:
1–8, 2000.
89. RW Wood, GH Dieke. J Chem Phys 6: 734–739, 1938.
90. RW Wood, GH Dieke. J Chem Phys 8: 351–361, 1940.
91. EA Colbourn, AE Douglas. J Mol Spectrosc 65: 332–333, 1977.
92. SP Reddy, CVV Prasad. Astrophys J 331: 572–582, 1988; 337: 579, 1989.
93. DL Albritton, AL Schmeltekopf, RN Zare. J Mol Spectrosc 67: 132–156, 1977.
94. JA Coxon. J Mol Spectrosc 72: 252–263, 1978.
95. R de L Kroning. J Physik 50: 347, 1928.
96. RJ Strutt. Proc R Soy Ser A 93: 254, 1917.
97. E Miescher, KP Huber. Int Rev Sci, Phys Chem 3: 37–73, 1976.
98. H Cisak, J Danielak, M Rytel. Acta Phys Pol A 37: 67–70, 1970.
99. R Engleman, Jr PE Rouse, HM Peek, VD Baiamonte. Los Alamos Scientific Labo-
ratory Report LA-4364, 1970.
100. R Engleman, Jr, PE Rouse. J Mol Spectrosc 37: 240–251, 1971.
101. R Freedman, RW Nicholls. J Mol Spectrosc 83: 223–227, 1979.

Copyright © 2002 by Taylor & Francis

102. RN Zare, AL Schmeltekopf, WJ Harrop, DL Albritton. J Mol Spectrosc 46: 37–
65, 1973.
103. JL Griggs, KN Rao, LH Jones, RM Potter. J Mol Spectrosc 22: 383–401, 1967.
104. C Amiot, R Bacis, G Guelachvili. Can J Phys 56: 251–265, 1978.
105. C Amiot, G Guelachvili. J Mol Spectrosc 76, 86–103, 1979.
106. A Valentin, A Henry, Ph. Cardinet, MF Le Maol, Da Wu Chen, KN Rao. J Mol
Spectrosc 70, 9–17, 1978.
107. A Henry, MF Le Moal, Ph. Cardinet, A Valentin. J Mol Spectrosc 70: 18–26, 1978.
108. JL Teffo, A Henry, Ph. Cardinet, A Valentin. J Mol Spectrosc 82, 348–363, 1980.
109. WL Meerts, A Dymanus. J Mol Spectrosc 44: 320–346, 1972.
110. CKN Patel, RJ Kerl. Opt Commun 24: 294–296, 1978.
111. RM Dale, JWC Johns, ARW McKellar, M Riggin. J Mol Spectrosc 67: 440–458,
1977.
112. DX Wang, C Haridass, SP Reddy. J Mol Spectrosc 175: 73–84, 1996.
113. C Amiot, JP Maillard, J Chauville. J Mol Spectrosc 87: 196–218, 1981.
114. A Gatterer, J Junkes. Astrophysical Laboratory of the Vatican Observatory, 2nd
ed. Cita Del Vaticano: Specola Vaticana, 1947.
115. AG Shenstone. J Opt Soc Am 45: 868, 1955.
116. FH Guezebroek, MG Tenenr, AW Kleyn, H Zacharias, S Stolte. Chem Phys Lett
187: 520–526, 1991.
117. M Douay, SA Rogers, PF Bernath. Mol Phys 64: 425–436, 1988.
118. R de Vivie, SD Peyerimhoff. J Chem Phys 90: 3660–3670, 1989.
119. RS Mulliken, A Christy. Phys Rev 38: 87–119, 1931.
120. JWC Johns, J Reid, DW Lepard. J Mol Spectrosc 65: 155–162, 1977.

Copyright © 2002 by Taylor & Francis

4
Photoabsorption Cross-Section
Measurements in the Ultraviolet
and Vacuum Ultraviolet

Kouichi Yoshino
Harvard–Smithsonian Center for Astrophysics,
Cambridge, Massachusetts

I. INTRODUCTION

Although wavelength measurements in spectroscopy have been very well per-

formed with photographic plates and films, photographic intensity measurements
were limited by poor dynamic range, nonlinearity, and technical difficulties.
However, after development of the photomultiplier there were many measure-
ments of photoabsorption coefficients in the Ultraviolet (UV) and Vacuum Ultra-
violet (VUV) region for many molecules during the period 1960 to 1980.
The ratio of the incident intensity I0(ν) to the intensity transmitted I(ν)
through a medium of column density N (cm⫺2) is related to the absorption cross-
section σ(ν) (cm2 mol⫺1) by the formula (Beer-Lambert law):

I0(ν)
ln ⫽ σN (1)
I(ν)

The quantity σN is called optical depth (also opacity or optical thickness). Two
other expressions also are used for absorption measurements: I(ν) ⫽ I0(ν)e⫺kpl
and I(ν) ⫽ I0(ν)10⫺εcl. The absorption coefficients κ and ⑀ are given in units of
atm⫺1 cm⫺1 and dm3 mol⫺1 cm⫺1, respectively. In the above expressions l is the
path length in units of cm, p is pressure in units of Torr or atm, and c is the
concentration in units of mol dm⫺3. Conversion factors between these absorption
coefficients (cross sections) are given in Okabe’s Table A-3 [1].

Copyright © 2002 by Taylor & Francis

II. CROSS-SECTION MEASUREMENTS BEFORE 1980

Absorption intensity measurements are the observation of the optical density un-
der known conditions such as the column density, path length, pressure of the
media, and temperature. Some earlier measurements presented the absorption
coefficients and the pathlength, but these data cannot be reduced to the absolute
scale without knowledge of the other parameters. Problems with these measure-
ments include estimation of I0(ν), the resolution of the spectrometer, and the
wavelength calibration. Most of instruments for the VUV region are single-path,
and the intensity of the incident light I0(ν) has to be assumed during the spectral
scan. This reflects on the uncertainty of the absolute base for absorption. Absorp-
tion cross-section (coefficient) measurements were performed using relatively
low resolution instruments (mostly 1 m spectrometer). As a result, the cross-
section measurements for molecular bands with fine structures were severely dis-
torted by the instrumental band widths, as pointed out by Hudson and Carter [2].
Wavelength calibration is essential to any spectroscopic measurements. Unfortu-
nately, the early workers did not have accurate calibration of wavelengths. There-
fore, we cannot simply compare the cross sections at the specific wavelength
because they may be at the shifted wavelengths. The cross sections for small
molecules up until 1977 were summarized by Okabe [1].

III. INSTRUMENTATION FOR CROSS-SECTION

MEASUREMENTS

It is important to have a clean and stable continuum background source to make

absorption cross-section measurements. We have available a xenon lamp oper-
ated by DC discharge in the UV region, with molecular hydrogen (or deuterium)
continuum from the bound to free molecular transitions in the wavelength region
above 170 nm. Occasionally the hydrogen source was used as the background
source below 170 nm, but the spectrum of hydrogen below 170 nm consists of
many molecular rotational lines. For low-resolution measurements, the spectrum
of molecular hydrogen looks like a continuum. The most convenient to use and
strongest continuum source in the VUV region is synchrotron radiation source.
Temperature-dependent absorption cross sections have become important
for applications in the atmospheric and astrophysical sciences. The absorbants
in cells can be cooled by a bath of precooled liquid such as methanol down to
their freezing point. Liquid nitrogen can used to cool down the gases to 78 K,
but this application is limited to selected molecules, which have sufficient vapor
pressure at this temperature. However, the supersonic expansion technique can
provide very cold gases around 20 K. The temperature reached by the supersonic

Copyright © 2002 by Taylor & Francis

expansion technique depends on the expansion rate, and the observation is limited
by small column density. Of course, the column density in the supersonic ex-
panded gases cannot be estimated directly, but can be deduced from the known
band oscillator strengths.
The absorbants can be separated from other medium using two windows.
The shortest wavelength reached by available windows material (LiF) is 105 nm.
The column density of gases can be obtained by
Pl
N ⫽ 9.65 ⫻ 1018 (2)
T
where P is pressure in torr, path length l in cm, and temperature T in K. The
absorption cell can be placed between the entrance slit and the background
source, or beyond the exit slit. The later application sometimes has an advantage
by limiting radiation to a monochromatic beam, and thus to gases that easily
dissociate. However, high-resolution instruments have focal plane scanning, and
therefore absorption cell must be placed at the entrance slit. Below 105 nm, the
main body of the spectrometer can be used as an absorption cell by sealing the
entrance slit to minimize leaking of gases through the slit. The relatively long
pathlengths of the spectrometer mean that the pressure is low and therefore the
accuracy of pressure measurements is also limited.
For the photoabsorption cross-section measurements of continuum and the
broad bands, low-resolution instruments can be used. The resolution should be
significantly better than the broad structures to observe. For line-by-line measure-
ments of bands with fine structures, it is an essential to use high-resolution in-
struments. Most such measurements have been done using a 6 or 10 m grating
spectrometer on photoelectric scan mode. Resolution of grating instruments is
0.3–0.4 cm⫺1, and they are larger than the Doppler-limited spectrum of simple
molecules. Only a few spectrometers remain available 6 m spectrometer at Center
for Astrophysics (Cambridge, MA), the Advance Light Source (Berkeley, CA),
the Photon Factory (KEK, Japan), and the 10 m spectrometer at Observatoire de
Paris-Meudon (France). Two institutions have combined the synchrotron radia-
tion and a high-resolution spectrometer into a facility for UV and VUV measure-
ments, (e.g., Photon Factory and the Advance Light Source).
The Fourier transform (FT) spectrometer is the most advanced instrument
for high-resolution cross-section measurements. A few commercial FT spectrom-
eters are available for measurements in UV region with resolution better than
0.1 cm⫺1, similar to the Doppler widths of simple diatomic molecules. The only
FT spectrometer capable of operating in the VUV region down to 140 nm is at
Imperial College (London, UK), with resolution down to 0.025 cm⫺1 [3]. Some
FT spectrometers have double detectors. One can be used for measurements of
the background intensity and the other for absorption measurements.

Copyright © 2002 by Taylor & Francis

IV. CROSS-SECTION MEASUREMENTS
OF THE PHOTOABSORPTION CONTINUUM

Cross-section measurements of photoabsorption continua can be performed using

an instrument with relatively lower resolution. It is convenient to use such instru-
ment to make a quick measurements to cover the entire wavelength region. Most
of the uncertainty in the cross-section measurements with a single path spectrom-
eter is estimating the background intensity level, I0(ν), since it may vary with
time and wavelength during the photoabsorption scan. It is also often difficult to
keep constant pressures of gases in the absorption cell during the scan, especially
with reactive or absorbent gases such as ozone and water. To overcome these
problems, absolute cross sections of the gases can be measured at several fixed
wavelength positions. At each position, the intensity can be recorded without
gases in the absorption cell, and then with gases in it. Background intensity is
estimated from the averages of two measurements (before and after) without
gases. The full ranges of scanned cross sections is then calibrated using these
absolute values at fixed wavelengths.
The wavelength scale of spectrometers, whether old or new (and now under
computer control), must be calibrated. The method of calibration depends on the

Figure 1 Absorption cross sections of H2O at 295 K in the wavelength region 140 nm
to 185 nm along with previous measurements [6]. The black circles show values from
Lee and Suto [7], gray circles from Laufer and McNesby [8], and open circles from Wata-
nabe and Zelikoff [9].

Copyright © 2002 by Taylor & Francis

background source. Traditional sources such as the hydrogen (deuterium) lamp,
Xe-lamp, and Ar-mini arc, often have impurity emission lines, which can be
used as wavelength standards. However, a very pure continuum source such as
synchrotron radiation has no lines. In this case, known absorption bands can be
used for wavelength calibrations such as the forth positive bands of CO in the
wavelength region 120–155 nm [4], and the Schumann-Runge bands of O2 from
175 to 190 nm [5]. In the UV region, atomic and ionic lines from hollow-cathode
lamps can also be used for the wavelength calibration.
Figure 1 illustrates the need for wavelength calibration and presents the
results of recent absorption cross-section measurements of H2O at 295 K [6]
compared with previous measurements. Measurements by Lee and Suto [7]
(black) agree completely with recent measurements [6] after a wavelength correc-
tion of ⫺0.57 nm. The results of Laufer and McNesby [8] (gray circles) are
shifted toward shorter wavelengths by 0.83 nm, and their cross sections around
170 nm are 5% lower than others. Early measurements of Watanabe and Zelikoff
[9] (open circles) agree in wavelength but their cross sections are 10% lower
than others. Some photoabsorption cross sections are pressure dependent, such
as the Herzberg continuum of O2. Pressure dependence usually suggests that the
absorption involves two molecules as dimers and/or pressure induced absorption.

V. CROSS-SECTION MEASUREMENTS OF THE BANDS

Cross-section measurements for bands with many fine rotational structures are
very limited, because of the lack of instrumental resolution. The best resolution
with grating instruments is 0.3–0.4 cm⫺1, compared with the Doppler widths of
around 0.1 cm⫺1 for most of the simple molecules. Hudson and Carter[2] pointed
out that plots of optical depth vs. column density are never linear, unless the
spectral features are very much broader than the instrumental widths. All mea-
sured cross sections are weighted averages over the instrumental widths, so that
the measured peak cross-section are lower limits to the true values; the measured
wing cross sections are upper limits. A more detailed discussion on the effect of
instrumental widths on cross-section measurements is found in the appendix of
work by Stark et al. [10].
The first measurements of line-by-line cross sections of molecules were of
the Schumann-Runge (S-R) bands of O2 in the wavelength region 175–200 nm.
The 6.6 m grating instrument used for these measurements was capable of a
resolution of 0.4 cm⫺1, but even this may not have been good enough to make
such line-by-line measurements. However, the upper levels of the S-R bands in
the VUV region are more or less predissociated, and so the rotational lines of
the S-R bands are broadened. These predissociation linewidths vary with the
vibrational and rotational quantum numbers, and are in the range 1–4 cm⫺1 [11].

Copyright © 2002 by Taylor & Francis

The broadened line widths make it possible to obtain absolute cross sections with
the grating instrument. The final results for cross sections of the S-R bands of
oxygen at 300 K are presented in Figure 3 of Yoshino et al. [12].
Photoabsorption cross sections depend on temperature, based on distribu-
tion of the rotational population. For atmospheric application of the measure-
ments of S-R bands, for example, cross sections at particular temperatures should

Figure 2 Temperature dependence of the photoabsorption cross sections. The (10,0)

band of the S-R system is plotted at 300 K and 130 K. (From Ref. 13.)

Copyright © 2002 by Taylor & Francis

be used. Minschwaner et al. [13] developed a model to calculate the cross sections
for temperatures between 130 and 500 K at 0.5 cm⫺1 steps based on high-resolu-
tion experimental measurements at 300K and 79K [14]. As an example, the calcu-
lated cross sections of the (10,0) band at 130K and 300K are presented in Figure
2, demonstrating their temperature dependency.
For astrophysical applications, temperatures of the interstellar media (10–
100 K) should be produced for laboratory measurement of spectra. By using the
supersonic expansion technique, cross sections of CO bands at wavelength be-
tween 92.5 and 97.4 nm have been observed [15] at ⬇ 20 K.
Even in transitions to highly excited levels, some bands (such as those of
N2, NO, and some O2 bands) appear with fine absorption lines that are close
to their Doppler widths (around 0.1 cm⫺1). In this case, the traditional grating
spectrometer cannot make reliable cross-section measurements because of limited
resolution. However the FT spectrometers can operate with resolution better than
0.1 cm⫺1 to measure bands with Doppler-limited lines. Cross-section measure-
ments of very weak Herzberg bands have been done in combination with FT
spectrometer and the White cell [16,17]. To obtain good signal-to-noise ratios,
FT spectrometry requires a stable and pure background source, and the ability
to limit the band widths. This combination was realized by taking the IC VUV-FT
spectrometer to the synchrotron radiation source at Photon Factory, thus making
possible ultra-high-resolution cross-section measurements of NO and O2 bands
[18] in the VUV region.
Laser spectroscopy contributes to very high resolution measurements in the
VUV region. However, the stability of the the laser sources makes it difficult to
measure cross sections. So far, the only contribution to cross-section measure-
ments is available from laser spectroscopy for A–X bands of CO [19].

VI. LINE AND BAND OSCILLATOR STRENGTHS

Integrated band cross sections are determined by summing σ(ν) over all of the
rotational lines belonging to the (v′, v″ ) band, and converting to the band oscilla-
tor strengths ( f-values) by the relation

mc2 1
fv′v″ ⫽ ∫ σ(ν)dν (3)
πe2 Ñ(v″ )

in which Ñ(v″) is the fractional Boltzmann population of the vibrational level

involved. The band oscillator strength is related to the line oscillator strength by

S J″J′
fv′v″ J′J″ ⫽ fv′v″ (4)
2J″ ⫹ 1

Copyright © 2002 by Taylor & Francis

where S J″J′ is the Hönl-London factor. The integrated cross-section is less affected
by the instrumental function than are the peak cross-sections. The true intensity
of the line comes from the the entire area of its absorption. Integrated cross sec-
tions of the line can be performed with the column density of the particular rota-
tional level (J″) obtained using Boltzmann statistics. We could also obtain band
oscillator strengths from integrated cross sections of the lines [19].
Band oscillator strengths are absolute values, and are independent of the
temperature. The integrated area over the entire bands at two temperatures is
the same. At the lower temperature, the rotational distribution is limited, and the
peak intensity of the low J lines become higher, as demonstrated in Figure 2.
This fact can be used when, for example, the column density cannot be measured,
as with a supersonic jet. Total band absorption at the jet temperature can be
combined with band oscillator strengths to deduce the column density [15].
Once band oscillator strengths are established, cross sections of any temper-
ature can in principle be obtained with known Hönl-London factors. However,
usually many of the excited levels are more or less perturbed by neighbor levels,
and their intensities affected. In these cases, calculated cross sections from band
oscillator strengths might not yield the correct values. Even after depertubation
treatments using energy shifts from the perturbation, the calculated intensity ef-
fected by the perturbation is not reliable, because the intensities are very sensitive
to interactions. Therefore, measurements of line oscillator strengths become im-
portant, especially in applications of astrophysical interest.

VII. RESULTS OF RECENT MEASUREMENTS

The results of UV cross sections are presented in Tables 2 and 3 of Chapter 14,
for broad-cross sections and line-by-line data in HITRAN, respectively. Most of
the VUV measurements are available from cfa-www.harvard.edu/amdata/
ampdata/cfamols.html.

REFERENCES

1. H Okabe. Photochemistry of Small Molecules. New York: John Wiley & Sons, 1978.
2. RD Hudson, VL Carter. Bandwidth dependence of measured UV absorption cross
sections of Argon. J Opt Soc Am 58:227–232, 1968.
3. AP Thorne, CJ Harris, I Wynne-Jones, RCM Learner, G Cox, A Fourier transform
spectrometer for the vacuum ultraviolet: design and performance. J Phys E: Sci In-
strum, 20:54–60, 1987.
4. SG Tilford, JD Simmons. Atlas of the observed absorption spectrum of Carbon Mon-
oxide between 1060 and 1900 Å. Phys Chem Ref Data 1:147–188, 1972.

Copyright © 2002 by Taylor & Francis

 5. K Yoshino, DE Freeman, WH Parkinson. Atlas of the Schumann-Runge absorption
bands of O2 in the wavelength region 175–205 nm. Phys Chem Ref Data 13:207–
227, 1984.
6. K Yoshino, JR Esmond, WH Parkinson, K Ito, T Matsui, Absorption cross section
measurements of water vapor in the wavelength region 120 nm to 188 nm. Chem
Phys 211:387–391, 1996.
7. LC Lee, M Suto. Quantitative photoabsorption and fluorescence study of H2O and
D2O at 50–190 nm. Chem Phys 110:161–169, 1986.
8. AM Laufer, JR McNesby. Deutrium isotope effect in vacuum ultra-violet absorption
coefficients of water and methane. Can J Chem 43:3487–3490, 1965.
9. K Watanabe, M Zelikoff. Absorption coefficient of water vapor in the vacuum ultra-
violet. J Opt Soc Am 43:753–755, 1953.
10. G Stark, K Yoshino, PL Smith, K Ito, WH Parkinson. High-resolution absorption
cross sections of carbon monoxide bands at 295 K between 91.7 and 100.4 nm.
Astrophys J 369:574–580, 1991.
11. AS-C Cheung, K Yoshino, JR Esmond, SS-L Chiu, DE Freeman, WH Parkinson.
Predissociation line widths of the (1,0)–(12,0) Schumann-Runge absorption bands
of O2 in the wavelength region 179–202 nm. J Chem Phys 92:842–849, 1990.
12. K Yoshino, JR Esmond, AS-C Cheung, DE Freeman, WH Parkinson. High resolu-
tion absorption cross sections in the transmission window region of the Schumann-
Runge bands and Herzberg continuum of O2. Planet Space Sci 40:185–192, 1992.
13. K Minschwaner, GP Anderson, LH Hall, K Yoshino. Polynomial coefficients for
calculating O2 Schumann-Runge cross sections at 0.5cm⫺1 resolution. J Geophys Res
97:10,103–10,108, 1992.
14. K Yoshino, DE Freeman, JR Esmond, WH Parkinson. High resolution absorption
cross sections and band oscillator strengths of the Schumann-Runge bands of oxygen
at 79K. Planet Space Sci 35:1067–1075, 1987.
15. K Yoshino, G Stark, JR Esmond, PL Smith, K Ito, M Matsui. High-resolution ab-
sorption cross section measurements of supersonic jet-cooled carbon monoxide be-
tween 92.5 and 97.4 nanometers. Astrophys J 438:1013–1016, 1995.
16. K Yoshino, JR Esmond, JE Murray, WH Parkinson, AP Thorne, RCM Learner, G
Cox. Band oscillator strengths of the Herzberg I bands of O2. J Chem Phys 103:
1243–1249, 1995.
17. M-F Mérienne, A Jenouvrier, B Coquart, M Carleer, S Fally, R Colin, AC Vandaele,
C Hermans. Fourier transform spectroscopy of the O2 Herzberg bands. II. Band oscil-
lator strengths and transition moments. J Molec Spectrosc 202:171–193, 2000.
18. T Imajo, K Yoshino, JR Esmond, WH Parkinson, AP Thorne, JE Murray, RCM
Learner, G Cox. The application of a VUV Fourier transform spectrometer and syn-
chrotron radiation source to measurements of: II. The δ(1,0) band of NO. J Chem
Phys 112:2251–2257, 2000.
19. G Stark, BR Lewis, ST Gibson, JP England. High-resolution oscillator strength mea-
surements for high-v′ bands of the A 1II(v′)-X 1 ∑⫹(v″ ⫽ 0) system of carbon monox-
ide. Astrophys J 505:452–458, 1998.

Copyright © 2002 by Taylor & Francis

5
Ultraviolet and Vacuum Ultraviolet
Laser Spectroscopy Using
Fluorescence and Time-of-Flight
Mass Detection
R. H. Lipson and Y. J. Shi
University of Western Ontario, London, Ontario, Canada

I. INTRODUCTION

The generation of tunable, monochromatic, and coherent short-wavelength light

by nonlinear optics is a continuing and important research pursuit. Many different
and novel chemical and physical applications using such sources have also been
reported. These two subjects, however, are so vast that the review here by neces-
sity will focus only on the principles and applications of ultraviolet (UV; 200
nm ⱕ λ ⱕ 300 nm), vacuum ultraviolet (VUV; 100 nm ⱕ λ ⱕ 200 nm), and
extreme ultraviolet (XUV; λ ⱕ 100 nm) so-called lasers generated by frequency
doubling in anisotropic crystals, nonresonant tripling or third harmonic genera-
tion (THG), and two-photon resonant four-wave sum- and difference-mixing
(FWSM and FWDM, respectively) in isotropic gases. The word laser is qualified
to serve as a reminder that the production of coherent short-wavelength radiation
is the result of parametric processes, and not population inversions. Furthermore,
applications will be limited to those in which single-photon excitations of jet-
cooled gas-phase molecules have been detected by either laser-induced fluores-
cence (LIF) or ion detection. This will unfortunately exclude a vast literature on
lamp and synchrotron-based experiments, as well as experimental results on high-
lying states in the UV, VUV, or XUV accessed by multiphoton absorption using
longer-wavelength sources.

Copyright © 2002 by Taylor & Francis

II. EXPERIMENTAL TECHNIQUES

A brief background review is provided here on a number of the different princi-

ples and experimental methods used in the UV and VUV laser spectroscopy of
jet-cooled molecules.

A. Light Sources
Nonlinear optics began in 1961 with the pioneering work of Franken and co-
workers, who doubled the red output of a ruby laser by second harmonic genera-
tion (SHG) in crystalline quartz [1]. Today, frequency doubling, using a battery
of different crystals, is commercially available and so commonplace that the fol-
lowing discussion will focus primarily on third-order nonlinear phenomena. The
theory of VUV generation is also well understood. A reiteration of some of the
major points is nevertheless warranted, particularly for those whose interests lie
not so much in laser physics but in potential applications. The interested reader
is directed to several comprehensive reviews on this topic [2–4].
All nonlinear optical signals originate from an oscillating nonlinear polar-
ization, P(ω) (dipole/unit volume), which can be induced in a material by the
strong external electric fields, E(ω), of one or more incident laser beams. Under
moderate laser powers, P(ω) can be written as a Taylor series in E(ω):
P(ω) ⫽ N(χ(1) ⋅ E(ω) ⫹ χ(2) :E(ω)E(ω) ⫹ χ(3) ⯗E(ω)E(ω)E(ω) ⫹ ⋅ ⋅ ⋅) (1)
where N, χ , and χ , n ⱖ 2, are the number density, linear susceptibility, and
(1) (n)

nth-order nonlinear susceptibilities tensors of the medium, respectively. Only the

first two nonlinear terms, which can serve as sources for coherent light at new
frequencies, will be considered.
The expression proportional to χ(2) is responsible for second harmonic gen-
eration in anisotropic crystals. However, isotropic gases are most often used for
VUV generation because nonlinear solid-state materials transparent to wave-
lengths shorter than the UV do not exist. The current limit is ⬇190 nm for β-
barium borate (BBO). Second harmonic generation in atomic media has also been
demonstrated, but the mechanism is now generally accepted to be a third-order
nonlinear effect involving two input laser photons and a third radial DC electric
field produced by atom ionization [5]. In the absence of DC electric fields, and
in isotropic media in which χ(2) is identically zero, the leading term that can yield
nonlinear effects, including THG, FWSM, and FWDM, involves the third-order
nonlinear susceptibility, χ(3).

1. Frequency Doubling
Wavelengths down to ⬇210 nm with a frequency bandwidth ⱕ0.5 cm⫺1 are rou-
tinely generated in many spectroscopy labs by SHG of pulsed fundamentals in

Copyright © 2002 by Taylor & Francis

anisotropic crystals. Several groups, however, are using continuous wave (cw)
lasers to obtain much higher resolution spectra. Two such systems are briefly
described here.
Pratt and co-workers have described an ultra-high-resolution UV laser they
have used to record fully rotationally resolved LIF excitation spectra of large
polyatomic molecules [6]. In their experiments, a single-mode ring dye laser
capable of stabilized scans of 90–120 GHz in the visible spectrum, and operating
on Rhodamine 6G (595–630 nm), Kiton Red (620–640 nm) and/or DCM (630–
700 nm) dye solutions, is pumped by ⬇1 W of the 514.5 nm line of an argon–
ion laser. UV light is generated by placing an LiIO3 doubling crystal in the auxil-
iary waist of the ring laser to produce up to ⬇1 mW output power in the UV.
Phase matching the SHG crystals doubles the scan range to 180–240 GHz or 6–
8 cm⫺1. The spectrometer has a frequency bandwidth resolution of 2.5 MHz at
an excitation wavelength of 300 nm, corresponding to an enviable resolving
power of ν/∆ν ⫽ 4 ⫻ 108.
A second intermediate resolution system reported by Bitto and Willmott
[7] involves pulsed dye amplifying the cw output of a ring laser with a typical
bandwidth of several MHz. Fourier transform limited pulses with up to a 100
mJ energy are obtained in this way using 550 mJ 532 nm light from a frequency-
doubled Q-switched Nd :YAG laser. Approximately 150 MHz frequency band-
width UV pulses with energies up to ⬇300 µJ are subsequently generated by
SHG in nonlinear crystals. Such a system is extremely versatile because it allows
both precise high-resolution frequency-resolved and time-resolved experiments
to be carried out.

2. Nonresonant Third Harmonic Generation

An energy level diagram for nonresonant tripling [8], where ωoutput ⫽ 3ω, and ω
is the input angular frequency of a single laser, is shown in Figure 1a. When a
Gaussian beam of frequency ω and input power Pω, is focused into an atomic
medium, the resultant third harmonic output power P3ω is given by [9]:
P(3ω) ⫽ N 2[χ(3) (3ω)]2 P 3ω F(b ∆k) (2)
The geometrical phase-matching function, F, describes the dispersive phase ve-
locity relationship between the input and output waves, and depends on the prod-
uct b ∆k, where b is the confocal beam parameter of the focused radiation and
∆k is the wave–vector mismatch between the input and output light beams:
2πn( λ3ω) 2πn(λω)
∆k ⫽ k3ω ⫺ 3kω ⫽ ⫺3 (3)
λ3ω λω
The indices of refraction, n(λ3ω) and n( λω), are those for the medium at the third
harmonic and input laser beam wavelengths, λ3ω and λω, respectively. The product

Copyright © 2002 by Taylor & Francis

Figure 1 (a) Energy level scheme indicates a one-color nonresonant third harmonic
generation process at the fundamental laser frequency ω (b) A four-level system indicates
the relevant energy levels for both four-wave sum mixing at 2ω1 ⫹ ω2, and four-wave
difference mixing at 2ω1 ⫺ ω2. IP, ionization potential.

N 2 |χ(3)(3ω)| is usually small for most media when the fundamental laser fre-
quency does not match a resonance condition for an allowed transition. P3ω in
general cannot, however, be increased simply by using a larger gas pressure,
because ∆k is also a function of the number density N. Instead, the product
N 2F(b ∆k) [or ∆k2F(b ∆k)] must be optimized.
THG is typically, generated using a pulsed laser output that is tightly fo-
cused so that b ⬍⬍ the length of the medium L. This enhances the conversion
efficiency by maximizing the P 3ω power dependence in Eq. 2. At the same time
the phase matching function F(b ∆k) must also be considered for this focusing
condition [9]:

F(b ∆k) ⫽ π 2 (b ∆k)2 eb∆k ∆k ⬍ 0

(4)
⫽0 ∆k ⱖ 0

Equation (4) intimates that nonresonant THG is only possible in spectral regions
in which the atomic medium is negatively dispersive. Gaps in the tuning curves
are found because ∆k is positive to the red of each allowed Rydberg state ←
ground state transition due to anomalous dispersion. The function, ∆k2F(b ∆k) is
a maximum when b ∆k ⫽ ⫺4. THG can therefore be optimized by changing
either the beam focus to vary b and/or changing N to alter ∆k. Since the refractive

Copyright © 2002 by Taylor & Francis

index is a strong function of λ near a resonance line, coherent VUV generation
at any allowed wavelength can be maximized by adding different amounts of a
positively dispersive buffer gas to the nonlinear medium.
The rare gases (Xe [10,11], Kr [12,13], Ar [14], and Ne [15]), or metal
vapors such as Hg [16], are most commonly employed for nonresonant THG
(Fig. 2). The latter, however, must be produced in high-temperature heat pipe
ovens, while the former can be housed in either a room-temperature gas cell with
a MgF2 or LiF output window for VUV generation, or expanded in a spatially
localized supersonic jet (described below) making it a near-ideal ‘‘windowless’’
environment for XUV generation [17] (Fig. 3). Typical conversion efficiencies
of ⬇10⫺6 –10⫺3% leading to intense coherent VUV outputs are possible using
commercial pulsed dye laser peak powers of 1–10 MW (⬇ 10 nsec pulse dura-
tion). It is also straightforward to triple the third harmonic output of a neodym-
ium: yttrium aluminum garnet (Nd :YAG) laser at ⬇355 nm to produce VUV
light at ⬇118 nm [18]. As described below, this source has been used extensively
for analytical detection of molecules, despite its fixed wavelength, due to its tech-
nical simplicity.

Figure 2 Main spectral regions in the VUV and XUV where coherent light has been
generated experimentally by nonresonant third harmonic generation, using Ne, Ar, Kr,
Xe, or Hg as nonlinear media.

Copyright © 2002 by Taylor & Francis

Figure 3 Illustration of a supersonic jet produced by expanding a gas from a region of
high pressure into vacuum. The orientation of the arrows on the dots (indicating atoms
and/or molecules) shows that the velocity distribution is narrower below the pinhole than
in the high-pressure reservoir and, therefore, that the resultant translational temperature
is lower. The spatially localized nature of the expansion located with the shock fronts make
it a near-ideal windowless environment for XUV generation with focused laser beams.

3. Two-Photon Resonance-Enhanced Four-Wave Mixing

As shown in Figure 1b, two-photon resonance-enhanced four-wave mixing is a
two-color variation of third-harmonic generation, where three input fundamental
photons from two lasers, two at angular frequency ω1, and one at angular fre-
quency ω2, are used to produce a fourth output photon at the sum (2ω1 ⫹ ω2)
and/or the difference (2ω1 ⫺ ω2) frequency, both of which can lie in the VUV
or XUV (Fig. 1b) [19,20]. The expression ‘‘four-wave mixing’’ is appropriate
because three input waves are used to generate a fourth output signal.
In an expression similar to Eq. (2), the intensity of the generated VUV
light, PVUV, for sum-mixing is governed by a P12P2 incident laser power depen-
dence for beams at frequencies ω1 and ω2, respectively, and by a phase-matching
function F(b ∆k) that is identical to Eq. (4) when a tight focusing geometry is
used. However, the wave vector mismatch here is now defined as ∆k ⫽ kVUV ⫺
2k1 ⫺ k2, where k1, k2, and kVUV are the wave vectors for the input and output
light beams, respectively. Although it may appear at first glance that resonant
four-wave sum-mixing, like nonresonant tripling, is only possible in spectral re-

Copyright © 2002 by Taylor & Francis

gions where the nonlinear medium is negatively dispersive, most atomic elements
used for this purpose are, to a good approximation, inherently negatively disper-
sive for virtually every VUV and XUV wavelength in their tuning ranges. Their
first resonance lines, which carry most of the oscillator strength, come to the blue
of the majority of the input laser frequencies used in the nonlinear optical process.
However, the resultant broad tunability is often only achieved at the expense of
the optimal phase-matching conditions.
The dispersion criterion for FWDM is less stringent in a tight focussing
situation since ∆k can be either positive or negative. The phase matching function
F(b ∆k) ⫽ π 2 e⫺b|∆k| (5)
is a maximum when ∆k ⫽ kvuv ⫺ 2k1 ⫹ k2 ⫽ 0 [9] and, ordinarily, the FWDM
conversion efficiency is about an order of magnitude better than that for FWSM.
One significant difference between resonance-enhanced four-wave mixing
and nonresonant THG is that an appreciable VUV output conversion efficiency
(on the order of 1 part in 107 –105) for the former can be achieved using moderate
incident pulsed laser powers by judiciously tuning frequencies ω1 and ω2 to be
either one- or two-photon resonant with allowed transitions (Ωag, Ωbg) in the me-
dium. This approach minimizes one or more of the resonance denominators in
the expression for the third-order nonlinear susceptibility [21]:

χ(3)(2ω1 ⫾ ω2; ω1, ω1, ⫾ ω2) ⫽ (6)

〈g|e4µ̂| a〉 〈a|e1µ̂| b〉 〈b|e2µ̂| c〉 〈c |e3µ̂| g〉
冱 (Ω
1
h3 abc ag ⫺ ω1) (Ωbg ⫺ 2ω1) (Ωcg ⫺ 2ω1 ⫿ ω2)

Here, ek is the unit vector identifying the polarization of wave vector k, µ̂ is the
electric dipole moment operator, Ωjg are the complex frequencies for the j ← g
transitions, and the summation is over all states.
Two-photon resonance enhancement is most commonly used experimen-
tally since it can substantially increase χ(3) with minimal two-photon absorption
in the medium, provided that the pulse durations of the fundamental laser beams
are reasonably short (ⱕ10 ns). The one-photon resonance condition, on the other
hand, can lead to substantial reduction in the coherent output due to strong linear
absorption of the fundamental. Additional resonance enhancement is also possi-
ble if the VUV output frequencies correspond to allowed transitions in the me-
dium, Ωcg. This effect is usually small when the unstructured continuum of an
atomic medium is used, due to the small magnitude of the bound-continuum 〈c |e3
µ̂| g〉 matrix element [22]. A better conversion efficiency, however, is possible if
the generated VUV wavelength is resonant with a transition to an autoionized
level embedded in the continuum. Likewise, VUV light generated at wavelengths
longer than the ionization potential of the nonlinear medium can often be in

Copyright © 2002 by Taylor & Francis

resonance with strongly allowed Rydberg state ← ground state transitions. While
the conversion efficiency can then be very high (up to 5%) [23], the possibility
of linear reabsorption is also large, and the generated light intensity can change
dramatically with wavelength. While this latter effect is the anithesis of what is
desired in a tunable spectroscopic light source, the resultant spectral fluctuations
can handled by appropriate normalization techniques. At all times, input laser
frequencies for atomic transitions leading to nonzero electric dipole matrix ele-
ments in the numerator of Eq. (6) must also be chosen.
In our laboratory, VUV light is routinely generated between ⬃174 and 66
nm by four-wave mixing in Mg and Hg vapor, as well as in Xe and Kr gas
[24,25]. Energy level diagrams and tuning curves for these elements are shown
in Figure 4.

Figure 4 Energy level diagrams of the relevant atomic levels of the nonlinear media
used at the University of Western Ontario for four-wave mixing. (a) Kr; (b) Xe; (c) Hg;
(d) Mg. The wavelength regions generated by either four-wave sum- and/or difference
mixing are depicted as hatched regions. The wavelengths (nm) of the fundamental laser
beams indicated are those providing two-photon resonance enhancement of the third-order
nonlinear susceptibility of the nonlinear media. IP, ionization potential of the nonlinear
medium.

Copyright © 2002 by Taylor & Francis

Table 1 Comparison of Second- and Third-Generation Synchrotrons and
VUV Lasers

Second-generation Third-generation Four-wave mixing

Properties synchrotron [84] synchrotron [85] sources

E-range 10⫺1 –3 ⫻ 104 eV Optimized for 10 eV to 6–19 eV

1 keV
Intensity ⬍109 Photons/pulse/ Up to 1013 photons/sc/ 1012 photons/pulse
0.1% band pass 0.01% band pass
Resolution 10⫺3 eV (0.03 Å) at 5– ⬍10 meV at 100 eV; ⱕ0.5 cm⫺1
40 eV; 10⫺2 eV (0.02 E/∆E as high as 105
Å) at 40–100 eV
Repetition ⬃1–500 MHz ⬍500 MHz; ⬎ 1.5 Typically ⬍50 Hz
rate MHz
Pulse duration ⬎100 ps ⬍70 ps ⱕ10 nsec

It is common to obtain VUV photon numbers exceeding 1010 per pulse by

generating the fundamental beams using commercial dye lasers. The monochro-
maticity of the VUV is then determined essentially by these sources. Current
commercial pulsed dye lasers have narrow output line widths, often ⬍0.07 cm⫺1,
which yield VUV line widths ⬍0.12 cm⫺1. The corresponding frequency-resolv-
ing power, υ/∆υ, is better than 106 at the shortest wavelengths. This, combined
with the resultant high spectral brightness due to the low divergence of the VUV
output, represents a considerable improvement over the best monochromators
available, and current synchrotron sources (Table 1). Gas-phase spectral resolu-
tion at the experimental level will be a convolution of the source line width,
Doppler broadening, and other broadening mechanisms such as predissociation
and/or autoionization. Individual rotational lines having a full-width at half maxi-
mum of ⱕ ⬇0.5 cm⫺1 in the VUV can usually be easily resolved.

B. Supersonic Jets
Supersonic jets have been established as a powerful tool for gas-phase spectros-
copy since they can produce rovibrationally cold molecules, radicals, and clusters
[26]. A free jet, shown schematically in Figure 3, arises when a gas is expanded
adiabatically from a region of high pressure (stagnation pressure Pstag) through a
pinhole of diameter D into vacuum. When D is larger than the mean free path
between the gas molecules, the large number of resultant binary collisions dra-
matically narrows the velocity distribution of the sample; that is, all the atoms
and/or molecules in the expansion travel in the same direction with the same
speed. The local speed of sound drops rapidly away from the pinhole, as the

Copyright © 2002 by Taylor & Francis

square root of the translational temperature, Ttrans , characterizing the width of the
gaseous velocity distribution. At the same time only a finite amount of thermal
energy in the high-pressure reservoir can be converted into a directed mass flow.
Thus, the maximum velocity a molecule can attain in a supersonic jet is actually
only slightly larger than the maximum velocity of a room-temperature Boltzmann
distribution. The term supersonic is used, however, because the ratio of the ap-
proximately constant flow velocity over the local speed of sound, M (Mach num-
ber) can easily exceed unity.
The relationship among temperature (T ), pressure (P), and density (ρ), in
the expansion to the Mach number and the stagnation temperature, pressure, and
density (Tstag, Pstag, and ρstag) behind the pinhole is given by [27]:

γ⫺1 ⫺1

冢 冣 冢 冣 冢 冣
(γ⫺1)/γ
T P ρ γ⫺1 2
⫽ ⫽ ⫽ 1⫹ M (7)
Tstag Pstag ρstag 2

where γ is the heat capacity ratio, Cp /Cv . Since calculations predict M to vary as
(x/D)γ⫺1, where x is the distance between a point on the flow axis and the nozzle
pinhole, it follows that the thermodynamic properties described by Eq. (7) are
strongly position-dependent within the isentropic core.
Molecules seeded in a monoatomic gas (typically He or Ar) will attempt
to equilibrate with Ttrans by collisionally transferring their energy stored in vibra-
tional and rotational degrees of freedom to the forward directional mass flow.
This can often result in highly nonequilibrium internal energy distributions, be-
cause the relaxation rates are different for different motions of the molecule. As
a rule of thumb, the translational, vibrational, (Tvib), and rotational (Trot) tempera-
tures of the molecules are ordered Ttrans ⱕ Trot ⬍ Tvib. At the same time, the gas
density also drops precipitously away from the pinhole. Phase transitions that
require three or more body collisions therefore become extremely improbable.
Accordingly, the resultant internal gas phase rotational temperatures are almost
always lower than the freezing point of a bulk sample, and often dramatically
(Trot ⱕ 10 K is not uncommon). The populations and, therefore, temperatures, of
the various degrees of freedom become ‘‘frozen’’ at that point in the expansion
where the number of collisions has effectively dropped to zero. Van der Waals
clusters can also be produced in a jet, but usually at Pstag conditions higher than
those needed simply to cool a stable monomer.
The position of the Mach disk, XM, formed when the jet pressure equals
the background gas pressure in the vacuum chamber, Pchamber, can be estimated
from

√
Pstag
XM ⫽ 0.67D (8)
Pchamber

Copyright © 2002 by Taylor & Francis

The gas beyond XM is much hotter than that in the isentropic core and, therefore,
the laser focus within the jet should be located between the nozzle pinhole and
XM. This condition, however, is very easy to achieve under typical but modest
experimental conditions where, for example, Pstag ⫽ 2 atm, Pchamber ⫽ 1.3 ⫻ 10⫺7
atm (⬇10⫺4 torr), D ⫽ 1.0 mm, and XM is ⬇2.6 m.
Although the discussion above is valid for continuous free jet expansions,
the same formalism is usually assumed to also hold for pulsed jets (often in the
200 µs range). These devices are in wide use since they provide lighter gas loads
for the pumping system due to their reduced duty cycle.

C. Detection Methods
A jet environment is poorly suited to direct absorption measurements because
the density of the absorbing molecules is generally very low and the absorption
path length is short. Thus, only a small fraction of the incident light intensity
can be absorbed, and that decrease must be detected over fluctuations in the UV
and VUV intensity, which can be quite pronounced due to the nonlinear processes
used to generate the light. It is fortunate that considerably more sensitive detec-
tion methods have been developed to overcome these problems. Here, the rudi-
ments of two such approaches, namely laser-induced fluorescence (LIF) and ion
detection by time-of-flight (TOF), mass spectrometry are presented briefly.

1. Laser-Induced Fluorescence
Laser-induced fluorescence involves the detection of optical emission from mole-
cules that have been excited by the absorption of laser radiation. Such an approach
for jet-cooled molecules requires excited states with a relatively short lifetimes
(ⱕµs). In this regard, unlike the UV and VUV, LIF in the infrared and microwave
regions is extremely impractical because single electronic vibrational and rota-
tional level lifetimes can be orders of magnitude longer.
UV and VUV photons usually have more than enough energy to excite
electronic transitions and, therefore, the resultant spectra can provide detailed
information about both electronic states involved. Two different experimental
arrangements are possible. In the first, emission from a specific level excited at
a fixed frequency is spatially separated in a monochromator and detected with
a photomultiplier tube to yield a dispersed fluorescence or emission spectrum.
Measurements of the resultant line spacings yield structural information about
the lower state. As an alternative, excitation spectra can be recorded by monitor-
ing the total undispersed emission as a function of the tunable exciting light
frequency. As the analysis below shows, the result is equivalent to an absorption
spectrum, and provides detailed excited state information [28].
Consider the experimental arrangement and energy level diagrams in Fig-
ure 5a and b, respectively. If collisions are neglected, each laser photon absorbed

Copyright © 2002 by Taylor & Francis

Figure 5 (a) Schematic diagram of a typical experimental arrangement required to rec-
ord laser-induced fluorescence excitation spectra. (b) Three-level energy level diagram
shows a single photon transition, and the possible resonant (⫽excitation frequency) and
nonresonant fluorescence decay channels.

will be emitted with unit quantum efficiency. The relationship between the num-
ber of photons detected per second at incident laser frequency ω, Np(ω), by an
optical system/photomultiplier combination having collection efficiency (Φ) and
quantum efficiency (QE), respectively, and the number of photons absorbed per
second within the focal volume ∆ V, NA (ω), is given by:
NP(ω) ⫽ NA(ω) ⋅ Φ ⋅ QE (9)
NA in turn can be expressed as
NA(ω) ⫽ σ(ω) ⋅ N ⋅ ∆V ⋅ F(ω) (10)
where σ(ω) is the absorption cross section at incident laser frequency ω, N is
the number of molecules per cm3, and F(ω) is the incident laser flux in photons
cm⫺2 s⫺1. An excitation spectrum involves recording
NP(ω)
⫽ σ(ω) ⋅ N ⋅ ∆V ⋅ Φ ⋅ QE (11)
F(ω)
If Φ and QE are independent of ω; (i.e., if they are insensitive to changes in
fluorescence over the range of absorption frequencies) the normalized ratio in
Eq. (11) represents the true absorption spectrum. As an example, consider a 1
mW continuous-wave UV laser operating at 303 nm (corresponding to ⬇2 ⫻ 1015
photons/s) with a beam cross-sectional area given by ∆V/∆x. With a collection
efficiency of Φ ⫽ 0.1 and a quantum efficiency of QE ⫽ 0.2,

Copyright © 2002 by Taylor & Francis

 NP ⫽ σ(ω) ⋅ N ⋅ ∆x ⋅ (4 ⫻ 1013) (12)
Assuming a photomultiplier dark count ND ⫽ 200 counts s , a signal-to-noise
⫺1

ratio NP /ND ⱖ 1 can be achieved for σ(ω) ⋅ N ⋅ ∆x ⱖ 5 ⫻ 10⫺12, which is sev-

eral orders of magnitude more sensitive than conventional absorption spectros-
copy. Although, in principle, excitation spectroscopy involves light detection
against a zero background, specific care must be taken to eliminate scattered
light that could be several orders of magnitude larger than the fluorescence. A
spatial pinhole filter before the photomultiplier can help significantly in this
regard.

2. Ion Formation and Detection by TOF Mass Spectrometry

Instead of monitoring absorptions through LIF, a second laser photon can be
used to ionize the excited species. A (1 ⫹ 1′) resonance-enhanced multiphoton
ionization (REMPI) scheme is shown in Figure 6. In general, the integers m and
n in (m ⫹ n) REMPI refer to the number of photons required to populate and
subsequently ionize the state of interest, respectively [29–31]. A prime associated
with the second integer indicates that the ionizing photon comes from a different
laser than that used to excite the species of interest. The resultant ions can be
collected with efficiency Φ ⫽ 1 and detected with an electron multiplier, chan-

Figure 6 Energy level scheme shows a (1 ⫹ 1′) resonance-enhanced multiphoton ion-

ization (REMPI) scheme leading to the production of a molecular ion, M⫹, from ground
state neutral M (| g〉), via the production of an excited-state M* (| a〉). Also indicated are
possible competitive spontaneous emission relaxation pathways for the resonantly excited
M* intermediate states with rate Aa .

Copyright © 2002 by Taylor & Francis

neltron, or microchannel plate (MCP) having a quantum efficiency (QE) also
close to unity. The ionizing laser need not be tunable as long as its energy hω2
⬎ IP ⫺ Ea , where IP is the ionization potential of the particle and Ea is the term
value of the excited state.
The pumping and relaxation scheme shown in Figure 6 can be readily ana-
lyzed using
dNa
⫽ (Ng ⫺ Na)σagn1 ⫺ NaσIn2 ⫺ AaNa (13)
dt
where Ng and Na are number densities in levels | g〉 and |a〉, respectively; σag is
the absorption cross section for the transition between levels |g〉 and | a〉 at fre-
quency ω1; σI is the ionization cross section out of level | a〉 at frequency ω2; n1
and n2 are the photon fluxes (in cm⫺2s⫺1); and Aa is the spontaneous relaxation
rate out of state |a〉. When Eq. (13) is solved for Na in the steady-state approxima-
tion, the ion signal rate S1 is then given by:
SI ⫽ Na ⋅ σI ⋅ n2 ⋅ Φ ⋅ QE ⋅ ∆V (14)
where, as before, ∆V is the interaction volume. Using the solution for Na from
Eq. (13), and identifying Ng ⋅ σag ⋅ n1 ⋅ ∆V as na, the number of photons absorbed
per second on the first transition, Eq. (14) can be rewritten as:
1
SI ⫽ na (15)
1 ⫹ [(σag ⋅ n1 ⫹ Aa)/(σI ⋅ n2)]
which shows that if σI ⋅ n2 ⬎⬎ σagn1 ⫹ Aa, nearly every absorbed photon can be
detected. It should be kept in mind, however, that ionization cross sections are
usually orders of magnitude smaller than transition cross sections. Therefore, the
intensity of the ionizing laser should be much larger than that of the exciting
laser, which can easily saturate the first step.
Recording the total ion yield as a function of ω1 will generate an excitation
spectrum, but mass selection prior to detection is more common. This additional
step becomes particularly important if there is more than one absorbing species
present in the jet or if the species has many naturally occurring isotopomers. In
this regard, a TOF mass spectrometer [32,33] offers several advantages that make
it the instrument of choice for jet spectroscopy.
A TOF mass spectrometer operates on the simple principle that an isoener-
getic packet of different ions will spatially separate according to their mass/
charge ratios (m/e) over a fixed distance. As a consequence, the mass of one ion
can be differentiated from another by measuring the time it takes to traverse that
fixed distance; that is, its time-of-flight. TOF instruments have almost unitary
transmission that is approximately independent of ion mass and mass resolution,

Copyright © 2002 by Taylor & Francis

and an infinite mass range. Their main deficiency is related to their ultimate mass-
resolving power. The two major limiting factors are the initial spread in the veloc-
ity of ions formed and the spatial spread of ions created in different volumes of
the ionization region. The former can be alleviated by using a jet with its flow
axis perpendicular to the flight tube axis. The velocity component along the flight
tube direction can be reduced further by skimming the free jet to make a low
divergent molecular beam.
The second problem can be minimized by operating the TOF mass spec-
trometer under the so-called Wiley-McLaren space focusing condition. Consider
the geometry of the linear TOF mass spectrometer shown schematically in Fig-
ure 7. It consists of three grids: a repeller plate, a ground plate, and an acceleration
plate. Space focusing can be understood by considering two ions of the same
mass initially with zero component of velocity along the flight tube direction
created at two different points S ⫾ δ. Although both ions enter the flight tube,
the ion at S ⫺ δ will actually gain more energy than the mass at S ⫹ δ, which
spends less time in the acceleration region. At some point down the flight tube,
therefore, the ion starting from S ⫺ δ will overtake the one starting from S ⫹
δ, even though the latter was initially ahead of the former. The optimum mass
resolution will be achieved if the detector is placed at this takeover position. In
practice, the flight tube distance is fixed and the acceleration and repeller plate
voltages are adjusted to obtain the optimum resolution. In a 1 m TOF mass spec-
trometer, where the separations between the repeller and acceleration plates and

Figure 7 Schematic diagram shows the geometry of a linear time-of-flight mass spec-
trometer needed to understand space-focussing. V1 and V2, voltages on the repeller plate
and acceleration grid, respectively.

Copyright © 2002 by Taylor & Francis

the ground plate are 2.16 cm and 2.41 cm, respectively, space focusing will be
achieved when V1 ⫽ ⫹360 V is applied to the repeller plate and V2 ⫽ ⫺2800
V is applied to the acceleration grid.
A low-duty cycle is also often cited as another limitation of TOF mass
spectrometers, because a new ion bunch can only be injected into the instrument
after the previous one has been completely detected. The use of relatively low
repetition rate laser systems (characteristically 10–20 Hz) essentially eliminates
this problem.

III. APPLICATIONS OF UV AND VUV LASERS

A. Spectroscopy
Although numerous examples exist, this chapter will use in part recent studies
of ion-pair state ↔ valence state transitions to illustrate the practical use of UV
and VUV LIF and TOF excitation spectra in jets. An ion-pair state is one that,
instead of producing two neutral fragments upon dissociation, diabatically yields
a pair of oppositely charged ions. Note that all molecules have ion-pair states in
numbers comparable to or even larger than their conventional valence states [34],
but specific examples will be drawn only from our studies of the group 12 (for-
merly IIB) metal monohalide radicals and the halogens.

1. Group 12 Monohalide Radicals

Unstable free radicals are some of the most challenging species to produce and
characterize spectroscopically. Yet their chemical significance becomes immedi-
ately apparent when one realizes that these highly reactive open shell molecules
are often the transient intermediates produced and consumed in the elementary
reaction steps of complex reaction mechanisms; for example, those associated
with combustion and atmospheric chemistry. As reviewed by Miller and co-work-
ers [35], a supersonic jet provides a near-requisite environment for cooling free
radicals and for minimizing their reactions, since by simple dilution all that they
will encounter are inert rare gas atoms.
The group 12 metal monohalide radical excimers, MX, M ⫽ Hg, Cd, Zn;
X ⫽ I, Br, Cl were first studied by Wieland over 60 years ago [36]. Interest in
these species became acute, however, when laser emission was found on their
ion-pair B 2⌺⫹ states → X 2 ⌺⫹ ground state transitions [37]. The mercuric mono-
halide radicals are particularly noteworthy because their coherent outputs lie in
the important blue–green spectral region. These laser systems are also attractive
because excited MX radicals can be produced by photolytically or electrically
dissociating an MX2 precursor, which subsequently reforms, resulting in a cycli-
cal production of the excited state [38].

Copyright © 2002 by Taylor & Francis

 MX radicals are analogues of the better-known and, in some instances,
commercially important rare gas monohalide excimers such as XeCl or KrF. Re-
placing Xe with its large ionization potential, IP ⫽ 97834.4 cm⫺1 [39], by Hg
(IP ⫽ 84184.1 cm⫺1 [39]), substantially decreases the separation between the
ion-pair state minimum and the ground state dissociation limit. The resultant
configurational mixing between B and X due to an avoided crossing between the
Coulomb branch of the ion-pair state and the repulsive ground state therefore
imparts a significant bound ionic character to the latter, which, for the most part,
is absent for the rare gas species. This mixing likewise decreases the binding
energies of the group 12 metal monohalide ion-pair states (⬇35000 cm⫺1) relative
to those for their rare gas monohalide counterparts (⬇44000 cm⫺1).
Due to the very different shapes of the potential energy curves involved,
the strongest Franck-Condon-allowed transitions originate from low v′-levels of
the HgX B-state and terminate on high vibrational levels of X 2⌺⫹. This expecta-
tion was quantitatively confirmed by Tellinghuisen and co-workers, who photo-
graphed emission spectra of single isotopomer samples excited in a mild Tesla
discharge [40]. They found, for example, that the potential energy curve of the
B 2⌺⫹ ion-pair state of HgBr dissociating to Hg⫹(2S1/2 ) ⫹ Br⫺(1S0) is very deep,
D′e ⬇ 39100 cm⫺1, due to the strong long-range Coulombic potential, V(R) ⬀ R⫺1,
binding the ion-pair together. Nevertheless, the long excited-state equilibrium
bond length (R′e ⬇ 3.2 Å) and small vibrational and rotational constants (ω′e ⬇
136 cm⫺1; B′e ⬇ 0.03 cm⫺1) observed are direct manifestations of its antibonding
nature, as predicted by simple molecular orbital theory. In converse fashion, the
X 2⌺⫹ ground state is considerably shallower (D″e ⬇ 5527 cm⫺1), even though
its equilibrium bond length is markedly shorter (by ⬇ 0.57 Å).
Experiments in which the unstable radical excimers are probed in absorp-
tion are rarer, and exceedingly more challenging. Franck-Condon considerations
dictate that transitions from lowest vibrational levels of the ground state will
probe the inner wing of the B-state potential energy curve in the UV, exciting
very high-lying closely spaced vibrational levels. Furthermore, many rovibronic
levels of the radicals are expected to be populated in a discharge. These factors,
combined with numerous naturally occurring isotopomers, conspire to yield near
continuum-like spectra that at best are difficult to interpret. This was demon-
strated by Eden and co-workers, who used transient absorption spectroscopy to
pump optically the B-state of HgBr radical from its ground state [41].
Our group found that these problems could be alleviated by generating
rotationally cold radicals in a corona-excited supersonic expansion (CESE) [42],
and using a tunable UV laser to excite LIF. In this way, we could obtain the first
precise ‘‘absorption’’ spectra involving the higher v′-levels of the HgX B-states.
Details about the experimental arrangement can be found elsewhere [43].
In brief, 1–2 mJ tunable UV radiation generated by frequency doubling the output
of a Nd:YAG-pumped dye laser was focussed below the pinhole of a quartz bulb

Copyright © 2002 by Taylor & Francis

nozzle. The nozzle was sealed into a flange mounted on a chamber evacuated to
⬍50 mtorr with a Roots blower/mechanical pump combination. HgX2 precursor
inside the nozzle was resistively heated to produce a vapor pressure above the
pinhole. The vapor was entrained in ⬇2 atm He and coexpanded in a continuous
supersonic jet. To conserve samples, the gas delivery could be pulsed by connect-
ing a solenoid valve outside the heater bulb nozzle inside the chamber.
HgX radicals were generated by exciting a discharge in the expanding gas.
Positive voltage (⬇400 V/10 mA) was applied to a stainless steel electrode run-
ning concentrically down the long axis of the bulb, while the chamber itself acted
as the grounded cathode. Undispersed fluorescence was collected, filtered, and
imaged onto the cathode of a photomultiplier. Care was taken not to image the
corona-discharge from both inside the nozzle and the brightest region of the jet.
Scattered light was reduced by temporally gating the data acquisition system ⬇
50 nsec after the peak of the scattered light.
Unlike the transient absorption spectra discussed above, the HgX B 2⌺⫹(v′)
← X2⌺⫹(v″) excitation spectra are highly structured due to rotational cooling
[43]. The (29,0) band of Hg35Cl shown in Figure 8, for example, exhibits well-
resolved peaks due to the naturally occurring isotopes of Hg. The intensity of

Figure 8 Close-up view of the (29,0) band of the Hg35Cl B 2⌺⫹ ← X 2⌺⫹ fluorescence
excitation spectrum shows the isotopic structure due to the numerous isotopes of Hg.

Copyright © 2002 by Taylor & Francis

each isotopomer peak within the band nicely reflects the normalized natural abun-
dance of the particular Hg atom involved relative to 202Hg (198Hg: 199Hg: 200Hg:
201
Hg: 202Hg: 204Hg ⫽ 0.335: 0.565: 0.776: 0.444: 1.00: 0.230) [44]. Isotopomers
containing 196Hg, however, were not detected for a (v′,0) Hg35Cl/Hg37Cl doublet,
since its normalized natural abundance is only 0.005.
The transition frequency of the Hg35Cl component for any (v′ ⬎⬎ 0,0) HgCl
band will always be larger than that for Hg37Cl due to simple mass scaling. Care
must, however, be exercised in the analysis of high vibrational levels of an ion-
pair state because vibrational isotope shifts can not only equal but often well
exceed the local vibrational spacings, ∆G′v ′⫹1/2. To a first approximation the vibra-
tional level, vo, where the isotope shift equals the vibrational spacings, can be
calculated from [45]:
ρi
vo ⫽ (16)
1 ⫺ ρi
where ρi is defined here as ρi ⫽ √µ(200Hg35Cl/µi) and µi is the reduced mass of
a particular HgCl isotopomer. The crossover vibrational level is calculated using
Eq. (16) to be vo ⬇ 42 for HgCl. As illustrated in Figure 9 the weaker Hg37Cl
doublet component will appear at a higher frequency than the nearest Hg35Cl for
v′ ⬍ 42, and at a lower frequency for v′ ⬎ 42. However, the vibrational quantum
number of the Hg37Cl band is actually one higher than its Hg35Cl.
Such crossovers in HgCl are expected for integer multiples of vo near v′
⫽ 84, 126, 168, and so on, with an additional unit increase in the relative number-
ing of the HgxCl bands per vo interval. The xHgX isotope shifts confirm the vo
value observed and, therefore, the absolute v′-numbering. Auxiliary information
of this type or the location of the electronic origin is usually required to establish
the crossover region being probed. This is because the standard deviation of ‘‘lo-
cal’’ least-squares fit for all choices of nvo ; n ⫽ 1, 2, . . . will be of similar
quality. Crossovers within a doublet component for adjacent HgCl isotopomers
are only expected to become noticeable for v′ ⬎ 2600, a spectral region far outside
the observed Franck-Condon envelope.
Although the HgCl spectra are rotationally cold, the frequency bandwidth
of the laser system was such that rotational structure could not be resolved. The
radicals produced in the CESE, however, were found to be vibrationally warm.
Thus, Franck-Condon minima in the hot-band vibrational progressions could be
used to establish B-state equilibrium bond lengths. B ← X transitions probe the
inner turning points of the excited state potential energy curve, where the high
v′ wave functions have their largest and ⬇ constant amplitudes. The intensity
minima therefore reflect the ground state wave function nodal structure superim-
posed on the electronic transition Franck-Condon factors.
Since a definite minimum in the v″ ⫽ 2 progression was found for HgCl
at (v′, v″) ⫽ (34,2), the B-state equilibrium bond length was established by first

Copyright © 2002 by Taylor & Francis

Figure 9 Overview of the HgCl B 2⌺⫹ ← X 2⌺⫹ fluorescence excitation spectrum in
the region of the first ‘‘crossover’’ point at (42, 0). Bands having the same vibrational
assignment but belonging to a radical with a different Cl isotope are connected by a diago-
nal line. The features labeled (a)–(e) are atomic transitions of He and Hg originating from
metastable levels. (i) B–X vibrational bands just to longer wavelengths of (42,0). In this
region the Hg35Cl-Hg37Cl isotope shifts are less than the local vibrational spacings. (ii)
B–X vibrational bands near (42,0). In this region the Hg35Cl–Hg37Cl isotope shifts are
approximately equal to the local vibrational spacings. (iii) B–X vibrational bands just to
shorter wavelengths of (42,0). In this region the Hg35Cl–Hg37Cl isotope shifts are greater
than the local vibrational spacings.

Copyright © 2002 by Taylor & Francis

constructing RKR-Morse potential curves [46] for both X 2⌺⫹ and B 2⌺⫹. Here
the derived vibrational constants were used in an RKR calculation [47] to deter-
mine the width of the potential as a function of v. The inner wings of the potentials
were then calculated using Morse functions that incorporated Re values from the
literature. Morse parameters were ascertained using the vibrational frequency and
dissociation energy for each electronic state. The outer turning points followed
immediately.
Vibrational wave functions were evaluated numerically [48] in both poten-
tials over the range of the observed levels and Franck-Condon factors derived.
A series of RKR-Morse/Franck-Condon factor calculations were then carried out
with Re(X) constrained, but Re(B) varied systematically until the predicted inten-
sity minima agreed with the experimental observations. The calculated positions
of the intensity minima were found to be very sensitive to the relative position
of the two potentials energy curves because, in absorption, transitions from v″
⫽ 2 probe an extremely steep portion of the B-state inner wing. For example, a
decrease of ⬇ 0.004 Å in Re(B) for HgCl dropped the minima location by ⬇
one quantum in v′. The best Re(B) for HgCl was found to be 2.960 Å, with an
estimated error of ⫾ 0.002 Å.

2. Diatomic Halogens
The ion-pair states of the diatomic halogens are arguably the best characterized
to date. A halogen ground state is found to be adequately described by the single
molecular orbital configuration: σ(g)2π(u)4π(g)4σ(u)0, where the exponents indicate the
number of electrons in each molecular orbital, and the g/u subscripts apply only
to the homonuclear molecules. By convention, only the exponents are listed and,
therefore, the ground state configuration is designated by 2440 [49]. There are
a total of 20 ion-pair states of an AB halogen. Half of that number can be orga-
nized into four tiers that dissociate diabatically to the A⫺(1S0) ⫹ B⫹(3P, 1D, 1S),
and, therefore, have separations that reflect the energy intervals of the positive
ion, B⫹. The remaining charge transfer states correlate with the much higher
energy A⫹ ⫹ B⫺ limit, provided the ionization potential, IP, of B ⬍ IP (A). The
20 states that dissociate to a common set of asymptotes for the homonuclear
species when A ⫽ B are delineated instead by inversion symmetry. Due to a
∆Ω ⫽ 0 propensity rule governing single-photon valence to ion-pair state excita-
tions [50], the most intense VUV spectral feature is the transition from the
⫹ ⫹
X0(g) halogen ground state to the first tier 0(u) ion-pair state having the strongly
antibonding 1441 configuration. After a long and often confusing history, it is
⫹
now widely accepted that the 0(u) state should be labeled E and D for the hetero-
nuclear and homonuclear halogens, respectively.
Like the group 12 monohalide radicals, the potential energy curve of the
1441 state of a given halogen (and all its other ion-pairs states) exhibits schizo-

Copyright © 2002 by Taylor & Francis

phrenic properties. It has a large dissociation energy relative to the ground state
(on the order of 4 eV), but a longer equilibrium bond length, and much smaller
vibrational and rotational energy level spacings. Spectroscopic difficulties arise
because Franck-Condon-allowed single-photon transitions probe the inner wing
of the 1441 ion-pair state potential energy curve, exciting very high, very closely
spaced vibrational levels in the VUV. Extensive room-temperature rotational con-
gestion and isotopic fine structure provide additional complications.
Many of these problems could be minimized by exciting a supersonically
cooled halogen with tunable, coherent, and monochromatic VUV radiation. De-
tailed analyses of the single-photon VUV spectra of IX, X ⫽ I [51], Br [52], Cl
[45], BrY, Y ⫽ Br [53], Cl [54], and Cl2 [55] at vibrational and, sometimes,
rotational precision have been completed by our group, in many instances, for
the first time. Initially, VUV LIF proved to be a successful detection scheme.
However, as our results for BrCl demonstrate, TOF mass detection provides an
alternative and an often more powerful method for unravelling complex excita-
tion spectra.
Until recently, the only quantitative information about the excited E0⫹
charge transfer state of BrCl was provided by single-isotopomer emission spectra
of the E0⫹ –B0⫹ band system excited in a mild Tesla discharge. That analysis was
based only on transitions involving vE ⱕ 2 [56]. Low-resolution VUV synchrotron
absorption and fluorescence excitation spectra of BrCl have also been reported
[57]. Near 145 nm both Rydberg state ← ground state and the E–X band systems
are observed in absorption, while only the ion-pair state is seen in fluorescence
excitation. Like the other interhalogens, its overall intensity was found to be
strongly modulated as a function of frequency, a phenomenon attributed to inter-
actions between E0⫹ and isoenergetic Rydberg states.
Details about our apparatus shown schematically in Figure 10, which com-
bines VUV LIF and TOF mass detection, have been published [25,54]. The sys-
tem was built to allow XUV generation below 105 nm by four-wave mixing in a
pulsed supersonic jet of Xe or Kr gas located in the first of two vacuum chambers.
Differential pumping effectively eliminates all window material in the path of
the XUV beam, allowing the generated light to enter the second chamber for
experiments. The entire vacuum system is maintained at a base pressure ⱕ 10⫺6
torr using diffusion pumps.
Since BrCl excitation wavelengths lie between 146.8 nm and 143.3 nm,
those four-wave mixing experiments were done not in a free jet but in ⬇ 5 torr
Kr gas housed in a static cell mounted to the front end of the first vacuum cham-
ber. BrCl molecules, formed by mixing Br2 and excess Cl2 prior to jet expansion
in He at a total stagnation pressure of 34 psia, were excited by one photon of
the VUV laser and subsequently ionized with a second photon from the funda-
mental beam in a (1 ⫹ 1′) REMPI process. The resultant ions were mass dispersed
in a linear TOF mass spectrometer (m/∆m ⱖ 500). Valve opening was synchro-

Copyright © 2002 by Taylor & Francis

Figure 10 Schematic diagram of the experimental apparatus used to record VUV and
XUV laser/time-of-flight mass spectra of jet-cooled molecules. Instruments indicated by
dashed boxes are optional. Their use depends on the range of VUV or XUV wavelengths
being generated.

nized to the arrival of the laser pulse inside the ionization volume with homemade
electronics. The timing was chosen to provide as many rotational lines as possible
without strong overlapping between vibrational bands.
Simultaneous VUV LIF detection with a solar-blind photomultiplier tube
mounted 180 degrees to the axis of the flight tube was accomplished by con-
structing the repeller plate from a fine mesh with ⬇90% optical transmission.
Scattered light into the photocathode was reduced using light baffles placed
around the phototube and the path of the laser beams. In other experiments that
will not be considered here, the photomultiplier was replaced with a dispersive
photoelectron spectrometer to measure the kinetic energies of electrons formed
by REMPI.

Copyright © 2002 by Taylor & Francis

Figure 11 A small portion of the VUV laser excitation spectrum of BrCl near 145.4
nm. The strongest vibrational band of the E0⫹ ← X0⫹ transition for the 79Br35Cl isotopomer
in this spectral region is assigned to (182,0). (a) VUV LIF excitation spectrum of a natu-
rally abundant isotopic sample of BrCl. (b) VUV laser/time-of-flight (TOF) mass spectrum
of the 79Br35Cl isotopomer. (c) VUV laser/TOF mass spectrum of the 81Br35Cl/ 79Br37Cl
isotopomer pair (d) VUV laser/TOF mass spectrum of the 81Br37Cl isotopomer. Additional
excitation spectra obtained by monitoring 79Br and 81Br atomic fragments produced by
predissociation confirm that the bands in trace (c) are due to 81Br35Cl.

Small segments of the VUV BrCl E0⫹ –X0⫹ LIF and TOF excitation spectra
near 145.4 nm are presented in Figure 11 [54]. Four naturally occurring isotopom-
ers contribute to the fluorescence excitation spectrum (Fig. 11a): 79Br35Cl (natural
abundance ⫽ 38.41%), 81Br35Cl (37.36%), 79Br35Cl (12.28%), and 81Br37Cl
(11.95%). Even though rotationally resolved vibronic bands are readily discerned,
their isotopomer identities are difficult to establish based on natural abundances
due to the strong perturbative intensity anomalies. The TOF spectra shown in
Figure 11 b–d are considerably simpler, although large fluctuations in the band
intensities are still evident in the TOF detection mode. Even though the mass
spectrometer could not resolve 79Br37Cl and 81Br35Cl, all the bands in Figure 11c
were subsequently assigned to 81Br35Cl by separately monitoring 79Br⫹ and 81Br⫹

Copyright © 2002 by Taylor & Francis

daughter fragment ions produced by excited state predissociation followed by
ionization, all within the VUV laser pulse duration.
The excited state vibrational numbering was initially estimated by fitting
the isotopic transition term values, νi , to the following mass-reduced mixed near-
dissociation expansion (NDE) appropriate for a R⫺1 long-range potential [58]:

νi ⫽ T′e ⫺ D′e ⫺
Ry
X2 冢a
1⫹ 3
X 冣 (17)

where X is defined as [(vD ⫹ 1/2) ⫺ ρi (v′ ⫹ 1/2)], ρi ⫽ √µ(79Br35Cl)/µi, where µi

is the reduced mass of a particular BrCl isotopomer, T e′ and D′e are the E-state
electronic origin and dissociation energy, respectively, and Ry is the Rydberg
constant for 79Br35Cl. The second and third terms in Eq. (17) represent the correct
limiting behavior for the term values in a Coulomb potential, while the fourth
term proportional to a3 is empirical and accounts for the fact that the vibrational
levels observed in emission lie well below the dissociation limit. It is not coinci-
dental that the limiting form of Eq. (17) is very similar to the Rydberg term
expression for H-like systems:
Ry
Tn ⫽ IP ⫺ (18)
(n ⫺ δ)2
where Tn is the electronic state term value, Ry, is the Rydberg constant, and n
is the principal quantum number. The quantum defect δ, which characterizes the
deviation of the atom from pure hydrogen behavior, decreases with increasing
orbital angular momentum ᐉ. In NDE theory vD is usually interpreted as the nonin-
teger vibrational quantum number of the highest level supported by the potential.
However, just as there are an infinite number of electronic states in the H-atom,
an ion-pair state potential energy curve supports an infinite number of vibrational
levels below its dissociation threshold. Therefore, |vD | can be viewed instead as
a vibrational quantum defect. Its magnitude is equal to the quantum number n
of a level in a fictional Coulomb potential having the same outer wing as the
molecular curve and the same term value as v′ ⫽ 0 of the ion-pair state. We
found that the high vibrational ion-pair state term values of the iodine containing
interhalogens could often be fitted to an NDE with fewer parameters than a Dun-
ham series.
Calculations based on a fit to Eq. (17) established that approximately 120
(v′, 0) isotopic bands with 179 ⱕ v′ ⱕ 213 are located in the excitation wavelength
region. Appreciable Franck-Condon factors for hot band transitions from v″ ⫽ 1
were also predicted at the short wavelength end of the experimental observations,
and for (v′, v″ ⫽ 2) transitions at the longer wavelengths. The lowest-energy
79
Br35Cl band from these calculations could be assigned to (180,0). Ultimately,
the best v′-numbering chosen was the one that minimized the standard deviation,

Copyright © 2002 by Taylor & Francis

σ, of a least-squares fit of the isotopic band origins and transition band head
transition wave numbers to a mass-reduced Dunham expansion. Then the lowest
79
Br35Cl energy band was assigned to (172,0), corresponding to a difference of
⫺8 from the estimate provided by the NDE calculation. This difference, however,
is not considered serious in light of the extensive frequency and intensity pertur-
bations observed. Furthermore, precise excited-state rotational constants for most
of the observed bands were also derived for the first time due to the monochroma-
ticity of the VUV laser used.
VUV spectroscopy of the 2 1⌺ u⫹(v′) ← X 1⌺ g⫹(v″ ⫽ 0) transition of elemen-
tal Cl2 provides an excellent example of LIF and TOF detection together solving
a difficult problem that can only be partially understood using one or the other
technique. Until recently, the 2 1⌺ u⫹ –X 1⌺ g⫹ band system defied quantitative analy-
sis. That situation has now been significantly rectified by the judicious application
of theory [59] and VUV laser experimentation [55]. In diabatical terms, the inner
wing of the 2-state is formed by a strong avoided crossing between the inner
wall of the third-tier ion-pair state that dissociates to Cl⫹(1Dg) ⫹ Cl⫺(1Sg) and
the outer wing of the 4pπ Rydberg state. Double-well formation results from an
additional perturbation at longer bond lengths, with the fourth tier ion-pair state
dissociating to Cl⫹(1Sg) ⫹ Cl⫺(1Sg). Due to the curvature of the potential near its
predicted minimum at R′e ⬇ 2.1 Å, the vibrational frequency of excited state,
ω′e was expected to be larger than that for a pure Rydberg state (⬇ 628–665
cm⫺1), valence state (⬇ 260 cm⫺1), or the ground state (⬇560 cm⫺1). VUV laser/
TOF mass spectra confirm this expectation (Fig. 12, ω′e ⬇ 760 cm⫺1). Further-
more, it can be concluded from measured vibrational isotope shifts that the fea-
tures previously attributed to the (v′, v″) ⫽ (0,0) band near 78125 cm⫺1 actually
belong to the (1,0) band.
It is interesting that the intensity distributions in the isotopically resolved
TOF spectra are anomalous due to what is believed to be an additional resonance
enhancement in the second step of the (1 ⫹ 1′) REMPI process by a superexcited
state embedded in the first ionization continuum of the molecule. As a conse-
quence, rotational analyses were carried out in large part by computer simulating
the LIF spectra. The results for 35Cl2 2-X (1,0) transition are shown in Figure 13.
An intensity alternation of 5:3 evident in the P-branch for the odd and even J-
levels, respectively, in Figure 13a, arises because the nuclear spin(I) of 35Cl ⫽ 3/2.
However, the R-branch was unresolved and therefore the absolute J-numbering
could not be established by ground state combination differences. Due to the
limited number of J-levels populated at room temperature, the J-numbering could
nevertheless be deduced almost by inspection to within ⫾2. The best J-numbering
used to determine constants was that producing a minimum least-squares fit stan-
dard deviation and the best simulation (δJ ⫽ 0; Fig. 13b). Changing the J-num-
bering by δJ ⫽ ⫹2 (Fig. 13 c) or by δJ ⫽ ⫺2 (Fig. 13d), as required by the
intensity alternation, produced simulations in much poorer agreement with the

Copyright © 2002 by Taylor & Francis

Figure 12 VUV laser excitation spectra of the 2 1⌺u⫹(v′ ⫽ 1) ← X 1⌺g⫹(v″ ⫽ 0) transition
of Cl2 (a) VUV laser-induced fluorescence excitation spectrum of a naturally abundant
isotopic sample of Cl2. VUV laser/TOF mass spectra for the (b) 35Cl2, (c) 35Cl37Cl, and
(d) 37Cl2 isotopomers are shown.

experimental observations. Different simulations were carried out over exactly

the same frequency range but using the new band origins and rotational B′-con-
stants derived from refitting the observed transition frequencies to a Dunham
expansion. As a consequence, a J-numbering was not considered acceptable if
the resultant simulation could only match the observed spectrum by frequency
shifting it wholesale to do so. The resultant Rv′⫽1 ⫽ 2.0991(46) Å is in very good
agreement with theoretical expectations.
Yamanouchi and Tsuchiya have reviewed the steadily growing VUV laser
spectroscopy literature up to 1994 [60]. Table 2 provides a revised catalogue of
atomic and molecular systems studied with VUV lasers since 1990. Even a cur-
sory inspection of our new list and the associated references shows that such
short-wavelength sources are now not only being used to obtain new spectra
and/or to rerecord known spectra with a higher-frequency resolution but also
extensively in reaction dynamics, for coherent control, and for ion spectroscopy
by pulsed field ionization methods.

Copyright © 2002 by Taylor & Francis

Figure 13 (a) A small portion of the VUV laser-induced fluorescence excitation spec-
trum assigned to the 2 1⌺u⫹(v′ ⫽ 1) ← X 1⌺g⫹(v″ ⫽ 0) transition of 35 Cl2. (b) Best simulation,
δJ ⫽ 0. (c) Simulation obtained by changing the best set {J} in the P-branch by δJ ⫽
⫹2. (d) Simulation obtained by changing the best set {J} in the P-branch by δJ ⫽ ⫺2.

B. Analytical Chemistry
Although lasers are becoming routine fixtures in many analytical chemistry labs
[61], the specific combination of VUV lasers and mass spectrometry provides
unique opportunities for organic molecule detection that are unavailable to other
experimental methods based on either longer wavelength coherent sources or
nonoptical strategies. For example, unambiguous molecular identification by con-

Copyright © 2002 by Taylor & Francis

Table 2 VUV and XUV Laser Studies of Atomic and Molecular Systems

Molecular or Spectroscopic
atomic species Electronic states method References
H(D) np Rydberg states, n ⫽ 3, 4, 7, 8 EIT 86
np Rydberg states, n ⫽ 4 ⬃ 8 EIT 87
3p (← 1s); 3p (← 2s) EIT 88–93
2p Rydberg states EIT 94–98
2p 2P (← 1s 2S) LIF 99–123
2P 2P (← 1s 2S) PAS 124
2p 2P (← 1s 2S) PTS 125–142
2p 2P (← 1s 2S) HRTOF 143–152
2p 2P (← 1s 2S) DSTOF 153, 154
H⫺ Near n ⫽ 2 threshold DS 155–157
H2, D2 H H 1⌺ g⫹ REMPI 158–160
B 1⌺ u⫹, B′ 1⌺ u⫹, C 1Πu , D 1Πu REMPI 161–164
I′ 1Πg , EF 1⌺ g⫹, GH 1⌺ g⫹ REMPI 165
EF 1⌺ g⫹, GK 1⌺ g⫹, H 1⌺ g⫹, I 1Πg , J 1∆g OODR 166
Singlet gerade s- and d-Rydberg states with J ⫽ 0, 1, 2 OODR 167
B′ 1⌺ u⫹, B 1⌺ u⫹, C 1Πu OODR 168–170
B 1⌺ u⫹, C 1Πu PTS 171
H⫹(1s0) ⫹ H⫺(1s2) TIPPS 172
Rydberg states: Q(1) np π, n ⫽ 5, 7, 14 ⬃ 22 SS 173
R(1)np1, n ⫽ 15 ⬃ 21
P(1)npσ, n ⫽ 8, 20 ⬃ 21
R(1)np3, n ⫽ 12 ⬃ 15
R(0)np2, n ⫽ 14, 15
R(0)np0, n ⫽ 14
Rydberg states, n ⫽ 30–49 PI 174
B′ 1⌺ u⫹, B1 ⌺ u⫹ LIF, PHOFEX 175–178

Copyright © 2002 by Taylor & Francis

Table 2 Continued

Molecular or Spectroscopic
atomic species Electronic states method References
H2⫹ X 2⌺ g⫹ PFI-ZEKE 179,180
HD H H 1⌺ g⫹, B″ B 1⌺ g⫹ REMPI 158, 159, 181
High gerade Rydberg states close to the first few rotational levels of REMPI 182
the ground vibronic state of HD⫹
B′ 1⌺⫹, B 1⌺⫹, C 1Π OODR 170
Rydberg states near second dissociation limit at ⬃84 nm LIF 183
He 2 1P REMPI 159
21S TPDFS 184
C 2p2 1D, 2p2 3P LIF 185, 186
CH3⫹ X̃ 1A1′ PFI-ZEKE 187
O 3s 1Do (← 2p 1D), 3s 3So (← 2p 3P2,1,0) LIF, PHOFEX 188–208
CH4, CD4 Ã 1T2 PEX 141
PEX 108
CH4⫹ X̃ 2B1 PFI-ZEKE 209
OH⫹, OD⫹ X 3⌺⫺ PFI-ZEKE 210
NH3 B̃ 1E″ LIF 211
NH3⫹ X̃ 2A″2 PFI-ZEKE 212
H2O C̃ 1B1 OODR 213
H2O, D2O B̃1A1 PEX, PHOFEX 142, 214
nd ← 1 b1 Rydberg series, n ⱖ 6; n ⫽ 6–11, converging to H2O⫹ REMPI 215
(000) and (100) vibrational states, respectively
H2O⫹, D2O⫹ X̃ 2B1 PFI-ZEKE 216
H2O⫹, H2S⫹ X̃ 2B1 PFI-ZEKE 217–219
CD2H2 Lowest energy 1T2 state PEX 129
CD2H2⫹ X̃ 2B2 PFI-ZEKE 220

Copyright © 2002 by Taylor & Francis

Ne 3s[3/2]1, 3s′[3/2]1 REMPI 221
HF⫹ X 2Π1/2, 3/2 PFI-ZEKE 222
ND3 B̃ 1E″, D̃ 1E′ LIF, REMPI 223
ND4⫹ X̃ 1A1 PFI-ZEKE, MATI 224
C2H2, C2D2 3R′2 10, 3R⵮2 10, 4R2 00, 4R2 10, 4R″2 00, 4R″2 10, E2 10, E4 20, E5 20 PEX 124, 143, 153, 154
HCN Ã 1A″ LIF 225
1
Π Rydberg state (. . . 1π33sσ1 dominant configuration) PEX 135, 138
HCN⫹ X̃ 2Π, Ã 2⌺⫹ PFI-ZEKE 226
CO X 1⌺⫹ LIF 185, 227
B 1⌺⫹ REMPI, LIF 163, 228–230
CC 231, 232
a′ 3⌺⫹ EX 233
A 1Π LIF 234–244
AS 245
C 1⌺⫹ REMPI 246
E 1Π REMPI 247
W 1Π, L 1Π, K 1⌺⫹, L′ 1Π REMPI 248–253
A 1Π, L 1Π FWM 254–255
j 3⌺⫹, a 3Π, k 3Π LRF 256–258
npσ Rydberg states, n ⫽ 5 ⬃ 8
ndσ, δ Rydberg states, n ⫽ 4 ⬃ 9
CO⫹ X 2⌺⫹ PFI-ZEKE 259
A 2Π3/2,1/2 PFI-ZEKE 260
N2 b′ 1⌺ u⫹, b 1Πu , o 1Πu REMPI 261, 262
c′4 1⌺ u⫹, c3 1Πu REMPI 263
N2⫹ X 2⌺ g⫹, A 2Πu , B 2⌺ u⫹ PFI-ZEKE 259, 264–267
HCO, DCO Ã 2A″ PHOFEX, LIF 268–270
HCO, DCO X̃ 2A′ LIF 237, 238

Copyright © 2002 by Taylor & Francis

Table 2 Continued

Molecular or Spectroscopic
atomic species Electronic states method References
NO B 2Π3/2, E 2⌺⫹, H 2⌺⫹, H′ 2Π, O 2⌺⫹, O′ 2Π FWM 271–274
ns Rydberg states, n ⫽ 7, 9⬃12 ARPES 275
nf Rydberg states, n ⫽ 5, 8⬃11, 13
NO⫹ a 3⌺⫹ PFI-ZEKE 276
X 1⌺⫹ PFI-ZEKE 277, 278
1
S D, 3P DS, LIF 279–283
3p3ns 3So, n ⫽ 5–7; 3p3np 3P0,1,2, n ⫽ 4,5 LIF 284
3p3nd 3Do n ⫽ 4,5
O2 O⫹ ⫹ O⫺(2P3/2, 1/2) TIPPS 285
B 3 ⌺u⫺, C′3Πu AS 286
f′ 1⌺ u⫹, f 1⌺ u⫹, j 1⌺ u⫹ AS 287
O2⫹ X 2Π1/2g,3/2g , a 4Πu PFI-ZEKE 218, 288–290
SiH4 1
T2 states (1a212a211t623a212t524p, 3d, 5s σ* (t2) configurations) PEX 291
H2S CC 292
HS⫹ ⫹ H⫺, HS⫺ ⫹ H⫹ TIPPS 293
H2S, D2S B̃ 1A1 PEX 134, 138, 140
Cl 4s 2P3/2, 1/2 ← (3p 2P3/2,1/2) LIF 111, 294
HCl, DCl Rydberg states in the energy range between their 2Π3/2 and 2Π1/2 ionic AI 295
thresholds
j 3⌺⫺(0⫹), H 1⌺ ⫹(0⫹), m 3Π(1) CC 232, 296
H⫹ ⫹ Cl⫺(1S0) TIPPS 297
HCl⫹ X̃ 2Π3/2,1/2 PFI-ZEKE 298
F2 C 1⌺ u⫹, H 1Πu REMPI 299

Copyright © 2002 by Taylor & Francis

Ar 4s′[1/2]1 REMPI 300
Rydberg states with n ⱖ 27 located below the 2P3/2 ionic threshold of PFI-ZEKE 301
Ar⫹.
Rydberg states with n ⫽ 60 ⬃ 200 ODDR, PFI-ZEKE 302
Ar⫹ 3p5 2P1/2, 3/2 PFI-ZEKE 174, 303, 304
CO2 3pπu 3⌺ u⫺ Rydberg states AS 305
CO2⫹ X̃ 2Π3/2g, 1/2g PFI-ZEKE 174, 306, 307
N2O⫹ Ã 2⌺⫹ PFI-ZEKE 308
X̃ 2Π PFI-ZEKE 309
CH3SH⫹, PFI-ZEKE 310, 311
CH3CH2SH⫹
CH3SCH3⫹
ClO C 2⌺⫺ LIF 199, 312–314
OCS 2 1⌺⫹ PHOFEX 315–317
Zn 3d94s24p 1Po1, 3d94s24p 3Do1 EIT 318
Cl2 1 1⌺ u⫹, 2 1⌺ u⫹, 2 3Πu LIF, REMPI 319, 55
CS2 Rydberg states [2Π1/2] npσu , n ⫽ 15 ⬃ 31 PI 320
[2Π1/2] npπu , n ⫽ 17 ⬃ 22
[2Π1/2] nfu , n ⫽ 14 ⬃ 25
CS2⫹ X̃ 2Π1/2,3/2 PFI-ZEKE 320
CH3PO2 PI 321
Ar2⫹ A2 ⌺1/2u
⫹
, C 2Π1/2u , C 2Π3/2u PFI-ZEKE 322, 323
HBr⫹ A2 ⌺ ⫹ PFI-ZEKE 324
Kr 4p55p[1/2]0, 4p55s[3/2]o1 EIT 325–327
Pb 6p2 3P0, 6p2 3P2, 6p7s 3P1, 6p9s 3P1 EIT 328
CH3OPO2 PI 321
CH2Cl2O PI 329
BrCl E 0⫹ LIF, REMPI 330

Copyright © 2002 by Taylor & Francis

Table 2 Continued

Molecular or Spectroscopic
atomic species Electronic states method References
HI Rydberg states converging to the 2Π1/2 ionic threshold of HI⫹ PI, ARPES 331, 332
ns, np, nd, nf, ng Rydberg states, n ⫽ 6, 7 ARPES 333
Q[3Π(1, 0⫹), 1Π(1)] DS 334
HI, DI 5sσ, 5dπ, 5dδ CC 335–340
CH3I CC 341
XeRg (Rg ⫽ Ne, Rg ⫹ Xe* (6s[3/2]o1, 6s′[1/2]0) LIF, REMPI 342–345
Ar, Kr)
XeRg (Rg ⫽ Ar, Rg ⫹ Xe*5d[3/2]o1 REMPI 346
Kr)
Kr2 Kr ⫹ Kr*(4d, 5p′, 6s) REMPI 347
Xe2 Xe ⫹ Xe*(5d, 6p, 6s′) REMPI 348, 349
Xe ⫹ Xe*(5d, 6p) PES 350
B′ 1u LIF 351
Xe ⫹ Xe*(5d[3/2]1, 5d[5/2]3, 7s[3/2]1, 7s[3/2]2) LIF 352
Xe2⫹ A 2⌺1/2u
⫹
, B2Π3/2g , C 2Π3/2u PFI-ZEKE 353
KrN, N ⫽ 2 near Kr* 5s 3P2 LIF 354
⫺2000
Studies are listed in approximate order of increasing mass; carried out from 1990 to date. AI, autoionization spectroscopy; AS, absorption spectroscopy;
ARPES, angle-resolved photoelectron spectroscopy; CC, coherent control; DS, Doppler spectroscopy; DSTOF, Doppler-selected time-of-flight spectrometry;
EIT, electromagnetically induced transparency; EX, excitation spectroscopy; FWM, four-wave mixing spectroscopy; HRTOFS, high-n Rydberg time-of-
flight spectrometry; LIF, laser-induced fluorescence spectroscopy; LRF, laser-reduced fluorescence spectroscopy; MATI, mass-analyzed threshold ionization
spectroscopy; OODR, Optical–optical double-resonance spectroscopy; PAS, photofragment action spectroscopy; PES, photoelectron spectroscopy; PEX,
photoexcitation spectroscopy; PFI-ZEKE, pulsed field ionization–zero kinetic energy photoelectron spectroscopy; PHOFEX, photofragment excitation spec-
troscopy; PI, photoionization spectroscopy; PTS, photofragment transnational spectroscopy; REMPI, resonance-enhanced multiphoton ionization spectros-
copy; SS, stark spectroscopy; TIPPS, threshold ion-pair production spectroscopy; TPDFS, two-photon Doppler-free spectroscopy.

Copyright © 2002 by Taylor & Francis

ventional methods such as electron ionization (EI) [62], chemical ionization (CI)
[63], and fast atom bombardment (FAB) [64] very often requires understanding
the complex fragmentation patterns that depend on the thermal energy in the
parent species, as well the excitation and ionization methods used. Furthermore,
ion yields from surface-adsorbed molecules using secondary ion mass spectrome-
try (SIMS) [65] and matrix-assisted laser desorption and ionization (MALDI)
[66] are strongly influenced by solid matrix effects. It should be appreciated,
however, that the number of neutral molecules and fragments produced by SIMS
and MALDI are expected to be orders of magnitude larger than the direct ion
yields. A significant enhancement in detection sensitivity could consequently be
achieved if these neutral species were postionized following thermal, particle-
induced, or laser-induced desorption. The combination of postionization follow-
ing surface desorption and TOF mass spectrometry has been termed ‘‘surface
analysis by laser ionization’’ (SALI) [67,68]. Its inherent value is that it decouples
surface desorption from postionization, thereby allowing the two steps to be inde-
pendently optimized.
Both strengths and weaknesses emerge when REMPI postionization is
compared to a nonresonant single-photon SALI process. REMPI has been shown
to be particularly successful for elemental analysis, and different schemes have
now been devised for almost every atom in the periodic table [69]. This approach
is highly sensitive, selective, and as noted above, uniformly efficient since a fo-
cussed laser can ionize almost every species in its focal volume with relative
ease. Nevertheless, surface composition measurements can be limited by large
ion signal variations due to laser shot-to-shot fluctuations.
Unlike elemental analysis, however, few good things can be said about the
application of REMPI to organic molecule detection. First, multiphoton absorp-
tion tends to result in facile and virtually unavoidable nonspecific fragmentation
[70]. Furthermore, since there are no guarantees that every organic molecule will
absorb a given UV/visible wavelength, electronic structures and oscillator
strengths for each individual molecule must be known. A useful illustration of
this problem is the facile detection of polycyclic aromatic hydrocarbons (PAHs)
generated by ionizing cigarette smoke at 266 nm [71]. Due to resonance enhance-
ment at that fundamental wavelength, PAH peaks dominate the resultant mass
spectrum even though they make up only a small percentage of the total mass
of the chemicals present. REMPI fundamentally yields the opposite of a structur-
ally ideal mass spectrum that has a single peak at the parent ion mass.
Nonresonant ionization with VUV lasers minimizes many of these prob-
lems. Prior spectroscopic knowledge of the molecule is not required, and quantita-
tive measurements become possible since single-photon ionization cross sections
tend to be more uniform from molecule to molecule than those for REMPI. Most
significant is that the first ionization potential of most organic molecules lies in
the 7–13 eV range and, therefore, VUV lasers sources can be viewed as a ‘‘uni-

Copyright © 2002 by Taylor & Francis

Table 3 Molecular Systems Detected by Nonresonant VUV Ionization

Compounds Methods References

Automotive exhaust (oxygenates and aromatics) and gasoline additives (methyl t-butyl ether and t- SPI-ITMS 355
amyl methyl ether)
Methyl radicals (from gas phase reactions of CO⫹ and N2O⫹ with CH4) SPI-TOFMS 356
C18 alkyl siloxane SAM on silicon; 1 : 1 mixture of polystyrene and p-fluoropolystyrene; pol- LD ⫹ SPI-ITMS 357
ystrene; glutathione tripeptide
Biotin; biotinylated SAM on gold LD ⫹ SPI-TOFMS 358
Thiolate, disulfide, and sulfide SAM on gold LD ⫹ SPI-TOFMS 359
Straight-chain alkanethilates SAM on gold LD ⫹ SPI-TOFMS 360
SiCl and SiCl2 etch products from the thermal etching of Si(100) surface with Cl2 SPI-TOFMS 361
Gramicidin S, (cyclic decapeptide); gramicidin D (linear peptide); fullerenes SPI-TOFMS 362
Fullerenes SPI-TOFMS; LD ⫹ 363, 364
SPI ⫹ FTMS
Rubber vulcanizates LD ⫹ SPI-TOFMS 364, 365
Di- and tripeptides containing glycine, alanine, leucine, and proline SPI-TOFMS 366
Tripeptide gylcine-alanine-leucine SPI-TOFMS 367
Linear, branched, and cyclic hexanes SPI-TOFMS 368
Aliphatic compounds: n-alkanes, alkenes, ketones, carboxylic acids, ethers, aldehyde, and amines SPI-TOFMS 369
Poly-methyl methacrylate, poly-ethylene glycol, polystyrene SALI 370
2,3,7,8-tetrachlorodibenzo-p-dioxin; 7-methylguanine LD ⫹ SPI-TOFMS 371
Teflon, Nylon 66 SALI 372
1,1,1-Trideuterio-2,5-dithiahex-3-yne IRMPD ⫹ SPI- 373
TOFMS

Copyright © 2002 by Taylor & Francis

Acetic anhydride IRMPD 374
⫹SPI-QMS
Disilane (Si2H6) SPI-TOFMS 375
Methanol dimer (CD3OH)2 SPI-TOFMS 376
Propyne (C3H6), allene (C3H4) photodissociation products SPI-TOFMS 377
Silicon LD ⫹ SPI-QMS 378
(C3 –C8) iodoalkanes PD ⫹ SPI-TOFMS 379
(C3 –C8) alkanethiols PD ⫹ SPI-TOFMS 380
Copper hexafluoroacetylacetonate (Cu(hfac)2) PD ⫹ SPI-TOFMS 381
(η5-Cyclopentadienyl)-Fe(CO)2R and (η5-Indenyl)-Fe(CO)2R (R ⫽ alkyl, aryl) PD ⫹ SPI-TOFMS 382
n-R-amines (R ⫽ heptyl, nonyl, undecyl, dipentyl) MALD ⫹ SPI- 383
TOFMS
(C5 –C8) Branched alkenes and dienes PD ⫹ SPI-TOFMS 384
Diacetylene (C4H2) and photochemical products PD ⫹ SPI-TOFMS 385–387
n ⫽ 30, 38, 40, 50, 60 alkanes and pentaerythritol tetraester oils SPI-TOFMS 388
Nitrobenzene (C6H5NO2) PD ⫹ SPI-TOFMS 389
As2, As4, Ga SPI-TOFMS 390
(C5 –C14) n-alkenes PD ⫹ SPI-TOFMS 391
1,2,4-octene PD ⫹ SPI-TOFMS 392
Ethylacetylene pyrolysis ⫹ SPI- 393
TOFMS
Tert-butyl nitrite flash pyrolysis ⫹ 394
SPI-TOFMS
Alkyl halides: XY, X ⫽ CH3, C2H5; Y ⫽ Cl, Br IPIS 395

FTMS, Fourier transform mass spectrometry; IPIS, ion photofragment imaging spectroscopy; IRMPD, infrared multiphoton dissociation; LD, laser desorption;
PD, photodissociation; SALI, surface analysis by laser ionization; SAM, self-assembled monolayers; SPI-ITMS, single-photon ionization–ion trap mass
spectrometry; SPI-TOFMS, single-photon ionization–time-of-flight mass spectrometry; SPI-QMS, single-photon ionization–quadrupole mass spectrometry.

Copyright © 2002 by Taylor & Francis

versal detector’’ [72]. This burgeoning area of analytical chemistry has recently
been reviewed [73].
Our preliminary examination of derivatized organic collectors used in min-
eral sulfide processing can demonstrate the virtues of soft nonresonant VUV laser
ionization [74]. Table 3 lists other organic systems probed in a similar way
(mainly but not exclusively with 118 nm light generated by nonresonant tripling
of the 355 nm output of a Nd :YAG laser in Xe).
Valuable trace metals are most commonly separated and concentrated from
crude ore by flotation [75]. Mineral grains (size ⱕ 150 µm), produced by crushing
and grinding, are rendered hydrophobic in aqueous solution by the addition of
organic molecules called collectors. The coated grains attach themselves to air
bubbles that float to the solution surface where they are removed and analyzed,
usually by mass spectrometry. As noted above, standard methods such as FAB,
CI, EI, and SIMS almost always exhibit numerous fragmentation peaks. It there-
fore becomes notoriously difficult to correlate a particular fragment with a given
parent molecule in mixtures of like compounds. This problem is not insignificant.
Mining companies will often empirically add many collector molecules with dif-
ferent relative concentrations to their flotation circuits to enhance the collection
efficiency. Given that billions of tons of ore are processed yearly by flotation,
any technique that can help directly to optimize this process could result in a
significant cost savings.
One of the most important classes of organic collectors are the alkali salts
of O-alkyl dithiocarbonates, ROC(S)S⫺, commonly known as xanthates [76].
Since the organic R-group is a short alkyl chain of 2–5 carbon atoms, xanthates
will selectively adsorb to the metal sulfide because either there is no adsorption
at the nonsulfide mineral/water interface or, if there is adsorption, the organic
groups are too short to impart a hydrophobic character to the grain. Selectivity
between different xanthate collectors can often only be achieved by the addition
of separate chemicals, called depressants or activation modifiers [77], and/or by
changes in solution conditions (pH, for example). For this reason, several collec-
tors are often added to flotation circuit, and the conditions changed empirically
to achieve the best results.
Xanthate salts have exceedingly low vapor pressures. Therefore, as proof-
of-principle, volatile xanthate ester derivatives [78], ROC(S)SR′, were ionized
using VUV wavelengths near 129.7 nm generated by four-wave difference mix-
ing in Kr. Here R and R′ are either identical or different alkyl chains. The esters
were chosen for study even though their utility as collectors has not been estab-
lished, to demonstrate the potential that the combination of VUV lasers and TOF
mass spectrometry has for the analysis of organic molecules by parent mass.
Typical results for CH3SC(S)OC2H5 under focused two-photon UV ( λ ⬇
216 nm; power density was ⬇108 W/cm2) and unfocused nonresonant VUV illu-
mination are shown in Figure 14a and b, respectively [74]. Since the UV excita-

Copyright © 2002 by Taylor & Francis

Figure 14 TOF mass spectrum of O-ethyl, S-methyl dithiocarbonate by (a) ultraviolet
(UV) two-photon ionization and (b) VUV laser single-photon ionization.

tion wavelength chosen is one-photon resonant with the sulfur n → σ* molecular

transition [79], the parent mass peak intensity in Figure 14a is markedly reduced
due to extensive fragmentation. The dominant feature in the VUV/TOF mass
spectrum shown in Figure 14b is the parent ion, and signals due to low molecular
mass ions are effectively absent. This result can also be compared with the mass
spectra for CH3SC(S)OC2H5 by obtained by EI and FAB (Fig. 15). Like the UV
REMPI/TOF spectrum, a weak parent ion peak and extensive fragmentation are
observed.
Individual VUV/TOF mass spectra of CH3SC(S)OC2H5, C2H5SC(S)OC2H5,
i-C3H7SC(S)OC2H5, and n-C3H7SC(S)OC3H7-i are shown in Figure 16. A similar
spectrum for a mixture of these esters can be compared in Figure 17. Here the
composition of the liquid mixture was established empirically to yield similar
gas phase intensities for each parent ion. It can be concluded immediately that
VUV single-photon ionization coupled with TOF mass spectrometry can readily
identify components of gas-phase xanthate mixtures by parent mass. Furthermore,
initial calculations suggest that the detection sensitivity of the experimental appa-

Copyright © 2002 by Taylor & Francis

Figure 15 (a) Fast atom bombardment (FAB) mass spectrum of O-ethyl, S-methyl dithi-
ocarbonate. (b) Electron impact (EI) mass spectrum of O-ethyl, S-methyl dithiocarbonate.
The electron energy was 70 eV.

ratus may well be in the femtomole range. Few other commercially available
analytical schemes match the performance hinted by our results. It may eventually
be possible to use VUV laser SALI methods to characterize mineral samples
from processing streams in situ. Our VUV laser/TOF mass spectrometer is cur-
rently being modified for true SALI experiments by the addition of thermal and
laser-induced desorption capabilities.

Copyright © 2002 by Taylor & Francis

Figure 16 Individual VUV laser/TOF mass spectra for (a) CH3SC(S)OC2H5; (b)
C2H5SC(S)OC2H5; (c) i-C3H7SC(S)OC2H5; and (d) n-C3H7SC(S)OC3H7-i.

Figure 17 VUV laser/TOF mass spectrum for a mixture of (1) CH3SC(S)OC2H5; (2)
C2H5SC(S)OC2H5; (4) i-C3H7SC(S)OC2H5; and (6) n-C3H7SC(S)OC3H7-i, in a volume ratio
of V1: V2 : V4 : V6 ⫽ 1.0: 4.4: 9.0: 24.8.

Copyright © 2002 by Taylor & Francis

IV. CONCLUSIONS

A goal of the literature reviews presented here is to chronicle the diverse and
often unexpected spectroscopic uses that have been found for VUV lasers. While
crystal ball gazing is, at best, fraught with danger, it is not unreasonable to expect
short-wavelength coherent sources to have a significant impact on emerging
‘‘new’’ research in the ‘‘old’’ field of classic chemical kinetics. Of course, state-
selective reaction dynamics must be viewed as a twinned discipline to spectros-
copy and, in this regard, VUV lasers are playing an important and growing role
(Table 2). However, the highly nonequilibrium conditions encountered in super-
sonic free jets are not well-suited to so-called conventional chemical kinetics
because critical parameters such as the temperature and concentration are very
difficult to characterize and control.
Rowe and co-workers at the University of Rennes I have overcome these
basic limitations by constructing a jet system known as a cinétique de réaction
en ecoulement supersonique uniforme (CRESU; reaction kinetics in uniform su-
personic flow) in which these properties can be well defined [80].
The heart of a CRESU is an axially symmetrical converging–diverging
(Laval) nozzle that is shown and compared schematically with a free jet and a
skimmed molecular beam in Figure 18 [81]. When such a nozzle is properly
designed, the true thermodynamic temperature of the continuous isentropic flow
obeys the relationship given by Eq. (7). Unlike the free jets described above,
however, the Mach number is very close to constant everywhere along the expan-
sion. Separate Laval nozzles are nevertheless needed for each buffer gas and each
desired flow temperature and number density.
Designing a Laval nozzle that can produce a uniform flow downstream of
the nozzle exit requires finding the inverse solution to a nonlinear second-order
Navier-Stokes equation with viscosity terms. This type of calculation can be car-
ried out perturbatively since uniform flow also implies that the nonlinear terms
are small.
In hydrodynamic terms, operation with small viscosity effects at the walls
of the nozzle means that the Reynolds number of the flow, Re, which is defined
as the ratio of the inertial to viscous terms in the Navier-Stokes equation, is large.
For a given reservoir temperature and Mach number, Re scales as √QM Pchamber
where QM is the mass flow.
A jet with minimal cluster formation and flow turbulence is produced at
Rennes from relatively large-diameter nozzles at a chamber pressure of ⬇ 10⫺1
torr with Re between 1000 and 2500. However, huge pumping capacities are
required. The experimental system uses two large Roots pumps (12000 m3 /hr)
in parallel backed by two additional Roots blowers with 8000 and 2000 m3 /hr
pumping speeds, respectively. Pulsed Laval nozzle systems have been designed

Copyright © 2002 by Taylor & Francis

Figure 18 Schematic diagrams of the nozzles used to generate (a) a very cold free jet
with large variations in density and pressure along the flow axis produced by expanding
a high-pressure gas through a small diameter pinhole; (b) a low divergent molecular beam
produced by placing a skimmer in the free jet; and (c) a uniform supersonic flow of con-
stant density and temperature produced by expanding a moderate pressure through a con-
vergent–divergent Laval nozzle.

to reduce these enormous demands [82]. The gas reservoir behind the Rennes
nozzle can also be cooled to liquid nitrogen temperatures (77 K), resulting in
supersonic flow temperatures as low as 13 K, and translated forward and back-
ward along the flow axis.
One of the most significant advantages of a CRESU system is that it pro-
vides an excellent environment in which to study reactions between neutral spe-

Copyright © 2002 by Taylor & Francis

cies by pulsed laser photolysis–laser-induced fluorescence (PLP-LIF). One tri-
pled Nd:YAG laser beam (⬇75 mJ per pulse) is typically used to generate an
unstable neutral radical or atom by the photolysis of a stable radical precursor.
The output of a second Nd:YAG pumped dye or optical parametric oscillator
(OPO) is then used to probe the radical concentration at a given time delay (jitter
ⱕ 1 ns) by laser-induced fluorescence after reaction with another reactant in the
expansion. A radical concentration–decay curve in the supersonic flow can be
obtained by systematically changing the delay time between the two laser pulses.
Due to the high velocities in the expansion, and because the supersonic flow is
only uniform over tens of centimeters, the time scale probed for any reaction in
a CRESU corresponds to ⬇100–500 µs. Experiments are by necessity limited by
these time and distance constraints to the study of processes having rate constants
exceeding ⬇10⫺12 cm3 molecule⫺1 s⫺1. A high-efficiency light collection system
downstream of the nozzle exit and coupled to a photomultiplier, is mounted in
such a way that the supersonic flow passes between them undisturbed.
Many important reactions in the intersteller medium involve H, O, C, and
N atoms that have strong single-photon-allowed transitions lying in the VUV.
Only one experiment to date has been reported that combines a CRESU and VUV
laser excitation, but its results are very encouraging. Reactions between ground
state C(3P) atoms with O2 and NO were studied by monitoring LIF at jet tempera-
tures between 295 K and 15 K from C(3P) atoms excited on their 3s 3P ← 2p 3P
transitions near 166 nm with a VUV laser generated by four-wave difference-
mixing in Xe [83]. Not only were precise temperature-dependent rate coefficients
deduced, but in the temperature range studied dramatic relaxation effects could
also be observed for the populations of the individual ground state spin-orbit
components. Results of this sort should significantly help to verify many of the
kinetic assumptions currently being used to model interstellar and atmospheric
chemistry.

ACKNOWLEDGMENTS

Research carried out at the University of Western Ontario (UWO) was supported
by the Natural Sciences and Engineering Research Council of Canada (NSERC)
and UWO. The authors thank Dr. S. S. Dimov, Dr. K. J. Jordan, Dr. P. Wang,
Dr. X. K. Hu, Dr. D. M. Mao, Dr. N. A. McDonald, and Mr. J. Vanstone for
their substantial contributions to the UWO research effort over the years. RHL
thanks CNRS France, B. R. Rowe, J. B. A. Mitchell, and the experimental astro-
physics group at the University of Rennes I for a visiting research fellowship,
their hospitality, and use of Figure 18. YJS is grateful for an Ontario Graduate
Scholarship.

Copyright © 2002 by Taylor & Francis

REFERENCES

1. PA Franken, AE Hill, CW Peters, G Weinreich. Generation of optical harmonics.

Phys Rev Lett 7: 118–119, 1961.
2. W Jamroz, BP Stoicheff. Generation of tunable coherent vacuum-ultraviolet radia-
tion. Progr Optics 20: 326–380, 1983.
3. CR Vidal. Four-wave frequency mixing in gases. In: LF Mollenauer, JC White,
eds. Tunable Lasers, Topics in Applied Physics 59. Berlin: Springer-Verlag, 1987,
pp 57–113.
4. JW Hepburn. Generation of coherent vacuum ultraviolet radiation: applications to
high-resolution photoionization and photoelectron spectroscopy. In: AB Meyers,
TR Rizzo, eds. Laser Techniques of Chemistry, Vol. XXIII. New York: John-Wi-
ley & Sons, 1995, pp 149–183.
5. W. Jamroz, PE LaRocque, BP Stoicheff. Resonantly enhanced second-harmonic
generation in zinc vapor. Opt Lett 7: 148–150, 1982.
6. WA Majewski, JF Pfanstiel, DF Plusquellic, DW Pratt. High resolution optical
spectroscopy in the ultraviolet. In: AB Meyers, TR Rizzo, eds. Laser Techniques
of Chemistry, Vol. XXIII. New York: John-Wiley & Sons, 1995, pp 101–148.
7. H Bitto, PR Willmott. The dynamics and its power dependence of S1(1B3u) pyrazine
studied with transform-limited nanosecond pulses. Chem Phys 165: 113–121, 1992.
8. R Hilbig, G Hilber, A Lago, B Wolff, R Wallenstein. Tunable coherent VUV radia-
tion generated by nonlinear optical frequency conversion in gases. Comments Atm
Mol Phys 18: 157–180, 1986.
9. GC Bjorklund. Effects of focusing on third-order nonlinear processes in isotropic
media, IEEE J Quantum Electron QE-11: 287–296, 1975.
10. R Hilbig, R Wallenstein. Enhanced production of tunable VUV radiation by phase-
matched frequency tripling in krypton and xenon. IEEE J Quantum Electron QE-
17: 1566–1573, 1981.
11. W Zapka, D Cotter, U Brackmann. Dye laser frequency tripling at 106 nm. Opt
Commun 36: 79–81, 1981.
12. D Cotter. Conversion from 3371 to 1124 Å by nonresonant optical frequency tri-
pling in compressed krypton gas. Opt Lett 4: 134–136, 1979; D Cotter. Tunable
narrow-band coherent VUV source for the Lyman-alpha region. Opt Commun 31:
397–399, 1979; H Langer, H Puell, H Rohr. Lyman-alpha (1216 Å) generation in
krypton. Opt Commun 34: 137–142, 1980.
13. R Mahon, TJ McIlrath, DW Koopman. Nonlinear generation of Lyman-alpha radia-
tion. Appl Phys Lett 33: 305–307, 1978.
14. R Hilbig, R Wallenstein. Tunable XUV radiation generated by nonresonant fre-
quency tripling in argon. Opt Commun 44: 283–288, 1983; EE Marinero, CT
Rettner, RN Zare, AH Kung. Excitation of H2 using continuously tunable coherent
XUV radiation (97.3–102.3 nm). Chem Phys Lett 95: 486–491, 1983.
15. R Hilbig, A Lago, R Wallenstein. Tunable XUV radiation generated by nonresonant
frequency tripling in neon. Opt Commun 49: 297–301, 1984; R Hilbig, G Hilber,
A Timmermann, R Wallenstein. Broadly tunable VUV radiation generated by fre-

Copyright © 2002 by Taylor & Francis

 quency mixing in gases. In: SE Harris, TB Lucatorto, eds. Laser Techniques in
the Extreme Ultraviolet. AIP Conference Proceedings 119. New York: American
Institute of Physics, 1984, pp 1–9.
16. R Hilbig, G Hilber, R Wallenstein. Nonresonant tripling and sum-frequency mixing
in Hg. Appl Phys B 41: 225–230, 1986.
17. CT Rettner, EE Marinero, RN Zare, AH Kung, Pulsed jets: novel nonlinear media
for generation of vacuum ultraviolet and extreme ultraviolet radiation, J Phys Chem
88: 4459–4465, 1984.
18. LJ Zych, JF Young. Limitation of 3547 to 1182 Å conversion efficiency in Xe.
IEEE J Quantum Electron QE-14: 147–149, 1978.
19. RT Hodgson, PP Sorokin, JJ Wynne. Tunable coherent vacuum-ultraviolet genera-
tion in atomic vapors. Phys Rev Lett 32: 343–346, 1973.
20. PR Herman, PE LaRocque, RH Lipson, W Jamroz, BP Stoicheff. Vacuum ultra-
violet laser spectroscopy III: laboratory sources of coherent radiation tunable
from 105 to 175 nm using Mg, Zn, and Hg vapors. Can J Phys 63: 1581–1588,
1985.
21. JA Armstrong, N Bloembergen, J Ducuing, PS Pershan. Interactions between light
waves in a nonlinear dielectric. Phys Rev 127: 1918–1939, 1962; BJ Orr, JF Ward.
Perturbation theory of the non-linear optical polarization of an isolated system. Mol
Phys 20: 513–526, 1971.
22. EU Condon, GH Shortley. The Theory of Atomic Spectra. Cambridge (UK): Cam-
bridge University Press, 1964.
23. CH Muller III, DD Lowenthal, MA DeFaccio, AV Smith. High-efficiency, energy-
scalable, coherent 130 nm source by four-wave mixing in Hg vapor. Opt Lett 13:
651–653, 1988.
24. RH Lipson. Laser spectroscopies in the VUV and XUV. Physics Can 55: 223–
232, 1999.
25. RH Lipson, SS Dimov, P Wang, YJ Shi, DM Mao, XK Hu, J Vanstone. Vacuum
ultraviolet and extreme ultraviolet lasers: principles, instrumentation, and applica-
tions. Instrum Sci Tech 28: 85–118, 2000.
26. RE Smalley, L Wharton, DH Levy. Molecular optical spectroscopy with supersonic
beams and jets. Accts Chem Res 10: 139–145, 1977.
27. DR Miller. Free jet sources. In: G Scoles, ed. Atomic and Molecular Beam Meth-
ods, volume I. New York: Oxford University Press, 1988, pp 14–53.
28. W Demtröder, Visible and ultraviolet spectroscopy: physical aspects. In: G Scoles,
ed. Atomic and Molecular Beam Methods. volume 2. New York: Oxford University
Press, 1992, pp 213–260.
29. PM Johnson, CE Otis. Molecular multiphoton spectroscopy with ionization detec-
tion. Annu Rev Phys Chem 32: 139–157, 1981.
30. SL Anderson. Multiphoton ionization state selection: vibrational mode and rota-
tional-state control. Adv Chem Physics 82: 177–212, 1992.
31. MNR Ashfold, JD Howe. Multiphoton spectroscopy of molecular species. Annu
Rev Phys Chem 45: 57–82, 1994.
32. WC Wiley, IH McLaren. Time-of-flight mass spectrometer with improved resolu-
tion. Rev Sci Instrum 26: 1150–1157, 1955.

Copyright © 2002 by Taylor & Francis

 33. DJ Auerbach. Velocity measurements by time-of-flight methods. In: G Scoles, ed.
Atomic and Molecular Beam Methods. volume 1. New York: Oxford University
Press, 1988, pp 362–379.
34. KP Lawley, RJ Donovan. Spectroscopy and electronic structure of ion-pair states.
J Chem Soc Faraday Trans 89: 1885–1898, 1993.
35. XQ Tan, TG Wright, TA Miller. Electronic spectroscopy of free radicals in super-
sonic jets. In: JM Hollas, D Phillips, eds. Jet Spectroscopy and Molecular Dynam-
ics. London: Blackie Academic & Professional, 1995, pp 74–117.
36. K Wieland. Bandenspektren der quecksilber-, cadmium- und zinkhalogenide. Helv
Phys Acta 2: 47–94, 1929; K Wieland. Absorptions- und Fluoreszenzspektren
dampfförmiger quecksilberhalogenide. I HgJ2. Z Phys 76: 801–813, 1932; K
Wieland. Absorptions- und Fluoreszenzsperktren dampfförmiger quecksilberhalo-
genide. II. HgBr2 and HgCl2. Z Phys 77: 157–1655, 1932; K Wieland. Das langwel-
lige emissions- ünd flüoreszenzspektrum (5700-3000 angstrom) von natürlichem
HgCl ünd von künstlich angereichertem HgC137. Helv Phys Acta 14: 420–464,
1941.
37. R Burnham. Discharge pumped mercuric halide dissociation lasers. Appl Phys Lett
33: 156–159, 1978; YE Gavrilova, VS Zrodnikov, AO Klementov, AS Podso-
sonny. Excimer HgI* laser excited by an electric discharge. Sov J Quantum Elec-
tron 10: 1457–1459, 1980; WT Whitney. Sustained discharge excitation of HgCl
and HgBr B 2⌺1/2 ⫹
→ X 2⌺1/2
⫹
lasers. Appl Phys Lett 32: 239–241, 1978; JG Eden.
VUV-pumped HgCl laser. Appl Phys Lett 33: 495–497, 1978; JH Parks. Laser
action on the B 2⌺1/2
⫹
→ X 2⌺1/2
⫹
band of HgCl at 5576 Å. Appl Phys Lett 31: 192–
194, 1977; JG Eden. Green HgCl (B 2⌺1/2 ⫹
← X 2⌺1/2
⫹
) laser. Appl Phys Lett 31:
448–450, 1977.
38. AW McCown, MN Ediger, JG Eden. Quenching kinetics and small signal gain
spectrum of the ZnI photodissociation laser. Opt Common 40: 190–194, 1982.
39. CE Moore. Atomic energy levels. Vol. III. Natl Bur Stand Circ No. 467. Washing-
ton, DC: USGPO, 1971.
40. J Tellinghuisen, JG Ashmore. The B → X transition in 200Hg 79Br Appl Phys Lett 40:
867–869, 1982; J Tellinghuisen, JG Ashmore. Mixed representations for diatomic
spectroscopic data: application to HgBr. Chem Phys Lett 102: 10–16, 1983; J Tell-
inghuisen, PC Davies, P Berwanger, KS Viswanathan. B → X transitions in HgCl
and HgI. Appl Phys Lett 41: 789–791, 1982; KS Viswanathan, J Tellinghuisen.
The B (2⌺⫹) → X (2⌺⫹) transition (4050–4500Å) in HgI. J Mol Spectrosc 98: 185–
198 (1983).
41. DP Greene, KP Killeen, JG Eden. X 2⌺ → B 2⌺ absorption band of HgBr: optically
pumped 502 nm laser. Appl Phys Lett 48: 1175–1177, 1986; DP Greene, KP Kil-
leen, JG Eden. Excitation of the HgBr B 2⌺1/2 ⫹
→ X 2⌺1/2⫹
band in the ultraviolet. J
Opt Soc Am B3: 1282–1287, 1986.
42. PC Engelking. Spectroscopy of jet-cooled ions and radicals. Chem Rev 91: 399–
414, 1991.
43. RH Lipson, KJ Jordan, HA Bascal. Fluorescence excitation spectra of jet-cooled
HgBr radicals. J Chem Phys 98: 959–967, 1993; KJ Jordan, HA Bascal, RH Lipson,
M Melchior. The B 2⌺⫹ ← X 2⌺⫹ transition of HgI. J Mol Spectrosc 159: 144–

Copyright © 2002 by Taylor & Francis

 155, 1993; HA Bascal, KJ Jordan, RH Lipson. Ion-pair spectroscopy of HgCl. Can
J Chem 71: 1615–1621, 1993.
44. P DeBièvre, M Gallent, NE Holden, IL Barnes. Isotopic abundances and atomic
weights of the elements. J Phys Chem Ref Data 13: 809–891, 1984.
45. RH Lipson, AR Hoy. Vacuum Ultraviolet Laser Spectra of ICl. J Chem Phys 90,
6821–6826, 1989.
46. J Tellinghuisen, SD Henderson. The use of Morse-RKR curves in diatomic calcula-
tions. Chem Phys Lett 91: 447–451, 1982.
47. RJ LeRoy. Program manual for RKR1: a computer program implementing the first-
order RKR method for determining diatom potential energy curves from spectro-
scopic constants. University of Waterloo, Chemical Physics Research Report CP-
425, 1992.
48. RJ LeRoy. Program manual for LEVEL: a computer program for solving the radial
Schrodinger equation for bound and quasibound levels and calculating (if desired)
expectation values and Franck-Condon intensity factors. University of Waterloo,
Chemical Research Report, CP-300, 1991.
49. RS Mulliken. Iodine revisited. J Chem Phys 55, 288–309, 1971; JCD Brand, AR
Hoy. Multiphoton spectra and states of halogens. Appl Spectrosc Rev 23: 285–
327, 1987.
50. RN Zare, DR Herschbach. Charge transfer model for alkali halide electronic transi-
tion strengths. J Mol Spectrosc 15: 462–472, 1965.
51. AR Hoy, RH Lipson. Reinvestigation of the Cordes band system of I2 using a vac-
uum ultraviolet laser. Chem Phys 140: 187–193, 1990.
52. RH Lipson, AR Hoy. A vacuum ultraviolet laser study of IBr. Mol Phys 68: 1311–
1319, 1989.
53. RH Lipson, AR Hoy, MJ Flood. Direct vibrational numbering of the D0u⫹ ion-pair
state of Br2. Chem Phys Lett 149, 155–160, 1988; RH Lipson, AR Hoy. VUV laser
spectroscopy of ion-pair states of Br2. J Mol Spectrosc 143: 183–198, 1989.
54. SS Dimov, RH Lipson, T Turgeon, JA Vanstone, P Wang, DS Yang. Vacuum
ultraviolet/time-of-flight mass spectroscopy: ion-pair spectra of 79Br35Cl. J Chem
Phys 100, 8666–8672, 1994; P Wang, SS Dimov, RH Lipson. Vibronic analysis
of the ion-pair (E0⫹)¸ ground state (X0⫹) transition of BrCl. J Chem Phys 107:
3345–3351, 1997.
55. P Wang, IV Okuda, SS Dimov, RH Lipson. Mass-resolved VUV laser spectra in
the vicinity of the Rydberg minimum of the 1 1⌺u⫹ state of Cl2. J Phys Chem 229:
370–376, 1994; P Wang, SS Dimov, G Rosenblood, RH Lipson. Vibrational reas-
signment of the 2 1⌺u⫹ ← X 1⌺g⫹ transition of Cl2. J Phys Chem 99: 3984–3989,
1995; P Wang, SS Dimov, RH Lipson. Rotational analyses for selected bands
of the 2 1⌺u⫹ ← X 1⌺g⫹ transition of Cl2. J Phys Chem A 101, 4555–4559, 1997;
P Wang, IV Okuda, SS Dimov, RH Lipson. Rotationally resolved vacuum ultravio-
let laser spectra of the 37Cl2 1 1⌺u⫹ ← X 1⌺g⫹ transition. J Mol Spectrosc 190, 213–
225, 1998.
56. DK Chakraborty, PC Tellinghuisen, J Tellinghuisen. The emission of BrCl: analysis
of the D′ → A′ and E → B transitions. Chem Phys Lett 141: 36–40, 1987; SW
Brown, CJ Dowd, Jr., J Tellinghuisen. The E → B transition in 81Br37Cl. J Mol
Spectrosc 132: 178–192, 1988.

Copyright © 2002 by Taylor & Francis

 57. A Hopkirk, D Shaw, RJ Donovan, KP Lawley, AJ Yencha. Vacuum-ultraviolet
absorption, fluorescence excitation, and dispersed fluoresence spectra of BrCl. J
Phys Chem 93: 7338–7342, 1989; A Kvaran, AJ Yencha, KP Lawley, RJ Donovan.
Analyses of the 290–400 nm oscillatory continua due to transitions from the E(O⫹)
ion pair state of BrCl. Mol Phys 75: 197–207, 1992.
58. RJ LeRoy, RB Bernstein. Dissociation energy and long-range potential of diatomic
molecules from vibrational spacings of higher levels. J Chem Phys 52: 3869–3879,
1970; RJ LeRoy, W-H Lam. Near dissociation expansions in the spectroscopic de-
termination of diatom dissociation energies in the spectroscopic determination of
diatom dissociation energies: method, and application to BeAr⫹. Chem Phys Lett
71, 544–548, 1980; KJ Jordan, RH Lipson, NA McDonald, RJ LeRoy. Jet emission
spectra of CdI and HgI and near dissociation theory analyses for CdI and ZnI. J
Phys Chem 96, 4778–4787, 1992; SH Pan, FH Mies. Rydberg-like properties of
rotational-vibrational levels and dissociation continuum associated with alkali-
halide charge-transfer states. J Chem Phys 89: 3096–3103, 1988.
59. SD Peyerimhoff, RJ Buenker. Electronically excited and ionized states of the chlo-
rine molecule. Chem Phys 57: 279–296, 1981.
60. K Yamanouchi, S Tsuchiya. Tunable vacuum ultraviolet laser spectroscopy: excited
state dynamics of jet-cooled molecules and van der Waals complexes. J Phys B:
Atm Mol Opt Phys 28: 133–165 (1995).
61. RN Zare. Laser chemical analysis. Science 226: 298–303, 1984; R Snook. Laser
techniques for chemical analysis. Chem Soc Rev 26: 319–326, 1997.
62. JH Futrell. In: TD Märk, GH Dunn, Eds. Electron Impact Ionization. Wien New
York: Springer-Verlag, 1985, pp 364–374.
63. HM Fales, Y Nagai, GWA Milne, HB Brewer, Jr., TJ Bronzert, LJ Pisano. Use of
chemical ionization mass spectrometry in analysis of amino acid phenylthiohydan-
toin derivatives formed during Edman degradation of proteins. Anal Biochem 43:
288–299, 1971; MA Baldwin, FW McLafferty. Direct chemical ionization of rela-
tively involatile samples. Application to underivatized oligopeptides. Org Mass
Spectrom 7: 1353–1356, 1973.
64. M Barber, RS Bordoli, GJ Elliot, RD Sedgwick, AN Taylor. Fast atom bombard-
ment mass spectrometry. Anal Chem 54: 645A–656A, 1982; RM Caprioli. Contin-
uous-flow fast atom bombardment mass spectrometry. Anal Chem 62: 477A–485A,
1990.
65. A Benninghoven, WK Sichtermann. Detection, identification and structural investi-
gation of biologically important compounds by secondary ion mass spectrometry.
Anal Chem 50: 1180–1184, 1978.
66. M Karas, U Bahr. Matrix-assisted laser desorption-ionization (MALDI) mass spec-
trometry of biological molecules. In: RM Caprioli, A Malorni, G Sindona, eds.
Mass Spectrometry in Biomolecular Sciences, Nato ASI Series Vol. 475. Dor-
drecht, The Netherlands: Kluwer Academic Publishers, 1996, pp 33–49.
67. CH Becker, KT Gillen. Surface analysis by nonresonant multiphoton ionization of
desorbed or sputtered species. Anal Chem 56: 1671–1674, 1984.
68. CH Becker, KT Gillen. Surface Analysis of contaminated GaAs: comparison
of new laser-based techniques with SIMS. J Vac Sci Technol A 3: 1347–1349,
1985.

Copyright © 2002 by Taylor & Francis

 69. GS Hurst, MG Payne, SD Kramer, JP Young. Resonance ionization spectroscopy
and one-atom detection. Rev Mod Phys 51: 767–819, 1979.
70. EW Schlag, HJ Neusser. Multiphoton mass spectrometry. Acc Chem Res 16: 355–
360, 1983.
71. BD Morrical, DP Fergenson, KA Prather. Coupling two-step laser desorption/ion-
ization with aerosol time-of-flight mass spectrometry for the analysis of individual
organic particles. J Am Soc Mass Spectrom 9: 1068–1073, 1998.
72. U Schühle, JB Pallix, CH Becker. Sensitive mass spectrometry of molecular ad-
sorbates by stimulated desorption and single-photon ionization. J Am Chem Soc
110: 2323–2324, 1988.
73. DJ Butcher. Vacuum ultraviolet radiation for single-photon mass spectrometry: a
review. Microchem J 62: 354–362, 1999.
74. YJ Shi, XK Hu, DM Mao, SS Dimov, RH Lipson. Analysis of xanthate derivatives
by vacuum ultraviolet laser-time-of-flight mass spectrometry. Anal Chem 70:
4534–4539, 1998.
75. I Persson. Absorption of ions and molecules to solid surfaces in connection with
flotation of sulphide minerals. J Coord Chem 32: 261–342, 1994; HL Shergold.
Flotation in mineral processing. In: JI Ives, ed. The Scientific Basis of Flotation.
The Hague: Martinus Nijhoff Publishers, 1984, pp 229–287; VI Klassen, VA Mpo-
krousov. An Introduction to the Theory of Flotation. London: Butterworth, 1963;
AM Gaudin. Flotation. New York: McGraw-Hill, 1957.
76. PK Ackerman, GH Harris, RR Klimpel, FF Aplan. Evaluation of flotation collectors
for copper sulfide and pyrite I. Common sulfhydryl collectors. Int J Min Process
21: 105–127, 1987.
77. JY Kim, SL Chryssoulis. Influence of lead ions in sulfide flotation—the application
of laser-ionization mass spectrometry. Min Metall Process 69–76, 1996.
78. I Degani, R Fochi. The phase-transfer synthesis of O,S-dialkyl dithiocarbonates
from alkyl halides and alkyl mthanesulfonates. Synthesis 5: 365–368, 1978.
79. KJ Rosengren. Electronic absorption spectra of unconjugated alkyl thials and thi-
ones. Acta Chem Scand 16: 2284–2292, 1962.
80. IR Sims, J-L Queffelec, A Defrance, C Rebrion-Rowe, D Travers, P Bocherel, BR
Rowe, IWM Smith. Ultralow temperature kinetics of neutral-neutral reactions. The
technique and results for the reactions CN ⫹ O2 down to 13 K and CN ⫹ NH3
down to 25 K. J Chem Phys 100: 4229–4241, 1994.
81. IWM Smith, BR Rowe. Reaction kinetics at very low temperatures: laboratory stud-
ies and interstellar chemistry. Accts Chem Res 33: 261–268, 2000.
82. DB Atkinson, MA Smith. Design and characterization of pulsed supersonic expan-
sions for chemical applications. Rev Sci Instrum 66: 4434–4446, 1995.
83. D Chastaing, SD Le Picard, IR Sims. Direct kinetic measurements on reactions of
atomic carbon, C(3P), with O2 and NO at temperatures down to 15 K. J Chem Phys
112: 8466–8469, 2000.
84. I Munro, AP Sabersky. In: H Winich, S Doniach, eds. Synchrotron Radiation Re-
search. New York: Plenum Press, 1980.
85. Characteristics of the Advanced Light Source: A Brief Summary, Lawrence Berke-
ley Labs. http: / /www-als.lbl.gov/als/workshops/alscharacter.html.
86. DW Tokaryk, GZ Zhang, BP Stoicheff, Phys Rev A 59: 3116–3119, 1999.

Copyright © 2002 by Taylor & Francis

 87. GZ Zhang, DW Tokaryk, BP Stoicheff. Phys Rev A 56: 813–819, 1997.
88. M Katsuragawa, GZ Zhang, K Kakuta. Opt Commun 129: 212–216, 1996.
89. RSD Sihombing, M Katsuragawa, GZ Zhang, K Hakuta. Phys Rev A 54: 1551–
1555, 1996.
90. RI Thompson, BP Stoicheff, GZ Zhang, K Hakuta. Appl Phys B 60: S129–S139,
1995.
91. GZ Zhang, M Katsuragawa, K Hakuta, RI Thompson, BP Stoicheff. Phys Rev A
52: 1584–1593, 1995.
92. RI Thompson, BP Stoicheff, GZ Zhang, K Hakuta. Quantum Opt 6: 349–358,
1994.
93. GZ Zhang, K Hakuta, BP Stoicheff. Phys Rev Lett 71: 3099–3102, 1993.
94. K Hakuta, L Marmet, BP Stoicheff. Phys Rev A 45: 5152–5159, 1992.
95. L Marmet, K Hakuta, BP Stoicheff. J Opt Soc Am B9, 1038–1046, 1992.
96. K Hakuta, L Marmet, BP Stoicheff. Phys Rev Lett 66: 1042–1045, 1991.
97. K Hakuta, L Marmet, BP Stoicheff. In: M Ducloy, E Giacobino, G Camy, eds.
Laser Spectroscopy X. Singapore: World Scientific, 1991, pp 301–306.
98. L Marmet, K Hakuta, BP Stoicheff. Opt Lett 16: 261–263, 1991.
99. RA Brownsword, M Hillenkamp, T Laurent, RK Vatsa, H-R Volpp, J Wolfrum.
J Phys Chem A, 101: 5222–5227, 1997.
100. RA Brownsword, M Hillenkamp, T Laurent, RK Vatsa, H-R Volpp. J Chem Phys
106: 4436–4447, 1997.
101. RA Brownsword, M Hillenkamp, T Laurent, RK Vatsa, H-R Volpp, J Wolfrum.
J Phys Chem A 101: 995–999, 1997.
102. RA Brownsword, M Hillenkamp, T Laurent, RK Vatsa, H-R Volpp, J Wolfrum.
J Chem Phys 106: 1359–1366, 1997.
103. RA Brownsword, M Hillenkamp, T Laurent, RK Vatsa, H-R Volpp. J Chem Phys
106: 9563–9569, 1997.
104. RA Brownsword, T Laurent, RK Vatsa, H-R Volpp, J Wolfrum. Chem Phys Lett
249: 162–166, 1996.
105. N Tada, K Tonokura, K Matsumoto, M Koshi, A Miyoshi, H Matsui. J Phys Chem
A 103: 322–329, 1999.
106. H-R Volpp, J Wolfrum. In: J Wolfrum, H-R Volpp, R Rannacher, J Warnatz, eds.
Gas Phase Chemical Reaction Systems: Experiments and Models 100 years after
Max Bodenstein. Springer Series in Chemical Physics 61. Heidelberg: Springer,
1996, pp 14–31.
107. RA Brownsword, T Laurent, RK Vatsa, H-R Volpp, J Wolfrum. Chem Phys Lett
258: 164–170, 1996.
108. RA Brownsword, M Hillenkamp, T Laurent, RK Vatsa, H-R Volpp, J Wolfrum.
Chem Phys Lett 266: 259–266, 1997.
109. S Koppe, T Laurent, PD Naik, H-R Volpp, J Wolfrum. Can J Chem 72: 615–624,
1994.
110. W Yi, R Bersohn. Chem Phys Lett 206: 365–368, 1993.
111. K Tonokura, Y Matsumi, M Kawasaki, S Tasaki, R Bersohn. J Chem Phys 97:
8210–8215, 1992.
112. S Koppe, T Laurent, PD Naik, H-R Volpp, J Wolfrum, T Arusi-Parpar, I Bar, S
Rosenwaks. Chem Phys Lett 214: 546–552, 1993.

Copyright © 2002 by Taylor & Francis

113. T Laurent, PD Naik, H-R Volpp, J Wolfrum, T Arusi-Parpar, I Bar, S Rosenwaks.
Chem Phys Lett 236: 343–349, 1995.
114. W Yi, S Satyapal, N Shafer, R Bersohn. J Chem Phys 99: 4548–4553, 1993.
115. A Tezaki, S Okada, H Matsui. J Chem Phys 98: 3876–3883, 1993.
116. Y Matsumi, K Tonokura, M Kawasaki, HL Kim. J Phys Chem 96: 10622–10626,
1992.
117. S Satyapal, R Bersohn. J Phys Chem 95: 8004–8006, 1991.
118. W Yi, A Chattopadhyay, R Bersohn. J Chem Phys 94: 5994–5998, 1991.
119. K-H Gericke, M Lock, FJ Comes. Chem Phys Lett 186: 427–430, 1991.
120. J Park, R Bersohn. J Chem Phys 93: 5700–5708, 1990.
121. S Satyapal, GW Johnston, R Bersohn, I Oref. J Chem Phys 93: 6398–6402,
1990.
122. GW Johnston, S Satyapal, R Bersohn, B Katz. J Chem Phys 92: 206–212, 1990.
123. M Koshi, F Tamura, H Matsui. Chem Phys Lett 173: 235–240, 1990.
124. P Löffler, D Lacombe, A Ross, E Wrede, L Schnieder, KH Welge. Chem Phys
Lett 252, 304–310, 1996.
125. DH Mordaunt, MNR Ashfold, RN Dixon, P Löffler, L Schneider, KH Welge. J
Chem Phys 108: 519–526, 1998.
126. MNR Ashfold, DH Mordaunt, SHS Wilson. Comm Atm Mol Phys 32: 187–196,
1996.
127. DH Mordaunt, MNR Ashfold, RN Dixon. J Chem Phys 104: 6460–6471, 1996.
128. SHS Wilson, CL Reed, DH Mordaunt, MNR Ashfold, M Kawasaki. Bull Chem
Soc Jpn 69: 71–76, 1996.
129. AJR Heck, RN Zare, DW Chandler. J Chem Phys 104: 3399–3402, 1996.
130. SHS Wilson, JD Howe, KNR Rosser, MNR Ashfold, RN Dixon. Chem Phys Lett
227: 456–460, 1994.
131. SHS Wilson, MNR Ashfold, RN Dixon. J Chem Phys 101: 7538–7547, 1994.
132. DH Mordaunt, MNR Ashfold. J Chem Phys 101: 2630–2631, 1994.
133. IR Lambert, GP Morley, DH Mordaunt, MNR Ashfold, RN Dixon. Can J Chem
72: 977–984, 1994.
134. GP Morley, IR Lambert, DH Mordaunt, SHS Wilson, MNR Ashfold, RN Dixon,
CM Western. J Chem Soc Faraday Trans 89: 3865–3875, 1993.
135. GP Morley, IR Lambert, MNR Ashfold, KN Rosser, CM Western. J Chem Phys
97: 3157–3165, 1992.
136. BA Balko, J Zhang, YT Lee. J Chem Phys 94: 7958–7966, 1991.
137. RE Continetti, BA Balko, YT Lee. Chem Phys Lett 182: 400–405, 1991.
138. MNR Ashfold, IR Lambert, DH Mordaunt, GP Morley, CM Western. J Phys Chem
96: 2938–2949, 1992.
139. X Xie, L Schnieder, H Wallmeier, R Boettner, KH Weige, MNR Ashfold. J Chem
Phys 92: 1608–1616, 1990.
140. L Schnieder, W Meier, KH Weige. J Chem Phys 92: 7027–7037, 1990.
141. DH Mordaunt, IR Lambert, GP Morley, MNR Ashfold, RN Dixon, CW Western,
L Schnieder, KH Welge. J Chem Phys 98: 2054–2065, 1993.
142. DH Mordaunt, MNR Ashfold, RN Dixon. J Chem Phys 100: 7360–7375, 1994.
143. P Löffler, E Wrede, L Schnieder, JB Halpern, WM Jackson, KH Welge. J Chem
Phys 109: 5231–5246, 1998.

Copyright © 2002 by Taylor & Francis

144. L Schnieder, K Seekamp-Rahn, E Wrede, KH Welge. J Chem Phys 107: 6175–
6195, 1997.
145. J Zhang, CW Riehn, M Dulligan, C Wittig. J Chem Phys 103: 6815–6818, 1995.
146. L Schnieder, K Seekamp-Rahn, J Borkowski, E Wrede, KH Welge, FJ Aoiz, L Ba-
ñares, MJ D’Mello, VJ Herrero, V Sáez Rábanos, RE Wyatt. Science 269: 207–
210, 1995.
147. J Zhang, M Dulligan, C Wittig. J Phys Chem 99: 7446–7452, 1995.
148. C Jaques, L Valachovic, S Ionov, E Böhmer, Y Wen, J Segall, C Wittig. J Chem
Soc Faraday Trans 89: 1419–1425, 1993.
149. J Segall, Y Wen, R Singer, C Wittig, A García-Vela, RB Gerber. Chem Phys Lett
207: 504–509, 1993.
150. J Segall, Y Wen, R Singer, M Dulligan, C Wittig. J Chem Phys 99: 6600–6606,
1993.
151. J Segall, Y Wen, R Lavi, R Singer, C Wittig. J Phys Chem 95: 8078–8081, 1991.
152. L Schnieder, K Seekamp-Rahn, F Liedeker, H Steuwe, KH Welge. Faraday Dis-
cuss. Chem Soc 91: 259–269, 1991.
153. J-H Wang, Y-T Hsu, K Liu. J Phys Chem A 101: 6593–6602, 1997.
154. L-H Lai, D-C Che, K Liu. J Phys Chem 100: 6376–6380, 1996.
155. P Balling, HH Andersen, CA Brodie, UV Pedersen, VV Petrunin, MK Raarup, P
Steiner, T Andersen. Phys Rev A 61: 022702-1–022702-11, 2000.
156. P Balling, P Kristensen, HH Andersen, UV Pedersen, VV Petrunin, L Praestegaard,
HK Haugen, T Andersen. Phys Rev Lett 77: 2905–2908, 1996.
157. HH Andersen, P Balling, P Kristensen, UV Pedersen, SA Aseyev, VV Petrunin,
T Andersen. Phys Rev Lett 79: 4770–4773, 1997.
158. E Reinhold, W Hogervorst, W Ubachs, L Wolneiwicz. Phys Rev A 60: 1258–1270,
1999.
159. W Hogervorst, KSE Eikema, E Reinhold, W Ubachs. Nucl Phys A 63: 353c–362c,
1998.
160. E Reinhold, W Hogervorst, W Ubachs. Phys Rev Lett 78: 2543–2546, 1997.
161. E Reinhold, W Hogervorst, W Ubachs. J Mol Spectrosc 180: 156–163, 1996.
162. PC Hinnen, W Hogervorst, S Stolte, W Ubachs. Can J Phys 72: 1032–1042, 1994.
163. RM Rao, J Dvorak, RJ Beuhler, MG White. J Phys Chem B 102: 9050–9060,
1998.
164. PC Hinnen, W Hogervorst, S Stolte, W Ubachs. Appl Phys B 59: 307–310, 1994.
165. E Reinhold, A de Lange, W Hogervorst, W Ubachs. J Chem Phys 109: 9772–9782,
1998.
166. K Tsuchiyama, J Ishii, T Kasuya. J Chem Phys 97: 875–882, 1992.
167. H Rottke, KH Welge. J Chem Phys 97: 908–926, 1992.
168. EF McCormack, EE Eyler. Phys Rev Lett 66: 1042–1045, 1991.
169. EE Eyler. Comments. Atm Mol Phys 24: 299–310, 1990.
170. EE Eyler, N Melikechi. Phys Rev A 48: R18–R21, 1993.
171. A Stolow, BA Balko, EF Cromwell, J Zhang, YT Lee. J Photochem Photobiol A.:
Chem 62: 285–300, 1992.
172. R Shiell, XK Hu, Q Hu, JW Hepburn. Faraday Discuss 115: 331–343, 2000.
173. HH Fielding, TP Softley. Chem Phys Lett 185: 199–205, 1991.
174. HH Fielding, TP Softley, F Merkt. Chem Phys 155: 257–265, 1991.

Copyright © 2002 by Taylor & Francis

175. A Balakrishnan, V Smith, BP Stoicheff. Phys Rev A 49: 2460–2469, 1994.
176. A Balakrishnan, V Smith, BP Stoicheff. Phys Rev Lett 68: 2149–2152, 1992.
177. A Balakrishnan, BP Stoicheff. J Mol Spectrosc 156: 517–518, 1992.
178. LM Dobeck, HM Lambert, W Kong, PJ Pisano, PL Houston. J Phys Chem A 103:
10312–10323, 1999.
179. F Merkt, TP Softley. J Chem Phys 96: 4149–4156, 1992.
180. F Merkt, SR Mackenzie, TP Softley. J Chem Phys 99: 4213–4214, 1993.
181. E Reinhold, W Hogervorst, W Ubachs. Chem Phys Lett 296: 411–416, 1998.
182. F Merkt, H Xu, RN Zare. J Chem Phys 104: 950–961, 1996.
183. A Balakrishnan, M Vallet, BP Stoicheff. J Mol Spectrosc 162, 168–171, 1993.
184. SD Bergeson, A Balakrishnan, KGH Baldwin, TB Lucatorto, JP Marangos, TJ
McIlrath, TR O’Brian, SL Rolston, CJ Sansonetti, J Wen, N Westbrook. Phys Rev
Lett 80: 3475–3478, 1998.
185. CEM Strauss, SH Kable, GK Chawla, PL Houston, IR Burak. J Chem Phys 94:
1837–1849, 1991.
186. D Chastaing, SD Le Picard, IR Sims. J Chem Phys 112: 8466–8469, 2000.
187. JA Blush, P Chen, RT Wiedmann, MG White. J Chem Phys 98: 3557–3559, 1993.
188. N Taniguchi, K Takahashi, Y Matsumi, SM Dylewski, JD Geiser, PL Houston. J
Chem Phys 111: 6350–6355, 1999.
189. LL Springsteen, S Satyapal, Y Matsumi, LM Dobeck, PL Houston. J Phys Chem
97: 7239–7241, 1993.
190. M Abe, Y Sato, Y Inagaki, Y Matsumi, M Kawasaki. J Chem Phys 101: 5647–
5651, 1994.
191. K Takahashi, Y Matsumi, M Kawasaki. J Phys Chem 100: 4084–4089, 1996.
192. K Takahashi, M Kishigami, Y Matsumi, M Kawasaki, AJ Orr-Ewing. J Chem Phys
105: 5290–5293, 1996.
193. K Takahashi, M Kishigami, N Taniguchi, Y Matsumi, M Kawasaki. J Chem Phys
106: 6390–6397, 1997.
194. K Takahashi, N Taniguchi, Y Matsumi, M Kawasaki, MNR Ashfold. J Chem Phys
108: 7161–7172, 1998.
195. J Miyawaki, K Yamanouchi, S Tsuchiya. Chem Phys Lett 180: 287–292, 1991.
196. J Miyawaki, T Tsuchizawa, K Yamanouchi, S Tsuchiya. Chem Phys Lett 165: 168–
170, 1990.
197. AMS Chowdhury, Y Matsumi, M Kawasaki. Chem Lett 77–78, 1997.
198. Y Matsumi, AMS Chowdhury. J Chem Phys 104: 7036–7044, 1996.
199. K Takahashi, R Wada, Y Matsumi, M Kawasaki. J Phys Chem 100: 10145–10149,
1996.
200. AMS Chowdhury, M Kawasaki. Laser Phys 6: 1175–1179, 1996.
201. AMS Chowdhury. Laser Phys 7: 1058–1062, 1997.
202. AMS Chowdhury, Y Matsumi, M Kawasaki. Laser Phys 7: 946–951, 1997.
203. Y Matsumi, SM Shamsuddin, Y Sato, M Kawasaki. J Chem Phys 101: 9610–9618,
1994.
204. A Miyoshi, K Tsuchiya, N Yamauchi, H Matsui. J Phys Chem 98: 11452–11458,
1994.
205. Y Matsumi, Y Inagaki, GP Morley, M Kawasaki. J Chem Phys 100: 315–324,
1994.

Copyright © 2002 by Taylor & Francis

206. M Abe, Y Sato, Y Inagaki, Y Matsumi, M Kawasaki. J Chem Phys 101: 5647–
5651, 1994.
207. SM Shamsuddin, Y Inagaki, Y Matsumi, M Kawasaki. Can J Chem 72: 637–642,
1994.
208. AMS Chowdhury. Laser Chem 17: 191–203, 1997.
209. R Signorell, F Merkt. J Chem Phys 110: 2309–2311, 1999.
210. RT Wiedmann, RG Tonkyn, MG White, K Wang, V McKoy. J Chem Phys 97:
768–772, 1992.
211. X Li, CR Vidal. J Chem Phys 101: 5523–5528, 1994.
212. B Niu, MG White. J Chem Phys 104: 2136–2145, 1996.
213. WL Glab. J Chem Phys 107: 5979–5982, 1997.
214. AH Zanganeh, JH Fillon, J Ruiz, M Castillejo, JL Lemaire, N Shafizadeh, F Rostas.
J Chem Phys 112: 5660–5671, 2000.
215. MJJ Vrakking, YT Lee, RD Gilbert, MS Child. J Chem Phys 98: 1902–1915, 1993.
216. RG Tonkyn, R Wiedmann, ER Grant, MG White. J Chem Phys 95: 7033–7040,
1991.
217. K Wang, MT Lee, V McKoy, RT Wiedman, MG White. Chem Phys Lett 219:
397–404, 1994.
218. F Merkt, R Signorell, H Palm, A Osterwalder, M Sommavilla. Mol Phys 95: 1045–
1054, 1998.
219. RT Wiedmann, MG White. Proc SPIE: Opt Methods State Time-Resolved Chem
1638: 273, 1992.
220. R Signorell, M Sommavilla, F Merkt. Chem Phys Lett 312: 139–148, 1999.
221. KSE Eikema, W Ubachs, W Hogervorst. Phys Rev A 49: 803–808, 1994.
222. A Mank, D Rodgers, JW Hepburn. Chem Phys Lett 219: 169–173, 1994.
223. X Li, CR Vidal. J Chem Phys 102: 9167–9173, 1995.
224. R Signorell, H Palm, F Merkt. J Chem Phys 106: 6523–6533, 1997.
225. DM Jonas, X Zhao, K Yamanouchi, PG Green, GW Adamson, RW Field. J Chem
Phys 92: 3988–3989, 1990.
226. RT Wiedmann, MG White. J Chem Phys 102: 5141–5151, 1995.
227. RL Miller, SH Kable, PL Houston, I Burak. J Chem Phys 96: 332–338, 1992.
228. L Fleck, RJ Beuhler, MG White. J Chem Phys 106: 3813–3816, 1997.
229. SS Dimov, CR Vidal. Chem Phys Lett 221: 307–310, 1994.
230. V Kleiman, K Trentelman, Y Huang, RJ Gordon. Chem Phys Lett 222: 161–166,
1994.
231. SP Lu, SM Park, Y Xie, RJ Gordon. J Chem Phys 96: 6613–6620, 1992.
232. RJ Gordon, S-P Lu, SM Park, K Trentelman, Y Xie, L Zhu, A Kumar, WJ Meath.
J Chem Phys 98: 9481–9486, 1993.
233. T Sykora, CR Vidal. J Chem Phys 108: 6320–6330, 1998.
234. GC McBane, SH Kable, PL Houston, GC Schatz. J Chem Phys 94: 1141–1149,
1991.
235. R Jimenez, SH Kable, JC Loison, CJSM Simpson, W Adam, PL Houston. J Phys
Chem 96: 4188–4195, 1992.
236. DW Neyer, SH Kable, J-C Loison, PL Houston, I Burak, EM Goldfield. J Chem
Phys 97: 9036–9045, 1992.
237. DW Neyer, X Luo, PL Houston. J Chem Phys 98: 5095–5098, 1993.

Copyright © 2002 by Taylor & Francis

238. DW Neyer, X Luo, I Burak, PL Houston. J Chem Phys 102: 1645–1657, 1995.
239. Y Inagaki, M Abe, Y Matsumi, M Kawasaki, H Tachikawa. J Phys Chem 99:
12822–12828, 1995.
240. YS Choi, CB Moore. J Chem Phys 103: 9981–9988, 1995.
241. M Abe, Y Inagaki, LL Springsteen, Y Matsumi, M Kawasaki, H Tachikawa. J Phys
Chem 98: 12641–12645, 1994.
242. SA Buntin, RR Cavanagh, LJ Richter, DS King. J Chem Phys 94: 7937–7950,
1991.
243. FJ Schlenker, F Bouchard, IM Waller, JW Hepburn. J Chem Phys 93: 7110–7118,
1990.
244. YS Choi, P Teal, CB Moore. J Opt Soc Am B 7: 1829–1834, 1990.
245. A Jolly, JL Lemaire, D Belle-Oudry, S Edwards, D Malmasson, A Vient, F Rostas.
J Phys B: Atm Mol Opt Phys 30: 4315–4337, 1997.
246. W Ubachs, PC Hinnen, P Hansen, S Stolte, W Hogervorst, P Cacciani. J Mol Spec-
trosc 174: 388–396, 1995.
247. P Cacciani, W Hogervorst, W Ubachs. J Chem Phys 102: 8308–8320, 1995.
248. W Ubachs, KSE Eikema, W Hogervorst, PC Cacciani. J Opt Soc Am B 14: 2469–
2476, 1997.
249. KSE Eikema, W Hogervorst, W Ubachs. Chem Phys 181: 217–245, 1994.
250. KSE Eikema, W Hogervorst, W Ubachs. J Mol Spectrosc 163: 19–26, 1994.
251. PF Levelt, W Ubachs, W Hogervorst. J Chem Phys 97: 7160–7166, 1992.
252. PC Hinnen, S Stolte, W Hogervorst, W Ubachs. J Opt Soc Am B 15: 2620–2625,
1998.
253. W Ubachs, KSE Eikema, PF Levelt, W Hogervorst, M Drabbles, WL Meerts, JT
Ter Meulen. Astrophys J 427: L55–L58, 1994.
254. K Tsukiyama, M Tsukakoshi, T Kasuya. Appl Phys B 50: 23–28, 1990.
255. K Tsukiyama, M Momse, M Tsukakoshi, T Kasuya. Opt Commun 79: 88–92, 1990.
256. A Mellinger, CR Vidal. J Chem Phys 101: 104–110, 1994.
257. A Mellinger, CR Vidal. Chem Phys Lett 238: 31–36, 1995.
258. A Mellinger, CR Vidal, Ch Jungen. J Chem Phys 104: 8913–8921, 1996.
259. W Kong, D Rodgers, JW Hepburn, K Wang, V Mckoy. J Chem Phys 99: 3159–
3165, 1993.
260. W Kong, JW Hepburn. J Phys Chem 99: 1637–1642, 1995.
261. W Ubachs, KSE Eikema, W Hogervorst. Appl Phys B 57: 411–416, 1993.
262. W Ubachs, I Velchev, A de Lange. J Chem Phys 112: 5711–5716, 2000.
263. PF Levelt, W Ubachs. Chem Phys 163: 263–275, 1992.
264. H Palm, F Merkt. Chem Phys Lett 284: 419–422, 1998.
265. F Merkt, TP Softley. Phys Rev A 46: 302–314, 1992.
266. JW Hepburn. J Chem Phys 107: 7106–7113, 1997.
267. ER Grant, MG White. Nature 354: 249–250, 1991.
268. J-C Loison, SH Kable, PL Houston, I Burak. J Chem Phys 94: 1796–1802, 1991.
269. SH Kable, J-C Loison, DW Neyer, PL Houston, I Burak, RN Dixon. J Phys Chem
95: 8013–8018, 1991.
270. SH Kable, J-C Loison, PL Houston, I Burak. J Chem Phys 92: 6332–6333, 1990.
271. K Tsukiyama, M Tsukakoshi, T Kasuya. J Chem Phys 94: 883–888, 1991.
272. K Tsukiyama, M Tsukakoshi, T Kasuya. J Chem Phys 92: 6426–6431, 1990.

Copyright © 2002 by Taylor & Francis

273. A Sugita, M Ikeda, K Tsukiyama. Appl Phys B 67: 253–256, 1998.
274. S Hayashi, T Suzuki, T Ichimura, K Tsukiyama. Appl Phys B 65: 555–561, 1997.
275. J Guo, A Mank, JW Hepburn. Phys Rev Lett 74: 3584–3587, 1995.
276. W Kong, D Rodgers, JW Hepburn. J Chem Phys 99: 8571–8576, 1993.
277. H Palm, F Merkt. Chem Phys Lett 270: 1–8, 1997.
278. RT Wiedmann, MG White, K Wang, V McKoy. J Chem Phys 98: 7673–7679,
1993.
279. G Nan, I Burak, PL Houston. Chem Phys Lett 209: 383–389, 1993.
280. G Nan, DW Neyer, PL Houston, I Burak. J Chem Phys 98: 4603–4609, 1993.
281. G Nan, PL Houston. J Chem Phys 97: 7865–7872, 1992.
282. GC McBane, I Burak, GE Hall, PL Houston. J Phys Chem 96: 753–755, 1992.
283. A Mank, C Starrs, MN Jego, JW Hepburn. J Chem Phys 104: 3609–3619, 1996.
284. U Berzinsh, L Caiyan, R Zerne, S Svanberg, E Biemont. Phys Rev A 55: 1836–
1841, 1997.
285. JDD Martin, JW Hepburn. Phys Rev Lett 79: 3154–3157, 1997.
286. BR Lewis, PM Dooley, JP England, K Waring, ST Gibson, KGH Baldwin, H Par-
tridge. Phys Rev A 54: 3923–3938, 1996.
287. BR Lewis, JP England, RJ Winkel, Jr., SS Banerjee, PM Dooley, ST Gibson, KGH
Baldwin. Phys Rev A 52: 2717–2733, 1995.
288. W Kong, JW Hepburn. Can J Phys 72: 1284–1293, 1994.
289. W Kong, D Rodgers, JW Hepburn. Chem Phys Lett 203: 497–502, 1993.
290. M Braunstein, V McKoy, SN Dixit, RG Tonkyn, MG White. J Chem Phys 93:
5345–5346, 1990.
291. Th Glenewinkel-Meyer, JA Bartz, GM Thorson, FF Crim. J Chem Phys 99: 5944–
5950, 1993.
292. VD Kleiman, L Zhu, X Li, RJ Gordon. J Chem Phys 102: 5863–5866, 1995.
293. R Shiell, XK Hu, Q Hu, JW Hepburn. J Phys Chem A 104: 4339–4342, 2000.
294. V Skorokhodov, Y Sato, K Suto, Y Matsumi, M Kawasaki. J Phys Chem 100:
12321–12328, 1996.
295. M Drescher, A Brockhinke, N Bowering, U Heinzmann, H Lefebvre-Brion. J Chem
Phys 99: 2300–2306, 1993.
296. SM Park, S-P Lu, RJ Gordon. J Chem Phys 94: 8622–8644, 1991.
297. JDD Martin, JW Hepburn. J Chem Phys 109: 8139–8142, 1998.
298. RG Tonkyn, RT Wiedmann, MG White. J Chem Phys 96: 3696–3701, 1992.
299. PF Levelt, KSE Eikema, S Stolte, W Hogervorst, W Ubachs. Chem Phys Lett 210:
307–314, 1993.
300. I Velchev, W Hogervorst, W Ubach. J Phys B: At Mol Opt Phys 32: L511–L516,
1999.
301. F Merkt, A Osterwalder, R Seiler, R Signorell, H Palm, H Schmutz, R Gunzinger.
J Phys B: Atm Mol Opt Phys 31: 1705–1724, 1998.
302. F Merkt, H Schmutz. J Chem Phys 108: 10033–10045, 1998.
303. JDD Martin, JW Hepburn, C Alcaraz. J Phys Chem A 101: 6728–6735, 1997.
304. F Merkt. J Chem Phys 100, 2623–2628, 1994.
305. XF Yang, J-L Lemaire, F Rostas, J Rostas. Chem Phys 164: 115–122, 1992.
306. RT Wiedmann, MG White, H Lefebvre-Brion, C Cassart-Magos. J Chem Phys 103:
10417–10423, 1995.

Copyright © 2002 by Taylor & Francis

307. F Merkt, SR Mackenzie, RJ Rednall, TP Softley. J Chem Phys 99: 8430–8439,
1993.
308. W Kong, D Rodgers, JW Hepburn. Chem Phys Lett 221: 301–306, 1994.
309. RT Wiedmann, ER Grant, RG Tonkyn, MG White. J Chem Phys 95: 746–753,
1991.
310. Y-S Cheung, J-C Huang, CY Ng. J Chem Phys 109: 1781–1786, 1998.
311. Y-S Cheung, CY Ng. Int J Mass Spectrom 187: 533–543, 1999.
312. Y Matsumi, S Nomura, M Kawasaki, T Imamura. J Phys Chem 100: 176–179,
1996.
313. Y Matsumi, SM Shamsuddin, M Kawasaki. J Chem Phys 101: 8262–8263, 1994.
314. Y Matsumi, SM Shamsuddin. J Chem Phys 103: 4490–4495, 1995.
315. CD Pibel, K Ohde, K Yamanouchi. J Chem Phys 101: 836–839, 1994.
316. A Hishikawa, K Ohde, R Itakura, S Liu, K Yamanouchi, K Yamashita. J Phys
Chem A 101: 694–704, 1997.
317. K Yamanouchi, K Ohde, A Hishikawa, CD Pibel. Bull Chem Soc Jpn 68: 2459–
2464, 1995.
318. KH Hahn, DA King, SE Harris. Phys Rev Lett 65: 2777–2779, 1990.
319. T Tsuchizawa, K Yamanouchi, S Tsuchiya. J Chem Phys 93: 111–120, 1990.
320. JC Huang, YS Cheung, M Evans, CX Liao, CY Ng, CW Hsu, P Heimann, H Lefeb-
vre-Brion, C Cossart-Magos. J Chem Phys 106: 864–877, 1997.
321. JH Werner, TA Cool. Chem Phys Lett 275: 278–282, 1997.
322. R Signorell, F Merkt. J Chem Phys 109: 9762–9771, 1998.
323. R Signorell, A Wuest, F Merkt. J Chem Phys 107: 10819–10822, 1997.
324. A Mank, T Nguyen, JDD Martin, JW Hepburn. Phys Rev A 51: R1–R4, 1995.
325. C Dorman, I Kucukkara, JP Marangos. Phys Rev A 61: 013802, 1999.
326. C Dorman, JP Marangos, JC Petch. J Mod Opt 45: 1123–1135, 1998.
327. C Dorman, JP Marangos. Phys Rev A 58: 4121–4132, 1998.
328. AJ Merriam, SJ Sharpe, H Xia, D Manuszak, GY Yin, SE Harris. Opt Lett 24:
625–627, 1999.
329. JH Werner, TA Cool. Chem Phys Lett 290: 81–87, 1998.
330. SS Dimov, RH Lipson, T Turgeon, JA Vanstone, P Wang, DS Yang. J Chem Phys
100, 8666–8672, 1994; P Wang, SS Dimov, RH Lipson. J Chem Phys 107: 3345–
3351, 1997.
331. A Mank, M Drescher, T Huth-Fehre, N Bowering, U Heinzmann, H Lefebvre-
Brion. J Chem Phys 95: 1676–1687, 1991.
332. A Mank, M Drescher, T Huth-Fehre, G Schonhense, N Bowering, U Heinzmann.
J Electron Spectrosc 52: 661–670, 1990.
333. A Mank, M Drescher, A Brockhinke, N Bowering, U Heinzmann. Z Phys D 29:
275–289, 1994.
334. DJ Gendron, JW Hepburn, J Chem Phys 109, 7205–7213, 1998.
335. L Zhu, V Kleiman, X Li, SP Lu, K Trentelman, RJ Gordon. Science 270: 77–80,
1995.
336. JA Fiss, L Zhu, K Suto, G He, RJ Gordon. Chem Phys 233: 335–341, 1998.
337. L Zhu, K Suto, JA Fiss, R Wada, T Seideman, RJ Gordon. Phys Rev Lett 79:
4108–4111, 1997.
338. JA Fiss, L Zhu, RJ Gordon. Phys Rev Lett 82: 65–68, 1999.

Copyright © 2002 by Taylor & Francis

339. RJ Gordon, L Zhu, T Seideman. Acc Chem Res 32: 1007–1016, 1999.
340. JA Fiss, A Khachatrian, L Zhu, RJ Gordon, T Siedeman. Faraday Discuss. 113:
61–76, 1999.
341. VD Kleiman, L Zhu, JA Fiss, RJ Gordon. J Chem Phys 103: 10800–10803, 1995.
342. CD Pibel, K Yamanouchi, J Miyawaki, S Tsuchiya, B Rajaram, RW Field. J Chem
Phys 101: 10242–10251, 1994.
343. CD Pibel, K Ohde, K Yamanouchi. J Chem Phys 105: 1825–1832, 1996.
344. T Tsuchizawa, K Yamanouchi, S Tsuchiya. J Chem Phys 92: 1560–1567, 1990.
345. S Liu, A Hishikawa, K Yamanouchi. J Chem Phys 108: 5330–5337, 1998.
346. DM Mao, XK Hu, YJ Shi, RH Lipson. J Chem Phys 111: 2985–2990, 1999.
347. DM Mao, XK Hu, JH Leech, YJ Shi, RH Lipson. J Chem Phys 114: 4025–4035,
2001.
348. DM Mao, XK Hu, YJ Shi, J Ma, RH Lipson. Can J Phys 78: 433–447, 2000.
349. DM Mao, XK Hu, SS Dimov, RH Lipson. J Mol Spectrosc 181: 435–445, 1997.
350. DM Mao, XK Hu, YJ Shi, RH Lipson. Chem Phys 257: 253–261, 2000.
351. CD Pibel, K Yamanouchi, S Tsuchiya. J Chem Phys 100: 6153–6159, 1994.
352. K Tsukiyama, T Kasuya. J Mol Spectrosc 151: 312–321, 1992.
353. RG Tonkyn, MG White. J Chem Phys 95: 5582–5589, 1991.
354. AV Kanaev, L Museur, MC Castex. J Chem Phys 107: 4006–4014, 1997.
355. DJ Butcher, DE Goeringer, GB Hurst. Anal Chem 71: 489–496, 1999.
356. J Li, VM Bierbaum, SR Leone. Chem Phys Lett 313: 76–84, 1999.
357. O Komienko, ET Ada, J Tinka, MBJ Wijesundara, L Hanley. Anal Chem 70: 1208–
1213, 1998.
358. JL Trevor, DE Mencer, KR Lykke, MJ Pellin, L Hanley. Anal Chem 69: 4331–
4338, 1997.
359. JL Trevor, KR Lykke, MJ Pellin, L Hanley. Langmuir 14: 1664–1673, 1998.
360. JL Trevor, L Hanley, KR Lykke. Rapid Commun. Mass Spectrom 11: 587–589,
1997.
361. N Materer, RS Goodman, SR Leone. J Vac Sci Technol A 15: 2134–2142, 1997.
362. CH Becker, KJ Wu. J Am Soc Mass Spectrom 6: 883–888, 1995.
363. P Wurz, KR Lykke. J Chem Phys 95: 7008–7010, 1991.
364. KR Lykke, P Wurz, DH Parker, MJ Pellin. Appl Opt 32: 857–866, 1993.
365. KR Lykke, DH Parker, P Wurz, JE Hunt, MJ Pellin, DM Gruen, JC Hemminger,
RP Lattimer. Anal Chem 64: 2797–2803, 1992.
366. C Koster, J Grotemeyer. Org Mass Spectrom 27: 463–471, 1992.
367. CH Becker, LE Jusinski, L Moro. Int J Mass Spectrom Ion Processes 95: R1–R4,
1990.
368. RJJM Steenvoorden, PG Kistemaker, AE De Vries, L Michalak, NMM Nibbering.
Int J Mass Spectrom Ion Processes 107: 475–489, 1991.
369. SE Van Bramer, MV Johnston. J Am Soc Mass Spectrom 1: 419–426, 1990.
370. JB Pallix, U Schühle, CH Becker, DL Huestis. Anal Chem 61: 805–811, 1989.
371. U Schühle, JB Pallix, CH Becker. J Am Chem Soc 110: 2323–2324, 1988.
372. U Schühle, JB Pallix, CH Becker, J Vac Sci Technol A6: 936–940, 1988.
373. S Ruhman, Y Hass, J Laukemper, M Preuss, H Stein, D Feldmann, KH Welge. J
Phys Chem 88: 5162–5167, 1984.
374. D Feldmann, J Laukemper, KH Welge. J Chem Phys 79: 278–282, 1983.

Copyright © 2002 by Taylor & Francis

375. K Tonokura, T Murasaki, M Koshi. Chem Phys Lett 319: 507–511, 2000.
376. S-T Tsai, J-C Jiang, YT Lee, AH Kung, SH Lin, C-K Ni. J Chem Phys 111: 3434–
3440, 1999.
377. C-K Ni, JD Huang, YT Chen, AH Kung, WM Jackson. J Chem Phys 110: 3320–
3325, 1999.
378. T Gonthiez, T Gibert, P Brault, C Boulmer-Leborgne, C Olivero, MC Castex. Appl
Phys A 69 [Suppl.]: S171–S173, 1999.
379. PL Ross, MV Johnston. J Phys Chem 99: 4078–4085, 1995.
380. PL Ross, MV Johnston. J Phys Chem 97: 10725–10731, 1993.
381. JA Bartz, DB Galloway, LG Huey, T Glenewinkel-Meyer, FF Crim. J Phys Chem
97: 11249–11252, 1993.
382. JA Bartz, TM Barnhart, DB Galloway, LG Huey, T Glenewinkel-Meyer, RJ McMa-
hon, FF Crim. J Am Chem Soc 115: 8389–8395, 1993.
383. CD Mowry, MV Johnston. J Phys Chem 98: 1904–1909, 1994.
384. SE Van Bramer, MV Johnston. Org Mass Spectrom 27: 949–954, 1992.
385. RE Bandy, C Lakshminarayan, RK Frost, TS Zwier. Science, 258: 1630–1633,
1992.
386. RK Frost, GS Zavarian, TS Zwier. J Phys Chem 99: 9408–9415, 1995.
387. RE Bandy, C Lakshminarayan, RK Frost, TS Zwier. J Chem Phys 98: 5362–5374,
1993.
388. E Nil, HE Hunziker, MS de Vries. Anal Chem 71: 1674–1678, 1999.
389. DB Galloway, JA Bartz, LG Huey, FF Crim. J Chem Phys 98: 2107–2114, 1993.
390. PG Strupp, AL Alstrin, RV Smilgys, SR Leone. Appl Opt 32: 842–846, 1993.
391. SE Van Bramer, PL Ross, MV Johnston. J Am Soc Mass Spectrom 4: 65–72, 1993.
392. SE Van Bramer, MV Johnston. Anal Chem 62: 2639–2643, 1990.
393. J Boyle, L Pfefferle. J Phys Chem 94: 3336–3340, 1990.
394. P Chen, SD Colson, WA Chupka, JA Berson. J Phys Chem 90: 2319–2321, 1986.
395. K Suto, Y Sato, CL Reed, V Skorokhodov, Y Matsumi, M Kawasaki. J Phys Chem
A 101: 1222–1226, 1997.

Copyright © 2002 by Taylor & Francis

6
Spectroscopy and Applications
of Diatomic and Triatomic
Molecules Assisted by Laser Light
at 157.6 nm
Alciviadis-Constantinos Cefalas and Evangelia Sarantopoulou
National Hellenic Research Foundation, Theoretical and Physical
Chemistry Institute, Athens, Greece

I. THE MOLECULAR FLUORINE LASER

Laser radiation from molecular fluorine transitions at 156.71, 157.48, and 157.59
nm was observed for the first time in 1978 by J. R. Woodworth and J. K. Rice
[1,2]. They used e-beam excitation to achieve population inversion in a mixture
of He (1500 mbar) and F2 (5.3 mbar). The observed stimulated emission was
assigned to the 3 Π g → 3 Π u transitions of the F2 molecule. The intensity of the
laser pulse was 7 MW/cm 2 and the efficiency was 3.5%. The experimental setup
used by Woodworth et al. was complicated and the system had to be pumped
and filled back after each shot. Plummer et al. [3] in 1979 used a different more
convenient technique by pumping mixtures of F2 and He in a fast ultraviolet
(UV)–preionized discharge. This technique proved successful for pumping of the
excimer lasers and it was more flexible. The cathode electrode whose position
was fixed consisted of a stainless steel mesh. The discharge volume could be
varied by moving one electrode in order to find the optimum electrode spacing
for higher laser output. In 1985 Cefalas et al. [4] developed a simple UV preion-
ized F2 laser of the fast-discharge type to measure the small signal gain by the
passive cell absorption method. The gain with this equipment at optimum work-
ing conditions of 2 atm total gas pressure and 2 cm electrode spacing was found

Copyright © 2002 by Taylor & Francis

to be 3.2% cm⫺1. The output energy was measured as 12 mJ per pulse and the
pulse duration was 10 ns despite the use of gases with high concentration of
impurities (0.1% impurities). With this experimental apparatus it was proved that
the molecular fluorine laser had the potential to be a powerful VUV laser similar
to the excimer lasers. Following its development, higher output energy of 15 mJ
per pulse was achieved by Ishchenko et al. [5]. The above values of output energ-
ies were half those predicted by the theoretical calculations of Ohwa and Obara
[6], if one considers only the dissociative collision of F2 molecules by ion–ion
recombination of energy transfer reaction and neglecting the direct excitation of
F2 molecules by electron impact or energy transfer from He*-, He**-, and
He *2 -excited atoms or clusters.
The predictions of Ohwa and Obara were confirmed by Yamada et al. [7],
who developed an F2 laser that delivered 112 mJ per pulse at 8 atm total gas
pressure. The electric circuit in this apparatus was similar to that developed by
Cefalas et al. (1985) of the fast charge transfer type with UV preionization. The
innovating point in the geometry of this cavity was the small distance between
the electrodes of 10 mm, providing stable discharge even at higher gas pressure
and hence improving the laser’s performance. In an effort to investigate the physi-
cal parameters and limits of the F2 molecular laser, Cefalas et al. used two dis-
charge laser heads driven by a spark-gap switch to measure the small-signal gain
and the saturation intensity in the oscillator–amplifier configuration [8,9]. The
small-signal gain coefficient was measured to be 5.2 ⫾ 0.4% cm⫺1 at 3 atm total
pressure and 1.5 cm electrode spacing. It was 4.1 ⫾ 0.4% cm⫺1 at 2 atm total
pressure and 2 cm electrode spacing. The values of saturation intensities were
found to be 5MW/cm 2 and 4.6 MW/cm 2, respectively. The dependences of the
energy output and efficiency of the F2 laser on the pump power (up to 40 MW/
cm 3 ) have been studied by Kuznetsor and Sulakshin [10]. The maximum lasing
efficiency was 0.05% and the laser radiation energy was 120 mJ at pressure of
the He–F2 mixture of 3 atm. A theoretical kinetics model successfully described
the characteristics of the F2 laser output. The authors concluded that at high gas
pressure operation, where dissociation of the F2 (A′) state is accelerated, higher
discharge pumping power is achieved [11]. The small-signal gain and the satura-
tion intensity of the discharge pumped F2 laser, operated at higher pressures (⬍10
atm) and higher excitation rates (7–39 MW/cm 3 ), were measured [12] with a
similar oscillator–amplifier configuration to that used by Cefalas et al. [8,9].
The small-signal net gain with this apparatus now reaches 37 ⫾ 4% cm⫺1 at an
excitation rate of approximately 26 MW/cm 3 for a 6-atm gas pressure. The satura-
tion intensity, which depends on a nonsaturation absorption coefficient, was esti-
mated. The output energy and the temporal behavior of a molecular F *2 laser
pumped by a coaxial electron beam have been measured in gas mixtures of He/
F2 and He/Ne/F2 [13]. The highest output energy of 172 mJ has been obtained
in a mixture of He/Ne/F2 (19.9%/80%/0.1%) at a pressure of 12 bar, correspond-

Copyright © 2002 by Taylor & Francis

ing to a specific output energy of 10.8 Joule/lt and an intrinsic efficiency of 2.6%.
An electron beam pumped molecular F*2 laser with pulse width up to 160 ns,
and output energy of 1.7 J (optical flux of 4.6 MW/cm 2 ) has been realized [14]
by the same group. The widths of the laser pulses seem to be limited by the
duration of the excitation pulse (160 ns). For specific output powers up to 100
kW/cm 3, no signs of self-terminating laser pulses, due to bottlenecking in the
lower laser level, have been observed. The application of a prepulse–main pulse
excitation scheme utilizing a saturable magnetic switch in combination with x-
ray preionization has resulted in the generation of long optical pulses from a
molecular fluorine laser [15]. Optimum laser pulse durations of 70 ns (full width
at half maximum) have been obtained in a gas mixture of helium and 3 mbar
fluorine at a total pressure of 2 bar. Laser pulse duration was limited by instabili-
ties in the electric discharge. The laser pulse duration was found to decrease with
increasing fluorine pressure and to saturate with increasing current density.

II. PHOTOLITHOGRAPHY AT 157 nm

Despite the fact that the molecular fluorine laser emits strong monochromatic
radiation at 157.6, its nontunability in the VUV is a serious drawback for a wide
range of applications. However, this disadvantage is balanced by the high energy
emitted per photon. The molecular fluorine laser has the potential to be used in
the near future in many areas of science and technology. Indeed, next-generation
microelectronics circuits will have minimum dimensions below 100 nm. It is
envisioned that 157 nm laser lithography will be the next step in optical lithogra-
phy. At 157 nm, under VUV illumination of the mask target, lithographic features
with dimensions of 0.10 µm on the photoresist have be achieved by an F2 laser
lithographic stepper developed by Exitech UK [16]. The high-repetition-rate mo-
lecular fluorine laser developed by Lambda Physik in Germany was likewise the
driving force for developing 157 nm photolithography. The first to propose the
F2 laser in photolithography was White et al. in 1984 [17]. A 2 kHz repetition
rate discharge-pumped molecular fluorine laser oscillating at 157 nm was devel-
oped. It has achieved an average power of 22 W at the repetition rate of 2 kHz
with a newly developed solid-state pulse power module with 6 J/pulse input
energy [18]. Single-line operation at 157.6 nm was achieved by means of a prism
assembly. Laser operation at repetition rates up to 1 kHz without signs of power
saturation results in an average power of 15 W. The energy stability with these
devices was comparable to the stability of the ArF laser [19].
Research in new photoresists for 157 nm photolithography [20–26] re-
quires the investigation of basic optical and photochemical studies at these wave-
lengths. The high value of the absorption coefficient of the polymeric materials
in the VUV, from 10 4 to 10 6 cm⫺1, imposes restrictions on selection of the pho-

Copyright © 2002 by Taylor & Francis

toresists for 157 nm photolithography, demanding high-purity materials and de-
fect-free thin films. There is a complete bond breaking of all the organic mole-
cules at these wavelengths. The parent molecule disintegrates into small
fragments, atomic, diatomic or triatomic, which fly apart with supersonic speeds.
This occurs because, for all the organic molecules, the dissociative excited states
of the small radicals occupy the energy range above 6.2 eV (200 nm). Therefore
the photodissociation process could impose serious problems on the optics of the
projection system by contaminating it.
To overcome this serious problem, complete microscopic description of
the photodissociation dynamics involved in the scission of the polymer chains,
and the dynamics of the extraction of the photoproducts, is required at 157 nm.
Until now various theories have been developed that analyze the role of the ex-
cited species in ultraviolet laser materials ablation. The complexity and diversity
of the processes involved in laser ablation dynamics, such as laser excitation of
the absorbing molecules and energy transfer from the excited parent molecules
into the internal and the translational degrees of freedom of the photofragments,
require for their description the analytical expression of the potential energy sur-
faces of the parent molecules and the photofragments, for both ground and the
excited states [27].
As the first contribution in this subject, and in order to investigate basic
photochemical mechanisms of the photodissociation dynamics of various poly-
meric materials at 157 nm, we have used potential photoresistive materials with
aliphatic and aromatic chains in their molecular structure [21–22]. Nylon 6.6 is
a good example. The nylon 6.6 monomer has the potential to dissociate into
small photofragments in which most of the photoresists used for 193 and 157
nm photolithography are likely to be dissociated.
Mass spectroscopy of nylon 6.6 at 157 nm [24] reveals that even at moder-
ate laser energy there was a complete breaking of the polymeric chain bonds.
The following fragments were observed for m/e larger than 30 atomic mass units
(amu). The molecular photofragments from the photodissociation of the parent
monomer were observed mainly between 20 and 30 amu (Fig. 1). Photofragments
with two carbon atoms have a relatively higher probability of being dissociated
from the parent monomer than heavier photofragments with four carbon atoms.
The polymeric material dissociates into fragments with the predominant mass at
28 amu for both laser wavelengths. Therefore the amide group is mainly involved
in the photodissociation process of nylon 6.6 in the VUV at 157 nm, which is
the case for wavelengths shorter than 248 nm [21].
Experimental findings suggest that the bound potential energy surfaces of
the excited states of the parent molecule correlate with the dissociative potential
energy surfaces of the excited states of the molecular photofragments over a wide
energy range above 5 eV. Photochemical dissociation of the aliphatic chain is

Copyright © 2002 by Taylor & Francis

Figure 1 a. Background mass spectrum of nylon 6.6 at 10⫺7 mbar b. Mass spectrum
of nylon 6.6 following photofragmentation at 157.6 nm.

the main ablative process at 157 nm. The same response to 157 nm photons has
been found in molecules with aromatic structure [22].
The experimental apparatus for obtaining the mass spectrum consists
mainly of the molecular fluorine laser source and the vacuum chamber, into which
the quadrupole mass spectrometer and sample under investigation were placed
(Figs. 2, 3). The sample was a membrane of pure amorphous nylon 6.6, 150 µm
thick. The density of the sample was 0.25 gr/cm 3 and it was placed 1 mm apart
from the quadrupole mass spectrometer (Baltzers QMG 311) at right angles to
its axis (Fig. 2). The all stainless steel 316 vacuum chamber was evacuated to
10⫺6 mbar using a turbomolecular pump.
Laser sources used in these experiments were of the fast-discharge type,
which have been described previously [8,9]. The laser head delivers 10 ⫾ 1 mJ
per pulse, and the pulse duration was ⬃15 ns at FWHM.
The laser beam was focused on the nylon 6.6 sample, using a quartz lens of
40 cm focal length. After photodissociation of the parent molecule, the molecular
photofragments were ionized using an electron gun inside an isopotential cham-
ber (Welnet). The molecular ions were focused with an Eizen lens, and eventually
directed alongside the quadrupole mass filter. The upper detection limit of this
filter was 300 amu. After entering the mass filter, ions were deflected at right
angles, and detected using a high-gain (⬃10 8 ) secondary electron multiplier

Copyright © 2002 by Taylor & Francis

Figure 2 Experimental setup for obtaining mass spectrum of nylon 6.6. LB, laser beam;
L, lens; VC, vacuum chamber; TM, turbomolecular pump; W, LiF, window; S, sample;
P, photofragments; QMS, quadrupole mass spectrometer.

Figure 3 157 nm F2 laser and mass spectrometer at the NHRF.

Copyright © 2002 by Taylor & Francis

(SEM). The signal was then amplified and registered using a boxcar integrator
and a computer. Using the same experimental techniques, several polymeric ma-
terials have been studied for both VUV absorption below 200 nm and photodisso-
ciation dynamic processes under 157 nm illumination. Si-based polymers seem
to have the right value of absorbance of 4 µm⫺1 at 157 nm and therefore they
can be imaged at the usual bilayer thickness of 0.1 µm. However, excessive
outgassing at 157 nm at the moment is a serious drawback for their use in 157
nm photolithography [20]. Mass spectroscopic and outgassing studies at 157 nm
suggest that a complete bond breaking of the aromatic and linear carbon chains
is taking place even at low values of laser energy [20]. However, fluorocarbon
polymeric materials seem to have the right absorption coefficient in the VUV
and limited outgassing at 157 nm.

III. 157 nm SPECTROSCOPY OF DIATOMIC

AND TRIATOMIC MOLECULES
A. OH
Photodissociation of the OH radical was studied at 157 nm via detection of the
product H atoms with the resonance fluorescence technique [28]. OH radicals
were produced in a fast-flow cell from the reaction between H and NO 2 and
subsequently photodissociated by an excimer laser operating on the F2 emission.
The quantum yield for photodissociation of OH was measured and the photodis-
sociation cross section calculated.

B. O 2
Direct photodissociation of molecular oxygen from the B 3 ∑ u⫺1 state have been
studied by photoexcitation at 157 nm. The fine structure branching ratios and
Doppler profiles of O( 3 Pj ) photofragments were measured using a resonance-
enhanced multiphoton ionization technique. Branching ratios of the O( 3 Pj ), j ⫽
2, 1, 0, photofragments were measured [29]. The fine structure population of
O(P3 ) produced in photodissociation of O 2 at 157 nm was measured in a pump
and probe experiment [30]. The population of the ground state (J ⫽ 2) was found
to be 93%, with only 6% in the J ⫽ 1 level and 1% in J ⫽ 0. The predominance
of J ⫽ 2 is consistent with direct dissociation on the B 3 ∑ u potential energy
surface. Based on available information in the literature, excitation of the 3 Π u
state and curve crossing to the 5 Π u state are ruled out as possible origins of the
population with J ⬍ 2. The populations of J ⬍ 2 are due to nonadiabatic transi-
tions at large distances. Mechanisms consistent with the data include a Demkov-
type interaction, which is caused by coupling by the radial kinetic energy opera-
tor. Photodissociation of O 2 at 157 nm has been studied using the photofragmen-

Copyright © 2002 by Taylor & Francis

tation translational spectroscopic technique [31]. Two product channels O 2 ⫹
hν → O( 1D) ⫹ O( 3P), O 2 ⫹ hν → O( 3P) ⫹ O( 3P) have been observed. The
relative yields and anisotropy parameters of both channels were determined. An-
isotropy mixing of dissociation resulting from a perpendicular excitation and a
parallel-type excitation has been observed in the dissociation channel O 2 ⫹ hν →
O( 3P) ⫹ O( 3P). The observed results can be used to look at the detailed dynamical
processes of O 2 dissociation through the Schumann-Runge band.

C. HCl, DCl, CH 3 Cl
Hydrogen chloride and methyl chloride were photodissociated at 157 nm [32].
Branching ratios for production of the resulting chlorine atoms in the 2 P1/2 state
relative to the 2 P3/2 state were determined at 157 nm for HCl and CH 3 Cl. Doppler
profiles of the chlorine fragments have been measured and are interpreted by
assuming their anisotropy parameters and translational energies. HCl undergoes
a perpendicular optical transition. For CH 3 Cl, the transition is a mixture of paral-
lel and perpendicular types at 157 nm.
The H, D, and Cl atoms from the photodissociation of HCl and DCl at 157
were detected by laser-induced fluorescence (LIF) in the vacuum ultraviolet re-
gion [33]. Doppler profiles of H and D resonance lines at 121.6 nm in the LIF
spectra indicate that the absorption of HCl and DCl at 157 nm is a mixture of
perpendicular and parallel transitions. Fine-structure branching ratios were mea-
sured for the chlorine atom by LIF. Results suggest that nonadiabatic couplings
during the break-up of HCl (DCl) in the excited states play an important role in
determining the fine-structure branching ratios.

D. H2
The photochemical desorption of molecular hydrogen was investigated by F2 laser
irradiation of a Si(111)-(1*1): H surface [34]. The photon energy of 7.9 eV used
was in the region of the broad sigma–sigma* optical transition of SiH centered
around 8.5 eV. Molecular dynamics calculations, based on a Tersoff-type interac-
tion potential between silicon and hydrogen, describe the reaction of a hydrogen
atom created in a direct bond-breaking process with a neighboring hydrogen atom
to form molecular hydrogen. This secondary surface reaction preserves the non-
thermal character of the desorption process.

E. OCS
The production of significant concentrations of electrons in high-intensity pulsed-
laser photolysis of carbonyl sulfide (OCS) to S( 1 S) at 157 nm has been observed
[35]. These results imply a photoionization cross section for S( 1 S) of 2.6 ⫻ 10⫺19

Copyright © 2002 by Taylor & Francis

cm 2. The authors have measured the quenching of S( 1 S) for a variety of condi-
tions, including incident laser fluences of 100 mJ/cm 2 and initial S( 1 S) densities
of 3 ⫻ 10 16 cm⫺3. Quenching by electrons can be minimized by adding diluents
such as SF6 or CF4 . They interpret their observed S( 1 S) decay rates in terms of
quenching by S( 3 P) with a rate constant of 3 ⫻ 10⫺11 cm 3 s⫺1 and by CS [and/
or O( 3 P)] with a rate constant near 2 ⫻ 10⫺10 cm 3 s⫺1. The measured S( 1 S) lifetime
decreases with increasing initial S( 1 S) density, reaching a value of about 1 µs at
a S( 1 S) density of 3 ⫻ 10 16 cm⫺3. The quantum yield for S( 1 S) production was
found to be high (near 0.8) and independent of incident laser flux or added diluent
pressure. These storage times and production efficiencies are consistent with pos-
sible use of the S( 1 S) to S( 1 D) transition as an efficient high-energy-storage laser.
A F2 laser at 157 nm was also used in a time of flight experiment to measure
the translational energy distribution of the fragments of OCS from which the
vibrational energy distribution was inferred [36]. It is highly inverted in spite of
the fact that the CO bond distances are similar in CO and OCS. Vibrational
population inversion, the authors believe, is related to the vibrational structure
seen in absorption. A symmetrical type of vibration is excited during the dissocia-
tion,which tends to pull both the O and S atoms away from the C atoms. The
quasivibrational trajectory ends with the sulfur atom separating, but in the process
the CO vibration is strongly excited.

F. H 2O
Polarized photofluorescence excitation spectroscopy of H 2 O was described by
Andresen et al. [37]. The H 2 O* state (A 1 B 1 ) 3 dissociates directly and rapidly
with 2.7 eV excess energy. Water molecules were photolyzed by an F2 laser at
157 nm [38,39]. The nascent OH 2 Π 3/2 , rotational state distribution, probed via
laser-induced fluorescence, reveals a strong preference for populating the upper
lambda-doublet component. The population inversion is found to be a function
of both the initial temperature of the H 2 O and the final rotational state in which
the OH radical is formed. These results may provide a simple mechanism for
the astronomical OH maser observed by others.
H 2 O was excited to the lowest electronic excited state, 1 B 1 , at 157.6 nm.
The OH product state distribution was completely analyzed by special LIF experi-
ments probing the fragment distribution [40]. The OH rotational excitation is
relatively low and can be described by a temperature parameter of approximately
500 K independently of the OH vibrational excitation. This is in accordance with
former measurements as well as with theoretical calculations. No selective popu-
lation of the electronic fine-structure levels was observed, which is in agreement
with the expectations for a room-temperature experiment. The observed magni-
tude of the vibrational excitation [P(υ″ ⫽ 0): P(υ″ ⫽ 1): P(υ″ ⫽ 2) :P(υ″ ⫽ 3) :
P(υ″ ⫽ 4) ⫽ 59.2: 33.1:6.1 :1.4 :0.2] is smaller than the one calculated. Calcula-

Copyright © 2002 by Taylor & Francis

tions predicted the υ″ ⫽ 6 level to be populated, whereas in the experiment,
transitions probing the υ″ ⫽ 5 state were not observed.
In similar experiments vibrationally mediated photodissociation, in excita-
tion of an overtone of an O–H stretching vibration, has been used. The vibra-
tionally excited molecule is dissociated with a second photon and initial vibra-
tional excitation alters the decomposition dynamics. For example, the amount of
vibrationally excited OH produced in vibrationally mediated photodissociation
of H 2 O depends strongly on the eigenstate prepared in the vibrational overtone
excitation step. This is in agreement with theoretical calculations [41]. The au-
thors describe experiments that exploit this sensitivity to control breaking of the
O-H bond in HOD.
Laser-induced fluorescence of microwave-stimulated OH molecules from
H 2 O photodissociation was investigated as a first step in a series of laboratory
experiments to understand features of astronomical OH masers [42]. The inver-
sion between the Π 3/2 J ⫽ 7/2 lambda-doublet states of OH generated via photo-
dissociation of cold H 2 O at 157 nm is shown to be 1.8:1. Within a microwave,
Fabry–Perot cavity tuned to the resonance of one of the main hyperfine transitions
in this lambda doublet, it is possible to stimulate all inverted OH molecules. The
linewidth of the two main microwave transitions is measured as a function of
microwave power and interaction time.

G. ICN
Dispersed CN B 2 ∑⫹ ⫺ X 2 ∑⫹ photofragment fluorescence polarization anisotrop-
ies measured [43] following iodo cyanide (ICN) dissociation at 157.6 nm vary
widely and apparently erratically with emission wavelength. They cannot be con-
verted directly to CN B 2 ∑⫹ rotational alignments because of spectral congestion.
A novel linear regression technique is used to extract CN B 2 ∑⫹ populations
and rotational alignments from fluorescence emission and polarization anisotropy
measurements. The authors present a flexible procedure that allows one to con-
sider many models for the population and alignment distributions. Criteria are
established to identify the best models. The CN B 2 ∑⫹ vibrational branching
ratios for υ′ ⫽ 0:1:2: 3: 4 are determined by linear regression to be 0.46 :0.25:
0.13 :0.09 :0.07, with a distinct rotational population dependence within each vibra-
tional level. Extracted CN B 2 ∑⫹ alignments for υ′ ⫽ 0,1,2, and 3 are presented,
and these range from ⫺0.31 to nearly 0.2. The alignments vary smoothly with
nuclear rotation N′ for each υ′, demonstrating that the scatter in the measured polar-
ization anisotropies results from vibrational band overlap at different wavelengths.
These results show the largest photofragment alignment variation with vibration
and rotation that has been measured following a single-photon dissociation process.
A model is presented to estimate partial channel CN B(υ′ ⫽ 0) product populations,
and a discontinuity in the experimental υ′ ⫽ 0 alignment is considered.

Copyright © 2002 by Taylor & Francis

H. HDO
Experimental values for the absorption coefficient and the branching ratio of
HDO at 157 nm were compared with theoretical values [44].

I. AsF3
Visible fluorescence from one-photon excitation of AsF3 at 157 nm was dispersed
and the fluorescing species were attributed to the excited photofragment of AsF2
[45]. The radiative lifetime of AsF*2 was measured to be 25.5 ⫾ 1.8 µs. The
quenching rate constant of AsF 2* by AsF3 was (1.51 ⫾ 0.05) ⫻ 10⫺10 cm 3 s⫺1.
Ultraviolet AsF (A-X) emission is observed from two-photon excitation of AsF3 .

J. O 3
Ozone was generated in pure oxygen (p ⬃ 5 kPa), synthetic air (p ⬃ 7 kPa),
and oxygen–argon mixtures (p ⬃ 3 kPa) by irradiation of these gases with the
VUV light of a repetitively pulsed (15 Hz) F2 laser at 157.6 nm with maximum
energy of about 4 mJ/pulse [46]. An absorption photometer measurement op-
erating at 253.7 nm (Hg line) determines the ozone concentration as a function
of oxygen and/or additive gas pressure, the repetition frequency of the laser, and
the wall temperature of the reaction chamber. The temporal development of ozone
concentration as a function of these parameters is calculated by means of rate
equations for the species O( 3 P), O 2 (X 3 ∑ g⫺ ), O 3 ( 1 A 1 ), O( 1 D), O 2 (a 1 ∆ g ),
O 2 (b 1 ∑ g⫹ ), vibrationally excited O 3* ( 1 A 1 ), and the photon distribution. The max-
imum concentration of O 3 in the sealed-off chamber reaches 1.6% in pure O 2 ,
4.1% in air, and 1.2% in a 1:5 O 2 /-Ar mixture at 3 kPa. The annihilation of O 3
by the wall and temperature-dependent volume processes (300 K ⬍ T ⬍ 395 K)
was studied and experimental and theoretical results were compared.

K. CO 2
The branching ratio was measured for the production of O( 3 P) in the photodisso-
ciation of CO 2 at 157 nm [47]. A gas mixture consisting of CO 2 , H 2 , and Ar was
irradiated with an F2 excimer laser, while the relative concentration of O( 3 P) was
monitored continuously using atomic resonance fluorescence. The O( 1 D) product
was removed by either reacting with H 2 or being quenched by CO 2 . At high H 2
CO 2 ratio, a residual O( 3 P) signal persisted due to the nascent photofragments
of CO 2 . Stern–Volmer analysis indicated that the fraction of O( 3 P) produced is
5.9%. Control experiments using O 2 and N 2 and O as precursor molecules con-
firmed this interpretation of the data. A mechanism is proposed based on curve
crossing from 2 B 2 to 3 B 2 potential energy surfaces of CO 2 . Since the 1 B 2 state is

Copyright © 2002 by Taylor & Francis

bent, a substantial fraction of absorbed energy is initially in bending motion,
resulting in a long-lived chaotic trajectory that has many opportunities to cross
over to the triplet surface.
O(2p 3 Pj ) ( j ⫽ 2, 1, and 0) fragments produced in the 157 nm photodissoci-
ation of CO 2 were detected by resonance-enhanced multiphoton ionization in a
molecular beam [48]. Doppler profiles and fine-structure branching ratios were
measured for the oxygen-atom photofragment in the 3 Pj states. Doppler profiles
were analyzed to give an anisotropy parameter of β ⫽ 2.0 and an internal energy
equivalent to 3.9 vibrational quanta of CO. The fine-structure populations were
found to be 0.70, 0.16, and 0.14 for j ⫽ 2, 1, and 0, respectively. A mechanism
is proposed in which complex on the 1 B 2 surface undergoes intersystem crossing
to the 3 B 2 surface. A phase-space model with a constraint on the impact parameter
is shown to be consistent with the observed energy release. The nonstatistical
fine-structure population could be caused by long-range interactions on the triplet
surface. In a bulb experiment, O( 3 P) was produced by quenching of O( 1 D). The
fine-structure populations of the resulting O( 3 Pj ) were 0.64, 0.25, and 0.11. This
state distribution is consistent with a long-lived complex that decays to give statis-
tical products.
Vibrational and rotational distributions of CO( 1∑ g⫹) produced in the 157
nm photodissociation of CO 2 have been determined by measuring vacuum-ultra-
violet laser-induced fluorescence spectra of the CO photoproduct [49]. The photo-
dissociation of CO 2 is known to occur via two pathways; one yielding O( 1 D) and
the other yielding O( 3 P). Spin conservation and previous experimental studies
confirm that dissociation via the O(1D) channel is the dominant process. The
available energy for this channel is sufficient to populate only the ground and
first excited vibrational levels of CO. The authors measured the rotational distri-
butions for CO in υ ⫽ 0 and υ ⫽ 1 and found them to be non-Boltzmann. In
fact, a highly structured distribution with distinct peaks at J ⫽ 10, 24, 32, and
39 is observed for CO in υ ⫽ 0. A less structured population is displayed by
molecules in υ ⫽ 1. The relative vibrational population (υ ⫽ 0/υ ⫽ 1) was
determined to be 3.7. Doppler spectra of individual rovibronic transitions were
also recorded. The profiles have widths in accord with the available translational
energy, display the expected v perpendicular to J correlation, and are best de-
scribed by an isotropic distribution of the velocity vectors with respect to the
polarization direction of the dissociation light.
Photodissociation of CO 2 at 157 nm was studied by the photofragment-
translational spectroscopy technique. Product time-of-flight spectra were re-
corded and center-of-mass translational energy distributions were determined
[50]. Two electronic channels were observed; one forming O( 1 D) and the other
O( 3 P). With previously determined anisotropy parameters of β ⫽ 2 for the O( 3 P)
channel and β ⫽ 0 for the O( 1 D) channel, an electronic branching ratio of 6%
O( 3 P) was obtained, consistent with previous results. The translational energy

Copyright © 2002 by Taylor & Francis

distribution for the CO(υ) ⫹ O( 3 P) channel was very broad (over 30 kcal/mol)
and appeared to peak near CO(υ ⫽ 0). The value of β ⫽ 2 for the O( 3 P) channel
was confirmed by comparing Doppler profiles, derived from the authors’ mea-
sured translational energy distribution, with previously measured Doppler pro-
files. This suggests that the O( 3 P) channel arises from a direct transition to an
excited triplet state. The O( 1 D) channel had a structured time-of-flight that related
to rovibrational distributions of the CO product. The influence of the excitation
of the CO 2 (V 2 ) bending mode was investigated and shown to have a small but
not negligible contribution. Based on a comparison of the authors’ data with a
previous VUV laser-induced fluorescence study, they obtain as their best estimate
of the vibrational branching ratio CO(υ ⫽ 0)/CO(υ ⫽ 1) ⫽ 1.9, for the CO(υ)
⫹ O( 1 D) channel.

L. SiH
SiH bonds on a hydrogenated Si(111) surface were directly broken by electronic
excitation with the 7.9 eV photons of a F2 laser [51]. Independence of the kinetic
energy of the desorbing hydrogen and the linear dependence of the desorption
yield on fluence prove the photochemical character of the photoprocess. Solution
of the heat diffusion equation indicates a negligible surface heating for the flu-
ences applied.

IV. VUV SPECTROSCOPY OF TRIATOMIC CLUSTERS

Investigation of the properties of small charged or neutral clusters is a subject

of growing interest because they bridge the gap between a solid and molecular
state of matter [52–59]. Experimental techniques developed to study the clusters
and their ions can be classified into two main categories: the optical and the
mass spectroscopic. The use of laser spectral techniques such as laser-induced
fluorescence or two-photon ionization has been proved unsuccessful for large
clusters. This is because both techniques require that the laser-induced transition
populate an excited electronic state that survives long enough to be detected by
either its fluorescence or its lowered ionization threshold. The large metal clusters
have hundred of electronic states within the first few electronvolts. Those states,
even if they interact weakly with each other, produce a very dense magnifold of
vibronic levels. As a result, very complicated spectra arise or radiationless transi-
tions between those levels that quickly lead to a degradation of electronic excita-
tion. On the other hand the detection of cluster ion using mass spectroscopy is
free of such problems for both large and small clusters. Mass spectroscopic stud-
ies provide basic information about the dissociation energy and kinetics of the

Copyright © 2002 by Taylor & Francis

clusters. This can be expressed by the appearance of strong cluster peaks in the
mass spectrum denoting the stability of the particularly cluster structures.
It has been observed experimentally that stable clusters can exist in various
sizes containing a definite number of atoms for a specific charge. This number is
called ‘‘magic number’’ [59–62]. A key problem is to find a sufficiently sensitive
detection scheme to detect them. Several models have been proposed to predict
or explain the experimentally observed ‘‘magic number’’ [63–68].
Investigation of the electronic structure and dynamics of clusters using vac-
uum ultraviolet spectroscopy is very important area in the study of clusters. VUV
experimental techniques have two advantages: systems with a large band gap can
be studied and excitation of inner shell electrons allow us to obtain element-
specific information.
The application of a tunable, intense, and coherent VUV light source to
molecular spectroscopy and photodissociation chemistry is proposed by K. Ya-
manouchi [69] to study the high-lying electronically excited states of small jet-
cooled molecules and van der Waals clusters. Combination of the VUV laser,
generated by two-photon resonance, four-wave sum, or difference frequency mix-
ing scheme, and the free jet expansion techniques is promising to derive precise
and substantial information about the level structure and the dynamic processes
characteristic of highly excited molecular species excited in the VUV wavelength
region.
Moller et al. [70] provide an overview of recent developments in the field
of cluster research using VUV radiation. Electronic excitation and relaxation dy-
namics in helium clusters, and the inner shell photoionization of NaCl clusters
produced with a new pick-up cluster source, are also discussed. The near edge
absorption structure at the Cl 2p edge contains information on the geometric
structure of clusters.
Clusters can be formed in a molecular beam apparatus [71–75], in an oven
using the gas aggregation technique [76–80], and by desorption from a surface
by ion beam bombardment [81–83], but they can also be the result of photodisso-
ciation of a parent molecule [84–87].
Ionization of clusters can be achieved by electron impact [88–90], by one-
photon ionization with VUV [91–93] light, by synchrotron radiation or multipho-
ton ionization with UV or visible laser light [94–96], by liquid metal ion source
LMIS [97–99], and by laser-induced field vaporization (LIFV) [68,100–101].
An interesting experimental setup for photoluminescence spectroscopy on van
der Waals clusters has been described by R. Karnbach et al. [102]. It consists of
a molecular beam apparatus with a cluster beam installed behind a high-intensity
VUV synchrotron radiation beamline. Special emphasis was given to the design
of a very intense cluster source that can also be used for preparation of quantum
clusters (He, H2 ). To determine the cluster size, a time-of-flight mass spectrome-
ter can be attached to the setup. In addition, an atomic cross jet is installed in

Copyright © 2002 by Taylor & Francis

the experimental chamber that can be used for mass separation or for doping of
the clusters. Luminescence light can be recorded with several different detectors
or spectrally analyzed with a secondary monochromator equipped with a position-
sensitive detector. The pulsed nature of synchrotron radiation provides the basis
for time-resolved measurements in the regime 100 ps–3 µs.
In the work of Wucher et al., neutral atoms and clusters desorbed from a
solid germanium surface by 5 keV Ar⫹ ion bombardment [74,103]. Charged clus-
ters were detected directly by time-of-flight mass spectrometry (TOF-MS). The
corresponding neutral species were postionized prior to mass and energy analysis.
These authors employed two different photoionization schemes. The first was a
single-photon ionization process using an F2 molecular laser as an intense VUV
source with photon energy in excess of all relevant ionization potentials. The
second was a nonresonant multiphoton ionization process using a high-intensity
laser delivering pulses of 250 femtosecond duration at a wavelength of 267 nm.
The two processes were compared, and in both cases the available laser pulse
energy was sufficient to saturate the ionization of Ge atoms and all detected Ge 2 ,
Ge 3 , and up to Ge 7 clusters. This is an indication that the results obtained with
both photoionization techniques closely reflect the true cluster sputtering yields
and, in particular, are not dominated by photon-induced fragmentation.
Relative yields of sputtered Ge n clusters are found to obey a power law
dependent on the cluster size n with an exponent around ⫺6.5 [74]. Kinetic en-
ergy distributions exhibit a shift of the maximum toward lower energies with
increasing cluster size. Asymptotic decay towards high energies is found to be
virtually identical for all measured atoms and clusters. The experiments were
repeated on two different single crystalline Ge surfaces ⬍100⬎and⬍111⬎ that
could be reproducibly amorphized by ion bombardment and reannealed. As a
result, the crystalline structure of the ion-bombarded surface was not found to
play a significant role in the formation of sputtered clusters.
Neutral clusters of Ag, Al, Nb, and Ta were also formed using the technique
of sputtering of polycrystalline surfaces. They were studied experimentally by
nonresonant single-photon postionization using a UV or VUV laser beam and
time-of-flight mass spectrometry [104]. The mass spectra were recorded at differ-
ent laser intensities. As a result, photoionization cross sections and the yields of
sputtered clusters were determined as a function of the cluster size. The cluster
yields roughly exhibit a power law dependence on the cluster size, the exponent
of which is found to be inversely correlated with the sputtering yield of the sam-
ple. This finding is of particular importance, since it rules out simple statistical
combination models to describe the formation of large sputtered clusters. From
the yield distributions it is inferred that, depending on the sputtering conditions,
up to 46% of the sputtered atoms may be emitted in a bound state. Experimental
results were compared with theoretical model descriptions of the cluster forma-
tion process.

Copyright © 2002 by Taylor & Francis

 Neutral silver atoms and small clusters Ag n (n ⫽ 1 . . . 4) were generated
by bombarding a polycrystalline silver surface with Ar⫹ ions of 5 keV [105,106].
The sputtered particles were ionized by a crossed electron beam or by the F2
single photon at 7.9 eV from a pulsed VUV laser that permits nonresonant single-
photon ionization (SPI) of all investigated species and is subsequently detected
by a quadrupole mass spectrometer. Photoionization cross sections were evalu-
ated from the laser intensity dependence of the measured mass spectrum. SPI
ionization cross sections do not vary dramatically between silver atoms and dif-
ferent clusters. As a consequence, fragmentation influences encountered in previ-
ous studies with longer-wavelength lasers are practically eliminated from the
determination of yields and kinetic energy distributions of the sputtered clusters.
The resulting relative cluster sputtering yields (normalized to the yield of silver
atoms) exhibit a power law dependence on the cluster size n according to n-delta
with exponents of delta ranging from 4.3 to 7.4 depending on the nature and the
bombarding energy of the primary ions. Kinetic energy distributions of the sput-
tered neutral atoms and clusters are evaluated up to clusters containing seven
atoms. Asymptotic decay of the energy distribution towards high-emission energ-
ies becomes steeper from Ag to Ag 3 and remains practically constant for larger
clusters. By in situ combination of both ionization mechanisms, absolute values
of the ratio σ(e) (Ag( n ))/σ(e)(Ag) between the electron impact ionization cross
sections of silver clusters and atoms could be determined for a fixed electron
energy of 46 eV. These values can then be used to calibrate previously measured
relative ionization functions. By calibrating the results using literature data mea-
sured for silver atoms, we present absolute cross sections for electron impact
ionization of neutral Ag 2 , Ag 3 , and Ag 4 as a function of the electron energy
between threshold and 125 eV.

V. LASER-INDUCED FLUORESCENCE AND MASS

SPECTROSCOPY OF MERCURY DIATOMIC
AND TRIATOMIC CLUSTERS ASSISTED
BY LASER LIGHT AT 157.6 nm

Laser-induced fluorescence of the mercury clusters Hg 2 and Hg 3 in the spectra

range from 300 to 510 nm has been obtained from the dissociation of HgBr 2 at
157.6 nm with an F2 molecular laser. The excitation process involves two-photon
absorption, which dissociates the molecule at 15.76 eV total photon energy with
the subsequent formation of metallic clusters [85].
Photodissociation of HgBr 2 at 157.5 nm using the F2 molecular laser was
proven to be an efficient method for producing small mercury clusters Hg 2 and

Copyright © 2002 by Taylor & Francis

Hg 3 in their excited states [107]. Although photodissociation of mercury halides
by the ArF excimer laser at 193 nm (and at other excimer and dye lasers wave-
lengths) has been extensively studied [108–113], there are no reports of metallic
cluster formation at this wavelength. In addition, a detailed study of the spectros-
copy of the triatomic mercury clusters in electric discharges has been investigated
well before VUV optical excitation [114,115].
The electronic structure of HgBr 2 and the HgBr molecules indicates that
three-dipole allowed transitions can be excited between the low-lying electronic
states of the HgBr 2 molecule, 1 1 Πu, 1 1 ∑ u⫹ , 2 1 ∑ u⫹ , and its ground state 1 1 ∑ g⫹
[116–119]. VUV excitation at 157.5 nm stimulates the 1 1 ∑ g⫹ → 2 1 ∑ u⫹ transition,
(Fig. 4).

HgBr 2 (1 1 ∑ g⫹ ) ⫹ hν(7.88 eV) → HgBr 2 (2 1 ∑ u⫹ ). (1)

The 2 1 ∑ u⫹ state of HgBr 2 at 7.88 eV correlates with the the D( 2 Π 3/2 ) and C( 2 Π 1/2 )
states of HgBr and the 2 P state of Br.

Figure 4 Energy pathway of formation of mercury atoms from the dissociation of HgBr 2
at 157.6 nm.

Copyright © 2002 by Taylor & Francis

 HgBr 2 (2 1 ∑ u⫹) → HgBr 2 ( 2 Π 1/2 ) ⫹ Br( 2 P)
(2)
HgBr 2 (2 1 ∑ u⫹) → HgBr 2 ( 2 Π 3/2 ) ⫹ Br( 2 P)
The 2 1 ∑ u⫹ electronic state is strongly correlated with D( 2 Π 3/2 ) and C( 2 Π 1/2 )
electronic excited states of the HgBr diatomic following dissociation to HgBr
and Br.
The small energy gap between the D state and the available energy from
one photon at 157.5 nm can be bridged thermally at high temperatures. Therefore
a fraction of the population is transferred to the D state [120] and the remaining
part to the C state of HgBr. From the C and D states the molecules can either
radiate to the X state or can absorb a second 157.5 nm photon and photoionize
the HgBr molecule. The C → X transition at 285 nm is the strongest transition
in the spectrum, while the D → X transition at 265 nm is ⬃10 times weaker.
The overall absorption cross section at 157.5 nm has been found to be (4.5 ⫾
2.6) ⫻ 10⫺16 cm 2 [107] in the temperature range between 40 and 90°C. This is
considerably higher than the value of the absorption cross section at 193 nm,
which has been found to be 3.8 ⫻ 10⫺17 cm 2 [121]. However, the theoretical
calculations of Wadt [116] predict that the absorption cross section between 1
∑ g and 2 1 ∑ u⫹ states at 157 nm should be smaller than the absorption cross
1 ⫹

section between 1 1 ∑ g⫹ and 1 1 ∑ u⫹ states at 193 nm by a factor of 35. The above

discrepancy between the experimental and theoretical values indicates that other
loss mechanisms than the absorption between the 1 1 ∑ g⫹ and 2 1 ∑ u⫹ states should
play a major part in the annihilation of VUV photons by the HgBr 2 molecules.
These absorption mechanisms include one-step ionization of HgBr from the C
and D states and two-step ionization of HgBr 2 from its ground state. The C and
D states of HgBr at 7.88 eV correlate with the 6 3 P0, 1 metastable states of Hg
either via one-photon absorption and subsequent deexcitation to the 6 3 P0, 1 states:

HgBr( 2 Π 1/2 , 2 Π 3/2 ) ⫹ hν(7.88 eV) → HgBr⫹(15.76 eV)

HgBr⫹(15.76 eV) → Hg(6D, 7S) ⫹ Br( 2 P) (3)
Hg(6D, 7S) → Hg(6P) ⫹ hν
or through thermal dissociation according to
HgBr( 2Π 1/2 , 6 3 P0, 1 ) ⫹ kT → Hg(6 3 P0 , 6 3 P1 ) ⫹ Br( 2 P) (4)
The energy gap between the 2 Π 3/2 state of HgBr and the Hg(6 3 P0 ) state of mercury
is between 0.58 and 0.53 eV [122,123]. Therefore, at 300°C less than ⬃0.1% of
the population of the diatomic is thermally dissociating to the Hg(6 3 P0 ) ⫹ Br( 2P)
atoms. The observed emissions at 404.6, 435.8, 407.8, 312, 313, 296, 365, 366,
and 254 nm are from the allowed dipole transitions from the 7S, 6D, and 6P
states of mercury

Copyright © 2002 by Taylor & Francis

 Hg(7 1 S 0 ) → Hg(6 3 P1 ) ⫹ hν(407.8 nm)
Hg(7 3 S 0 ) → Hg(6 3 P1 ) ⫹ hν(4358 nm)
Hg(7 1 S 0 ) → Hg(6 3 P0 ) ⫹ hν(404.6 nm)
Hg(6 1 D 2 ) → Hg(6 3 P2 ) ⫹ hν(365.5 nm) (5)
Hg(6 3 D 2 ) → Hg(6 3 P1 ) ⫹ hν(312.6 nm)
Hg(6 3 D 1 ) → Hg(6 3 P1 ) ⫹ hν(313.1 nm)
Hg(6 3 P 1 ) → Hg(6 1 S0 ) ⫹ hν(253.6 nm)
Hg(6 3 D 1 ) → Hg(6 3 P0 ) ⫹ hν(296.0 nm)
The kinetic processes associated with the formation of Hg 2 (AO g⫾) from 6 3 P0, 1
and 6 1 S 0 states have been studied in detail [124]. Spectroscopy of the Hg 2 mole-
cule has been given by the work of Niefer et al. [125], Callear et al. [126], and
Drullinger et al. [127]. A collision between a metastable mercury atom 6 3 P0 and
a ground state atom 6 1 S 0 can create the mercury dimer Hg 2 in its excited states
AO g⫾. Collision between a 6 3 P0 mercury atom and a 6 1 S 0 ground state atom can
form a mercury dimer either in the AO g⫹ state or in the Dl u state, depending on
the thermal energy available. The AO g⫾ states 2800 cm⫺1 below the optically
active Dl u state either can act as an energy reservoir for the formation of the Hg 3
cluster or can thermally populate the Dl u state of Hg 2 , which gives the 335 nm
band through de-excitation to the dissociative ground state.
Hg(6 3 P0 ) ⫹ Hg(6 1 S 0 ) → Hg 2 (AO g⫺ )
Hg(6 3 P0 ) ⫹ Hg(6 1 S 0 ) → Hg 2 (AO g⫹ )
→ Hg 2 (Dl u )
Hg(AO g⫾ ) ⫹ kT(0.34 eV) → Hg 2 (Dl u )
Hg(Dl u ) → 2Hg(6 1 S 0 ) ⫹ hν(335 nm)
The spectral lines detected in the region between 410 and 426 nm originate
from the de-excitation of the Hg 2 (Pl u ) Rydberg state to the AO g⫹ state.
The Hg 2 (Pl u ) state is created through a collision between an Hg(7 3 S 1 ) atom
with a ground state atom Hg(6 1 S 0 )
Hg(7 3 S 1 ) ⫹ Hg(6 1 S 0 ) → Hg 2 (Pl u )
Hg 2 (Pl u ) → Hg 2 (AO g⫹ ) ⫹ hν(410–426 nm)
This de-excitation channel is another pathway of populating the reservoir of the
AO g⫾ states.
The 335 nm band is much narrower than the one previously reported from
electric discharges [128]. Transition from the 6 3 D 1, 2 manifold to the 6 3 P 1 state
is located at 313 nm. Transitions from the same 6 3 D 1, 2, 3 manifold to the 6 3 P 2
state were recorded as well. Two new progressions of lines at 349 nm, 353 nm,

Copyright © 2002 by Taylor & Francis

and 357 nm, and at 371 nm, 375 nm, and 380 nm had been observed. The first
manifold has been observed previously in HgBr discharges and was attributed
to the presence of N 2 . However, the spectra have been recorded alongside of and
sideways to the laser beam and the presence of N 2 should be excluded from our
laser system.
The intensity of these spectral lines varies with temperature (they are very
intense above 250°C). Their temporal behavior suggests that they are transitions
from excited states of the Hg 2 cluster formed by the highly excited states to Hg
without attempting any assignments. The dependence of the intensity of mercury
lines at 404.6 nm and 366 nm on temperature, together with rate of formation
arguments, suggest that these lines might originate from emissions from the
highly excited states of Hg 2 : for example, (Hg(7 1 D 2 ) ⫹ Hg(6 1 S 0 )) to lower ones.
As far as the progression of lines around 380 nm is concerned, there is a depen-
dence between the intensity of these lines and the intensity of mercury lines at
365 nm that originate from the 6 3 D manifold. The remaining part of the spectrum
between 385 and 400 nm consists of a series of lines. Temporal evolution of
these spectral lines suggests that they originate from transitions from the excitated
states of the mercury triatomic Hg 3 . Similar long-lived states have been observed
previously around 446 nm and have been attributed to the complex potential
surfaces of Hg 3 . The lifetime of excited states of the Hg 2 cluster is believed to
be short [129], and the long decay rates observed are attributed instead to clusters’
rates of formation in their excited states. This argument is also supported by the
temporal evolution of the 335 nm band. Therefore, considering that the rates of
formation are small, the long lifetimes could be attributed to Hg 3 clusters because
of the complicated structure of their potential surfaces. Strong perturbations pres-
ent in the Hg 3 molecule, alongside tunneling effects, could be the source of long
decay rates. The broad continuum in the blue-green region of the spectrum taken
in a mercury vapour column with maximum at 495 nm has been assigned to
emission from the Hg 3 molecules [127]. The same band has been observed previ-
ously by Cefalas et al. [107] by laser-induced fluorescence. Niefer et al. [125],
using selective excitation in mercury vapor, had assigned the observed absorption
between 415 and 510 nm to the presence of the mercury cluster Hg 3 . The side-
ways fluorescence of the 495 nm band between 480 nm and 510 nm indicates
some vibrational structure with a maximum at 494 nm. The formation of Hg 3
originates from the collision of an Hg 2 molecule in its first excited state with an
Hg atom in its ground state, since the band is observed even without the use of
a buffer gas.

Hg 2 (AO g⫾) ⫹ Hg(6 1 S 0 ) → Hg 3* (AO g⫾ )

→ Hg 3 (XO g⫾ ) → 3Hg ⫹ hν(495 nm band) (6)

The emitted radiation at 495 nm is highly directional and concentrated. Its super-
fluorescence nature [107] was manifested by the appearance of blue–green spots

Copyright © 2002 by Taylor & Francis

on a screen even 50 cm away from the fluorescence cell. Similar kinds of blue–
green spots have been observed by Shingenari et al. [130] at 404 and 407 nm
during the photodissociation of HgBr 2 at 193 nm. The appearance of these spots
suggests that at 495 nm the gain exceeds the absorption for this excitation scheme.
The blue–green spots do not appear on a regular basis, indicating that thermal
gradients inside the heat pipe oven might cause strong refractive index variations.
For excitation schemes that involve electrical discharges or selective excitation
of mercury atomic levels at high temperatures, extension of the band towards
the red is due to the formation of mercury clusters Hg n (n ⬎ 3).

VI. BISTATE QUANTUM SWITCHING IN MERCURY

TRIATOMIC CLUSTERS

Quantum beats with a period from 25 to 100 ns have been observed in mercury
triatomic clusters following de-excitation from excited electronic states of volca-
nic shape to the ground electronic state (Fig. 5). Observation of the quantum
beats can be explained if we assume that the dynamic surface of the Hg 3 cluster
has a symmetry of a double well. This is true for a large number of X 3 systems,
including Cu 3 [131], Na 3 [132], P 3 , As 3 , Sb 3 Bi 3 [133] Ca 3 , and Sc 3 [134]. Forma-
tion of the Hg 3 clusters was also observed by optical spectroscopy by J. Koperski
et al. [135].
Because the vibrational and the electronic parts of the total wavefunctions
are not any longer separable for the Hg 3 molecular system, the excited electronic
states have a complex volcanic structure. The presence of potential energy barri-
ers is the source of tunneling effects of the optical electron. In the simplest case,
potential energy barriers exhibit saddle points with double minima. A double
barrier can exhibit resonances where total transmission of electrons is taking
place even for energies below the height of the potential barrier. This accounts
for zero dispersion of the particle wave packet. It is responsible for the observed
long duration of the quantum beats because the molecule oscillates, trapped be-
tween a symmetrical and an antisymmetrical wavefunction without any reflection
losses. The presence of the double minima causes splitting of vibrational eigen-
functions in a symmetrical and an antisymmetrical component. The reason is the
introduction of an additional translational symmetry in the total Hamiltonian of
the system [136]. The amount of the energy split between the two parts depends
on the molecular parameters and energy separation of the optical electron from
the top of the potential barrier (Fig. 6). Energy splitting is related to the beat
period. Hence the period of the quantum beat depends on the electron energy
relative to the top of the barrier. It increases with increased energy separation
from the top of the double barrier because total transmission probability of the
optical electron through the energy potential barrier depends on its energy separa-
tion from the top of the potential barrier as well. The potential barrier over which

Copyright © 2002 by Taylor & Francis

Figure 5 a. Mode competition at 445 nm. b. Mode competition at 446.9 nm.

quantum beats were observed was found to be 0.05 eV deep [137]. As a result,
selective excitation in Hg 3 clusters could modulate the period of the quantum
beat and the electron transmission probability through the potential barrier of the
excited electronic states. This process simulates the switching function of a quan-
tum transistor with its gate driven by light.
Double-charged Hg 3 clusters following the photodissociation of the HgBr 2
were observed using the mass spectroscopic technique. The parent HgBr 2 was
expanded from a molecular beam apparatus, after being heated in the molecular
beam chamber. Parent molecules were crossed with the F2 molecular laser fo-
cused in the molecular beam with a MgF2 lens with focal length of 20 mm (Fig.
7). The Hg 3 clusters were formed according to the mechanism described previ-

Copyright © 2002 by Taylor & Francis

Figure 6 Potential energy surface for the 445 nm band. Symmetry splitting is indicated.
Optical transitions are taking place between the excited and ground electronic states of
the mercury trimer.

ously, in their excited electronic states and were observed with the quadruple
mass spectrometer (Fig. 8). Double-charged triatomic mercury clusters were the
smallest ever observed (Fig. 7). Berchignac at al. [77], using a different experi-
mental technique, observed double-charged mercury clusters containing up to
five atoms of mercury. The theoretical calculations performed by Tomanek [67]
give for the number n in Hg n2⫹ the value of 300; Jentsh et al. found n ⫽ 8 [68],
and Sattler et al. found n ⫽ 9 [80].
To interpret the observed quantum beats and make the proper assignments,

Figure 7 Geometry of detection of trimer mercury clusters. Direction of the laser light
is perpendicular to the direction of the molecular beam and quadrupole mass filter.

Copyright © 2002 by Taylor & Francis

a simple formula was derived that relates the experimentally measured splitting
of the vibrational levels of the excited triatomic molecule to the shape of the
potential surfaces showing double minima. The presence of the double minima
imposes translational symmetry in the potential energy surfaces. We will consider
the simplest case: the potential energy surface is symmetrical with respect to an
axis that passes from the point equidistant from the two minima (Fig. 6). For the
sake of simplicity, we will consider only one dimension x. The symmetry of the
potential surface implies the relation

U(x ⫹ a) ⫽ U(x) (7)

where a is the distance between the two minima of the potential surface U(x).
Therefore we can introduce the translation operator T̂(a), which translates
the space coordinate of a point at x by a distance a, and for a wave function Ψ(x)
we have

T̂(a)ψ(x) ⫽ ψ(x ⫹ a) (8)

Expanding ψ(x ⫹ a) in a Taylor series around the point x we have

冢冱 冣
∞ ∞
1 d m ψ(x) m
冱
1 m m
ψ(x ⫹ a) ⫽ a ⫽ a 䉮 ψ(x) ⫽ T̂(a)ψ(x) (9)
0
m! dx m 0
m!

and the translation operator T̂(a) is of the form

T̂(a) ⫽ exp(a䉮) (10)

The action of the translational operator on the potential function is of the form

T̂(a)U(x) ⫽ U(x ⫹ a) ⫽ U(x) (11)

Figure 8 Mass spectrum of mercury doubly ionized trimer clusters.

Copyright © 2002 by Taylor & Francis

Using the relationship above it is easy to see that the translation operator com-
mutes with the potential energy of the system and the kinetic part of the Hamilto-
nian Ĥ, where
h2 d2
Ĥ ⫽ ⫺ ⫹ U(x) (12)
2m dx 2
Therefore the translational operator commutes with the Hamiltonian of the system
and the operators Ĥ and T̂(a) have the same set of wave functions, ψ(x).
Suppose now that the eigenfunction ψ(x) of the translation operator T̂(a)
corresponds to the eigenvalue T(a)
T̂(a)ψ(x) ⫽ T(a)ψ(x) (13)
We can prove easily that
T̂ m (a) ⫽ T(ma) (14)
and if we write T(a) in the form
T(a) ⫽ exp(iφ(a)) (15)
it can be proved that
φ(a) ⫽ ka (16)
T̂(a)ψ(x) ⫽ exp(ika)ψ(x) ⫽ ψ(x ⫹ a)
(17)
ψ(x ⫹ a) ⫽ exp(ika)ψ(x)
Therefore the eigenfunctions ψ(x) of the Hamiltonian and the translation operator
obey the so-called translation property of Eq. (17) (Bloch equations). The zero-
order approximation now of the wave function ψ(x), ψ 0 (x) enables the calculation
of the energy eigenvalues of Eq. (12) to the first order.

Ĥ ⫽ ⫺ 冤
h2 d2
2m dx 2 冥
⫹ U(x) ψ 0 (x) ⫽ E 1 ψ 0 (x) (18)

where the subscripts 0 and 1 refer to the zero and first-order approximations for
the wave function and the energy. From Eq. (18) we find
∫ ψ*0 (x)[⫺(h 2 /2m)(d 2 /dx 2 ) ⫹ U(x)]ψ 0 (x)dx
E1 ⫽ (19)
∫ ψ 0* (x)ψ 0 (x)dx
The potential function can be expanded now around x in a Taylor series

U(x) ⫽ 冱 U (x)exp(ika)
a
a (20)

Copyright © 2002 by Taylor & Francis

Therefore Eq. (19) becomes

冤
E 1 ⫽ I ∫ ψ 0* (x) ⫺
h2 d 2
2m dx 2
⫹ 冱 U (x) exp(ika)冥ψ (x)dx
a
a 0 (21)

where

I ⫽ {∫ ψ 0* (x)ψ 0 (x)dx}⫺1
Now we have

冦 冤
E 1 ⫽I ∫ψ*0 (x) ⫺
h2 d 2
2m dx 2
⫹ 冱 U (x)exp(ika′)⫹U (x) exp(ika)冥ψ (x)dx冧
a′≠a
a′ a 0 (22)

taking into account the translational properties of the potential function U(x) and
the wave function ψ(x) Eq. (17), we have

冦 冤
E 1 ⫽ I ∫ ψ *0 (x) ⫺
h2 d 2
2m dx 2
⫹ 冱 U (x ⫺ a′)冥ψ (x)dx
a′≠a
a′ 0

冧
⫹ U a (x ⫺ a) ∫ψ *0 (x)ψ 0 (x)dx

冦 冤
E 1 ⫽ I ∫ ψ *0 (x) ⫺
h2 d 2
2m dx 2
⫹ 冱 U (x ⫺ a′)冥ψ (x ⫺ a′) exp(ika)dx
a′≠a
a′ 0

⫹ U a (x ⫺ a) ∫ψ*0 (x)ψ 0 (x)dx 冧 (23)

⫽ E a ⫹ F(a, k) exp(ika)

where we have denoted

E a ⫽ U a (x ⫺ a)
and

冦 冤
F(a, k) ⫽ I ∫ ψ *0 (x) ⫺
h2 d 2
2m dx 2
⫹ 冱 U (x ⫺ a′ ⫺ a)冥ψ (x ⫺ a)dx冧
a′≠a
a′ 0 (24)

The last integral describes the exchange of energy between the two parts of the
barrier: right of the symmetry axis where the wave function is ψ 0 (x) and left of
the symmetry axis where the wave function is ψ 0 (x ⫺ a).
The energy E 1 therefore depends quasicontinuously on the wave vector k
and its real part changes from E 1 min to E 1 max as k changes:

Copyright © 2002 by Taylor & Francis

 E 1 ⫽ E a ⫹ F(a, k) cos(ka)
E 1 min ⫽ E a ⫹ F(a, k) (25)
E 1 max ⫽ E a ⫹ F(a, k)
Hence the amount of the energy splitting as the particle moves through the double
barrier is ∆E ⫽ 2F(ak). The mass of the cluster m is related to the energy E 1
through the equation

m⫺1 ⫽ 冷 1 d 2E
h 2 dk 2 冷 (26)

and therefore

m⫺1 ⫽ 冷 1 dF
h 2 dk 2
cos(ka) ⫺ 2
dF
h 2 dk
a sin(ka) ⫺
Fa 2
h2
cos(ka) 冷 (27)

For a heavy particle like the triatomic mercury trimer, we do not expect the
exchange integral to change a great deal with the wave vector k. Therefore, Eq.
(27) becomes

冷∆Ea 2hcos(ka)冷
2
m⫽ 2
(28)

Equation 28 is our final result and relates the energy-splitting ∆E of the

vibrational levels to the mass (m) of the triatomic and the distance (a) between
the two minima of the potential surface. The observed period of the quantum
beats (Fig. 9) is related to the energy-splitting ∆E of the vibrational levels by

Figure 9 Time evolution of transitions between excited and ground electronic states of
mercury trimer.

Copyright © 2002 by Taylor & Francis

Table 1 Period T of Quantum Beats for the Three Emission
Bands of Mercury Triatomic Cluster

Band λ (nm) Emission λ (nm) Q B Period T (ns)

385 385, 387, 389 75, 50, 25

445 445, 447, 449, 451, 453 100, 75, 50, 25
485 482, 485, 488 300, 100, 75, 50, 25

The second column shows the wavelength of emission from different vibronic
levels.

the uncertainty relation ∆E ∆T ⫽ h. For a typical beat distance of 75 ns, ∆E ⫽

5 ⫻ 10⫺6 eV. From Eq. 28 we find that this value corresponds to a distance a
between the minima of the potential surface of 0.2 nm for k ⫽ 53 ⫻ 10 10 m⫺1.
A formula that relates the molecular parameters to the amount of splitting
∆E is given by

冢 冣
β
∆E ⫽
∆Eo
π
exp ⫺ 冮⫺β
|k|dx (29)

where β is the distance between two turning points (Fig. 6), ∆Eo the energy
difference between two neighboring vibronic levels near the top of the barrier,
and k is the wave vector related to the reduced cluster mass and it is given by
k ⫽ (2m/h 2 (E ⫺ U))1/2. (E ⫺ U) is the distance of a vibronic level from the top
of the barrier. Eq. (29) can be used to test the validity of our approximations,
which have been derived from Eq. (28).
From spectroscopic data the energy difference between two neighboring
vibrational levels is 0.04 eV. Hence, the first vibronic level below the top of the
potential barrier will be situated below the top of the barrier at the most by the
same amount of energy of 0.04 eV. Therefore, the wave vector k associated with
the vibronic level near the top of the potential barrier is ⬃3.6 ⫻ 10 11 m⫺1. For
a splitting ∆E of 5 ⫻ 10⫺6 eV, the distance 2β is estimated to be 0.02 nm and
the distance α between the two minima is 0.2 nm. The period of the observed
quantum beats for the three emission bands is indicated in Table 1. Fourier trans-
form analysis of the signals indicates multiple frequencies over the same vibra-
tional level. This is due to the fact that the potential energy surfaces are not
symmetrical in space. Quantum beats are observed only for the vibrational levels
indicated in Table 1 because symmetry splitting and tunneling is taking place
only for those levels close to the top of the potential barrier, in agreement with
the theoretical model. The transmission coefficient T of a particle for a double

Copyright © 2002 by Taylor & Francis

barrier described by the U 0 cosh⫺2 (x/2β) dependence in a region near its top is
given by the equation

sinh 2 (2πkβ)
T⫽ (30)
sinh (2πkβ) ⫹ cosh 2 [1/2 π√32mU 0 η⫺2 β 2 ⫺ 1]
2

From Eq. (30) if 1 ⫺ 32mU 0 η⫺2 β 2 ⫽ (2n ⫹ 1) 2, then the transmission

coefficient is equal to 1. In this case the particle is transmitted through the barrier
with zero reflection losses. For n ⫽ 1, U 0 ⫽ 10⫺3 eV, in agreement with our
experimental results regarding the high of the potential barrier over which beat
oscillations were observed. The process simulates the switching properties of a
molecular transistor with its gate driven by light.

REFERENCES

1. JR Woodworth, JK Rice. An efficient, high-power F2 laser near 157 nm. J Chem

Phys 69:2500–2504, 1978.
2. JK Rice, AK Hays, JR Woodworth VUV emissions from mixtures of F2 and the
noble gases-A molecular F2 laser at 1575 A. Appl Physics Letts 31:31–33, 1977.
3. H Pummer, K Hohla, AM Diegelmann, JP Reilly. Discharge pumped F2 laser at
1580 A. Optics Communics 28:104–106, 1979.
4. AC Cefalas, C Skordoulis, M Kompitsas, CA Nicolaides. Gain measurements at
157 nm in an F2 pulsed discharge molecular laser. Optics Communics 55:423–426,
1985.
5. VN Ishchenko, SA Kochubei, AM Razhev. High-power efficient vacuum ultravio-
let F2 laser excited by an electric discharge. Sov J Quantum Electron. 16:707–709;
translated from Kvantov Elektron Moskva 13:1072–1075, 1986.
6. M Ohwa, M Obara. Theoretical evaluation of high-efficiency operation of dis-
charge-pumped vacuum-ultraviolet F2 lasers. Appl Physics Letts. 51:958–960,
1987.
7. K Yamada, K Miyazaki, T Hasama, T Sato. High-power discharge-pumped F2 mo-
lecular laser. Appl Physics Letts 54:597–599, 1989.
8. C Skordoulis, S Spyrou, AC Cefalas. Gain and saturation measurements in a dis-
charge excited F2 laser using an oscillator amplifier configuration. Appl Physics B
51:141–145, 1990.
9. C Skordoulis, E Sarantopoulou, S Spyrou, AC Cefalas. Amplification characteris-
tics of a discharge excited F2 laser. J Modern Optics 37:501–509, 1990.
10. AA Kuznetsov, SS Sulakshin. An F2 laser-excited by a high-power proton-beam.
Kvantov Elektron 18:22–22, 1991.
11. M Kakehata, T. Uematsu, F. Kannari, M. Obara. Efficiency characterization of
vacuum ultraviolet molecular fluorine (F2 ) laser (157 nm) excited by an intense
electric-discharge. IEEE J Quantum Electron 27:2456–2464, 1991.

Copyright © 2002 by Taylor & Francis

 12. M Kakehata, CH Yang, Y Ueno, F Kannari. Gain and saturation intensity measure-
ments of a discharge pumped F2 laser at high-excitation rates. Appl Physics Letts
61:3089–3091, 1992.
13. FTJL Lankhorst, HMJ Bastiaens, PJM Peters, WJ Witteman. High specific laser
output energy at 157-nm from an electron-beam-pumped He/Ne/F2 gas-mixture.
Appl Physics Letts 63:2869–2871, 1993.
14. FTJL Lankhorst, HMJ Bastiaens, PJM Peters, WJ Witteman. Long pulse electron
beam pumped molecular F2 laser. J Appl Physics 77:399–401, 1995.
15. HMJ Bastiaens, SJM Peeters, X Renard, PJM Peters, WJ Witteman. Long pulse
operation of an x-ray preionized molecular fluorine laser excited by a pre pulse-
main pulse system with a magnetic switch. Appl Physics Letts 72:2791–2793,
1998.
16. N Rizvi, D Milne, P Rumsby, M Gower. Laser micromachining—new develop-
ments and applications. Proc SPIE 3933:261–270, 2000.
17. JC White, HG Craighead, RE Howard, LD Jackel, OR Wood. VUV laser photoli-
thography. AIP Conf Proc 119:324–329, 1984.
18. S Nagai, T Enami, T Nishisaka, J Fujimoto, O Wakabayashi, H Mizoguchi. Devel-
opment of kHz F2 laser for 157 nm lithography. In: Digest of Papers, Micro-
processes and Nanotechnology ’99. International Microprocesses and Nanotechnol-
ogy Conference. Tokyo, Japan: Japan Society of Applied Physics, 234:64–65 1999.
19. T Hofmann, JM Hueber, P Das, S Scholler. Revisiting F2 laser for DUV micro-
lithography. Proc SPIE 3679:541–546, 1999.
20. AC Cefalas, E Sarantopoulou, E Gogolides, P Argitis. Absorbance and outgassing
of photoresist polymeric materials for UV lithography below 193 nm including 157
nm lithography. Microelectron Engin 53:123–126, 2000.
21. AC Cefalas, N Vassilopoulos, Z Kollia, E Sarantopoulou, C Skordoulis. Mass spec-
troscopic studies and ablation characteristics of nylon 6.6 at 248 nm. Appl Phys
A 70:21–28, 2000.
22. AC Cefalas, E Sarantopoulou, P Argitis, E Gogolides. Mass spectroscopic and de-
gassing characteristics of polymeric materials for 157 nm photolithography. Appl
Physics A 69: 929–933, 1999.
23. AC Cefalas, P Argitis, Z Kollia, E Sarantopoulou, TW Ford, AD Stead, A Maranka,
CN Danson, J Knott, D Neely. Laser plasma x-ray contact microscopy of living
specimens using a chemically amplified epoxy resist. Appl Phys Letts 72:25–27,
1998.
24. AC Cefalas, E Sarantopoulou. 157 nm photodissociation of polyamides, Mi-
croelect. Engin 53:464–468, 2000.
25. TM Bloomstein, MW Horn, M Rothschild, RR Kunz, ST Palmacci, RB Goodman,
Lithography with 157 nm lasers. J Vac Sci Technol B 15:2112–2116 1997.
26. RR Kunz, TM Bloomstein, DE Hardy, RB Goodman, DK Downs, JE Curtin, Out-
look for 157 nm resist design. SPIE 3678:13–23, 1999.
27. I. Petsalakis, G Theodorakopoulos, Electronic states of CF⫹. Chem Phys 254:181–
186, 2000.
28. RJ Cody, C Moralejo, JE Allen Jr. Photodissociation of the hydroxyl radical (OH)
at 157 nm. J Chem Phys 95: 2491–2496, 1991.
29. Y Matsumi, M Kawasaki. Fine-structure branching ratios of the O(3Pj) atomic frag-

Copyright © 2002 by Taylor & Francis

 ments from photodissociation of oxygen molecules at 157 and 193 nm. J Chem
Phys 93:2481–2486, 1990.
30. YL Huang, RJ Gordon. The multiplet state distribution of O(3Pj) produced in the
photodissociation of O 2 at 157 nm. J Chem Phys 94:2640–2647, 1991.
31. JJ Lin, DW Hwang, YT Lee, XM Yang. Photodissociation of O 2 at 157 nm—
experimental-observation of anisotropy mixing in the O 2 ⫹ hv⫺)O(P ⫺ 3) ⫹
O(P ⫺ 3) channel. J Chem Phys 109:1758–1762, 1998.
32. Y Matsumi, PK Das, M Kawasaki. Doppler spectroscopy of chlorine atoms gener-
ated from photodissociation of hydrogen-chloride and methyl-chloride at 157-nm
and 193-nm. J Chem Phys 92:1696–1701, 1990.
33. K Tonokura, Y Matsumi, M Kawasaki, S Tasaki, R Bersohn. Photodissociation of
hydrogen-chloride at 157 and 193 nm—angular-distributions of hydrogen-atoms
and fine-structure branching ratios of chlorine atoms in the 2Pj levels. J Chem Phys
97:8210–8215, 1992.
34. U Wetterauer, A Pusel, P Hess. VUV photodesorption of molecular-hydrogen from
the hydrogenated silicon(III) surface. Chem Phys Lett 300:397–402, 1999.
35. JK Rice, JR Woodworth. High-intensity laser photolysis, of OCS at 157 nm: S 1/S
production photoionization, and loss. J Appl Phys 50:4415–4421, 1979.
36. GS Ondrey, S Kanfer, R Bersohn. The collinear photodissociation of OCS at 157
nm. J Chem Phys 79:179–184, 1983.
37. P Andresen, GS Ondrey, EW Rothe, B Titze. The photodissociation of H 2 O at 157
nm: full internal state distributions and alignment of nascent OH(X 2 Π) radicals.
In: J Eichler, W Fritsch, IV Hertel, N Stolterfoht, U Wille, eds. 13th International
Conference on the Physics of Electronic and Atomic Collisions. Berlin, West Ger-
many: 1983.
38. P Andresen, GS Ondrey, B Titze. Creation of population inversions in the Lambda
doublets of OH by the photodissociation of H 2 O at 157 nm: a possible mechanism
for the astronomical maser. Phys Rev Lett 50:486–488, 1983.
39. P Andresen, EW Rothe. Polarized LIF spectroscopy of OH formed by the photodis-
sociation of H 2 O by polarized 157 nm light. J Chem Phys 78:989–990, 1983.
40. K Mikulecky, K Heinz-Gericke, FJ Comes. Decay dynamics of H 2 O(1 B 1 ): full char-
acterization of OH product state distribution. Chem Phys Lett 182:290–296, 1991.
41. RL Vander-Wal, JL Scott, FF Crim. Selectively breaking the O-H bond in HOD.
J Chem Phys 92:803–805, 1990.
42. H Wurps, H Spiecker, JJ ter-Meulen, P Andresen. Laser induced fluorescence mea-
surements of microwave stimulated OH molecules from H 2 O photodissociation.
J Chem Phys 105:2654–2659, 1996.
43. JA Guest, F Webster. Strong fragment alignment variation with internal state from
ICN dissociation at 157.6 nm: linear regression modelling of CN B 2 sigma⫹ prod-
ucts J Chem Physics 86:5479–5490, 1987.
44. N Shafer, S Satyapal, R Bersohn. Isotope effect in the photodissociation of HDO
at 157.5 nm. J Chem Phys 90:6807–6808, 1989.
45. C Ye, M Suto, LC Lee, TJ Chuang. Radiative lifetime and quenching rate constant
of AsF*2 fluorescence. J Phys B 22:2527–2529, 1989.
46. B Freisinger, U Kogelschatz, JH Schafer, J Uhlenbush, W Viol. Ozone production
in oxygen by means of F2 laser at X ⫽ 157.6 nm. Appl Phys B 49:121–129, 1989.

Copyright © 2002 by Taylor & Francis

 47. Zhu Yi-Fei, RJ Gordon, The production of O(3 P) in the 157 nm photodissociation
of CO 2 . J Chem Physics 92:2897–2901, 1990.
48. Y Matsumi, N Shafer, K Tonokura, M Kawasaki, H Yu Lin, RJ Gordon. Doppler
profiles and fine structure branching ratios of O ( 3Pj ) from photodissociation of
carbon oxide at 157 nm. J Chem Phys 95:7311–7316, 1991.
49. RL Miller, SH Kable, PL Houston, I Burak. Product distributions in the 157 nm
photodissociation of CO 2 . J Chem Phys 96:332–338, 1992.
50. A Stolow, YT Lee. Photodissociation dynamics of CO 2 at 157.6 nm by photofrag-
ment translational spectroscopy. J Chem Phys 98:2066–2076, 1993.
51. A Pusel, U Wetterauer, P Hess. Direct breaking of SiH-surface bonds with 157 nm
laser radiation. Laser Optom 31:48–50, 1999.
52. JM Soler, MR Beltran, K Michaelian, IL Garzon, P Ordejon, D Sanchez-Portal,
E Artacho. Metallic bonding and cluster structure. Phys Rev B 61:5771–5780,
2000.
53. K Michaelian, N Rendon, IL Garzon. Structure and energetics of Ni, Ag, and Au
nanoclusters. Phys Rev B 60:2000–2010, 1999.
54. U Meier, SD Peyerimhoff, F Grein. Ab initio MRD-CI study of neutral and charged
Ga 2 Ga 3 and Ga 4 clusters and comparison with corresponding boron and aluminum
clusters. Z Phys D 17:209–224, 1990.
55. Surface Science 156:1986; Z Phys D 3:1986.
56. A Dixon, C Colliex, R. Ohana, P Sudraud, J Van-de-Walle. Field-ion emission
from liquid tin. Phys Rev Letts 46:865–868, 1981.
57. K Gamo, Y Inomoto, Y Ochiai, S Namba. B, Sb and Si liquid metal alloy ion
sources for submicron fabrication. Jpn J Appl Phys 21:415–420, 1982.
58. C Brechignac, M Broyer, P Cahuzac, G Delacretaz, P Labastie, L Woste. Stability
and ionization threshold of doubly charged mercury clusters. Chem Phys Lett 118:
174–178, 1985.
59. AR Waugh. Current-dependent energy spectra of cluster ions from a liquid gold
ion source. J Phys D 13: L203–208, 1980.
60. S Mukherjee, KH Bennmann. On the stability of charged metallic clusters. Surface
Sci 156:580–583, 1985.
61. G Durand, JP Daudey, JP Malrieu. Singly and doubly charged clusters of (S 2 ) mel-
als. Ab initio and model calculations on Mg n⫹ and Mgn⫹⫹ (n ⱕ 5). J Phys 47:1335–
1346, 1986.
62. S Martrenchard, C Jouvet, C Lardeux-Dedonder, D Solgadi, Observation of very
small doubly charged homogeneous para difluorobenzene clusters. J Chem Phys
94:3274–3275, 1991.
63. K Sattler. The physics of microclusters. Adv Solid State Phys 23:1–12, 1983.
64. O Echt, K Sattler, E Recknagel. Magic numbers for sphere packings: experimental
verification in free xenon clusters. Phys Rev Lett 47:1121–1124, 1981.
65. JG Gay, BJ Berne. Computer simulation of Coulomb explosions in doubly charged
Xe microclusters. Phys Rev Lett 49:194–198, 1982.
66. P Pfau, K Sattler, R Pflaum, E Recknagel. Observation of doubly charged lead
clusters below the critical size limit for Coulomb explosion of sphere configura-
tions. Phys Lett A 104A:262–263, 1984.

Copyright © 2002 by Taylor & Francis

 67. D Tomanek, S Mukherjee, KH Bennemann. Simple theory for the electronic and
atomic structure of small clusters. Phys Rev B 28:665–673, 1983.
68. Th Jentsch, W Drachsel, JH Block. Stability of doubly charged homonuclear tri-
meric metal clusters. Chem Phys Lett 93:144–147, 1982.
69. K Yamanouchi. Tuneable vacuum ultraviolet laser ideal light for spectroscopy of
atoms, molecules, and clusters. Electron Spectrosc Rel Phenom 80:267–270,
1996.
70. T Moller, ARB Decastro, K Vonhaeften, A Kolmakov, T Laarmann, Lofken, C
Nowak, F Picucci, M Riedler, C Rienecker, A Wark, M Wolff. Electronic-structure
and excited-state dynamics of clusters—what can we learn from experiments with
synchrotron radiation. Electron Spectrosc Rel Phenom 103:185–191, 1999.
71. OF Hagena. Formation of silver clusters in nozzle expansions. Z Phys D 20:425–
8, 1991.
72. C Brechignac, M Broyer, P Cahuzac, G Delacretaz, P Labastie, L Woste. Size
dependence of linear-shell autoionization lines in mercury clusters. Chem Physics
Letts 120:559–563, 1985.
73. C Brechgnac, M Broyer, P Cahuzac, G Dalacretaz, P Labastie, L Woste, Stability
and ionization thresholod of double charged mercury clusters. Chem Phys Lett 118:
174–278, 1985.
74. R Heinrich, A Wucher. Yields and energy distributions of sputtered semiconductor
clusters. Nuclear Instrum Methods Physics Res 140:27–38, 1998.
75. MA Ratner. Kinetics of cluster growth in expanding rare-gas jet. Low Temp Physics
25:266–273, 1999.
76. W Schulze, B Winter, J Urban, I Goldenfeld. Stability of multiply charged silver
clusters. Z Phys D 4:379–381, 1987.
77. W Schulze, B Winter, I Goldenfeld. Generation of germanium clusters using the
gas aggregation technique: stability of small charged clusters. J Chem Physics 87:
2402–2403, 1987.
78. W Schulze, B Winter, I Goldenfeld. Stability of multiply positively charged homo-
nuclear clusters. Phys Rev B 38:12937–12941, 1988.
79. A Hoareau, P Melinon, B Cabaud, D Rayane, B Tribollet, M Broyer. Fragmentation
of singly and doubly charged lead clusters. Chem Phys Lett 143:602–608, 1988.
80. K Sattler, J Muhlbach, O Echt, P Pfau, E Rechnagel. Evidence for Coulomb explo-
sion of doubly charged microclusters. Phys Rev Lett 47:160–163, 1981.
81. VI Matveev, PK Khabibullaev. Sizes and charges of clusters produced in the pro-
cess of ion sputtering of a metal. Doklady Physics 44:789–792, 1999.
82. A Wucher, Z Ma, WF Calaway, MJ Pellin. Yields of sputtered metal clusters: the
influence of surface structure. Surface Sci 304:L439–444, 1994.
83. A Wucher, M Wahl, H Oechsner. The mass distribution of sputtered metal clusters.
Exp Nuclear Instrum Methods Physics Res 83:73–78, 1993.
84. S Nishigaki, W Drachsel, JH Block. Photon-induced field ionization mass spec-
trometry of ethylene on silver. Surface Sci 37:389–409, 1979.
85. C Skordoulis, E Sarantopoulou, SM Spyrou, C Kosmidis, AC Cefalas. Laser in-
duced fluorescence (LIF) of Hg 2 and Hg 3 via dissociation of HgBr 2 at 157 nm.
Z Phys D 18:175–180, 1991.

Copyright © 2002 by Taylor & Francis

 86. L Wiedeman, H Helvajian. Laser photodecomposition of sintered Yba 2 Cu 3 /O 6⫹x :
ejected species population distributions and initial kinetic energies for the laser
ablation wavelengths 351, 248, and 193 nm. J Appl Phys 70:4513–4523, 1991.
87. Proceedings Laser Induced Chemistry Conference, 18–22 Sept. 1989, Bechyne,
Czechoslovakia. Spectrochim Acta 46A(4):1990.
88. M Ovchinnikov, BL Grigorenko, KC Janda, VA Apkarian. Charge localization and
fragmentation dynamics of ionized helium clusters. J Chem Phys 108:9351–9361,
1998.
89. C Brechgnac, M Broyer, P Cahuzac, G Dalacretaz, P Labastie, L Woste. Stability
and ionization threshold of double charged mercury clusters. Chem Phys Lett 118:
174–278, 1985.
90. D Rayane, P Melinon, B Cabaus, A Hoareau, B Tribollet, M Broyer. Electronic
properties and fragmentation processes for singly and doubly charged indium clus-
ters. J Chem Phys 90:3295–3299, 1989.
91. E Ruhl, PG F Bisling, B Brutschy, H Baumgartel. Photoionization of aromatic van
der Waals complexes in a supersonic jet. Chem Phys Lett 126:232–237, 1986.
92. JR Grover, EA Walters, ET Hui. Dissociation energies of the benzene dimer and
dimer cation. J Phys Chem 91:3233–3237 1987.
93. W Kamke, B Kamke HU Kiefl, IV Hertel. Photoionization and fragmentation of
(N 2 O) N clusters. J Chem Phys 84:1325–1334, 1986.
94. B Bescos, B Lang, J Weiner, V Weiss, E Wiedenmann, G Gerber. Real-time obser-
vation of ultrafast ionization and fragmentation of mercury clusters. Eur Phys J D
9:399–403, 1999.
95. T Baumert, R Thalweiser, V Weiss, G Gerber. Time-resolved studies of neutral
and ionized Na n clusters with femtosecond light pulses. Z Phys D 26:131–134,
1993.
96. JA Booze, T Baer. On the determination of cluster properties by ionization tech-
niques. J Chem Physics 96:5541–5543, 1992.
97. A Dixon, C Colliex, R Ohana, P Sudraud, J Van de Walle. Field-ion emission from
liquid tin. Phys Rev Lett 46:865–868, 1981.
98. J Van de Walle, P Joyes. Study of Bi np⫹ ions formed in liquid-metal ion sources.
Phys Rev B 35:5509–5513, 1987.
99. Y Gotoh, T Kashiwagi, H Tsuji, J Ishikawa. Empirical relation between electric
field at the ionization point and emission current of liquid copper, gold, germanium,
and tin ion sources. Appl Phys A 64:527–532, 1997.
100. W Drachsel, T Jentsch, KA Gingerich, JH Block. Observation of doubly charged
triatomic cluster ions in field evaporation. Surface Sci 156:173–182, 1985.
101. K Hata, M Ariff, K Tohji, Y Saito. Selective formation of C 20 cluster ions by field
evaporation from carbon nanotubes. Chem Phys Lett 308:343–346, 1999.
102. R Karnbach, M Joppien, J Stapelfeldt, J Wormer, T Moller. CLULU—an experi-
mental setup for luminescence measurements on van der Waals clusters with syn-
chrotron radiation. Rev Sci Instrum 64:2838–2849, 1993.
103. A Wucher, R Heinrich, RM Braun, KF Willey, N Winograd. Vacuum ultraviolet
single-photon versus femtosecond multiphoton ionization of sputtered germanium
clusters. Rapid Commun Mass Spectrom 12:1241–1245, 1998.

Copyright © 2002 by Taylor & Francis

104. A Wucher, M Wahl. The formation of clusters during ion-induced sputtering of
metals. Nucl Instrum Methods Phys Res B 115:581–589, 1996.
105. D Koch, M Wahl, A Wucher. Electron-impact and single-photon ionization cross-
sections of neutral silver clusters. Z Phys D 32:137–144, 1994.
106. M Wahl, A Wucher. VUV Photoionization of sputtered neutral silver clusters. Nucl
Instrum Methods Physics Res B 94:36–46, 1994.
107. AC Cefalas, C Skordoulis, C Nicolaides. Superfluorescent laser action around
495nm in the blue–green band of the mercury trimer Hg 3 . Optics Commun 60:49–
54, 1986.
108. TA Cool, JA McGarvey Jr, AC Eriandson. Two-photon excitation of mercury atoms
by photodissociation of mercury halides Chem Phys Lett 58:108–113, 1978.
109. W Luthy, R Schiimiele, T Gerber. Spectral properties of a HgBr 2 photodissociation
laser. Phys Lett 88:450–452, 1982.
110. Nai-Ho Cheung, JA McGarvey Jr, AC Eriandson, TA Cool. The vibrational distri-
bution of HgBr (X 2 Sigma) molecules formed by photodissociation of HgBr 2 at
193 nm J Chem Phys 77:5467–5474, 1982.
111. T Efthimiopoulos, D Zevgolis, J Katsenos, D Zigos. Laser action from (B2E-
X2E)HgBr induced by UV laser multiphoton dissociation of HgBr 2 measurements
and experimental results. Proc SPIE 3423:68–71, 1998.
112. EJ Schimitschek, JE Celto, JA Trias. Mercuric bromide photodissociation laser
Appl Phys Lett 31:608–610, 1977.
113. JH Schloss, JG Eden. Excited-state absorption in HgBr Appl Phys Lett 55:1282–
1284, 1989; see also C Whitehurst, TA King. Multiphoton excitation of mercury
atoms by photodissociation of HgX 2 (X ⫽ Cl, Br, I). J Phys B 20:4053–4051, 1987.
114. RE Drullinger, MM Hessel, EW Smith. Experimental study of mercury molecules.
J Chem Phys 66:5656–66, 1977.
115. AB Callear, KL Lai. Electronic absorption spectra of metastable dimers and trimers
in optically excited mercury vapour. The carrier of the green emission. Chem Phys
Lett 64:100–104, 1979.
116. WR Wadt. The electronic structure of HgCl 2 and HgBr 2 and its relationship to
photodissociation J Chem Phys 72:2469–2478, 1980.
117. C Duzy, HA Hyman. Radiative lifetimes for the B to X transition in HgCl, HgBr,
and HgI. Chem Phys Lett 52:345–348, 1977.
118. J Tellinghuisen, JG Ashmore. The B to X transition in 200 Hg 79 Br. Appl Phys Lett
40:867–869, 1982.
119. J Husain, JR Wiesenfeld, RN Zare. Photofragment fluorescence polarization fol-
lowing photolysis of HgBr 2 at 193 nmJ. Chem Phys 72:2479–2483, 1980.
120. AK Rai, SB Rai, OK Rai. Ultraviolet bands in diatomic mercury bromide. J Phys
B 16:1907–1913, 1983.
121. BE Wilcomb, R Burnham, N Djew. UV absorption cross section and fluorescence
efficiency of HgBr 2 . Chem Phys Lett 75:239–242, 1980.
122. SH Linn, WB Tzeng, JM Jr Brom, CY Ng. Molecular beam photoionization study
of HgBr 2 and HgI 2 . J Chem Phys 78:50–61, 1983.
123. J Maya. Fluorescence yields of metal halide vapors excited by photodissociation.
Appl Phys Lett 32:484–486, 1978.

Copyright © 2002 by Taylor & Francis

124. M Stock, EW Smith, RE Drullinger, MM Hessel. Relaxation of the mercury 6 2 P0
and 6 3 P states J Chem Phys 67:2463–2469, 1977.
125. RJ Niefer, J Supronowicz, JB Atkinson, L Krause. Laser-induced fluorescence of
Hg 2 H1 u excimer molecules. Phys Rev A 34:1137–1142, 1986; J Supronowicz, RJ
Niefer, JB Atkinson, L Krause. Laser-induced fluorescence spectrum arising from
G0u⫹ X0g⫹ bound-free transitions in Hg 2 L. J Phys B19:1153–1164, 1986; RJ Niefer,
J Supronowicz, JB Atkinson, L Krause. Laser-induced fluorescence from the H, I,
and J states of Hg 2 . Phys Rev A 35:4629–4238, 1987; RJ Niefer, J Supronowicz,
JB Atkinson, L Krause. Laser-induced fluorescence spectroscopy of the Hg 3 exci-
mer. Phys Rev A 34:2483–2485, 1986; RJ Niefer, JB Aktinson, L Krause. Vibra-
tional structure of some Hg 2 fluorescent bands. J Phys B16:3767–3773, 1983.
126. AB Callear, KL Lai. Electronic absorption spectra of metastable dimers and trimers
in optically excited mercury vapour: the carrier of the green emission. Chem Phys
Lett 64, 100, 1979; AB Callear, KL Lai. Vibrational structure of Rydberg transitions
of metastable mecury dimers. Chem Phys Lett 75:234, 1980; AB Callear, KL Lai.
Transient immediates in the flash photolysis of mercury vapour. Dimers, trimers
and the enigmatic 2480 AA bands. Chem Phys 69:1, 1982; AB Callear, DR Kendall.
A violet emission from optically excited mercury vapour: kinetics of the formation
and decay of excited dimers and trimers. Chem Phys Lett 64:401, 1979; AB Callear,
DR Kendall. Combination of metastable and ground-state mercury atoms into un-
gerade dimer states. Chem Phys Lett 70:215, 1980; AB Callear, DR Kendall. Lumi-
nescence from mixtures of mercury vapour and nitrogen following pulsed excitation
with resonance radiation. Chem Phys 57:65, 1981.
127. RE Drullinger, MM Hessel, EW Smith. A theoretical analysis of mercury mole-
cules. J Chem Phys 66:5656–5666, 1977; EW Smith, RE Drullinger, MM Hessel,
J Cooper. A theoretical analysis of mercury molecules. J Chem Phys 66:5667–
5681, 1977.
128. ER Mosburg, MD Wilke. Excimer densities and destruction mechanisms in a high
pressure pure mercury positive column. J Chem Phys 66:5682–5693, 1977.
129. PJ Hay, TH Dunning, RC Raffeneti. Electronic states of Zn 2 ab initio calculations
of a prototype for Hg 2 . J Chem Phys 65:2679–2689, 1976.
130. T Shigenari, F Uesugi, H Takuma. Superfluorescent laser action at 404.7 nm from
the transition of a mercury atom produced by photodissociation of HgBr 2 by an
ArF laser. Opt Lett 7:362–364, 1982.
131. SR Langhoff, CW Bauschlicher Jr, SP Walch, BC Laskowski. Ab initio study of
the ground state surface of Cu 3 . J Chem Phys 85:7211–7215, 1986.
132. M Broyer, G Delacretaz, P Labastie, JP Wolf, L Woste. Spectroscopy of vibrational
ground-state levels of Na 3 . J Phys Chem 91:2626–2630, 1987.
133. K Balasubramanian, K Sumathi, D Dai. Group V trimers and their positive ions:
the electronic structure and potential energy surfaces. J Chem Phys 95:3494–34505,
1991.
134. SP Walch, CW Bauschlicher Jr. On 3d bonding in the transition metal trimers: the
electronic structure of equilateral triangle Ca 3 , Sc 3 , Sc3⫹, and Ti3⫹. J Chem Phys 83:
5735–5742, 1985.
135. J Koperski, JB Atkinson, L Krause. First observations of laser-excited Hg 3 and Hg 2
Rg spectra in a supersonic expansion beam. J Mol spectrosc 187:181–192, 1998.

Copyright © 2002 by Taylor & Francis

136. LD Landau, EM Lifshitz. Quantum mechanics. Oxford, England: Pergamon Press,
1981.
137. AC Cefalas, SM Spyrou, E Sarantopoulou. Observation of vibronic–electronic in-
teraction in mercury triatomic clusters via real time quantum beats. Laser Ultrafast
Proc 4:175–181, 1991.

Copyright © 2002 by Taylor & Francis

7
Tunable Solid-State Ultraviolet
and Vacuum Ultraviolet Lasers

Ilko K. Ilev and Ronald W. Waynant

U.S. Food and Drug Administration, Rockville, Maryland

I. INTRODUCTION

The ultraviolet (UV) and vacuum ultraviolet (VUV) spectral range is of great
interest for many attractive industrial and biomedical applications. The research
efforts in this range are focused on development of alternative laser methods and
materials providing widely tunable short-wavelength laser generation. The all-
solid-state laser concept for UV/VUV tunable generation is a superior alternative
to the conventional excimer-gas-based UV/VUV lasers. This concept includes
the development of all-solid-state nonlinear optical techniques as well as tunable
UV/VUV solid-state active materials such as rare-earth-doped laser crystals. Sev-
eral important factors have recently significantly contributed to the all-solid-state
nonlinear techniques being recognized as one of the most effective and easy ways
of getting UV- and VUV-tunable generation.
These factors can be summarized in the following main groups. First is the
development of all-solid-state ultrashort-pulse nonlinear techniques based on
high-intensity picosecond and femtosecond lasers that allow, for instance, use of
only two sequential cascade processes to obtain UV/VUV tunable laser genera-
tion. Second is the development of new nonlinear upconversion crystals possess-
ing high conversion efficiency and broad spectral transparency in the UV and
VUV range down to 155 nm, which is below the shortest wavelength of 157 nm
achieved by F2-excimer laser. Third is the development of all-solid-state diode-
pumped lasers that allow extremely compact and reliable laser systems to be
commercially produced. The final group is the development of new fiber-optic
materials for UV laser delivery and generation. Most of the fundamental nonlin-

Copyright © 2002 by Taylor & Francis

ear effects contribute directly to the process of generating UV- and VUV-tunable
laser radiation that is the subject of this chapter. Most of it will therefore address
the physical principles of nonlinear frequency conversion effects as well as com-
monly used nonlinear conversion techniques, including frequency upconversion
and fiber Raman lasers. The final section briefly presents the basic features of
other known solid-state sources of tunable UV/VUV radiation, including rare-
earth-doped wide-band-gap crystals for UV/VUV fluorescence.

II. ALL-SOLID-STATE NONLINEAR OPTICAL TECHNIQUES

FOR UV AND VUV TUNABLE LASER GENERATION

Among modern photonics technologies, nonlinear optics is one of the most inten-
sively studied, especially since the invention of ultrashort high-intensity lasers.
Short-pulse lasers provide high-intensity optical interactions and the nonlinear
conversion effects acquire an important influence on the mechanism of these
interactions. The first part of this section provides an analytical overview of the
general principles and features of basic nonlinear effects in light of their applica-
tion for development of all-solid-state UV/VUV-tunable lasers. It reviews the
fundamental properties of nonlinear susceptibilities and main second- and third-
order nonlinear frequency conversion processes. The second and third parts con-
sider the main all-solid-state nonlinear techniques and optical materials used for
UV/VUV-tunable generation. Significant results achieved by the nonlinear fre-
quency upconversion techniques are also considered.

A. Nonlinear Frequency Upconversion Effects

1. Nonlinear Susceptibilities
In principle, the intensity of a laser source is significantly higher (orders of magni-
tude) than could be produced by any conventional light source. Laser intensity
is comparable with the typical field strength of optical materials (109 –1010V/m).
Hence, the laser intensity affects the optical medium and creates conditions for
nonlinear optical processes. This effect has a bidirectional character. First, the
laser emission affects the optical medium. As a consequence, the material re-
sponse ceases to be a linear function of the applied electromagnetic field. In fact,
optical constants that characterize the properties of optical material (for instance,
refractive indexes and dielectric constants) are no longer constants and become
nonlinear functions of the light intensity. Second, at the same time, there is ob-
served an opposite effect: of the optical medium on the incidence light field. As
a result, some important parameters of the light field are changed for instance,
frequency, intensity, phase, propagation direction, among others.

Copyright © 2002 by Taylor & Francis

 In an analytical representation, the appearance of nonlinear optical effects
is related to a conversion of the material equation that describes the dependence
between the macroscopic polarization (P) induced by light propagating in the
medium and the optical electric field (E) [1–8]. This equation is converted from
a linear form:
P ⫽ χ(1) E (1)
to a nonlinear form:
P ⫽ PL ⫹ PNL ⫽χ(1) E ⫹ χ(2) EE ⫹ χ(3) EEE ⫹ ⋅ ⋅ ⋅ (2)
Here,
PL ⫽ χ(1) E (3)
is the linear polarization;
PNL ⫽χ(2) EE ⫹ χ(3) EEE ⫹ ⋅ ⋅ ⋅ (4)
is the nonlinear polarization; χ(1) is the linear susceptibility, a second-order tensor
with typical values χ(1) ⬇ 1; χ(2) is the quadratic (second-order) nonlinear suscepti-
bility, a third-order tensor with typical values χ(2) ⬇ 10⫺11 –10⫺12 m/V; and χ(3) is
the cubic (third-order) nonlinear susceptibility, a fourth-order tensor with typical
values χ(3) ⬇ 10⫺21 –10⫺22 m2 /V2.
With increasing laser intensity, the high-order nonlinear polarization terms
P(n⬎1) become more and more important and will lead to a large variety of nonlin-
ear optical effects. The study of the nonlinear susceptibilities χ(2), χ(3), χ(4) . . . of
the excited optical medium gives rich information about the energetic spectra of
the materials. It also provides alternative techniques for development of effective
laser wavelength converters including UV and VUV ranges. The nonlinear opti-
cal effects can be classified by the specific properties of the nonlinear susceptibili-
ties as follows:
The linear susceptibility χ(1) defines the complex refractive index of the
optical medium and is connected with the dielectric constant of the free space
⑀0 by the expression:
ε1 ⫽ ε0(1 ⫹ χ(1)) (5)
The second-order nonlinear susceptibility χ(2) describes second-order non-
linear effects and it is related to the asymmetrical-type crystals, which have no
center of inversion symmetry. Nonlinear effects included in this category are:
Second-harmonic generation associated with |χ(2)(2ω ⫽ ω ⫹ ω)| 2
Sum or difference frequency generation—χ(2)(ω 3 ⫽ ω1 ⫹ ω 2)
Parametric fluorescence
Optical rectification

Copyright © 2002 by Taylor & Francis

 The third-order nonlinear susceptibility χ(3) describes third-order nonlinear
effects and is related to the symmetrical-type crystals, liquids, and gases for which
χ(2) ⫽ 0. These nonlinear optical effects are:
Third-harmonic generation— | χ(3)(3ω ⫽ ω ⫹ ω ⫹ ω)| 2
Four-photon parametric mixing—χ(3)(ω 4 ⫽ ω1 ⫹ ω 2⫺ω 3)
Two-photon absorption—Im χ(3)(ω 2 ⫽ ω1 ⫹ ω 2⫺ω 2)
Stimulated Raman scattering—Im χ(3)(ω AS,S ⫽ ω p ⫾ ω R)
Optical Kerr effect—Re χ(3)(ω ⫽ 0 ⫹ 0 ⫹ ω)
Self-phase modulation
A simple illustration of the above-listed nonlinear optical effects is shown
in Figure 1. We consider two waves ω1 and ω 2 interacting in a quadratic nonlinear
medium characterized by the second-order nonlinear susceptibility χ(2) (Fig. 1a)
or in a cubic nonlinear medium characterized by the third-order nonlinear suscep-
tibility χ(3) (Fig. 1b).
Since most of the mentioned nonlinear effects contribute directly to the
process of generation of tunable UV and VUV laser emission [9–11] that is the
subject of this chapter, the basic nonlinear effects will be discussed in some detail
below. At the beginning, harmonic generation and sum frequency mixing will
be considered. Next, nonlinear effects such as stimulated Raman scattering, four-
photon mixing, and self-phase modulation will be discussed in view of their appli-
cation for development of UV-tunable fiber Raman lasers.

2. Harmonic Generation
Second-Harmonic Generation. Second-harmonic generation (SHG) oc-
curs in optical materials without an inversion center of symmetry, that is, in
quadratic nonlinear media when χ(3) ⫽ 0 [1,2,12]. It is a nonlinear process
whereby two pumping optical fields of the same frequency interact in the nonlin-
ear material to produce a third field with a double higher frequency. In Figure
1a the SHG process is associated with the case when the frequencies of the inter-
acting waves ω1 and ω 2 are equal: ω1 ⫽ ω 2 ⫽ ω. Analytically, the SHG process
can be described by a simple consideration. Let us assume to have a pumping
electromagnetic field in the standard form for harmonic electromagnetic fields:
E ⫽ A sinωt (6)

Figure 1 Basic nonlinear optical effects observed at nonlinear interaction of two elec-
tromagnetic waves with frequencies ω1 and ω 2 in an optical medium possessing quad-
ratic- (a) or cubic- (b) nonlinear susceptibility. SHG, second harmonic generation; SFG,
sum-frequency generation; DFG, difference-frequency generation; THG, third-harmonic
generation; FPM, four-photon mixing.

Copyright © 2002 by Taylor & Francis

Copyright © 2002 by Taylor & Francis
If we put this field in the nonlinear material Eq. (2), we get:
P ⫽ χ(1) A sinωt ⫹ 1/2χ(2) A2 ⫺ 1/2χ(2) A2 cos(2 ωt) (7)
Hence, when we act with a harmonic electromagnetic field [Eq. (6)] on a qua-
dratic nonlinear optical medium, a polarization wave P arises that contains three
components:
P ⫽ Pconst ⫹ Pω ⫹ P2ω (8)
where Pconst ⫽ 1/2χ(2)A2 is the constant polarization component; Pω ⫽ χ(1)A sin
ωt is the polarization component with a frequency equal to the pumping wave
frequency; and P2ω ⫽ 1/2χ(2) A2cos(2ωt) is the polarization component with a fre-
quency twice that of the pumping wave frequency.
The SHG process is most efficient when the fundamental and second-
harmonic phase velocities are matched. Collinear phase matching occurs when
the interacting waves have a common wave vector direction. In the case of SHG,
the phase matching requires that the propagation constant of the source polariza-
tion 2kω be equal to the propagation constant k2ω of the second harmonic or:
∆k ⫽ 2k ω ⫺ k 2ω ⫽ 0 (9)
If we present the propagation constants by the refractive indexes (nω, n2ω) and
wavelengths (λω, λ 2ω) of the fundamental and second-harmonic waves, respec-
tively, the condition [Eq. (9)] can be rewritten in the form:
∆k ⫽ 2π[(n 2ω /λ 2ω) ⫺ (2n ω / λ ω) ⫽ 0 (10)
Thus, the phase-matching condition reduces to:
n 2ω ⫽ n ω (11)
In practice, the most effective and commonly used phase-matching technique is
to utilize anisotropic crystals possessing natural birefringence that can compen-
sate for material dispersion. For example, in a negative (ne ⬍ n0) uniaxial crystal
(see Fig. 2), the fundamental wave is directed into the crystal as an ordinary wave
(o-wave) and the second-harmonic wave is generated as an extraordinary wave
(e-wave). As can be seen from Figure 2, in the direction OA (so-called phase-
matching direction), which forms angle θm (so-called phase-matching angle) with
the crystal axis z, the phase-matching condition [11] is satisfied in the form:
n 02ω ⫽ nωe (12)
The dependence [Eq. (12)] corresponds to the so-called type I phase-matching
condition (ooe-phase-matching) that is one of the possible varieties. Table 1
shows entire possibilities for phase-matching at SHG in optical crystals. It is
essential that while, at Type I matching, two fundamental waves have identical
linear polarizations and the second-harmonic wave has perpendicular polariza-

Copyright © 2002 by Taylor & Francis

Figure 2 Phase matching at SHG in a negative (ne⬍no) uniaxial crystal. The condition
n o2ω ⫽ n eω [see Eq. (12)] is satisfied at θ ⫽ θm .

tion, at Type II matching, two fundamental waves should have reciprocally per-
pendicular polarizations.
To maximize the conversion efficiency in SHG, various techniques usually
based on focusing of the fundamental gaussian laser beams or intracavity SHG
are applied [2,4,5,7]. Figure 3 shows a typical optical arrangement of a continu-
ous-wave intracavity SHG. The SHG crystal is placed at the beam waist of the
intracavity laser mode. This technique provides a high fundamental intensity and
SHG conversion efficiency. However, special measures should be taken to pre-
vent SHG instability caused by the laser mode impurity.
In addition, the material equation for second-order nonlinear processes in
the cartesian coordinate system can be presented in the form:

Table 1 Possible Phase-Matching

Conditions at SHG in Nonlinear Crystals

Nonlinear crystal
Type of phase
matching Negative Positive

Type I ooe eeo

Type II oee eoo

Copyright © 2002 by Taylor & Francis

Figure 3 Typical optical arrangement for intracavity SHG of a solid-state laser. Rmax ,
Rlow, maximum and low reflectance, respectively, of the cavity mirrors.

i ⫽
P (2) 冱ε χ
jk
(2)
0 ijk Ej Ek (13)

In general case, χ (2)

ijk is a three-dimensional tensor having 27 independent coeffi-
cients. However, when one applies a perturbation symmetry condition between
the electromagnetic fields Ej and Ek, the number of independent coefficients re-
duces to 18. Hence, the three-dimensional tensor χ (2)ijk can be replaced by a more
simple and convenient two-dimensional 3 ⫻ 6 tensor, well known as the
Kleinman d-tensor. The use of the d-tensor coefficients is a commonly applied
method for characterizing the properties of nonlinear optical media.
Third- and Higher-Order Harmonic Generation. Third-harmonic genera-
tion (THG) is a third-order nonlinear effect defined by the third-order nonlinear
susceptibility χ(3) [1,3]. It can occur in both acentric and centric material, includ-
ing liquids and gasses. The THG is described by the four-rank tensor | χ(3)(3ω ⫽
ω ⫹ ω ⫹ ω)| . In the terms of Figure 1b, the THG process is achieved when one
of the interacting frequencies ω1 or ω 2 is equal to the double value of the other
frequency: ω 2 ⫽ 2ω1 or ω1 ⫽ 2ω 2. Thus, the interaction is ω1 ⫹ ω1 ⫹ ω1 ⫽ 3ω1
or ω 2 ⫹ ω 2 ⫹ ω 2 ⫽ 3ω 2 (see Fig. 1b). This is the case of so-called direct THG
when the THG process is realized entirely in a cubic nonlinear medium described
by χ(3). Figure 4 illustrates the THG mechanism by an energy-level diagram.
However, the THG process is also possible in a quadratic nonlinear medium
characterized by the second-order nonlinear susceptibility χ(2) that is not com-
pletely zero. In this case THG is realized by frequency mixing of the output
frequency and SHG frequency. This kind of THG is called the quasicubic or
cascade cubic nonlinear process.
In an identical manner, higher-order harmonic generation processes can be
interpreted. The energy diagrams in Figure 4 show also the mechanism for direct
realization of fourth-harmonic generation (FOHG) and fifth-harmonic generation
(FIHG). These processes are characterized by high-order nonlinear susceptibilit-

Copyright © 2002 by Taylor & Francis

Figure 4 Energy-level diagrams illustrate mechanism of direct third-(THG), fourth-
(FOHG), and fifth- (FIGH) harmonic generation.

ies, χ(4) for FOHG and χ(5) for FIHG. They can be described by the expressions
(ω ⫹ ω ⫹ ω ⫹ ω) ⫽ 4ω (FOHG) and (ω ⫹ ω ⫹ ω ⫹ ω ⫹ ω) ⫽ 5ω (FIHG).
However, the FOHG, for example, can be realized as a result of various cascade
processes related to the lower-order susceptibilities χ(2) and χ(3). In this case, the
FOHG can be considered as a cascade process defined by the dependence

ω ⫹ ω ⫹ ω → 3ω → 3ω ⫹ ω → ⫽ 4 ω (14)

This means that the first stage of the cascade FOHG process is the third harmonic
to be generated and the THG frequency to be mixed with the fundamental fre-
quency in order for the FOHG frequency to be obtained. In the next sections,
we will consider concrete examples of practical application of these high-order
harmonic generation processes For effective frequency conversion, the third- and
higher-order harmonic generation processes require fulfilling phase-matching
conditions similar to these for the SHG [see Eqs. (9) to (12)].

Copyright © 2002 by Taylor & Francis

3. Frequency Upconversion
According to the consideration in Sec. II.A.1. (see Fig. 1), the upconversion or
sum-frequency generation (SFG) can be described as a second-order nonlinear
effect at which two lower-energy photons, ω1, ω 2, combine in a nonlinear medium
to generate a higher-energy photon ω 3 where:

ω1 ⫹ ω2 ⫽ ω3 (15)

The SFG process can be analyzed in a similar way to SHG (see Sec. II.A.1.a)
[2–5,7]. Let us assume that two electromagnetic waves propagate in a quadratic
nonlinear medium (χ(3) ⫽ 0) with different frequencies ω1 and ω 2:

E1(z,t) ⫽ E01 cos(ω1t ⫺ k1z) (16)

E2(z,t) ⫽ E02 cos(ω2t ⫺ k2z) (17)

If we replace expression (16) and (17) in the nonlinear material Eq. (4), we obtain

PNL ⫽ χ(2)[E 201 cos2(ω1t ⫺ k1z) ⫹ E 202 cos2(ω2t ⫺ k2z)

(18)
⫹ 2E01E02 cos(ω1t ⫺ k1z) cos(ω2t ⫺ k2z)

Eq. (18) includes the following components:

P0 ⫽ 1/2χ(2) (E 201 ⫹ E 202) (19)

P2ω1 ⫽ 1/2χ(2) cos[2(ω1t ⫺ k1z)] (20)

P2ω2 ⫽ 1/2χ(2) cos[2(ω2t ⫺ k2z)] (21)

Pω1⫹ω2 ⫽ χ(2) E01E02 cos[(ω1 ⫹ ω2)t ⫺ (k1 ⫹ k2)z] (22)

Pω1⫺ω2 ⫽ χ(2) E01E02 cos[(ω1 ⫺ ω2)t ⫺ (k1 ⫹ k2)z] (23)

The availability of the polarization component [Eq. (22)] confirms the fact that
an SFG with a higher frequency ω 3 ⫽ ω1 ⫹ ω 2 and a propagation constant k3
⫽ k1 ⫹ k2 are generated in the medium. Similarly to the case with SHG (see
Table 1 and Fig. 2), the SFG process also requires satisfaction of phase-matching
conditions. In this case, these conditions have the following forms:

k3 ⫽ kω1⫹ω2 ⫽ k1 ⫹ k2 (24)

n2ω 3 ⫽ n1ω1 ⫹ n2ω2 (25)

Copyright © 2002 by Taylor & Francis

Figure 5 A basic SFG configuration in which a signal laser beam at ω1 and a strong
pumping laser beam at ω 2 are combined by a beam splitter in a nonlinear crystal to generate
a laser beam at the sum frequency ω 3 ⫽ ω1 ⫹ ω 2.

A typical experimental SFG arrangement is demonstrated by Figure 5. A

signal laser beam at ω1 and a strong pumping laser beam at ω 2 are combined by
a beam splitter in a nonlinear crystal to generate a laser beam at the sum frequency
ω 3 ⫽ ω1 ⫹ ω 2.
Of a significant importance for the development of widely tunable (includ-
ing in the UV and VUV) all-solid-state lasers is an SFG variation called optical
parametric amplification (OPA) [2,4,5,7]. It can be considered as a reverse SFG
process in which low-energy photons at frequency ωs (the signal photons) are
amplified by a nonlinear interaction with high-energy photons at frequency ωp
(the pump photons). To be valid according to the law of energy conservation,
additional photons at a frequency ωi (the idler photons) should be simultaneously
generated. Thus, the SFG process can be represented by the dependence:
ωp ⫽ ωs ⫹ ωi (26)
The basic practical applications of the SFG nonlinear processes are associated
with development of tunable optical parametric oscillators (OPO). A principal
OPO configuration is shown in Figure 6. The nonlinear crystal is commonly
placed within an optical Fabry–Perot resonator. It forms by two dichroic dielec-
tric mirrors that have either maximum transmittance at the pump wavelength and
maximum reflectance at the signal wavelength or both at the signal and idler
wavelengths. At a given threshold intensity of the pumping laser, the OPO cavity
provides resonances for the signal or idler (or both) waves.
An important advanced OPO property is its potential for wide spectral tun-
ing of the output laser emission. A conventional and simple method of OPO
tuning is to rotate the crystal or to change its orientation angle θp (see Fig. 6)
relative to the axis of the Fabry–Perot cavity. Concrete examples of widely tun-

Copyright © 2002 by Taylor & Francis

Figure 6 Principal OPO arrangement in which a nonlinear crystal, placed in a Fabry-
Perot cavity with resonance frequencies ω S and ω1 satisfying the OPO condition ω p ⫽ ω S
⫹ ωI [see Eq. (26)], is pumped by a laser beam at ω p . As a result, simultaneous amplifica-
tion of the signal (at ω S) and idler (ar ωI ) wave is obtained. The angle θp relates to the
OPO crystal orientation.

able spectra obtained using SFG and OPO techniques are presented in the next
section.

B. UV and VUV Solid-State Laser Sources Based

on Harmonic Generation
In the last decade when high-intensity ultrashort and diode-pumped solid-state
laser systems have become commercially available, development of tunable all-
solid-state UV and VUV lasers represents an attractive alternative to the well-
established excimer UV/VUV laser [9]. The reason for the dramatically increas-
ing interest in the solid-state concept for short-wavelength lasers can be attributed
to some of its advanced features: functional simplicity and compact construction;
turnkey reliability; high average laser power and repetition rate; low operating
cost; and removal of toxic gases. Among the known solid-state laser systems,
the nonlinear frequency conversion, based on fundamental nonlinear effects de-
scribed in Sec. II.A., are the most common short-wavelength laser techniques.
Usually these techniques include generation of tunable UV and VUV laser radia-
tion using harmonic generation or frequency mixing in nonlinear crystals. In this
section, we will summarize the main results achieved by nonlinear frequency
conversion techniques.

1. UV/VUV Material Limitations

With respect to the basic material characteristics of the nonlinear crystals as well
as supporting optical elements, the UV and VUV ranges impose some specific

Copyright © 2002 by Taylor & Francis

limitations that can be summarized in the following main groups [4,6]: relatively
short spectral transmittance; relatively low lifetime of the laser optics; availability
of a temporal walk-off effect caused by the different group velocities in the case
of ultrashort laser pulses; and small available sizes of some nonlinear crystals.
An additional limitation exists for laser radiation with wavelength shorter than
185 nm: air absorption caused mainly by absorption in the Schumann–Runge
bands of oxygen. Due to this absorption, laser radiation in this spectral range
should be propagated in vacuum. In fact, the wavelength of 185 nm defines the
boundary between the UV (185–400 nm) and VUV (105–185 nm) ranges. As
an illustration of one of the basic UV/VUV material limitations, Figure 7 shows
the temperature dependence of the short-wavelength transmission limit of various
conventional UV/VUV window materials [11,13]. The cutoff wavelength of all
materials drops with temperature, but no material has a cutoff below those of
LiF at 105 nm. Despite the mentioned UV/VUV limitations, nonlinear crystals
and optical materials are now widely commercially available in both the UV and
VUV. Research efforts are directed to the development of new nonlinear crystals
and optical materials with improved parameters.

2. Nonlinear Crystals for Harmonic Generation

The generation of optical harmonics is one of the simplest ways to obtain short-
wave length laser emission. By using various kinds of nonlinear crystals for direct
SHG or cascade harmonic generation processes, we easily can convert laser radia-
tion from the visible and near-infrared (IR) range to the UV and VUV. At present,
the most commonly used SHG nonlinear crystals transparent in the UV and par-
tially in the VUV are KDP, ADP, KTP, LBO, and BBO. These crystals have
been studies and fully characterized both theoretically and experimentally. Basic
properties of the SHG crystals are summarized in Table 2. Some of the newly
developed crystals such as CLBO, CBO, and LB4, which can be also used in
harmonic generation processes, are presented in Table 3 (see Sec. II.C.2.) with
regard to their application in nonlinear mixing processes. In accordance with the
analysis described in Sec. II.A., the most important condition for a nonlinear
process to be maximally effective is fulfilling the phase-matching conditions [see
Eqs. (9), (11), (24), (25)]. In fact, these are the most important requirement char-
acterizing the quality of the nonlinear crystal. For this reason, one of the most
conventional methods for testing the nonlinear crystal quality is to study phase-
matching-angle and temperature dependencies of the output laser power produced
by nonlinear optical interactions. Each of the nonlinear crystals described in Table
2 possesses its own advanced features that may have significant importance in
a given situation. For example, KDP is the crystal that can be grown to large
sizes. However, it has relatively low nonlinear coefficients and small hygroscop-
icity. KTP, on the other hand, has about an order of magnitude higher nonlinear

Copyright © 2002 by Taylor & Francis

Figure 7 Short-wavelength transmission cutoff vs. temperature for various widely used
UV optical materials. (From Refs. 11 and 13.)

Copyright © 2002 by Taylor & Francis

Table 2 Basic Properties of Commonly Used Crystals for Harmonic Generation in the UV and VUV Range

Cutoff of UV SHG Nonlinearity Damage Max aperture

Nonlinear Type of transmission cutoff deff threshold diameter
crystal interaction (nm) (nm) (pm/V) (GW/cm2) (mm) References

KDP ooe 177 487 d36 ⫽ 0.39 23 ⬃400 4, 12, 22, 29, 55
(KH2PO4) oee
ADP ooe 180 525 d36 ⫽ 0.47 25 ⬃100 4, 12, 29, 30, 55
(NH4H2PO4) oee
KTP eoe 350 990 d32 ⫽ 5.0 1 ⬃10 4, 16, 22, 29, 30
(KTiO2PO3) d31 ⫽ 6.5
LBO ooe 155 555 d32 ⫽ 1.16 25 ⬃15 4, 19, 22, 30, 55,
(LiB3O5) eoo 56
BBO ooe 189 411 d32 ⫽ 16.0 10 ⬃15 4, 20, 22, 39, 55
(β-BaB2O4) eoe d31 ⫽ 0.08

Copyright © 2002 by Taylor & Francis

Table 3 Properties of Nonlinear Optical Crystals Used for SFG Processes in UV and VUV Ranges
Minimum wavelength (nm)
achievable at:
Nonlinear Crystal Transparency Nonlinearity
crystal type [µm] deff [pm/V] SHG FOHG SFG References

LBO Negative 0.160–3.2 d32 ⫽ 1.16 277 242.5 160 4, 19, 22, 30, 55,
(LiB3O5) Biaxial 56
BBO Negative 0.189–3.5 d32 ⫽ 16.0 204.8 189 189 4, 20, 22, 39, 55
(β-BaB2O4) Uniaxial d31 ⫽ 0.08
CBO Positive 0.167–3.0 d14 ⫽ 1.08 272.8 236.3 167 53, 55, 56, 63
(CsB3O5) Biaxial
CLBO Negative 0.180–2.75 d36 ⫽ 0.86 236.7 211.7 180 23, 24, 54, 55,
(CsLiB6O10) Uniaxial 56
KB5 Positive 0.182–1.5 d31 ⫽ 0.04 217 194.8 162 18, 55, 56, 60,
(KB5O84H2O) Biaxial d32 ⫽ 0.003 64, 65
LB4 Negative 0.160–3.3 d31 ⫽ 0.15 243.8 218.3 160 55, 56
(Li2B4O7) Uniaxial
KBBF Negative 0.155–3.66 d11 ⫽ 0.76 164 155 155 55, 56, 57
(Kbe2BO3F2) Uniaxial

Copyright © 2002 by Taylor & Francis

coefficient and is nonhygroscopic. Among the other crystals presented, it has the
longest cutoff wavelength in the UV, but only can be grown to relatively small
sizes. The situation with both BBO and LBO is quite similar. They have high
nonlinearity, but relatively small aperture sizes. In the case of ultrashort pumping
pulses, the short length of the nonlinear crystal has some advantages related to
neglecting the influence of such unwanted nonlinear effects such as SFM and
two-photon absorption, as well as the ultimate use of the crystal transparency
window. In the next section, we will present some of the basic results on the
UV/VUV frequency conversion obtained by harmonic generation.

3. UV/VUV Frequency Conversion Using

Harmonic Generation
UV and VUV laser radiation has been generated by means of either direct SHG
or cascade high-order harmonic generation processes. The range of pumping laser
sources used includes both solid-state lasers with fixed wavelength (such as nano-
second [14–30], picosecond [31–34] and cw [35–41] Nd: doped lasers) and tun-
able lasers (such as dye [42–46], Ti: Sapphire [47–52], etc.).
Harmonic Generations of Nd:Doped Lasers. Since the invention of laser
four decades ago, and especially recently with the extensive introduction of laser-
diode pumped systems, neodymium-doped lasers have played a basic role in
solid-state laser technology. In combination with nonlinear frequency conversion
techniques, these lasers provide one of the most effective and low-cost methods
of short-wavelength laser generation. An enormous number of papers have been
written on the generation of harmonics using Nd: YAG, Nd: YLF, Nd: YAB, Nd:
glass, and other Nd :doped lasers. These works cover the range of laser parame-
ters including pulsed [14–34] and continuous wave (cw) [35–41] regimens, nano-
second [14–30] and picosecond [31–34] pulses, and external [35,36] and intra-
cavity [37–41] techniques for harmonic generation.
In the nanosecond range, a highly efficient SHG of a Q-switched Nd :YAG
laser (λ ⫽ 1.064 µm, 9-ns pulse duration, 10-Hz repetition rate, and 200-MW/
cm2 pump-power density) was reported by Xie et al. [14]. The authors used a 14
mm LBO crystal and achieved more than 70% energy conversion efficiency. Up
to 59% energy conversion efficiency of SHG has been obtained at frequency
doubling of a Q-switched Nd:YAG laser (λ ⫽ 1.064 µm, 45-ns pulse duration,
10-Hz repetition rate, and 380-MW/cm2 pump-power density) using a 3.9 mm
KTP crystal [15]. Chen et al. [16] demonstrated a cascade generation of the sec-
ond (λ ⫽ 532.1 nm), third (λ ⫽ 354.7 nm), fourth (λ ⫽ 266.1 nm) and fifth (λ
⫽ 212.8 nm) harmonics of either nanosecond Q-switched Nd :YAG laser (λ ⫽
1.064.2 µm, 8 ns pulse duration, 10-Hz repetition rate, and 250-MW/cm2 pump-
power density) or picosecond mode-locked Nd:YAG laser (λ ⫽ 1.064.2 µm, 1
ns pulse duration, and 2-GW/cm2 pump-power density). The authors used differ-

Copyright © 2002 by Taylor & Francis

ent kinds of BBO crystals and obtained various energy conversion efficiencies
as follows: 58% (68%) for SHG; 23% (no measurement) for THG; 12% (29%)
for FOHG; and 4% (15%) for FIHG when the nanosecond (picosecond) Nd:YAG
laser is used, respectively. Komatsu and colleagues [17] reported the growth and
UV applications of a large-sized (over 50 mm diameter) LBO crystal. The crystal
was tested at generation of the second (achieved 20% conversion efficiency),
fourth, and fifth harmonics of a Q-switched Nd: YAG laser. Umenura and Kato
[18] presented the first experimental study on the sixth-harmonic generation (λ ⫽
177.4 nm) of a Q-switched Nd:YAG laser (λ ⫽ 1.0642 µm, 12 ns pulse duration,
and 80-MW/cm2 pump-power density). They used a 10 mm KBO crystal (see
Table 3 for its basic parameters) and achieved an approximate 1 kW output peak
power.
In the picosecond range, many authors, starting from one of the earliest
works of Shapiro [31] and Glenn [32], have studied SHG process both theoreti-
cally and experimentally. A high-efficiency SHG of a picosecond Nd :YAG laser
(λ ⫽ 1.064 µm, 35-ps pulse duration, and 1.5–4 GW/cm2 pump-power density)
was demonstrated by Hung et al. [33]. In this experiment, an energy conversion
efficiency of 60–65% is achieved when a 15 mm LBO crystal is used. Ebrahimza-
deh and colleagues [34] used a frequency-doubled, mode-locked, Q-switched,
diode-laser-pumped Nd: YLF laser (λ ω ⫽ 1.047 µm, λ 2ω ⫽ 523.5 nm, 182-MHz-
mode-locked and 500-Hz-Q-switched repetition rate, 55 ps pulse duration of a
single pulse in a train of 105 ns) to pump an LBO-based parametric oscillator.
In the cw regime of Nd: YAG lasers, harmonic generation processes have
been studied using both external [35,36] and intracavity [37–41] techniques for
harmonic generation. An external-cavity SHG of a cw Nd :YAG laser (λ ⫽ 1.064
µm, 18-W pump power) in a 6 mm LBO crystal has reported by Yang et al. [35].
In this case, the conversion efficiency was 36%. Ou et al. [36] obtained a much
higher conversion efficiency (85%) for frequency doubling of a cw TEM00-mode
Nd :YAIO3 laser (λ ω ⫽ 1.08 µm, λ 2ω ⫽ 540 nm, 700-mW pump power). As a
nonlinear crystal, a 10 mm KTP crystal inside an external ring cavity is used
Hemmati [37] demonstrated a cw diode-pumped self-frequency-doubled Nd:
YAB laser (λ ω ⫽ 1.063 µm, λ 2ω ⫽ 531 nm, 1-W pump power). The conversion
efficiency obtained from a self-frequency-doubling Nd:YAB crystal is 4%. The
first all-solid-state cw UV laser source with an output power more than 1 W was
presented by Oka et al. [38]. The authors used an intracavity-frequency-doubled
cw Nd :YAG laser (λ ω ⫽ 1.064 µm, λ 2ω ⫽ 532 nm, 2.9-W SHG output power)
as a pumping source for the fourth-harmonic generation (λ 4ω ⫽ 266 nm, 1.5-W
output pump) in a 5 mm long BBO crystal placed inside an external ring cavity.

Harmonic Generations of Tunable Lasers. Using a tunable dye laser (λ

⫽ 410–430 nm, 8 ns pulse duration, and 120 MW/cm2 pump-power density)
pumped by a Q-switched Nd :YAG laser (λ p ⫽ 532.1 nm), Kato [42] obtained

Copyright © 2002 by Taylor & Francis

a SHG-based UV laser radiation tunable in the range 204.8–215 nm. The SHG
process is realized in a 9 mm BBO crystal whose phase-matching angle is
changed from 90 to 72 degrees to obtain a continuously tunable output emission.
Shorter-wavelength UV radiation (up to 201.1 nm) is generated by an SFG pro-
cess when the dye laser’s is SHG mixed with the Nd: YAG pumping wavelength.
Kato [43] mixed the SHG of a tunable dye laser (λ1 ⫽ 613.1–595 nm) and the
fundamental wavelength (λ 2 ⫽ 1.0642 µm) of a Q-switched Nd :YAG laser by
an SFG process in an 8-mm LBO crystal. Tunable UV radiation in the range
232.5–238 nm was thus obtained. Clab and Hessler [44] used an 8.2 mm BBO
crystal to generate continuously tunable UV laser radiation near 200 nm (λ ⫽
197.4–203.2 nm) by SFG mixing of tunable dye laser radiation (λ ⫽ 592.2–608
nm, 5 ns pulse duration, and 50 MW/cm2 pump-power density) with its SHG (λ
⫽ 296.1–304.8 nm, 5 ns pulse duration, and 50 MW/cm2 pump-power density).

Harmonic Generations of Ti :Sapphire Lasers. Among modern laser tech-

nology systems, Ti:sapphire lasers are one of the most suitable laser sources for
the development of all-solid-state nonlinear-conversion-based laser systems in
the UV and VUV. Typically, Ti-sapphire lasers are tunable near 800 nm. How-
ever, using frequency quadruplers based on two sequential SHG stages or other
cascade nonlinear processes, it is easy to move to the 200 nm spectral region,
providing high peak powers and high conversion efficiencies.
Generation of optical harmonics of Ti:sapphire lasers has been realized for
both pulsed (including nanosecond, picosecond, and femtosecond pulses) and cw
Ti:sapphire lasers. The first SHG of a nanosecond Ti:sapphire laser (λ ⫽ 700–
900 nm tuning range; 12 ns and 25 ns pulse duration at λ ⫽ 780 nm and λ ⫽
715 nm, respectively; 25 Hz repetition rate; and 35 mJ output energy at λ ⫽ 780
nm) was demonstrated by Skripko et al. [47]. The authors used a 5 mm LBO
crystal and obtained an SHG tuning range in the interval 350–450 nm with a
conversion efficiency of 30%. In the picosecond range, Nebel and Beigang [48]
reported on the generation of second, third, and fourth harmonic of a mode-locked
Ti:sapphire laser (λ ⫽ 720–850 nm tuning range; 1.5 ps pulse duration; 82-MHz
repetition rate; and 1.75 W output power at λ ⫽ 790 nm). They used three types
of nonlinear crystals (LBO, BBO and LiJO3). By tuning over the whole range
of the Ti: sapphire laser, they obtained a broad tunable range from 205 to 525
nm. In the femtosecond range, an intracavity UV SHG in a 0.057 mm BBO crystal
was realized by Edelstein and co-workers [49]. The authors used a femtosecond,
passively mode-locked, dye laser (λ max ⫽ 620 nm, 49 fs pulse duration) as pump-
ing source and obtained UV SHG femtosecond radiation with the following pa-
rameters: 43 fs pulse duration, 100 MHz repetition rate, 20 mW power. Roter-
mund and Petrov [50] described a high-repetition femtosecond laser system
operating in the 200 nm spectral range based on the FOHG of a femtosecond
Ti:sapphire laser. They used a Tsunami Ti: sapphire laser (Spectra-Physics) with

Copyright © 2002 by Taylor & Francis

an 85 fs pulse duration that provides a tunable range from 760 to 845 nm with
an average power of 2 W at 82-MHz repetition rate. The shortest wavelength
achieved with this system is 193.7 nm [51] Bourzeix et al. reported UV generation
at 205 nm by two frequency-doubling steps of a cw Ti:sapphire laser (λ ⫽ 820
nm, 2.2 W power). The first SHG (λ ⫽ 410 nm) was realized by a 10.7-mm-
long LBO crystal and the second (λ ⫽ 205 nm) by a 13.8 mm BBO crystal.

C. UV and VUV Tunable Solid-State Laser Sources Based

on Nonlinear Frequency Upconversion
The all-solid-state concept for generation of UV/VUV tunable laser radiation is
maximally comparable with the nonlinear frequency upconversion techniques
based on the second-order nonlinear optical effects. Basic physical principles of
these nonlinear effects are presented in Sec. II.A.1. and II.A.3. In practice, the
SFG and OPO frequency upconversions are the most extensively applied fre-
quency upconversion techniques [53–75]. In this section, we summarize the basic
results achieved by these techniques on UV/VUV tunable laser generation.

1. Nonlinear Crystals for Frequency Upconversion

Basic parameters of both conventionally used (BBO, LBO, and KB5) and some
newly developed (CBO, CLBO, and LB4) nonlinear crystals applicable for UV/
VUV SFG and OPO are shown in Table 3, similarly to Table 2 (see Sec. II.B.2).
In fact, the well-known nonlinear crystals KDP and ADP presented in Table 2
have been also widely used for UV SFG since they have a cutoff wavelength of
UV transmission below 200 nm. However, because their near-IR transmission is
limited to about 1.55 µm, they can not be used for SFG processes below 190
nm. BBO and LBO, both discovered by Chen and co-workers in 1984 and 1988,
respectively [20,19], are among the most commonly used frequency upconversion
crystals. BBO possesses a large spectral transmission range with a small absorp-
tion, a large temperature phase-matching bandwidth and relatively high break-
down threshold. It has the largest birefringence among all-VUV transparent non-
linear crystals. LBO combines a very high damage threshold and a high
nonlinearity with a relatively large acceptance angle bandwidth and low birefrin-
gence. It has a small walk-off between fundamental and generated radiation that
allows the use of long crystals and the generation of laser beams with good spatial
mode quality.
LBO and BBO are especially suitable for frequency conversion of high-
intensity laser radiation. Recently developed borate nonlinear crystals CBO (in-
troduced by Wu et al. in 1993 [53]) and CLBO (introduced by Mori et al. [54])
offer advanced features in nonlinear upconversion processes. The CBO crystal
is transparent far into the UV (up to 167 nm) and has relatively high nonlinearity

Copyright © 2002 by Taylor & Francis

and damage threshold. However, the tuning range of this crystal is limited to
about 184 nm due to the small birefringence and the vanishing nonlinear optical
constant. Because of its adequate birefringence and larger effective nonlinear
constant, CLBO has greater potential than LBO for effective frequency conver-
sion below 190 nm. Unfortunately, it has a cutoff wavelength at 180 nm. In
addition, CLBO is hydroscopic. That creates serious troubles also during crystal
fabrication. KB5 has been used as a nonlinear crystal for frequency conversion
in the UV and VUV for more than three decades. It possesses one of the shortest
cutoff wavelengths of UV transmission, which allows it to be used for SFG pro-
cesses below 190 nm. The nonlinearity of KB5 is smaller than that of BBO and
the crystal is hydroscopic. LB4 has relatively lower nonlinear coefficients. How-
ever, its effective nonlinearity is maximum at an angle of 90 degrees. Moreover,
this crystal has a higher damage threshold than LBO and its hygroscopicity is
smaller than for other nonlinear borate crystals. LB4 has a significant potential
in the VUV region below 180 nm. The minimum estimated SFG wavelengths
that can be achieved by LB4 and KB5 are 170.3 nm and 164.6, respectively
[55,56]. In addition, one of the most recently developed nonlinear crystals is also
included in Table 3. It is the KBBF crystal introduced by Chen et al. in 1996
[57]. Among all the crystals presented, KBBF has the shortest cutoff wavelength
UV transmission. It transmits down to 155 nm, which is below 157 nm, the
shortest wavelength achieved by the F2-excimer laser. Although KBBF is not yet
commercially available, the initial experiments with this crystal, when VUV SHG
in the range 184.7–200 nm is observed, as well as some estimated minimum
SFG wavelengths [55] confirm its promising SFG features.

2. UV/VUV Laser Generation Using Nonlinear

Frequency Upconversion
In accordance with the previous analyses on short-wavelength frequency conver-
sion processes and techniques (see Sec. II.A.1, II.A.3., II.C.2.), among the most
suitable and practical techniques leading to UV/VUV-tunable generation are SFG
and OPO frequency upconversions. Here, we will consider the main results
achieved by these techniques in both nanosecond and ultrashort pulse ranges.
In the earlier frequency upconversion experiments conducted in the nano-
second range [59–67], Kato [59] has demonstrated a UV-tunable SFG at 207.3–
217.4 nm by a type I frequency mixing of the fundamental (λ ⫽ 622 nm, 10 ns
pulse duration, 10 Hz repetition rate, and 4.2 MW pump-power) and second har-
monic (λ ⫽ 311 nm and 1.4 MW pump power) of a visible dye laser in a KB5
crystal. Kato [60], using the same type KB5 crystal, obtained a tunable UV SFG
up to 196.6 nm when the fourth harmonic of a Q-switched Nd:YAG laser (λ ⫽
266 nm, 12 ns pulse duration, 10 Hz repetition rate and 20 MW power) is mixed
with a near-IR dye laser (λ ⫽ 721–770 nm and 1.2 MW power). Muckenheim

Copyright © 2002 by Taylor & Francis

et al. [61] used an 8 mm BBO crystal to generate tunable (188.9–197 nm) UV
SFG radiation down to the end of the transparency range below 190 nm. They
realized type I SFG mixing between a tuned dye laser (λ ⫽ 780–950 nm, 15 ns
pulse duration and 4.5 mJ energy) and a frequency-doubled dye laser at fixed
wavelength λ ⫽ 497 nm. A sum-frequency mixing [ω 3 ⫽ ω1 ⫹ ω 2, see Eq. (26)]
of the fourth (λ1 ⫽ 266 nm) or the fifth (λ1 ⫽ 213 nm) harmonic of a pulsed
Nd :YAG laser and the tunable infrared radiation from an OPO using a BBO
crystal (BBO-OPO, λ 2 ⫽ 1.22–1.6 µm) has been presented by Borsutsky et al.
[62]. The SFG is realized in a LBO crystal and produces tunable laser radiation
in the range λ 3 ⫽ 188–242 nm. A similar SFG arrangement, but using a CBO
crystal, has been presented by Kato [63]. The authors obtained a tunable SFG
up to 185 nm by sum-frequency mixing of the fifth harmonic (λ ⫽ 213 nm) of
a Nd:YAG laser and the output of a KTP-OPO tuned in the mid-IR range (2.8–
3.2 µm).
In the range of short (picosecond [68,69] and femtosecond [55,56,70–74])
laser pulses, Nebel and Beigang [68] have employed a cw mode-locked tunable
Ti: sapphire laser (λ ⫽ 720–850 nm, 1.4 ps pulse duration, 82 MHz repetition
rate, and 1.5 W power at λ ⫽ 790 nm) for a sum-frequency mixing of the funda-
mental and third harmonic (λ ⫽ 240–285 nm, 150 mW power) of this laser.
Figure 8 shows the experimental arrangement of the SFG system. The authors
used an 8-mm-long LBO crystal for the SHG (40% conversion efficiency), a 9
mm BBO crystal for the THG (30% conversion efficiency), and an 8.5 mm BBO
crystal for the SFG (0.7% conversion efficiency from the fundamental radiation).
The shortest wavelength they obtained is 192.5 nm. An all-solid-state UV pico-
second optical parametric generator using a type II phase-matched LBO crystal
is described by Izawa et al. [69]. As a pumping laser, the fourth harmonic of an
actively mode-locked Q-switched Nd :YAG laser (λ 4ω ⫽ 266 nm, 70 ps pulse
duration, 3 GW/cm2 pumping intensity) was used. The optical parametric genera-
tor produced a narrow linewidth (0.1 nm FWHM) UV radiation at λ ⫽ 326.7
nm. The first experimental realization of a tunable femtosecond fourth harmonic
generation in the near-VUV range (189–200 nm) by sum-frequency mixing be-
tween the fundamental and third harmonic was demonstrated by Ringling et al.
[70]. The nonlinear SFG scheme is shown in Figure 9. The authors used an ampli-
fied femtosecond Ti: sapphire laser (λ ⫽ 756–800 nm, 150 fs pulse duration, 300
µm single-pulse energy at 1 kHz repetition rate) and obtained tunable UV pulses
with energies as much as 4 µJ at λ ⫽ 200 nm. Utilizing the same type of femtosec-
ond amplified Ti:sapphire laser, Seifert et al. demonstrated a SFG of 200 fs VUV
(172.7–187 nm) pulses in a 1 mm LBO crystal [71]. The sum-frequency mixing is
realized between the Ti: sapphire’s fourth harmonic and parametrically generated
tunable IR pulse (λ ⫽ 1.65–2.15 µm). Petrov et al. [56] have presented a tunable
VUV SFG of 100 fs pulses down to 170 nm employing a 107 µm LB4 crystal.
They used a sum-frequency mixing of the fourth harmonic (λ 4ω ⫽ 189–210 nm)

Copyright © 2002 by Taylor & Francis

Figure 8 A typical experimental configuration of frequency upconversion system based
on sum-frequency mixing (SFM) of the second and third harmonic of a cw mode-locked
Ti: sapphire laser. D, compensation of the group-velocity-difference-caused temporal de-
lay; P, polarization adjustment. (From Ref. 68.)

of a commercial femtosecond Ti: sapphire laser (λ ω ⫽ 756–840 nm, 100 fs pulse

duration, 1 kHz repetition rate, 600 µJ energy) and tunable laser radiation of a
near-IR optical parametric generator (λ ⫽ 1.6–2.5 µm). In this experiment, a
VUV SFG in the range 170–185 nm with conversion efficiency of 4% is
achieved.

D. UV Tunable Fiber Raman Lasers

Optical fibers (OFs) are an effective solid-state medium for broadband optical
frequency conversions in the spectral range covering the UV, VIS, and near-
IR. When powerful laser emission is propagated in OFs, conditions for effective
development of such basic nonlinear processes as stimulated Raman scattering

Copyright © 2002 by Taylor & Francis

Figure 9 An experimental noncollinear SFG set up. BS, beam splitter; F, fundamental,
SH, second harmonic; TH, third harmonic; DC, dichroic mirror; θ, φ, crystal angles. (From
Ref. 70.)

(SRS), self-phase modulation (SFM), and four-photon mixing (FPM) are created
[76–78]. The high efficiency of these processes in OFs can be attributed to the
simultaneously achievement of a long effective length of interaction (10 2 –104 m)
and a high density of the exciting power (higher than 108 W/cm2) because of the
microscopic fiber core diameters. As a result of the mutual operation of the non-
linear effects, broadband continuum spectra are generated in OFs. In principle,
multicascade SRS and, to a lesser degree FPM, form the discrete structure of
the separate spectral components. SPM and FPM lead to broadening of these
components and thus the spectrum is filled and a broadband continuum is ob-
tained. The continuum spectra provides attractive tunable (discrete or continuum)
laser sources for practical applications.
Nonlinear effects in OFs, in particular, SRS and its application for the de-
velopment of fiber Raman lasers, have been investigated considerably more inten-
sively in the near-IR and visible spectral regions, in comparison with the UV,

Copyright © 2002 by Taylor & Francis

mainly due to very low attenuation losses and low material dispersion of OFs in
these regions. Nevertheless, investigations of SRS and its applications in the UV
region have been motivated by continuously increasing interest in applications
in various research areas associated with shorter wavelength range, as well as
the availability of conventional UV fibers with low losses reduced to Rayleigh
scattering limit.
Nonlinear properties of the silicon-based OFs (on the base of SiO2) are
determined by the third-order nonlinear susceptibility χ(3). These media are cen-
trosymmetrical and the quadratic nonlinear susceptibility χ(2) is zero (see Sec.
II.A.). The various nonlinear effects are related to the two χ (3)
S components ac-
cording to the expression:

χ (3) ⫽ χ (3)
R ⫺ iχ I
(3)
(27)

where the real (nonresonance) part χ (3)

R determines SFM, FPM, and optical Kerr
effect, and the imaginary part χ (3)
I determines SRS and stimulated Brillouin scat-
tering. In comparison with typical conventional nonlinear media having non-
waveguide structure, the value of the nonlinear susceptibility χ(3) for SiO2 is sig-
nificantly lower. For instance, while χ(3) for SiO2 is 1.4 ⫻ 10⫺22 m2 /V2, χ(3) for
the nitrobenzene is 39 ⫻ 10⫺22 m2 /V2, and for sulfur-carbonate it is 159 ⫻ 10⫺22
m2 /V2 [76]. This obvious disadvantage of OFs is compensated multiply by their
unique property to maintain high levels of exciting power density at long dis-
tances.

1. Stimulated Raman Scattering in Optical Fibers

Stokes and Anti-Stokes Raman Scattering. Raman scattering is an inelas-
tic photon–phonon interaction of exciting light emission with vibrations (optical
phonons) of the crystal lattice. The incidence electromagnetic field with optical
frequency ω p causes changes in the electron configuration as well as displace-
ments of the molecular nuclei. As a result, molecules of the optical medium start
to vibrate with their own frequencies ω R (so-called Raman frequencies), which
are specific unique features of each optical medium. An opposite effect is incurred
too. The molecular vibrations with frequencies ω R modulate the incidence light
field with frequency ω p . In this way, light emission with combined frequencies
ω S ⫽ ω p ⫺ ω R and ω AS ⫽ ω p ⫹ ω R occurs. Here, the emission with lower-
frequency ω S corresponds to Stokes scattering (red satellite) and the emission
with higher-frequency ω S to anti-Stokes scattering (blue satellite).
From a quantum-mechanical point of view, we can consider each molecule
as a quantum system that contains discrete electron levels and vibration sublevels.
Consider two vibration levels of an electron level (Fig. 10). The energy level k
is assumed to be a basic (nonexcited) level, the level n is a higher (excited) level,
and the energy difference between k and n is hω R , where h is Planck’s constant

Copyright © 2002 by Taylor & Francis

Figure 10 Energy-level diagrams of (a) Stokes (ω S ⫽ ω p ⫺ ω R ) and (b) anti-Stokes
(ω AS ⫽ ω p ⫹ ω R ) Raman scattering.

and ω R is the Raman frequency. In this case, the Stokes Raman component with
frequency ω S ⫽ ω p ⫺ ω R is obtained (Fig. 10a) when a light energy quantum
hω p excites the molecule from basic level k to a virtual higher level v possessing
a relatively short relaxing time. Then the molecule relaxes to the other level n
and a Stokes light quantum with smaller energy hω S ⫽ hω p ⫺ hω R is emitted.
As can be seen, the Stokes component is obtained from molecules initially only
on the basic level. The anti-Stokes Raman component with frequency ω AS ⫽
ω p ⫹ ω R is formed (Fig. 10b) only from molecules that are already excited to
the upper level n by a preliminary Stokes process. An energy quantum hω p falls
on a molecule excited to the level n. After its fast relaxation from a higher virtual
level v, the molecule returns to the basic level k, emitting an anti-Stokes light
quantum with higher energy hω AS ⫽ hω p ⫹ hω R. The intensity of the Stokes
Raman components is significantly stronger than that of the anti-Stokes compo-
nents, because the molecule population of the basic level k is much higher than
of the excited level n.

Copyright © 2002 by Taylor & Francis

 Spontaneous Raman Scattering. Spontaneous Raman scattering in optical
media is observed at weak exciting light fields. In this case, the molecule vibra-
tions have a fluctuation (incoherent) character. Hence, the scattering light waves
are also incoherent and have 106 –108 times smaller intensity than that of the
exciting light.
The most frequently used OFs are produced on a base of oxide glasses
(SiO2). These fibers possess an amorphous structure. In such optical media, fre-
quencies of the molecule vibrations form nonhomogeneous broadened bands that
are mutually covered and create a continuum frequency distribution. Because of
this, the spontaneous Raman spectrum of silicon glasses has a continuum charac-
ter and covers a frequency shift range from 0 to 1000 cm⫺1. Figure 11 shows

Figure 11 Typical spontaneous Raman scattering spectra of some oxide glasses such
as SiO2, GeO2, and P2O5, which are used in OF production. (From Refs. 77 and 79.)

Copyright © 2002 by Taylor & Francis

typical spontaneous Raman spectra of oxide glasses used for fiber production
[77,79]. The intensity maxima of the spontaneous Raman scattering for both
SiO2- and GeO2- doped fibers are in the range 440–460 cm⫺1. The Raman spec-
trum of P2O5-doped fibers has two maxima, at 650 cm⫺1 and 1330 cm⫺1, respec-
tively.
Stimulated Raman Scattering. SRS is observed at strongly (laser) excited
light fields. It can be characterized by the following features. First, the incidence
light wave excites phase-matched stimulated vibrations and the scattered light is
coherent (see Fig. 12a). Second, stimulated vibrations are accumulated, the vibra-
tion level population is changed, and conditions for anti-Stokes Raman scattering
are created. Third, besides the Stokes and anti-Stokes first-order components, the
scattered light contains components of higher orders (ω p ⫾ n ω R), that is, a cas-
cade (or multi-Stokes) conversion of optical frequencies is developed (Fig. 12b).
Fourth, with regards to the level of the exciting light power, the SRS is a threshold
process. Finally, the spectral width of the SRS generation is narrower and the
intensity of the SRS components is much stronger (about e10 –e20 times) than at
spontaneous Raman scattering.
The process of SRS in OFs has specific behavior due mainly to the fact that
the OF, as a waveguide medium with long effective length and small transverse
dimensions, provides maintenance of high pump power density over a long dis-
tance. Spatial evolution of the SRS along the fiber length can be described in

Figure 12 Energy-level diagrams of (a) stimulated and (b) cascade (or multi-Stokes)
Raman scattering.

Copyright © 2002 by Taylor & Francis

the following manner. In the initial part of the OF length, a spontaneous Raman
scattering signal is obtained. This signal is weak; however, it covers the whole
frequency region corresponding to the spontaneous Raman scattering. When the
pumping power reaches the SRS threshold, the weak signal from the spontaneous
scattering is amplified according to an exponential law that can be expressed [76]
as;
S(ωS) ⬃ exp[gR(ωS)] (28)
Here S(ω S) is the spectrum observed at Stokes frequencies and gR(ω S) is the
Raman gain. The exponential amplification is strongly nonuniform and it is maxi-
mally for those lines at the maximum of spontaneous spectra, that is, with maxi-
mum Raman gain. This process of nonuniform amplification of SRS lines leads
to a significant reduction of their spectral linewidth. On the other hand, because
of the exponential increasing power of the first Stokes order (S1), an effect of
pump power depletion appears. Then, the S1 power reaches the threshold level
needed for excitation of the second Stokes order (S2). At this moment, saturation
of the Raman gain is achieved and the S2 power increases exponentially. Hence,
here S1 is in the role of pumping emission. The processes leading to the S2 genera-
tion (and to the generation of higher-order Stokes components S3, S4, S5, etc.,
respectively) are identical with those for the S1 generation. Thus, a multi-Stokes
generation is realized (see Fig. 12b) and the Stokes part of the SRS spectrum is
formed. Simultaneously, each Stokes component prepares the optical medium for
generation of the corresponding anti-Stokes component from the SRS spectrum.
For analytical description of the SRS in OFs, two basic theoretical models
developed by Smith [80] and AuYeung and Yariv [81,82] can be applied. Smith’s
model is used at relatively weak nonlinear interactions when the pump power
threshold for the SRS generation should be estimated. AuYeung and Yariv’s
model, improved by Yijiang and Snyder [83], is applied at an effective Stokes
(especially at multi-Stokes) conversion and includes the effects of pump depletion
and Raman gain saturation. According to Smith’s model, the evolution of a Stokes
power signal PS(L) along the fiber length can be presented in the form:
PS(L) ⫽ PS(0) exp[(⫺αL) ⫹ (gRPp(0)/Aeff)Leff] (29)
Here, Pp(0) is the pump power at the fiber beginning, gR is the Raman gain of
the SRS in the medium, α is the linear attenuation coefficient. Leff is the effective
length of the OF and can be characterized by the expression [76]:
Leff ⫽ [1 ⫺ exp(⫺αL)]/α. (30)
Aeff is the effective transverse area of the OF that, in the case of multimode OF,
coincides with the transverse area of the OF core. However, in the case of single-
mode or graded-index, OF is given by the formula Aeff ⫽ 2π(ωeff )2, where the
effective radius of the OF core ωeff can be estimated by the dependence:

Copyright © 2002 by Taylor & Francis

 ωeff ⫽ a[0.65 ⫹ 1.619/(Vp)3/2 ⫹ 2.879/(Vp)6] (31)
Here, a is the OF core radius and Vp is the normalized frequency at pumping
laser wavelength λ p . The Raman threshold power PRth is defined as the input pump
power Pp(0) at which the power of the Stokes component PS(L) and the pump
power Pp(L) at the OF end are equal. Using the above dependencies, we can
obtain a practical formula that allows the Raman threshold power P Rth to be esti-
mated:
P Rth ⫽ 16(bAeff)/gR Leff) (32)

where b is a coefficient depending on the polarizing properties of both the pump-

ing laser emission and OF.
The efficiency of the SRS in OFs depends on the properties of both optical
medium and exciting laser emission. These properties can be summarized in three
main groups: (1) material parameters, (2) waveguide parameters, and (3) parame-
ters of the pump laser emission. The first group includes such material parameters
of the fiber-optic medium as spontaneous Raman spectra (see Fig. 11), Raman
gain, material absorption among others. From Figure 11’s Raman spectra of
glasses doped with SiO2, GeO2, and P2O5, it can be seen that the maximum-
frequency Raman shift (1330 cm⫺1) has P2O5-doped OF. Larger Raman shifts of
3000 cm⫺1 and 4200 cm⫺1 are achieved when D2- and H2-doped OFs are used,
respectively [84,85]. Moreover, according to the same Raman spectra, the Raman
gain of GeO2-doped OFs is about 10 times higher than that of SiO2-doped OFs.
However, the SiO2-doped OFs have low attenuation losses (below 1 dB/km),
while the losses in GeO2-doped OFs exceed 50 dB/km. P2O5-doped OFs have
similar behavior. Hence, depending on the concrete requirements of the output
Raman spectra, optimum material OF parameters should be found. The second
group of OF parameters includes such waveguide OF parameters as attenuation
losses (α), effective OF length (Leff ), and transverse area (Aeff ), optical birefrin-
gence, among others. In accordance with the dependencies [29–32], a decreasing
of attenuation losses leads to increasing effective OF length. In combination with
a decreasing of the OF transverse dimensions this leads to decreasing SRS thresh-
old power and increasing of SRS output power. Here, as with the material OF
parameters, we have to optimize waveguide properties of the OFs depending on
our purposes. The third group of parameters includes properties of the pumping
laser emission: pump power, pulse duration, and spectral linewidth. For an effec-
tive SRS process in OF, the pump laser power should be within the interval
between two limitations. It should be higher than the SRS threshold power and
lower than the critical laser power required to damage the OF material. Typical
experimental data for pump power densities needed for SRS in OF are power
densities in the interval 108 –1011 W/cm⫺1 [76,77]. Depending on the laser pulse
duration, two basic regimens of the SRS in OFs are observed [76]. If the pulse

Copyright © 2002 by Taylor & Francis

duration is longer than a typical time for transverse relaxation of excited states,
a quasistationary (or quasi-cw) SRS regime in the OF medium is settled. It is a
situation typical for pump laser pulses with nanosecond (1–100 ns) duration gen-
erated by lasers with Q-switched modulation. When the pulse duration is shorter
than the specific relaxation time, we have a nonstationary SRS regimen in the
OF. This regimen is typical for ultrashort laser pulses (⬍100 ps). In this case,
an effect of increasing of the group-velocity dispersion occurs and special tech-
niques for compensation of this effect need to be applied.

2. Four-Photon Mixing in Optical Fibers

Four-photon mixing (FPM) is a nonlinear optical process in which two exciting
photons with frequencies ω p1 and ω p2 are converted into a Stokes ω S and an anti-
Stokes ω AS photon by analogy with SRS, because ω AS ⬎ ω p1 ⬎ ω S (see Sec.
II.A.1). In basic FPM variant, the exciting photons possess equal frequencies (ω p1
⫽ ω p2 ⫽ ω p) that, in combination with the signal frequencies ω AS and ω S, satisfy
the law of energy conservation

2ωp ⫽ ω AS ⫹ ωS (33)

The FPM frequencies ω AS and ω S are symmetrically disposed towards ω p and

are shifted by the interval ∆ω F that can be expressed by the dependence

∆ωF ⫽ ω p ⫺ ωS ⫽ ω AS ⫺ ωp (34)

The FPM frequency shifts of ∆ω F in OFs cover a broad spectral range between
100 cm⫺1 and 5500 cm⫺1 [76,77,86,87].
The principal difference between the SRS and FPM is that while the SRS
is a self-matched process, the FPM needs a phase-matching condition to be satis-
fied. This condition is necessary because of differences existing between the prop-
agation constants of the interacting waves and, in the case of fiber-optic media,
it can be described by the expression (see Fig. 13):

∆k ⫽ 2kp ⫺ kAS ⫺ KS ⫽ 0 (35)

Figure 13 Collinear phase-matching condition [∆k ⫽ 2kp ⫺ kAS ⫺ kS ⫽ 0, Eq. (35)]

in an OF required at four-photon mixing (FPM) between the pumping, anti-Stokes, and
Stokes waves having wave vectors kp, kAS , and kS , respectively.

Copyright © 2002 by Taylor & Francis

Here, kp, kAS, and kS are wave vectors of exciting, anti-Stokes, and Stokes waves,
respectively. In nonwaveguide media with normal material dispersion, collinear
phase matching is impossible because always (kAS ⫹ kS) ⬎ 2kp . However, in the
case of waveguide media (including OFs), it is possible for the material dispersion
to be compensated and, therefore, collinear phase matching conditions over long
interaction lengths can be created. In multimode OFs, phase-matching conditions
are realized when the interacting waves (exciting, Stokes, and anti-Stokes) are
propagated in waveguide modes for which the material dispersion is compensated
by the mode dispersion. In single-mode OFs, the material dispersion is compen-
sated by variations of the waveguide dispersion.

3. Self-Phase Modulation in Optical Fibers

SPM is a nonlinear effect (similar to the optical Kerr effect in the time scale)
defined by the intensity dependence of the refractive index of the optical medium.
If a light pulse with intensity I(t) is propagated in an OF, this dependence can
be written in the form [8]:
n(t) ⫽ n0 ⫹ n2I(t) (36)
where n0 and n2 are linear and nonlinear parts of the refractive index n (t), respec-
tively. SPM is a threshold process and the minimum exciting power P Sth necessary
for obtaining it is given by an expression similar to the Eq. (32) related to the
SRS threshold power P Rth [8,76]:
P Sth ⫽ (λ pAeff)/(2 πn2Leff) (37)
Using dependencies from Eqs. (29–32) and (37), a comparable estimation of the
threshold powers P Sth and P Rth can be made. If we assume the following typical
values of the input parameters: λ p ⫽ 1.064 µm; 2a ⫽ 50 µm; L ⫽ 1 km; α ⫽
4.6 ⫻ 10⫺6 cm⫺1 (2 dB/km); gR ⫽ 9.20 ⫻ 10⫺12 cm/W; and n2 ⫽ 3.2 ⫻ 10⫺16
cm2 /W [8,76], we obtain P Rth ⫽ 850 W and P Sth ⫽ 12 W, and for the relation
P Rth /P Sth ⫽ 70.
Spectral broadening at SPM is observed predominantly at short picosecond
pump pulses when the combined action of SPM and group-velocity dispersion
occurs. Then, an abnormal dispersion occurs in OFs, conditions for propagation
of optical solitons are created, or, at normal dispersion, pulse shortening occurs.
SPM is observed also at nanosecond pump pulses. In this case, the spectral broad-
ening is attributed to the existence of a fine structure in the pump pulse containing
subnanosecond pulses.

4. UV Raman Scattering in Optical Fibers

The broadband continuum spectra generated in OFs as result of mutual action
of such nonlinear effects described in the previous section as SRS, FPM, and

Copyright © 2002 by Taylor & Francis

SPM provide an attractive basis for development of tunable fiber Raman lasers
(FRLs). These laser sources offer a highly efficient and low-threshold means of
nonlinear frequency conversion and widely tunable generation covering the UV,
visible, and near-IR spectral ranges. Two basic operating FRL regimens have
been established. First is the single-pass FRL (Fig. 14a) in which there is no
cavity feedback and multimode cascade Stokes emission is generated. Second is
the fiber Raman oscillator (Fig. 14b), in which an OF is placed inside a Fabry-
Perot cavity. The laser emission can be widely tuned if a frequency-selective
element is used, such as prism–mirror combination, grating, or intracavity inter-
ferometer [76,77]. As a nonlinear medium for the FRLs, small-core single-mode
fibers are usually used. Nevertheless, multimode graded-index fibers possess sev-
eral essential advantages, from the nonlinear frequency conversion point of view,
such as more efficient launching and maintenance of high pump power, higher
overall Stokes power generation, and broadband frequency conversion. They can
therefore, be applied in FRL systems for the generation of widely tunable emis-
sion, especially in the nanosecond range in which effects such as SPM and group-
velocity dispersion are negligible [76,77]. In the visible range, broadband nonlin-
ear frequency conversion in both single-mode and multimode OFs is obtained
by pumping with various laser sources such as the second harmonic (λ p ⫽ 532
nm) of Q-switched or mode-locked Nd: YAG laser or the dye and argon (λ p ⫽
514) laser [76,77,88]. The study of nonlinear frequency conversion in OFs is
especially intensive in the near-IR around the fiberoptic windows at wavelengths
0.85 µm, 1.3 µm, and 1.55 µm, where the material dispersion and attenuation
losses have minimal values. In this range, broadband spectra by both single-pass

Figure 14 Two basic optical arrangements of conventional fiber Raman lasers (FRLs):
(a) single-pass FRL and (b) fiber Raman oscillator.

Copyright © 2002 by Taylor & Francis

and oscillation generation in single-mode or multimode OFs are demonstrated
[76,77]. As a conventional pumping source, Nd: YAG laser at λ p ⫽ 1.064 µm
or λ p ⫽ 1.32 µm is used.
The development of FRLs in the UV range is more complicated, due mainly
to the relatively higher attenuation losses and material dispersion of OFs in this
range. However, the recent progress in manufacturing UV OFs with low attenua-
tion losses comparable to those of near-IR OFs, as well as the increasing interest
in various laser applications associated with shorter wavelength range, have stim-
ulated research efforts directed toward development of UV FRLs.
The first experimental results with FRLs on the short wavelength side have
been demonstrated by Lin and Stolen [88]. They used a 20 kW 10 ns blue dye
laser to pump a 7 µm core-diameter 19.5 m silica-core OF. A combination of
SRS and FPM produced a broad continuum spectrum in the interval 392–537
nm. Furthermore, early experiments on the first-order and second-order Stokes
generation in UV OFs has been presented also by Rothschild and Abad [89], and
by Pini et al. [90], respectively. In Rothschild and Abad’s experiment, a 4-ns
nitrogen laser pulse (λ p ⫽ 337.1 nm) is used for the first-order Stokes generation
in a 200 µm-core-diameter 50 m fused silica OF. The output spectrum contains
a broad asymmetrical Raman band with a peak at λ S1 ⫽ 342.3 nm, corresponding
to a Raman shift of 450 cm⫺1. Pini et al. [90] used both nitrogen (λ p ⫽ 337.1)
and excimer (λ pl ⫽ 308 nm and λ p2 ⫽ 351 nm) pump UV lasers. They obtained
the first and second Stokes components with Raman shifts between 431 cm⫺1
and 499 cm⫺1. After these initial experiments, several consequence papers on
multiple-order Stokes generation [91] and broadband frequency conversion [92]
in multimode UV OFs have been reported. A broadband frequency conversion
by SRS in a 200-µm silica core UV OF has been demonstrated by Pini and
colleagues [91]. As a pumping laser source, a 4 MW peak-power 7 ns pulse-
duration XeCl excimer laser (λ p ⫽ 308 nm) is used. The broadband UV spectrum
contains nine Stokes orders in the 308–350 nm spectral range. Raman spectrum
parameters, including the wavelength of the registered Stokes orders and the cor-
responding Raman shifts, are shown in Table 4. Efficient SRS has been obtained
when a 200 µm core-diameter 200 m fused-silica multimode UV OF is pumped
by an 8 ns pulse-duration XeCl excimer laser (λ p ⫽ 351 nm) [92]. Up to 20th
Stokes orders are observed, which corresponds to a total frequency shift of about
8800 cm⫺1. Pini and co-workers have suggested a combined Raman oscillator–
amplifier system. In an XeCl (λ p ⫽ 308 nm) excimer laser-pumped UV
multimode OF, a power conversion efficiency of 80% is achieved at a pump
power density of 1.1 ⫻ 109 W/cm2.

5. UV Tunable Double-Pass Fiber Raman Laser

In principle, because of the relatively higher total attenuation losses in the UV
OFs, one of the main problems in the UV spectral region is that investigations

Copyright © 2002 by Taylor & Francis

Table 4 Experimentally Obtained Raman
Spectra Parameters

Wavelength Shift
(nm) (cm⫺1)

L 307.9–308.1
S1 312.4 463
S2 316.7 431
S3 320.8 403
S4 325.3 430
S5 329.9 428
S6 334.7 434
S7 339.5 428
S8 344.5 428
S9 349.1 383

Results derived from 105 m long, 100 µm core-

diameter multimode UV OF pumped by the third
harmonic (λ ⫽ 355 nm) of a Q-switched Nd:YAG
laser.

are directed to the development of single-pass FRLs, rather than to fiber Raman
oscillators, has been very successfully realized in the near-IR and visible [76,77].
A new double-pass fiber Raman laser (DFRL) concept was recently suggested
by Ilev and coauthors first in the visible and near-IR [94–96]. This laser provides
a wide tunable range and permits powerful spectral components to be generated,
due to the double passing of the nonlinear converted laser emission through the
OF, the maximum cavity feedback, the use of a multimode OF, and powerful
pumping. Profiting from these important advantages of the DFRL concept, an
alternative all-solid-state UV and blue tunable DFRL was demonstrated in Refs.
97 and 98.
Principle of Operation and General DFRL Properties. The principal opti-
cal arrangement of the DFRL is shown in Figure 15a. It is based on the application
of a simple fiber-optic autocollimation technique with Littrow-prism-tuned emis-
sion [94–96]. The operating principle of the experimental scheme is more similar
to that for the oscillation regime of the FRL than to the single-pass FRL [76,77].
In fact, this scheme corresponds to a double-pass fiber Raman laser regime, be-
cause of both double passing of the nonlinear converted laser emission through
the OF (called Raman OF) and the absence of an output cavity mirror. The dou-
ble-pass regime of the DFRL is realized experimentally by two sequential stages
(see Fig. 15b). First, an initial broadband single-pass continuum is generated
(single-pass generation) after the forward passing of the pump laser emission
through the OF. Second, the broadband continuum is spectrally dispersed and

Copyright © 2002 by Taylor & Francis

Figure 15 (a) Basic experimental configuration and (b) principal idea of the double-
pass fiber Raman laser (DFRL).

reflected from the autocollimation (Littrow) prism. Next, the formatted narrow-
band emission is launched again into the OF for the second passing. This emission
passes through the forward single-pass emission along a spatial overlap length
inside the Raman OF and the useful backward signal is amplified by the powerful
forward pumping. Thus, the basic double-pass regimen (double-pass generation,
Fig. 15b) of the DFRL is realized and narrow-band tunable spectral components
are generated.
The key optical element in the DFRL scheme is the Raman OF used in a
multifunctional regime (Fig. 15). It acts simultaneously as a point laser source
for the formation of collimated input (to the Littrow prism) emission, as a highly
sensitive point receiver for the autocollimation backreflectance signal, and as a
medium for nonlinear optical frequency conversion and generation of a broad-
band continuum spectrum. In order to be realized as a fiber laser generating high-

Copyright © 2002 by Taylor & Francis

power and widely tunable emission, whose parameters are comparable with those
of a conventional broadband tunable laser source, we used a large-core and low-
loss multimode OF. It ensures, compared to small-core single-mode OFs, more
efficient launching and supporting of powerful laser emission, higher damage
threshold of the fiberoptic material, and higher power of overall Raman Stokes
generation [76,77]. A 30 degree Littrow prism (PRL) with a wideband total reflec-
tance coating is used both to achieve autocollimation back reflectance (maximum
cavity feedback) and to disperse the broadband continuum spectrum generated
in the OF. In some cases, to increase the spectral resolution for the autocollima-
tion back reflectance after the first passing of the pump laser emission through
the OF, additional dispersing prisms in the DFRL cavity may be used. The useful
double-pass laser emission is reflected by a beam splitter (BS) that is a dichroic
filter with a maximum reflectance for the Stokes part of the initial single-pass
continuum.
On the basis of the principal DFRL arrangement presented (Fig. 15), an
experimental setup for UV tunable laser generation has been designed (Fig. 16).
As a pumping source, the third harmonic (λ p ⫽ 355 nm) of an 8 ns pulse duration
Q-switched Nd: YAG laser is used. The pump pulse duration corresponds to the
so-called quasi-cw (or long-time) regime [76] and defines a spatial overlap length
(Fig. 15b) between the forward (single-pass) and backward (double-pass) laser
pulses exceeding 1.5 m. This overlap length value is close to the experimentally
established minimal OF length required to create multicascade stimulated Raman
scattering at powerful pumping. It also ensures significant amplification of the
output DFRL emission. As an active Raman gain medium, a multimode UV all-
silica high-temperature OF is used, which has the following parameters: 100 µm

Figure 16 Experimental setup of the UV-tunable DFRL. Nd:YAG, Q-switched Nd:

YAG pumping laser; 3ω, third-harmonic generator; BS, beam splitter; L1, L2, L3, lenses;
D, diaphragm; PR d , additional dispersing prism; PRL , Littrow prism; F, color and neutral
filters; CCD, computerized CCD-camera-based spectrometer.

Copyright © 2002 by Taylor & Francis

core diameter; 110 µm cladding diameter; 105 m length; NA ⫽ 0.22 numerical
aperture; 250–1100 nm operating wavelength region; 130 dB/km, 13 dB/km,
and 14 dB/km attenuation losses at λ ⫽ 355 nm, 488 nm, and 515 nm, respec-
tively. With these features, the OF ensures efficient launching and maintaining
of powerful pump laser emission, high damage threshold at power pumping, high
power, and multicascade Raman Stokes generation. The 30 degree apex angle
Littrow prism (PRL), used both to achieve a maximum laser cavity feedback and
to tune the fiber Raman laser emission, as well as the additional dispersing prism
(PR d ) and focusing lenses L1 and L 2 (with focal length 50 mm), are produced
from UV-grade fused silica with high UV transmission. The cavity objective (O)
is a 13⫻ Newport UV microscope objective. The widely tunable laser emission,
formed after the second backward pass through the OF, is reflected by the beam
splitter (BS; a special dichroic mirror with maximum reflectivity for the Stokes
part of the spectrum) and is collected by a fiberoptic spectrometer (Ocean Optics,
S1000) with 0.6 nm spectral resolution in the 340–550 nm range.
The double-pass fiber Raman laser generates UV and blue laser emission
in two basic regimes: a broad-band multicascade Stokes single-pass generation
and a narrow-band tunable double-pass generation. The output laser emission is
formed by sequential action of the single-pass and double-pass regimes.
Single-Pass DFRL Generation. As a result of the single-pass of the pump
laser emission through the multimode UV Raman OF, an initial broad-band stim-
ulated spectrum (from 360 nm to 527 nm) is generated, shown in Figure 17.

Figure 17 A multi-Stokes Raman spectrum in the 360–527 nm region generated as a

result of a single pass of the pump laser emission (λ p ⫽ 355 nm, 150 kW pump power)
through a 105 m, 100 µm core-diameter UV Raman OF.

Copyright © 2002 by Taylor & Francis

 The following features are observed in the spectrum. First, it clearly shows
the development of multicascade Raman Stokes nonlinear conversion into the
UV Raman OF. As can be seen, up to 21 well-defined discrete maxima corre-
sponding to Raman Stokes orders are obtained. The Raman Stokes shifts are in
the interval 412–486 cm⫺1 in agreement with the results obtained by other authors
[76,77,91]. Second, the Raman Stokes spectrum has a clear, discrete structure.
This result, which was also observed by other authors [91,92,99], represents, in
the field of nonlinear fiber optics, an essential difference in the development of
nonlinear processes in the near-IR and visible regions (where continuum spectra
have been conventionally obtained in many experiments) from that in the UV
region. The discrete structure of the registered multicascade Stokes spectrum
(Fig. 17) can be explained by the relatively different contributions of basic nonlin-
ear processes that developed in the UV fiber. Under the present experimental
conditions (Fig. 16), the dominant nonlinear process in the multimode UV OF
is the SRS, which, as a self-phase-matched process, comprises the energy of
many fiber modes and leads to formation of the discrete spectrum structure. With
regard to self-phase modulation and four-photon mixing, which are the funda-
mental nonlinear processes in OFs that cause a broadening of the Stokes compo-
nents and lead to wideband continuum generation, there are no suitable conditions
for their development for two basic reasons. One is that it is the nanosecond
pump pulse duration (or so-called quasi-cw regimen [76]) that determines that
the influence of self-phase modulation on continuum formation is negligible. The
other is that the relatively high level of attenuation losses and material dispersion
in the UV fibers leads to a considerable reduction in both the typical coherent
length required for effective phase-matching parametric processes and the laser
power carried by separate OF transverse modes in the framework of which para-
metric processes occur. Third, although powerful laser pumping (greater than
150 kW peak pump power corresponding to greater than 1.9 GW/cm2 input pump
intensity) near the damage threshold of the fiber-optic material is applied, the
experimental conditions for the realization of a coherent regimen of broad-band
SRS [100] are not fulfilled. This is due to the high attenuation and dispersion
that increase the power threshold for the achievement of this regimen. We con-
clude that SRS plays a dominant role in the nonlinear processes that developed
in the UV Raman OF and is responsible for the formation of the discrete structure
of the registered broad-band single-pass multiple-Stokes spectrum.

Double-Pass DFRL Generation. Using the double-pass regimen of the

fiber Raman laser, a UV and blue discretely tunable narrow-band emission corre-
sponding to the various Stokes orders over the initial single-pass multicascade
Stokes spectrum (see Table 5; Fig. 17) in the range 360–493 nm is generated.
Figure 18 shows typical spectral distributions of the double-pass fiber Raman
laser emission in the UV regions, where Figure 18a corresponds to λ max ⫽ 371

Copyright © 2002 by Taylor & Francis

Table 5 Raman Spectra Parameters

P S1 S2 S3 S4 S5 S6 S7 S8 S9 S10

λ(nm) 355 361.2 367.4 373.1 379.2 385.4 392.2 398.8 406.1 413.3 420.5
Shift 0 486 469 412 437 426 445 428 448 430 414
(cm⫺1)

S11 S12 S13 S14 S15 S16 S17 S18 S19 S20 S21

λ(nm) 428.3 436 444.2 452.9 461.6 471.4 482.2 493 504.2 515.4 527.1
Shift 429 413 425 435 417 449 475 453 452 430 429
(cm⫺1)

Parameters were obtained in a 180 m long UV OF pumped by a 650 kW power XeCl excimer laser
(λ ⫽ 308 nm).

nm (third Stokes order) and Figure 18b to λ max ⫽ 392 nm (sixth Stokes order).
The spectral linewidth of the generated components is from 70 cm⫺1 (for the low-
order Stokes components) to 380 cm⫺1 (for the high-order Stokes components).
A potential method for more effective laser emission generation using the DFRL
scheme in a narrower spectral range (in comparison to the spectrum in Fig. 17)
involves using a shorter UV Raman fiber. In this case, because of the significant
reduction of the total attenuation losses in the OF, the initial single-pass spectrum
contains more powerful Stokes components. The Raman amplification of the use-
ful laser emission at the second pass through the OF is also much higher. Thus,
we obtain a discrete tunable laser emission with higher (compared with the use
of a longer OF, Fig. 18) output power and, in addition, the relative influence of
the Raman backscattering and competition between the Stokes orders is strongly
reduced. Figure 19 illustrates this scheme’s potential by showing typical spectral
distributions obtained from a double-pass fiber Raman laser with 25 m 100-µm
core-diameter UV fiber at 150 kW peak pump power (the same power level as
when the spectra from Figs. 17 and 19 are obtained), where Fig. 19a corresponds
to λ max ⫽ 361.2 nm (first Stokes order) and Fig. 19b to λ max ⫽ 373.1 nm (third
Stokes order). However, the use of a shorter OF leads to the generation of a
narrower initial single-pass spectrum and, therefore, to a decrease in the laser
tunable range.
It can be summarized that the presented UV DFRL possesses several basic
advantages. (1) It provides a wide tunable range in the UV-blue spectral region,
mainly due to the use of large-core low-loss multimode OFs as well as high
pump power level. (2) It permits relatively powerful spectral components to be

Copyright © 2002 by Taylor & Francis

Figure 18 Typical spectral distributions of the UV DFRL generation.

generated, because of the double-pass laser regimen, maximum cavity feedback,

use of a multimode OF, and powerful pumping. (3) The principal optical scheme
is simplified, compact, and contains very conventional and low-cost optical ele-
ments. Moreover, compared to conventional optical parametric oscillator
schemes, the principal DFRL scheme excludes critical adjustment and relatively
complicated laser cavities and optical elements, as well as strongly temperature-
and phase-matching-dependent optical elements such as nonlinear crystals. Also,

Copyright © 2002 by Taylor & Francis

Figure 19 Spectral distributions of UV DFRL at (a) λ max ⫽ 361.2 nm and (b) λ max ⫽
373.1 nm when a 25 m, 100 µm core-diameter UV Raman OF is used.

the same Raman OF can be used with a wide range of pump wavelengths at
room temperature. The generation of powerful and widely tunable laser emission
covering the UV, visible, and near-IR, DFRL can be used for optical signal pro-
cessing that includes optical pulse shaping and compression, ultrashort pulse gen-
eration, and amplification.
To extend the spectral range of UV FRL generation toward the 200 nm
range, research efforts are directed to development of low-loss UV-transmitting
OFs and high-efficient UV fiber-based delivery systems. UV fibers with attenua-
tion losses below 0.3 dB/m and 1 dB/m for the spectral range around 300 nm
and 250 nm, respectively, are already commercially available. Ilev and Waynant
[101] recently presented a simple alternative technique for UV laser delivery
and high-efficiency direct laser-to-waveguide coupling using an uncoated glass
hollow taper. The operating principle of the uncoated hollow taper is based on
the light grazing-incidence effect. The grazing-incidence-based taper technique

Copyright © 2002 by Taylor & Francis

produces a high-quality smooth profile of the output laser beam and ensures high
UV laser-to-taper and laser-to-fiber coupling efficiencies [102,103].

III. UV AND VUV GENERATION USING SOLID-STATE

RARE-EARTH-ACTIVATED LASER CRYSTALS

In addition to the all-solid-state nonlinear techniques for UV/VUV tunable laser

generation considered in the previous section, another promising alternative has
been the development of solid-state UV/VUV lasers using rare-earth-activated
wide-band-gap fluoride dielectric crystals as the base. The concept of this solid-
state UV/VUV laser includes use of the broadband emission features of the inter-
configurational d–f transmissions of trivalent rare-earth-activated ions (such as
Nd3⫹, Er3⫹, Pr 3⫹, Ho3⫹, Tm3⫹, Yb3⫹, Ce3⫹, Sm3⫹, Eu3⫹, Gd3⫹, Tb3⫹, and Tb3⫹).
These are doped in dielectric host crystals (such as LaF3, YF3, LiYF4, LuF3, Li-
LuF4, LiCaAlF6, etc.) [104–118]. Figure 20 shows a typical energy-level diagram
of Nd3⫹ ions in LaF3. The excitation spectra centered at λ ⫽ 159 nm originate

Figure 20 Energy-level diagram of Nd3⫹ ions in LaF3. The excitation spectra is centered
at λ p ⫽ 157 nm and the VUV 4f 25d→4f 3 fluorescence is observed at λ fl ⫽ 172 nm.

Copyright © 2002 by Taylor & Francis

from transitions from the 4f 3 ground state to the crystal field split 4f 25d levels
of the Nd3⫹ ions in LaF3. The VUV fluorescence observed at λ ⫽ 172 nm origi-
nates from the 4f 25d → 4f 3 electric dipole-allowed transition levels and transfers
to the 5d → 4I11/2 and 5d → 4I13/2 interconfigurational transition levels.
In earlier experiments investigating rare-earth-activated fluoride dielectric
materials, Yang and DeLuca [104] observed broadband UV/VUV (from 165 to
260 nm) fluorescence in Nd3⫹-, Er3⫹-, and Tm3⫹-doped LaF3, YF3, LiYF4, and
LuF3. The first VUV solid-state laser was experimentally constructed by Waynant
and Klein in 1986 [105]. The authors obtained VUV laser radiation at λ ⫽ 172
nm produced by pumping of Nd3⫹ :LaF3 crystal with incoherent Kr *2 radiation at
λ ⫽ 146 nm. Figure 21 shows absorption and fluorescence spectra obtained with
Nd3⫹ : LaF3 crystals containing Nd3⫹ concentration of 0.1%, 0.5%, and 1.0%. Due
mainly to the higher absorption at higher Nd3⫹ concentration, the greatest fluo-
rescence intensity as well as laser action is observed in the lightest (0.1%) doped
crystal. As a pumping technique in this experiment, the authors used a source of
fast electrons that excite high-pressure Kr gas and generate Kr *2 dimers.
Several research groups have been working intensively on UV/VUV fluo-
rescence spectroscopy of rare-earth-activated fluoride dielectric materials. Dubin-
skii, Cefalas, Sarantopoulou, and their coauthors have presented a broad spectrum
of experimental realizations of solid-state VUV laser systems using fluoride di-
electric crystals doped with various rare-earth-activated ions such as Nd3⫹ [106–

Figure 21 Typical absorption and fluorescence spectra of LaF3 crystals doped with 0.1,
0.5, and 1.0% Nd3⫹ under pumping by 146 nm radiation from Kr *2.

Copyright © 2002 by Taylor & Francis

110], Er3⫹ [111], Pr3⫹ [112], Ho3⫹ [113,114], and Tb3⫹ and Tm3⫹ [115,116]. In
the usual case, the authors used as a pumping laser source a pulsed-discharged
molecular F2 laser with the following parameters: λ p ⫽ 157 nm, 12 ns pulse
duration, 12-mJ maximum energy, and 120-mJ/cm2 maximum energy density.
Dubinskii et al. obtained an efficient (14% conversion efficiency) laser action at
λ ⫽ 172 nm from a Nd3⫹ :LaF3 crystal pumped by a molecular F2 laser at: λ p ⫽
157 nm [106]. VUV laser-induced fluorescent spectra of K2YF5 : Nd3⫹ (PFYK:
Nd) and LiYF4 :Nd3⫹ (YLF:Nd) single crystals pumped by the F2 laser were
obtained by Kollia and colleagues [110]. The authors observed eleven and nine
dipole transitions, between the 4I9/2 ground level of the 4f 3 configuration and the
levels of the 4f 25d configuration of the Nd3⫹ ion 9 (see Fig. 1), in YLF and
PFYK crystal hosts, respectively. In the most recent papers [113,114], researchers
obtained VUV laser-induced fluorescent spectra in the spectral range between
120 and 170 nm using an LiLuF4 : Ho3⫹ (LIF) single crystal.
Spectroscopic study of rare-earth-activated fluoride dielectric materials us-
ing other doped ions has been reported by another authors. Wegh et al. [117]
presented an investigation of the 4f 7 energy levels of Gd3⫹ in LiF4 in the VUV
spectral range. The authors carried out high-resolution excitation measurement
using synchrotron radiation. Ehrlich et al. [118] reported on the first observation
of stimulated emission from 5d-4f transition in Ce3⫹ ions in LiYF4. They applied
optical pumping using a KrF laser (λ p ⫽ 249 nm, 25 ns pulse duration) and
obtained a Ce3⫹ : LiYF4 stimulated emission at λ p ⫽ 325.5 nm.
Further research efforts in the field of solid-state rare-earth-activated laser
materials for UV and VUV laser generation are focused on two main directions:
(1) development of laser materials with improved optical properties for effective
UV/VUV stimulated generation; and (2) the search for and development of effec-
tive and relatively simple optical techniques for laser pumping. Detailed consider-
ation of the specific features and potential applications of the rare-earth-activated
fluoride dielectric materials for UV and VUV laser generation is given in the
following chapters.

REFERENCES

1. N Bloembergen. Nonlinear Optics. New York: WA Benjamin, Inc., 1965, pp 1–

119.
2. A Yariv. Quantum Electronics 3rd ed New York: John Wiley & Sons, 1989, pp
378–494.
3. DH Auston. Picosecond nonlinear optics. In: SL Shapiro, ed. Ultrashort Light
Pulses. New York: Springer-Verlag, 1977, pp 123–201.
4. CL Tang. Nonlinear optics. In: M Bass, ed. Handbook of Optics. 2nd ed. New
York: McGraw-Hill, 1995, pp 38.1–38.26.

Copyright © 2002 by Taylor & Francis

 5. CC Davis. Lasers and Electro-Optics. Cambridge: University Press, 1996, pp 508–
560.
6. GL Wood, EJ Sharp. Nonlinear optics. In: RW Waynant, MN Ediger, eds. Electro-
Optics Handbook. 2nd ed. New York: McGraw-Hill, 2000, pp 13.1–13.28.
7. A Yariv. Introduction to Optical Electronics. 2nd ed. New York: Holt, Rinehart
and Winston, 1976, pp 198–244.
8. AE Siegman. Lasers. Mill Valley, CA: University Science Books, 1986, pp 369–
397.
9. RW Waynant, MN Ediger, eds. Selected Papers on UV, VUV, and X-Ray Lasers.
Bellingham, WA: SPIE Optical Engineering Press, 1993, pp 253–322.
10. CL Rhodes. Review of ultraviolet laser physics. IEEE J Quantum Electron QE-10:
153–174, 1974.
11. RW Waynant, RC Elton. Review of short wavelength laser research. IEEE J Quan-
tum Electron 64:1059–1092, 1976.
12. F Zernike, JE Midwinter. Applied Nonlinear Optics. New York: John Wiley &
Sons, 1973, pp 43–74.
13. WR Hunter. Optics in the vacuum ultraviolet. Electro-Optical Systems Design 5:
16–23, 1973.
14. F Xie, B Wu, G You, C Chen. Characterization of LiB3O5 crystal for second-
harmonic generation. Opt Lett 16:1237–1239, 1991.
15. TY Fan, CE Huang, BQ Hu, RC Eckardt, YX Fan, RL Byer, RS Feigelson. Second
harmonic generation and accurate index of refraction measurements in flux-grown
KTiOPO4. Appl Opt 26:2390–2394, 1987.
16. C Chen, Y Fan, R Eckardt, R Byer. Recent developments in barium borate. Proc
SPIE 681:12–19, 1986.
17. R Komatsu, T Sugawara, K Sassa, N Sarukura, Z Liu, S Izumida, Y Segawa, S Uda,
T Fukuda, K Yamanouchi. Growth and ultraviolet application of Li2B4O7 crystals:
generation of the forth and fifth harmonics of Nd:Y3Al5O12 lasers. Appl Phys Lett
70:3492–3494, 1997.
18. N Umemura, K Kato. Phase-matched UV generation at 0.1774 µm in KB5O8 4H2O.
Appl Opt 35:5332–5335, 1996.
19. C Chen, Y Wu, A Jiang, B Wu, G You, R Li, S Lin. New nonlinear-optical crystal:
LiB3O5. J Opt Soc Am B 6:616–621, 1989.
20. C Chen, B Wu, G You, A Jiang, Y Huang. High-efficiency and wide-band second-
harmonic generation properties of new crystal β-BaB2O4. Proceedings, International
Quantum Electronics Conference, Washington, D.C., 1984, paper MCC5.
21. RC Eckardt, H Masuda, YX Fan, RL Byer. Absolute and relative nonlinear optical
coefficients of KDP, KD*P, BaB2O4, LiIO3, MgO:LiNbO3, and KTP measured
by phase-matched second-harmonic generation. IEEE J Quantum Electron 26:922–
933, 1990.
22. WK Jang, Q Ye, D Hammons, J Eichenholz, J Lim, M Richardson, BHT Chai, EW
Vab Stryland. Improved second-harmonic generation by selective Yb ion doping in
a new nonlinear optical crystal Yca4O(BO3)3. IEEE J Quantum Electron 35:1826–
1833.
23. YK Yap, M Inagaki, S Nakajima, Y Mory, T Sasaki. High-power fourth- and fifth-

Copyright © 2002 by Taylor & Francis

 harmonic generation of a Nd:YAG laser by means of a CsLiB6O10. Opt Lett 21:
1348–1350, 1996.
24. N Umemura, K Kato. Ultraviolet generation tunable to 0.185 µm in CsLiB6O10.
Appl Opt 36:6794–6796, 1997.
25. S Imai, T Yamada, Y Fujimori, K Ishikawa. Third-harmonic generation of an alex-
andrite laser in β-BaB2O4. Appl Phys Lett 54:1206–1208, 1989.
26. GC Bhar, S Das, U Chatterjee. Noncollinear phase-matched second-harmonic gen-
eration in beta barium borate. Appl Phys Lett 54:1383–1384, 1989.
27. B Wu, N Chen, C Chen, D Deng, Z Xu. Highly efficient ultraviolet generation at
355 nm in LiB3O5. Opt Lett 14:1080–1081, 1989.
28. G Feugnet, JP Pocholle. 8-mJ TEM00 diode-end-pumped frequency-quadrupled Nd:
YAG laser. Opt Lett 23:55–57, 1998.
29. DN Nikogosyan. Beta Barium Borate (BBO): a review of its properties and applica-
tions. Appl Phys A52:359–368, 1991.
30. DN Nikogosyan. Lithium Triborate (LBO): a review of its properties and applica-
tions. Appl Phys A58:181–190, 1994.
31. S Shapiro. Second-harmonic generation in LiNBO3 by picosecond pulses. Appl
Phys Lett 13:19–21, 1968.
32. WH Glenn. Second-harmonic generation by picosecond optical pulses. IEEE
J Quantum Electron QE-5:284–290, 1969.
33. JY Huang, YR Shen, C Chen, B Wu. Noncritically phase-matched second-harmonic
generation and optical parametric amplification in a lithium triborate crystal. Appl
Phys Lett 58:1579–1581, 1991.
34. M Ebrahimzadeh, GJ Hall, AI Ferguson. Singly resonant, all-solid-state, mode-
locked LiB3O5 optical parametric oscillator tunable from 652 nm to 2.65 µm. Opt
Lett 17:652–654, 1992.
35. ST Yang, CC Pohalski, EK Gustafson, RL Byer, RS Feigelson, RJ Raymakers, RK
Route. 6.5-W, 532-nm radiation by cw resonant external-cavity second-harmonic
generation of an 18-W Nd:YAG laser in LiB3O5. Opt Lett 16:1493–1495, 1991.
36. ZY Ou, SE Pereira, ES Polzik, HJ Kimble. 85% efficiency for cw frequency dou-
bling from 1.08 to 0.54 µm. Opt Lett 17:640–642, 1992.
37. H Hemmati. Diode-pumped self-frequency-doubled neodymium yttrium aluminum
borate (NYAB) laser. IEEE J Quantum Electron 28:1169–1171, 1992.
38. M Oka, LY Liu, W Wiechmann, N Eguchi, S Kubota. All solid-state continuous-
wave frequency-quadrupled Nd:YAG laser. IEEE J Quantum Electron 1:859–866,
1995.
39. DW Anthon, DL Sipes, TJ Pier, MR Ressl. Intracavity doubling of cw diode-
pumped Nd:YAG lasers with KTP. IEEE J Quantum Electron 28:1148–1157,
1992.
40. LR Marshall, AD Hays, A Kaz, RL Burnham. IEEE J Quantum Electron 28:1158–
1163, 1992.
41. M Oka, S Kubota. Stable intracavity doubling of orthogonal linearly polarized
modes in diode-pumped Nd:YAG lasers. Opt Lett 13:805–807, 1988.
42. K Kato. Second-harmonic generation to 2048 A in β-BaB2O4. IEEE J Quantum
Electron QE-22:1013–1014, 1986.

Copyright © 2002 by Taylor & Francis

 43. K Kato. Tunable UV generation to 0.2325 µm in LiB3O5. IEEE J Quantum Electron
26:1173–1175, 1990.
44. WL Glab, JP Hessler. Efficient generation of 200-nm light in β-BaB2O4. Appl Opt
26:3181–3182, 1987.
45. HJ Dewey. Second-harmonic generation in KB5O84H2O from 217.1 to 315.0 nm.
IEEE J Quantum Electron QE-12:303–306, 1976.
46. K Miyazaki, H Sakai, T Sato. Efficient deep-ultraviolet generation by frequency
doubling in β-BaB2O4 crystals. Opt Lett 11:797–799, 1986.
47. GA Skripko, SG Bartoshevich, IV Mikhnyuk, IG Tarazevich. LiB3O5: a highly
efficient frequency converter for Ti: sapphire lasers. Opt Lett 16:1726–1728, 1991.
48. A Nebel, R Beigang. External frequency conversion of cw mode-locked Ti:Al2O3
laser radiation. Opt Lett 16:1729–1731, 1991.
49. DC Edelstein, ES Wachman, LK Cheng, WR Bosenberg, CL Tang. Femtosecond
ultraviolet pulse generation in β-BaB2O4. Appl Phys Lett 52:2211–2213, 1988.
50. F Rotermund, V Petrov. Generation of the fourth harmonic of a femtosecond Ti :
sapphire laser. Opt Lett 23:1040–1042, 1998.
51. S Bourzeix, B de Beauvoir, F Nez, F de Tomasi, F Biraben. Ultra-violet light gener-
ation at 205 nmby two frequency doubling steps of a cw titanium-sapphire laser.
Opt Commun 133:239–244, 1997.
52. S Cussat-Blanc, RM Rassoul, A Ivanov, E Freysz, A Ducasse. Influence of cascad-
ing phenomena on a type I second-harmonic wave generated by an intense femto-
second pulse: application to the measurement of the effective second-order coeffi-
cient. Opt Lett 23:1585–1587, 1998.
53. Y Wu, T Sasaki, S Nakai, A Yokotani, H Tang, C Chen. CsBS3O5: a new nonlinear
optical crystal. Appl Phys Lett 62:2614–2615, 1993.
54. Y Mori, I Kuroda, S Nakajima, T Sasaki, S Nakai. New nonlinear optical crystal:
cesium lithium borate. Appl Phys Lett 67:1818–1820, 1995.
55. V Petrov, F Rotermund, F Noack, J Ringling, O Kittelmann, R Komatsu. Frequency
conversion of Ti: sapphire-based femtosecond laser systems to the 200-nm spectral
region using nonlinear optical crystals. IEEE J Sel Topics Quantum Electron 5:
1532–1542, 1999.
56. V Petrov, F Rotermund, F Noack, R Komatsu, T Sugawara, S Uda. Vacuum ultravi-
olet application of Li2B4O7 crystals: generation of 100 fs pulses down to 170 nm.
J Appl Phys 84:5887–5892, 1998.
57. C Chen, Z Xu, D Deng, J Zhang, GKL Wong, B Wu, N Ye, D Tang. The vacuum
ultraviolet phase-matching characteristics of nonlinear optical KBe2BO3F2 crystal.
Appl Phys Lett 68:2930–2931, 1996.
58. K Kato. Temperature-tuned 90° phase-matching properties of LiB3O5. IEEE
J Quantum Electron 30:2950–2952, 1994.
59. K Kato. Efficient ultraviolet generation of 2073–2174 A in KB5O8 /4H2O. IEEE
J Quantum Electron QE-13:544–546, 1977.
60. K Kato. Tunable uv generation in KB5O84H2O to 1996 A. Appl Phys Lett 30:583–
584, 1977.
61. W Muckenheim, P Lokai, B Burghardt, D Basting, L Physik. Attaining the wave-
length range 189–197 nm by frequency mixing in β-BaB2O4. Appl Phys B 45:259–
261, 1988.

Copyright © 2002 by Taylor & Francis

 62. A Borsutzky, R Brunger, R Wallenstein. Tunable UV radiation at short wavelengths
(188-240 nm) generated by sum-frequency mixing in lithium borate. Appl Phys B
52:380–384, 1991.
63. K Kato. Tunable UV generation to 0.185 µm in CsB3O5. IEEE J Quantum Electron
31:169–171, 1995.
64. JA Paisher, ML Spaeth, DC Gerstenberger, IW Ruderman. Generation of tunable
radiation below 2000 A, by phase-matched sum-frequency mixing in KB5O84D2O.
Appl Phys Lett 32:476–478, 1978.
65. RE Stickel Jr, FB Dunning. Generation of tunable coherent vacuum uv radiation
in KB5. Appl Opt 17:981–982, 1978.
66. B Wu, F Xie, C Chen, D Deng, Z Xu. Generation of tunable coherent vacuum
ultraviolet radiation in LiB3O5 crystal. Opt Commun 88:451–454, 1992.
67. GC Bhar, P Kumbhakar, U Chatterjee, AM Rudra, Y Kuwano, H. Kouta. Efficient
generation of 200–230-nm radiation in beta barium borate by noncollinear sum-
frequency mixing. Appl Opt 37:7827–7831, 1998.
68. A Nebel, R Beigang. Tunable picosecond pulses below 200 nm by external fre-
quency conversion of cw modelocked Ti:Al2O3 laser radiation. Opt Commun 94:
369–372, 1992.
69. J Izawa, K Midorikawa, M Obara, K Toyoda. Picosecond ultraviolet optical para-
metric generation using a type-II phase-matched lithium triborate crystal for an
injection seed of vuv lasers. IEEE J Quantum Electron 33:1997–2000, 1997.
70. J Ringling, O Kittelmann, F Noack, G Korn, J Squier. Tunable femtosecond pulses
in the near vacuum ultraviolet generated by frequency conversion of amplified Ti:
sapphire laser pulses. Opt Lett 18:2035–2037, 1993.
71. F Seifert, J Ringling, F Noack, V Petrov, O Kittlemann. Generation of tunable
fentosecond pulses to as low as 172.7 nm by sum-frequency mixing in lithium
triborate. Opt Lett 19:1538–1540, 1994.
72. GPA Malcolm, M Ebrahimzadeh, AI Rerguson. Efficient frequency conversion of
mode-locked diode-pumped lasers and tunable all-solid-state laser sources. IEEE
J Quantum Electron 28:1172–1178, 1992.
73. MT Reiten, RA Cheville, NJ Halas. Phase matching and focussing effects on non-
collinear sum frequency mixing in the near VUV region. Opt Commun 110:645–
650, 1994.
74. J Ringling, O Kittelmann, F Noack. Efficient generation of subpicosecond seed
pulses at 193 nm for amplification in ArF gain modules by frequency mixing in
nonlinear optical crystals. Opt Lett 17:1794–1796, 1992.
75. C Chen, Y Wang, B Wu, K Wu, W Zeng, L Yu. Design and synthesis of an ultravio-
let-transparent nonlinear optical crystal Sr2Be2B2O7. Nature 373:322–323,
1995.
76. G Agrawal. Nonlinear Fiber Optics. 2nd ed. London: Academic Press, 1995.
77. C Lin. Nonlinear optics in fibers for fiber measurements and special device func-
tions. J Lightwave Technol LT-4:1103–1115, 1986.
78. S Sudo, H Itoh. Efficient non-linear optical fibres and their applications. Opt Quan-
tum Electron 22:187–212, 1990.
79. F Galleener, J Mikkelsen, R Geils, W Mosby. The relative Raman cross sections
of vitreous SiO2, GeO2, B2O3, and P2O5. Appl Phys Lett 32:34–36, 1978.

Copyright © 2002 by Taylor & Francis

 80. R Smith. Optical power handling capacity of low-loss optical fibers as determined
by stimulated Raman and Brillouin scattering. Appl Opt 11:2489–2498, 1972.
81. J AuYeung, A Yariv. Spontaneous and stimulated Raman scattering in long-loss
fibers. IEEE J Quantum Electron QE-14:347–352, 1978.
82. J AuYeung, A Yariv. Theory of CW Raman oscillation in optical fibers. J Opt Soc
Am 69:803–810, 1079.
83. C Yijiang, A Snyder. Saturation and depletion effect of Raman scattering in optical
fibers. J Lightwave Technol 7:1109–1119, 1989.
84. A Chraplyvy, J Stone, C. Burrus. Optical gain exceeding 35 dB at 1.56 µm due
to stimulated Raman scattering by molecular D2 in a solid silica optical fiber. Opt
Lett 8:415–417, 1983.
85. J Stone, A Chraplyvy, C Burrus. Opt Lett 7:297–299, 1982.
86. C Lin, M Bosch. Large-Stokes-shift stimulated four-photon mixing in optical fibers.
Appl Phys Lett 38:479–681, 1981.
87. E Dianov, E Zahidov, A Karasik, P Mamishev, A Prohorov. Stimulated parametric
four-photon mixing in glass fibers. Sov Phys JETF Lett 34:38–42, 1981.
88. C Lin, R Stolen. New nanosecond continuum for excited-state spectroscopy. Appl
Phys Lett 28:216–218, 1976.
89. M Rothschild, H Abad. Stimulated Raman scattering in fibers in the ultraviolet.
Opt Lett 8:653–655, 1983.
90. R Pini, M Mazzoni, R Salimbeni, M Matera, C Lin. Ultraviolet stimulated Raman
scattering in multimode silica fibers pumped with excimer lasers. Appl Phys Lett
43:6–8, 1983.
91. R Pini, R Salimbeni, M Matera, C Lin. Wideband frequency conversion in the UV
by nine orders of stimulated Raman scattering in a XeCl laser pumped multimode
silica fiber. Appl Phys Lett 43:517–518, 1983.
92. K Liu, E Garmire. Role of stimulated four-photon mixing and efficient Stokes gen-
eration of stimulated Raman scattering in excimer-laser-pumped UV multimode
fibers. J Opt Soc Am 16:174–176, 1991.
93. R Pini, R Salimbeni, M Vannini, AFMY Haider, C Lin. High conversion efficiency
ultraviolet fiber Raman oscillator-amplifier system. Appl Opt 25:1048–1050, 1986.
94. I Ilev, H Kumagai, K Toyoda. A widely tunable (0.54–1.01 µm) double-pass fiber
Raman laser. Appl Phys Lett 69:1846–1848, 1996.
95. I Ilev, H Kumagai, K Toyoda. A fiber-optic autocollimation refractometric method
for dispersion measurement of bulk optical materials using a widely tunable fiber
Raman laser. IEEE J. Quantum Electron 32:1897–1902, 1996.
96. I Ilev, H Kumagai, K Toyoda. A powerful and widely tunable double-pass fiber
Raman laser. Opt Commun 138:337–340, 1997.
97. I Ilev, H Kumagai, K Toyoda. Ultraviolet and blue discretely tunable double-pass
fiber Raman laser. Appl Phys Lett 70:3200–3202, 1997.
98. I Ilev, R Waynant, H Kumagai, K Toyoda. Double-pass fiber Raman laser—a pow-
erful and widely tunable in the ultraviolet, visible and near-infrared fiber Raman
laser for biomedical investigations. IEEE J Sel Topics Quantum Electron 5:1013–
1018, 1999.
99. K Liu, E Garmire. Understanding the formation of the SRS Stokes spectrum in
fused silica fibers. IEEE J Quantum Electron 27:1022–1030, 1991.

Copyright © 2002 by Taylor & Francis

100. I Ilev, H Kumagai, K Toyoda, I Koprinkov. Highly-efficient wideband continuum
generation in a single-mode optical fiber by powerful broadband laser pumping.
Appl Opt 35:2548–2553, 1996.
101. I Ilev, R Waynant, M Ediger, M Bonaguidi. Ultraviolet laser delivery using an
uncoated hollow taper. IEEE J Quantum Electron 36:944–948, 2000.
102. I Ilev, RW Waynant. Grazing-incidence-based hollow taper for infrared laser-to-
fiber coupling. Appl Phys Lett 74:2921–2923, 1999.
103. I Ilev, R Waynant. Uncoated hollow taper as a simple optical funnel for laser deliv-
ery. Rev Sci Instrum 70:3840–3843, 1999.
104. K Yang, J DeLuca. VUV fluorescence of Nd3⫹-, Er3⫹-, and Tm3⫹-doped trifluorides
and tunable coherent sources from 1650 to 2600 A. Appl Phys Lett 29:499–501,
1976.
105. RW Waynant, PH Klein. Vacuum ultraviolet laser emission from Nd⫹3:LaF3. Appl
Phys Lett 46:14–16, 1985.
106. MA Dubinskii, AC Cefalas, E Sarantopoulou, SM Spyrou, CA Nicolaides. Efficient
LaF 3:Nd3⫹-based vacuum-ultraviolet laser at 172 nm. J Opt Soc Am B 9:1148–
1150, 1992.
107. MA Dubinskii, AC Cefalas, E Sarantopoulou, R Yu. Abdulsabirov, SL Korableva,
AK Naumov, VV Semashko. On the interconfigurational 4f 25d-4f 3 VUV and UV
fluorescence fratures of Nd3⫹ in LiYF 4 (YLF) single crystals under F 2 laser pump-
ing. Opt Commun 94:115–118, 1992.
108. E Sarantopoulou, AC Cefalas, MA Dubinskii, CA Nicolaides, R Yu, Abdulsabirov,
SL Korableva, AK Naumov, VV Semashko. VUV and UV fluorescence and absorp-
tion studies of Nd3⫹ and Ho3⫹ ions in LiYF4 single crystals. Opt Commun 107:
104–110, 1994.
109. Z Kollia, E Sarantopoulou, AC Cefalas, CA Nicolaides, AK Naumov VV Sem-
ashko, R Yu Abdulsabirov, SL Korableva, MA Dubinskii. Vacuum-ultraviolet in-
terconfigurational 4f 3 → 4f 25d absorption and emission studies of the Nd3⫹ ion in
KYF, YF, and YLF crystal hosts. J Opt Soc Am B 12:L782–785, 1995.
110. Z Kollia, E Sarantopoulou, AC Cefalas, AK Naumov, VV Semashko, RY Abdul-
sabirov, SL Korableva. On the 4f 25d → 4f 3 Interconfigurational transitions of Nd3⫹
ions in K2YF5 and LiYF 4 crystal hosts. Opt Commun 149:386–392, 1998.
111. E Sarantopoulou, Z Kollia, AC Cefalas, MA Dubinskii, CA Nicolaides, RY Abdul-
sabirov, SL Korableva, AK Naumov, VV Semashko. Vacuum-ultraviolet and ultra-
violet fluorescence and absorption studies of Er3⫹-doped LiLuF 4 single crystals.
Appl Phys Lett 65:813–815, 1994.
112. E Sarantopoulou, AC Cefalas, MA Dubinskii, CA Nicolaides, RY Abdulsabirov,
SL Korableva, AK Naumov, VV Semashko. VUV and UV fluorescence and absorp-
tion studies of Pr3⫹-doped LiLuF 4 single crystals. Opt Lett 19:499–501, 1994.
113. AC Cefalas, E Sarantopoulou, Z Kollia. Vacuum ultraviolet 4f 95d → 4f10 intercon-
figurational transitions of Ho3⫹ ions in LiYF 4 single crystals. J Opt Soc Am B 16:
625–630, 1999.
114. E Sarantopoulou, Z Kollia, AC Cefalas. 4f 95d → 4f10 spin-allowed and spin-forbid-
den transitions of Ho3⫹ ions in LiYF 4 single crystals in the vacuum ultraviolet. Opt
Commun 169:263–274, 1999.
115. E Sarantopoulou, Z Kollia, AC Cefalas, VV Semashko, RY Abdulsabirov, AK

Copyright © 2002 by Taylor & Francis

 Naumov, SL Korableva. On the VUV and UV 4f 7(8S)5d → 4f 8 interconfigurational
transitions of Tb3⫹ ions in LiLuF 4 single crystal hosts. Opt Commun 156:101–111,
1998.
116. E Sarantopoulou, AC Cefalas, MA Dubinskii, Z Kollia, CA Nicolaides, RY Abdul-
sabirov, SL Korableva, AK Naumov, VV Semashko. VUV and UV fluorescence
and absorption studies of Tb3⫹ and Tm3⫹ trivalent ions in LiLuF 4 single crystal
hosts. J Mod Opt 41:767–775, 1994.
117. RT Wegh, H Donker, A Meijerink, RJ Lamminmaki, J Holsa. Vacuum-ultraviolet
spectroscopy and quantum cutting for Gd3⫹ in LiYF 4. Phys Rev B 56:13841–13848,
1997.
118. DJ Ehrlich, PF Moulton, RM Osgood Jr. Ultraviolet solid-state Ce:YLF laser at
325 nm. Opt Lett 4:184–186, 1979.

Copyright © 2002 by Taylor & Francis

8
VUV Laser Spectroscopy
of Trivalent Rare-Earth Ions in
Wide Band Gap Fluoride Crystals

Evangelia Sarantopoulou and Alciviadis-Constantinos Cefalas

National Hellenic Research Foundation, Theoretical and Physical
Chemistry Institute, Athens, Greece

I. INTRODUCTION

The absorption and the excitation spectroscopic characteristics of trivalent rare-

earth (RE) ions in the vacuum ultraviolet (VUV) and ultraviolet (UV) spectral
regions, activated in wide band gap fluorine dielectric crystals, are due to transi-
tions between the levels of the 4f n single electronic configuration of the trivalent
RE ion, and the levels of the 4f n⫺15d mixed electronic configuration, where a 4f
electron is promoted to a 5d localized level [1–3]. The 4f n ↔ 4f n⫺15d electronic
transitions are characterized by strong Frank–Condon factors with broadband
absorption and emission spectra in the VUV and UV. In contrast, the intracon-
figurational 4f n → 4f n transitions are parity forbidden. They are forced by the
crystal field configuration mixing and they appear to be weak and sharp.
There are two different experimental methods for exciting a trivalent RE
ion in wide band gap dielectric crystals. The first method uses VUV lasers [4]
and/or x-ray or synchrotron radiation [5,6]. This pumping arrangement has the
advantage of populating the levels of the 4f n⫺15d electronic configuration directly
from the ground state of the trivalent RE ion, via one photon transition only. The
subsequent de-excitation mechanisms within the 4f n⫺15d electronic configuration
efficiently populate the levels of the 4f n single electronic configuration of the
trivalent RE ion [7–9]. They allow one to study the excitation dynamics and the
structure of the levels of the 4f n⫺15d electronic configuration of the trivalent RE

Copyright © 2002 by Taylor & Francis

ions. The second method applies upconversion pumping arrangement with all
solid state laser elements [10–12], which greatly simplifies the experimental ob-
stacles arising from the VUV pumping. The radiative interconfigurational d–f
transitions of the RE activated ions, in the wide band gap of dielectric crystals,
offer the possibility that these materials can be used in a variety of applications
and for generating coherent VUV or UV light. This is attractive due to its relative
simplicity in comparison to existing nonlinear methods using gases and mole-
cules. In the case of optically pumped LiYF 4 :Ce 3⫹ and LaF3: Ce3⫹ crystals [13–
15], laser emission at 325 and 286 nm was obtained, respectively. Waynant and
Klein [16,17] have reported the first laser action in the VUV from solid-state
dielectric crystals. They used the LaF 3 : Nd 3⫹ dielectric crystal to generate laser
action at 172 nm when it was optically pumped by incoherent light (emitted from
excited Kr*2 molecules). With a different pumping arrangement using an F 2 pulsed
discharge molecular laser operating at 157.6 nm, coherent light from the same
crystal at 172 nm was generated [18]. Fluorescent materials based on RE ions
can also be used as high quantum efficiency phosphors [19], for plasma display
screens and mercury-free light tubes [20–22], fast scintillators [23,24] and for
light wave communications [25,26].
In addition to the above-mentioned applications, next-generation micro-
electronics circuits will have minimum dimensions below 100 nm. It is envi-
sioned that 157 nm laser lithography [27,28,28a] will be the next step in optical
lithography. At 157 nm, under VUV illumination of the mask target, lithographic
features with dimensions less then 0.10 µm on the photoresist could be achieved.
However, there are problems related to the design of the optical projection sys-
tem, due to the fact that the absorption coefficient of most materials in the VUV
is large, yet, their optical properties degrade constantly with time under VUV
irradiation. Up to now, only calcium fluoride seems to be promising as optical
material for 157 nm photolithography, and the possibility of using wide band
gap fluoride dielectric crystals, such as YF 4 and LiCaAlF 6, as optical elements
for 157 nm photolithography was investigated. These materials could be grown
from melts at lower temperature than CaF 2 and they have similar physical proper-
ties, despite the fact that they are forming crystals of different symmetry than
cubic. Doped or nondoped wide band gap crystals can be used as passive or
active optical elements in the VUV: i.e., prisms, lenses, filters of variable and
controllable attenuation [28, 28a].
All these applications depend on the development of new dielectric wide
band gap materials, and the structure of the levels of the 4f n⫺15d electronic con-
figuration that specify the d–f transitions of the RE ions. However, despite the
early spectroscopic measurements of the f–d transitions [29], only limited infor-
mation is available in this field. A small amount of data has been taken and
analyzed with the use of VUV lasers, x-ray, or synchrotron light sources by excit-
ing a common RE impurity in different dielectric fluoride crystal hosts. Investiga-

Copyright © 2002 by Taylor & Francis

tion of fundamental physical interactions in the crystal environment is of consid-
erable importance and will determine the response of these materials to VUV
light.

II. VACUUM ULTRAVIOLET SPECTROSCOPY:

METHODS AND TECHNIQUES
A. Molecular Fluorine Laser
Investigating physical processes in the VUV requires the use of intense sources
of photons in this spectral range. For a detailed investigation of the absorption
spectrum, the emission characteristics of the sources are such that either the radia-
tion occurs as a continuum or the emission spectrum consists of closely packed
lines. In addition, the continuous light sources should be stable with respect to
time and the pulsed ones should have good pulse to pulse stability. In most experi-
mental cases, the available continua source suitable for ultraviolet absorption
studies is molecular hydrogen. Its molecular spectrum extending from the visible
to 100 nm is obtainable from a low-pressure positive column discharge. Continua
spectra have likewise been taken from positive column discharges in rare gases,
from 60 to 200 nm. At higher pressures, the continuum in helium discharges
could be extended up to 400 nm. Tunable VUV radiation can be generated from
nonlinear methods using frequency mixing in rare gases and metal vapors [30–
33] and nonlinear crystals [34,35]. However, the nonlinear methods are relatively
complex and characterized by low conversion efficiency and low-output energy.
The major breakthrough in this direction was made in 1973 with the intro-
duction of the excimer laser sources based on rare gas dimers. Strong dipole
transitions between the bound electronic excited state and the dissociative ground
state of diatomic clusters of Xe 2, Kr 2, and Ar 2 in liquid, gas, or solid phase [36–
39] provided a new powerful radiation source in the VUV. These systems were
characterized by high quantum efficiency in the emitted wavelength, high gain,
and conversion efficiency. However, limited tunability and e-beam excitation to
achieve population inversion imposes serious experimental restrictions on their
use for spectroscopic applications.
At the present time, very few laser sources are available in the VUV (Table
1). Among them the ArF excimer laser with limited tunability at 193 nm [40,41],
and the molecular fluorine laser at 157 nm [42–50], are the only efficient laser
sources available on the laboratory scale. Laser transitions at 156.71, 157.48, and
157.59 nm have been observed for the first time by Woodworth and Rice [42,43]
in F 2 /He mixtures excited with an electron gun with high efficiency and in elec-
tric discharge-pumped lasers. The laser transition has been identified as of the
3
Π 2g → 3 Π 2u type and is the same as for the laser transition of the 301.5–346 nm
band in molecular iodine. The energy difference between the first two transitions

Copyright © 2002 by Taylor & Francis

Table 1 VUV Laser Sources

Laser Emission wavelength (nm) Excitation

H2 109.8–164.4 EB
D2 111.3–161.6 PD
Ar 2 126 EB
Kr 2 145.7 EB
F2 157.6 PD
ArCl 169, 175 PD
Xe 2 170 EB
Kr 3⫹ 175.6 CW
Cu 181 PD
Kr 4⫹ 183.2 CW
Ar 4⫹ 184.3 CW
CO 187.8, 189.7, 197 PD
ArF 193 PD
Kr 3⫹ 195 CW
Kr 3⫹ 196 CW

Population inversion is achieved commonly by electron beam

(EB) or pulsed (PD) and continued (CW) plasma discharges.

was attributed to de-excitation from two different rotational levels of the same
vibrational level of the 3 Π 2g state to the 3 Π 2u, and the energy difference between
the first and the third level to de-excitation from two different vibrational levels.
The energy position of the 3 Π u electronic state was found to be 3 eV above the
X 1 ⌺ g ground state from electron-scattering experiments [51]. This state is
slightly bound by 0.15 eV [52].
The energy position of the lower vibrational level of the 3 Π 2g state was
found to be 11.62 eV above the X 1 ⌺ g ground state [53], which is in agreement
with theoretical calculations [54].
The kinetic scheme proposed [44] to explain the above transition in the
plasma discharge assumes the production of metastable helium atoms He* and
e⫺ /He⫹ pairs. The He⫹ ions at high pressure are forming He 2⫹ and He 3⫹ ionized
molecular clusters in the ground or excited states, which, finally, through effective
collisions, either ionize the fluorine molecules or form atomic species:
He n⫹ ⫹ F 2 → F 2⫹ ⫹ nHe (n ⫽ 1,2,3)
He*n ⫹ F 2 → F 2⫹ ⫹ nHe (n ⫽ 1,2) ⫹ e
He*n ⫹ F 2 → F ⫹ F* ⫹ nHe (n ⫽ 1,2,3)
Low-energy electrons can efficiently generate atomic fluorine negative ions
e ⫹ F 2 → F ⫹ F⫺ or e ⫹ F 2⫹ → F ⫹ F*

Copyright © 2002 by Taylor & Francis

and then
F⫺ ⫹ F 2⫹ → F* ⫹ 2F or
F⫺ ⫹ Hen⫹ → F* ⫹ nHe
The final result of the atomic and molecular collisions is the formation of the F 2
molecules in their excited 3 Π 2g electronic state:
F* ⫹ F2 → F2( 3 Π 2g ) ⫹ F, F⫺ ⫹ He → F 2 ( 3 Π 2g ) ⫹ He
Once the F 2 ( 3 Π 2g ) molecular species have been formed, they can be de-
excited by spontaneous and stimulated emission and by collision quenching.
F 2 ( 3 Π 2g ) → F 2 ( 3 Π 2u ) ⫹ hν(157.6 nm)
F 2 ( 3 Π 2g ) ⫹ e⫺ → F 2 ( 3 Π 2u ) ⫹ e⫺
F 2 ( 3 Π 2g ) ⫹ F 2 → 2F 2
Electron excitation [42] is a relatively complicated method in comparison
to the pulsed fast discharge, using LC-inversion or charge transfer circuits (Fig.
1) together with preionization of the gas mixture. Preionization is achieved by
UV light, which is triggered a few nanoseconds before the main discharge
[44,55,56]. As soon as the fast switch S (thyratron or spark-gap) closes (Fig. 1),
the stored energy in the capacitor C 2 is transferred into the discharge volume
through the capacitors C 1. Double preionization (2 ⫻ 80 pins) of the laser volume
ensures a uniform and stable main discharge. The electrodes were semicylidrical
with round surfaces and no sharp edges, fabricated from stainless steel 316, 1
cm wide and 80 cm long, spaced 2 cm apart.
Good preionization in the circuit is essential to shape together with Rogow-
ski profile electrodes, which are relatively difficult to machine in comparison to
the semicylindrical profile. Stability of the discharge main mode was also ensured
by the low inductance of the discharge circuit of 60 nH, which gives a fast rise

Figure 1 Schematic layout of the electric circuit of the ‘‘charge transfer type’’ F 2 laser.
HV, high voltage; R, resistive load; S, spark gap; C 1, C 2, capacitors; A, preionization gap;
G, ground.

Copyright © 2002 by Taylor & Francis

time of the main charging voltage of 50 ns. The discharge volume is 100 ⫻ 1.5 ⫻
0.3 cm3, and this device can deliver 10 mJ at 157 nm per pulse at 2–3 atm total
helium pressure. A typical pulse duration is of the order of 10–20 ns, and its
spectral linewidth at FWHM is less than 0.05 nm [44]. The small signal gain
coefficient of the previous pulsed discharge F 2 laser has been measured using
the passive cell absorption method at 2 atm of the helium buffer gas. It was found
to be 3.2% cm⫺1. This value was half that predicted by theory [47] considering
only dissociative collision of the F 2 molecules by either ion–ion recombination
or energy transfer reaction and neglecting direct excitation of the F 2 molecules
by either electron impact or energy transfer from He* and He*2 molecules.
The small signal gain coefficient using the oscillator–amplifier method was
found to be of the same order of magnitude and the saturation intensity was on
the order of 4.5 MW/cm 2 [49,50]. Despite the fact that the value of the small
signal gain coefficient at 157 nm is comparable to the value of the rare gas halide
transitions of the excimer lasers up to 1988, the output energy at 157 nm never
exceeded 15 mJ. This was mainly due to the presence of small amounts of organic
impurities and air in the commercial gases and the experimental restrictions in
raising the total gas pressure above 3 atm. The absorption coefficient of the air
at 157 nm is higher than 200 cm⫺1 and therefore the net gain competed with the
absorption losses. By increasing the pressure of the helium buffer gas to 8 atm,
112 mJ per pulse were obtained by Yamada et al. [48].

B. VUV Absorption Spectroscopy

Absorption cross-sectional investigations in the VUV require adequate measure-
ment of the ratio of the incident to transmitted radiation. For atomic or molecular
systems that exist in the gas state, accurate measurement of pressure and tempera-
ture is required. Measurement of the absorption coefficient of metal vapors is
made even more difficult by the rapidity with which the reactive vapor attacks
the windows of the absorption cell and by the high temperatures often necessary
for the production of significant concentrations of the atoms.
For many experimental systems, measurement of the absorption coefficient
is prohibited in the wavelength region below the cutoff wavelength of the LiF
window (Fig. 2). For other experimental configurations, the source discharge
gases are separated from the detector and the sample areas by differential pump-
ing slits. In this case the cutoff wavelength is determined by the efficiency of
the gas discharge and the cutoff wavelength of the monochromator and the photon
detector (Fig. 3).
The absorption spectra of trivalent RE ions in the VUV in wide band gap
dielectric crystals are due to the 4f n → 4f n⫺15d electronic transitions between
the ground state of the 4f n single electronic configuration and the Stark levels
of the 4f n⫺15d mixed electronic configuration. Experimental apparatus for ob-

Copyright © 2002 by Taylor & Francis

Figure 2 Emission spectrum from a longitudinal stabilized low-pressure hydrogen col-
umn recorded with a secondary electron multiplier (SEM), of open configuration. The
hydrogen lamp was sealed with an LiF crystal window and it was the only optical element
between the VUV lamp and the SEM. With this experimental configuration, the edge of
the conduction band of wide band gap dielectric crystals can be determined accurately.

taining the absorption spectra consists mainly of a hydrogen VUV light source,
the vacuum chamber where the crystal sample is placed, and the optical and
electronic detection equipment (Fig. 4) [7]. The hydrogen light source operates
in a longitudinally stabilized discharge mode. The discharge’s high stability pro-
vides good signal-to-noise ratio (better than 2000). The optical paths of the light
source and the fluorescence light beams are inside vacuum lines of stainless steel
at 10⫺6 mbar pressure.
The vacuum chamber where the crystal samples are placed is equipped
with a cryogenic facility. The optical detection system consists of a VUV mono-
chromator, a solar blind photomultiplier, or a secondary electron multiplier
(SEM). When a VUV photon is absorbed within the volume of the crystal sample,
new VUV photons are re-emitted within the crystal at the same or different wave-
length due to spontaneous emission from the levels of the 4f n5d electronic con-
figuration of the trivalent RE ions. The intensity of the radiation within the crystal
volume V depends on the position of the atoms within the crystal and the wave-
length [57].

Copyright © 2002 by Taylor & Francis

Figure 3 Emission spectrum from a longitudinal stabilized hydrogen lamp. No optical
element was placed between the hydrogen discharge column and the secondary electron
multiplier (SEM). The hydrogen flow through the discharge cell was controlled with differ-
ential pumping at different stages between the discharge and the SEM.

∂I(λ, x)
∂x
⫽ [N 2 ⫺ N 1]F(λ)
Bhλ
Vnc 2 冢 冣
I(λ, x) (1)

where N 2 and N 1 are the excited and the ground state populations of the atomic
transition at λ, F(λ)dλ is the fraction of the transitions in which the photon fre-
quency lies in a small range dλ about the wavelength λ, V is the total crystal
volume where the interaction of radiation with matter occurs, B is Einstein’s
coefficient, and n is the refractive index of the crystal. The solution of this equa-
tion is more complicated than it first appears because N 2 and N 1 themselves de-
pend on the intensity of the radiation and hence on the position of the atoms
inside the crystal and the wavelength of light. Eq. (1) takes the form

∂I(λ, x)
⫽ [a(λ, x) ⫺ β(λ, x)]I(λ, x) (2)
∂x

where a(λ, x) is the gain coefficient that describes photon generation through
spontaneous emission and β(λ, x) is the absorption coefficient.

Copyright © 2002 by Taylor & Francis

(a)

(b)

Figure 4 a. Schematic layout of the VUV absorption spectrometer. H, hydrogen supply;

HV, high voltage; L, longitudinal stabilized hydrogen discharge W 1,2: LiF windows; C,
vacuum chamber; CF, cold finger; CS, crystal sample; CR, cryostat; MO, VUV monochro-
mator; S, slits; TM, turbo molecular pump; DE, detection electronics; SB, photomultiplier
or secondary electron multiplier. b. X-UV and VUV absorption spectrometer at the Na-
tional Hellenic Research Foundation (Athens, Greece). Crystal samples could be cooled
down to 10K with a closed circle helium cryostat. Further cooling to 4.2 K can be achieved
with a specially constructed liquid helium cryostat.

Copyright © 2002 by Taylor & Francis

 The solution of Eq. (2) is
x
I(λ, x) ⫽ I(λ, x ⫽ 0)e ∫ 0 [a(λ,x)⫺β(λ,x)] dx (3)

Taking into consideration that the gain coefficient a(λ, x) is proportional

to the number density of the excited atoms N 2 (x) and thus proportional to the
intensity of light
a(λ, x) ⫽ σ(λ)N 2 (λ, x) N 2 (λ, x) ⫽ cI(λ, x) (4)
Equation (3) then becomes

冦冮 [σ(λ)cI(λ, x) ⫺ β(λ, x)] dx冧

x
I(λ, x) ⫽ I(λ, x ⫽ 0) exp (5)
0

This equation can be solved using successive iterations. As zero order ap-
proximation we consider the case where there is no secondary photon generation
and the absorption coefficient is independent on the position of the interacting
atoms inside the crystal. In this case the solution of the equation takes its usual
form (Lambert–Beer law)

冤冮 ⫺ β(λ, x)] dx冥

x
I (0) (λ, x) ⫽ I(λ, x ⫽ 0) exp (6)
0

Substituting Eq. (6) into Eq (3) we get I (1) (λ, x) in first-order approximation.

冦冮 [σ(λ)cI(λ, x′ ⫽ 0) e 冧
x
I (1) (λ, x) ⫽ I(λ, x ⫽ 0) exp ⫺β(λ)x′
⫺ β(λ)] dx (7)
0

Integrating the above equation and setting

冤
F (1) (λ, x) ⫽ exp cσ(λ)
I(λ, x ⫽ 0)
β(λ)
(1 ⫺ e⫺β(λ)x ) 冥 (8)

Equation (7) becomes

I (1) (λ, x) ⫽ I(λ, x ⫽ 0) F (1) (λ, x) e⫺β(λ)x (9)
The F (1)(λ, x) term that describes deviation of the absorption processes from the
law of Lambert and Beer is a function of wavelength and the crystal thickness.
It describes photon emission through spontaneous emission within the crystal
volume in first-order approximation. By measuring the photon intensity emitted
from the crystal I (1) (l, x) as a function of the crystal thickness, the F (1) (λ, x), can
be determined.
In the case of ‘‘optically thin’’ samples, βx ⬍⬍ l, Eq. (9) takes the form
I (1) (λ, x) ⬇ I(λ, x ⫽ 0) exp(ax) ⬇ I(λ, x ⫽ 0)(1 ⫹ ax) (10)

Copyright © 2002 by Taylor & Francis

 Under these experimental conditions, the emitted photon intensity from the
crystal is taking place through spontaneous emission processes and varies linearly
with crystal thickness. Similarly when βx ⬎⬎ 1, ‘‘optically thick sample’’ Eq.
(9) takes the form

冤
I (1) (λ, x) ⬇ I(λ, x ⫽ 0) e[α(λ)/β(λ)] ⬇ I(λ, x ⫽ 0) 1 ⫹
a(λ)
β(λ)
⫹O
冢 冣冥
a(λ)
β(λ)
(11)

In this case the emitted intensity from the crystal is independent on its thickness.
For cases between Eqs. (10) and (11), the emission is reduced by self-absorption
and care has to be taken to ensure that all the terms in the transfer equation are
properly allowed for individual cases. When a single RE ion absorbs one VUV
photon, the 4f n⫺15d electronic configuration is populated and competition starts
between radiative and nonradiative transitions. The excited ion decays to the
different levels of the 4f n⫺15d electronic configuration and subsequently relaxes
to the ground level of the 4f n electronic configuration with VUV or UV photon
emission. It is therefore expected that the F (1) (λ, x) factor will be equal to unity
for the spectral region where there is no photon emission. In this case, absorption
processes inside the crystal sample fall within the validity of the Lambert and
Beer’s law [Eq. (9)]. For example, in the case of the LiYF 4:Nd 3⫹ crystal, sponta-
neous emission modulates the value of the absorption coefficient only in the spec-
tral range from 180 to 182 nm, since the crystal emits strongly in this spectral
range when it is excited with VUV laser light at 157 nm [7].

III. OPTICAL AND ELECTRONIC PROPERTIES OF WIDE

BAND GAP FLUORIDE DIELECTRIC CRYSTALS
DOPED WITH TRIVALENT RE IONS
A. Electronic Properties of Wide Band Gap Materials
The optical properties of the nondoped wide band gap dielectric crystals in the
VUV are mainly due to transitions from electronic states between the valance
and the conduction band. The high value of the absorption coefficient, for photon
energies higher than the crystal’s band gap, is due to transitions involving delocal-
ized electronic states in the conduction band, and arise from the crystal symmetry.
The crystals’ energy bands are formed from the energy levels of atoms when
they are brought close together. On the other hand, local lattice imperfections,
vacancies and other point defects, and dislocations bring about the formation of
allowed states connected with the perturbation area. The electronic wavefunctions
of such states are nonzero in approximately the same area where the perturbation
exists. In other words, the electron is localized in the perturbation area. The
smaller the perturbation energy, the greater the localization area. In this case,

Copyright © 2002 by Taylor & Francis

transitions are taking place between the valance band and the localized states in
the conduction band. The width of the band gap is proportional to the value of
the exchange integral, the origin of which is the fact that the electrons may,
with some probability, be located near any atom. The exchange integral can be
calculated with perturbation theory and accepting electron states of an isolated
atom as zero approximation for the solution of the Schrödinger equation for the
electron in a periodic crystal field U(r).

Ĥψ(r) ⫽ Eψ(r)
(12)
h2
Ĥ ⫽ ⫺ ∆ ⫹ U(r); U(r) ⫽ U(r ⫹ n)
2m

denoting the Hamiltonian of an isolated atom by

h2
Ĥa ⫽ ⫺ ∆ ⫹ Va(r)
2m (13)
Ĥa ψa(r) ⫽ Eaψa(r)

where V a (r) is the potential energy of the electron in an isolated atom, E a is the
a energy level, and ψ a (r) is the wavefunction corresponding to E a. Solution of
the equation for the atom is supposed to be known. The zero-order electron
wavefunction in the crystal ψ (0)(r) should be the sum of the atomic wavefunctions
at the point r ⫺ m that satisfy the translational condition at the point r ⫾ m, and
n is the coordinate of an ion.

ψ(0) (r) ⫽ 冱 exp(ikm)ψ (r ⫺ m)

m
a (14)

Zero approximation of the wavefunction ψ(0) (r) enables the first approximation
for the energy of the electron in the crystal field E (1) to be calculated

E (1)
⫽
冤 冱 exp(ikp)S ⫹ C ⫹ 冱 exp(ikp)A(p)冥
N Ea
p
p
p
(15)
N ∑ exp(ikp)Sp

where Sp is the overlapping integral that depends not on the coordinates of two
ions, but also on the distance p between them

∫ψa (r ′)ψa (r′ ⫺ p) ⫽ Sp (16)

Copyright © 2002 by Taylor & Francis

and A(p) is the exchange energy

A( p) ⫽ ∫ψ *a (r ′) 冤冱V (r′ ⫺ n) ⫹ W(r ′)冥 ψ (r ′ ⫺ p)dτ ′

n′≠p
a a (17)

where W(r ′) is the periodic self-consistent lattice field

W(r) ⫽ U(r) ⫺ 冱V (r ⫺ n)
n
a (18)

A( p) is made up of the wave functions of two atoms separated by the distance

p, with the physical meaning that the two atoms at distance p from each other
may exchange electrons. The exchange proceeds through the field of all other
atoms and the periodic self-consistent part of the lattice field W(r). Due to the
exponential form of the atomic wavefunctions, the exchange takes place mainly
between neighboring atoms. Electron exchange between any two atoms of the
crystal takes place by way of chain of neighbor exchanges. Therefore, electrons
are not localized near individual atoms but move freely through the crystal jump-
ing from one atom to another by the exchange process. From Eq. (15), the energy
of the atomic electrons in a crystal field is revealed. When the interaction of the
electrons with the ions and with the other electrons is taken into account, the
energy levels are lowered by the amount C and split into a band of definite width.
Because S p ⬇ 0 when p ≠ 0, we may write

E (1) ⫽ Ea ⫹ C ⫹ 冱 exp(ikp)A
p
p (19)

For a simple cubic lattice like the LiF crystal, every atom has six nearest neigh-
bors with the coordinates α[(1,0,0), (⫺1,0,0), (0,1,0), (0,⫺1,0), (0,0,1), (0,0,
⫺1)]. Assuming exchange energy to be isotropic and neglecting electron ex-
change between distant atoms we obtain

E (1) ⫽ Ea ⫹ C ⫹ 2A(cos k x a ⫹ cos kya ⫹ cos k z a) (20)

The energy depends quasicontinuously on the wave vector k and changes from
E min to E max
E min⫽ Ea ⫹ C ⫹ 6A
(21)
E max⫽ Ea ⫹ C ⫺ 6A
Thus, as a result of the atomic interaction, the energy level of an isolated
atom in a cubic crystal lattice drops by the amount C and splits into a band 12
A wide. The energy band gap depends on the exchange energy A. But the ex-
change energy itself is dependent on the area of the overlapping of the wave
functions: the more the atomic wave functions overlap, the greater will the ex-

Copyright © 2002 by Taylor & Francis

change energy be. It follows from here that energy levels corresponding to inner
shell electrons do not split as intensely as those of the outer shells, since the
inner shell electrons are localized in smaller areas of space. As the energy in-
creases, bands become wider and the gaps narrower. The higher a level, the lower
it drops and the wider it spreads. This result is basic for explaining the dependence
of the band gap width on the atomic number of the elements belonging to the
same group. For example the width of the forbidden band between the valance
and the conduction band for the metals Li, Na, K, Sc, and the halogens F, Cl,
Br I should decrease in the same order.
Following the above arguments it is expected that the LiF crystal will have
the widest band gap, among all the dielectric crystals, a fact that is verified from
experiments. Optical transitions in the VUV might take place between the valance
and the core band of the metal cations, which are formed of electron orbitals
lying lower in energy than the ones forming the valance band (2p F⫺ ). In this
case an electron is excited from the core band to the conduction band. The hole
created in the core band relaxes very fast to the band edge and is subsequently
annihilated by electron capture from the filled valance band.
The conservation of energy in this case requires emission of a VUV or UV
photon. The process is known as cross-relaxation [58] and it has been observed
in different wide band gap dielectric crystals [59]. Band-to-band excitonic transi-
tions in the VUV have been observed in various wide band gap materials such
as LiCaAlF 6 [60] and LiYF 4 [61] crystals doped with Nd 3⫹ and Ho3⫹ ions, respec-
tively. In these cases, the peaks around 114, 118, and 112, 116, 119 nm might
correspond to excitonic transitions between the valence and the conduction band
of the host lattice. The exitonic transitions imply a large value of the index of
refraction of fluoride crystals in the VUV, in agreement with previous measure-
ments [62–70].
Considering that the cut-off wavelength of the LiF window (which is placed
in front of the hydrogen lamp and the solar blind photomultiplier) is at 110 nm
(for commercial LiF windows 5 mm thick) (Fig 2.), the absorption spectrum
between 113 and 120 nm indicates only the energy of the corresponding transi-
tions. The edge of the conduction band is extended over a relatively wide spectral
range (113–120 nm). This situation reflects that the electron wavefunctions of
the fluorine and the metal ligands are extended over a large range as is expected
from the strong electrostatic nature of the corresponding bonds. The main contri-
bution at the value of the absorption coefficient in the spectral range from 110
to 120 nm is from the absorption of light within the conduction band of the
LiCaAlF 6 crystal. The absorption within the conduction band of the LiCaAlF 6
crystals is at least one order of magnitude higher than the absorption within the
4f 2 5d electronic configuration. As a consequence of the large value of the absorp-
tion coefficient in this spectral region, it falls out of the dynamic range of the
absorption spectrometer and hence deviates significantly from its mean value.
Optical transitions in the VUV are likewise taking place between the valance

Copyright © 2002 by Taylor & Francis

and charge transfer bands situated inside the band gap of the host crystal. They
are formed by strong electrostatic interactions between the d electron of the triva-
lent RE ions and the anions of the host crystal [71–74].

IV. VUV ABSORPTION SPECTROSCOPY OF TRIVALENT

RE IONS IN WIDE BAND GAP FLUORIDE CRYSTALS

When a particular atom or ion is surrounded by a symmetrical distribution of

atoms or ions (ligands), many new properties emerge that were not present in
the free atom case. There are two models for the treatment of such systems. In
the first model the central atom is assumed to be surrounded by a distribution of
charges that produces an electrostatic field at the position of the central atom. In
this model (crystal field) there is no sharing of electronic charge between the
central atom and the ligands. The second model treats the central atom and its
ligands as if they were a single molecule (molecular–orbital method ). In this
case electrons are permitted to overlap. In most cases this model is treated by
the methods of molecular–orbital theory. In both cases symmetry considerations
are of great importance. The symmetry of the optical site determines the degener-
acy of the electronic states and the optical properties of the materials. In the case
of wide-band-gap dielectric crystals doped with trivalent RE ions, the crystal
field model is usually applied since it is simpler, and the optical properties of
the RE ions are determined by the influence of the crystal field on the Hamiltonian
of the free ion.
In order to assign the dipole-allowed transitions between the Stark compo-
nents of the 4f n⫺15d electronic configuration and the levels of the 4f n electronic
configuration, the exact solution of the secular equation of the 4f n⫺1 5d electronic
configuration in the presence of the crystal field should be known [75,76]. How-
ever up to now, there have been no detailed theoretical calculations of the energy
position of the levels of the 4f n⫺15d electronic configuration. Crystal field calcula-
tions are carried out only to a low order of perturbation theory where the strongest
perturbation is applied first, followed by the next strongest, etc. [77]. The exact
theoretical treatment of the energy level problem of the 4f n⫺15d electronic con-
figuration is difficult to address. Even for the simplest case of the Ce3⫹ ion, the
crystal field for a d electron in a cubic symmetry requires 10 parameters for its
characterization. In order to identify these from the experimental data, one must
resolve and identify at least 11 spectral lines within the configuration. The number
of lines within the configuration is large (20 for Pr 3⫹, 3106 for Tb3⫹ ions) and
as a consequence attempts to evaluate crystal-field parameters by fitting an experi-
ment are difficult.
In principle, if one knows the radial wave functions of the ion in question,
then the crystal field perturbation can be calculated and the energy levels can
be compared with the experimental values. However, it is tedious to construct

Copyright © 2002 by Taylor & Francis

wavefunctions of the electrons in the 4f n⫺15d mixed configurations, and only
qualitative interpretation of the absorption spectra has been available up to now.
The early attempt of Dieke [29] to deduce the extent of the 4f n⫺15d electronic
configuration for each of the RE ions was made using experimental data and
some observed regularities in the lanthanide spectra. The effect of the crystal
field on the 4f n⫺1 5d electronic configuration is to lower its energy in comparison
to the free ion case (Table 2).
Various empirical methods have been developed [77] to interpret the VUV
absorption spectra of the 4f n⫺15d electronic configuration. These are based on
the following principle. The energy difference ∆ between two configurations dif-
fering by a single electron in the 4f configuration (e.g., 4f n and 4f n⫺15d) is con-
stant for all the lanthanides. This allows the evaluation of the energies of different
configurations for free ions with different ionization. Brewer [77] used this to
estimate the energy of the lowest energy level of a given configuration, beginning
with measured energies for neutral rare earths. For the 5d electronic configuration
of the RE ions, the crystal field perturbation is stronger than the electron–electron
interaction, due to the fact that the d electron is extended over a large distance
and been far apart from the nucleus of the ion. In this case the effect of the crystal
field is calculated first or simultaneously with the imposition of the Coulomb
interaction followed by the spin–orbit interaction. For example for the D 4 tetrago-
nal symmetry, the energy levels are the following: Γ (4), Γ (5), Γ (3), and Γ (1).
When the spin and the spatial parts of the wave functions are coupled within
the framework of the symmetry group of the RE ion, the number of the energy
levels is described by the decomposition of the product representation of Γ (6) (spin
part) with Γ (4), Γ (5), Γ (3), and Γ (1). The new energy levels now are the following:
Γ (7) (2B 2 ), Γ (6) (2E), Γ (7) (2E), Γ (7) (2B 1 ), Γ (6) (2A 1 ). In addition, when only one elec-
tron is present in the 5d orbital, as concerns the symmetry of the states produced
by the combined action of the crystal field and the spin–orbit interaction, it
makes no difference in which order the two perturbations are applied. The symme-
try group of the atom is no longer the three-dimensional rotation group because
all directions in space are no longer equivalent. Instead, the symmetry is that of a
point group containing a finite number of elements and is a subgroup of the three-
dimensional rotation group. An irreducible representation of the latter becomes
reducible with respect to the lower symmetry point group. The spherical harmon-
ics are not any longer energy eigenstates and are transformed under the irreduc-
ible representations of the symmetry operators of the three-dimensional rotation
group to its linear combinations. The effect on the states of the free atom is to
remove degeneracy. The first-order energies are sensitive to the magnitude of the
interactions and hence to the order in which they are applied.
In the case where the Coulomb repulsion of the electrons or the spin–orbit
interaction is the strongest perturbation, as in the case of the 4f n or the 4f n⫺1
electronic configuration [78], which are well shielded by the 6p and the 6s elec-

Copyright © 2002 by Taylor & Francis

Table 2 Energy Position of the Edge of Levels of 4f n⫺15d Electronic Configuration of Trivalent RE Ions

Free RE 3⫹
ions ⫻ 10 3 YF 3 ⫻ LuF 3 ⫻ LiYF 4 ⫻ LaF 3 ⫻ BaY 2F 8 ⫻ CaF 2 ⫻ KY 3F 10 ⫻
RE 3⫹ ion cm⫺1 10 3 cm⫺1 10 3 cm⫺1 10 3 cm⫺1 10 3 cm⫺1 10 3 cm⫺1 10 3 cm⫺1 10 3 cm⫺1

Nd 3⫹ 70.1 58.4 57.1 54.6 60.0 53.3 53.8

Er 3⫹ 75.4 64.3 63.8 62.5 64.7 61.9 63.3
Tm 3⫹ 74.3 63.4 63.0 61.5 64.4 63.0
Ce 3⫹ 49.7 34.4 38.8 34.0 31.9
Ho 3⫹ 58.9 63.3 63.3 63.1 64.1
Values are given as measured from the ground level of the 4f n electronic configuration in different crystal hosts.

Copyright © 2002 by Taylor & Francis

tronic configurations, the transformation properties of the electronic wavefunc-
tions are well described by the three-dimensional rotation group O⫹ (3) and the
spherical harmonics are basic functions for the representations D(1)(α, β, γ) of
the O⫹ (3). The subsequent application of the crystal field will further split the
energy levels.
Vacuum absorption and emission spectra of the trivalent RE ions in wide
band gap fluoride crystals are due to the interconfigurational 4f n⫺15d ↔ 4f n transi-
tions of the RE ions. They have been interpreted by fitting the crystal field split-
ting of the d electronic configuration to the energy gaps in the ground multiplets
of the 4f n⫺1 core [79]. The high complexity of the energy level systems of the
trivalent RE ions in the 4f n⫺15d electronic configuration makes a detailed interpre-
tation of the observed spectra still impossible. This is mainly due to the fact that
the wave functions of the two electronic configurations are transformed differ-
ently under the symmetry operations and the classification of spectral terms is
rather difficult, since the selection rules for the angular momentum are no longer
valid.
A theoretical interpretion of the atomic spectra of the 4f n⫺15d electronic
configuration was based on the finding, that the interaction energy of the crystal
field with the d electron exceeds the Coulomb interaction energy between the d
and the 4f n⫺1 electrons [80]. Energy states of the d electron are characterized by
irreducible representations of the corresponding crystal symmetry group Γ (2). For
the 4f n⫺1 electronic configuration, L–S coupling is predominant and further split-
ting of energy levels is taking place (Table 3). The energy levels are characterized
by the Γ (1) index of the irreducible representation of the three-dimensional rotation
group.
For this method the electronic wave functions can be written in the form
| SLJΓ (1), Γ (2), αΓµ⬎, where µ is the number of the basis function of the irreduc-
ible representation Γ, and the index α identifies the representation Γ that appears
several times in the direct product representation Γ (1) ⫻ Γ (2).
A qualitative interpretation of the 4f n⫺15d ↔ 4f n interconfigurational transi-
tions of trivalent RE ions in wide band gap dielectric crystals was made by
Szczurek and Schlesinger [81]. They constructed spectral terms first (the selection
rules for angular momentum are valid in this case), and then allowed for crystal
field splitting. Because the electronic wave functions of the 4f n⫺1 and 5d elec-
tronic configurations have different transformation properties, this approximation
is expected to be valid for the situation in which a more or less uniform charge
distribution of the ligands takes place within the unit crystal cell. This is possible
provided that the radius of the d orbital is smaller than lattice constants. In this
case, the electric field around the origin for a given distribution of ligands is more
or less spherical. This situation partially restores the three-dimensional rotation
symmetry of the 5d electronic configuration. As a first approximation, the 5d and
4f n⫺1 electronic configurations might well have the same transformation proper-
ties under the three-dimensional rotational group.

Copyright © 2002 by Taylor & Francis

Table 3 Crystal Field Splitting of the Main Levels of the 4f 2 Electronic
Configuration in the Presence of Crystal Field with Octahedral (O) and Tetragonal (D 4)
Symmetry
Configuration Terms Levels O D4

f 2 1
S 1
S0 1
Γ1 1
Γ1
3
P 3
P2 3
Γ 3 ⫹ 3Γ 5 1
Γ 1 ⫹ 1Γ 3 ⫹ 1 Γ 4 ⫹ 1 Γ 5
3
P1 3
Γ4 1
Γ 2 ⫹ 1Γ 5
3
P0 3
Γ1 3
Γ1
1
I 1
I6 1
Γ 1 ⫹ 1 Γ 2 ⫹ 1 Γ 3 ⫹ 1Γ 4 ⫹ 21Γ 1 ⫹ 1Γ 2 ⫹ 21Γ 3 ⫹
2 1Γ 5 21Γ 4 ⫹ 31Γ 5
1
D 1
D2 1
Γ 3 ⫹ 1Γ 5 1
Γ 1 ⫹ 1Γ 3 ⫹ 1 Γ 4 ⫹ 1 Γ 5
1
G 1
G4 1
Γ 1 ⫹ 1 Γ 3 ⫹ 1 Γ 4 ⫹ 1Γ 5 21Γ 1 ⫹ 1Γ 2 ⫹ 1Γ 3 ⫹ 1Γ 4
⫹ 21Γ 5
3
F 3
F4 3
Γ 1 ⫹ 3 Γ 3 ⫹ 3 Γ 4 ⫹ 3Γ 5 2 Γ 1 ⫹ 3 Γ 2 ⫹ 3 Γ 3 ⫹ 3Γ 4
3

⫹ 23Γ 5
3
F3 3
Γ 2 ⫹ 3Γ 4 ⫹ 3Γ 5 3
Γ 2 ⫹ 3Γ 3 ⫹ 3Γ 4 ⫹ 23Γ 5
3
F2 3
Γ 2 ⫹ 3Γ 5 3
Γ 1 ⫹ 3Γ 3 ⫹ 3 Γ 4 ⫹ 3 Γ 5
3
H 3
H6 3
Γ 1 ⫹ 3 Γ 2 ⫹ 3 Γ 3 ⫹ 3Γ 4 ⫹ 23Γ 1 ⫹ 3Γ 2 ⫹ 23Γ 3 ⫹
2 3Γ 5 23Γ 4 ⫹ 33Γ 5
3
H5 3
Γ 3 ⫹ 23Γ 4 ⫹ 3Γ 5 3
Γ 1 ⫹ 23Γ 2 ⫹ 3Γ 3 ⫹
3
Γ 4 ⫹ 33Γ 5
3
H4 3
Γ 1 ⫹ 3 Γ 3 ⫹ 3 Γ 4 ⫹ 3Γ 5 2 Γ 1 ⫹ 3 Γ 2 ⫹ 3Γ 3 ⫹
3

3
Γ 4 ⫹ 23Γ 5

A uniform charge distribution could be the result of charge compensation

from the F⫺ ligands, or of the localized character of the d orbitals. For example,
for a hydrogen-like atom, the expectation value of the radius of the electronic
clound 〈r〉 is given by the formula
α0
〈r〉 ⫽ [3n2 ⫺ l(l ⫹ 1)] (22)
2Z
where n is the principal quantum number equal to 5, α 0 is the Bohr radius of the
atom, α 0 ⫽ 0.053 nm, Z is the charge of the nucleus, and l ⫽ 2. For Nd 3⫹, Tb3⫹,
and Ho3⫹ ions, the calculated values of the ionic radii are 〈r〉 ⬃ 0.028, 0.027,
and 0.030 nm. The experimentally measured values for the three trivalent ions
in CaF 2 are 0.11, 0.10, and 0.97 nm. Since the ionic radii of the F⫺ ions are 0.136
nm, and taking into consideration that the distance between two neighboring
atoms is of the order of ⬃0.5 nm (as in LiYF 4 crystal), the spherical symmetry
is expected to be restored partially by charge compensation from F⫺ ions and the
localized character of the d orbital. Using this approximation, the number of
experimentally observed transitions is in excellent agreement with the number

Copyright © 2002 by Taylor & Francis

of theoretically expected 4f n → 4f n⫺1 5d dipole-allowed transitions for Nd 3⫹ and
Tb3⫹ ions [82,83]. (See Table 4.)

A. Cerium (Ce3ⴙ, 4f 1 )
The Ce3⫹ ion has only one electron in the d electronic configuration and therefore
the spectroscopic assignment of the energy levels is simple. The free Ce3⫹ ion
in the 4f electronic configuration has two energy levels— 2F 5/2, 2F 7/2,—which are
separated by 2253 cm⫺1. In the d electronic configuration it has also two levels—
2
D 3/2, 2 D 5/2 —which are separated by 49,733 and 52,226 cm⫺1 from the ground
state, respectively. The crystal field splitting of the energy levels of the 5d elec-
tronic configuration depends on the site symmetry of the Ce3⫹ ions. The fivefold
degenerate d level split into two levels with degeneracies 2 and 3. The d orbitals
(z 2, dx2 ⫺ y 2, d yz, d zx, and d xy ) are the basis functions for the D(2) representations
of O⫹ (3) group and the reduction of the D(2) into irreducible representations of the
cubic group is E(Γ (3) ) ⫹ T 2 (Γ (5) ). Since E and T 2 are two- and three-dimensional
respectively, the energy levels associated with E and T 2 are doubly and triply
degenerate. When the site symmetry of the Ce3⫹ ion is further reduced to a tetrag-
onal D 4 one, degeneracy of the energy levels is removed since the irreducible
representations of the cubic group are reducible representations of the D 4 group.
The energy levels are split further and the degeneracy is removed: E (Γ (3) ) ⫽ 2A 1
(Γ (1) ) ⫹ 2B 1 (Γ (3) ), T2 (Γ (5) ) ⫽ 2A 1 (Γ (1) ) ⫹ 2B 1 (Γ (3) ). The absorption spectra of the
Ce3⫹ ions in dielectric crystal hosts are attributed to the dipole-allowed transitions
between the 2F 5/2 ground state of the 4f electronic configuration and the energy
levels of the 5d electronic configuration observed early mainly due to experimen-
tal reasons, since the Ce3⫹ ions absorb and emit in the UV [84–99]. Six strong
interconfigurational dipole-allowed transitions have been observed [92,96]. Four
transitions have been assigned to the interconfigurational 4f → 5d electronic tran-
sitions of the Ce3⫹ ions of tetragonal site symmetry, and two to clusters formed
by two or more Ce3⫹ ions. One more transition of low intensity corresponds to
the e g level, which is doubly degenerate [97]. The remaining e g level indicates
a strong localized vibrational structure [81]. In this case the electronic wavefunc-
tions of the Ce3⫹ ions do not overlap with the wavefunctions of the ions of the
host lattice [99] and the strongest phonon lines are those corresponding to local-
ized phonon frequencies.

B. Praseodymium (Pr 3ⴙ, 4f 2 )

Energy levels of the free Pr 3⫹ ion in various electronic configurations such as
4f 2, 4f5d, 4f6s, and 4f6p have been analyzed by Sugar [100]. A comparison of
all the third and the fourth spectra of the lanthanide ions show a similarity

Copyright © 2002 by Taylor & Francis

throughout the series [101]. Energy levels of the 4f 2 electronic configurations
have been identified by comparing spectra in the vapor phase and in LaCl 3 :Pr 3⫹
crystals [102]. The lines arising from the 4f 2 → 4f5d transitions were identified
in a series of exposures of sparks made with successively increasing current. The
energy levels were assigned by comparing the level structure with that of the
isoelectronic Ce2⫹ ion and from relative line intensities [103]. The free Pr 3⫹ ion
in the 4f5d electronic configuration has 18 energy levels, the lowest is the 1G 4
at 61,171 cm⫺1. The 3G 3,4,5, 3 H 4,5 levels are excited through dipole-allowed transi-
tions from the 3 H 4 ground state of the 4f 2 electronic configuration. The absorption
spectra of the 4f 2 and the 4f5d electronic configurations of the Pr 3⫹ ions were
studied previously in a number of crystal lattices [8,104–110]. If the Pr 3⫹ ion is
located at a C 4v symmetry center, the 3G 3,4,5 and the 3 H 4,5 energy levels are split
for a total number of 35 Stark levels [81].
The number of absorption lines in CaF 2:Pr 3⫹ crystals is about half that, but
some lines may be weak or superimposed on others. If only the strongest lines
are considered, an additional selection rule ∆J ⫽ ∆L is applied in the C4v symme-
try, the number of Stark levels is limited to 12, which is close to what was experi-
mentally observed [81]. In the case of a high concentration of the Pr 3⫹ ions
(higher than 0.01 at.%), the appearance of additional lines in the absorption spec-
trum is attributed to Pr 3⫹ clusters and sites of additional symmetries. A typical
VUV absorption spectrum for the KY3F 10 :Pr 3⫹ crystal for 0.1 at.% concentration
of the Pr 3⫹ ions at room temperature is indicated in Figure 5.
The 4f5d electronic configuration in the free ion case is extended between
61,170 and 78,776 cm⫺1. The edge of the levels of the 4f5d electronic configura-
tion in various fluoride dielectric crystals is extended down to 44,440 cm⫺1 and
depends on the host material. For example, the edge of the levels of the 4f5d
electronic configuration in LiLuF 4 crystal hosts is at 45,100 cm⫺1 [8], in CaF 2 at
45,450 cm⫺1 [81], and in LaF 3 at 49,995 cm⫺1 [110]. The 4f 2 electronic configura-
tion of the Pr 3⫹ ions consists of 13 energy levels. 1S 0 is the highest one, which
lies very close to the 4f5d electronic configuration. The energy position of the
1
S 0 level is within the 4f5d electronic configuration for most of the cases. For
the LaF 3 crystal, the 1S 0 level is situated 3000 cm⫺1 below the edge of the band
[111], and hence it can be highly populated from the 4f5d electronic configura-
tion. A similar position of the 1S 0 energy level is valid for the YF 3 [24] and most
of the oxide crystal hosts [112–114].

C. Neodymium (Nd 3ⴙ, 4f 3 )

The absorption spectrum of the trivalent Nd 3⫹ ion in the VUV has been studied
previously in different dielectric crystal hosts [4,7,10,61,81,82,109,115,116]. The
energy position of the levels of the 4f 2 5d electronic configuration of the free Nd 3⫹
ion can be estimated from data available for the isoelectronic Pr 2⫹. According to

Copyright © 2002 by Taylor & Francis

Table 4 Number of Electric Dipole-Allowed Interconfigurational Transitions of Trivalent RE Ions Between the Ground State of the
4f n Electronic Configuration and the Excited Terms of the 4f n⫺15d Electronic Configuration

4f n⫺15d configuration
Dipole allowed transitions
4f n configuration ∆S ⫽ 0, ∆L ⫽ 0 ⫾ 1, ∆L ⫽ ∆J
Number Ground Number Lowest Terms in tetragonal
Ion n of levels state of levels level Excited terms crystal field Transitions
3⫹ 1
La 0 1 S0 0 — — — 0
Ce3⫹ 1 2 2
F 5/2 2 2
D 3/2 2
D 3/2 Γ6 ⫹ Γ7 2
Pr3⫹ 2 13 3
H4 20 1
G4 3
G3 Γ 2 ⫹ Γ 3 ⫹ Γ 4 ⫹ 2Γ 5 5
3
H4 2 Γ 1 ⫹ Γ 2 ⫹ Γ 3 ⫹ Γ 4 ⫹ 2Γ 5
Nd3⫹ 3 41 4
I 9/2 107 4
I 9/2 4
H 7/2 (2) 2Γ 6 ⫹ 2Γ 7 9
4
I 9/2 3 Γ 6 ⫹ 2Γ 7
4
K 11/2 3Γ 6 ⫹ 3Γ 7
Pm3⫹ 4 107 5
I4 386 5
L6 5
H 3 (3) Γ 2 ⫹ Γ 3 ⫹ Γ 4 ⫹ 2Γ 5 14
5
I 4 (2) 2Γ 1 ⫹ Γ 2 ⫹ Γ 3 ⫹ Γ 4 ⫹ 2Γ 5
5
K5 Γ 1 ⫹ 2Γ 2 ⫹ Γ 3 ⫹ Γ 4 ⫹ 3Γ 5
Sm3⫹ 5 198 6
H 5/2 977 6
L 11/2 6
G 3/2 (4) Γ6 ⫹ Γ7 20
6
H 5/2 (3) Γ 6 ⫹ 2Γ 7
6
I 7/2 (2) 2Γ 6 ⫹ 2Γ 7
Eu3⫹ 6 295 7
F0 1876 7
K4 7
D 1 (2) Γ2 ⫹ Γ5 7
7
F 1 (3) Γ2 ⫹ Γ5
7
G 1 (2) Γ2 ⫹ Γ5
Gd3⫹ 7 327 8
S 7/2 2725 8
H 3/2 8
P 9/2 3Γ 6 ⫹ 2Γ 7 3

Copyright © 2002 by Taylor & Francis

Tb3⫹ 8 295 7
F6 3106 9
D2 7
D 5 (5) Γ 1 ⫹ 2Γ 2 ⫹ Γ 3 ⫹ Γ 4 ⫹ 3Γ 5 30
7
F 6 (5) 2Γ 1 ⫹ Γ 2 ⫹ 2Γ 3 ⫹ 2Γ 4 ⫹ 3Γ 5
7
G 7 (5) Γ 1 ⫹ 2Γ 2 ⫹ 2Γ 3 ⫹ 2Γ 4 ⫹ 4Γ 5
Dy3⫹ 9 198 6
H 15/2 2725 8
G 15/2 6
G 13/2 (11) 3Γ 6 ⫹ 4Γ 7 58
6
H 15/2 (10) 4Γ 6 ⫹ 4Γ 7
6
I 17/2 (9) 5Γ 6 ⫹ 4Γ 7
Ho3⫹ 10 107 5
I8 1876 7
H8 5
H 7 (18) Γ 1 ⫹ 2Γ 2 ⫹ 2Γ 3 ⫹ 2Γ 4 ⫹ 4Γ 5 79
5
I 8 (14) 3Γ 1 ⫹ 2Γ 2 ⫹ 2Γ 3 ⫹ 2Γ 4 ⫹ 4Γ 5
5
K 9 (11) 2Γ 1 ⫹ 2Γ 2 ⫹ 2Γ 3 ⫹ 2Γ 4 ⫹ 4Γ 5
Er3⫹ 11 41 4
I 15/2 977 6
L 17/2 4
H 13/2 (18) 3Γ 6 ⫹ 4Γ 7 79
4
I 15/2 (14) 4Γ 6 ⫹ 4Γ 7
4
K 17/2 (11) 5Γ 6 ⫹ 3Γ 7
Tm3⫹ 12 13 3
H6 386 5
L6 3
G 5 (13) Γ 1 ⫹ 2Γ 2 ⫹ Γ 3 ⫹ Γ 4 ⫹ 3Γ 5 62
3
H 6 (11) 2Γ 1 ⫹ Γ 2 ⫹ 2Γ 3 ⫹ 2Γ 4 ⫹ 3Γ 5
3
I 7 (9) Γ 1 ⫹ 2Γ 2 ⫹ 2Γ 3 ⫹ 2Γ 4 ⫹ 4Γ 5
Yb3⫹ 13 2 2
F 7/2 107 4
H 9/2 2
D 5/2 (5) Γ 6 ⫹ 2Γ 7 25
2
F 7/2 (5) 2Γ 6 ⫹ 2Γ 7
2
G 9/2 (5) 3Γ 6 ⫹ 2Γ 7
Lu3⫹ 14 1 1
S0 20 3
P2 1
P1 Γ2 ⫹ Γ5 1

The number of dipole-allowed transitions is calculated taking into consideration partial restoration of the spherical symmetry for the 5d electronic configura-
tion. The theoretical predictions are in excellent agreement with experimental results.

Copyright © 2002 by Taylor & Francis

(a)

(b)

Figure 5 a. VUV absorption spectrum of the KY3F 10:Pr 3⫹ crystal. b. VUV absorption
spectrum of LaF 3:Pr 3⫹ crystal.

Copyright © 2002 by Taylor & Francis

Sugar [117], 107 possible energy levels exist in the free Pr 2⫹ ion of the 4f 2 5d
electronic configuration. Among them, only nine levels, 4f 2 ( 3F)5d [4 H 7/2, 4 H 9/2,
4
H 11/2] and 4f 2 ( 3 H)5d [4 H 7/2, 4 H 9/2, 4 H 11/2, 4 I 9/2, 4 I 11/2, 4 K 11/2], are excited from the
ground level 4 I 9/2 of the 4f 3 electronic configuration. In order to assign the dipole-
allowed transitions between the Stark components of the 4f 2 5d electronic con-
figuration and the levels of the 4f 3 electronic configuration, the exact solution of
the secular equation of the 4f 2 5d electronic configuration in the presence of the
crystal field should be known. However up to now, there have been no theoretical
calculations regarding the energy position of the levels of the 4f 2 5d electronic
configuration.
To find the total number of interconfigurational dipole-allowed transitions
of the Nd 3⫹ ions, spectral terms that arise from the 5d and the 4f orbital are
formed, followed by calculation of the effect of the crystal field on the system’s
Hamiltonian. The crystal field divides these nine levels into a total number of
40 Stark components for a C 4v or S 4 site symmetry. However if the selection
rules for dipole transitions ∆S ⫽ 0, ∆l ⫽ ⫹1, ⫺1, ∆L ⫽ 0, ⫹1, ⫺1, ∆J ⫽ 0,
⫹1, ⫺1 (J ⫽ 0 → J ⫽ 0 forbidden) and ∆J ⫽ ∆L are applied, only 11 transitions
remain after excitation from the ground level 4 I 9/2 and the 4f 2 ( 3F)5d [4 K 11/2] level
is the lowest. The 4f 2 ( 3 H)5d [4 H 7/2] level is the highest. In the case of Nd 3⫹ ion
in LiYF 4 crystal hosts (C64h space symmetry group, S 4 point symmetry group of
the Nd 3⫹ ions), 11 Stark components are populated through dipole-allowed transi-
tions from the ground level 4 I 9/2 of the 4f 3 electronic configuration, which is in
excellent agreement with the experiments [82].
Therefore, the charge compensation and localized character of the d orbitals
partially restore the spherical symmetry of angular momentum. In this case, the
main effect of a strong crystal field is to lower the energy levels of the 4f n⫺15d
electronic configuration relative to the free ion case. The amount of shifting is
directly correlated to the magnitude of the crystal field. The absorption spectrum
was slightly broadened at room temperature compared to lower temperature (77
K). The structure of the spectrum verifies the mixed character of the 4f 2 5d elec-
tronic configuration and partial restoration of the spherical symmetry. The absorp-
tion peak around 66,578 cm⫺1 probably has a weak additional contribution from
the highly excited levels of the Nd 3⫹ ion of the 4f 3 (2F) electronic configuration
[29]. In this experiment the band gap of the LiYF 4 crystal host was estimated to
be 86,900 cm⫺1 wide (for a 0.5 mm thick crystal). However, the value of the
band gap is estimated to be wider for a thinner sample. The absorption band from
115 to 118 nm has been observed for various host lattices. It was attributed to
transitions between the valence and the conduction band of the host lattice. In
condensed matter at room and low temperatures, dipole transitions between the
ground level of the Nd 3⫹ ion and the levels of the 4f 2 5d electronic configuration
usually have a broad band structure, due to the wide spatial distribution of the
electronic cloud. A similar response is expected for crystal hosts of different

Copyright © 2002 by Taylor & Francis

symmetry. For example, in the case of the rhombic PFYK:Nd crystal (space group
Pnam) [118], absorption spectrum showed that the 4f 2 5d configuration levels
split into nine main Stark components with maximum absorption at 56,212;
58,423; 59,180; 62,606; 65,789; 66,906; 70,061; 74,174; and 79,878 cm⫺1, re-
spectively [82]. The absorption spectrum of the LiCaAlF 6:Nd 3⫹ crystal in the
spectral range from 120 to 200 nm is likewise indicated in Figure 6. The peaks
were assigned to strong dipole-allowed transitions between the 4 I 9/2 ground level
of the 4f 3 electronic configuration and the Stark components of the levels of the
4f 2 5d electronic configuration.
When the crystal field of trigonal symmetry is applied, on the 4 K 11/2 level
of the Nd 3⫹ free ion, it splits it into three Stark components: 2(Γ (4) ⫹ Γ (5) ⫹
Γ (6) ), Γ (4), Γ (4). The 4 I 9/2 level likewise splits into two Stark components by the
trigonal crystal field: 2(Γ (4) ⫹ Γ (5) ⫹ Γ (6) ), Γ (4). The 4 H 7/2 splits into three Stark
components: the Γ (4), Γ (4), (Γ (4) ⫹ Γ (5) ⫹ Γ (6) ). Some of these components are
degenerate and the degeneracy is removed within the 4f 2 5d electronic configura-
tion. In the case of trigonal symmetry, the dipole-allowed transitions between
the ground state 4 I 9/2 of the 4f 3 electronic configuration and the Stark components

Figure 6 Absorption spectrum of LiCaAlF 6:Nd 3⫹ crystal in the spectral region from
120 to 250 nm.

Copyright © 2002 by Taylor & Francis

of the 4f 2 5d electronic configuration are 20, in agreement with experimental ob-
servations. The peak at 115 nm corresponds to excitonic transitions between the
valence and conduction band of the host lattice, suggesting a large value of the
index of refraction, which is in agreement with previous measurements of fluoride
crystals’ than films in the VUV.

D. Promethium (Pm3ⴙ, 4f 4)
Promethium is an artificial element that is highly radioactive. The 4f 4 electronic
configuration of Pm3⫹ ions has been investigated in LaCl 3 crystals [119,120].
There are 107 and 386 levels of the 4f 4 and the 4f 35d electronic configurations
respectively. In the free ion case, 5L 6 is the lowest level of the 4f 35d electronic
configuration. Six terms—the 5 H (3), 5 I, and 5 K—are excited through dipole-
allowed transitions from the ground state 5 I 4 of the 4f 4 electronic configuration.
Fourteen dipole-allowed transitions are expected in the crystal field with tetrago-
nal symmetry. Free-ion calculations put the edge of the 4f 35d electronic configu-
ration at 73,300 cm⫺1 and the 4f 34s electronic configuration at 111,000 cm⫺1 [2].

E. Samarium (Sm3ⴙ, 4f 5)
The energy position of the levels of 4f 4 5d electronic configuration of free Sm3⫹
ion can be estimated based on the data available for isoelectronic Nd ⫹ [120,121].
The free Sm3⫹ ion has 977 levels of 4f 45d electronic configuration. Among them,
only nine terms [four 6G, three 6 H, and two 6 I] are excited from the ground level
6
H 5/2 of the 4f 5 electronic configuration, giving 20 dipole-allowed transitions ex-
cited from the ground state. The lowest level of the 4f 45d electronic configuration
is the 6L 11/2 one at 73,700 cm⫺1 above the ground state. The 4f 5 electronic config-
uration of the free ion consists of 198 levels [81]. It has been analyzed in crystals
[122,123]. Early measurements of crystal field splitting of single and mixed elec-
tronic configurations of Sm ions in various dielectric crystal hosts were attributed
to the divalent Sm2⫹ ions [124–126]. The edge of the levels of 4f 45d electronic
configuration in LaF 3 was at 58,000 cm⫺1 and extends well beyond the cut-off
of the crystal host. The VUV absorption spectra of Sm ions in various dielectric
crystal hosts, CaF 2 [121,127] and LiYF 4 [128], were interpreted as 4f 5 → 4f 45d
interconfigurational transitions of the trivalent Sm3⫹ ions. Absorption spectra of
CaF 2:Sm3⫹ are unique in that some spectral lines arising from thermally excited
Stark levels of the 4f 5 electronic configuration were observed at low temperatures
[81]. The average spacing between zero phonon and vibronic lines is ⬃486 cm⫺1.
This frequency has been attributed to a local mode between the Sm3⫹ ion and its
nearest-neighbor fluorine ions [129].
As already mentioned, the effect of the crystal field on the 4f n⫺15d elec-
tronic configuration energy would be to lower its edge; various empirical methods

Copyright © 2002 by Taylor & Francis

have been established [77], based on the following principle: The energy differ-
ence ∆ between two configurations differing by a single electron in the 4f config-
uration (e.g., 4f n and 4f n⫺15d) is constant for all the lanthanides. Martin [130]
estimated the ∆ energy difference for neutral monovalent and divalent ions from
a few measured values. Sugar and Reader [131] applied this principle to estimate
trivalent ion differences using Martin’s results. In the free ion case, the energy
difference between the 4f 45d and the 4f 5 levels was estimated to be 76,000 ⫾
2000 cm⫺1 [77,130–131]. In the presence of the CaF 2 crystal field, a depression
of the upper configuration by about 18000 cm⫺1 towards the ground state is taking
place [92]. Hence the edge of the levels of the 4f 45d electronic configuration in
the presence of the CaF 2 crystal field is expected to be at 58,000 ⫾ 2000 cm⫺1,
in agreement with the experiments [121].

F. Europium (Eu 3ⴙ, 4f 6)

The free Eu3⫹ ion has 1878 levels [81] of 4f 55d electronic configuration. Among
them, seven terms [two 7G, three 7F, and two 7 D] are excited from the ground
level 7F 0 of the 4f 6 electronic configuration, giving seven dipole allowed transi-
tions excited from the ground state. The lowest level of the 4f 55d electronic
configuration is the 7 K 4, at 81,800 cm⫺1 above the ground state 7F 0 of the 4f 6
electronic configuration. The 4f 6 electronic configuration of the free ion consists
of 295 levels [81]. Comparison of the absorption spectrum of CaF 2: Eu3⫹ ion
with an energy level system of the free isoelectronic Sm2⫹ has been given by
Schlesinger and Szczurek [121]. The so-called system difference 4f 55d–4f 6 in
the free Eu3⫹ ions was estimated to be 85,500 ⫾ 1000 cm⫺1. In the presence of
the CaF 2 crystal field, the first level of the 4f 55d electronic configuration is ex-
pected to be at 67,500 ⫾ 1000 cm⫺1, which is in agreement with experimental
observations [121]. The vibrational structure has been analyzed from the ab-
sorption spectrum and the separation from zero-phonon lines. Their vibronics
were found to be ⬃452 cm⫺1 and 486 cm⫺1 [121]. The two values were attributed
to the formation of centers of different site symmetry and clustering at higher
concentration of the RE ions. Strong absorption of Eu3⫹ ions in LaF 3 crystals
between 200 and 300 nm has been attributed to interconfigurational transitions
from divalent ions [2]. The 5d electronic configuration of the Eu2⫹ ion is calcu-
lated to begin at 33,900 cm⫺1 and the 6s electronic configuration at 44,300 cm⫺1
[131].

G. Gadolinium (Gd3ⴙ, 4f 7)
Crosswhite and Kielkopf have studied the free ion spectrum and identified several
lines from the 4f 65d electronic configuration occurring between 104,000 and
122,000 cm⫺1 [132,133]. Despite the fact that Gd 3⫹ ions in the 4f 65d electronic

Copyright © 2002 by Taylor & Francis

configuration have 2725 levels [81], only three among them are important if
electric dipole-allowed transitions from the ground state are taken into considera-
tion, the 8P 5/2, 8P 7/2, and 8P 9/2. In free Gd 3⫹ ions these levels are expected to be sep-
arated from the lowest energy level of the 4f 65d electronic configuration, 8 H 3/2,
by about 8000, 9600, and 11,200 cm⫺1, respectively [132]. These three levels
are spread over the energy range of 3200 cm⫺1. Separation between the extreme
nonphonon lines in the absorption spectrum is about the same, namely 3600 cm⫺1
[134]. The lowest energy level of the 4f 65d electronic configuration 8 H 3/2, in CaF 2
is situated at 77,760 cm⫺1, in agreement with Loh’s estimation at 78,000 cm⫺1
[92]. In this work, the absorption lines at 77,890 and 79,710 cm⫺1 were assigned
to spin-forbidden transitions. The spectrum of Gd 3⫹ ions in CaF 2 is exceptional
in that it does not seem to exhibit the type of vibronic feature at about 490 cm⫺1
from the parent zero phonon line that appears in the 4f → 5d absorption spectra
of all 10 other triply ionized rare earth ions in CaF 2.

H. Terbium (Tb3ⴙ, 4f 8)
The absorption spectrum of the trivalent Tb3⫹ ion has been studied previously in
different crystal lattices [2,5,83,135]. The electric crystal field splits all the levels
of single and mixed configuration. The number and the spacing of the components
depend on the symmetry and intensity of the crystal field. Such splitting has been
observed previously for the Tb3⫹ ion in other dielectric crystal host materials as
well. For example, regarding the LiYF 4:Tb3⫹ crystal (site symmetry C 4h or S 4 ),
it was found that the electric crystal field splits the 4f 75d electronic configura-
tion into five main Stark components with maximum absorption at 46,700,
51,500, 55,000, 60,600, and 80,900 cm⫺1, respectively [135]. For the LaF3:Tb3⫹
crystal (C 2 or D 3h site symmetry), the electric crystal field likewise splits the
levels of the 4f 75d electronic configuration into seven main Stark components
with maximum absorption at 54,000, 55,500, 57,800, 58,800, 61,300, 65,800,
and 73,000 cm⫺1 respectively [2]. In the case of the free Tb3⫹ ion, Dieke and
Crosswhite indicated that the 4f 75d electronic configuration splits into two groups
of levels, and the energy gap between them extends from 55,700 to 65,000 cm⫺1
[29]. For the LaF 3 crystal, the edge of the levels of the 4f 7 5d configuration is at
50,000 cm⫺1 [19]; for the CaF 2 at 46,500 cm⫺1 [19]; and around 44,900 cm⫺1 for
the LiYF 4 crystal [18]. The edge of the levels of the 4f 75d electronic configuration
in the case of the free Tb3⫹ ion is at 54,900 cm⫺1 [23].
According to Szczurek and Schlesinger [81], the 4f 7 electronic configura-
tion of the Tb3⫹ ion is half-filled, and the lower energy level of the 4f 75d elec-
tronic configuration is the 8S 7/2 one, which has zero total angular momentum.
Therefore, in a pure L–S coupling the 5d electron does not interact with the 4f 7
shell. The interaction is taking place only through the crystal field. In this aspect
the Tb3⫹ ion is similar to the Ce3⫹ ion, although the last one in the 5d electronic

Copyright © 2002 by Taylor & Francis

configuration has an empty 4f shell. As a result, the low-energy part of the absorp-
tion spectrum of the Tb3⫹ ions in the 4f 75d electronic configuration should be
quite similar to that of Ce3⫹ ions. The absorption spectra of LiLuF 4: Tb3⫹ in the
140–220 nm spectral range consists of a band displaying similar structure to
the spectrum of Ce3⫹ ions in crystal field in the 180–320 nm spectral range. The
separation between the extreme lines in the spectra is about 20,600 cm⫺1 for
terbium and about 21,600 cm⫺1 for cerium. This experimental evidence supports
the argument that in both cases the lowest energy-level system in the 4f n⫺15d
electronic configuration is formed through the interaction between the 5d electron
and the crystal field. Following this argument and taking into consideration that
spherical symmetry is partially restored, the 4f 7(8S)5d electronic configuration
consists of two terms, the 9 D and the 7 D one, and only the 7 D 5 is populated
through the electric dipole transitions from the 4f 8 ( 7F 6 ) ground level of the 4f 8
electronic configuration. The crystal field of tetragonal symmetry should split the
7
D 5 level into eight Stark components ( 7 D 5 ⫽ Γ1 ⫹ 2Γ2 ⫹ Γ3 ⫹ Γ4 ⫹ 3 Γ5 ).
In order to find the relative energy position of the Stark components, it is
necessary to construct the basic functions for the Γ1– 5 irreducible representations,
and then to calculate the matrix elements of the electrostatic potential on the Tb3⫹
ion, for each one of the eight Stark components. In this approximation, the total
angular momentum, J, is the good quantum number (J–J coupling). However in
order to avoid the lengthy calculations, a different approach can be applied that
considers the orbital angular momentum of the d electron (L ⫽ 2) to be the good
quantum number. In this first-order approximation, the Γ4 irreducible representa-
tion has the lowest energy in a tetragonal site symmetry. The next irreducible
representation with higher energy to the Γ4, is the Γ5 one, followed by the Γ3 and
the Γ1. The Γ2 and the Γ5 Stark components are double and triple degenerate.
The degeneracy can be raised when the L–S coupling strength is strong enough
within the levels of the 4f 75d electronic configuration. The splitting of the levels
of the 4f 75d electronic configuration due to the L–S coupling is only few hundred
wavenumbers, and this value is considerably smaller than the crystal field split-
ting. In the case of the LiYF 4:Tb3⫹ ions, the eight Stark components of the 7 D 5
level were assigned to the levels of the 4f 75d electronic configuration at 47,000,
52,000, 54,400, 55,500, 55,900, 62,200, 64,300, and 67,500 cm⫺1 respectively
[135]. We therefore assign the triple degenerate Γ5 Stark component to the energy
levels at 54,400, 55,500, and 55,900 cm⫺1, and the double degenerate Γ2 Stark
component to the energy levels at 62,200 and 64,300 cm⫺1, respectively. The
remaining three Stark components Γ1, Γ3, and Γ4 were assigned to the energy
levels at 67,400, 52,000, and 47,000 cm⫺1, respectively. The next to the lowest
8
S level of the 4f 75d electronic configuration is the 4f 7 (6P)5d one. The exact
value of the separation between the 6S and the 6P levels of the 4f 7 electronic
configuration is not known exactly, but it might be assumed to be of the same
order of magnitude as in the case of the Gd 3⫹ ions, which is 32,000 cm⫺1 [24].

Copyright © 2002 by Taylor & Francis

Figure 7 VUV absorption spectrum of K 2YF 5:Tb3⫹.

The 4f 7 (6P)5d electronic configuration has four terms that are excited from
the F 6 ground level of the 4f 8 electronic configuration: the 7F 6, 7F 5, 7G 7 and the
7

7
G 6 ones. The crystal field splits all these levels to a total number of 26 Stark
components for a C 4v or an S 4 site symmetry. However, if the selection rules for
dipole transitions ∆S ⫽ 0, ∆L ⫽ ⫾1, ∆L ⫽ 0, ⫾1, ∆J ⫽ 0, ⫾1, (J ⫽ 0 → J ⫽
0 forbidden), and ∆J ⫽ ∆L applied, only 14 transitions remain after the excitation
from the ground level 7F 6, which occupies the spectral region from 120 to 240 nm.
A similar structure has been observed previously for the LiYF 4: Tb3⫹ crystal
in this spectral range [135]. The band gap of the LiLuF 4 crystal was estimated to
be 77,600 cm⫺1 wide. It is possible the levels of the 4f 75d electronic configuration
overlap with the levels of the 4f 76s electronic configuration below 120 nm [2].
The VUV spectrum of the Tb3⫹ ions in K 2YF 5 crystal hosts is indicated in Figure
7. The spectrum is similar to that of the LiYF 4:Tb3⫹ crystal [135].

I. Dysprosium (Dy3ⴙ, 4f 9 )
The 4f 9 electronic configuration of trivalent dysprosium in LaF 3 indicated a com-
plex structure. The edge of the levels of the 4f 85d electronic configuration of
free trivalent dysprosium ions is calculated to be at 66,300 cm⫺1 and at 99,100
cm⫺1 for the 6s electronic configuration. The absorption spectrum of Dy3⫹ ions

Copyright © 2002 by Taylor & Francis

in LaF 3 indicates a complex structure [136] and accurate assignment of states
has not been made above 24,000 cm⫺1. The VUV absorption spectrum of the
Dy3⫹ ions in CaF 2 begins at 175 nm and consists of many sharp zero-phonon
lines and elaborately structured phonon side bands. There are two peaks in the
VUV absorption spectrum of Dy3⫹ ions in CaF 2 situated at 490 and 100 cm⫺1
from their parent zero-phonon lines [137]. The peak at 490 cm⫺1 is attributed to
a localized vibrational mode of the eight nearest fluorine neighbors of the Dy3⫹
ions. The vibration mode at 490 cm⫺1 is present in CaF 2 crystals doped with
different RE ions such as Ce3⫹, Tb3⫹, and Ho3⫹ ions.

J. Holmium (Ho3ⴙ, 4f 10 )
The absorption spectrum of the trivalent Ho3⫹ ion was studied previously in a
number of crystal lattices [2,7,81,138,139]. The electric crystal field splits all the
levels of the single and the mixed configurations. Three Stark components of the
levels of the 4f 95d electronic configuration were observed previously in the case
of the BaY 2F 8 crystal host, with maximum absorption at 63,900 cm⫺1 (156.4 nm),
67,800 cm⫺1 (147.4 nm), and 71,800 cm⫺1 (139.2 nm). The edge of the levels of
the 4f 95d electronic configuration was found to be at 63,100 cm⫺1 (158.4 nm)
[139]. For the LaF 3 and the CaF 2 crystal hosts, the edges of the levels of the
4f 95d electronic configuration for the spin-allowed transitions were found to be
at 63,300 cm⫺1 (158.7 nm) and 64,100 cm⫺1 (156.0 nm), respectively [2,138].
For the free ion case, the edge of the levels of the 4f 95d electronic configuration
was found to be at 58,900 cm⫺1 (169.5 nm) [29]. In the case of the single LiYF 4
crystal and for 0.1 at. % concentration of the Ho3⫹ ions, the edge of the levels
of the 4f 95d electronic configuration for the spin-allowed transitions was found
to be at 63,300 cm⫺1 (158 nm) [174]. In the spectral range from 120 to 158 nm,
five main Stark components of the 4f 95d electronic configuration with maximum
absorption at 64,800 cm⫺1 (153.4 nm), 69,300 cm⫺1 (143.5 nm), 73,200 cm⫺1
(136.6 nm), 78,900 cm⫺1 (126.7 nm), and 84,100 cm⫺1 (118.8 nm) were observed.
In the spectral range from 158 to 170 nm, four main Stark components of
the 4f 95d electronic configuration with maximum absorption at 59,700 cm⫺1
(167.5 nm), 60,200 cm⫺1 (166 nm), 61,200 cm⫺1 (163 nm), and 61,900 cm⫺1 (162
nm) were likewise observed. They were assigned previously to spin-forbidden
transitions [128]. This assignment was justified because of the low value of the
absorption coefficient, in comparison to the value of the absorption coefficient
for wavelengths less than 158 nm, and on the other hand because of the long
lifetime (300 ns) of the corresponding transitions. The transitions between 58,800
cm⫺1 (170 nm) and 54,400 cm⫺1 (184 nm) were assigned to the 4f 10 → 4f 10
intraconfigurational dipole-forbidden transitions.
The local maximum around 116 nm [175] is attributed to the transition
between the valence and the conduction bands of the host lattice. From the ab-

Copyright © 2002 by Taylor & Francis

sorption spectrum it is estimated that the band gap of the LiYF 4 crystal host is
86,200 cm⫺1 (116 nm) wide. This value is in agreement with the value of the
band gap of 10.7 eV (115 nm) of the LiYF 4 crystal, estimated previously from
reflection spectra [3]. Indeed, the cut-off wavelength of the LiF window, which
is placed in front of the hydrogen lamp and the solar blind photomultiplier, is at
110 nm. The transmittance of the LiF window drops successively from its highest
value at 120 nm, to zero at 110 nm when it is recorded using a secondary electron
multiplier of open geometry without the LiYF 4:Ho3⫹ crystal in the optical path.
In this spectral region, there is only one local maximum at 112 nm. When the
transmittance is recorded with the the LiYF 4:Ho3⫹ crystal in the optical path, two
new additional local maxima appear in the absorption spectrum from 110 to 120
nm, one at 116 nm and the other at 118.8 nm. A local maximum around 116 nm
appears as well in the absorption spectrum of LiYF 4 crystals doped with Nd 3⫹
ions [3,82]. Hence the position of the local maximum at 116 nm should be inde-
pendent of the RE dopands in LiYF 4 crystal hosts and it is assigned to a transition
between the valance and the conduction band of the LiYF 4 crystal.
There is not a sharp transition between the valance and the conduction
band due to the delocalized nature of the conduction band. From the absorption
spectrum, the oscillation strength of the transition between the valance and the
conduction band increases as one shift in the wavelength scale from 122 to 116
nm. For wavelengths less than 122 nm, the levels of the 4f 95d electronic configu-
ration of the Ho3⫹ ions are mixed with the delocalized levels of the conduction
band of the crystal host, and the transmittance of the crystal decreases rapidly
in this spectral region. Hence, the absorption spectrum between 110 and 122 nm,
indicates only the position of the levels of 4f 95d electronic configuration inside
the conduction band. In this case the value of the absorption coefficient cannot
be calculated, because of the low value of the signal-to-noise ratio of the light
source, which approaches unity. The absorption peak at 122 nm corresponds to
the transition between the 5 I 8 ground level of the 4f 10 electronic configuration
and the edge of the conduction band. Therefore the position of the 5 I 8 ground
level of the Ho3⫹ ion is 4200 ⫾ 800 cm⫺1 above the top of the valance band of
the LiYF 4 crystal host, provided that the band gap of the LiYF 4 crystal is 86,200
cm⫺1 wide. The local maximum at 118.5 nm likewise corresponds either to the
transition between the ground 5 I 8 level and a Stark component of the 4f 95d elec-
tronic configuration inside the conduction band or to a transition from the ground
5
I 8 level to a localized level inside the conduction band.

K. Erbium (Er 3ⴙ, 4f 11 )

The edge of the levels of the 4f 105d electronic configuration of the Er 3⫹ ions was
calculated to be at 75,400 cm⫺1 [131]. The lowest level of the 4f 105d electronic

Copyright © 2002 by Taylor & Francis

configuration is the 6L 17/2 and 79 dipole-allowed transitions are excited from the
ground state 4 I 15/2 of the 4f 11 electronic configuration. The absorption spectrum
of the trivalent Er 3⫹ ion in the VUV has been studied previously in a number of
crystal lattices [1–3,9,81,140–141] and it was found to be rather simple, mainly
because very few energy levels of the 4f 105d electronic configuration are within
the band gap of the host material. Intraconfigurational VUV dipole-forbidden
4f 11 → 4f 11 transitions of Er 3⫹ ions were observed in various dielectric crystals
as well [142]. The edge of the levels of the 4f 105d electronic configuration in
KY 3F 10 host was found to be at 60,850 cm⫺1. Five absorption bands of the 4f 105d
electronic configuration have been observed in BaY 2F 8 hosts with the maximum
of the absorption at 63,700, 67,900, 70,500, 73,400, and 74,600 cm⫺1. The edge
of the levels of the 4f 105d electronic configuration in this crystal host was found
to be at 61,900 cm⫺1, 58,000 cm⫺1 in LaF 3 [2], and at 63,000 cm⫺1 in LiLuF 4
[9]. Five main Stark components of the 4f 105d electronic configuration have been
observed as well in LiLuF 4 hosts with the maximum of the absorption at 60,980,
69,198, 71,422, 74,343, and 76,916 cm⫺1. The presence of an additional weak
absorption band with maximum of absorption at 163 nm in LiLuF 4 host has been
observed as well by Devyatkova et al. [141] in KY 3F 10 matrices at 164.4 nm.

Figure 8 VUV absorption spectrum of K 2YF 5:Er 3⫹ crystals.

Copyright © 2002 by Taylor & Francis

The VUV absorption spectrum of the Er 3⫹ ions in K 2YF 5 crystal hosts is indicated
in Figure 8.

L. Thulium (Tm3ⴙ, 4f 12 )
The edge of the levels of the 4f 115d electronic configuration of the free Tm3⫹
ions was estimated to be at 74,300 cm⫺1, and that of the 4f 116s electronic configu-
ration at 101,300 cm⫺1 [29]. The relative simplicity of the spectrum in various
dielectric hosts arises because only some of the levels of the 4f 115d electronic
configuration lay below the edge of the conduction band. The absorption spec-
trum of Tm3⫹ ions in LiLuF 4 [135] indicates the presence of four bands with
maximum of absorption at 63,300, 69,000, 73,000 and 76,500 cm⫺1, respectively
and for the K 2YF 5 host the VUV absorption spectrum of the Tm3⫹ ions is indi-
cated in Figure 9. VUV absorption spectra of Tm3⫹ ions in different fluoride
dielectric hosts have been analyzed by Szczurek and Schlesinger [143–145]. The
main result of these investigations was that different possible transitions and

Figure 9 VUV absorption spectrum of the K 2YF 5:Tm3⫹ crystal.

Copyright © 2002 by Taylor & Francis

relaxation paths are dependent upon lattice temperature. The temperature-
dependent lifetimes can be expected to occur as a result of phonon-assisted radia-
tive transitions.

M. Ytterbium (Yb3ⴙ, 4f 13 )
The free Yb3⫹ in the 4f 12 5d electronic configuration has 107 energy levels.
Among these, some 20 levels have been measured from gaseous spectra [146]
including the lowest 4 H 9/2 level. This enables one to make some limited compari-
son of the observed spectra of Yb3⫹ in different crystal hosts with the free Yb3⫹
energy levels [147–149]. The crystal field of the C 4v symmetry splits the lowest
4
H 9/2 level into five Stark levels. The separation between the extreme of these
five lines is aboute 2100 cm⫺1. This value is about one order of magnitude smaller
than the corresponding 4f n⫺15d level splitting observed, for instance, in cerium,
praseodymium, or terbium. This suggests that the interaction between the 5d
electron and the crystal field is weak.

IV. LASER-INDUCED FLUORESCENCE OF RE IONS

IN WIDE BAND GAP FLUORIDE CRYSTALS AT 157 nm

When the RE ion is excited from its ground level of the 4f n electronic configura-
tion to a given level of the 4f n⫺15d electronic configuration, it populates the levels
of the 4f n electronic configuration from the edge and the Stark components of the
levels of the 4f n⫺15d electronic configuration. This is a rather common response
to photoexcitation of the trivalent RE ions in fluoride dielectric crystals, and it
has been observed previously for other dielectric crystals doped with RE ions
[9,82,150]. In this case competition starts between radiative and nonradiative
transitions. The nonradiative transitions are faster and usually the ion will decay
to the lowest of the 4f n⫺15d levels. The transitions arising from the edge and the
levels of the 4f n⫺15d electronic configuration and the laser induced fluorescence
(LIF) spectrum could be interpreted on the basis of four different processes:

1. The formation of optically active RE centers of different site symme-

tries, which always are present at high concentration of the RE ions
[151].
2. Direct 4f n⫺15d → 4f n emission from the levels of the 4f n⫺15d electronic
configuration due to large energy separation of the 4f n⫺15d levels
and/or weak electron [4f n⫺15d]–phonon interaction.
3. Emission from the levels of the 4f n⫺15d electronic configuration due
to the repopulation of these levels via phonon trapping and reabsorp-

Copyright © 2002 by Taylor & Francis

 tion of the lattice vibrations within the levels of the 4f n⫺15d electronic
configuration.
4. Spin-forbidden transitions.

Direct emission from the Stark components of the levels of the 4f n⫺15d
electronic configuration should be taking place only when: (1) two successive
electronic levels are well separated by the combined energy of a few phonons
(e.g., the phonon energy of LiYF 4 crystal is between 140 and 500 cm⫺1 [152]);
(2) the energy separation is larger than kT; and (3) the strength of the electron
[4f 95d]–phonon interaction is weak.
For example, in the case of the Ho3⫹ ions, there are 1878 electronic levels
of the 4f 95d electronic configuration in the energy range from 80,000 to 60,000 cm⫺1.
Therefore, two successive electronic levels are separated by 11 cm⫺1 (assuming an
even distribution of the electronic energy levels over the spectral range). Because the
phonon energies of the LiYF 4 crystal are higher than 150 cm⫺1 [153], there is a small
probability of direct transitions between the Stark components of the levels of the
4f 95d electronic configuration and the levels of the 4f 10 electronic configuration.
The interconfigurational 4f 95d → 4f 10 transitions in this case are taking place from
the edge of the levels of the 4f 95d electronic configuration.
For those of the RE ions with weak electron [4f n⫺15d]–phonon interaction,
the coupling strength of the interaction is described within the validity of the
Born-Oppenheimer approximation. In this case, radiationless transitions within
the levels of the 4f n⫺15d electronic configuration with a different set of good
quantum numbers are forbidden. Therefore, the probability for interconfigura-
tional radiative transitions, which originate directly from the levels of the 4f n⫺15d
electronic configuration, is expected to be high for those of the RE ions where
the electron [4f n⫺15d]–phonon interaction is weak. This situation is confirmed
by experimental observations in the case of the Tb3⫹ [83], Gd 3⫹ [22], and Er 3⫹
[9,128] ions. On the contrary, in the case of strong electron–phonon interaction,
as for the Pr 3⫹ [8], Nd 3⫹ [82], and Ho3⫹ [7] ions, the rate of the internal relaxation
within the levels of the 4f n⫺15d electronic configuration is expected to be high.
In this case the dipole-allowed transitions are expected to originate mainly from
the edge of the levels of the 4f n⫺15d electronic configuration.
In addition, for those of the RE ions with strong electron–phonon interac-
tion [5,153–155], dipole-allowed transitions directly from inside the levels (Stark
components) of the 4f n⫺15d electronic configuration could be explained on the
basis of the repopulation of these levels through phonon reabsorption and trap-
ping within the levels of the 4f n⫺15d electronic configuration [82]. The repopulat-
ing process will cease when the trapped phonons escape from the active volume
by frequency shifting through the dipole-allowed 4f n⫺15d → 4f n interconfigura-
tional transitions. Finally, the spin allowed and the spin-forbidden [150] intercon-

Copyright © 2002 by Taylor & Francis

figurational transitions can be explained on the basis of selection rules and rates
of relaxation within the levels of the 4f n⫺15d electronic configuration.

A. Cerium
There is much interest in Ce3⫹-doped ionic crystals for applications in UV scintil-
lators [150–160] and lasers [13–15,26]. Cerium trivalent ions in various dielec-
tric crystal hosts are characterized by excellent photochemical stability, high
conversion efficiency, and low excited-state absorption coefficient. They also ex-
hibit broadband tunable positive gain in UV, such as in LiCaAlF 6 crystal hosts
[161,162]. Polarized luminescence and absorption spectroscopy of Ce3⫹ ions in
dielectric fluoride crystals reveal the crystal field splitting of the 5d electronic
configuration [93,163]. The luminescence spectra of Ce3⫹ in different dielectric
fluoride crystal hosts consists of two bands due to transitions from the lowest
Stark component of the 5d electronic configuration to the 2F 5/2 and 2F 7/2 levels of
the 4f electronic configuration.

B. Praseodymium
The LIF spectrum of Pr 3⫹ ions in LiLuF 4 single-crystal hosts [8] excited in the
4f5d electronic configuration with the F 2 molecular laser at 157 nm indicated
that interconfigurational transitions mainly originate from the edge of the band
of the 4f5d electronic configuration at 46400 cm⫺1 and spans the spectral range
from 218 to 280 nm. Despite the fact that electron–phonon interaction within the
4f5d electronic configuration is strong, the LIF spectrum indicated the presence of
two weak bands around 200 and 170 nm. These bands were assigned to intercon-
figurational transitions from the Stark components of the 4f5d electronic configu-
ration at 53,200 and 63,400 cm⫺1, respectively, and taking into consideration the
energy gaps of the absorption spectrum (which are similar to the energy gaps in
the absorption spectrum of the Tb3⫹ ions [83,135]). The two weak bands have
been observed from the same crystal using x-rays as the excitation source [162].
UV fluorescence spectra using two photon excitation techniques have been re-
corded as well [163]. Luminescence spectra of Pr 3⫹ ions in YF3 and LaF 3 crystals,
where the edge of the levels of the 4f5d electronic configuration is at 50,000
cm⫺1, indicated that radiative interconfigurational transitions in the UV and the
visible are taking place between the 1S 0 energy level at 48,000 cm⫺1 and the lower
energy levels of the 4f 2 electronic configuration [24,111].

C. Neodymium
In the case of VUV excitation of the YLF:Nd crystal, it was observed experimen-
tally that the emission spectrum depends on the excitation wavelength [115]. Two

Copyright © 2002 by Taylor & Francis

peaks in this case were situated at 181.5 and 185.3 nm. The two peaks were
observed in multiphoton and single photon laser excitation of the YLF:Nd crys-
tals as well [7,10,82], at right angle to the excitation axis. The two peaks were
assigned to transitions from the low Stark component of the lower 4f 2 5d (4 K 11/2 )
level of the Nd 3⫹ ion to the levels of the 4f 3 configuration.
Indeed, when the crystal field is applied on the 4 K 11/2 level of the Nd 3⫹ ion
it splits it into three Stark components, Γ (8), Γ (7), and Γ (6), for a tetragonal site
symmetry of the Nd 3⫹ ion. The Γ (8) one is quadratic degenerate, 2(Γ (6) ⫹ Γ (7) )
and its degeneracy is removed when the L–S coupling is strong enough within
the 4f 2 5d electronic configuration. These levels occupy the spectral range from
453 to 1700 cm⫺1 above the low energy limit of the levels of the 4f 2 5d electronic
configuration, in agreement with findings by Thogersen et al. [10].
The broad band distribution of the electronic cloud near the edge of the
band reflects the wide electronic character of the 4f 2 5d (4 K 11/2 ) → 4f 3 dipole
transition, with the maximum of the emission at 181.5 and 185.3 nm, respectively.
The narrow emission peaks of the 180 nm band correspond to transitions from
the Stark component of the 4f 2 5d (4 K 11/2 ) level to the Stark components of the
4
I 9/2, and 4 I 11/2 levels of the 4f 3 configuration. The 4 I 9/2 level of the 4f 3 configura-
tion is split into five Stark components at 1, 139, 180, 236, and 529 cm⫺1 [82].
The 4 I 11/2 level of the 4f 3 configuration is split into five Stark components at 1979,
2015, 2056, 2221, and 2254 cm⫺1. At room and low temperatures all these levels
should be populated from transitions that originate from the 4 K 11/2 electronic level
of the 4f 2 5d configuration, at 55,648, 55,096, and 54,495 cm⫺1, respectively. In
these experiments the linewidth of the 4f 2 5d → 4f 3 atomic transitions, should
be less than 0.1 nm. This fact suggests that the emission cross section of the
4f 2 5d → 4f 3 transitions along the optical axis of the crystal is ⬃10⫺15 –10⫺16 cm2,
which is two or three orders of magnitude higher than the emission cross section
of the 4f 2 5d → 4f 3 transitions observed along the directions perpendicular to the
optical axis of the crystal. In the later case, the linewidth of the 4f 2 5d → 4f 3
transitions was 8.5 nm [10,115], giving a value for the corresponding emission
cross section of ⬃10⫺18 cm2.
A possible mechanism to explain the structure of the LIF spectrum of the
180 nm band could be phonon trapping within the levels of the Stark component
of the 4 K 11/2 level within the active volume of the crystal. This assignment is
supported by the fact that the energy difference of 552 and 600 cm⫺1 between
the Stark component of 4 K 11/2 level reflects the energy difference (within the ex-
perimental error) between successive vibronic modes of the YLF:Nd lattice [82].
This process is taking place because of the efficient coupling between sites
of the Nd 3⫹ ions, through multipole or exchange interactions. The repopulating
process of the higher vibronic modes will cease when the trapped phonons escape
from the active volume by frequency shifting via the dipole-allowed 4f 2 5d →
4f 3 interconfigurational transitions.

Copyright © 2002 by Taylor & Francis

Figure 10 LIF spectrum of LiLuF 4:Tb3⫹ following excitation at 157.6 nm. (From Ref.
83.)

The radiative lifetime of the transitions within the 180 nm band was found
to be ⬃30 ns in agreement with previous measurements [4,6,164], using different
crystal samples and measurements on VUV decay and energy transfer dynamics
of Nd 3⫹ ions [165] in various dielectric crystal hosts under x-ray and synchrotron
excitation [166–168] for scintillation and laser [169,170] applications.

D. Terbium
Terbium trivalent ions are an interesting case since in a pure L–S coupling there
is not interaction between the 4f 7 and the 5d electronic configuration, the interac-
tion is only through the crystal field. On the other hand, partial restoration of the
spherical symmetry of the 5d electronic configuration in the crystal field of the
surrounding ligands is taking place. The LIF spectrum in the spectral range from
150 to 270 nm under excitation at 157.6 nm, with the F 2 laser, for 0.01 at.%
concentration of the Tb3⫹ ion at liquid nitrogen temperature, is in Figure 10. The
fluorescence peaks were assigned to the dipole transitions between the levels of
the 4f 7 5d electronic configuration and the levels of the 4f 8 electronic configura-
tion of the Tb3⫹ ion [83]. When the Tb3⫹ ion is excited from its 4f 8 ( 7F 6 ) ground
level to a given level of the 4f 75d electronic configuration, the experimental evi-
dences indicate that it populates the levels of the 4f 8 electronic configuration,

Copyright © 2002 by Taylor & Francis

from both the levels and the edge of the 4f 75d electronic configuration. Direct
emission from the levels of the 4f 75d electronic configuration could take place
only when two successive electronic levels are well separated by the combined
energy of few phonons, as mentioned previously. However, according to
Szczurek and Schlesinger [81], there are 3106 electronic levels of the 4f 75d elec-
tronic configuration in the energy range from 80,000 to 45,000 cm⫺1. This number
corresponds to the energy level density of 13 electronic levels per wavenumber
(considering an even distribution of the energy levels over the spectral range).
Hence there is a small probability of direct transitions between the Stark compo-
nents of the levels of the 4f 75d electronic configuration and the levels of the 4f 8
electronic configuration. Therefore the LIF spectrum can be interpreted provided
that repopulation of the levels of the 4f 75d electronic configuration is taking
place. The repopulation occurs via phonon reabsorption and trapping within the
levels of the 4f 75d electronic configuration. These kind of processes have their
origin in the strong ionic coupling between two different Tb3⫹ ions through
multipole or exchange interactions, and they have been observed previously for
different crystal samples as well [172]. The repopulating process will cease when
the trapped phonons escape from the active volume by frequency shifting via the
interconfigurational transitions. Phonon trapping is usually detected by observing
the fluorescent radiation from the excited levels to the ground state. In the case
of weak electron [4f n⫺1]–phonon interaction and/or spin-forbidden 4f n⫺15d tran-
sitions, the phonon trapping and reabsorption processes compete with the nonra-
diative relaxation process within the levels of the 4f n⫺15d electronic configura-
tion. In this case, the interconfigurational transitions from inside the levels of the
4f n⫺15d electronic configuration are strong enough, as for example in the case
of the Tb3⫹, Gd 3⫹, and Er 3⫹ ions [172]. In the case of strong electron–phonon
interaction, for example for the Nd 3⫹ ion, the rate of the internal relaxation within
the levels of the 4f n⫺15d electronic configuration is higher than the rate of the
phonon reabsorption within the levels of the 4f n⫺15d electronic configuration. In
this case, the dipole transitions originate mainly from the edge of the levels of
the 4f n⫺15d electronic configuration.
As one finds from the LIF spectrum in Figure 10, the dipole transitions,
between the 7 D 5 Stark components of the 4f 75d electronic configuration and the
levels of the 4f 8 electronic configuration are due to the dipole transitions 7 D 5 →
7
F 4,5,6, which satisfy the selection rules for electric dipole transitions ∆S ⫽ 0,
∆l ⫽ ⫹1, ⫺1, ∆L ⫽ 0, ⫹1, ⫺1, ∆J ⫽ 0, ⫹1, ⫺1 and ∆J ⫽ ∆L.
The emission band, which covers the spectral range from 220 to 253 nm,
reflects the wide electronic character of the 4f 75d( 7 D 5 ) → 4f 8 ( 7F 0–6 ) dipole transi-
tions of the Γ4 Stark component of the levels of the 4f 75d ( 7 D 5 ) electronic config-
uration. The interconfigurational transitions in this spectral range originate from
the edge of the levels of the 4f 75d electronic configuration. From the LIF spec-
trum of [83], it was found that the energy position with the higher Frank–Condon

Copyright © 2002 by Taylor & Francis

factor is at (45.2 ⫾ 0.2) ⫻ 103 cm⫺1. In this case the following assignments can
be made regarding the observed dipole transitions (Fig. 11).
4f 75d[7 D 5 (Γ4 ), 45.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 6, 0) ⫹ hν(221.2 nm)
4f 75d[7 D 5 (Γ4 ), 45.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 5, 1.5 ⫻ 103 cm⫺1 ) ⫹ hν(228.8 nm)
4f 75d[7 D 5 (Γ4 ), 45.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 4, 3.1 ⫻ 103 cm⫺1 ) ⫹ hν(237.7 nm)
Besides the above-observed dipole-allowed transitions, dipole-forbidden transi-
tions [173] (∆J ≠ 0, ⫾1) were observed from the Γ4 Stark component of the
levels of the 4f 75d ( 7 D 5 ) electronic configuration.
4f 75d[7 D 5 (Γ4 ), 45.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 3, 4.0 ⫻ 103 cm⫺1 ) ⫹ hν(242.5 nm)
4f 75d[7 D 5 (Γ4 ), 45.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 2, 4.2 ⫻ 103 cm⫺1 ) ⫹ hν(244.2 nm)
4f 75d[7 D 5 (Γ4 ), 45.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 1, 5.0 ⫻ 103 cm⫺1 ) ⫹ hν(248.5 nm)
4f 75d[7 D 5 (Γ4 ), 45.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 0, 5.6 ⫻ 103 cm⫺1 ) ⫹ hν(252.7 nm)
The position of the levels of the 4f 8 electronic configuration of the Tb3⫹ ion in
the LiLuF 4, single-crystal host, is not known exactly. The position of these levels
in the free ion case, for the 7F 0–5 terms is at 58,000, 51,000, 48,000, 40,000, 31,000,
and 19,000 cm⫺1 respectively. From the LIF spectrum and for the Γ4 Stark compo-
nent, the experimental position of the 7F 0–5 terms in the LiLuF 4 crystal was found
to be at 56,000, 50,000, 42,000, 40,000, 31,000, and 15,000 cm⫺1, respectively.
Taking into consideration the fact that the electrons of the 4f 8 electronic configura-
tion are well screened by the electrons of the 5d electronic configuration, the posi-
tion of the levels of the 4f 8 electronic configuration inside the LiLuF 4 crystal should
be shifted relative to the position of the levels in the free ion case, by few wavenum-
bers, and this situation is reflected well in the experimental data.
The emission spectrum that covers the spectral region from 193 to 220 nm
can likewise be assigned to the dipole transitions between the Γ5 Stark compo-
nent of the 4f 75d( 7 D 5 ) electronic configuration, with maximum of absorption at
54,400 cm⫺1 and the levels of the 4f 8 electronic configuration. In this case the
Frank–Condon factor of the corresponding transitions has its highest value at
(52.9 ⫾ 0.2) ⫻ 103 cm⫺1. For these transitions, the following assignments can
be made within the limits of the experimental error of 200 cm⫺1 for these wave-
lengths.
4f 75d[7 D 5 (Γ5 ), 52.9 ⫻ 103 cm⫺1] → 4f 8 ( 7F 6 ) ⫹ hν(188.9 nm)
4f 75d[7 D 5 (Γ5 ), 52.9 ⫻ 103 cm⫺1] → 4f 8 ( 7F 5, 1.5 ⫻ 103 cm⫺1 ) ⫹ hν(194.7 nm)
4f 75d[7 D 5 (Γ5 ), 52.9 ⫻ 103 cm⫺1] → 4f 8 ( 7F 4, 3.0 ⫻ 103 cm⫺1 ) ⫹ hν(200.2 nm)
4f 75d[7 D 5 (Γ5 ), 52.9 ⫻ 103 cm⫺1] → 4f 8 ( 7F 3, 4.0 ⫻ 103 cm⫺1 ) ⫹ hν(204.5 nm)
From the LIF spectrum [83], it was found that the position of the Stark compo-
nents of the levels of the 4f 8 electronic configuration 7F 3– 5 in the LiLuF 4 crystal

Copyright © 2002 by Taylor & Francis

Figure 11 Simplified energy diagram shows various transitions of the LIF spectra of
the LiLuF4 :Tb3⫹ crystal. The position of levels of the 4f 75d electronic configuration of
the Tb3⫹ ion is where the maximum absorption is taking place.

Copyright © 2002 by Taylor & Francis

is at 40,000, 30,000, and 15,000 cm⫺1, respectively, in agreement with the energy
position of these levels in the spectral range from 220 to 253 nm.
The transition around 194 nm could originate as well from the Γ3 Stark
component, with maximum of absorption at 52,000 at 51,600 cm⫺1.
4f 75d[7 D 5 (Γ3 ), 51.6 ⫻ 103 cm⫺1] → 4f 8 ( 7F 6 ) ⫹ hν(193.8–194.7 nm)
The emission spectrum, which covers the spectral range from 159 to 174
nm, can be assigned to the dipole transitions between the Γ2 Stark component
of the 4f 75d( 7 D 5 ) electronic configuration, and the levels of the 4f 8 electronic
configuration. However, for this spectral range, it is difficult to make any specific
assignment for the 4f 75d → 4f 8 transitions, due to the presence of intense scat-
tered laser light. Taking into consideration the position of the levels of the Tb3⫹
ions of the 4f 8 electronic configuration in the LiLuF 4 crystal host, and the emis-
sion spectrum, the following assignments can be made for the 4f 75d → 4f 8 transi-
tions:
4f 75d[7 D 5 (Γ2 ), 62.6 ⫻ 103 cm⫺1] → 4f 8 ( 7F 6 ) ⫹ hν(159.7 nm)
4f 75d[7 D 5 (Γ2 ), 62.6 ⫻ 103 cm⫺1] → 4f 8 ( 7F 4 or 7F 3 ) ⫹ hν(171 nm)
4f 75d[7 D 5 (Γ2 ), 59.3 ⫻ 103 cm⫺1] → 4f 8 ( 7F 6 ) ⫹ hν(168.6 nm)
4f 75d[7 D 5 (Γ2 ), 59.3 ⫻ 103 cm⫺1] → 4f 8 ( 7F 4 ) ⫹ hν(174.1 nm)
The transitions at 278, 256.2, and 287.2 nm were assigned to the spin-forbidden
transitions [173], taking into consideration that the energy position of the 5 D 4
and 5 D 3 levels of the 4f 8 electronic configuration in the free Tb3⫹ ion case is at
20,500 and 26,000 cm⫺1, respectively.
4f 75d[7 D 5 (Γ1 ), 60,200 cm⫺1] → 4f 8 (5 D 3 ) ⫹ hν(278 nm)
4f 75d[7 D 5 (Γ2 ), 59.3 ⫻ 103 cm⫺1] → 4f 8 (5 D 4 ) ⫹ hν(256.2 nm)
4f 75d[7 D 5 (Γ2 ), 59.3 ⫻ 103 cm⫺1] → 4f 8 (5 D 3 ) ⫹ hν(287.2 nm)
The transitions at 166.1 and 174.8 nm [83] can be assigned to the interconfigura-
tional transitions originating from the energy position at 60,200 cm⫺1.
4f 75d[7 D 5 (Γ2 ), 60.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 6 ) ⫹ hν(166.1 nm)
4f 75d[7 D 5 (Γ2 ), 60.2 ⫻ 103 cm⫺1] → 4f 8 ( 7F 4 ) ⫹ hν(174.8 nm)

E. Holmium
The LIF spectrum of the LiYF 4:Ho3⫹ crystal, under excitation at 157.6 nm with
the F 2 laser, for 0.1 at.% concentration of the Ho3⫹ ion at liquid nitrogen tempera-
ture includes both spin-allowed and spin-forbidden interconfigurational transi-
tions and intraconfigurational transitions [7,174,175]. The emission peak at 171.2
nm is assigned to the interconfigurational spin-allowed transition from the lower

Copyright © 2002 by Taylor & Francis

Stark component of the 5 H 7 term at 63,450 cm⫺1 (157.6 nm) to the 5 I 7 level of
the 4f 10 electronic configuration:

4f 95d(63,450 cm⫺1 ) → 4f 10 (5 I 7 ) ⫹ hν(171.2 nm)

The spin-allowed transition to the ground 5 I 8 level (which is expected at 157.6

nm) was not observed due to the presence of the intense laser scattered light.
The lowest 7 H 8 level of the 4f 95d electronic configuration should likewise
split into four doubly degenerate Stark components in the LiYF 4 crystal field. The
transitions that originate from the Stark components of the 4f 95d ( 7 H 8 ) electronic
configuration are spin-forbidden and therefore they are expected to be of low
intensity. The transitions at 167.5 and 165.6 nm are assigned to the spin-forbidden
transitions as follows:

4f 95d( 7 H 8, 60,380 cm⫺1 )→ 4f 10 (5 I 8 ) ⫹ hν(165.6 nm)

4f 95d( 7 H 8, 59,700 cm⫺1 )→ 4f 10 (5 I 8 ) ⫹ hν(167.5 nm)

The LIF transitions from 172.5 to 189 nm were assigned to the 4f 10 → 4f 10

intraconfigurational transitions. From both emission and absorption spectra, the
dipole-forbidden transitions at 172.5, 174, 177.5, 178.5, 181, 185, 188, and 189
nm were assigned to the transitions from the high excited levels of the 4f 10 elec-
tronic configuration to the ground 5 I 8 level [174,175].

F. Erbium
The LIF spectrum of the Er 3⫹ ions in LiYF 4 crystal hosts of Figure 12 includes
both spin-allowed and spin-forbidden transitions. The emission bands at 158.6,
157, and 195 nm were assigned to dipole-allowed transitions from the edge of the
lower energy level of the 4f 105d electronic configuration to the energy levels of
the 4f 11 electronic configuration [9]. Emission at 167–169 nm from an absorption
band situated at 163 nm [115] was observed for the first time by Sarantopoulou
et al. [9]. It was assigned to spin-forbidden transitions from the Stark component
of the 4f 11 electronic configuration to the ground level 4 I 15/2 of the 4f 11 electronic
configuration [9,150]. VUV emission of Er 3⫹ ions in various dielectric hosts using
synchrotron and x-ray excitation has been reported as well [176–179].

G. Thulium
Due to the fact that the laser photons at 157 nm populate the 4f 115d electronic
configuration of the Tm3⫹ ions in LiLuF 4 crystal hosts just above the edge of the
band [135], the LIF spectrum of the LiYF 4:Tm3⫹ monocrystals, under the F 2 laser
excitation, exhibits strong interconfigurational transitions in the VUV (Fig. 13).

Copyright © 2002 by Taylor & Francis

Figure 12 VUV LIF spectrum of LiLuF 4 :Er 3⫹ crystal under F 2 laser excitation.

Figure 13 VUV LIF spectrum of LiYF 4:Tm3⫹ crystals under F2 laser excitation.

Copyright © 2002 by Taylor & Francis

REFERENCES

1. KH Yang, JA DeLuka. VUV fluorescence of Nd 3⫹, Er 3⫹ and Tm3⫹ and Tm3⫹ doped
trifluorides and tuneable coherent sources from 1650 to 2600 nm. Appl Phys Lett
29: 499–501, 1976.
2. S Heaps, LR Elias, WM Yen. Vacuum-ultraviolet bands of trivalent lanthanides in
LaF 3. Phys Rev B 13: 94–104, 1976.
3. KH Yang, JA DeLuca. Vacuum ultraviolet excitation studies of 5d 1 4f n⫺1 to 4f n
and 4f n to 4f n transitions of Nd 3⫹, Er 3⫹ and Tm3⫹-doped trifluorides. Phys Rev B
17: 4246–4255, 1978.
4. MA Dubinskii, AC Cefalas, E Sarantopoulou, RY Abdulsabirov, SL Korableva,
AK Naumov, VV Semashko. On the interconfigurational 4f 2 5d–4f 3 VUV and UV
fluorescence features of Nd 3⫹ in YLF single crystals under fluorine laser pumping.
Opt Commun 94: 115–118, 1992.
5. JP Krupa, I Gerard, A Mayoler, P Martin. Electronic structure of f-element system
in the UV and VUV energy range. Acta Phys Pol A84: 843–848, 1993.
6. R Visser, P Dorenbos, CWE Van Eijk, A Meijerink, HW den Hartog. The scintilla-
tion intensity and decay from Nd 3⫹ 4f 2 5d and 4f 3 excited states in several fluoride
crystals. Phys Cond Matt 5: 8437–8460, 1993.
7. E Sarantopoulou, AC Cefalas, MA Dubinskii, CA Nicolaides, RY Abdulsabirov,
SL Korableva, AK Naumov, VV Semashko. VUV and UV fluorescence and absorp-
tion studies of Nd 3⫹ and Ho3⫹ ions in LiYF 4 single crystals. Opt Commun 107:
104–110, 1994.
8. E Sarantopoulou, AC Cefalas, MA Dubinskii, CA Nicolaides, RY Abdulsabirov,
SL Korableva, AK Naumov, VV Semashko. VUV and UV fluorescence and absorp-
tion studies of Pr 3⫹-doped LiLuF 4 single crystals. Opt Lett 19: 499–501, 1994.
9. E Sarantopoulou, Z Kollia, AC Cefalas, MA Dubinskii, CA Nicolaides, RY Abdul-
sabirov, SL Korableva, AK Naumov, VV Semashko. Vacuum ultraviolet and ultra-
violet fluorescence and absorption studies of Er 3⫹-doped LiLuF 4 single crystals.
Appl Phys Lett 65: 813–815, 1994.
10. J Thogersen, JD Gill, HK Haugen. Stepwise multiphoton excitation of the 4f 2 5d
configuration in Nd 3⫹:YLF. Opt Commun 132: 83–88, 1996.
11. S Guy, MF Joubert, B Jacquier. Photon avalance and mean-field approximation.
Phys Rev B 55: 8240–8248, 1997.
12. M Malinowski, B Jacquier, M Bouazaoui, MF Joubert, C Linares. Laser induced
fluorescence and up conversion processes in LiLuF 4: Nd 3⫹ laser crystals. Phys Rev
B 41: 31–40, 1990.
13. DJ Ehrlich, PF Moulton, RM Osgood Jr. Ultraviolet solid state Ce:YLF laser at
325 nm. Opt Lett 4: 184–186, 1979.
14. K-S Lim, DS Hamilton. Optical gain and loss studies in Ce3⫹: YLiF 4. J Opt Soc
Am B6: 1401–1406, 1989.
15. DJ Ehrlich, PF Moulton, RM Osgood Jr. Optically pumped Ce:LaF 3 laser at 286
nm. Opt Lett 5: 339–341, 1980.
16. RW Waynant, PH Klein. Vacuum ultraviolet emission from Nd 3⫹: LaF 3. Appl Phys
Lett 46: 14–15, 1985.
17. RW Waynant. Vacuum ultraviolet laser emission from Nd 3⫹: LaF 3. Appl Phys B
28: 205, 1982.

Copyright © 2002 by Taylor & Francis

 18. MA Dubinskii, AC Cefalas, CA Nicolaides. Solid state Nd 3⫹ doped LaF 3 VUV laser
pumped by a pulsed discharge fluorine-molecular laser at 157 nm. Opt Commun 88:
122–124, 1992.
19. JC van’t Spijker, P Dorenbos. CWE van Eijk, JEM Jacobs, HW den Hartog, N
Korolev. Luminescence and scintillation properties of BaY 2F 8:Ce3⫹, BaLu 2F 8 and
BaLu 2F 8:Ce3⫹. J Lumin 85: 11–19, 1999.
20. S Kubodera, M Kitahara, J Kawanaka, W Sasaki, K Kurosava. A vacuum ultraviolet
flash lamp with extremely broadened emission spectra. Appl Phys Lett 69: 452–
454, 1996.
21. G Blasse, BC Grabmayer. Lumininescent Materials. Berlin: Springer Verlag,
1996.
22. RT Wegh, H Donker, A Meijerink, RJ Lamminmaki, J Holsa. Vacuum-ultraviolet
spectroscopy and quantum cutting for Gd 3⫹ in LiLuF 4. Phys Rev B 56: 13841–
13848, 1997.
23. CM Combes, P Dorenbos, CWE Vaneijk, C Pedrini, HW Denhartog, JY Gesland,
PA Rodnyi. Optical and scintillation properties of Ce3⫹ doped LiYF 4 and LiLuF 4
crystals. J Lumin 71: 65–70, 1997.
24. AM Srivastava, SJ Duclos. On the luminescence of YF 3-Pr 3⫹ under vacuum ultravi-
olet and X-ray emission. Chem Phys Lett 275: 453–456, 1997.
25. MJF Digonnet, RW Sadowski, HJ Shaw, RH Pantell. Resonantly enhanced nonline-
arity in doped fibers for low power all optical switching. Opt Fib Techn 3: 44–64,
1997.
26. CD Marshall, JA Speth, SA Payne, WF Krupke, GJ Quarles, V Castillo, B Chai.
Ultraviolet laser emission properties of Ce3⫹ doped LiSrAlF 6 and LiCaAlF 6. J Opt
Soc Am B 11: 2054–2065, 1994.
27. TM Bloomstein, M Rothschild, RR Kunz, DE Hardy, RB Goodman, ST Palmacci.
Critical issues in 157 nm lithography. J Vac Sci Techn B 16: 3154–3157,
1998.
28. AC Cefalas, E Sarantopoulou, P Argitis, E Gogolides. Mass spectroscopic and de-
gassing characteristics of polymeric materials for 157 nm photolithography. Appl
Phys A 69: 5229–5234, 2000.
28a. E Sarantopoulou, Z Kollia, AC Cefalas. YF3 : ND3⫹, Pr3⫹, Gd3⫹ wide band gap crys-
tals as optical materials for 157 nm photolithography. Optical Materials 18: 23–
26, 2001.
29. GH Dieke, HM Crosswhite. The spectra of the doubly and triply ionized rare earths.
Appl Opt 2: 675–686, 1963.
30. R Hilbic, R Wallenstein. Enhanced Production of Tunable VUV Radiation by
phase-matched frequency tripling in krypton and xenon. IEEE J Quantum Electron
17: 1566–1573, 1981.
31. W Jamroz, PE LaRocque. BP Stoicheff. Generation of continuously tunable coher-
ent vacuum-ultraviolet radiation (140–106 nm) in zinc vapor. Opt Lett 7: 617–
619, 1982.
32. JC Miller, RN Compton, CD Cooper. Vacuum ultraviolet spectroscopy of mole-
cules using third-harminic generation in rare-gases. J Chem Phys 76: 3967–3973,
1982.
33. Y Hirakawa, K Nakai, T Obara, M Maeda, K Muraoka. Generation of coherent

Copyright © 2002 by Taylor & Francis

 XUV radiation by 2-photon resonant 4-wave-mixing combined with stimulated Ra-
man-scattering. Opt Commun 92: 215–218, 1992.
34. K Miyazaki, H Sakai, T Sato. Efficient deep-ultraviolet generation by frequency
doubling in beta-Bab2O4 crystals. Opt Lett 11: 797–799, 1986.
35. BC Wu, F Xie, C Chen, D Deng, Z Xu. Generation of tunable coherent vacuum
ultraviolet-radiation in LiB 3O 5 crystal. Opt Commun 88: 451–454, 1992.
36. NG Basov, VA Danilychev. Condensed- and compressed-gas lasers. Sov Phys Usp
29:31–56, 1986.
37. RH Lipson, PE Larocque, BP Stoicheff. Vacuum ultraviolet laser-excited spectra
of Xe 2. Opt Lett 9: 402–404, 1984.
38. T Eftimiopoulos, BP Stoicheff, RI Thomson. Efficient population-inversion in exci-
mer states by supersonic expansion of discharge plasmas. Opt Lett 14: 624–626,
1989.
39. H Nahme, N Schwentner. Luminescence of rare-gas crystals at high-excitation den-
sities for VUV laser applications. Appl Phys B 51: 177–191, 1990.
40. AC Cefalas, TA King. Injection locking of ArF excimer lasers. Appl Phys B37:
159–164, 1985.
41. AC Cefalas, TA King. Phase conjugation by four wave mixing in an ArF excimer
amplifier. Opt Commun 51: 105–110, 1984.
42. JR Woodworth, JK Rice. An efficient high-power F 2 laser near 157 nm. J Chem
Phys 69: 2500–2504, 1978.
43. JR Woodworth, JK Rice. High intensity laser photolysis of OCS at 157 nm. IEEE
J Quantum Electron 15: 88D–89D, 1979.
44. H Pummer, OK Hohla, M Digelman, JP Reilly. Discharge pumped F 2 pulsed laser
at 1580 A. Opt Commun 28: 104–106, 1979.
45. AC Cefalas, C Skordoulis, M Kompitsas, CA Nikolaides. Gain measurements at
157 nm in an F 2 pulsed discharge molecular laser. Opt Commun 55: 423–426,
1985.
46. VN Ishchenko, SA Kochubei, AM Razhev. High-power efficient vacuum ultravio-
let F 2 laser excited by an electric discharge. Sov J Quantum Electron 16:707–709,
1986.
47. M Ohwa, M Obara. Theoretical evaluation of high-efficiency operation of dis-
charge-pumped vacuum ultraviolet F 2 lasers. Appl Phys Lett 51: 958–960,
1987.
48. K Yamada, K Miyazaki, T Hasama, T Sato. High-power discharge-pumped F 2 mo-
lecular laser. Appl Phys Lett 54: 597–599, 1989.
49. C Skordoulis, E Sarantopoulou, S Spyrou, AC Cefalas. Amplification characteris-
tics of a discharge excited F 2 laser. J Mod Opt 37: 501–509, 1990.
50. C Skordoulis, S Spyrou, AC Cefalas. Gain and saturation measurements in a dis-
charge excited F 2 laser using an oscillator amplifier configuration. Appl Phys
B 51: 141–145, 1990.
51. H Nishimura, DC Cartwright, J Trajmar. Electron energy loss spectroscopy of mo-
lecular fluorine. J Chem Phys 71: 5039–5041, 1979.
52. DJ Cartwright, PJ Hay. Theoretical studies of the valence electronic states and the
1
Π u → X 1 ∑⫹g absorption spectrum of the F 2 molecule. J Chem Phys 70: 3191–
3203, 1979.

Copyright © 2002 by Taylor & Francis

 53. K Hoshiba, Y Fujita, SS Kano, H Takuma, T Takayanagi, K Wakiya, H Suzuki.
Experimantal observation of the F 2 VUV laser levels. J Phys B Atom Mol Phys
18: L875–L879, 1985.
54. RG Wang, ZW Wang, MA Dillon, D Spense. Electron energy loss spectroscopy
of molecular fluorine. J Chem Phys 80: 3574–3579, 1984.
55. AC Cefalas, TA King. A doubly preionized ArF laser. J Phys E: Sci Instr 17: 760–
764, 1984.
56. RC Caro, MC Gower, CE Webb. A simple tunable KrF laser system with narrow
bandwidth and diffraction limited divergence. J Phys D Appl Phys 15: 767–773,
1982.
57. R Loundon. The Quantum Theory of Light. Oxford: Oxford University Press, 1973.
58. NN Ershov, NG Zakharov, PA Rodnyi. Spectral-kinetic study of the intrinsic lumi-
nescence characteristics of a fluorite-type crystal. Opt Spectrosk 53: 51–54 1982.
59. CWE van Eijk. Cross-luminescence. J Lumin 60: 936–941, 1994.
60. E Sarantopoulou, Z Kollia, AC Cefalas. LiCaAlF 6:Nd 3⫹ crystal as optical material
for 157 nm photolithography. Opt Commun 177: 377–382, 2000.
61. Z Kollia, E Sarantopoulou, AC Cefalas, CA Nikolaides, AK Naumov, VV Sem-
ashko, RYu Abdulsabirov, SL Korableva, MA Dubinskii. Vacuum-ultraviolet inter-
configurational 4f 3 → 4f 2 5d absorption and emission studies of the Nd 3⫹ ion in
KYF, YF and YLF crystal hosts. J Opt Soc Am B12: 782–785, 1995.
62. V Dauer. Optical constants of lithium fluoride thin films in the far ultraviolet. J
Opt Soc Am B 17: 300–303, 2000.
63. ED Palik, ed. Handbook of Optical Constants of Solids. 1st ed. Orlando, FL: Aca-
demic Press, 1985.
64. A Milgram, M Parker Givens. Extreme ultraviolet absorption by lithium fluoride.
Phys Rev 125: 1506–1509, 1962.
65. P Laporte, JL Subtil, M Courbon, M. Bon. Vacuum ultraviolet refractive index of
LiF and MgF 2 in the temperature range 80–300K. J Opt Soc Am 73: 1062–1069,
1983.
66. G Stephan, YLe Calvez, JC Lemonier, Mme S Robin. Properties optiques et spectre
electronique du MgF 2 et du CaF 2 de 10 a 48 eV. J Phys Chem Solids 30: 601–
608, 1969.
67. OR Wood, HG Craighead, JE Sweeney, PJ Maloney. Vacuum ultraviolet loss in
magnesium fluoride films. Appl Opt 23: 3644–3649, 1984.
68. MW Williams, RA MacRae, ET Arakawa. Optical properties of magnesium fluo-
ride in the vacuum ultraviolet. J Appl Phys 38: 1701–1705, 1966.
69. ET Hutcheson, G Hass, JT Cox. Effect of deposition rate and substrate temperature
on the vacuum ultraviolet reflectance of MgF 2- and LiF-overcoated aluminum mir-
rors. Appl Opt 11: 2245–2248, 1972.
70. AS Barriere, A Lachter. Optical transitions in disordered thin films of the ionic
compounds MgF 2 and AlF 3 as a function of their conditions of preparation. Appl
Opt 16: 2865–2871, 1977.
71. A Mayolet, JC Krupa, I. Gerard, P. Martin. Luminescence of Eu3⫹ doped materials
excited by VUV synchrotron radiation. Materials Chem Phys 31: 107–109, 1992.
72. AN Belsky, JC Krupa. Luminescence excitation mechanisms of rare earth doped
phosphors in the VUV range. Displays 19:185–196, 1999.

Copyright © 2002 by Taylor & Francis

 73. G Ionova, JC Krupa, I. Gerard, R. Guillaumont. Systematics in electron-transfer
energies for lanthanides and actinides. N J Chem 19: 677–689, 1995.
74. I Gerard, JC Krupa, E Simoni, P Martin. Investigation of charge transfer O2⫺ →
Ln3⫹ and F⫺ → Ln3⫹ in LaF 3: (Ln3⫹, O2⫺ ) and YF 3: (Ln3⫹, O2⫺ ) systems. J Alloys
Comp 207/208: 120–127, 1994.
75. HR Moser, G Wendin. Theoretical-models for intensities of d-f transitions in elec-
tron-energy-loss spectra of rare-earth and actinide metals. Phys Rev B 44: 6044–
6061, 1991.
76. BF Aull, HP Jenssen. Impact of ion-host interactions on the 5d to 4f spectra of
lanthanide rare-earth–metal ions. I. A phenomenological crystal field model. Phys
Rev B 34: 6640–6646, 1986.
77. L Brewer. Energies of the electronic configurations of the singly, doubly and triply
ionized lanthanides and actinides. J Opt Soc Am 61: 1666–1682, 1971.
78. BR Judd. Optical absorption intensities of rare-earth ions. Phys Rev B127: 750–
761, 1962.
79. E Loh. 4f n → 4f n⫺15d spectra of rare earth ions in crystals. Phys Rev B 175: 533–
536, 1968.
80. NV Starostin. On the theory of composite f n⫺1d configuration of lanthanides in
crystalline fields of cubic symmetry. Opt Spec 23: 260–261, 1964.
81. T Szczurek, M Schlesinger. Vacuum ultraviolet absorption spectra of CaF 2:RE 3⫹
crystals. In: Proceedings of the International Symposium Rare Earth Spectroscopy.
Singapore: World Scientific, pp 309–330, 1985.
82. Z Kollia, E Sarantopoulou, AC Cefalas, AK Naumov, VV Semashko, RYu Abdul-
sabirov, SL Korableva. On the 4f 2 5d → 4f 3 interconfigurational transitions of Nd 3⫹
ions in K 2YF 5 and LiYF 4 crystal hosts. Opt Commun 149: 386–392, 1998.
83. E Sarantopoulou, Z Kollia, AC Cefalas, VV Semashko, R Yu Abdulsabirov, AK
Naumov, SL Korableva. On the VUV and UV 4f 7 (8S)5d → 4f 3 interconfigurational
transitions of Tb3⫹ ions in LiLuF 4 single crystal hosts, Opt Commun 156: 101–
111, 1998.
84. P Feofilov. On absorption and luminescence spectra of Ce3⫹ ions. Opt Spectr 6:
150–151, 1959.
85. AA Kaplyanski, VN Medvedev, PP Feofilov. The spectra of trivalent cerium ions
in alcaline earth fluoride crystals. Opt Spect 14: 351–356, 1963.
86. MH Crozier. Zeeman effect in the 4f → 5d spectrum of Ce3⫹ in CaF 2. Phys Rev
137: A1781–A1783, 1965.
87. AA Kaplyanski, VN Medvedev. Piezospectroscopic determination of the symmetry
of the crystal field acting on trivalent ions of rare-earth in fluorite lattice. Opt Spect
18: 451–456, 1965.
88. AA Kaplyanski, VN Medvedev. Linear Stark effect in the spectra of local centers
in cubic crystals. Opt Spect 23: 404–410, 1967.
89. SK Gayen, DS Hamilton. Two photon excitation of the lowest 4f-5d near ultraviolet
transition in Ce3⫹: CaF 2. Phys Rev B28: 3706–3711, 1983.
90. MA Dubinskii, VV Semashko, AK Naumov, R Yu Abdulsabirov, SL Korableva.
Spectroscopy of new active medium of a solid-state UV laser with broadband single
pass gain. Laser Phys 3: 216–217, 1993.
91. IT Jacobs, GD Jones, K Zdansky, RA Satten. Electron phonon interaction of hydro-

Copyright © 2002 by Taylor & Francis

 genated, deuterogenated and tritiated crystals of calcium and strontium fluoride
containing cerium. Phys Rev B3: 2888–2910, 1971.
92. E Loh. Lowest 4f-5d transition of trivalent rare-earth ions in CaF 2 crystals. Phys
Rev 147: 332–335, 1966.
93. M Yamaga, D Lee, B Henderson, T Han, H Gallagher, T Yosida. The magnetic
and optical properties of Ce3⫹ in LiLuCaAlF 6. J Phys Cond. Matter 10: 3223–3237,
1998.
94. CD Marshall, JA Speth, SA Payne, WF Krupke, GJ Quarles, V Castillo. Ultraviolet
laser emission properties of Ce3⫹-doped LiSrAlF 6 and LiCaAlF 6. J Opt Soc Am B
11: 2054–2065, 1994.
95. E Loh. Ultraviolet absorption spectra of Ce3⫹ ion in alkaline–earth fluorides. Phys
Rev 154: 270–276, 1967.
96. WJ Manthey. Ultraviolet absorption spectra of Ce3⫹ ion in alkaline–earth fluorides.
Phys Rev B 8: 4086–4098, 1973.
97. E Loh, G Samoggia, E Reguzzoni, L Nosenzo. Thermaly modulated ultraviolet
absorption spectra of CaF 2:Ce3⫹ crystals. Phys Status Solidi B 79: 795–799 (1977).
98. M Schlesinger, PW Whippey. Investigations of 4f-5d transitions of Ce3⫹ in CaF 2.
Phys Rev 171: 361–364, 1968.
99. T Szczurek, GWF Drake, M Schlesinger. Vibronic structure in the absorption spec-
trum of the Ce3⫹ ion in CaF 2. Phys Rev B8: 4910–4912, 1973.
100. J Sugar. Analysis of the spectrum of triply ionized praseodymium (Pr IV). J Opt
Soc Am 55: 1058–1061, 1965.
101. GH Dieke, HM Crosswhite, B Dunn. Emission spectra of the doubly and triply
ionized rare earths. J Opt Soc Am 51: 820–827, 1961.
102. R Sarup, M Crozier. Analysis of the eigenstates of Pr 3⫹ in LaCl 3 using the Zeeman
effect in high fields. J Chem Phys 42: 371–376, 1965.
103. J Sugar. Description and analysis of the third spectrum of cerium. J Opt Soc Am
55: 33–58, 1965.
104. E Loh. 1S 0 level of Pr 3⫹ in crystals of fluorides. Phys Rev A 140: 1463–1466, 1965.
105. WT Carnall, PR Fields, R Sarup. 1S 0 level of Pr 3⫹ in crystal matrices and energy-
level parameters for the 4f 2 configuration of Pr 3⫹ in LaF 3. J Chem Phys 51: 2587–
2591, 1969.
106. EY Wong, OM Stafsudd, DR Johnston. Absorption and fluorescence spectra of
several praseodymium-doped crystals and the change of covalence in the chemical
bonds of the praseodymium ion. J Chem Phys 39: 786–793, 1963.
107. HH Caspers, HE Rast, RA Buchanan. Energy levels of the Pr 3⫹ in LaF 3. J Chem
Phys 43: 2124–2128, 1965.
108. E Loh. Ultraviolet absorption spectra of Pr 3⫹ ion in alkaline-earth fluorides. Phys
Rev 158: 273–279, 1967.
109. WM S Heaps, LR Elias, WM Yen. Vacuum-ultraviolet absorption bands of trivalent
lanthanides in LaF 3. Phys Rev B13: 94–104, 1976.
110. LR Elias, WM S Heaps, WM Yen. Excitation of uv fluorescence in LaF 3 doped
with trivalent cerium and praseodymium. Phys Rev B8: 4989–4995, 1973.
111. WW Piper, JA De Luca, FS Ham. Cascade fluorescent decay in Pr 3⫹-doped fluo-
rides: achievement of a quantum yield greater than unity for emission of visible
light. J Lumin 8: 344–348, 1974.

Copyright © 2002 by Taylor & Francis

112. AM Srivastava, DA Doughty, WW Beers. On the vacuum-ultraviolet excited lumi-
nescence of Pr 3⫹ in LaB 3O 6. J Electrochem Soc 144: L190–L192, 1997.
113. AM Srivastava, WW Beers. Luminescence of Pr 3⫹ in SrAl 12O 19: observation of two
photon luminescence in oxide lattice. J Lumin 71: 285–290, 1997.
114. AM Srivastava, DA Doughty, WW Beers. Photon cascade luminescence in Pr 3⫹
in LaMgB 5O 10. J Electrochem Soc 143: 4113–4116, 1997.
115. KM Devyatkova, ON Ivanova, SA Oganesyan, KB Seiranyan, SP Chernov. Lumi-
nescent properties of single crystals of LiYF 4:Nd 3⫹ in the vacuum ultraviolet region.
Sov Phys Dokl 35: 56–57, 1990.
116. GJ Quarles, GE Venikouas, RC Powell. Sequential two-photon excitation processes
of Nd 3⫹ ions in solids. Phys Rev B 31: 6935–6940, 1985.
117. J Sugar. Analysis of the third spectrum of praseodymium. J Opt Soc Am 53:831–
839, 1963.
118. MA Dubinskii, NM Khaidukov, IG Garipov, LN Dem’yanets, AK Naumov, VV
Semashko, VA Malysov. Spectral–kinetic and laser characteristics of new Nd 3⫹-
activated laser hosts of the KF-YF 3 system. J Mod Opt 37: 1355–1360, 1990.
119. W Baer, JG Conway, SP Davis. Crystal spectrum of promethium3⫹ in LaCl 3. J
Chem Phys 59: 2294–2302, 1973.
120. SIJ Weissman. Intermolecular energy transfer. The fluorescence of complexes of
Euro J Chem Phys 10: 214–217, 1942.
121. M Schlesinger, T Szczurek. Vacuum ultraviolet spectra of CaF 2:Eu3⫹ and CaF 2:
Sm3⫹. J Opt Soc Am 70: 1025–1029, 1980.
122. HE Rast, JL Fray, HH Caspars. Energy levels of Sm3⫹ in LaF 3. J Chem Phys 46:
1460–1466, 1967.
123. JD Axe, GH Dieke. Calculation of crystal field splittings of Sm3⫹ and Dy3⫹ levels
in LaCl 3 with inclusion of J mixing. J Chem Phys 37: 2364–2371, 1962.
124. BP Zakharchenya, VP Makarov, A Ya Ruskin. Zeeman effect for d-f transitions
in the spectra of Sm2⫹ activated alkali-earth fluoride crystals. Opt Spect 17: 116–
120, 1964.
125. ZJ Kiss, HA Weakliem. Stark effect of 4f states and linear crystal field in BaClF:
Sm2⫹. Phys Rev Lett 15: 457–460, 1965.
126. BP Zakharchenya, AYa Ruskin. Zeeman effect in the absorption and luminescence
spectrum of CaF 2:Sm⫹⫹ and SrF 2:Sm⫹⫹ crystals. Opt Spect 13: 501–502, 1962.
127. E Loh. 4f n → 4f n⫺15d spectra of rare-earth ions in crystals. Phys Rev 175: 533–
536, 1968.
128. RT Wegh, H Donker, A. Meijerink. Vacuum ultraviolet excitation and emission
studies of 4f n → 4f n⫺15d transitions for Ln3⫹ in LiYF 4. Proc Elect Soc 97: 284–
295, 1988.
129. M Schlesinger, GFW Drake. On the vibronic spectrum of rare-earths in calcium
fluoride. Can J Phys 54: 1699–1701, 1976.
130. WC Martin, R Zalubas, L Hagan. Atomic Energy Levels—The Rare Earth Ele-
ments. Washington, DC: National Bureau of Standards, 1978.
131. J Sugar, J Reader. Ionization energies of doubly and triply ionized rare earths. J
Chem Phys 59: 2083–2089, 1973.
132. JF Kielkopf, HM Crosswhite. Preliminary analysis of the spectrum of triply ionized
gadolinium (Gd IV). J Opt Soc Am 60: 347–351, 1970.

Copyright © 2002 by Taylor & Francis

133. GH Dieke. In: Crosswhite HM, Crosswhite H, eds. Spectra and Energy Levels of
Rare Earth Ions in Crystals, New York: Wiley, 1968, pp 249–253.
134. M Schlesinger, T Szczurek, GWF Drake. The lowest energy 4f → 5d transition of
the triply ionized gadolinium in CaF 2. Solid State Commun 28: 165–166, 1978.
135. E Sarantopoulou, AC Cefalas, MA Dubinskii, Z Kollia, CA Nicolaides, RY Abdul-
sabirov, SL Korableva, AK Naumov, VV Semashko. VUV and UV fluorescence
and absorption studies of Tb3⫹ and Tm3⫹ trivalent ions in LiLuF 4 single crystal
hosts. J Mod Opt 41: 767–775, 1994.
136. JL Fray, HH Caspars, HE Rast, SA Miller. Optical absorption and fluorescence
spectra of Dy3⫹ in LaF 3. J Chem Phys 48: 2342–2348, 1968.
137. M Schlesinger, T Szczurek, MCK Wiltshire. 4f → 5d transition studies of Tb3⫹
and Dy3⫹ in calcium fluoride. Can J Phys 54: 753–756, 1975.
138. T Szczurek, M Schlesinger. 4f → 5d transition studies of Ho3⫹ in calcium fluoride.
Phys Rev B 9: 3938–3940, 1974.
139. AA Vlasenko, LI Devyatkova, ON Ivanova, VV Mizailin, SP Chernov, T Uvarova,
BP Sobolev. Transmission spectra of single crystals of the type BaLn 2F 8 in a wide
spectral region (from 12 to 0,12 µm). Sov Phys Docl 30: 395–397, 1985.
140. LI Devyatkova, ON Ivanova, VV Mikhailin, SN Rudnev, BP Sobolev, TV Uranova,
SP Chernov. Experimental study of 4f-5d transitions in Ho3⫹, Er 3⫹, Tm3⫹, and Yb3⫹
in BaY 2F 8. Sov Phys Dolk 30: 687–689, 1985.
141. KM Devyatkova, ON Ivanova, KB Seiranyan, SA Tamazyan, SP Chernov. Vacuum
ultraviolet properties of a new fluoride matrix. Sov Phys Dokl 35: 40–41, 1990.
142. LI Devyatkova, ON Ivanova, VV Mikhailin, SN Rudnev, SP Chernov. High energy
4f states of Er 3⫹ and Ho3⫹ ions in fluoride crystals. Sov Phys Dolk 62: 275–276,
1985.
143. T Szczurek, M Schlesinger. Spectroscopic studies of excited Tm3⫹ ions in CaF 2
crystals. Phys Rev B34: 6109–6111, 1986.
144. M Schlesinger, T Szczurek. Spectroscopic studies of excited Tm3⫹ ions in alkaline-
earth fluorides. Phys Rev B35: 8341–8347, 1986.
145. T Szczurek, M Schlesinger. Temperature dependent rare-earth impurity-site sym-
metries in CaF 2. Phys Rev B36: 8263–8267, 1987.
146. BW Bryant. Spectra of doubly and triply ionised Ytterbium YbIII and Yb IV. J
Opt Soc Am 55: 771–779, 1965.
147. M Schlesinger, T Szczurek, MK Wade, GWF Drake. Anomalies in the vacuum uv
absorption spectrum of Yb3⫹ in CaF 2. Phys Rev B18: 6388–6390, 1978.
148. E Loh. Ultraviolet-absorption spectra of europeum and ytterbium in alkaline earth
fluorides. Phys Rev B184: 348–352, 1969.
149. E Loh. Strong-field assignment on 4f 135d levels of Yb2⫹ in SrCl 2. Phys Rev B7:
1846–1850, 1972.
150. RT Wegh, H Donker, A Maijerink. Spin-allowed and spin-forbidden fd emission
from Er 3⫹ and LiYF 4. Phys Rev B57: R2025–R2028, 1998.
151. MB Seelbinder, JC Wright. Site selective spectroscopy of CaF 2:Ho3⫹. Phys Rev B
20:4308–4320, 1979.
152. E Sarantopoulou, YS Raptis, E Zouboulis and C Raptis. Pressure and temperature-
dependent Raman study of YLiF 4. Phys Rev B59: 4154–4162, 1999.
153. P Dorenbos, CWE van Eijk, AJJ Bos, CL Melcher. Scintillation and thermolumi-

Copyright © 2002 by Taylor & Francis

 nescence properties of Lu 2SiO 5:Ce fast scintillation crystals. J Lumin 60&61: 979–
982, 1994.
154. O Guillot-Noël, JTM de Haas, P Dorenbos, CWE van Eijk, K Krämer, HU Güdel.
Optical and scintillation properties of cerium-doped LaCl 3, LuBr 3 and LuCl 3. J
Lumin 85:21–35, 1999.
155. JC van’t Spijker, P Dorenbos, CWE van Eijk, K Krämer, HU Güdel. Optical and
scintillation properties of Ce3⫹ doped K 2 LaCl 5. J Lumin 85: 1–10, 1999.
156. CM Combes, P Dorenbos, CWE van Eijk, K Krämer, HU Güdel. Optical and scin-
tillation properties of pure and Ce3⫹-doped Cs 2LiYCl 6 and Li 3 Ycl 6:Ce3⫹ crystals.
J Lum 82: 299–305, 1999.
157. O Guillot-Noël, JC van’t Spijker, JTM de Haas, P Dorenbos, CWE van Eijk, K
Krämer, HU Güdel. Scintillation properties of RbGd 2Br 7:Ce: advances and limita-
tions. IEEE Trans Nucl Sci 46: 1274–1284, 1999.
158. EG Devitsin, N Yu Kirikova, VA Kozlov, VN Makhov, S Yu Potashov, LN Dmi-
truk, MA Terekhin, IH Munro, C Mythen, DA Shaw, KW Bell, RM Brown, PS
Flower, PW Jeffreys, JM Parker. Time-resolved studies of emission properties of
cerium-doped fluoro-hafnate glasses under VUV synchrotron radiation excitation
Nucl Instr Methods Phys Res A405: 418–422, 1998.
159. P Dorenbos, JC van’t Spijker, OWV Frijns, O Guillot-Noël, CWE van Eijk, K
Krämer, HU Güdel, A Ellens. Scintillation properties of RbGd 2Br 7:Ce3⫹ crystals;
fast efficient and high density scintillators. Nucl Instr Methods Phys Res B132:
728–731 1999.
160. CM Combes, P Dorenbos, CWE van Eijk, C Pedrini, HW Den Hartog, JY Gesland,
PA Rodnyi. Optical and scintillation properties of Ce3⫹ doped LiLuF 4 and LiLuF 4
crystals. J Lumin 71: 65–70, 1997.
161. MA Dubinskii, VV Semashko, AK Naumov, R. Yu Abdulsabirov, S Korableva.
Ce3⫹ doped colquirite. A new concept of all-solid state tunable ultraviolet laser. J
Mod Opt 40: 1–5, 1993.
162. C Combes. Scintillation properties of 6Li-based materials for thermal-neutron detec-
tion. PhD thesis, Technical University of Delft, 1999.
163. S Nicolas, Y Guyot, VV Semashko, R Yu Abdulsabirov, E Descoix, MF Joubert.
Spectroscopie de l’ion Pr 3⫹ dans LiLuF 4. Proceedings, Phenomenes Luminescents
des Materiaux Isolants, Lyon, France, 1999.
164. SA Payne, GD Wilke. Transient gratings by 4f to 5d excitation of rare earth impuri-
ties in solids. J Lumin 50: 159–168, 1991.
165. RB Barthem, R Buisson, JC Vial, H Harmand. Optical properties of Nd 3⫹ pairs in
LiYF 4-existence of a short range interaction. J Lumin 34: 295–305, 1986.
166. M Gruwe, S Tavernier. Determination of the scintillation light yield of neo-
dymium doped LaF 3 scintillator. Nucl Instr Methods Phys Res A311: 301–305,
1992.
167. P Dorenbos, JTM de Haas, CWE van Eijk. The intensity of the 173 nm emission
of LaF 3:Nd 3⫹ scintillation crystals. J Lumin 69: 229–233, 1996.
168. AN Belsky, P Chevallier, JY Gesland, N Yu Kirikova, JC Krupa, VN Machov, P
Martin, PA Orekhanov, M Queffelec. Emission properties of Nd 3⫹ in several fluo-
ride crystals. J Lumin 72/74: 146–148, 1997.
169. MA Dubinskii, R Yu Abdulsabirov, SL Korableva, AK Naumov, VV Semashko.

Copyright © 2002 by Taylor & Francis

 On the possibility of ultraviolet lasing on f-f transitions in Nd 3⫹ ion. Laser Phys
2: 239–241, 1992.
170. MA Dubinskii. Light-driven optical switch, based on excited state absorption in
activated dielectric crystals. J Mod Opt 38: 2323–2326, 1991.
171. WM Yen, PM Selzer. Laser Spectroscopy of Solids. Berlin: Springer Verlag, 1981.
172. RT Wegh, H Donker, A Meijerink. Spin allowed and spin-forbidden fd emission
from Er 3⫹ in LiYF 4. Phys Rev B 57: R2025–R2028, 1998.
173. RT Wegh, A Meijerink. Spin-allowed and spin-forbidden 4f n → 4f n⫺15d transitions
for heavy lanthanides in fluoride hosts. Phys Rev B 60: 10820–10830, 1999.
174. AC Cefalas, Z Kollia, E Sarantopoulou. Vacuum Ultraviolet 4f 95d → 4f 10 intercon-
figurational transitions of Ho3⫹ ions in LiLuF 4 single crystals. J Opt Soc Am B
V16: 625–630, 1999.
175. E Sarantopoulou, Z Kollia, AC Cefalas. 4f 95d → 4f 10 spin-allowed and spin-forbid-
den transitions of Ho3⫹ ions in LiYF 4 single crystals in the vacuum ultraviolet. Opt
Commun 169: 263–274, 1999.
176. J Becker, JY Gesland, M Yu Kirikova, JC Krupa, VN Makhov, M Runne, M Quef-
felec, TV Uvarova, G Zimmerer. Fast VUV emission of rare earth ions (Nd 3⫹, Er 3⫹,
Tm3⫹ ) in wide band gap crystals. J Alloys Comp 275–277: 205–208, 1998.
177. A Meijerink, RT Wegh. VUV spectroscopy of lanthanides: extending the horizon.
Materials Sci Forum 315–317: 11–26, 1999.
178. JC Krupa, M Queffelec. UV and VUV optical excitations in wide band gap materi-
als doped with rare earth ions:4f-5d transitions. J Alloys Comp 250: 287–292, 1997.
179. P Dorenbos. The 5d4f n⫺1 → 4f n luminescence of trivalent rare earth ions in inor-
ganic crystals. Materials Sci Forum 315–317: 222–227, 1999.

Copyright © 2002 by Taylor & Francis

9
Spectroscopy of Broad-Band
UV-Emitting Materials Based on
Trivalent Rare-Earth Ions

Richard Moncorgé
Université de Caen, Caen, France

I. INTRODUCTION

There are a number of scientific and technological applications for which tunable
ultraviolet (UV) solid-state laser sources are already or will soon be well suited.
These include in environmental sciences, for the detection of pollutants and bio-
logical hazards in the atmosphere as well as measurements and control of various
climatic parameters (see Chapter 13). In medicine the applications are in ophthal-
mology and biotechnologies for which broad-band laser sources can be used for
the production of ultra-short excitation light pulses. Other important needs also
exist in the fields of photolithography (see Chapter 6); material processing with,
for example, the deposition of thin films; and in information technology, with
high-density optical storage.
For some of these, such as high-density information storage, compact size
and energy efficiency are more important than high power or energy storage.
It is thus likely that diode lasers, such as those based on III-N wide bandgap
semiconductors, will dominate these applications. For most of the other needs,
however, the laser sources based on rare-earth doped materials offer important
advantages. We will comment on just two examples.
It is important to monitor some atmospheric pollutants over broad areas,
perhaps even globally. Laser techniques such as differential absorption lidar (the
optical equivalent of radar) are well suited to this task. In this technique one
sends out light beams at two different wavelengths, one of which is absorbed by

Copyright © 2002 by Taylor & Francis

the chemical species of interest and the other is not. Each is reflected off the
ground or some other feature, or is scattered off particles in the atmosphere, such
that returning light is detected at the emitting site. The difference in signals at
the two wavelengths gives the absorption strength, and hence, the concentration
of the species of interest. If distance information is required, the travel time of
the laser pulses can provide it. Such applications require substantial pulse energ-
ies and high beam quality from a compact system with high efficiency and reli-
ability. Solid-state lasers based on rare-earth-doped materials are exceptionally
well suited to such tasks. For example, the National Aeronautics and Space Ad-
ministration (NASA) uses such systems for ozone detection, which requires UV
wavelengths in pairs such as 289 nm for absorption and 300 nm for nonabsorption
in the troposphere, or 301 and 311 nm, respectively, in the stratosphere [1]. There
has also been interest in UV solid-state lasers for the excitation of tryptophan
fluorescence to detect biological hazards in the atmosphere at a distance [2].
As the efficiency and lifetime of features in semiconductor devices continue
to improve and their size to shrink, photolithography at shorter wavelengths is
also becoming very important [3]. Considerable effort has already been put into
193 nm technology [4], with the expectation of using XeF excimer lasers. How-
ever, the greater reliability and potentially lower life-cycle costs of solid-state
lasers have attracted interest in the development of rare-earth-doped crystalline
systems to reach the same wavelength [5]. This approach combines an Nd laser,
frequency multiplication, and optical parametric oscillator (OPO) technology to
achieve the desired wavelength and beam quality [6]. It is believed that the requi-
site average power can be achieved. This is an application potentially of very
great importance.

II. STATE OF THE ART FOR BROAD-BAND UV LASER

SYSTEMS

Most of the tunable laser sources operating in the near UV that are now commer-
cially available remain either very complicated and/or very expensive. As men-
tioned above, these are based on frequency upconversion: frequency-doubling,
tripling, and/or mixing in nonlinear crystals of tunable visible or near-infrared
laser sources. These sources can be dye lasers pumped by frequency-doubled or
tripled Nd:YAG lasers, which have good efficiencies but pose those problems
of maintenance, lifetime, dimensions, and toxicity. They can be solid-state Ti-
sapphire or Cr: LiSAF lasers themselves, pumped by flash-lamps, frequency-
doubled Nd :YAG, or laser diodes. Finally, they can be OPOs based on nonlinear
crystals such as beta-baryum-borate (BBO). However, in this case, one comes
up against a problem of cost, since visible OPOs necessitate high-beam-quality
frequency-tripled Nd:YAG laser sources, which often means injection by diode-

Copyright © 2002 by Taylor & Francis

pumped minilasers. The global optical conversion efficiency does not exceed a
few percent and the quality of the resulting laser beam is often very disappointing.
In addition, there are problems of maintenance and dimensions and, more scien-
tifically, of wavelength adjustment and versatility, and of spectral bandwidth.
For these reasons, the most attractive solution for obtaining a tunable UV
laser source that is efficient, reliable, compact, and of reasonable cost is to take
advantage of the 5d → 4f interconfigurational broad-band laser transitions of
rare-earth ions such as Ce 3⫹ in crystals.
This also appears to be the only way to obtain a really good-quality laser
beam, controlled adjustment of the spectral width of the laser emission (which
is useful for some application), and a multiwavelength operation from only one
laser oscillator.
However, because of serious problems of solarization, due to the formation
of transitory or permanent color centers consecutive to the absorption of UV
excitation and laser radiation in the rare-earth-ion-emitting state, this type of
solid-state laser has been discarded for many years. Only recently, with the dis-
covery [7] of the Ce 3⫹ doped LiCaAlF 6 (LiCAF) laser crystal (see Chapter 11),
has this type of laser really been understood. It was demonstrated, for example,
that starting from a careful spectroscopic analysis (that of the excited state absorp-
tion as a function of polarization) one could deduce optimal operating conditions
and obtain a spectacular improvement in the laser performance of the materials
[8,9]. It was even shown that solarization could be reduced and laser efficiencies
increased by pumping crystals with the frequency-quadrupled radiation of a
pulsed Nd :YAG laser, and exposing them simultaneously to residual green light
resulting from frequency-doubling of the same pump laser. The laser conversion
efficiency of this type of material now reaches nearly 30–40 % with respect to
the UV pump radiation, and about 10% with respect to the infrared (IR) radiation
at 1.064 µm of the Nd:YAG pump laser, which is quite good. The tunability
domain is moderately wide (285–310 nm) but it already allows several applica-
tions and a complete commercial system is now available from Lambda-Physics
[10]. It includes a 1 kHz, frequency-quadrupled diode-pumped Nd: YAG (Star
Line series) pumping a Ce :LiCAF laser that delivers 500 mW average power at
gain maximum.
Research is now very active in this field, including investigations of several
other materials doped with Ce 3⫹ and other rare-earth ions such as Pr 3⫹ and Nd 3⫹,
which present similar broad-band UV emissions in complementary wavelength
domains. For example, Ce :LiLuF 4 is now competing seriously with Ce: LiCAF,
since the former recently proved to have exceptional laser efficiencies (around
45–50%) and a very interesting wavelength tunability extending from 305 up to
335 nm [11–13]. In the case of the other ions, however, the situation seems less
favorable since, up to now, only one really led to laser action: Nd :LaF 3 [14,15].
This crystal was pumped directly at 157 nm with the aid of an F 2 molecular laser

Copyright © 2002 by Taylor & Francis

into the first absorption band of the 4f 25d-excited electronic configuration of the
Nd 3⫹ ions. Laser action occurred at 172 nm (4f 25d → 4f 3 transition), which
makes Nd :LaF 3 the solid-state laser with the shortest laser wavelength demon-
strated so far. Attempts have been made with several Pr 3⫹-doped materials, by
using pump-probe techniques, to detect gain at particular emission wavelengths
[16]. None of these attempts have been successful [16,17]. However, considering
the complexity of the problem and referring to the case of Ce 3⫹, this does not
mean that it cannot be resolved. Materials doped with ions such as Pr 3⫹, Nd 3⫹,
or Tm 3⫹ might be more interesting, in the end, than Ce 3⫹-doped ones, because
of their energy level structures (see in Fig. 1). Indeed, they are characterized not
only by 4f ↔ 5d interconfigurational optical transitions in the UV domain but
also by 4f ↔ 4f intraconfigurational transitions in the visible. This means that
their UV emissions can be excited via multiple absorptions of visible photons
within their 4f energy levels. This mechanism has only been studied so far as an
optical loss for laser transitions in the visible and the near-infrared between the
4f energy levels. It has not yet been really studied as a population mechanism
of the UV emission bands in view of their potentials as tunable laser sources.
Such a mechanism, however, could be very interesting, since multistep excitation

Figure 1 Approximate positions of 4f n and 4f n⫺15d energy levels Ce 3⫹, Pr 3⫹, Nd 3⫹, and
Tm 3⫹ in YLiF 4 .

Copyright © 2002 by Taylor & Francis

with the aid of visible and, eventually, infrared photons would likely reduce the
solarization effects associated with a UV pumping light.
The following sections examine various mechanisms that can be involved
in the excitation and emission processes of these rare-earth doped materials.
Based on our present understanding of the question, we then derive some guide-
lines for future work, with particular emphasis on the spectroscopy and dynamics
of materials doped with Pr 3⫹.

III. INTEREST AND LIMITATIONS OF Pr 3ⴙ , Nd 3ⴙ , Er 3ⴙ , AND

Tm 3ⴙ COMPARED TO Ce 3ⴙ

The positions of the 4f n ↔ 4f n⫺15d absorption and emission transitions do not

only depend on the type of rare-earth ion to be considered, but also on the host
matrix. This is due to the strong crystal field effect on the 5d outer-shell orbital,
which splits into various components depending on the symmetry of the crystal-
line environment around the rare-earth active ion. This can be used to advantage
since emissions can occur in very different, and thus complementary, spectral
domain. This is shown in Figure 2 for Ce 3⫹ and Pr 3⫹ in various materials.
A strong crystal field may also lead to strong electron–phonon interaction
and to increased Stokes shifts between the potential curves associated with energy

Figure 2 Emission domains spanned by various Ce 3⫹ and Pr 3⫹-doped materials.

Copyright © 2002 by Taylor & Francis

levels of various electronic configuration. It can thus lead to increased nonradia-
tive relaxations, which has been studied in great detail by researchers investigat-
ing these materials for phosphor and scintillator applications.
Depending on the host materials, sharp line emissions may occur rather
than broad bands. Strong ligand-to-metal charge transfer absorption transitions
(LMCT) may overlap with the 4f n⫺15d ↔ 4f n interconfigurational absorption and
emission transitions.
Even in the case of strong 4f n⫺15d → 4f n emission transitions, laser gain
may not be observed because of excited-state absorption losses occuring in the
wavelength domain of excitation or of UV emission, and/or the subsequent transi-
tory or permanent formation of absorbing color centers due to so-called solariza-
tion effects.
This means that several parameters have to be considered before one may
discover a new laser material.

A. Position and Nature of the 4f nⴚ1 5d-Emitting Level

As seen in Figures 1 and 2, in the same crystal host, the Pr 3⫹ emission bands lie
at higher energies (shorter wavelengths) than the Ce 3⫹ ones. They are shifted
from each other by some 12,000 cm ⫺1. This agrees with numerous previous esti-
mations [18]. For example, Pr:YLF (YLiF 4) emission starts around 220 nm,
while Ce: YLF starts around 300 nm, which leads to a difference in energy of
12,100 cm ⫺1. In the case of YAP (YAlO 3) and YAG (Y 3 Al 5 O 12), which give
very distinct emission bands, they are found as 11,500 and 12,250 cm ⫺1, respec-
tively. This also agrees with the observation reported recently by Dorenbos [19]
that knowledge of the position of the lowest energy level of the 4f n⫺15d configu-
ration of a particular rare-earth ion in a specific host allows one to predict the
position of the lowest energy level of the other 13 Ln 3⫹ ions in that same host.
This can be of considerable help in predicting the position of the emission
bands.
According to the positions of the free ion energy levels of the trivalent
rare-earth ions and their positions in a fluoride host such as LaF 3 , as reported by
Dieke and Crosswhite and Heaps et al. [20,21] as well as by deVries and Blasse
[18], and based on results collected in the literature of the last 20 years, in addition
to Ce 3⫹ and Pr 3⫹, only Nd 3⫹, Er 3⫹, and Tm 3⫹-doped compounds would lead to
efficient 4f n⫺15d → 4f 2 emissions in the near UV [see 22–25]. Among these,
only Tm 3⫹ would lead to emission bands in oxide hosts; Nd 3⫹ and Er 3⫹ would
only emit in wide-bandgap materials such as fluorides. In particular, Er 3⫹ and
Tm 3⫹ are the only ions having more than half-filled 4f shells (n ⫽ 11 and 12,
respectively) for which there might be no close 4f n energy levels lying below
the 4f n⫺15d-emitting one, and thus for which nonradiative quenching might be
negligible.

Copyright © 2002 by Taylor & Francis

 According again to the same literature, the emission bands of Nd 3⫹ would
start some 10,000 cm ⫺1 above those for Pr 3⫹, thus some 22,000 cm ⫺1 above those
for Ce 3⫹. In the case of Nd :YLF that would mean an emission spectrum starting
around 180 nm, which is in perfect agreement, for example, with the spectrum
reported by Thogersen et al. [23]. Using the same rule of thumb and data reported
by Yang and Deluca [26], the 4f n⫺15d → 4f n emission spectra for Er 3⫹ and Tm 3⫹
should start 8300 cm ⫺1 above those for Nd 3⫹ and 17,850 cm ⫺1 above those for
Pr 3⫹. This would be at around 158 nm in the case of YLF and 200 nm in the
case of YAG, for example. These results are in good agreement with the spectra
of Er: YLF and Tm: YLF reported by Wegh and Meijerink [25] and the spectrum
of Tm :YAG reported by Lips et al [22].
When there are no spectroscopic data available in the literature, it is worth
noting that a rough estimate of the energy of the lowest 5d absorption band can
be obtained by calculating the change of the 4f –5d centroid energy difference
from its free-ion value due to the polarizability of the ligands. This was found
for Pr 3⫹, Nd 3⫹, Er 3⫹, and Tm 3⫹ to be roughly equal to the amount of energy by
which the crystal field depresses the 4f n⫺15d configuration. The procedure was
introduced by Morrison [27], based on a Judd suggestion and used, for example,
by Aull and Jenssen [17] and Merlde et al [28]. According to these authors, the
4f –5d centroid energy difference is given by

(E 5d ⫺ E 4f ) solid ≅ (E 5d ⫺ E 4 f ) free–ion ⫺ σ 2 冱 (α e /R )
i
i
2 6
i (1)

where α i is the polarizability of the i th ligand, R i is the RE–ligand distance, and

σ 2 is given by

σ 2 ⫽ 〈r 2〉 5d ⫺ 〈r 2〉 4f (2)

that is, the difference in expectation values of r 2 between the 5d and 4f wave
functions.
By treating σ 2 as an adjustable parameter, it was found to be a constant if
slightly decreasing value (lanthanide contraction effect) ranging from 2.3 to 1.9
(Å)2 for Nd 3⫹, Er 3⫹, and Tm 3⫹-doped fluorides such as YLiF 4 and LaF 3 , and about
3.1 (Å)2 for Pr 3⫹-doped oxides such as YAG, YAP, and SAM (SrAl 12 O 19). These
values assume polarizabilities of about 1 and 3.8 (Å) 3 for fluorine and oxygen
ions, respectively. Using, for example, the calculated and measured 4f 2 –4f 5d
energy differences for the free Pr 3⫹ ion and for Pr:YLiF 4 (i.e., 61,170 cm ⫺1 [29]
and 45,450 cm ⫺1, respectively), a σ 2 value of 2.3 is found by using the same
other parameters used by Morrison [27] and, thus in good agreement with the
above values. This would mean, for example, that a σ 2 value of about 3 should
be used for Tm 3⫹-doped YAG. According to Morrison [27] and Merkle et al.
[28], it would lead to a 4f 12 ⫺ 4f 115d minimum energy gap of about 49,500 cm ⫺1,

Copyright © 2002 by Taylor & Francis

which is in perfect agreement with the onset of the 4f 115d → 4f 12 UV fluorescence
observed in this material around 200 nm and reported by Lips et al. [22].
For the more compact (tighter) octahedral complexes such as elpasolite
Cs 2 NaYF 6 and borate ScBO 3 , however, the situation is more complex. This is
due to a stronger crystal field as well as more significant metal–ligand covalency
effects. It partially results in greater effective ligand polarizabilities, which further
shrink the 4f –5d centroı̈d energy difference. In the case of the Ce :Rb 2 NaYF 6
fluoro-elpasolite, for example, an effective fluoride polarizability α ⫽ 1.83 (Å)3
was reported [17] and the value σ 2 ⫽ 1.25. These values give, according to Eqs.
(1) and (2), a 4f –5d centroı̈d shift of 14,000 cm ⫺1. With a free-ion centroı̈d differ-
ence of 50,000 cm ⫺1, a 5d crystal-field splitting of the order of 18,000 cm ⫺1, and
a 4f spin-orbit splitting of 2000 cm ⫺1, this locates the onset of UV emission
around 350 nm, which is in good agreement with the so-called blue-site emission
reported by Aull and Jenssen [17]. In the case of the chloro-elpasolites [30–32],
because of the greater polarizability of the chlorine ions (α ⬇ 3.6 according to
[33]), the 4f –5d centroı̈d shift should be even larger than in fluorides and lead
to much lower emission bands. This is not as dramatic, however, because of the
smaller crystal field experienced by the rare-earth dopant. The onset of the emis-
sion band in a Ce 3⫹-doped chloro-elpasolite such as Cs 2 NaYCl 6 (onset at about
360 nm [30–32]) is only shifted down by about 800 cm ⫺1.
It is worth noting that for materials with the same type of ligands, oxygen
O 2⫺ for oxides and fluor F ⫺ for fluorides, for example, the higher the coordination
number (CN) the longer the metal–ligand distances and the weaker the local
crystal field [34]. With CN ⫽ 8 for YAG and CN ⫽ 12 for YAP, the crystal
field is stronger in the case of YAG. It has an emission spectrum starting at much
lower energy than that of YAP, because of a larger crystal field splitting of the
5d bands. The same type of argument is applicable in the case of materials with
different types of ligands but the same coordination number, such as YAP and
YLF (or BYF) for both of which CN ⫽ 8. In this case, the higher the ligand
valency the stronger the crystal field and the lower the emitting state.
The question then arises as to the nature of the lowest energy-emitting level
of the 4f n⫺15d electronic configuration. This is of great importance, since it gov-
erns the radiative emission probability of the emission transition and thus the
value of the stimulated emission cross section. To discuss that question we refer
to two very recent works, by Wegh and Meijerink [25] and by Laroche et al.,
[35] which are discussed in more detail below. According to these authors, the
case of the more than half-filled 4f shell ions such as Er 3⫹ and Tm 3⫹ might be
very different from that of the less than half-filled 4f shell ions, such as Ce 3⫹,
Pr 3⫹, and Nd 3⫹. Indeed, because of Hund’s rule, if spin still can be considered
as a good quantum number (which is not proven), the high-spin states of the
ground and first-excited configurations 4f n and 4f n⫺15d of more than half-filled
4f shell ions have the lowest energies but their spins differ by unity. Moreover,

Copyright © 2002 by Taylor & Francis

the high-spin state in the first excited configuration could lie just below a low-
spin one. For Er 3⫹, for example, the high-spin ground state is a quartet (4 I 15/2)
whereas in the 4f 105d-excited configuration the lowest energy level is likely to
come from a high-spin sextet. The next higher-lying one could be again a quartet.
This means that all the emitting transitions from the sextet energy level to the
lower energy levels of the 4f 11 ground state configuration (levels 4 I 15/2, 4 I 13/2, etc.
. . .) will have a spin-forbidden character. All the emitting transitions coming
from the above-lying quartet (provided that radiative probability competes favor-
ably with nonradiative decay) are spin-allowed. This is what was observed and
reported recently for Er 3⫹ and Tm 3⫹-doped fluorides [24,25], especially at low
temperature. The first consequence is a fluorescence lifetime of a few µs. It was
3 and 8 µs in the case of Er 3⫹ and Tm 3⫹-doped YLF, respectively, which is two
orders of magnitude longer than what is usually found with Ce 3⫹ or Pr 3⫹. This
is not good for scintillator or laser applications, since it means much smaller
stimulated emission cross sections, and thus smaller laser gains. Wegh and Meij-
erink [25] compared this to the situation encountered in organic molecules; it
occurs with the singlet and triplet states of organic dyes. The situation would be
more favorable if the energy mismatch between high and low spin-excited states
(3000 cm ⫺1 in the case of Er :YLiF 4 [25]) could be reduced to a few hundred
cm ⫺1. Thermalization could then occur at room temperature, resulting in a situa-
tion comparable to that found with the 2 E → 4 A 2 and 4 T 2 → 4 A 2 spin-forbidden
and spin-allowed transitions of Cr 3⫹ in alexandrite (BeAl 2 O 4) and other medium-
field materials [36].
In the case of the less than half-filled 4f shell ions such as Ce 3⫹, Pr 3⫹, and
3⫹
Nd , the situation is different. Assuming, as previously, that LS coupling holds
and that the 4f –5d electrostatic interaction dominates over spin-orbit coupling,
the ground and first-excited electronic configurations would have the same high-
spin states: doublets, triplets, and quartets in the case of Ce 3⫹, Pr 3⫹, and Nd 3⫹,
respectively. As a consequence, for all three ions, the 4f n⫺15d → 4f n emission
transitions should be parity as well as spin-allowed transitions. Although the real
situation is likely to be more complex, at least in the case of Pr 3⫹ and Nd 3⫹, as was
proved recently with Pr:YLiF 4 [35], this is consistent with measured fluorescence
lifetimes, which usually range between 10 and 40 ns, depending on the ion and
the crystal host. For fluorescence quantum efficiencies nearly equal to one, such
fluorescence lifetimes also mean stimulated emission cross sections on the order
of 10 ⫺18 cm 2.
This is illustrated in Figure 3, in which we have reported the stimulated
emission spectra of Pr 3⫹-doped LiYF 4 , BaY 2 F 8 , KY 3 F 10 , and Cs 2 NaYCl 6 . It also
accounts for the various wavelength domains and emission band shapes that can
be observed by changing host materials. In the case of Pr 3⫹ (and the same observa-
tion could be made with Nd 3⫹), the most important emission peaks correspond
to the lowest Stark levels 3 H 4,5,6 and 3 F 2,3,4 of the 4f 2 ground-state configuration

Copyright © 2002 by Taylor & Francis

Figure 3 Stimulated emission spectra of some Pr 3⫹-doped crystals.

(see Fig. 1). These levels are more or less split by the crystal field and the spin-
orbit interaction, and the transitions are more or less sensitive to electron–phonon
coupling. In the case of Ce 3⫹, observed emission peaks come from transitions to
the two spin-orbit levels 2 F 5/2 and 2 F 7/2 that are split and eventually mixed by
the local crystal field, depending on its characteristics (strength and degree of
distortion).

B. Fluorescence Quantum Efficiency

Fluorescence quantum efficiency is a very important parameter that might depend
on various nonradiative channels. An extensive literature has been produced on

Copyright © 2002 by Taylor & Francis

the subject. However, the situation is far from satisfactory, probably because the
measurement of absolute fluorescence quantum efficiencies is not simple, and
interpretation of the observed phenomena varies according to the type of host.
For example, in the case of Pr 3⫹-doped materials, the authors often refer
to work by Piper et al. [37], according to which the fluorescence quantum effi-
ciencies of the 4f 25d → 4f 2 UV emissions of Pr 3⫹-doped KYF 4 , KY 3 F 10 , BaY 2 F 8 ,
and YPO 4 would be considerably lower than that of Pr 3⫹ : YLiF 4 . However, noth-
ing has been said about the quality of the samples, so it would be educative to
do these measurements again with high-quality and well-characterized samples.
It was demonstrated in our laboratory, and also reported by Dubinskii et al. [38],
that Ce 3⫹ :LiLuF 4 was much more fluorescent than Ce 3⫹ :YLiF 4 . We do not know
yet for sure if this is due to the quality of the crystals or to more physical reasons,
such as the smaller size and the larger weight of Lu 3⫹ compared to that of Y 3⫹.
It may also be related to excitation or relaxation processes involving the top of
the valence band, with its 4f(Lu 3⫹) character on the one hand and its 2p(F ⫺)
character on the other.
Three types of mechanisms are usually involved in the nonradiative relax-
ation of the 4f n⫺15d excited configuration and subsequent reduction of fluores-
cence quantum efficiency (see in [39,40], for example). There are photoionization
(or metal → ligand charge/electron transfer), when the lowest 4f n⫺15d levels lie
just inside or just below the conduction band (CB) of the host material; absorption
by overlapping charge transfer (ligand → metal charge/electron transfer) bands;
and 4f n⫺15d → 4f n intersystem crossing and/or multiphonon relaxation.
Of course, the former cause of nonradiative quenching can be avoided eas-
ily by working with host materials such as fluorides, which only absorb at high
energy in the VUV spectral domain. However, for different reasons, including
material fabrication and robustness, oxides may be more suitable. This is because,
despite their lower optical absorption edge, many produce very efficient UV lumi-
nescence. Ce 3⫹ and Pr 3⫹ :YAG [41,42], with an intrinsic absorption edge around
52,000 cm ⫺1 and 5d and 4f5d lowest levels around 20,500 and 32,000 cm ⫺1,
respectively, are good examples.
In many materials, however, it is clear that direct photoionization will be
more of a problem with Pr 3⫹ and Tm 3⫹ than with Ce 3⫹, because of higher-energy
4f n⫺15d absorption bands. The other ions of interest here (i.e., Nd 3⫹ and Er 3⫹),
never led to 4f n⫺15d UV luminescence other than in a fluoride crystal. Even
though photoionization prevents UV luminescence in Ce 3⫹-doped materials such
as nitrates, vanadates, and carboxylates with absorption edges below 40,000
cm ⫺1, many other Ce 3⫹-doped oxides, borates, phosphates, silicates, and alumi-
nates with absorption edges above 50,000 cm ⫺1 give intense luminescences. Ac-
cording to Yen et al. [43], the crossover from nonluminescent to luminescent
Ce 3⫹-doped systems would occur for host materials with intrinsic absorption en-
ergy ranging between 45,000 and 50,000 cm ⫺1. In the case of Pr 3⫹, the number

Copyright © 2002 by Taylor & Francis

of UV-emitting oxide materials is not only more restricted but the interpretation
of the quenching mechanism may be more subtle.
The case of Pr 3⫹-doped (Sc, Lu, Y)2 O 3 , discussed by Blasse et al. [39] and
Aumuller et al. [44], is educative. The fast nonradiative 3 P 0 → 1 D 2 relaxation as
well as the 4f 5d UV luminescence quenching observed in these materials are
accounted for by 4f 5d–4f 2 intersystem crossing. The explanation is based on the
fact that the Pr 3⫹ ion substitutes for smaller-size cations, such that the Pr-O dis-
tance is usually short and the covalent character of the metal–ligand bonding is
increased. This causes the 4f 5d configuration to shift to lower energies, close to
the 3 P 0 level of the underlying 4f 2 configuration, together with a strong crystal
field splitting. This is in accordance with the observations made in the Pr 3⫹-doped
(Sc, Lu, Y, La)BO 3 series [45]. The 4f5d lowest emitting state was observed to
shift to lower energies with decreasing substituted ion radius (Pr 3⫹ : 0.99 Å, Sc 3⫹ :
0.74 Å, Lu 3⫹ :0.86 Å, Y 3⫹ :0.90 Å, for coordination 6 [46]), an effect sometimes
termed nephelauxetic [47]. There is, however, some disagreement as to the effect
of the increased metal–ligand covalent character resulting from tighter metal–
ligand distances, and thus from a stiffer crystalline environment, on the Stokes
shift. Usually in ionic crystals, the tighter the crystal environment, the smaller
the Stokes shift, which is in good agreement with observations reported [45]
concerning Pr 3⫹-doped borates. DeMello Donega and Colleagues [40], however,
argue that the fast 3 P 0 → 1 D 2 relaxation observed even at low temperature in the
Pr 3⫹ :Ln 2 O 3 series could not be accounted for. In case of increased overlap be-
tween the metal and the ligand wavefunctions, the 4f5d state would acquire some
charge-transfer character. Thus as the 5d wavefunction becomes more delocal-
ized, the 4f5d state would be more contracted (greater force constant leading to
a steeper potential curve), thus resulting in a larger Stokes shift. This apparent
discrepancy was already noted in the case of the Ce 3⫹-doped fluoro-elpasolites
[17] mentioned above, since only a relatively strong covalent character (compared
to the other strongly ionic fluorides) could account for both a 4f –5d band at low
energy (strong crystal field) and a large Stokes shift.
Some attention can be paid to LMCTs such as F ⫺ → Ln 3⫹ and O 2⫺ →
Ln in fluoride and oxide host lattices containing Ln 3⫹ trivalent lanthanides.
3⫹

According to Jorgensen [47], the position of an LMCT transition can be estimated

by :

E CT ⬇ 3 ⫻ 10 4 [χ(L) ⫺ χ(M)] cm ⫺1 (3)

where χ stands for the electronegativities of the considered ligand and metal ions.
According to work by Jorgensen, van Vught et al., Ropp and Carrol, and Gérard
et al. [47–50],

χ(C1 ⫺) ⬇ 3, χ(O 2⫺) ⬇ 3.2, χ(F ⫺) ⬇ 3.9

Copyright © 2002 by Taylor & Francis

and
χ(Ce 3⫹) ⬇ 2, and χ(Pr 3⫹) ⬇ 1.1
so that it can be found that
E CT (F ⫺ → Ce 3⫹) ⬇ 57,000 cm ⫺1 and E CT (F ⫺ → Pr 3⫹) ⬇ 84,000 cm ⫺1
and
E CT (O 2⫺ → Ce 3⫹) ⬇ 36,000 cm ⫺1 and E CT (O 2⫺ → Pr 3⫹) ⬇ 63,000 cm ⫺1
which also means
E CT (fluorides) ⬇ E CT (oxides) ⫹ 21,000 cm ⫺1
Experimentally, it has been found, for example, that
E CT (YLF) ⬍ E CT (YPO 4 , YA1O 3 , YBO 3) ⬍ E CT (YAG)
Based on Figures 2 and 3, it is worth noting here that LMCT transitions
should not be a problem for any Ce 3⫹ or Pr 3⫹-doped materials of good quality.
There might be some problem, however, if oxygen traces are present, for exam-
ple, in Ce 3⫹-doped fluoride materials (which is a topic often discussed in the
literature). Indeed, in this hypothesis, the Ce 3⫹ emission bands of fluorides such
as (La,Lu)F 3 might overlap with a O 2⫺ → Ce 3⫹ LMCT absorption band located
around 275 nm.
To summarize, it is likely that photoionization and intersystem crossing
are more serious problems than LMCT absorption, especially in the case of the
Pr 3⫹-doped oxide materials. Among all the emitting ones, it is clear that those
with the higher optical absorption edge, such as Pr 3⫹ :YAlO 3 and Pr 3⫹ :(Sc, Y,
Lu)BO 3 , might be the best candidates for UV laser operation. The other Pr 3⫹-
doped materials are subject either to nonradiative relaxations or excited-state ab-
sorption (ESA).

IV. 4 f 2 → 4 f 5d ESA TRANSITIONS AND 4 f 5d ENERGY

LEVELS OF Pr 3ⴙ

Because of the spin-forbidden character of emission transitions, there seems to

be little hope of finding a UV or VUV laser system based on Er 3⫹ or Tm 3⫹-doped
materials. Apart from Ce 3⫹, which will not be discussed here, and for Nd 3⫹, which
already lased in LaF 3 , only Pr 3⫹ may lead to laser action and probably also in a
fluoride host.
For that reason we will concentrate on Pr 3⫹ by presenting the results of a
recent experimental and theoretical analysis [51,52] evaluating the possibility of

Copyright © 2002 by Taylor & Francis

Figure 4 Energy levels and optical transitions involved in Pr 3⫹-doped materials.

pumping the 4f5d UV emission band with the aid of two photons (see Fig. 4).
One would bring the Pr 3⫹ ion into its 3 P 0 or 1 D 2 excited state. Another photon
would bring the system from this excited level into the 4f5d absorption band,
following work done in the past with Pr 3⫹ : YAG [53,54].
The materials used for these investigations included Perovskite Pr:YAlO 3 ,
because good-quality single crystals with various dopant concentrations were
available. Among the oxide systems, Pr:YAlO 3 is presently the most serious
candidate material for UV laser emission. Also used were the fluorides Pr:YLiF 4 ,
LiLuF 4 , BaY 2 F 8 , KY 3 F 10 , and KYF 4 , all grown in our laboratory. Although few
data can be found on Pr:KY 3 F 10 and KYF 4 , most of the basic optical properties
of the other selected materials have been reported in the literature.
Positions of energy levels of the 4f 2 ground configuration for Pr 3⫹-doped
yttrium orthoaluminate (YAlO 3) are available [55]. The emission spectrum of the

Copyright © 2002 by Taylor & Francis

4f 5d first excited configuration exhibits two broad bands at room temperature
centered at 247 and 282 nm [56,57]. Their excitation spectrum exhibits, at energ-
ies below the band gap energy (8 eV [58]), two groups of bands located around
46,080 and 55,550 cm ⫺1, respectively, assigned to the 4f 2 → 4f 5d transitions
[56].
The basic luminescence properties of Pr:YLiF 4 (4f 2 energy level positions
and intraconfigurational transition intensities) can be found elsewhere [59–61].
Unpolarized data on 4f 2-4f 5d optical transitions were reported by Lawson and
Payne [16] and Piper et al. [37]. In the case of Pr:BaY 2 F 8 , 4f 2 luminescence
and stimulated emission data have been reported [62,63], as have 4f 2-4f 5d unpo-
larized spectra [16,64]. Information on Pr: KY 3 F 10 , is also available [51,65].
In the case of Pr: KYF 4 , no information is available on the positions of the
4f 2 energy levels, and on its optical transitions between 4f 2 and 4f 5d configura-
tions.
We report in Table 1 the positions of 1 D 2 and 3 P 0 energy levels and their
fluorescence lifetimes measured at room temperature. Based on these data, ab-
sorption in the spin-triplet state is more efficient (spin-allowed transition) but
more energy can be stored into the singlet. Both metastable levels can be interest-
ing as intermediate states in the two-step pumping process mentioned above.

A. Excited-State Absorption Data

Excited-state absorption (ESA) measurements were performed using a pump-
probe experimental set-up previously described [51]. The pump source was pro-
vided by a broad-band optical parametric oscillator (OPO GWU model C355)
widely tunable in the visible and infrared domains and delivering pump pulses
of about 10 mJ with a pulse duration of 10 ns at a repetition rate of 10 Hz. The
probe beam was a continuous wave (CW) high pressure Xe arc-lamp (Osram
100 W). The probe light propagated colinearly to the pump beam through the
sample, but in the opposite direction. It was collimated and focused successively
onto the sample and the entrance slit of a monochromator with the use of concave
mirrors to avoid chromatic aberrations.

Table 1 Approximate Positions and Fluorescence Lifetimes of 1 D 2 and 3 P 0

Metastable Levels in Pr-Doped YAlO 3 , KY 3 F 10 , YLiF 4 , BaY 2 F 8 , and KYF 4

Materials Pr: YAlO 3 Pr: KY 3 F 10 Pr: LiYF 4 Pr: BaY 2 F 8 Pr:KYF 4

ν(3 P 0) 20,410 cm ⫺1 20,730 cm ⫺1 20,860 cm ⫺1 20,840 cm ⫺1 21,000 cm ⫺1

τ(3 P 0) 11 µs 33.5µs 43.5 µs 42.5 µs 63 µs
ν(1 D 2) 16,380 cm ⫺1 16,670 cm ⫺1 16,740 cm ⫺1 16,650 cm ⫺1 17,540 cm ⫺1
τ (1 D 2) 165 µs 92.5 µs 205 µs 175 µs 300 µs

Copyright © 2002 by Taylor & Francis

 Polarized and calibrated ESA spectra could be recorded from 450 to 220
nm (22,000–45,000 cm ⫺1). Those obtained in the case of Pr:YLiF 4 are shown
in Figures 5a and 6a. Those for the other materials can be found elsewhere
[51,52]. These spectra are reported in units of wavenumbers (cm ⫺1). They can
be shifted by the energies of the respective 3 P 0 and 1 D 2-absorbing excited levels
(see Table 1) to yield the positions of the involved 4f 5d levels :

ν(4f 5d) ⫽ ν(3 P 0) ⫹ ν(3 P 0 → 4f 5d)

(4)
ν(4f 5d) ⫽ ν(1 D 2) ⫹ ν(1 D 2 → 4f 5d)

As can be seen in Figures 5 and 6, ESA cross sections are on the order of
10 ⫺18 cm 2, which is typical for 4f-5d parity-allowed electric dipole transitions
(see, for example, the absorption cross section of Ce 3⫹ ions in LiCa(Sr)AlF 6 [8]).
Inspection of these 3 P J → 4f 5d and 1 D 2 → 4f 5d ESA bands and compari-
son, especially in the case of Pr:YLiF 4 , with the corresponding 3 H 4 → 4f 5d
ground-state absorption (GSA) or excitation spectra reported elsewhere [51,52]
leads to the following conclusions:

ESA spectra strongly depend on polarization.

ESA spectra from the 3 P J and 1 D 2 levels strongly differ from each other.
4f 2(3 P J, 1 D 2) → 4f 5d ESA spectra are much more structured than 4f 2(3 H 4)
→ 4f 5d GSA spectra.

The first observation is to be expected but this is the first time it has been
demonstrated. The usually reported 4f 2 → 4f 5d UV and vacuum ultraviolet
(VUV) spectra are not polarized, for technical reasons, and their spectral resolu-
tion is poor. For that reason, 4f 2 → 4f 5d ESA spectra, which are recorded in
the near-UV domain, should supply more information on the structure of the
4f 5d levels. They show first that the 4f 5d level structure is much more compli-
cated than the usually reported 5d splitting into e g and t 2g levels, for example, in
the case of cubic crystal-field symmetry, and that these observed 4f 5d levels
should not be interpreted solely on the basis of the local crystal-field environment.
They should also most likely be interpreted in terms of electrostatic interaction
and spin-orbit coupling. On the other hand, these levels should not be interpreted
considering electrostatic interaction as the major source of splitting, as for the
free-ion, and then selecting the only levels for which selection rule ∆S ⫽ 0 is
verified [68,69]. A better method, as suggested previously [70] for the 4f 65d
configuration of Eu 2⫹ and the 4f 75d configuration of Tb 3⫹ [71], would be to con-
sider the 4f n⫺1 electron core and 5d electron as separate systems with internal
interactions (electrostatic ⫹ spin-orbit interactions for the former, and crystal-
field for the latter) that exceed their mutual electrostatic and spin-orbit interac-
tions.

Copyright © 2002 by Taylor & Francis

 This is relatively easy to do in the case of Tb 3⫹ since the 4f 7 core leads to
the isolated orbital singlet 8 S (ground state of Gd 3⫹). In the case of Pr 3⫹, which
immediately follows the relatively trivial case of Ce 3⫹, the levels of the 4f 5d
excited configuration could be constructed from the separate states of 4f and
5d electrons (i.e., by coupling 2 F 5/2(4f ) and 2 F 7/2(4f ) states). This presupposes
negligible crystal-field splitting of these 4f states (which might not be accurate
in certain strong-field systems such as the fluoro-elpasolites [17]), and the 5d
states resulting from crystal field splitting (E g and T 2g in case of T d or O h symme-
try, for example). In the case of Nd 3⫹, the levels of the first excited electronic
configuration 4f 25d could be constructed from the separate states of 4f 2 and 5d
electrons, i.e., by coupling (since we are more interested in the lower energy
levels) the states 4f 2(3 H J, 3 F J, 1 G 4) of lower energies with the 5d states resulting
from crystal-field splitting. This is not done exactly the same way as in previous
works [69]. An important question, in particular, is that concerning spin–orbit cou-
pling: Should we treat spin–orbit coupling in the 4f n⫺1 separate state before or on
the same level as the 5d crystal-field splitting and 4f –5d electrostatic interaction?
The answer was given recently in the case of the 4f 5d configuration of Pr 3⫹ [35].

B. 4 f 5d Energy Level Analysis

Regarding the energy levels of the 4f 5d electronic configuration for the free ion
[72], the energy spacings (a few thousands cm ⫺1) between the different spectral
terms resulting from the 4f-5d electrostatic interaction have comparable magni-
tude to that usually produced by the local crystal field on the 5d orbitals. It is
likely as a consequence that the electrostatic and local distortion hamiltonians
should be considered together. Without complete calculations or a knowledge of
their respective energies, we can only speculate on the nature and the symmetry
of the resulting states.
Full calculation of the 4f 5d sublevels and of the electric dipole transitions
from different states of the 4f 2 configuration towards these sublevels were re-
cently performed for Pr:YLiF 4 . In addition to very detailed polarized ESA spec-
tra (from 4f 2 states 1 D 2 and 3 P 0 ⫹ 1 I 6) recorded at low as well as room temperature
[35,51], complete ground state absorption data were available [59,66]. Calcula-
tions were made by choosing pure LS coupling wavefunctions /f 2 SLJM⬎ and /
fdLSJM⬎ and the hamiltonian H ⫽ Hee ⫹ VSO ⫹ Vcryst , sum of the electrostatic,
spin–orbit, and crystal-field hamiltonians, respectively. For the free-ion parame-
ters, Sugar’s data were used [72] and some small correction factors introduced.
The 4f crystal field parameters were those previously determined experimentally
[59]. Among the three 5d crystal-field parameters, two of them were taken as
adjustable parameters, the third one B 44(5d) being linked to the others. It was
also assumed that the ratio B kq(5d)/B kq(4f ) should be independent on q, but not

Copyright © 2002 by Taylor & Francis

on k, as assumed by Apaev et al. [73]. The 〈r〉 f d radial integral involved in the
electric dipole transition matrix elements was adjusted to yield the best overall
agreement between observed and calculated cross sections at absorption maxima.
The results of a simulation of the observed ESA spectra of Pr:YLiF 4 based
on these calculations are reported in the Figures 5b and 6b. A very good agree-
ment is thus obtained with the experimental data, at least in the wavelength do-
mains in which they have been recorded.

Figure 5 (3 P 0,1,2, 1 I 6) → 4f 5d polarized ESA spectra of Pr: YLiF 4 (a) recorded and (b)
calculated at room temperature (after M. Laroche, J. Margerie, et al.).

Copyright © 2002 by Taylor & Francis

Figure 6 1 D 2 → 4f 5d polarized ESA spectra of Pr :YLiF 4 (a) recorded and (b) calcu-
lated at room temperature.

These calculations show clearly that the main interaction within the 4f 5d
configuration is the crystal-field splitting of the 5d orbital into four levels with
crystallographic quantum numbers [74] µ d ⫽ 0, µ d ⫽ ⫾1 (doubly degenerate),
and twice µ d ⫽ 2 labeled 2s and 2a, for symmetrical and antisymmetrical, respec-
tively (or Γ 2 , Γ 3,4, Γ 2 , and Γ 1 respectively, in Bethe notation [75]). More exactly
(Fig. 7), the calculations show that the local crystal-field of symmetry S 4 in YLiF 4
creates three groups of sublevels (and not four) because /µ d ⫽ ⫾1⬎ and /µ d ⫽
2s⬎ are too close to be resolved. This is in very good agreement, for example,

Copyright © 2002 by Taylor & Francis

Figure 7 Position and overall nature of the 4f 5d energy levels in Pr: YLiF 4 .

with the three peaks observed in the ground-state absorption spectrum [67]
around 215 nm (46,500 cm ⫺1), 185 nm (54,000 cm ⫺1), and 160 nm (62500 cm ⫺1)
for transitions to states /µ d ⫽ 2a⬎, /µ d ⫽ 0⬎, and /µ d ⫽ ⫾1,2s⬎, respectively.
This splitting is very similar to that found in the case of Ce :LiLuF 4 [73].
The calculations also show that the 4f crystal-field parameters are more
than one order of magnitude smaller than the 5ds, that spin–orbit coupling of
the 5d electron is quenched, and that at least in the domain below 60,000 cm ⫺1,
which was more closely investigated via ESA, only three potentially significant
interactions remain. These three are the crystal field on the 5d electron, the elec-
trostatic interaction of the 4f and 5d electrons, and the spin–orbit coupling of the
4f electron. All of these are necessary to yield a reasonable description of the
GSA and of the polarized ESA spectra. It was also demonstrated, however, that
although the electrostatic interaction is important to account for the description
of the high-energy part of the spectra (⬎58,000 cm ⫺1), the two other interactions
were sufficient to explain roughly the two groups of bands of lower energies
ranging between 45,000 cm ⫺1 and 50,500 cm ⫺1 (222–198 nm) and between
51,000 cm ⫺1 and 58,000 cm ⫺1 (196–172 nm). Moreover, especially for the first
group, the observed splitting of about 3200 cm ⫺1 (seen more clearly on the 1 D 2 →
4f 5d ESA spectra) would be predominantly due to the spin–orbit coupling
7ξ 4f /2 of the 4f electron with j ⫽ 5/2 and 7/2 values. The difference in the
shapes of the absorption spectra, with much more complicated ones derived from
ESA than GSA, is thus explained by the percentages of pure j ⫺ j states in the
starting multiplets. For example, the only multiplet for which the j monoelectronic
quantum number has a strong preference for a particular value (either 5/2 or 7/2)

Copyright © 2002 by Taylor & Francis

is the 3 H 4 ground state, with 75% (5/2, 5/2) pure character and 22% mixed (5/2,
7/2) character.
These unexpected results should be related to the observation made by
Lawson and Payne [76], referring to other authors [77,78], following which inter-
actions between 5d and 4f electrons can be small due to a fortuitous cancellation
of the multipole–multipole Coulomb interactions and the anisotropic exchange
interactions.
At least for Pr 3⫹, the interpretation is far from that consisting of separating
the energy levels of the 4f 5d configuration in purely low- and high-spin states.

V. 4f 3 → 4f 25d ESA TRANSITIONS OF Nd 3ⴙ

As shown in Figure 1, the lowest energy levels for the first excited configuration
4f 25d of Nd 3⫹ are relatively close to the higher-lying ones 2 G 7/2,9/2 of the 4f 3
fundamental configuration. They become closer as the crystal field becomes
stronger. In the case of oxide materials, this means very reduced energy gaps,
and, due to phonon energies of the order of 600–900 cm ⫺1, this also means strong
nonradiative relaxations. On the other hand, in the case of fluorides, with energy
gaps of the order of 7500 cm ⫺1 for YLiF 4 and BaY 2 F 8 and of 14,500 cm ⫺1 for
LaF 3 and KY 3 F 10 [79], and phonon energies of the order of 450 cm ⫺1, nonradia-
tive relaxations are strongly reduced and efficient 4f 25d → 4f 3 UV emissions
occur corresponding to transitions ending on levels 4 I 9/2,11/2,13/2 (between 0 and
5000 cm ⫺1), 4 F 3/2 – 4 F 9/2 (between 11,000 and 14,500 cm ⫺1), and 4 G 5/2,7/2 – 4 G 7/2,9/2
(between 17,500 and 20,500 cm ⫺1). In the case of Nd :YLiF 4 for example, this
means three groups of bands located around 180–200 nm, 220–240 nm, and
255–280 nm [23].
Two-step and three-step excitation pumping were considered (see in Fig.
8). Three-step excitation pumping was successfully realized, for example, to ex-
cite the VUV emissions of Nd:YLiF 4 [23,79] by using the same pump photons
at 532 nm of a frequency-doubled Nd :YAG laser. The pumping sequence was
4
I 9/2 → 4 G 7/2,9/2 , 2 K 13/2 (⬃20000 cm ⫺1) → 2 F 5/2 (⬃38000 cm ⫺1) → 4f 5d

As in the case of Pr 3⫹, two pathways were considered for two-step excitation
[79]. Pumping was produced either into the 4 D 3/2 multiplet with the frequency-
tripled radiation of a Nd :YAG laser at 355 nm or into the 4 F 3/2 metastable level
around 800 nm with an OPO, a Ti:Sa, or a diode laser. Then the excited ions
were brought into the 4f 25d excited configuration bands with the aid of a second
photon. In this case, the pumping sequences were as follows:
4
I 9/2 → 4 D 3/2 (⬃28500 cm ⫺1) → 4f 2 5d
or 4
I 9/2 → 4 F 3/2 (⬃11000 cm ⫺1) → 4f 2 5d

Copyright © 2002 by Taylor & Francis

Figure 8 Energy levels and optical transitions involved in Nd 3⫹-doped materials.

Whatever the excitation process is, only a detailed knowledge of the relax-
ation times of the involved intermediate excited levels and of the ESA cross
section spectra can determine the more efficient one. Table 2 shows characteristic
lifetimes of the levels involved in some fluorides. From these values, it is clear
that three-step excitation pumping involving short-lived excited states such as
the 4 G 7/2 multiplet [80,81] is likely to be more efficient by using short, nanosec-

Copyright © 2002 by Taylor & Francis

Table 2 Lifetimes of Some Energy Levels of Nd-Doped Fluoride Crystals [80,81]

Materials Nd: LaF 3 Nd : YLiF 4 Nd : BaY 2 F 8 Nd :NaYF 4

τ(2 F 5/2) 23 µs 14.5 µs 13 µs 6µs

τ(4 G 7/2) 56 ns 8.5 ns
τ (4 D 3/2) 13.5 µs 1.3 µs 2.7µs 7µs
τ(2 P 3/2) 225 µs 35 µs 50 µs 62 µs
τ(4 F 3/2) 700 µs 550 µs 660 µs 305 µs

Source: Refs. 80, 81.

ond-type, laser excitation pulses. For two-step pumping, involving intermediate

long-lived levels with lifetimes of several microseconds, longer excitation pulses
may allow larger energy storage and larger excitation pump powers.
The preliminary experiments mentioned above [79] have shown that ESA
into the lowest-energy 4f 25d absorption bands is much more efficient after pump-
ing around 800 nm (i.e., after populating the 4 F 3/2 metastable level) than after
pumping around 355 nm in the 4 D 3/2. Indeed, ESA spectra obtained from these
two excited states exhibit cross sections on the order of 10 ⫺18 cm 2 and 10 ⫺20 cm 2,
respectively. Such a difference indicates, as in the case of Pr 3⫹, that selection
rules apply in some way and that they are substantially different, at least in the
ESA wavelength domain investigated, for levels such as 4 D 3/2 and 4 F 3/2, although
they are both spin-quartets.
As in the case of Pr 3⫹, it would be worth performing the calculations to
clarify this situation. At the moment, without adopting Szczurek and Schleising-
er’s procedure [68] in its entirety, the Russel-Saunders approximation scheme
can be used to give a crude interpretation of the above-mentioned observation.
Let us assume that the 4f 25d levels resulting from spin-orbit, electrostatic, and
crystal field interactions keep some memory of their free-ion structure and posi-
tions [82]. In this case, if we use selection rules ∆S ⫽ 0, ∆L ⫽ 0, ⫾1, ∆J ⫽ 0,
⫾1 (except 0 ↔ 0) for transitions between 2S⫹1L J levels, the 4f 3 → 4f 25d allowed
transitions of lower energy from the above considered energy levels would be
as follows:
4
I 9/2 → 4 K 11/2, 4 I 9/2, and 4 H 7/2, up to levels located at frequencies σ 0 ⫹ 500
cm ⫺1, σ 0 ⫹ 1700 cm ⫺1, and σ 0 ⫹ 2600 cm ⫺1, respectively
4
F 3/2 → 4 G 5/2 at σ 0 ⫹ 2200 cm ⫺1
4
D 3/2 → 4 F 3/2 at σ 0 ⫹ 7300 cm ⫺1
where σ 0 stands for the energy of the lowest level 2 H 9/2 of the 4f 25d configuration.
This is very crude, but begin to interpret the ESA results and the observa-
tion of strong ESA from state 4 F 3/2 and weak ESA from state 4 D 3/2 in the investi-

Copyright © 2002 by Taylor & Francis

gated wavelength domain, corresponding to the lower energy levels of the 4f 25d
electronic configuration.
Consequently, as expected, efficiency of multistep excitation of the VUV
emission bands of Nd 3⫹ should be strongly dependent on the wavelength (and
polarization) of the pumping photons.

VI. SIMULATION OF TWO-STEP EXCITATION PUMPING

AND TENTATIVE GAIN MEASUREMENTS

The different ESA spectra reported above suggest that two-step excitation, be-
cause of a second step with a very high cross section, should be an efficient
upconversion process to populate 4f 5d energy levels in the Pr 3⫹-doped fluoride
and oxide crystals.
This was demonstrated by Laroche et al. [51] in the case of Pr:KY 3 F 10 .
Two-step excitation efficiency was investigated by estimating theoretically the
single-pass gain that should be obtained after excitation with 5 ns pump pulses
at 445 and 355 nm for the first and second steps 3 H 4 → 3 P 2 and 3 P 0 → 4f 5d,
respectively (see in Fig. 4). Using the fluorescence lifetimes of 30 µs and 26 ns
of the 3 P 0 and 4f 5d-emitting states and the adequate rate equation model, 20%
gain per pass was found with pump fluences of 0.5 J/cm 2. Such a result is cer-
tainly too optimistic, considering that the simulation was made by assuming neg-
ligible losses due to ESA of the pump or of the UV-emitted photons in the 4f 5d
emitting state and/or due to solarization effects (see Sec. VII.).
Measurements of UV laser gain were also attempted in all the materials at
a wavelength of 266 nm and by pumping them, as shown in Figure 4, with two
consecutive photons at around 470 nm (to excite the 3 P 0,1,2 , 1 I 6 multiplets) and
355 nm (to reach the 4f 5d band from the 3 P 0,1,2 , 1 I 6-excited multiplet). Gain was
attempted at 266 nm because it was the only probe wavelength that was not too
difficult to produce with our Nd :YAG-pumped OPO laser (by frequency dou-
bling the residual of the 532 nm radiation). This probe was at a longer wavelength
than that already tried unsuccessfully [16]: around 225 nm in Pr:YLiF 4 and Pr:
BaY 2 F 8 (but also in Pr:CaF 2 and several other similar fluorides). Unfortunately,
either the emission cross section was too weak at this probe wavelength (in the
case of Pr: LiYF 4 or Pr: BaY 2 F 10) or this wavelength was well adapted (in the
case of Pr:KY 3 F 10 or Pr:YAlO 3), but the crystals solarized so much that it was
impossible to detect any laser gain.

VII. SOLARIZATION EFFECTS

As mentioned previously, solarization (the creation of color centers under UV

light excitation) is a critical problem in most of these UV-emitting materials.

Copyright © 2002 by Taylor & Francis

However, it has been demonstrated, at least in the case of the now famous Ce :
LiCAF and Ce:LiSAF laser systems, that these solarization effects depend not
only on the characteristics of the optical transitions such as polarization but also
on the purity and the dopant substitution process of the crystals. Concerning the
first aspect, solarization was attributed to ESA (of the pump and/or the emitted
photons) from the 5d UV-emitting state up into the conduction band of the crys-
tals) followed by trapping of the resulting free electrons by impurity traps (traps
resulting from incomplete charge compensation in the doping process, in the case
of Ce: LiCAF and Ce :LiSAF). We have not yet tried to pump our Pr doped
crystals directly into their 4f 5d band to check because the 213 nm Nd :YAG
laser pump source necessary was not available. However, we could observe some
solarization effects in the course of the two-photon excitation experiments as
well as during the tentative gain measurements reported in the previous section
[83]. These solarization effects manifest themselves by the appearance of broad
absorption bands in the blue/UV domain extending, for example, from 230 to
320 nm and from 350 to 550 nm in the case of Pr:KY 3 F 10 (see Fig. 9) and from
250 to 650 nm in the case of Pr:YAlO 3 (Fig. 10). Concerning these systems, it is
worth noting the two following points. Solarization effects in Pr:KY 3 F 10 depend
strongly on crystal quality and, in the case of Pr:YAlO 3 , on dopant concentration.
In the case of Pr: KY 3 F 10 , poor optical quality also means a larger number of

Figure 9 Color center absorption bands (dashed curve) produced in a Pr 3⫹-doped

KY 3 F 10 single crystal.

Copyright © 2002 by Taylor & Francis

Figure 10 Deconvoluted color center absorption band (UV/visible solarization) pro-
duced in a Pr 3⫹-doped YA1O 3 single crystal.

lattice defects, and therefore a larger number of potential trap centers for the free
electrons produced in the conduction band. In the case of Pr: YAlO 3 , solarization
effects could be reduced significantly by increasing the Pr dopant concentration.
This is in agreement with previous work [84] showing that the color center forma-
tion in YAlO 3 comes mainly from the formation of electron-hole pairs followed
by trapping of holes by Pr 3⫹ ions, thus by the formation of Pr 4⫹ (Pr 3⫹ ⫹ e ⫹ →
Pr 4⫹). However, the reverse electron capture process at Pr 4⫹ ions also occurs
(Pr 4⫹ ⫹ e ⫺ → Pr 3⫹). So, Pr 3⫹ ions, serving as additional traps, change the direction
of processes of charge carriers’ release and capture, which results in an increase
of the material’s stability to short wavelength radiation.
These results both mean that better-quality crystals need to be grown and
that dopant concentration can be a very important parameter.

VIII. DISCUSSION AND CONCLUSION

As mentioned above and already discussed in the cases of Ce 3⫹ [42,85–88], Pr 3⫹

[16,54] and Eu 2⫹ and Sm 2⫹ [76] doped materials, solarization effects would likely
result from ESA from the 5d emitting level to the host conduction band (CB).
According to Griffith [75], the large oscillator strength of this localized-to-delo-

Copyright © 2002 by Taylor & Francis

calized 5d → CB transition would be due to an ‘‘intensity borrowing’’from the
valence band (VB) to the conduction band (CB) transition. This would be due
to the large spatial extent of the 5d orbital and its mixing, in case of fluorides
for example, with the 2p orbitals of the F ⫺ ligands. Such an intensity borrowing
mechanism was also involved recently to account for the large oscillator strength
of the Nd 3⫹ absorption transitions found in YVO 4 [89]. It turns out that the posi-
tion and strength of this 5d → CB ESA transition depend on several parameters,
and that only the position of this band can be estimated correctly at the moment.
This can be made in two ways, depending on the available data: first via
photoionization threshold measurements and using the electrostatic model devel-
oped in the past by Pedrini et al. [90]; and, second, via XPS measurements.
Photoionization measurements give EPI, (the energy difference between the
4f n ground state of the rare-earth dopant and the conduction band [see Fig. 4 for
Pr 3⫹]) and, according to the electrostatic model, E PI can be also obtained from
the ionization potential I p of the free ion by using the relation:
E pi ⫽ I p ⫺ C (5)

where C is a correction factor due to the crystal environment. Knowing the energy
of the 5d-emitting level with respect to the 4f ground-state, defined as E df , the
5d → CB ESA threshold would occur at the energy:
E esa ⱕ E pi ⫺ E df ⫽ I p ⫺ C ⫺ E df (6)

On the other hand, XPS data give the approximative position of the 4f ground
state within the bandgap above the top of the valence band, which we denote as
E fv . In this case, the 5d → BC ESA threshold would be obtained from:
E esa ⱖ E g ⫺ E df ⫺ E fV (7)

Using the data of Pedrini et al. [90] for divalent rare-earth ions in fluorides and
assuming, according to Sugar and Reader [91], that the ionization energies for
the trivalent ions have the same relative values, it means that for Ce 3⫹, Pr 3⫹, and
Nd 3⫹, photoionization energies can be deduced from each other by using the
respective ratios 1:1.2 :1.31.
Take YLiF 4 , for example, for which the top of the valence band VB is
predominantly made up of 2p F orbitals and the bottom of the BC, of 4d Y, with
an energy gap of about 84,500 cm ⫺1 (10.1 eV). According to the XPS results of
Guillot-Noel et al. [89], E fV(Nd 3⫹) ⬇ 9700 cm ⫺1 ⬇ 1.1 eV, thus EPI(Nd 3⫹) ⬇
74,800 cm ⫺1. This means that EPI(Pr 3⫹) ⬇ 74,800 ⫻ (1.2/1.31) ⬇ 68,500 cm ⫺1
and EPI(Ce 3⫹) ⬇ 74,800 ⫻ (1/1.31) ⬇ 57,000 cm ⫺1, thus Ef V(Pr 3⫹) ⬇ 16,000
cm ⫺1 ⬇ 1.9 eV, and Ef V(Ce 3⫹) ⬇ 27,500 cm ⫺1 ⬇ 3.3 eV. These results are in
perfect agreement with the XPS measurements made recently on CeF 3 and PrF 3
crystals using synchrotron radiation [92].

Copyright © 2002 by Taylor & Francis

 The same procedure can be used in the case of YAlO 3 , for which XPS
spectra indicate Ef V values of Ef V(Pr 3⫹) ⬇ 0 and Ef v(Ce 3⫹) ⬇ 1 eV [93], the top
of the valence band being made predominantly of 2p 0 orbitals. Indeed, with an
energy gap Eg ⬇ 61,700 cm ⫺1, it is found EPI(Pr 3⫹) ⱖ 61,700 cm ⫺1 and
EPI(Ce 3⫹) ⱖ 61,700 ⫻ (1/1.2) ⬇ 51,400 cm ⫺1 thus Ef V(Ce 3⫹) ⱕ 10,300 cm 1 (1.2
eV).
Consequently, it seems reasonable to use the above Ef V values along with
the known Eg and Edf values to give estimates of the positions of the 5d → CB
ESA band thresholds, at least in fluorides and oxides. A complication may occur,
however, in the case of the Lu-based fluorides such as LiLuF 4 , since XPS mea-
surements [94] proved that the 4f orbital of Lu 3⫹ lies just above the 2p for F ⫺.
In this case, the above EfV values for fluorides would be reduced by about 1 eV,
giving values of about 0, 1, and 2 eV for Nd 3⫹, Pr 3⫹, and Ce 3⫹, respectively.
Once we are able to estimate the onset for the 5d → CB ESA transition, we are
faced with the question of the width of the CB (∆ CB). To answer that question,
knowledge of the band structure of the material is necessary, which is not always
easy to determine.
Let us see the consequences of the above considerations in the case of the
known laser materials: first Ce: LiCaAlF 6 and Ce :YLiF 4 with emissions in the
ranges 280–320 nm and 300–340 nm, respectively. With 5d metastable emitting
levels located around 280 and 300 nm (i.e., Efd ⬇ 35,700 and 33,300 cm ⫺1) and
4f ground states around 3.3 eV above the top of the valence band (i.e., EfV ⬇
27,500 cm ⫺1) ESA of the laser emissions in the 5d levels could occur between
91,700 and 98,500 cm ⫺1, in the case of Ce: LiCAF, and between 90,200 and
94,100 cm ⫺1, for Ce :YLF (i.e., between 11 and 12 eV and between 10 and 11
eV, respectively, above the top of the valence band). These energies coincide
with the edge of the VB–CB energy gaps Eg in these materials. It explains why
ESA-induced solarization effects were observed in these systems, although they
are reduced compared to other lower-energy-gap materials. This may explain
why solarization effects seem slightly more important in the case of Ce:LiSAF
than in the case of Ce: LiCAF, since optical absorption edges occur around 10.4
and 10.7 eV [95], respectively. This also agrees well with our own observation
in the case of Ce:LiLuF 4 . No color center formation was ever observed in our
samples by irradiating them at 248 nm with a KrF laser producing up to 4J/cm 2
at 100 Hz repetition rate, whereas weak and strong parasitic absorption bands
could be easily observed, in the same pumping conditions, in the case of Ce
doped YLiF 4 and LuF 3 , respectively (see Figure 11).
From these observations, it can be concluded that Ce:YLiF 4 solarizes more
efficiently than Ce :LiLuF 4 for two reasons. Band calculations [96] have shown
that the band gap energy Eg of LiLuF 4 is larger than that of LiYF 4 by a factor
of 1.2, which means some 12.5 eV values instead of 10.5 eV. Moreover, ac-
cording to XPS measurements [94], the top of the valence band of LiLuF 4 would

Copyright © 2002 by Taylor & Francis

Figure 11 Color center absorption bands produced in a Ce 3⫹-doped LuF 3 single crystal.

have a 2p F 4f Lu mixed orbital character, which means that the 4f ground state of
the Ce 3⫹ dopant ion should be closer to the top of the valence band. Both effects
thus act favorably to relegate the edge of the 5d → CB ESA to higher energies,
at least beyond the emission domain of this Ce 3⫹-doped material.
Based on these results, some prospective arguments can be derived for the
search for new Ce 3⫹ laser systems and a good Pr 3⫹ candidate.
Let us consider first the case of Pr 3⫹ in YAP (YAlO 3) and LuAP (LuAlO 3)
with Eg ⬇ 7.4 eV, in YPO 4 and LuPO 4 with Eg ⬇ 8.1 and 8.3 eV, and in YLiF 4
and LiLuF 4 with Eg ⬇ 10.5 and 12.5 eV. In the case of Pr:YAP and LuAP and
of Pr:YPO 4 and LuPO 4 , emissions occur between about 240 and 300 nm. With
a 4f ground-state located near the top of the VB, ESA in the emission domain
would thus occur between about 9 and 10 eV (i.e., directly from the 5d-emitting
level into the conduction band). In the case of Pr:LiYF 4 and LiLuF 4 , with E fV
values of about 1.9 and 0 eV, respectively, and emissions ranging between 220
and 270 nm, ESA would occur between about 10.7 and 11.7 eV, which is some-
what more favorable. Consequently, it seems difficult to find a Pr-doped material
for which ESA in the region of the emission does not fall within the conduction

Copyright © 2002 by Taylor & Francis

band. This does not mean that laser operation in these Pr-doped crystals is hope-
less. Indeed, some good conditions (CB structure, polarization of the transitions)
may be encountered for which positive gain may be obtained. Such a hope comes
from the fact, as mentioned by Lawson and Payne [16], that the ESA cross sec-
tions seem barely to exceed the emission cross sections (e.g., YLiF 4), which
means that some portion of the emission band could exhibit laser action.
In the case of Ce 3⫹ and Pr 3⫹, another alternative is to search for materials
with large bandgaps that induce large crystal-field splittings (from a strong local
crystal-field). With a Ce 3⫹ emission extending from about 320 up to 400 nm in
Rb 2 NaYF 6 , for example [17], and thus a Pr 3⫹ emission starting around 230 nm
and extending to about 300 nm, ESA in the emission domain would reach levels
located somewhere between 76,800 and 87,000 cm ⫺1 (9.2 and 11.5 eV). Elpasol-
ite systems such as (Rb,Cs,K)2 Na(Y,Sc,Lu)F 6 would be interesting to study.
The last question concerns the potentiality of new Nd 3⫹-doped materials.
In this case, ESA in the emission domain of these materials would occur, for
example, in the case of Nd:YLiF 4 (see Fig. 8), with a value E f V ⬇ 1.1 eV and
an emission domain ranging between 180 and 280 nm, between about 12 and
14.5 eV. This would thus eventually be above the CB of the material [97], in a
region where levels should be found from the 4f 26s second excited electronic
configuration, the next higher-lying one 4f 26p being located around 17.5 eV, and
thus at much higher energies [98]. However, because of the Laporte selection
rule, ESA transitions from the emitting level of the 4f 25d electronic configuration
to the next 4f 26s should not be very intense, and most probably less intense than
the parity-allowed emission itself.
In conclusion, several directions of research still need to be explored. These
include efforts to find new crystalline materials with good enough optical quality,
characterized by large bandgaps of (10–12 eV), a strong local crystal field (tight
lattices), and reduced Stokes shifts (especially in the case of Pr 3⫹) to avoid nonra-
diative quenching (stiff lattices). Much attention must be payed to the spectro-
scopic aspects, such as the structure of the conduction band of the materials and
the intensity and state of polarization of the 4f n⫺15d → CB ESA transitions, which
will require further calculations and experiments.

ACKNOWLEDGMENTS

Most of the work reported in this Chapter and quoted in Refs. 13, 28, 32, 50,
51, 78, and 81 was done under the auspices of the US Army European Research
Office (contract N68171-97-M-5764) and of CNRS (PICS ⫹ GDR Lasmat), and
with the collaboration of colleagues from the University of Caen (Prof. J.
Margerie, Dr. S. Girard, and PhD student M. Laroche), the University of Houston-
Downtown (Dr. L. D Merkle), and the Universities of Lyon and St Etienne (Drs

Copyright © 2002 by Taylor & Francis

Y. Guyot, M.F. Joubert, S. Guy, C. Pédrini, Prof. E. Descroix, and PhD student
S. Nicolas). Many thanks are expressed to all of them.

REFERENCES

1. EV Browell. In: BHT Chai, SA Payne, eds. OSA Proceedings on Advanced Solid-
State Lasers. Washington, DC: Optical Society of America, 1995, vol 24, pp 2–4.
2. J Fox, C Swim. In: BHT Chai, SA Payne, eds. OSA Proceedings on Advanced Solid-
State Lasers. Washington, DC: Optical Society of America, 1995, vol 24, pp 13–
15.
3. J Wallace. Laser Focus World 35(8): 139–142, 1999.
4. Special Report. Laser Focus World. July: 82–96, 1997.
5. RD Mead, CE Hamilton, DD Lowenthal. Proc SPIE 3051: 882, 1997.
6. AA Buj, AV Kachinsky, AV Ermakov, AS Grabchikov, IA Khodasevich, VA
Orlovich. Conference on Lasers and Electro-Optics CLEO/Europ’2000. Nice,
France, 2000.
7. MA Dubinskii, VV Semashko, AK Naumov, R Yu. Abdulsabirov, SL Korableva.
Laser Physics 3: 216–217, 1993.
8. CD Marshall, JE Speth, SA Payne, WF Krupke, GJ Quarles, V Castillo, BHT Chai.
J Opt Soc Am B11 2054–2065, 1994. AJ Bayramian, CD Marshall, JH Wu, JA
Speth, SA Payne, GJ Quarles, V Castillo. In: SA Payne, C Pollock, eds. OSA Trends
in Optics and Photonics. Washington, DC: Optical Society of America, 1: 60–65,
1996.
9. JF Pinto, L Esterowitz, GJ Quarles. Electron Lett 31: 2009–2011, 1995.
10. SV Govorkov, AO Wiessner, Th Schroder, U Stamm, W Zschocke, D Bastings.
OSA Trends Optics Photonics 19: 2–5, 1998.
11. N Sarukura, MA Dubinskii, Z Liu, V Semashko, AK Naumov, SL Korableva, RY
Abdulsabirov, K Edamatsu, Y Suzuki, T Itoh, Y Segawa. IEEE J Select Topics
Quant Electron 1: 792–804, 1995.
12. P Rambaldi, R Moncorgé, JP Wolf, C Pédrini, JY Gesland. Opt Commun 146: 163–
166, 1998. Laser Focus World 33: 26, 1997.
13. AJS McGonigle, S Girard, DW Coutts, R Moncorgé. Electron Lett 35: 1640–1641,
1999.
14. RW Waynant, Ph H Klein. Appl Phys Lett 46: 14–16, 1985.
15. MA Dubinskii, AC Cefalas, E Sarantpoulou, SM Spyrou, CA Nicolaides, R Yu.
Abdulsabirov, SL Korableva, VV Semashko. J Opt Soc Am B9: 1148–1150, 1992.
16. JK Lawson, SA Payne. Opt Mater 2: 225–232, 1993.
17. BF Aull, HP Jenssen. Phys Rev B34: 6640–6646, 1986. Phys Rev B34: 6648–6655,
1986.
18. AJ deVries, G Blasse. Mater Res Bull 21: 683–694, 1986.
19. P Dorenbos. International Conference on Luminescence, (Osaka, Japan 1999), paper
BO4–5.
20. GH Dieke, HM Crosswhite. Appl Opt 2: 675–686, 1963.
21. WS Heaps, LR Elias, WM Yen. Phys Rev B 13: 94–104, 1975.

Copyright © 2002 by Taylor & Francis

22. HV Lips, N Schwentner, G Sliwinski, K Petermann. J Appl Spectr 62: 803–814,
1995.
23. J Thogersen, JD Gill, HK Haugen. Opt Comm 132: 83–88, 1996.
24. J Becker, JY Gesland, N Yu. Kirikova, JC Krupa, VN Makhov, M Runne, M Quef-
felec, TV Uvarova, G Zimmerer. J Lumin 78: 91–96, 1998.
25. RT Wegh, A Meijerink. Phys Rev B60: 10820–10830, 1999.
26. KH Yang, JA Deluca. Phys Rev B 17: 4246–4255, 1978.
27. CA Morrison. J Chem Phys 72: 1001–1002, 1980.
28. LD Merkle, B Zandi, Y Guyot, HR Verdun, B McIntosh, BHT Chai, JB Gruber, M
Seltzer, CA Morrison, R Moncorgé. In: TY Fan, BHT Chai, eds. OSA Proceedings
on Advanced Solid State Lasers. Washington, DC: OSA, 20: 361–366, 1994.
29. J Sugar. J Opt Soc Am 55: 1058–1061; 1965. Phys Rev Lett 14: 731–732, 1965.
30. S Mroczkowski, P Dorain. J Less Common Met 110: 259–265, 1985.
31. JC van’t Spijker, P Dorenbos, CWE van Eijk, MS Wickleder, HU Gudel, PA Rodnyi.
J Lumin 72–74: 786–788, 1997.
32. M Laroche, M Bettinelli, S Girard, R Moncorgé. Chem Phys Lett 311: 167–172,
1999.
33. L Pauling. Proc Soc (London) A114: 181, 1927.
34. GA Slack, SL Dole, V Tsoukala, GS Nolas. J Opt Soc Am B11: 961–974, 1994.
35. M Laroche, JL Doualan, S Girard, J Margerie, R Moncorgé. J Opt Soc Am B17:
1291–1303, 2000.
36. JC Walling, OG Peterson, HP Jenssen, RC Morris, EW O’Dell. IEEE J Quant Elec-
tron 16: 1302–1315, 1980.
37. WW Piper, JA DeLuca, FS Ham. J Lumin 8: 344–348, 1974.
38. MA Dubinskii, VV Semashko, AK Naumov, R Yu. Abdulsabirov, SL Korableva.
Laser Physics 4: 480–484, 1994.
39. G Blasse, W Schipper, JJ Hamelink. Inorg Chim Acta 189: 77–85, 1991.
40. C De Mello Donega, A Meijerink, G Blasse. J Phys Cond Matter 4: 8889–8902;
1992. J Phys Chem Sol 56: 673–685, 1995.
41. MJ Weber. Solid Stat Commun 12: 741–744, 1973.
42. DS Hamilton, SK Gayen, GJ Pogatshnik, RD Ghen, WJ Miniscalco. Phys Rev B39:
8807–8815, 1989.
43. WM Yen, M Raukas, SA Basun, W van Schaik, U Happek. J Lumin 69: 287–294,
1996.
44. GC Aumüller, W Köstler, BC Grabmaier, R Frey. J Phys Chem Sol 55: 767–772,
1994.
45. G Blasse, JP van Vliet, JWM Verwey, R Hoogendam, M Wiegel. J Phys Chem Sol
50: 583–585, 1989.
46. RD Shannon. Acta Crystal A 32: 75, 1976.
47. CK Jorgensen. Modern Aspects of Ligand-Field Theory. Amsterdam: North-Hol-
land, 1971; R Reisfeld, CK Jorgensen. Lasers and Excited State of Rare-Earths.
Berlin: Springer, 1977.
48. N van Vugt, T Wigmans, G Blasse. J Inorg Nucl Chem 35: 2602, 1973.
49. RC Ropp, B Carrol. J Phys Chem 81: 1699–1700, 1977.
50. I Gérard, JC Krupa, E Simoni, P Martin. J Alloys Comp 207/208: 120–127, 1994.

Copyright © 2002 by Taylor & Francis

51. M Laroche, A Braud, S Girard, JL Doualan, R Moncorgé, M Thuau, LD Merkle.
J Opt Soc Am B 16: 2269–2277, 1999.
52. S Nicolas, M Laroche, S Girard, R Moncorgé, Y Guyot, MF Joubert, E Descroix,
AG Petrosyan. J Phys Cond Matter 11: 7937–7946, 1999.
53. J Ganem. WM Dennis, WM Yen. J Lumin 54: 79–87, 1992.
54. YM Cheung, SK Gayen. Phys Rev B 49: 14827–14835, 1994.
55. M Malinowski, C Garapon, MF Joubert, B Jacquier. J Phys Condens Matter 7: 199,
1995.
56. EG Gumanskaya, MV Korzhik, SA Smirnova, VB Pavienko, AA Fedorov. Opt
Spectrosc 72: 86–88, 1991.
57. C Pédrini, D Bouttet, C Dujardin, B Moine, I Dafinei, P Lecoq, M Koselja, K Blazek.
Opt Mat 3: 81–88, 1994.
58. T Tomiki, F Kaminao, M Fujisawa, Y Tanahara. J Phys Soc Japan 55: 2090; 1986.
T Tomiki, F Kaminao, M Fujisawa, Y Tanahara and T Futemna. J Lumin 40&41:
379, 1988.
59. L Esterowitz, FJ Bartoli, RE Allen, DE Wortman, CA Morrison, RP Leavitt. Phys
Rev B 19: 6442–6455, 1979.
60. GM Renfro, JC Windscheif, WA Sibley, RF Belt. J Lumin 22: 51–68, 1980.
61. JL Adam, WA Sibley, DR Gabbe. J Lumin 33: 391–407, 1985.
62. BM Antipenko. Opt Spektrosc 56: 72–77, 1984.
63. AA Kaminskii, SE Sarkisov. Phys Stat Sol 97: K163–168, 1986.
64. SP Chernov, LI Devyatkova, ON Ivanova, AA Kaminskii, VV Mikhailin, SN Rud-
nev, TV Uvarova. Phys Stat Sol 88: K169–173, 1985.
65. CA Morrison, DF Wortman, RP Leavitt, HP Jenssen. Harry Diamond Labs report,
HDL-TR 1897, 1980.
66. E Sarantopoulou, AC Cefalas, MA Dubinskii, CA Nicolaides, R Yu. Abdulsabirov,
SL Korableva, AK Naumov, VV Semashko. Opt Lett 19: 499–501, 1994.
67. JC Krupa, M Queffelec. J Alloys Comp 250: 287–292, 1997.
68. T Szczurek, M Schleisinger. In: B Jerowska-Trzebiatowska, J Legendziewicz, W
Strek, eds. Rare Earth Spectroscopy, Proceedings, International Symposium on RE
Spectroscopy, Wroclaw, Poland, Sept. 1984, pp 309–330.
69. Z Kollia, E Sarantopoulou, AC Cefalas, AK Naumov, VV Semashko, RY Abdulsabi-
rov, SL Korableva. Opt Comm 149: 368–392, 1998.
70. AA Kaplyanski, AK Przhevuskii. Opt Spectrosc 19: 331–338, 1965.
71. T Hoshina, SZ Kuboniwa. J Phys Soc Jpn (171): 828–840.
72. HM Crosswhite, GH Dieke, WJ Carter. J Chem Phys 43: 2047–2054, 1965.
73. RA Apaev, MV Eremin, AK Naumov, VV Semashko, R Yu. Abdulsabirov, SL
Korableva. Opt Spectrosc 84: 735–737, 1998.
74. KH Hellwege. Ann Phys 4: 95, 1948.
75. JS Griffith. The Theory of Transition-Metal Ions. Cambridge, UK: Cambridge Uni-
versity Press, 1964.
76. JK Lawson, SA Payne. Phys Rev B47: 14003–14010; 1993. J Opt Soc Am 8: 1404–
1411, 1991.
77. MJ Freiser, S Methfessel, F Holtzberg. J Appl Phys 39: 900–902, 1968.
78. A Yanase, T Kasuga. Prog Theor Phys Suppl 46: 388–410, 1970.

Copyright © 2002 by Taylor & Francis

79. Y Guyot, S Guy, MF Joubert. J Alloys Comp 323/324: 722, 2001.
80. TT Basiev, A Yu. Dergachev, YV Orlovski, AM Prokhorov. In: A Dubé, L Chase,
eds. OSA Proceedings, Advanced Solid State Lasers. Washington, DC: Optical Soci-
ety of America, vol 10, 358–362, 1991.
81. MF Joubert, B Jacquier, C Linares, R Macfarlane. J Lumin 53: 477–482, 1992.
82. J Sugar. J Opt Soc Am 53: 831–839, 1963.
83. M Laroche. University of Caen, private communication, 2000.
84. TI Butaeva, KL Ovanesyan, AG Petrosyan. Cryst Res Tech 23: 849, 1988.
85. KS Lim, DS Hamilton. J Opt Soc Am 6: 1401, 1989.
86. RR Jacobs, WF Krupke, MJ Weber. Appl Phys Lett 33: 410, 1978.
87. WJ Miniscalco, JM Pellegrino, WM Yen. J Appl Phys 49: 6109, 1978.
88. JF Owen, PB Dorain, T Kobayasi. J Appl Phys 52: 1216, 1983.
89. O Guillot-Nöel, B Bellamy, B Viana, D Gourier. Phys Rev B60: 1668, 1999.
90. C Pédrini, F Rogemond, DS McClure. J Appl Phys 59: 1196; 1986. C Pédrini, DS
McClure, CH Anderson. J Chem Phys 70: 4959, 1979.
91. J Sugar, J Reader. J Chem Phys 59: 2083–2089, 1973.
92. D Bouttet, C Dujardin, C Pédrini, W Brunat, D Tran Minh Duc, JY Gesland. Pro-
ceedings, International Conference on Inorganic Scintillators and their Applications.
Delft, pp 111–113, 1996.
93. C Pédrini, D Bouttet, C Dujardin. Proceedings, International Conference on Inor-
ganic Scintillators and their Applications, Delft 106–108, 1996.
94. B Moine, C Dujardin, H Lautesse, C Pédrini, CM Combes, A Belski, P Martin, JY
Gesland. Mater Sci Forum 239–241: 245, 1997.
95. K Shimamura, SL Baldochi, N Mujilatu, K Nakano, Z Liu, N Sarukura, T Fukyda.
J Cryst Growth 2000.
96. A Hairie, M Laroche, S Girard, R Moncorgé (unpublished).
97. RA Heaton, CC Lin. Phys Rev B22: 3629–3638, 1980.
98. L Brewer. J Opt Soc Am 61: 1666–1682, 1971.

Copyright © 2002 by Taylor & Francis

10
Generation of Coherent Ultraviolet
and Vacuum Ultraviolet Radiation
by Nonlinear Processes in Intense
Optical Fields
A. L. Oldenburg and J. G. Eden
University of Illinois, Urbana, Illinois

For the rest of my life, I will reflect on what light is.

A. Einstein (ca. 1917)

I. INTRODUCTION

The year 2001 marks the 200th anniversary of the discovery of the ultraviolet
(UV). In 1801, Johann Wilhelm Ritter reported [1] in Annalen der Physik that,
on the 22nd of February of that year, he had detected solar radiation ‘‘on the
side of the violet of the color spectrum, outside of the same . . .’’ (emphasis
added). A more detailed paper, published by Ritter in 1803, states: ‘‘Dafs auch
aufserhalb des Violett des Newtonschen Spectrums unsichtbare strahlen anzutref-
fen find . . .’’ (roughly translated: ‘‘That outside of the violet of Newton’s spec-
trum, invisible radiation is found . . .’’) and concludes that ‘‘ . . . these chemically
active rays must be completely different from the colored ones’’ [2]. Tousey [3]
has recounted the pioneering work of Stokes and Schumann, who discovered the
vacuum ultraviolet (VUV) later in the 19th century and extended the frontiers
of spectroscopy to wavelengths below 130 nm.
One hundred and sixty-two years following Ritter’s announcement, H. G.
Heard ushered in a new era for optical physics in the UV and beyond when he

Copyright © 2002 by Taylor & Francis

stated, in a 30 line letter published in Nature, that ‘‘Ultraviolet coherent light
has been generated directly at room temperature in a pulsed nitrogen-gas laser
. . . Despite the short pulse duration, the output spot of the laser is exceedingly
brilliant’’ [4]. The peak power generated at 337.1 nm in these first experiments
was a modest 10 W but D. A. Leonard [5] succeeded less than 2 years later
in producing 200 kW pulses from an oscillator–amplifier system employing a
segmented transverse discharge. In the early 1970s, the discharge-pumped N 2
laser was the only widely available coherent UV source for the laboratory. Capa-
ble of producing single-pulse energies of several mJ at repetition frequencies of
tens of Hz, commercial N 2 laser systems were indispensable for pumping dye
lasers. Nevertheless, because their peak (instantaneous) pulse powers are on the
order of 1 MW, the N 2 laser proved to be of limited value in exploring nonlinear
phenomena.
The discovery of the rare gas–halide lasers in 1975 introduced a family of
oscillators offering peak powers of tens of MW at several discrete wavelengths
in the UV and VUV (193–353 nm) [6]. Despite the enormous impact that these
lasers have had in several areas such as photochemistry and medical procedures,
the difficulty in generating laser pulse widths below ⬃0.5 ps, in addition to the
beam quality, limits the peak intensities readily attainable from a system of ac-
ceptable complexity. Four decades after the advent of the laser, therefore, few
primary sources offering peak powers above 1 MW exist in the UV and VUV.
Those systems that are available are characterized by limited tunability (typically
⬍150 cm ⫺1) and are incapable of generating intensities above ⬃10 12 W-cm ⫺2 in
a compact (tabletop), high pulse repetition frequency (PRF) system.
The second avenue to generating coherent UV/VUV radiation, nonlinear
frequency upconversion from a fixed or tunable primary source, has a history that
also extends back prior to 1970. Over the past 30 years, a wide variety of nonlinear
processes have been demonstrated to yield coherent UV or VUV radiation (tunable
single frequency or broadband) and it is this subject to which the present chapter
is devoted. For the sake of brevity, we choose to limit our remarks to three pro-
cesses that have proven particularly valuable in the advance of spectroscopy and
optical physics in atoms and small molecules: continuum generation, four-wave
mixing, and a source of more recent origin—high-order harmonic generation.

II. CONTINUUM GENERATION

Continuum generation is a convenient and versatile process by which UV radia-

tion can be produced and, accordingly, has been applied to time-resolved absorp-
tion spectroscopy and the production of ultrashort laser pulses. Because of the
large body of literature devoted to this subject, the present review will discuss
only the fundamental mechanisms of continuum generation and the practical as

Copyright © 2002 by Taylor & Francis

pects of the production of broad continua while maximizing efficiency.
When a sufficiently intense laser pulse excites a nonabsorbing medium, it
will invariably, in the absence of competing processes, exhibit continuum genera-
tion. Continuum generation is characterized by spectral broadening of the incident
laser light and, near threshold, has the important feature of preserving the tempo-
ral coherence of the laser. Broadening occurs as a result of the nonlinear refractive
index (n 2) of the medium, which gives rise to self-phase modulation (SPM) and
higher-order processes. As a result, continuum generation is typically achievable
only with laser pulses ⱗ 10 ps in duration and ⲏ 10 10 W/cm 2 in intensity.

A. Theory and Experimental Considerations

To understand the mechanisms that drive continuum generation in a particular
medium, it is useful to write the refractive index in terms of the average linear
refractive index n 0 and the nonlinear refractive index n 2 :
n(t) ⫽ n 0 ⫹ n 2 |A(t) | 2 (1)
where A is the complex electric field envelope. We have assumed here that the
response time of n 2 is sufficiently fast to follow the pulse envelope, but not on
the same time scale as an optical cycle. Contributions to the magnitude and re-
sponse time of n 2 (which is ⬀ χ (3) in lowest order) can come from various physical
processes such as electronic polarization, molecular orientation, and electrostric-
tion. For practical purposes, it is a general rule that vapors have a lower n 2 than
do condensed media such as liquids and glasses.
To illustrate the SPM process, one can write the solution for a linearly
polarized laser pulse propagating in a nonlinear medium (neglecting processes
such as Raman scattering and absorption) as follows [7]:

A(z, t) ⫽ a 0 F (t)exp i冤 ω2cn a F (t)z冥

0 2 2
0
2
(2)

where F (t) is the laser pulse envelope, a 0 is its amplitude, and ω 0 is the fundamen-
tal frequency. As the laser pulse arrives, it induces a change in the refractive
index of the medium (as quantified by n 2), which, in turn, induces a phase change
in the pulse envelope that is proportional to the field strength. This phase change
causes the instantaneous frequency (ω(t) ⫽ ω 0 ⫺ dφ(t)/dt) to be first red-shifted
and subsequently blue-shifted, giving rise to an upchirp in the central portion of
the pulse. Computing the spectrum from this solution illustrates that not only
spectral broadening but also spectral modulation occurs, both of which are spec-
trally symmetrical about ω 0 in the pure SPM case. It is also important to note
that, in this ideal case, the temporal coherence of the pulse is preserved. Although
the temporal pulse width may increase following SPM, it is theoretically possible

Copyright © 2002 by Taylor & Francis

to recompress the pulse to durations shorter than that for the pre-SPM pulse by
compensating for the spectral phase.
By assuming a gaussian temporal pulse envelope with a FWHM of τp, we
can estimate the maximum spectral extent of the SPM-broadened pulse to be:

√
8 ln 2 ω 0 n 2 a 20 z
∆ω max ⬇ (3)
e cτp
Although the broadening is proportional to the interaction length z and 1/τp, for
femtosecond driving pulses one must also consider the detrimental effects of
group velocity dispersion (GVD) which increase with z and the laser bandwidth
(typically, ⬀1/τp). GVD gives rise to temporal broadening (in the form of laser
chirp), which reduces the peak intensity, thereby quenching the SPM effect. For
this reason, vapors may be more desirable, due to their reduced GVD, than con-
densed media.
Another result of the above expression [Eq. (3)] is that broadening is pro-
portional to the laser intensity, given by a 20 . A second limitation, aside from the
availability of intense laser sources, is breakdown of the medium, which occurs
at much lower intensities for condensed media than for vapors. Also, vapors and
liquids are self-healing, whereas solid samples generally are not.
It should be noted that, in the above treatment, other nonlinear effects such
as self-focusing and self-steepening, which are often inextricably linked with
continuum generation, were neglected. Self-focusing arises as a result of the
change in index over the spatial profile of the electric field E(r) associated with
the laser pulse laser; this mechanism facilitates the SPM process by confining
the laser pulses to a smaller cross-sectional area of higher intensity. However,
self-focusing also gives rise to potentially undesirable effects such as conical
emission (see color insert, Fig. 1A), which appears in the far-field pattern, and
the breakup of the laser profile into microfilaments within the interaction medium.
Both are capable of destroying the spatial and temporal coherence of the radia-
tion. The process of self-steepening is characterized by a time delay in the peak
of the pulse with respect to the edges, caused by the larger index of refraction
induced by the peak intensity. GVD and self-steepening can be responsible for
asymmetries in the spectrum of an otherwise symmetrical, purely SPM-broad-
ened continuum.

B. Review of Experimental Results

Although self-phase modulation was observed in liquid CS 2 in 1967 by Shimizu
[8], the first demonstration of continuum generation over thousands of cm ⫺1(later
dubbed supercontinuum generation) was reported by Alfano and Shapiro in 1970
[9–11]. These supercontinua, which in some cases extended from 400 to 700

Copyright © 2002 by Taylor & Francis

nm, were produced in such media as glass, crystals, and condensed rare gases
with 4 ps laser pulses at 530 nm. Microfilaments 5–20 µm in diameter were
observed and credited with increasing the laser power density to ⬃10 13 W/cm 2.
Since, ideally, the continuum spectrum is symmetrical about the pump fre-
quency, high-intensity (⬎10 13 W/cm 2) femtosecond laser pulses are necessary
when one is working with visible or near-infrared pump sources to produce exper-
imentally useful continua extending into the ultraviolet. Since stimulated Raman
scattering (SRS) becomes of increasing importance with excitation of increasing
duration, [7] femtosecond lasers are the logical choice for realizing the required
pumping intensities and minimizing the detrimental effects of SRS.
To the authors’ knowledge, Fork et al. were the first to report, in 1982, the
generation of continua extending significantly into the ultraviolet, [12,13] using
a colliding pulse mode-locked (CPM) laser generating 65 fs pulses centered at
627 nm. By focusing ( f /12) the laser pulses into an ethylene glycol jet, a super-
continuum having a spectral range from 190 nm to 1.6 µm was produced. In this
experiment, the effect of chirp due to GVD was essentially negligible, owing to
the small thickness of the jet (500 µm), giving rise to delays as low as 10 fs/
100 nm of spectral width.
Because of the lower nonlinear index n 2 generally exhibited by gases and
vapors as compared to condensed media (typically differing by four orders of
magnitude), the former were not considered seriously as a source of efficient
continuum generation until 1985, when Glownia et al. [14] found that intense
pulses gently focused into air exhibited a surprising degree of SPM. Using XeCl
amplifier modules to produce 10 mJ, 350 fs pulses at 308 nm, it was found that
pulses exiting the first amplifier with an energy of 1.5 mJ were spectrally broad-
ened in air from 60 cm ⫺1 into a continuum of ⬃1000 cm ⫺1 breadth. It was found
independently [15] that supercontinua extending over 10,000 cm ⫺1 could easily
be produced in many high-pressure (1–40 atm) gases, such as most of the rare
gases, H 2 , N 2 , and CO 2 . In retrospect, gases are the medium of choice for most
applications in the UV, due to their low GVD and, in many cases, transparency
in the UV.
Upon further study of continuum generation in gases, [15,16] it was sug-
gested by Corkum and co-workers [15] that the threshold for continuum genera-
tion is closely associated with that of self-focusing. In support of this concept,
Glownia et al. [17,18] reported that the onset of continuum generation coincides
with the formation of filaments within the medium. As a result, the continuum-
generation process cannot be properly described by the simplified theory pre-
sented above. The reader is referred to several theoretical papers [19–22] which
take self-focusing and higher-order effects into account. In fact, these papers
illustrate that GVD or another process, which quenches self-focusing, must be
present to prevent total collapse of the light filaments to a diameter on the order
of a wavelength.

Copyright © 2002 by Taylor & Francis

 Subsequent to the discovery of continuum generation in gases and vapors,
a number of experiments were reported. The studies carried out by Corkum et
al. [15,16,23] utilized a 600 nm laser producing pulses with energies of 0.5 mJ
and temporal widths of 70 fs and 2 ps (Table 1). For the gases studied (Ar, Kr,
Xe, H 2 , N 2 , and CO 2), the ‘‘cutoff’’ wavelength on the blue side ranged from
290 nm in Xe to 330 nm in CO 2 , and was independent of pump intensity, but
did increase slightly with pressure. In general, the position of the blue edge of
the spectral profile appeared to be fixed, independent of the gas used and the
intensity applied above threshold. Glownia and co-workers [17,18] utilized a 308
nm source generating 350 fs pulses (as mentioned above) to produce a continuum
with a spectral range from 235 nm to 450 nm, for the purposes of time-resolved
absorption spectroscopy experiments. Their best results were achieved in 33 atm
of argon by focusing the pump light with a 1.5 m focal length lens to a point 20
cm before the exit window of a cell, 1 m in length. Approximately 10 filaments
were observed in the near field. Several single-shot spectra were obtained that
exhibited a widely varying frequency of modulation, illustrating the problem of

Table 1 Summary of the Characteristics of Continua Generated in Several

Selected Experiments
Pump laser Continuum Continuum
characteristics Medium spectral range spectral density References

627 nm/70 fs/250 µJ Ethylene 190 nm–1.6 µm ⬃2 nJ/nm @ 400 nm 12, 13

glycol
600 nm/2 ps/500 µJ Xe 290 nm-IR NA 15, 16, 23
and CO2
600 nm/70 fs/350 µJ Ar, Kr, N2, H2 330 nm-IR
⬃300 nm-IR
308 nm/300 fs/2.5 mJ Ar 235–450 nm ⬃2 µJ/nm @ 280 nm 17, 18
H2 mostly SRS
CO2 295–360 nm
248 nm/700 fs/10 mJ Ne 190–400 nm NA 24
Ar 200–340 nm
Kr 200–320 nm
405 nm/165 fs/5 mJ Liquid H2O 200–600 nm ⬃5 µJ/nm @ 300 nm 25
790 nm/125 fs/167 mJ Ne 700 nm-IR — 26
Ar 150 nm-IR ⬃20 µJ/nm @ 300 nm
Kr 220 nm-IR ⬃100 µJ/nm @ 300 nm
Xe 280 nm-IR ⬃2 µJ/nm @ 300 nm
(see Fig. 2)

The spectral densities given are order-of-magnitude values for the sake of comparison.

Copyright © 2002 by Taylor & Francis

shot-to-shot noise for this highly nonlinear experiment. Also, it is interesting to
note that, in H 2 , both SRS and continuum generation were present, with the con-
tinuum generation contributing less to the spectrum at lower pressures.
At shorter wavelengths, continuum generation has been studied by Gosnell
et al. [24] in Ne, Ar, and Kr at high pressures (ⱖ40 atm). With 700 fs pump
pulses at 248 nm, they found the threshold energy for continuum generation to
be as low as 60 µJ. They also determined that the shot-to-shot noise can be
reduced to a reasonable level (0.2% absorption sensitivity when averaging 100
shots) by using Kr at 10 atm. Figure 1 illustrates the results obtained for neon,
argon, and krypton, which produce continua spanning from roughly 200 nm to
320 nm. Neon is recommended for shorter-wavelength applications because it
exhibits the broadest spectrum, extending from the violet to 190 nm.
Although the feasibility of using gases for ultraviolet continuum generation
was accepted and generally preferred by the early 1990s, an experiment by Rodri-
guez and co-workers [25] illustrates the potential advantages of a liquid medium.
The frequency-doubled output of a Ti:sapphire laser, yielding 5 mJ, 165 fs pulses,
was used to excite H 2 O. The continuum generated was extremely broad (⬃33,000
cm ⫺1), extending from 200 nm (limited by the cell transmission) to 600 nm. A

Figure 1 Supercontinuum spectra obtained with neon, argon, and krypton at 40 atm.
The input pulse energy is approximately 10 mJ. (From Ref. 24, reprinted by permission.)

Copyright © 2002 by Taylor & Francis

surprising finding was that several SPM signatures such as microfilamentation,
conical emission, and spectral modulation were not observed in this experiment,
resulting in good shot-to-shot stability and spatial coherence.
As the availability of strong optical fields (intensities ⬎ 10 13 –10 15 W/cm 2)
produced by near-IR femtosecond lasers became more commonplace in the
1990s, XUV production through high-order harmonic generation in the rare gases
(see Sec. IV) drew considerable attention. However, using one of these highly
intense lasers strictly for continuum generation can also produce interesting re-
sults. In 1995, for example, Nishioka et al. [26] demonstrated that a pulse at 790
nm, having a peak power of 1.6 TW, will break down 1 atm of rare gas, even
if focusing of the beam is weak ( f ⫽ 3 m). For a focal length of 5 m, however,
fine trapping channels formed, which were measured in air to be 3 m in length.
The resulting spectra spanned the region from the IR down to 150 nm for argon
(see Fig. 2). These are, to our knowledge, the broadest continua reported to date.
The threshold for continuum generation (i.e., at which a single filament would
form) occurred at 10 mJ. As the laser power was increased by an order of magni-
tude, the spatial mode pattern actually improved, possibly due to the presence
of a higher number of microfilaments.

Figure 2 Spectral intensity as a function of wavelength for continua generated in atmo-

spheric-pressure rare gases. The laser power is 1.6 TW. (From Ref. 26, reprinted by per-
mission.)

Copyright © 2002 by Taylor & Francis

 For those not having access to a terawatt laser, however, the next best
option for UV continuum generation might be the use of a gas-filled hollow fiber.
It combines the advantages of the gaseous medium (low GVD, UV transmittance)
with a waveguide that maintains the optical field intensity over an interaction
region of extended length. For applications in femtosecond pulse generation, the
hollow fiber is also desirable because it improves the spatial coherence of the
output radiation, which, in turn, minimizes loss of the temporal coherence. Within
the femtosecond laser community, it is well known that preserving temporal co-
herence (for the purpose of pulse compression) requires that continua be gener-
ated near threshold so as to minimize the number of microfilaments. The hollow
fiber, therefore, affords a means of increasing the field–gaseous medium interac-
tion length without the loss of coherence.
Recent experiments that took advantage of the hollow fiber were conducted
by Nibbering et al. [27,28] to produce continua spanning the 360–430 nm spectral
region. The emitted light was subsequently phase compensated to produce pulses
as short as 8 fs with energies of 15 µJ. The hollow fiber used had a bore diameter
of 220 µm, and was filled with Ar. Argon was chosen specifically because it
balances the ionization threshold with a moderate nonlinear susceptibility. In con-
trast, neon and helium have larger ionization thresholds, but the nonlinear suscep-
tibilities are considerably smaller than that for Ar. Sokolowski-Tinten et al. [29]
have also shown that the hollow fiber provides a degree of spectral broadening
comparable to that obtainable for glass. Their experiments, conducted with 1.2
mJ pulses at 800 nm launched into a 70 cm long fiber of 170 µm bore diameter
filled with up to 5 atm of Ar, also demonstrated that the spatial quality of the
radiation emerging from a hollow fiber is preferable to that generated in BK7
glass.

III. FOUR-WAVE MIXING

Four-wave mixing (FWM) is a parametric process that owes its origin to the
third-order nonlinear susceptibility (χ (3)) of the medium. In essence, a coherent
fourth wave is generated with a frequency (ω 4) that is one of the following linear
combinations of the input waves: ω 4 ⫽ 3 ω 1 (third harmonic generation; THG),
ω 4 ⫽ 2ω 1 ⫹ ω 2 , and ω 4 ⫽ ω 1 ⫹ ω 2 ⫹ ω 3 (sum-frequency generation; SFG),
and ω 4 ⫽ 2ω 1 ⫺ ω 2 and ω 4 ⫽ ω 1 ⫺ ω 2 ⫾ ω 3 (difference-frequency generation;
DFG). As outlined below, four-wave mixing can be used for efficient frequency
conversion (⬎50%) into the near UV as well as the VUV and XUV. This can
be achieved by optimizing the effects of phase matching in combination with
resonant enhancement of the third-order susceptibility (χ (3)).

Copyright © 2002 by Taylor & Francis

 A comprehensive treatment of FWM theory and an overview of the consid-
erable corpus of experimental work in the literature would itself require several
chapters. Rather, the fundamental theoretical principles and a sampling of experi-
mental techniques and results are presented here with the intention of providing
an introduction to the subject. For a more detailed discussion, the reader is en-
couraged to consult Refs. 30–33.

A. Background
Before delving into the experimental details of FWM, it may be useful to review
the basic concepts and theoretical considerations associated with this process. If
ground-state depletion is ignored, three sufficiently intense electromagnetic
waves will induce a third-order polarization (in a nonlinear medium) that oscil-
lates at one of the mixing frequencies of the input waves. This polarization facili-
tates the growth of a fourth wave over the length of the medium. For incident
plane waves, the intensity of the generated fourth wave can be written as [30]

| χ (3) | 2 L 2 sin 2 (∆kL/2)

I4 ⬀ I 1 I 2 I 3 (4)
n 1 n 2n 3 n 4 λ 24 c 2 (∆kL/2) 2
where the Ij are the intensities of the incident waves, nj are the indices of refraction
at the wave frequencies ωj , L is the length of the medium, and ∆k is the ‘‘phase
mismatch’’ of the four wave vectors [∆k ⫽ k 4 ⫺ (k 1 ⫾ k 2 ⫾ k 3)]. The last term
in the above expression is essentially a sinc function, which is responsible for a
significant (if not total) loss of power delivered to the fourth wave, unless the
phase-matching condition ∆k ⫽ 0 is satisfied. For sum-frequency generation, for
example, the phase-matching condition can be written as:
2πc∆k ⫽ n(ω 4)ω 4 ⫺ [n(ω 1)ω 1 ⫹ n(ω 2)ω 2 ⫹ n(ω 3)ω 3] ⫽ 0 (5)
Since, in this instance, ω 4 ⫽ ω 1 ⫹ ω 2 ⫹ ω 3 , the presence of positive or negative
dispersion in the medium (dn(ω)/dω ⬎ 0 or ⬍ 0, respectively) gives rise to a
positive or negative ∆k, respectively.
The phase-matching condition, presented above in the plane-wave approxi-
mation, is sensitive to the tightness of focusing, as well as the positioning of the
focal point within the medium. Table 2 summarizes the optimum value of the
vector ∆k (which has been computed for gaussian beams by Bjorklund [34]) in
terms of the confocal parameter b ⫽ 2πω 20 n/λ (where ω 0 is the beam-waist ra-
dius) of the incident pump waves. As illustrated by Table 2, for certain cases a
‘‘perfect’’ phase-match (∆k ⫽ 0) may actually result in no output when the beams
are tightly focused (b ⬍⬍ L). Also, the desired sign of ∆k depends on the type
of FWM process. Specifically, a normally dispersive medium induces a positive
∆k and is, therefore, well suited for DFG of the type ω 4 ⫽ ω 1 ⫺ ω 2 ⫺ ω 3 .

Copyright © 2002 by Taylor & Francis

Table 2 Optimal Values for the Wave-Vector Mismatch ∆k in Terms
of the Confocal Parameter b for Various Focusing Schemes and FWM Processes
Optimum ∆k

Plane Focus far

Focus in center of medium wave outside medium

Tight focusing Loose focusing

FWM process (b ⬍⬍ L) (b ⬎⬎ L)

THG and SFG ∆k ⫽ ⫺2/b ∆k ⫽ ⫺4/b

(No FWM for
∆k ⱖ 0)
ω4 ⫽ ω1 ⫹ ω2 ⫺ ω3 ∆k ⫽ 0 ∆k → 0 as ∆k → 0
and b→ ∞
ω 4 ⫽ 2ω 1 ⫺ ω 2
ω4 ⫽ ω1 ⫺ ω2 ⫺ ω3 ∆k ⫽ 2/b ∆k⫽ 4/b
and (No FWM for
ω 4 ⫽ ω 1 ⫺ 2ω 2 ∆k ⱕ 0)

Source: Adapted from Ref. 34.

This is in contrast to THG or SFG, for which a negatively dispersive medium is

desirable.
Various techniques can be used to achieve optimal phase matching. In solid
media, the length of the medium (and, if the medium is a crystal, the angle of
orientation) is often optimized for the wavelengths of interest. Also, birefringence
(the dependence of n on the angle of polarization) can be exploited when the gener-
ated wave is driven with a polarization perpendicular to one or more of the incident
waves. In isotropic media where birefringence does not occur, however, the disper-
sion is typically positive in the photon energy region below the first excited state,
with anomalous dispersion occurring at each single-photon resonance. It is often
possible to achieve phase-matching by simply tuning or choosing the input frequen-
cies carefully, in order to take advantage of the anomalous dispersion effect. One
technique for phase-matching in vapors involves using a mixture of positively and
negatively dispersive gases, in which the partial pressure of each gas can be tuned
so as to achieve phase-matching. Typically, one constituent of the vapor mixture
will actually contribute to the interaction (through χ (3)), while both contribute to
the dispersion n(ω). Of course, since n(ω) is also dependent upon the number
density, pressure and temperature tuning, in general, are important to the FWM
process in gases and vapors. Finally, the angle of the incident beams can be varied,
which takes advantage of the vector nature of ∆k. This is particularly useful in
liquids and solids, where pressure tuning is not possible.
Again considering Eq. (4), it is evident that the FWM process is critically

Copyright © 2002 by Taylor & Francis

dependent upon the third-order susceptibility, χ (3). The lowest-order nonlinear
term for isotropic media, χ (3)(ω), is greatly enhanced at frequencies near reso-
nance. However, tuning an input wave near a single photon resonance is accom-
panied by resonant absorption, which quenches FWM. Also, anomalous disper-
sion near a resonance becomes steep (i.e., | dn/dω | is large), so that the detuning
requirements for phase-matching can be quite strict. This is also true for three-
photon resonances, where the output wave is resonant with the medium (and so
is absorbed by the medium). Tuning near a two-photon resonance, however, is
often ideal in that it effectively enhances χ (3) without exhibiting anomalous dis-
persion (again assuming negligible depletion of the ground state). In this way,
χ (3) can be enhanced without affecting the phase-matching condition.

B. Experiments
1. FWM in Alkali Metal Vapors
Because of the large dipole moments associated with optical transitions in the
alkali atoms, one expects χ (3) to be large and conducive to efficient FWM. Indeed,
Harris and Miles [35] in 1971 predicted χ (3) for the alkali atoms to be larger than
that for the rare gases by six orders of magnitude. In experiments reported that
year, Young and co-workers [36] demonstrated the potential of FWM in the alkali
vapors by frequency-tripling the output of the Nd :YAG laser to 354.7 nm. Their
results showed for the first time that FWM could be a tool for generating coherent
radiation at new wavelengths and for upconverting fixed wavelength primary
sources, in particular. A key aspect of these experiments was the introduction of
mixing two (or more) vapors to realize phase-matched third harmonic generation.
A factor of 33 improvement in the efficiency for generating UV photons was
obtained when a 412:1 Xe/Rb vapor mixture (rather than pure Rb vapor) served
as the conversion medium. The Xe/Rb ratio was chosen such that the index of
refraction at the fundamental wavelength (1.064 µm) was the same as that at
the third harmonic. As one might expect, phase matching is equally sensitive to
temperature. Optimum conversion of power at the fundamental into the third
harmonic was measured for a temperature of 265°C. Deviations of the tempera-
ture from this value as small as ⫾ 10°C yielded virtually no output.
The experiments of Young and colleagues [36] were designed such that,
as illustrated in Figure 3a, the THG process in Rb vapor is enhanced by a three-
photon near-resonance. That is, the combined energies of three photons, at fre-
quency ω 1 , differs from the energy of a three photon-accessible state by the detun-
ing δ ⬃ 300 cm ⫺1. However, as discussed earlier, a more favorable situation for
FWM is that in which a two-photon excitation step in the overall process is reso-
nantly enhanced. As an example, Leung et al. [37] tripled the frequency of the
694 nm output of a ruby laser in Cs vapor by a three-photon process having a

Copyright © 2002 by Taylor & Francis

Figure 3 Generalized atomic energy level diagrams illustrating a variety of four wave-
mixing processes that have been employed in metal vapors and gases. Solid and dashed
horizontal lines denote stationary and virtual states, respectively, and the shaded region
in (b) indicates the ionization continuum. An arrow with a heavy line indicates a photon
incident on the nonlinear medium; the double-line arrows represent photons generated
within the nonlinear medium. Examples of experiments that have demonstrated each of
these processes include (a) Refs. 36, 52; (b) Refs. 37, 38, 53; (c) Ref. 39; (d) Refs. 42,
50; (e) Refs. 48, 49; (f) Refs. 38, 39; (g) Ref. 40; (h) Refs. 55, 57–60; (i) Refs. 61, 62.

two-photon intermediate resonance (cf. Fig. 3b). Although χ (3) for this process
is roughly a factor of 70 larger than that for the frequency-tripling experiments
of Young et al., [36] the combined energies of the three-pump photons exceed
the ionization potential for Cs, with the result that the conversion efficiency is
insensitive to phase matching with the rare gases [38].
An alternative, and considerably more efficient, process was reported by
Taylor [38] whose experiments in sodium vapor demonstrated an ω 3 ⫽ 2ω 1 ⫺
ω 2 DFG process (schematically illustrated in Fig. 3f). Occurring simultaneously
with THG, the DFG process produced 330.2 nm photons in the UV—in the
vicinity of the second principal series of Na (4 2 P J → 3 2 S, J ⫽ 1/2, 3/2)—with an
efficiency of 5 ⫻ 10 ⫺2%, or a factor of ⬃500 greater than that for the THG
process at the same fundamental wavelength. In this process, also known as four-
wave parametric oscillation (FWPO), only ω 1 is supplied externally whereas ω 2
and ω 3 are produced within the medium. Two possible mechanisms for generation
of the idler wave are stimulated electronic Raman scattering (SERS), and opti-
cally pumped stimulated emission (OPSE), which can occur as a result of the
initial two-photon excitation process. The idler then acts in concert with the pump
radiation (ω 1) to generate the signal wave near the np → ground transition by

Copyright © 2002 by Taylor & Francis

FWM. An interesting set of experiments carried out by Smith and Ward [39]
examined FWPO in Cs vapor. With pulses from a ruby laser, they were able to
excite near-resonantly the 9 2 D 3/2 ←← 6 2 S 1/2 two-photon transition and observed
coherent emission in the vicinity of the np 2 P 1/2 (7 ⱕ n ⱕ 10) → 6 2 S 1/2 transitions.
As much as 3 W of peak power generated at the signal wavelengths was observed.
The comparatively recent development of femtosecond laser systems offers
a new tool with which to examine FWM. Because of the large spectral bandwidths
associated with ⱗ100 fs pulses, for example, it is now possible to drive simulta-
neously more than one two-photon transition in the alkalis (see Fig. 3(g)). Figure
4 displays the near-UV spectrum observed when Rb vapor is excited by ⬃620
nm, 120 fs pulses from a colliding pulse, mode-locked (CPM)-dye amplifier sys-
tem [40]. Phase matching dictates that coherent emission is produced just to the
‘‘blue’’ side of the Rb np 2 P → 5s 2 S 1/2 transitions. Pump-probe experiments
[40] demonstrate that this emission serves as a means of detecting an atomic
wavepacket. A photograph of the violet signal radiation produced by two-photon
excitation of Rb vapor with ⬃100 fs pulses at 770 nm (Ti :Al 2 O 3 laser system)
is shown in Figure 5.

Figure 4 UV spectrum of the coherent emission produced by Rb vapor when excited

by 620 nm, ⬃120 fs pulses. Each of the sharp features observed arises from four-wave
parametric oscillation near the terms of the np 2 P J → 5s 2 S 1/2 (7 ⱕ n ⱕ 11) principal
series of Rb. (Data from Ref. 40.)

Copyright © 2002 by Taylor & Francis

Figure 5 Photograph of the violet emission at 420 nm produced when 2 Torr of Rb
vapor (in the heat pipe at right) is irradiated with 1 mJ, 120 fs pulses at 770 nm produced
by a Ti: sapphire laser. The pump radiation is incident on the heat pipe from the right and
the coherent violet/UV radiation emerges from the left end of the heat pipe. The violet
emission is the result of FWPO in the vicinity of the 6p → 5s transition of Rb and the
divergence of the coherent radiation matches that of the incident (pump) laser radiation.
(See also color insert.)

2. Other Metal Vapors

Although the highest conversion efficiencies for FWM have generally been
achieved with the alkali metal vapors, the available wavelengths are typically
constrained to narrow spectral regions in the near-UV. The desire to generate
tunable VUV radiation prompts us to turn to other nonlinear media and the group
II metal vapors, in particular. In the early 1970s, Kung et al. [41] demonstrated
the ability of parametric processes in metal vapor/rare gas mixtures to produce
coherent VUV. With phase-matched mixtures of Cd vapor and Ar, they generated
radiation at 118, 152, and 177 nm and peak powers as high as 7 kW. Various
SFG processes were invoked, the most efficient of which was tripling frequency-
doubled Nd: YAG radiation (532 nm) to yield 177 nm photons.

Copyright © 2002 by Taylor & Francis

 Hodgson, Sorokin, and Wynne [42] subsequently reported that resonant
enhancements of the FWM process of ⬃10 4 could be realized by tuning a dye
laser near an atomic two photon-allowed transition. Utilizing two dye lasers, they
were able to generate, in Sr vapor, coherent radiation over broad regions (1200–
3500 cm ⫺1) in the VUV by the 2ω 1 ⫹ ω 2 SFG process illustrated in Figure 3d.
McKee and colleagues [43] extended this approach to the important spectral re-
gion encompassing the Lyman-α transition of hydrogen (121.6 nm). A photon
flux ⱖ 10 8 photons/pulse was produced at the Lyman-α wavelength with two
KrF laser (248 nm)-pumped dye lasers to excite Mg vapor, yielding a narrow-
linewidth source that is well suited for the detection of H atoms.
Four-wave mixing in Hg vapor was reported in 1981 by Mahon and Tomp-
kins [44,45], who showed that conversion efficiencies and peak powers up to 1%
and 10 kW, respectively, were obtainable but the tuning range was initially lim-
ited to 2–8 Å wide windows in the 121.9–125.5 nm region. Hilbig and Wal-
lenstein [46] subsequently extended the lower limit of the tuning range for four-
wave sum mixing in Hg vapor to 109.0 nm. Tunability down to the LiF cutoff
(⬃104.5 nm) was achieved by Herman and co-workers [47] by using the 6s 8s 1 S
state of Hg for the two-photon intermediate resonance. Unprecedented conversion
efficiencies as high as 5% were reported by Muller et al. [48] in the late 1980s
for the generation of 130 nm photons in Hg vapor using a collimated beam geom-
etry and optical field-nonlinear medium interaction lengths greater than 1 m. Re-
cently, FWM in Hg vapor has successfully produced CW radiation at the hydro-
gen Lyman-α wavelength [49]. Because of the narrower linewidths offered by
CW FWM, this process is attractive for high-resolution spectroscopy. First ob-
served by Freeman et al. [50] in 1978, the CW generation of coherent VUV
radiation by FWM places stringent requirements on the frequencies of the pump
lasers. As illustrated qualitatively by Figure 3e, efficient energy conversion dic-
tates that both ω 1 and ω 3 ⫹ ω 2 ⫹ ω 1 be nearly resonant with the atomic medium
and ω 1 ⫹ ω 2 must be on resonance.

3. The Rare Gases

Although the initial experiments in THG [51] were performed in the rare gases,
it was the pioneering work of Harris, Young, and colleagues in the alkalis [35,36]
that considerably heightened interest in FWM as a practical technique for wave-
length conversion. However, the corrosive properties of the alkalis prompted a
re-examination of the rare gases. In 1973, Kung and co-workers [52] reported
an efficiency approaching 3% for the conversion of 355 nm energy into its third
harmonic (118 nm) in phase-matched Xe/Ar gas mixtures. Subsequent work in
the 1970s extended the limit of nonlinear frequency upconversion to 57 nm, well
into the extreme ultraviolet (XUV) [33,53]. Most of these experiments were fixed-

Copyright © 2002 by Taylor & Francis

frequency THG or SFG performed in the rare gases, using either an excimer laser
pump or the frequency-quadrupled output of an Nd :YAG or Nd: glass laser [33].
It was found that the rare gases are also suitable for generating intense
Lyman-α radiation. Langer and colleagues [54] demonstrated conversion effi-
ciencies up to 1.4 ⋅ 10 ⫺4 in a mixture of Kr and Ar, by tripling the frequency-
doubled output of a dye laser. Recently, Meyer and Faris [55] were able to pro-
duce 7 µJ of Lyman-α radiation in pulses having a peak power of 1.3 kW. They
chose a phase-matched gas mixture of Kr and Ar as the nonlinear medium and
generated the desired wavelength by DFG with an ArF (193 nm) excimer laser
and a dye laser.
Broadly tunable coherent radiation in the VUV has also been produced by
FWM in the rare gases and Kr and Xe, in particular [56]. Marangos et al. [57]
demonstrated one practical scheme for generating tunable VUV radiation span-
ning the entire 120–200 nm spectral region. They locked the frequency of one
dye laser onto a two-photon resonance of Kr to take advantage of the resonance
enhancement, while tuning a second dye laser in order to achieve DFG of the
form 2ω 1 ⫺ ω 2 (illustrated schematically in Fig. 3(h).
Finally, with the development of femtosecond lasers in the last decade, the
generation of ultrafast pulses in the UV and VUV has become a reality. One
scheme advanced by Glownia et al. [58] produced pulses in the VUV at 156 nm
having energies as high as 60 µJ. An amplified KrF laser served as the source
for ω 1 , a femtosecond Ti:Al 2 O 3 laser generated ω 2 , and the VUV radiation was
produced in Xe gas by the 2ω 1 ⫺ ω 2 DFG process of Figure 3h. It is interesting
to note that they also observed simultaneous FWPO at 148 nm, both in the pres-
ence and absence of pumping at ω 2 . Subsequent work [59,60] using techniques
similar to those introduced in Ref. 58 but with other rare gases and different
excimer laser wavelengths, have produced femtosecond pulses elsewhere in the
VUV. Another route was taken by Durfee and co-workers [61,62], who intro-
duced a hollow core fiber to facilitate phase-matching and increase the interaction
length. In these experiments, femtosecond pulses in the deep UV were produced
from pulses at 800 nm and 400 nm, the fundamental and frequency-doubled out-
puts of a Ti:sapphire amplifier system, by the process 3ω 1 ⫽ 2ω 1 ⫹ 2ω 1 ⫺ ω 1
(Fig. 3i). Experiments performed in rare gases both with and without the fiber
produced 3ω 1 light with conversion efficiencies of 13% and 0.1%, respectively,
illustrating the substantial improvement achieved with the hollow fiber. The UV
light generated within the hollow fiber was subsequently phase-compensated to
produce 8 fs pulses, centered at 270 nm, with energies of 1 µJ. Although this
DFG experiment was nonresonant, the use of the hollow fiber might be expected
to achieve similarly spectacular results for resonant FWM.
Before concluding this section, we would be remiss in not mentioning the
recent experiments conducted at Stanford [63] in which VUV radiation was pro-

Copyright © 2002 by Taylor & Francis

duced efficiently by adapting electromagnetically induced transparency (EIT) to
enhance the FWM process resonantly, without the need to resort to phase-match-
ing. Merriam and co-workers reported that 233 nm photons could be converted
in lead vapor to a wavelength of 186 nm with a photon conversion efficiency
approaching unity. The application of EIT to short-wavelength generation is in
its infancy but, when coupled with existing nonlinear frequency upconversion
processes, is likely in the future to yield sources of significantly improved effi-
ciencies and beam quality.

IV. HIGH-ORDER HARMONIC GENERATION

Over the past 10–15 years, experimentalists have observed a remarkable variety
of nonlinear effects that are a result of the optical field intensities and, hence,
field strengths that can now be generated. The impact of the mode-locked Ti-
doped sapphire (Ti :Al 2 O 3) laser, in particular, on atomic and optical physics over
the past several years rivals that of the tunable dye laser more than two decades
ago. From its discovery in 1982, the Ti: Al 2 O 3 laser has been developed rapidly
and commercially available, mode-locked oscillator/regenerative amplifier sys-
tems are now capable of generating sub-20 fs pulses having energies beyond 1
mJ at kHz pulse repetition frequencies (PRF). Consequently, this laser offers
access to a temporal and intensity (⬎10 16 W-cm ⫺2) realm likely to yield a number
of surprising nonlinear optical effects. One such phenomenon is high-order har-
monic generation (HG), the production of odd harmonics when an intense optical
field interacts with an atom or molecule in the gas phase.

A. Experimental Observations
It has been known for more than three decades that odd harmonics of the optical
field frequency are generated when an intense laser pulse interacts with an atom
in the gas phase. In 1967, New and Ward [51] reported producing the third har-
monic of the ruby laser in the rare gases. The ensuing years have witnessed
steady improvements in the number of harmonics produced. Early experiments
by Reintjes and colleagues [64] in He and Ne produced harmonics having wave-
lengths as short as 38 nm with ⬃30 ps pulses at 266 nm (fourth harmonic of
Nd :YAG) and, following the introduction of the excimer laser, several studies
[65,66] were carried out at 308 nm (XeCl) and 248 nm (KrF) with pulse widths
as low as 15 ps [66] and, in one experiment, coherent radiation at the seventh
harmonic of 248 nm (35.5 nm) was observed.
A turning point in the field occurred when McPherson et al. [67] detected
radiation corresponding to the 17th harmonic of 248 nm when Ne was exposed

Copyright © 2002 by Taylor & Francis

Figure 6 Relative efficiency for the conversion of 248 nm fundamental radiation to the
odd harmonics 5 ⱕ N ⱕ 17. The data shown are for Ne and fundamental intensities of
10 15 ⫺ 10 16 W-cm ⫺2. (Adapted from Ref. 67.)

to driver field intensities in the 10 15 –10 16 W-cm ⫺2 range. More importantly, the
rate of decline of the relative intensity of the harmonics slowed dramatically
beyond the 11th harmonic (in Ne). Figure 6 shows the data reported in Ref. 67
in 1987. Similar results were observed for the other rare gases. Ferray and co-
workers [68] and Li et al. [69] of Saclay subsequently confirmed the presence
of higher harmonics (harmonic order N ⱕ 33 for Ne) for optical fields of longer
wavelength as well (λ ⫽ 1.06 µm). An intriguing characteristic of harmonic
spectra is a region, known as the plateau, in which the harmonic intensity is
relatively insensitive to harmonic order N, followed by cutoff, a rapid decline in
the harmonic response. For the lowest harmonics (N ⱗ 7), intensity also falls
rapidly with increasing harmonic order (as had been observed previously),
prompting Li et al. [69] to comment that ‘‘The first harmonics behave . . . ac-
cording to theoretical predictions derived from lowest-order perturbation theory.
. . .’’ However, ‘‘neither the extended plateau . . . nor the cutoff at high order
can be explained by the traditional theory of nonlinear optics in [a] weak field.
. . .’’ [69] Therefore, the experimental results demonstrate that HG cannot be
described by conventional perturbation theory. A vivid illustration of the nonper-

Copyright © 2002 by Taylor & Francis

Figure 7 Harmonic spectrum reported by Macklin et al. for 125 fs, 15 mJ pulses from
a Ti: Al 2 O 3 laser (λ ⬃ 800 nm) focused to an intensity of ⬃1.3 ⫻ 10 15 W-cm ⫺2 in Ne
gas. (From Ref. 70, reprinted by permission.)

turbative behavior of HG is provided by the data of Macklin et al. [70] illustrated

in Figure 7. With 125 fs, 15 mJ pulses produced by a Ti:sapphire laser (λ ≅ 800
nm), they produced a spectrum in Ne extending out to the 109th harmonic. The
authors noted that the roll-off in intensity for the lower harmonics is an artifact
of the spectral response of the monochromator. In reality, the harmonic response
declines slowly (with increasing harmonic order) up to approximately n ⫽ 85
and then falls rapidly. In addition to the harmonic spectra reported by several
groups over the past decade [71–84], extensive measurements of the polarization
characteristics [85–93], temporal coherence [94], and spatial properties [95] of
the harmonic radiation have appeared in the literature.
Perhaps the primary motivation for the experimental and theoretical interest
in HG is its potential for providing a table-top source of coherent XUV or soft
X-ray radiation. Recent experiments [70–84] have produced coherent radiation
at wavelengths at the edge of the UV to below 5 nm (soft x-ray region) from
both atoms and molecules for instantaneous optical intensities from 10 15 to ⬃10 18
W-cm ⫺2. With pulses as short as 5 fs produced by mode-locked Ti:sapphire la-
sers, harmonics having photon energies beyond 500 eV have been generated at
kHz repetition frequencies. In the UV and VUV, average powers in the µW range
can now be readily produced for individual harmonics, given the current capabili-
ties of commercial Ti: Al 2 O 3 oscillator/amplifier systems. This reliable source
of coherent short wavelength radiation is a tremendous asset for experimentalists
and promises future applications in fields such as biochemistry and medical diag-

Copyright © 2002 by Taylor & Francis

nostics as the pulse energy per harmonic continues to rise with improvements in
optical driver technology.

B. Theoretical Treatment of Harmonic Generation

Not only is HG of interest for the generation of short-wavelength coherent radia-
tion but it also offers a unique tool for exploring the interaction between an
atom and a strong optical field. Two competing views of the physical processes
responsible for HG have emerged and will be briefly described here. For further
detail, the reader is urged to consult the references cited.
Describing the harmonic response of the atom at high optical fields, and
the presence and characteristics of the higher harmonics comprising the plateau
and cutoff, in particular, requires (as noted earlier) nonperturbative theoretical
methods. Several approaches have been explored [84,96–115]. By direct numeri-
cal integration of the time-dependent Schrödinger equation, Krause, Schafer and
Kulander [98–100] calculated the response of a single atom to a strong field,
assuming the single electron approximation to be valid. However, the computa-
tional capacity required by this approach and the desirability for insight into the
physical processes underlying HG have spawned the pursuit of other nonperturba-
tive theoretical models. Several groups [92,97,102–107] have solved Schröding-
er’s equation by approximating the atomic potential, the number of bound atomic
states, or ground state depletion in different ways. In each instance, the descrip-
tion of the optical field and its interaction with an electron were treated classically.
The physical picture that best describes these calculations and the underly-
ing assumptions is, perhaps, the two-step model proposed by Kulander et al. [100]
and Corkum [101]. In this view, the atom is first ionized and the photoelectron is
accelerated by the optical field. When the electric field reverses direction in the
next half-cycle, the electron is accelerated back toward the parent atom where
recombination occurs and radiation is produced. The role of the optical field in
this semiclassical model, therefore, is the ionization of the atom and acceleration
(or deceleration) of the free electron. For this reason, the common theme of semi-
classical theoretical treatments of HG is the goal of determining the temporal
behavior (and, hence, dipole moment) of the electron oscillating in the optical
field. The semiclassical formulation proposed by Lewenstein et al. [102] has been
widely applied to describing HG experimental results, including emission spectra,
and polarization and harmonic linewidth data. It has been successful in describing
the general characteristics of HG data acquired with both long (⬎100 fs) and short
(⬍10 fs) optical pulse widths. A detailed description of semiclassical theory, the
single atom response to an intense optical field and the propagation of the re-
sulting spectrum through a plasma channel, can be found elsewhere [84,102–
111].

Copyright © 2002 by Taylor & Francis

 Recently, a second theoretical view of HG was introduced [112–115]. This
theory quantizes the optical field and makes use of the stationary solutions to
Schrödinger’s equation for an electron interacting with an optical field [116].
Known as quantized field (quantum) Volkov states, these stationary states are
separated by integral multiples of the fundamental photon energy. An analytical
expression for the matrix element for transitions between Volkov states was deter-
mined [112,114]. Consequently, the picture proposed is analogous to that for the
generation of radiation by an atom: namely, transitions between stationary states.
In this case, however, the eigenstates arise from the interaction of an electron
with an optical field as opposed to an electron interacting with an ion core.
The second theoretical picture of HG, therefore, suggests that once an elec-
tron is released from the atom via ionization, the electron interacts with the funda-
mental photon field, resulting in the formation of quantum Volkov states. Optical
transitions between Volkov states generate harmonic radiation, and the relative
intensity of each harmonic is governed by the transition matrix element. The
quantum theory also predicts that the velocity of the photoelectron is unchanged
by Volkov-to-Volkov state transitions. Comparisons of theoretical predictions
with a variety of experimental data show that the quantum theory also describes
experiments well [112–115]. To differentiate between the semiclassical and
quantum models of HG will require additional experiments over a range of pa-
rameters and, possibly, including multicolor driving fields or an x-ray prepulse
to provide a background of energetic photoelectrons.

V. CONCLUSIONS

The two centuries following the discovery of the UV have witnessed enormous
progress in both the development and application of new sources in the UV and
VUV. Today, primary and nonlinear (secondary) sources of coherent radiation
allow one to access any portion of the UV and VUV and to do so with pulses
of at least a few kW of peak power. At several discrete wavelengths in this
spectral region, peak powers beyond tens of MW are available from the F 2 (157
nm) and rare gas–halide lasers as well as high-order harmonic generation and
wave-mixing processes in the rare gases and alkali vapors.
Challenges for future work continue to be the conception and demonstration
of CW sources in the VUV as well as pulsed sources of higher efficiency and
yet reduced cost. If the events of the past 40 years are any guide to the future,
we can expect major advances to be made as the intricacies of the interaction of
radiation with atoms, molecules, and artificial (synthetic) structures are probed
to ever deeper levels.

Copyright © 2002 by Taylor & Francis

ACKNOWLEDGMENTS

This work was supported by the U.S. Air Force Office of Scientific Research.

REFERENCES

1. JW Ritter. Ann Phys 7: 527, 1801.

2. JW Ritter. Ann Phys 12: 409, 1803.
3. R Tousey. Appl Opt 1: 679, 1962.
4. HG Heard. Nature 200: 667, 1963.
5. DA Leonard. Appl Phys Lett 7: 4, 1965.
6. CK Rhodes, ed. Excimer Lasers, 2nd ed. Berlin: Springer-Verlag, 1984.
7. RR Alfano, ed. The Supercontinuum Laser Source. New York: Springer-Verlag,
1989.
8. F Shimizu. Phys Rev Lett 19: 1097, 1967.
9. RR Alfano, SL Shapiro. Phys Rev Lett 24: 592, 1970.
10. RR Alfano, SL Shapiro. Phys Rev Lett 24: 584, 1970.
11. RR Alfano, SL Shapiro. Phys Rev Lett 24: 1217, 1970.
12. RL Fork, CV Shank, C Hirlimann, R Yen, WJ Tomlinson. Opt Lett 8: 1, 1983.
13. RL Fork, CV Shank, RT Yen, C Hirlimann, WJ Tomlinson. In: KB Eisenthal, RM
Hochstrasser, W Kaiser, A Laubereau, eds. Picosecond Phenoma III. New York:
Springer-Verlag, 1982, pp 10–13.
14. JH Glownia, G Arjavalingam, PP Sorokin, JE Rothenberg. Opt Lett 11: 79, 1985.
15. PB Corkum, C Rolland, T Srinivasan-Rao. Phys Rev Lett 57: 2268, 1986.
16. PB Corkum, C Rolland, T Srinivasan-Rao. In: GR Fleming, AE Siegman, eds.
Ultrafast Phenomena V. New York: Springer-Verlag, 1986, pp 149–152.
17. JH Glownia, J Misewich, PP Sorokin. J Opt Soc Am B 3: 1573, 1986.
18. JH Glownia, J Misewich, PP Sorokin. In: GR Fleming, AE Siegman, eds. Ultrafast
Phenomena V, New York: Springer-Verlag, 1986, pp 153–156.
19. JE Rothenberg. Opt Lett 17: 583, 1992.
20. JE Rothenberg. Opt Lett 17: 1340, 1992.
21. D Strickland, PB Corkum. J Opt Soc Am B 11: 492, 1994.
22. AA Zozulya, SA Diddams, AG Van Engen, TS Clement. Phys Rev Lett 82: 1430,
1999.
23. PB Corkum, C Rolland, IEEE J Quantum Electron 25: 2634, 1989.
24. TR Gosnell, AJ Taylor, DP Greene. Opt Lett 15: 130, 1990.
25. G Rodriguez, JP Roberts, AJ Taylor. Opt Lett 19: 1146, 1994.
26. H Nishioka, W Odajima, K Ueda, H Takuma, Opt Lett 20: 2505, 1995.
27. ETJ Nibbering, O Duhr, G Korn. Opt Lett 22: 1335, 1997.
28. O Duhr, ETJ Nibbering, G Korn, G Tempea, F Krausz. Opt Lett 24: 34, 1999.
29. K Sokolowski-Tinten, M Werner, P Zhou, D von der Linde. In: Conference on
Laser and Electro-Optics. OSA Technical Digest. Washington, DC: Optical Society
of America, pp 324–325, 1999.

Copyright © 2002 by Taylor & Francis

 30. JF Reintjes. Nonlinear Optical Parametric Processes in Liquids and Gases. New
York: Academic Press, 1984.
31. CR Vidal. In: LF Mollenauer, JC White, eds. Tunable Lasers. Berlin: Springer-
Verlag, pp 57–113, 1987.
32. R Hilbig, G Hilber, A Lago, B Wolff, R Wallenstein. In: P Yeh, ed. Nonlinear
Optics and Applications. Bellingham, WA: SPIE, pp 48–55, 1986.
33. J Reintjes. Appl Opt 19: 3889, 1980.
34. GC Bjorklund. IEEE J Quantum Electron QE-11: 287, 1975.
35. SE Harris, RB Miles. Appl Phys Lett 19: 385, 1971.
36. JF Young, GC Bjorklund, AH Kung, RB Miles, SE Harris. Phys Rev Lett 27: 1551,
1971.
37. KM Leung, JF Ward, BJ Orr. Phys Rev A 9: 2440, 1974.
38. JR Taylor. Opt Comm 18: 504, 1976.
39. AV Smith, JF Ward. IEEE J Quantum Electron QE-17: 525, 1981.
40. HC Tran, PC John, J Gao, JG Eden. Opt Lett 23: 70, 1998.
41. AH Kung, JF Young, GC Bjorklund, SE Harris. Phys Rev Lett 29: 985, (1972).
42. RT Hodgson, PP Sorokin, JJ Wynne. Phys Rev Lett 32: 343, 1974.
43. TJ McKee, BP Stoicheff, SC Wallace. Opt Lett 3: 207, 1978.
44. FS Tomkins, R Mahon. Opt Lett 6: 179, 1981.
45. R Mahon, FS Tomkins. IEEE J Quantum Electron QE-18: 913, 1982.
46. R Hilbig, R Wallenstein. IEEE J Quantum Electron QE-19: 1759, 1983.
47. PR Herman, PE LaRocque, RH Lipson, W Jamroz, BP Stoicheff. Can J Phys 63:
1581, 1985.
48. CH Muller III, DD Lowenthal, MA DeFaccio, AV Smith. Opt Lett 13: 651, 1988.
49. KSE Eikema, J Walz, TW Hänsch. Phys Rev Lett 83: 3828, 1999.
50. RR Freeman, GC Bjorklund, NP Economou, PF Liao, JE Bjorkholm. Appl Phys
Lett 33: 739, 1978.
51. GHC New, JF Ward. Phys Rev Lett 19: 556, 1967; see also JF Ward, GHC New,
Phys Rev 185: 57, 1969.
52. AH Kung, JF Young, SE Harris. Appl Phys Lett 22: 301, 1973.
53. MHR Hutchinson, CC Ling, DJ Bradley. Opt Comm 18: 203, 1976.
54. H Langer, H Puell, H Rohr. Opt Comm 34: 137, 1980.
55. SA Meyer, GW Faris. Opt Lett 23: 204, 1998.
56. R Hilbig, R Wallenstein. Appl Opt 21: 913, 1982.
57. JP Marangos, N Shen, H Ma, MHR Hutchinson, JP Connerade. J Opt Soc Am B
7: 1254, 1990.
58. JH Glownia, DR Gnass, PP Sorokin. J Opt Soc Am B 11: 2427, 1994.
59. O Kittelmann, J Ringling, G Korn, A Nazarkin, IV Hertel. Opt Lett 21: 1159, 1996.
60. M Wittmann, MT Wick, O Steinkellner, P Farmanara, V Stert, W Radloff, G Korn,
IV Hertel. Opt Comm 173: 323, 2000.
61. CG Durfee III, S Backus, MM Murnane, HC Kapteyn. Opt Lett 22: 1565, 1997.
62. CG Durfee III, S Backus, HC Kapteyn, MM Murnane. Opt Lett 24: 697, 1999.
63. AJ Merriam, SJ Sharpe, H Xia, DA Manuszak, GY Yin, SE Harris. IEEE J Select
Topics Quantum Electron 5: 1502, 1999.
64. J Reintjes, C-Y She, RC Eckardt. IEEE J Quantum Electron QE-14: 581, 1978.
65. J Reintjes, LL Tankersley, R Christensen. Opt Comm 39: 334, 1981.

Copyright © 2002 by Taylor & Francis

 66. J Bokor, PH Bucksbaum, RR Freeman. Opt Lett 8: 217, 1983.
67. A McPherson, G Gibson, H Jara, U Johann, TS Luk, I McIntyre, K Boyer, CK
Rhodes. J Opt Soc Am B 4: 595, 1987.
68. M Ferray, A L’Huillier, XF Li, LA Lompré, G Mainfray, C Manus. J Phys B 21:
L31, 1988.
69. XF Li, A L’Huillier, M Ferray, LA Lompré, G Mainfray. Phys Rev A 39: 5751,
1989.
70. JJ Macklin, JD Kmetec, CL Gordon III. Phys Rev Lett 70: 766, 1993.
71. N Sarukura, K Hata, T Adachi, R Nodomi, M Watanabe, S Watanabe. Phys Rev
A 43: 1669, 1991.
72. K Miyazaki, H Sakai. J Phys B 25: L83, 1992.
73. JK Crane, MD Perry, S Herman, RW Falcone. Opt Lett 17: 1256, 1992.
74. A L’Huillier, Ph Balcou. Phys Rev Lett 70: 774, 1993.
75. A L’Huillier, M Lewenstein, P Salières, Ph Balcou, M Yu Ivanov, J Larsson, CG
Wahlström. Phys Rev A 48: R3433, 1993.
76. Y Nagata, K Midorikawa, M Obara, K Toyoda. Opt Lett 21: 15, 1996.
77. I Mercer, E Mevel, R Zerne, A L’Huillier, Ph Antoine, CG Wahlström. Phys Rev
Lett 77: 1731, 1996.
78. Z Chang, A Rundquist, H Wang, MM Murnane, HC Kapteyn. Phys Rev Lett 79:
2967, 1997.
79. Ch Spielmann, NH Burnett, S Sartania, R Koppitsch, M Schnürer, C Kan, M Len-
zner, P Wobrauschek, F Krausz. Science 278: 661, 1997.
80. Ch Spielmann, C Kan, NH Burnett, T Brabec, M Geissler, A Scrinzi, M Schnürer,
F Krausz. IEEE J Select Topics Quantum Electron 4: 249, 1998.
81. M Schnürer, Ch Spielmann, P Wobrauschek, C Streli, NH Burnett, C Kan, K Fer-
encz, R Koppitsch, Z Cheng, T Brabec, F Krausz. Phys Rev Lett 80: 3236,
1998.
82. A Rundquist, CG Durfee III, Z Chang, C Herne, S Backus, MM Murnane, HC
Kapteyn. Science 280: 1412, 1998.
83. C Altucci, R Bruzzese, C deLisio, M Nisoli, S Stagira, S DeSilvestri, O Svelto, A
Boscolo, P Ceccherini, L Poletto, G Tondello, P Villoresi. Phys Rev A 61: 021801,
1999.
84. P Villoresi, P. Ceccherini, L Poletto, G Tondello, C Altucci, R Bruzzese, C deLisio,
M Nisoli, S Stagira, G Cerullo, S DeSilvestri, O Svelto. Phys Rev Lett 85: 2494,
2000.
85. KS Budil, P Salières, A L’Huillier, T Ditmire, MD Perry. Phys Rev A 48: R3437,
1993.
86. NH Burnett, C Kan, PB Corkum. Phys Rev A 51: R3418, 1995.
87. FA Weihe, SK Dutta, G Korn, D Du, PH Bucksbaum, PL Shkolnikov. Phys Rev
A 51: R3433, 1995.
88. K Miyazaki, H Takada. Phys Rev A 52: 3007, 1995.
89. FA Weihe, PH Bucksbaum. J Opt Soc Am B 13: 157, 1996.
90. M Yu Ivanov, T Brabec, N Burnett. Phys Rev A 54: 742, 1996.
91. M Kakehata, H Takada, H Yumoto, K Miyazaki. Phys Rev A 55: R861, 1997.
92. Ph Antoine, A L’Huillier, M Lewenstein, P Salières, B Carré. Phys Rev A 53:
1725, 1996.

Copyright © 2002 by Taylor & Francis

 93. Ph Antoine, B Carré, A L’Huillier, M Lewenstein. Phys Rev A 55: 1314, 1997.
94. M Bellini, C Lyngå, A Tozzi, MB Gaarde, TW Hänsch, A L’Huillier, C-G Wahl-
ström. Phys Rev Lett 81: 297, 1998.
95. P Salières, T Ditmire, KS Budil, MD Perry, A L’Huillier. J Phys B 27: L217, 1994,
and refs. cited therein.
96. JH Eberly, Q Su, J Javanainen. J Opt Soc Am B 6, 1289 1989; Phys Rev Lett 62:
881, 1989.
97. W Becker, S Long, JK McIver. Phys Rev A 41: 4112, 1990.
98. JL Krause, KJ Schafer, KC Kulander. Phys Rev Lett 68: 3535, 1992; Phys Rev A
45: 4998, 1992.
99. KC Kulander, KJ Schafer, JL Krause. In: M Gavrila, ed. Atoms in Intense Radiation
Fields. New York: Academic Press, 1992.
100. KC Kulander, KJ Schafer, JL Krause. In: B Piraux, A L’Huillier, K Rza̧żewski,
eds. Super-Intense Laser-Atom Physics. NATO ASI Ser B, Vol 316. New York:
Plenum, 1993.
101. PB Corkum. Phys Rev Lett 71: 1994, 1993.
102. M Lewenstein, Ph Balcou, M Yu Ivanov, A L’Huillier, PB Corkum. Phys Rev A
49: 2117, 1994.
103. W Becker, S Long, JK McIver. Phys Rev A 50: 1540, 1994.
104. SC Rae, K Burnett, J Cooper. Phys Rev A 50: 3438, 1994.
105. M Lewenstein, KC Kulander, KJ Schafer, P Bucksbaum. Phys Rev A 51: 1495,
1995.
106. M Lewenstein, P Salières, A L’Huillier. Phys Rev A 52: 4747, 1995.
107. W Becker, A Lohr, M Kleber, M Lewenstein. Phys Rev A 56: 645, 1997.
108. C Kan, NH Burnett, CE Capjack, R Rankin. Phys Rev Lett 79: 2971, 1997.
109. MB Gaarde, Ph Antoine, A L’Huillier, KJ Schafer, KC Kulander. Phys Rev A 57:
4553, 1998.
110. W Becker, S Long, JK McIver. Phys Rev A 50: 1540, 1994.
111. E Priori, G Cerullo, M Nisoli, S Stagira, S DeSilvestri, P Villoresi, L Poletto, P
Ceccherini, C Altucci, R Bruzzese, C deLisio. Phys Rev A 61: 063801, 2000.
112. J Gao, F Shen, JG Eden. Phys Rev Lett 81: 1833, 1998.
113. J Gao, F Shen, JG Eden. J Phys B 32: 4153, 1999.
114. J Gao, F Shen, JG Eden. Phys Rev A 61: 043812, 2000.
115. J Gao, F Shen, JG Eden. Int J Mod Phys B 14: 889, 2000.
116. D-S Guo, GWF Drake. J Phys A 25: 3383, 1992.

Copyright © 2002 by Taylor & Francis

11
All-Solid-State, Short-Pulse,
Tunable, Ultraviolet Laser Sources
Based on Ce 3⫹-Activated Fluoride
Crystals
Zhenlin Liu
Japan Science and Technology Corporation, Kawasaki, Kanagawa,
Japan

Nobuhiko Sarukura
Okazaki National Research Institutes, Myodaiji, Okazaki, Japan

Mark A. Dubinskii
Magnon, Inc., Reisterstown, Maryland

I. INTRODUCTION

Ultraviolet (UV) tunable lasers have become the most important tool in many
fields of science and technology. Their most impressive applications include envi-
ronmental sensing, engine combustion diagnostics, semiconductor processing,
micromachining, optical communications, and medicinal and biological applica-
tions.
For example, the behavior of trace constituents in the earth’s upper at-
mosphere, governed by chemical, dynamic, and radiative processes, is of par-
ticular importance for the overall balance of the stratosphere and mesosphere. In
particular, ozone plays a dominant role by absorbing short-wavelength UV radia-
tion that might damage living organisms and by maintaining the radiative budget
equilibrium. Measurement of the total ozone column content and vertical profile
by a ground-based UV spectrometer network or by satellite-borne systems re-
mains the fundamental basis for global observations and trend analysis. Remote

Copyright © 2002 by Taylor & Francis

measurements of trace constituents using an active technique such as lidar have
been made possible by the rapid development of powerful tunable laser sources.
These have opened a new field in atmospheric spectroscopy by providing sources
that can be tuned to characteristic spectral features of atmospheric constituents
[1].
Recently, National Aeronautics and Space Administration (NASA) mis-
sions used tunable UV laser sources for atmospheric differential absorption lidar
(DIAL) measurements from the airplanes to analyze the global distribution of
O 3 radicals, which is directly relevant to the ‘‘ozone hole’’ and global climate
formation problems [2]. In the airborne UV DIAL system, two frequency-doubled
Nd :YAG lasers are used to pump two high-conversion-efficiency, frequency-
doubled, tunable dye lasers. NASA used the UV lasers in the wavelength region
from 289 to 311 nm.
Considerable effort has been devoted since the early 1970s to the develop-
ment of new vibronic crystals for tunable solid-state lasers. However, almost all
the efficient and successful vibronic materials, including Ti: sapphire [3], alexan-
drite [4] and, more recently, Cr 3⫹:LiCaAlF 6 (Cr:LiCAF) [5] and Cr 3⫹ :LiSrAlF 6
(Cr :LiSAF) [6] are emitting in the same near-infrared (IR) spectral region (700–
1000 nm).

A. UV Laser Systems Using Frequency Conversion

The existing commercially available tunable UV laser sources (comprising subse-
quent steps of nonlinear frequency conversion: doubling, tripling, and/or mixing
of tunable radiation obtained from the primary traditional tunable visible or near
infrared lasers) have been extremely complicated and expensive (also bulky, inef-
ficient, inconvenient, and unreliable for airborne measurements in flight or aboard
spacecraft applications [2] up to now. For example, UV-tunable lasers based on
dye-laser tunability, in addition to a pumping source (usually an Ar ⫹ ion laser
or an Nd:YAG laser with an attached nonlinear frequency-doubler), need a tun-
able dye laser in the orange–red region of the spectrum. The laser should be
provided with a dye circulation system as well as a system for frequency-doubling
of the dye laser radiation with a servotuning system (often called ‘‘Autotracker’’)
to follow the wavelength changes, and also a system for separation of the visible
and UV beams emanating from it. Solid-state tunable UV lasers using Ti:sap-
phire lasers with subsequent mixing of its tunable output with 532-nm pumping
radiation should also be supplied by ‘‘Autotracker’’ [7].
Recent development of frequency-tripled flash lamp-pumped Q-switched
Cr : LiSAF tunable laser holds promise for devising more reliable tunable solid-
state UV lasers. Cr: LiSAF is of particular interest due to its strong absorption
in the 670 nm wavelength region, allowing for potential pumping with AlGaInP/
GaAs diode lasers. Harmonic generation in the UV region from 260 to 320 nm

Copyright © 2002 by Taylor & Francis

was demonstrated using an LiB3O5 (LBO)/β-BaB2O4 (BBO) sum-frequency mix-
ing scheme. The maximum output was 6 mJ at the wavelength of 300 nm [8].
On the other hand, remarkable progress in high-power ultrashort pulse la-
sers has been made since chirped-pulse amplification [9] was applied to solid-
state lasers such as Ti:sapphire lasers along with the generation of ultrashort
pulses. These success, however, did not extend to UV-wavelength region, be-
cause the frequency upconversion limits the bandwidth, resulting in a long pulse
width and low conversion efficiency. For higher peak power or high average
power in the UV region, UV gain media with broad gain bandwidth are necessary.

B. Direct Generation of UV Coherent Light from Lasers

To provide tunable or ultrashort UV laser radiation in a reliable and efficient
way, the version with the best prospects at the moment would be to use directly
pumped solid-state UV active media, based on the electrically dipole-allowed
interconfigurational 5d–4f transitions of rare-earth ions in wide band-gap fluoride
crystals: YLiF 4 (YLF) [10], LaF 3 [11], and more recently LuLiF 4 (LLF) [12,13],
LiCaAlF 6 (LiCAF) [14–16], and LiSrAlF 6 (LiSAF) [17,18]. In fact, this is the
only version that also allows independent control of tunable radiation bandwidth,
or when necessary, even provides multiwavelength UV output in one laser beam
from the oscillator [12,13]. This version was proposed in 1977 from purely spec-
troscopic considerations [19] and then confirmed experimentally on Ce 3⫹ ion in
UV region by Ehrlich et al. [10,11] and on Nd 3⫹ ion in the VUV region by
Waynant [20].
As early as 1977, Yang and Deluca proposed a simple way of implementing
a tunable laser capable of producing radiation directly in the UV and even VUV
spectral ranges [19]. For this purpose, they proposed to use interconfiguration
5d–4f transitions of rare-earth ions in wide band-gap dielectric crystals. Because
of the strong lattice interaction with 5d electrons, the fluorescence that results
from 5d to 4f transitions of trivalent rare-earth ions in solid hosts is characterized
by broad bandwidths and large Stokes shifts. Such fluorescence is particularly
attractive for the development of tunable lasers. Powder samples of Ce 3⫹ :LaF 3
and Ce 3⫹ : LuF 3 were excited by the 253.7 nm radiation transmitted through a
narrow-band interference filter inserted in front of an Hg lamp source. Broad-
band UV fluorescence was reported for Ce 3⫹ :LaF 3 (276–312 nm) and Ce 3⫹ :LuF 3
(288–322 nm). The fluorescence quantum yields account for the fact that not all
of the atoms raised to the pump bands subsequently decay to the upper laser
level. Some of these atoms can in fact decay from the pump bands straight back
to the ground state or perhaps to other levels, which are not useful. The pump
quantum efficiency or fluorescence quantum yields η q(λ) are defined as the ratio
of the number of atoms that decay to the upper laser level to the number of
atoms raised to the pump band by a monochromatic pump at wavelength λ. The

Copyright © 2002 by Taylor & Francis

fluorescence quantum yields of LaF 3 :1%Ce 3⫹ and LuF 3 :0.1%Ce 3⫹ are 0.9 and
0.82, respectively. Estimates of the threshold power for lasing action suggested
that a laser system tunable from 276 to 322 nm is feasible with noble-gas–halide
lasers as pumping sources [19].
After that, Ce 3⫹ : Y 3 Al 5 O 12 (YAG) was investigated as a model system for
a 5d–4f solid-state tunable laser [21]. This system was chosen since YAG had
been extensively studied as a laser host and good-quality crystals were readily
available. Despite providing apparently adequate conditions to achieve stimulated
emission, it was unable to detect laser action in Ce 3⫹ : YAG. It was found that
there was strong excited-state absorption (ESA) in this material at the wave-
lengths of its fluorescence. The ESA was sufficiently strong to quench completely
any possible laser action. The crystal showed a net optical loss instead of optical
gain at the wavelength of the fluorescence transition. This self-absorption may
explain the failure of all attempts to obtain stimulated emission in this material.
A laser of the type, which has a 5d–4f transition, was originally imple-
mented with Ce 3⫹ :YLiF 4 (Ce: YLF) as a laser medium [10]. It should be noted
that Ce 3⫹ ions are the most promising activators for the UV spectral range. How-
ever, in spite of a large number of studied Ce-activated materials [22,23], the
investigations performed before 1992 revealed only two laser-active media
[10,11].
In 1979, Ehrlich et al. reported the first observation of stimulated emission
from a 5d–4f transition in triply ionized rare earth-doped crystal Ce: YLF, opti-
cally pumped at 249 nm, and emitted at 325.5 nm [10]. Since the Ce 3⫹ ion has
only one electron in the 4f state, the impurity energy levels of Ce-doped crystals
are particularly simple. The ground state is split into a 2 F 5/2 and a 2 F 7/2 levels by
the spin-orbit interaction. The first excited state is a 5d state, which interacts
strongly with the host lattice because of the large spatial extent of the 5d wave-
function. Thus the crystal-field interaction dominates over the spin-orbit interac-
tion and the 5d state splits into four levels as a result of the S 4 site symmetry
for the rare-earth ion in YLF.
The broad absorption bands that peak at 195, 205, 240, and 290 nm result
from transitions from the 4f ground state to the crystal-field split 5d levels of the
Ce 3⫹ ion. The fluorescence spectrum has two peaks, the result of transitions from
the lowest 5d level to the two spin-orbit ground states 4f(2 F 5/2) and 4f(2 F 7/2). The
40 ns radiative lifetime of the 5d level results from the electric-dipole-allowed
character of the 5d → 4f transition. The potential tuning range of the Ce: YLF
laser estimated from the half-power points of the fluorescence spectrum is from
305 to 335 nm. The maximum output energy observed was ⬃1 µJ in a pulse
width of 35 ns for an absorbed pump energy of 300 µJ.
However, the operation of the Ce: YLF laser was hampered by several poor
performance characteristics. These include an early onset of saturation and rolloff

Copyright © 2002 by Taylor & Francis

in the above-threshold gain and power output as well as a drop in the output for
pulse repetition rates above 0.5 Hz. It has been shown that an excited-state ab-
sorption of the UV pump light is responsible for a photoionization of the Ce 3⫹
ions, which in turn leads to the formation of transient and stable color centers.
The color centers have a deleterious effect on the lasing characteristics of Ce:
YLF since they absorb at the cerium emission wavelengths. The growth and
relaxation of these centers influence the gain saturation and pump rate limitation
of the Ce: YLF laser [24,25]. This experiment is of historic significance, but it
is short of practical use due to the existence of solarization.
In 1980, the same group mentioned above reported the operation of an
optically pumped Ce 3⫹ :LaF 3 laser [11]. Because of the rapid internal relaxation
to the lowest 5d state, the fluorescence spectrum was not noticeably different for
ArF (193 nm), KrF (248 nm), or frequency-doubled Ar ⫹ ion laser (257 nm) pump-
ing. The fluorescence lifetime was likewise identical for 248 nm or 193 nm excita-
tion. For the 0.05%-doped crystal, the lifetime was 18 ⫾ 2 ns. The approximate
potential tuning range is from 275 to 315 nm. The output of a small commercial
excimer laser, producing 40 mJ at 248 nm or 10 mJ at 193 nm in a 25 ns full width
at half maximum (FWHM) pulse, was used for optical pumping. The primary
difficulties encountered (i.e., low output power and high threshold) can probably
be ascribed to initial difficulties in crystal growth.
Subsequent studies showed that the attempts to find laser-active media
among Ce-activated materials failed because of absorption from the excited state
of the 5d configuration of Ce 3⫹ ions in Ce 3⫹ :YAG [26,27] and Ce 3⫹ :CaF 2 [28],
formation of stable or transient color centers in Ce-activated samples [23,24,28],
and other complicated processes occurring in such media under the high-power
UV pumping.
The spectrally broad vibronic emission bands in impurity-doped solids
serve as the basis for wavelength-tunable laser operation in these materials. Be-
cause of the broad emission and absorption bands, however, these materials are
susceptible to ESA, which can significantly reduce the performance characteris-
tics of the laser materials. The ESA is a two-step process, by which the first
photon absorbed promotes an electron from the 4f ground state to the lowest 5d
state of the trivalent cerium ion. Within the lifetime of this excited 5d state, a
second photon is absorbed, which then photoionizes the ion by promoting that
electron to the conduction band. The free electron subsequently traps out at an
electron acceptor site, forming a stable color center. These color centers are ab-
sorptive at the wavelengths for stimulated emission of the trivalent cerium ions,
and hence they serve as a quenching mechanism for laser gain in this crystal.
The color centers produced are photochromic in that they can be optically
bleached [23]. Over a long period of time, difficulties in overcoming these prob-
lems, which are inherent in well-known materials used for producing UV light

Copyright © 2002 by Taylor & Francis

[10,11], made investigators believe that this scheme of UV lasers was of little
promise.
However, recent investigations showed that by an appropriate choice of
activator–matrix complexes and active medium–pump source combinations, one
can create efficient tunable lasers using d–f transitions in Ce 3⫹ ion in the UV
spectral range [12–16] and in Nd 3⫹ ion in the VUV range [29]. Furthermore,
such lasers proved to be stable under the pumping.
In 1992, Dubinskii et al. reported UV laser medium, Ce 3⫹ :LuLiF 4 (Ce:
LLF) crystal, which can be pumped by KrF excimer laser [12,13]. This crystal
has almost the same optical properties as Ce :YLF. But Ce :LLF has a smaller
solarization effect, and it shows more promise in practical use. In 1993, the same
group reported Ce 3⫹ :LiCaAlF 6 (Ce:LiCAF) crystal, which can be pumped by
the fourth harmonic of Nd: YAG lasers. No solarization effect of this crystal was
observed [14–16].
Ce 3⫹ :LiSrAlF 6 (Ce:LiSAF) crystal was reported in 1994, and it also can
be pumped by the fourth harmonic of Nd:YAG lasers [17,18]. It has similar laser
properties to Ce :LiCAF crystal.
Since these new Ce-doped crystals were reported, studies on the solid-state
ultraviolet tunable lasers have become popular again. All these new Ce-doped
crystals have a broad gain-bandwidth in the ultraviolet region, which is especially
attractive for ultrashort-pulse generation and amplification. Figure 1 shows the
tunable wavelength regions of the five known Ce-doped laser crystals.

Figure 1 Various tunable lasers in ultraviolet region. Solid lines and dots indicate the
confirmed tunable wavelength region; dotted lines show potential tunable wavelength re-
gion.

Copyright © 2002 by Taylor & Francis

II. NEW UV LASER MATERIALS: Ce: LLF AND Ce:LiCAF
CRYSTALS
A. Basic Properties of Ce: LLF Laser Medium
Ce :LLF is a tunable solid-state laser material in the UV region that was first
reported in 1992 [12,13]. The choice of this material for the experiments in UV
is more reasonable because of its structural and chemical similarity to comprehen-
sively studied and commercially grown YLF single crystal. The LLF crystals
belong to the scheelite structural type. Similar to all rare-earth ions, Ce 3⫹ ions
take part in activation, substituting for Lu 3⫹ ions in the position with the point
group S 4 .
Ce :LLF has a potential tuning region of 305–340 nm, so it is especially
attractive for use in spectroscopy of wide band-gap semiconductors for blue laser
diodes, such as GaN [30]. The fluorescence spectrum of Ce :LLF crystal has two
peaks at 311 nm and 328 nm (Fig. 2). KrF excimer laser (248-nm) can be used
as the pumping source for Ce: LLF crystal. The fluorescence lifetime was found
to be about 40 ns.

B. Gain Spectrum of Ce: LLF crystal

A Ce: LLF single crystal was grown from a carbon crucible using the Bridgman-
Stockbarger method in a fluorinated atmosphere by Dubinskii et al. [31,32]. A

Figure 2 (a) π-Polarized absorption and normalized fluorescence (b: π-polarization, c:

σ-polarization) spectra of Ce: LLF single crystal.

Copyright © 2002 by Taylor & Francis

seeded crucible was used to obtain oriented boule. The seed orientation was such
that the c-axis was perpendicular to the cylindrical axis of the crucible. The corre-
sponding orientation of c-axis of grown crystal was checked by observing the
conoscopic picture of the boule. The sample obtained contains about 0.5 at.%
Ce 3⫹ ions.
The sample for the amplification experiment (1 cm long and 5 mm diame-
ter) was cut cylindrically. A flat window was polished on the side of the cylinder
parallel to its axis. The end surfaces were polished parallel within 5 arcmin. The
sample was oriented in such a manner that the c-axis of the crystal was parallel
to the side window and perpendicular to the axis of the cylinder. The pump and
probe experiment was done in order to measure the gain spectrum for the 1 cm
long Ce :LLF crystal under the randomly polarized KrF excimer laser excitation
with side-pumping geometry (Fig. 3).
The probe beam was the second harmonic of a tunable mode-locked DCM
(4-Dicyanmethylene-2-methyl-6-(p-dimethylaminostyryl)-4H-pyran) dye laser,
which was synchronously pumped by the second harmonic of a cw mode-locked
Nd:YAG laser. The tuning range of the DCM dye laser is from 645 to 680 nm.
The second harmonic of the DCM laser was generated using an LBO nonlinear
crystal in type I phase-matching condition. When the wavelength of the DCM dye
laser was changed, it was necessary to rotate the LBO crystal to satisfy the phase-
matching condition. The pumping source to excite Ce:LLF crystal was a conven-
tional randomly polarized KrF excimer laser. The pumping pulses were focused
onto the side window of the Ce:LLF crystal by a cylindrical lens.
A biplanar phototube was used to detect the signal. A Tektronix 7934 stor-
age oscilloscope was used to display the gain traces. The data were taken for
two times (with and without probe beam) and saved on the oscilloscope screen.
Then the amplified pulse train and the fluorescence pulse train could be seen at

Figure 3 Gain spectrum measurement using the second harmonic of a mode-locked

DCM dye laser as a probe.

Copyright © 2002 by Taylor & Francis

the same time. The difference between the upper signal and the lower signal will
be the net gain. With the use of the mode-locked pulse train as a probe beam,
the fluorescence background is easily identified.
The probe beam was σ-polarized (the polarization is perpendicular to the
optical axis or c axis of the Ce: LLF crystal). In order to check the gain-polariza-
tion dependence, the polarization direction of the DCM dye laser was changed
by rotating a half-wave plate. No noticeable anisotropy of the gain was observed
within the accuracy of the measurements (at least for 325 nm probe radiation).
The gain was demonstrated from 323 nm to 335 nm (Fig. 4), due to the available
probe laser tunability. This gain bandwidth is large enough for amplification of
tunable femtosecond pulses.

C. Small-Signal Gain and Saturation Fluence of Ce:LLF

Crystal
Small-signal gain and saturation fluence are very important parameters for de-
signing lasers [31]. The small-signal gain and saturation fluence of Ce: LLF were
evaluated using the second harmonic (325 nm) of a nanosecond DCM-dye laser
as a probe as shown in Figure 5. A KrF excimer laser was used to pump the
Ce :LLF from its side window. The frequency conversion scheme used an LBO
nonlinear crystal to double the dye laser output. The DCM dye laser was pumped
by the second harmonic of a Q-switched Nd :YAG laser. The relative timing of
the probe dye laser and the pumping excimer laser were controlled by a synchro-
nizer to obtain the largest possible gain. The single-pass gain in the small-signal
region (⬃1 mJ/cm 2) was over 6-dB/cm (4.3 times) with 0.5-J/cm 2 pumping flux
(Fig. 6).
The 10 ns pulsewidth of the probe light was short enough in comparison
with a 40 ns fluorescence lifetime of Ce :LLF. Therefore, the evaluation of the

Figure 4 Gain spectrum of Ce :LLF pumped by randomly polarized KrF laser using
the second harmonic of a picosecond CW mode-locked DCM dye laser as a σ-polarized
probe (pumping fluence ⬃ 0.1 J/cm 2).

Copyright © 2002 by Taylor & Francis

Figure 5 Small-signal gain and saturation fluence were evaluated using the second har-
monic (325 nm) of a nanosecond DCM dye laser as a probe.

Figure 6 Gain dependence of Ce: LLF on pumping power at 325 nm. The single-pass
gain in the small-signal region was over 6 dB/cm (4.3 times) with 0.5 J/cm 2 excitation
fluence.

Copyright © 2002 by Taylor & Francis

saturation fluence and emission cross section is possible under the assumption
of the Frantz-Nodvik relation modeled for slow-decay gain medium [33].
G ⫽ E s log e {1 ⫹ G 0 [exp(E i /E s) ⫺ 1]}/E i
where E i is the input fluence, E s is the saturation fluence, G 0 is the single-pass
small-signal gain, and G is the gain corresponding to E i . The saturation fluence
of Ce :LLF, assuming the Frantz-Nodvik relation, was estimated to be 50 mJ/
cm 2 at 325 nm (Fig. 7), which is about two orders of magnitude higher than that
of organic dyes (⬃1 mJ/cm 2). The emission cross section estimated from this
saturation fluence was 10 ⫺17 cm 2 by the relation of E s ⫽ hν/σ. These results
indicate a high potential for Ce :LLF as a power amplification medium.

D. Basic Properties of Ce: LiCAF Laser Medium

In contrast to excimer laser-pumped UV solid-state laser media, such as Ce: LLF
and Ce: YLF, the Ce :LiCAF crystal is the first known tunable UV laser directly
pumped by the fourth harmonic of a standard Nd:YAG laser [14–16].
The samples of Ce :LiCAF were also grown in a fluorinated atmosphere
using the Bridgeman-Stockbarger technique from carbon crucibles by Dubinskii
et al. The crystals grown had the colquiriite structure and the space group P31c.
The nonpolarized absorption spectrum of a 2.3 mm thick Ce :LiCAF sample,
containing about 0.1% of Ce 3⫹ ions, is shown in Figure 8a. Due to the first 4f–
5d absorption peak with a half width of 3000 cm ⫺1 centered at about 37,000

Figure 7 Ce : LLF gain dependence on input fluence of 10-ns, 325-nm probe pulses.
The dotted line indicates the gain saturation curve assuming the Frantz-Nodvik relation,
which was fitted by the least square method. The saturation fluence and the small signal
gain were fitted to be 50 mJ/cm 2 and 5.1 (pumping fluence ⬃ 0.5 J/cm 2).

Copyright © 2002 by Taylor & Francis

Figure 8 (a) Nonpolarized absorption spectrum of a Ce: LiCAF sample (0.1 at.%; 2.3
mm in length). (b) Nonpolarized fluorescence spectrum of Ce: LiCAF (0.9 at.%). (c) Sin-
gle-pass small signal gain dependence on the probe beam wavelength for a Ce: LiCAF
sample (0.9 at.%; 2.3 mm in length). (From Ref. 34.)

cm ⫺1, Ce :LiCAF can be pumped by the fourth harmonics of various commer-

cially available Nd-lasers (e.g., YAG, YAP, YLF, and GSGG). The Ce: LiCAF
fluorescence spectrum (Fig. 8b) displays the nearly two-humped shape character-
istic of Ce 3⫹ ions in most known hosts, due to the allowed 5d–4f transitions
terminating at the 2 F 7/2 and 2 F 5/2 components of the spin–orbit split ground term.
Ce :LiCAF has a potential tuning range from 280 to 320 nm. Ce :LiCAF has
sufficiently higher effective gain cross section (6.0 ⫻ 10 ⫺18 cm 2) compared with
Ti: sapphire [3], which is favorable for designing laser oscillators. Ce :LiCAF
has also larger saturation fluence (115 mJ/cm 2) [18] than organic dyes, which is
attractive for designing power amplifiers. The fluorescence lifetime was reported
to be 30 ns, which is too short for constructing regenerative amplifiers. However,
it is long enough for designing multipass amplifiers. The nonpolarized small-
signal single-pass gain dependence on the probe beam wavelength for a 2.3-mm
thick Ce :LiCAF sample with Ce 3⫹ ion concentration of 0.9 at.% is shown in
Figure 8c. The sample optical axis orientation with respect to the direction of
observation in this experiment was the same as for obtaining the fluorescence
spectra (Fig. 8b). The pump fluence of 0.3 J/cm 2 was used to obtain the above
dependence. The probe fluence was less than 1 mJ/cm 2. From a comparison of
Figure 8b and 8c, it is evident that the small-signal gain curve shape is similar
to the fluorescence spectrum observed. This means that induced absorption is
small in the gain spectral region. The small-signal gain reaches a value of 2.5
in the vicinity of the main fluorescence peak [34].

Copyright © 2002 by Taylor & Francis

III. GENERATION OF SUBNANOSECOND PULSES FROM
Ce 3ⴙ-DOPED FLUORIDE LASERS

Direct ultrashort pulse generation has not been obtained from ultraviolet solid-
state lasers as it has for near-infrared tunable laser materials like Ti:sapphire and
Cr :LiSAF crystals. This is due to the difficulty in obtaining continuous-wave
(cw) UV laser operation, which is required for Kerr lens mode-locking (KLM)
schemes utilizing spatial or temporal Kerr type nonlinearity [35,36].

A. Short Pulse Generation with Simple Laser Scheme

A general technique for subnanosecond pulse generation from laser-pumped dye
laser has been described [37]. The technique makes use of the resonator tran-
sients, are in the form of damped relaxation oscillation or ‘‘spiking.’’ These reso-
nator transients are the consequences of the interaction between the excess popu-
lation inversion and the photons in the cavity. Their durations can be controlled
by proper choices of photon cavity decay time and pumping level. The decay
lifetime (photon lifetime) t c of a cavity mode is defined as

dE/dt ⫽ ⫺E/t c (1)

where E is the energy stored in the mode. If the fractional (intensity) loss per
pass is L and the length of the resonator is l c, then the fractional loss per unit
time is cL/nl c ; therefore,

t c ⫽ ⫺nl c /cL (2)

For the case of a resonator with mirrors’ reflectivities R 1 and R 2,
nl c
tc ⬇ (3)
c(1 ⫺ √R 1 R 2)
The quality factor of the resonator is defined universally as:
Q ⫽ ωE/P ⫽ ⫺ωE/(dE/dt) (4)

where E is the stored energy, ω is the resonant frequency, and P ⫽ ⫺dE/dt is

the power dissipated. By comparing Eqs. (4) and (1), we obtain

Q ⫽ ωt c (5)

Let us consider a laser resonator with two mirrors pumped by a Q-switched

laser. Normally the pulse duration from the Q-switched laser is approximately
10 ns. During the pump pulse, inversion will build up in the laser active medium;
after threshold inversion is reached, a delayed laser pulse will develop. The pro-
posed method of short-pulse generation is based on the fact that this pulse may

Copyright © 2002 by Taylor & Francis

be shorter than the pump pulse, due to the transient characteristics of the laser
oscillator.
The transient behavior of such a laser can be understood by treating numeri-
cal examples with a computer. The laser is adequately described by the well-
known rate equations for four-level systems as follows:
dn/dt ⫽ W(t) ⫺ Bnq (6)
dq/dt ⫽ Bnq ⫺ q/t c (7)
where n is the population inversion (n 2 ⫺ n 1) and nearly equal to the upper state
population n 2, W(t) is the pumping rate, B is the Einstein B coefficient for stimu-
lated emission, q is the total number of photons in the cavity, and t c is the resona-
tor (photon) lifetime.
The pumping rate was assumed to have the form of a gaussian pulse with
full half-width T 1 and integrated photon number N. A large reduction of the over-
all pulse duration demands a high value of T 1 /t c. The ratio T 1 /t c can easily be
made large, since the laser resonator needs no switching elements and therefore
need not be longer than the material itself. Furthermore, it is possible to use a
laser resonator of low mirror reflectivity, whose resonator lifetime is not markedly
longer than the single-pass transit time l c /c. To obtain short single pulses, it is
necessary to use resonator lifetimes that are small compared to the pump duration
in combination with controlling the level of pumping [38].

B. Subnanosecond Ce:LLF Laser

1. Subnanosecond Ce: LLF Laser Pumped by KrF Excimer
Laser
Based on the technique mentioned above, a subnanosecond Ce: LLF laser was
made with a low-Q, short-cavity configuration [39]. The experimental setup is
simple. A Ce :LLF sample (2.5 cm long, 5 mm diameter) was cut cylindrically.
A flat window was polished on the side of the cylinder parallel to its axis. The
sample was oriented so that the c-axis of the crystal was parallel to the side
window and perpendicular to the axis of the cylinder. No dielectric coatings were
deposited on the polished surfaces.
The resonator was formed by a 30%-transmission flat output coupler and the
surface reflection of the crystal (⬃5% reflectivity) as shown in Figure 9 (t c ⬃ 0.2
ns ⬍⬍ τ p ⫽ 10 ns). The cavity spacing is 3 cm. KrF laser pumping pulses of 40
mJ, 10 ns, 1 Hz were focused on the side window by a 20 cm focal-length cylindri-
cal lens. A single 3 µJ pulse from the side of 30% transmission output coupler
was obtained at the sacrifice of the extracted energy. The pulse duration observed
by a streak camera was 950 ps. The pulse did not have any satellite pulses because
the pulse duration was longer than the cavity round trip time (200 ps).

Copyright © 2002 by Taylor & Francis

Figure 9 Low-Q, short-cavity Ce: LLF laser. The cavity is formed by a 30% transmission
flat output coupler and the surface reflection of the crystal. The cavity spacing is 3 cm.

At higher pumping levels, the pulse develops double or triple peaks in the
form of damped relaxation oscillation. The pumping energy dependence of the
output performance has been investigated. As shown in Figure 10, single-pulse
operation of 1 ns duration was observed for a sufficiently wide pumping energy
region from the lasing threshold up to 60 mJ pumping energy. Above this pump-

Figure 10 Pumping-energy dependence of output performance for two Ce: LLF lasers
with a 70% reflection end mirror (squares: 20%-reflection output coupler; circles: 5%
crystal-surface reflection coupler). Single-pulse operation was achieved in wide pumping-
energy region.

Copyright © 2002 by Taylor & Francis

ing level, multiple-pulse operation was observed. For the higher-Q cavity con-
figuration (the cavity is formed by a 70% reflection end mirror and a 20% reflec-
tion output coupler), a wide single-pulse operation region was obtained with a
pulse duration of few nanoseconds. In these experiments, the output energy in-
creases linearly with the pumping energy.
Figure 11 shows a typical single-pulse operation under the appropriate
pumping energy (Fig. 11a) and a typical multiple pulse operation under higher
pumping energy (Fig. 11b). The laser pulse was observed using a Tektronix 7934
storage oscilloscope with a biplanar phototube as the detector.
A 3 cm cavity-length oscillator composed of a flat high-reflection mirror
and a 70% transmission flat output coupler was also tested. KrF laser pumping
pulses 50 of mJ, 10 ns, 1 Hz were focused on the side window by a 20 cm focal-
length cylindrical lens up to 200 mJ/cm 2 pumping fluence. A 2 mJ output at 325
nm was obtained. The obtained pulse had multiple temporal spikes.
From these results, it can be concluded that the conditions for single-pulse
operation are not critical, either to the pumping energy or to the coupling constant.
In this way, a single short pulse can be easily generated from a low-Q, short-
cavity Ce :LLF laser pumped by a standard 10 ns KrF laser.

2. Subnanosecond Ce: LLF Laser Pumped by the Fifth

Harmonic of Nd: YAG Laser
Figure 12 indicates the absorption spectrum (nonpolarized) of a Ce: LLF sample
containing 1 at.% of Ce 3⫹ ions (in the melt). Considering the Ce: LLF absorption,
one can see that while this material does not noticeably absorb at 266 nm, it will
absorb pumping radiation at the fifth harmonic of Nd: YAG laser wavelength

(a) (b)

Figure 11 (a) Typical single pulse operation under appropriate pumping energy. (b)
Typical multiple pulse operation under higher pumping energy.

Copyright © 2002 by Taylor & Francis

Figure 12 Nonpolarized absorption spectrum of Ce 3⫹ ions in LiLuF 4 single crystal (Ce-
rium concentration in melt: 1 atomic %).

(213 nm) almost as efficiently as at the KrF-excimer pumping wavelength of 248

nm [40]. With the recent development of new nonlinear borate crystal materials
such as CsLiB 6 O 10 [41] and Li 2 B 4 O 7 [42], the fifth harmonic of Nd:YAG lasers
has become usable and practical due to the significantly improved efficiency and
stability, close to those typical of the fourth harmonic of Nd:YAG laser.
From Fig. 12, it is evident that even the off-peak absorption at 213 nm is
strong enough to consider the crystal suitable for an all-solid-state approach using
the fifth harmonic of an Nd: YAG laser for side pumping. To prove this, the Ce :
LLF was tested under side pumping conditions by the fifth harmonic of an Nd :
YAG laser. The 213 nm, 25 mJ, 5 ns, horizontally polarized pulses (σ-pumping)
for pumping Ce: LLF laser were stably obtained in Li 2 B 4 O 7 crystal using the
type-I sum frequency generation process between 1064 nm and 266 nm pulses
from a conventional Q-switched (Q-sw) Nd:YAG laser [42]. The optical layout
for a short-cavity Ce: LLF laser is shown schematically in Figure 13. A Ce:LLF
single crystal with 0.2 at.% doping level was cut to form a cylinder (5 mm diame-
ter and 25 mm length) with a flat polished window on the side. The sample was
oriented so that its optical axis was parallel to the side window and perpendicular
to the cylinder longitudinal axis. No antireflection coatings were applied to the
rod ends for this experiment. To increase the efficiency of side pumping, the
effective pumping penetration depth was geometrically reduced using the novel
tilted-incidence-angle side-pumping scheme instead of conventional normal-inci-
dence side pumping. Using a 20 cm focal length cylindrical lens that was also
tilted to be parallel to the side window of the laser crystal, the pumping pulse
was focused down to a 1.2 ⫻ 0.15 cm 2 line-shaped area to provide the 140 mJ/

Copyright © 2002 by Taylor & Francis

Figure 13 Short-cavity, tilted-incidence-angle (approximately 70 degrees) side-pump-
ing, tunable Ce : LLF laser layout. The pumping pulse was focused by a 20 cm focal length
cylindrical lens tilted so that the cylinder axis is parallel to the side window of the laser
crystal.

cm 2 pumping fluence at nearly 70 degree incidence angle. Observed efficient

penetration depth was estimated to be ⬃1 mm. In a conventional normal-inci-
dence side pumping scheme, the pumping penetration depth was over 3 mm for
the Ce :LLF crystal used here, which is too deep to obtain a good output beam
pattern. Obvious advantages of the above-mentioned tilted-incidence-angle, side-
pumping scheme are very simple: focused geometry, reduced pumping fluence
at the rod surface, reduced risk of damaging optics, and better matching of the
excited rod volume and the laser cavity mode volumes (similar to the coaxial
pumping scheme). This better matching resulted in a better output beam quality.
The low-Q, short-cavity Ce: LLF laser consisted of a Littrow prism used as an
end mirror, and a low-reflection (20%) flat output coupler. The total length of
the laser cavity was 6 cm. Typical spectrally and temporally resolved streak-
camera images of the Ce :LLF laser output pulse are shown in Figure 14. Using
the 213 nm, 5 ns, 16 mJ pumping pulses, we were able to obtain 880 ps, 77 µJ,
σ-polarized, and satellite-free reproducible pulses at 309 nm. It is worth men-
tioning here that pumping at 213 nm does not cause noticeable laser rod solariza-
tion or laser performance degradation during several hours of continuous opera-
tion at a 10 Hz repetition rate.

C. Short Pulse Generation from Ce: LiCAF Laser

1. Nanosecond Pumping
Ce :LiCAF is a tunable UV laser medium that can be directly pumped by the
fourth harmonic of a standard Nd:YAG laser. A 1% doped, 5 mm cubic Ce:
LiCAF sample was used without any dielectric coatings on the polished surfaces

Copyright © 2002 by Taylor & Francis

Figure 14 Temporally and spectrally resolved streak camera image of a UV short pulse
from the low-Q, short-cavity Ce: LLF laser.

[43]. The experimental setup of the subnanosecond Ce: LiCAF laser is shown in
Figure 15. The 1.5 cm long laser cavity was formed by a flat high-reflection
mirror and an 80% transmission flat output coupler. The 20 mJ, 10 ns, 1 Hz, 266
nm, horizontally polarized pumping pulses (the fourth harmonic of a conventional
10 ns Q-sw Nd: YAG laser) were focused longitudinally from the high-reflection

Figure 15 Experimental setup of the low-Q, short-cavity Ce :LiCAF oscillator with

nanosecond pumping.

Copyright © 2002 by Taylor & Francis

mirror side by a 30 cm focal-length lens with ⬃300 mJ/cm 2 pumping fluence
inside the active medium. The c-axis of the Ce: LiCAF laser crystal sample was
parallel to the direction of the pumping polarization. The absorbed energy was
5 mJ. The single-output pulse has energy of 45 µJ. The pulse duration was mea-
sured to be 600 ps using a streak camera. In this simple way, subnanosecond
pulse can be generated from an all-solid-state Ce: LiCAF laser pumped by a 10
ns Q-sw Nd :YAG laser.

2. Picosecond Pumping
Six hundred picosecond pulses have been generated from a low-Q, short-cavity
Ce :LiCAF laser pumped by the fourth-harmonic of a Q-sw 10 ns Nd: YAG laser
above. For shorter pulse generation, a shorter pumping source was tried: the
fourth harmonic of a mode-locked Nd: YAG oscillator and regenerative amplifier
system operated with the repetition rate of 10 Hz [44]. A low-Q, short-cavity
Ce :LiCAF oscillator was formed by a flat high-reflection mirror and a 30% re-
flection flat output coupler. The cavity length was 1.5 cm. A 10 mm Brewster-
cut, 1% doped (in the melt) Ce :LiCAF crystal was used. A typical spectrally
and temporally resolved streak camera image of the output pulse of the Ce:
LiCAF laser is shown in Figure 16. The Ce :LiCAF laser pulse width was mea-
sured to be 150 ps; the pumping pulse width was 75 ps.

Figure 16 Streak camera trace of the pulse from the low-Q, short-cavity Ce: LiCAF
laser with picosecond pumping.

Copyright © 2002 by Taylor & Francis

IV. PASSIVE SELF-INJECTION-SEEDED PULSE TRAIN
LASERS

Liu [45,46] and, independently, Ewart [47] presented a novel and useful tech-
nique for the production of nanosecond and subnanosecond high-power laser
pulses in a stable, reproducible, and efficient way: the so-called self-injection
technique. Brito Cruz et al. obtained the single or double, highly stabilized nano-
second pulses with large efficiency by the application of self-injection and cavity
dumping techniques [48]. They also demonstrated the high-power (0.9 GW) pico-
second pulse generation by intracavity nonlinear compression in a self-injected
Nd :YAG laser [49]. The technique employs the principle of regenerative ampli-
fication of a seed pulse. It differs from previous work [50] using a regenerative
amplifier in that it does not require an external oscillator to produce the seed
pulse. As a result, the operation is greatly simplified.

A. Passive Self-Injection-Seeded Pulse Train Laser

Scheme
Passive self-injection-seeded pulse train laser (SSPT laser) employs the principle
of subnanosecond pulse generation, making use of the resonator transient de-
scribed in Sec. III, and the self-injection, regenerative amplification process.
There is no Pockels cell in our laser cavity and no timing controller, so it is
completely passive.
Figure 17 shows conceptual design of the passive SSPT laser. The laser
consists of a pulse-seeding laser cavity and a feedback laser cavity. In the pulse-

Figure 17 Conceptual design for the self-injection-seeded pulse train laser. The laser
consists of a pulse-seeding laser cavity and a feedback laser cavity. A single short pulse
is generated in the pulse-seeding laser cavity. The short pulse is injected into the feedback
laser cavity and reflected from the end mirror, and then amplified regeneratively in the
same gain medium.

Copyright © 2002 by Taylor & Francis

seeding laser cavity, a single short pulse is generated without any deep gain
quenching. The seeding laser can generate short pulse even without any coupling
with the feedback laser cavity part. This is an advantage over the previously
reported active self-injection seeding scheme, which cannot operate without the
coupled cavity. The short pulse is injected into the feedback laser cavity, reflected
from the end mirror, and then amplified in the gain medium. Some portion of
the amplified pulse will be coupled out; the other part will be injected into the
feedback laser cavity again. This regenerative amplification continues until the
gain is quenched completely. Such a cavity is, therefore, expected to generate
pulse trains similar to those from Q-switched mode-locked lasers, even though
the generation mechanism differs.
There is no requirement for CW-operation capability or an external short-
pulse seeding laser [51] to generate such short pulse train. The pulse duration is
determined by the seeding pulse, and the pulse separation can be controlled by
the length of the feedback laser cavity as in a mode-locked laser (Fig. 18).
The operation conditions for the passive SSPT scheme are as follows. The
effective lifetime (T Geff ) of the gain medium, considering the excitation
pulsewidth (T ex) and relaxation time (T 1), should exceed the round-trip time (T rFB)
of the feedback laser cavity. A single pulse should be generated from the pulse-
seeding laser cavity without any deep gain quenching. The intracavity fluence
of the seeding pulse (F SP) should be much smaller than the saturation fluence
(F sat) of the gain medium. The round-trip time (T rPS) of the pulse-seeding laser
cavity should be shorter than the duration of seeding pulse (T p), to avoid multiple
satellite pulses. Under this condition (T rPS ⬍⬍ T p), the reflected portion of the
main pulse will overlap the trailing edge of the main pulse itself. The limiting

Figure 18 Output of passive SSPT Ce : LLF laser for different feedback cavity length
(a) L ⫽ 115 cm and (b) L ⫽ 65 cm. The separation of pulses can be changed as expected.

Copyright © 2002 by Taylor & Francis

factor of pulse duration in this passive SSPT laser is round-trip time or cavity
length of the pulse-seeding laser. For an active self-injection seeding or cavity-
flipping technique [45–49], it is limited by the transit time of the Pockels cell.
All of these requirements (T rPS ⬍⬍ T p ⬍⬍ T rFB ⬍⬍ T Geff, F SP ⬍⬍ F sat) can be easily
satisfied in any Ce-doped tunable laser material, and will be adequate for other
solid-state laser media, organic dyes, or semiconductors if appropriate cavity pa-
rameters are selected.

B. Self-Injection-Seeded Pulse Train Ce: LLF Laser

The experimental setup of the passive SSPT Ce :LLF laser is shown in Figure
19. The seeding laser cavity was the same as the low-Q, short-cavity Ce :LLF
laser described in Sec. III.B.1 (the resonator was formed by a 30% transmission
flat output coupler and the end surface reflection of the Ce: LLF crystal) [39]. A
confocal lens pair (f ⫽ 10 cm) with antireflection coating was inserted in the
feedback laser cavity for better mode matching with the pulse-seeding laser cav-
ity. The lens pair was adjusted to obtain the maximum output energy. The pump-
ing source was a KrF excimer laser. The 3 µJ, 325 nm, 950 ps seeding pulse
was amplified up to 60 µJ energy in the total pulse train with ⬃200 mJ/cm 2
pumping fluence.
A spectrometer was inserted before the streak camera to measure the spec-
trum. The spectrum scale was calculated using a CW He-Cd laser with wave-
length at 325 nm. Observation with a streak camera during this self-injection
seeding amplification process revealed that neither the pulsewidth nor the spec-
trum width changed, as shown in Figure 20. This result shows that the output
properties of this laser can be totally controlled by the pulse-seeding laser.

Figure 19 Experimental setup of the passive self-injection-seeded pulse train UV Ce :

LLF laser.

Copyright © 2002 by Taylor & Francis

Figure 20 (a) Spectrally and temporally resolved streak camera image of the injection
seeding pulse. The pulse duration in FWHM was 950 ps, and the center wavelength was
325 nm. (b) Spectrally and temporally resolved streak camera image of pulse train. This
image shows that neither the pulsewidth nor the spectrum width changed during this self-
injection seeding amplification process.

The parameters for the passive SSPT Ce:LLF laser in this case are T rPS
(200 ps) ⬍⬍ T p (950 ps) ⬍⬍ T rFB (7.7 ns) ⬍⬍ T Geff (⬃50 ns), F SP (⬃1 mJ/cm 2) ⬍⬍
F sat (⬃50 mJ/cm 2 (the spot size in the gain medium was ⬃1 mm). The perfor-
mance of this laser can be improved after these parameters are optimized. In this
way, UV short-pulse trains were directly generated from a Ce :LLF laser pumped
by a standard 10 ns KrF excimer laser using the simple passive self-injection-
seeded pulse train laser scheme.

C. Self-Injection-Seeded Pulse Train Ce: LiCAF Laser

To obtain a short pulse train, the passive SSPT laser scheme was employed in
the Ce :LiCAF laser as it was for the Ce :LLF laser described above [43].
The experimental setup of the SSPT Ce: LiCAF laser is shown in Figure
21. The 1.5 cm long seeding laser cavity was formed by a flat high-reflection
mirror and an 80% transmission flat output coupler. The 20 mJ, 10 ns, 1 Hz, 266
nm, horizontally polarized pumping pulses (the fourth harmonic of a conventional
10 ns Q-sw Nd :YAG laser) were focused longitudinally from the high-reflection
mirror side by a 30 cm focal-length lens with ⬃300 mJ/cm 2 pumping fluence
inside the active medium. The c-axis of the Ce: LiCAF laser crystal sample was
parallel to the direction of the pumping polarization. The absorbed energy was
5 mJ. In the feedback laser cavity, a confocal lens pair (f ⫽ 10 cm) with antire-

Copyright © 2002 by Taylor & Francis

Figure 21 Experimental setup of the passive self-injection-seeded pulse train UV Ce :
LiCAF laser.

Figure 22 (a) Oscilloscope trace of seeding pulse. (b) Oscilloscope trace of pulse train.

Copyright © 2002 by Taylor & Francis

flection coatings was inserted for better mode matching with the pulse-seeding
laser cavity. A 1 mm thick fused silica plate as an intracavity output coupler was
placed at an angle of 45 degrees to the resonator axis. The estimated coupling
constant was 3.6%.
As can be seen from the oscilloscope traces (Fig. 22) of the single seeding
pulse and the pulse train, this SSPT scheme is quite efficient with a Ce: LiCAF
laser. The ⬃45 µJ, 290 nm, horizontally polarized, 600 ps seeding pulse was
amplified up to ⬃230 µJ in the total pulse train inside the cavity, and 8 µJ was
coupled out.
This Ce :LiCAF laser meets all the requirements for the efficient passive
SSPT scheme operation as follows: T rPS (100 ps) ⬍⬍ T p (600 ps) ⬍⬍ T rFB (7.4
ns) ⬍⬍ T Geff (⬃40 ns), F SP (⬃5 mJ/cm 2) ⬍⬍ F sat (⬃100 mJ/cm 2). Because of the
shorter length of the gain medium (0.5 cm) and pulse-seeding laser cavity (1.5
cm), a shorter pulse duration (600 ps) from the Ce :LiCAF laser than that from
the SSPT Ce :LLF laser (crystal length: 2.5 cm; cavity length: 3 cm; pulse dura-
tion: 950 ps; see Sec. IV.B) was obtained.

V. ULTRAVIOLET-TUNABLE SUBNANOSECOND PULSE

GENERATION

In most lasers, all of the energy released via stimulated emission by the excited
medium is in the form of photons. Tunability of the emission in solid-state lasers
is achieved when the stimulated emission of photons is intimately coupled to the
emission of vibrational quanta (phonons) in a crystal lattice. In these ‘‘vibronic’’
lasers, the total energy of the lasing transition is fixed, but can be partitioned
between photons and phonons in a continuous fashion. The result is broad wave-
length tunability of the laser output. In other words, the existence of tunable
solid-state lasers is due to the subtle interplay between the Coulomb field of the
lasing ion, the crystal field of the host lattice, and electron–phonon coupling
permitting broadband absorption and emission. Therefore, the gain in vibronic
lasers depends on transitions between coupled vibrational and electronic states;
that is, a phonon is either emitted or absorbed with each electronic transition.
Rare earth ions doped in appropriate host crystals exhibit vibronic lasing.
The main difference between transition metal and rare earth ions is that the former
is crystal-field-sensitive and the latter is not. As distinct from transition metal
ions, the broad-band transitions for rare earth ions are quantum mechanically
allowed and therefore have short lifetimes and high cross sections. The Ce 3⫹ ion
laser, using a 5d–4f transition, has operated in the host crystals. Such a system
would be an alternative to the excimer laser as a UV source, with the added
advantage of broad tunability.
Due to the vibronic nature of the Ce 3⫹ ion laser, the emission of a photon

Copyright © 2002 by Taylor & Francis

is accompanied by the emission of phonons. These phonons contribute to ther-
malization of the ground-state vibrational levels. The laser wavelength depends
on which vibrationally excited terminal level acts as the transition terminus; any
energy not released by the laser photon will then be carried off by a vibrational
phonon, leaving the Ce 3⫹ ion at its ground state. The terminal laser level is a set
of vibrational states well above the ground state. So the Ce 3⫹ ion lasers belong
to four-level lasers.

A. Tunable Ce: LLF Laser

The fluorescence spectrum of Ce :LLF crystal excited by the fifth harmonic of
an Nd: YAG laser is shown in Figure 23 [40]. The Ce :LLF crystal was the one
used in the experiment described in Sec. III.B.2. Two peaks around 308 nm and
325 nm are seen from the fluorescence spectrum. The fluorescence lifetime of
the Ce :LLF crystal was estimated to be 40 ns. To test the Ce: LLF laser tunability
under 213 nm pumping conditions, we employed a high-Q cavity by replacing
the output coupler shown in Figure 12 with an 80% reflection flat mirror. Tunable
operation was realized by rotating the prism about the vertical axis. The Ce: LLF
laser tunability obtained at the pumping energy level of 22 mJ at 213 nm is shown

Figure 23 Spectrum of Ce : LLF pumped by the fifth harmonic of an Nd: YAG laser.
There are two peaks around 308 nm and 325 nm. The fluorescence lifetime is 40 ns.

Copyright © 2002 by Taylor & Francis

Figure 24 Short-cavity, high-Q Ce : LLF laser tunability obtained at a pumping energy
of 22 mJ at 213 nm.

in Figure 24. In a subnanosecond pulse regimen, the tuning was achieved from
309.5 nm to 312.3 nm and from 324.5 nm to 327.7 nm. The gap in the tuning
curve is attributed to the close-to-the-threshold operation regimen necessary to
maintain a single subnanosecond pulse operation. This tuning behavior resembles
that for Ce :YLF laser [52].

B. Tunable Ce: LiCAF Laser

The experimental setup of the tunable, short-cavity Ce :LiCAF laser is shown
schematically in Figure 25. The cavity length was 25 mm. To study the tuning
performance of this laser without consideration of its temporal characteristics, a
high-Q laser with a low-transmission, flat-output coupler (T ⫽ 20%) was de-
signed [53,54]. The laser consisted of a Littrow prism with a high reflection
coating at one face used as the end mirror and a 10 mm long Brewster cut at the

Figure 25 Experimental setup of the tunable, short-cavity Ce: LiCAF oscillator pumped
by the fourth harmonics of a Q-switched Nd: YAG laser.

Copyright © 2002 by Taylor & Francis

end faces, 1% doped (in the melt) Ce :LiCAF crystal used as the gain medium
without any special cooling. The 266 nm, horizontally polarized pumping pulses
from a Q-sw Nd: YAG laser in short-pulse operation mode were focused longitu-
dinally from the output mirror side using a 20 cm focal-length lens. In a single
pumping shot, there were four short pulses with pulse durations of about 1 ns
separated by about 5 ns as shown in Figures 28 and 29. In most cases, the laser
operated at 2 Hz to reduce any possible thermal problems in the Ce: LiCAF crys-
tal without mandatory cooling, and to obtain higher extraction efficiency. The c-
axis of the Ce: LiCAF crystal was parallel to the direction of the pumping polar-
ization. The tuning operation was realized by rotating the prism horizontally.
Tuning using a Ce:LiCAF crystal with Brewster-cut end faces is very efficient
because it can increase the dispersion of the laser beam. The output pulse was
separated from the pumping beam by a dichroic mirror, which has high reflection
in the region of 280–320 nm at a 45 degree incidence angle and high transmission
for the pumping beam (266 nm). The output pulse energy tuning curve obtained
is shown in Figure 26, with the pumping energy of 8 mJ. As illustrated in Figure
26, the pulses were multipeak at some points in the form of damped relaxation
oscillation, and the main peaks had pulse durations of 2–3 ns. The corresponding
tuning was accomplished from 282 nm to 314 nm, sacrificing the capability of
single short-pulse generation.
To obtain shorter pulses and a single-pulse output, a low-Q, short-cavity
Ce :LiCAF laser was constructed by changing the output coupler of the laser
mentioned above to a 75% transmission flat coupler. The output pulse energy
dependence on wavelength is shown in Figure 27. The pumping energies for
different wavelengths were varied between 2 and 4 mJ to obtain single-pulse
generation. The demonstrated tuning range was 281 nm to 297 nm. The pulse

Figure 26 For the high-Q (T c ⫽ 20%), short-cavity Ce: LiCAF laser, tunability between
282 and 314 nm was obtained.

Copyright © 2002 by Taylor & Francis

Figure 27 From the low-Q (T c ⫽ 75%), short-cavity Ce :LiCAF laser with a Littrow-
prism end mirror, tunability from 281 to 297 nm was achieved while maintaining single
short-pulse properties.

durations at different wavelengths observed by a streak camera are also shown

in Figure 27. Under the appropriate pumping fluence control, no satellite pulse
was observed because the pulse duration exceeded the cavity round trip time
(170 ps). The typical spectrally and temporally resolved streak camera image of
the output pulse at a wavelength of 289 nm, including the image of the pumping
pulse, is given in Figure 28. They were taken in a single shot. Figure 29 was

Figure 28 Spectrally and temporally resolved streak camera image of the output pulse
at 289 nm from the low-Q, short-cavity Ce: LiCAF laser. The image of the pumping pulse
can also be observed.

Copyright © 2002 by Taylor & Francis

Figure 29 Temporally resolved streak camera traces of the output pulse and the pump-
ing pulse processed from the image in Figure 28. The pulse width (at half maximum) of
the output pulse was 0.7 ns, and the center wavelength was 289 nm with a spectrum width
(at half maximum) of about 2 nm.

derived from Fig. 28 and describes the temporally resolved streak camera traces
of the output pulse and the pumping pulse [53].
To increase the output energy while maintaining the single short output
pulse, a 55% reflection flat-output coupler was used. A single-pulse output was
generated with the pumping energy of 15 mJ. The demonstrated tuning range
was 282–314 nm (Fig. 30). The maximum single subnanosecond pulse output

Figure 30 Tuning curve for the short-cavity Ce: LiCAF laser with 55% reflection output
coupler.

Copyright © 2002 by Taylor & Francis

was 1 mJ with the pulse width of 0.9 ns observed by a streak camera. In this
way, single short pulses with broad tuning regions can be easily generated from
a low-Q, short-cavity, tunable Ce :LiCAF laser pumped by the fourth harmonic
of a Q-sw Nd :YAG laser.

C. Tunable Subnanosecond Pulse Generation Around 230

nm by Sum-Frequency Mixing
To obtain tunable subnanosecond UV pulses around 230 nm, subnanosecond
pulses from a tunable Ce :LiCAF master oscillator and power amplifier (MOPA)
system were mixed with the fundamental radiation of a Q-sw Nd :YAG laser
[54]. The spectral region covered by this sum-frequency generation lies between
the fourth and fifth harmonics of Nd lasers and also between the third and fourth
harmonics of tunable near-infrared lasers such as Cr :LiSAF. The spectral region
near 230 nm is important practically because it spans the signature absorption
features of chemical and biological species such as nitric oxide and tryptophan.
The experimental setup is shown schematically in Figure 31. The oscillator
stage of the MOPA system is the same as mentioned above (see Sec. V. B). The
double-side-pumping, double-pass-amplification configuration was applied in the
amplifier stage. A Q-sw Nd:YAG laser was used to pump the oscillator and

Figure 31 Experimental setup of sum-frequency mixing of Ce: LiCAF and Nd :YAG

laser beams through a BBO crystal in type I phase matching condition.

Copyright © 2002 by Taylor & Francis

amplifier stages with optimized optical delay. The 266 nm pulses with 12 mJ
and 15 mJ energies from each side were slightly focused on the Ce :LiCAF crystal
of the amplifier. The pump energy for the Ce: LiCAF oscillator was adjusted to
obtain a better beam pattern so as to increase the efficiency of the amplifier. An
energy-tuning curve for the Ce: LiCAF MOPA system is shown in Figure 32.
Continuous tunability of the MOPA system was obtained from 284 to 299 nm
with the maximum pulse energy up to 4 mJ.
A BBO nonlinear crystal was utilized for frequency mixing of the Ce :
LiCAF MOPA system output with the fundamental output of another Q-sw Nd:
YAG laser synchronized to the Nd: YAG laser for pumping the Ce :LiCAF
MOPA system. The BBO crystal (5 mm ⫻ 5 mm ⫻ 5 mm) was cut at θ ⫽ 45
degree for type I phase matching. The Ce :LiCAF laser beam was focused with
a 100 cm focal-length lens. Using a dichroic beamsplitter, the Ce :LiCAF and
Nd :YAG laser beams were spatially overlapped and subsequently input into the
BBO crystal. The delay between the Ce :LiCAF and Nd :YAG laser beams
was electrically controlled to ensure temporal overlapping of the two input
beams. Pulse energy of the Nd: YAG laser fundamental beam was 30 mJ. The
tunability after the sum frequency mixing is shown in Figure 33. The obtained
tuning region was from 223 to 232 nm. Sum-frequency generation was optimized
at each wavelength by tuning the BBO crystal to maintain the proper phase
matching. The peak of the tuning occurred at 227 nm with 0.5 mJ of output
energy. The spectrally and temporally resolved streak camera image of the output
pulse at this wavelength is shown in Figure 34, and the pulse width was measured
to be 1.0 ns.

Figure 32 Tuning curve for the Ce : LiCAF MOPA system for the sum frequency mix-
ing with Nd : YAG laser.

Copyright © 2002 by Taylor & Francis

Figure 33 Tuning curve for sum-frequency mixing of Ce : LiCAF and Nd :YAG laser
beams.

Figure 34 Spectrally and temporally resolved streak camera image of the sum-
frequency generation at the peak point of the tuning curve of Figure 33.

Copyright © 2002 by Taylor & Francis

VI. EFFICIENT UV LASER AMPLIFICATION USING Ce:LLF
AND Ce:LiCAF CRYSTALS

The generation of high-energy pulses is based on the combination of a master

oscillator and multistage power amplifiers. In an oscillator–amplifier system,
pulse width, beam divergence, and spectral width are primarily determined by
the oscillator, whereas pulse energy and power are determined by the amplifier.
Operating an oscillator at relatively low energy levels reduces beam divergence
and spectral width. Therefore, from an oscillator-amplifier combination one can
obtain either a higher energy than is achievable from an oscillator alone or the
same energy in a beam with smaller beam divergence and narrower linewidth.
Generally speaking, the purpose of adding an amplifier to a laser oscillator is to
increase the brightness B [W cm ⫺2 sr ⫺1 ] of the output beam
B ⫽ P/AΩ
where P is the power of the output beam emitted from the area A and Ω is the
solid-angle divergence of the beam.

A. Confocal Four-Pass CW Ce: LLF Amplifier

To demonstrate a practical device with high small-signal gain, the Ce :LLF ampli-
fier was designed with a side-pumped confocal four-pass configuration as shown
in Figure 35. The Ce: LLF sample used for the amplifier contains about 0.5 at.%

Figure 35 Side-pumped confocal four-pass Ce :LLF amplifier. The pumping area was
3 mm wide and 10 mm long. A confocal lens pair and prisms with antireflection coatings
were used to reduce the loss.

Copyright © 2002 by Taylor & Francis

Ce 3⫹ ions [31]. The sample (1 cm long and 5 mm diameter) for the amplification
experiment was prepared for the transverse pumping. The amplifier consisted of
a gain medium located at the beam waist of a confocal lens pair (15 cm focal
length) and two turning prisms for a small displacement of each pass. This con-
figuration allowed the beams from different passes to overlap well in a small
pumped region. Side pumping is possible due to the high absorption (3 cm ⫺1) at
the pumping wavelength. KrF-laser pulses (10 mJ, 10 ns) were weakly focused
by a 10 cm focal-length cylindrical lens. The pumping laser operated at a 3 Hz
repetition rate to reduce the possible influence of color-center formation [24].
A Cw He-Cd laser was used as the probe light with the wavelength at 325
nm. A biplanar phototube was used as the detector. The oscilloscope trace of the
gain profile is shown in Figure 36. The upper trace is with the Cw He-Cd laser
probe; the lower trace is only the fluorescence without the probe. The difference
between the two traces is the real amplification. Considering this difference and
the probe level measured separately, the four-pass differential gain was estimated
to be 20 dB for the transmitted probe light. This small signal gain is large enough
for any practical use.

B. Picosecond Pulse Ce: LLF Amplifier

Using the above laser material and the same four-pass amplifier configuration,
a solid-state UV picosecond laser system was designed (Fig. 37). The seeding

Figure 36 Temporal gain profile of the 4-pass amplifier. The upper trace is with the
cw He-Cd laser probe beam, and the lower trace is only fluorescence without the probe.
Considering the difference of the two traces and the probe level without pumping measured
separately, the gain is estimated to be 20 dB for the transmitted probe light.

Copyright © 2002 by Taylor & Francis

Figure 37 Experimental set-up of solid-state UV picosecond laser system. This system
consisted of a Ti: sapphire oscillator and amplifier system, a frequency mixer, and a Ce:
LLF amplifier.

picosecond UV pulses at 325 nm were generated by the sum-frequency mixing

of the 835 nm picosecond pulses from a Ti: sapphire oscillator and amplifier
system and the second harmonic of an Nd :YAG laser [31]. The 10 Hz Q-sw
Nd :YAG laser was synchronized to an 82-MHz mode-locked Ti:sapphire laser
oscillator, and delivered 50 mJ pulse energy (532 nm) for pumping the Ti:sap-
phire amplifier and 10 mJ pulse energy for frequency mixing. The 10 nJ, 1.5 ps
pulses in the mode-locked pulse train from the Ti:sapphire oscillator without any
single-pulse selection were amplified typically up to a 100 µJ level by a double-
side-pumped confocal four-pass Ti: sapphire amplifier. For efficient sum-fre-
quency generation, the peak position of the 532 nm, 7 ns pulse was adjusted to
the largest pulse among the amplified picosecond pulse train by an optical delay
(40 ns). In this way, a single 1 µJ, 325 nm seeding pulse was generated at a 10
Hz repetition rate through a 5 mm thick BBO crystal cut at 35 degrees. The
duration of the UV pulses was evaluated to be less than 3 ps using a streak
camera.
The confocal four-pass Ce :LLF amplifier as described in Sec. VI. A was
pumped by 100 mJ KrF-laser pulses at a 0.5-Hz repetition rate synchronized to
the 10 Hz UV seed pulses with 0.5 J/cm 2 pumping fluence. To calibrate the gain
factor, a reference pulse separated from the input pulse was detected together

Copyright © 2002 by Taylor & Francis

Figure 38 Streak camera trace of the amplified ultraviolet pulse. The pulse duration
was measured to be 2.9-ps (resolution limit).

with the attenuated amplified pulse. A 17 dB gain was demonstrated for the trans-
mitted 325 nm ps pulses.
Using a streak camera, the duration of amplified UV pulses was evaluated
to be less than 2.9 ps, but the measurement was limited by the resolution of the
apparatus (Fig. 38). This result shows that there is no significant pulse width
broadening in the amplifier under the conditions of the experiments.

C. Efficient UV Short-Pulse Amplification in a Ce: LiCAF

MOPA System
The experiment setup of a Ce :LiCAF MOPA system is shown in Figure 39. A
low-Q, short-cavity Ce: LiCAF master oscillator with 15 mm cavity length was
formed by a flat high-reflection mirror and a 30% reflection flat output coupler
[55]. A 10 mm Brewster-cut, 1% doped (in the melt) Ce: LiCAF crystal was used
as the oscillator gain medium. The 15 mJ, 10 ns, 266 nm, horizontally polarized
pumping pulses from a Q-sw Nd: YAG laser were focused longitudinally from
the high-reflection mirror side by a 20 cm focal-length lens to obtain a 1 J/cm 2
pumping fluence inside the active medium. In most cases, the laser operated at
2 Hz to avoid possible thermal problems in the Ce: LiCAF crystal and to obtain
higher extraction efficiency. The c-axis of the Ce :LiCAF crystal was parallel to
the direction of the pumping polarization. A single 1 mJ, horizontally polarized
pulse at 289 nm was obtained at the sacrifice of the extracted energy from this
master oscillator. The pulse duration observed by a streak camera was 1.0 ns as
shown in Figures 40 and 41a. Under the appropriate pumping fluence control,

Copyright © 2002 by Taylor & Francis

Figure 39 Experiment setup of the Ce : LiCAF MOPA system pumped by fourth har-
monics of two conventional 10-ns Q-switched Nd: YAG lasers. This system consists of
a low-Q, short-cavity master oscillator, and a confocal double-pass power amplifier.

Figure 40 Spectrally and temporally resolved streak-camera image of the UV short

pulses from the oscillator stage and after the amplifier stage. A part of the oscillator-
pumping pulse image can also be observed. All these images are taken in a single shot.

Copyright © 2002 by Taylor & Francis

Figure 41 Temporally resolved streak-camera traces of the pulses processed from the
image in Figure 40. (a) The temporal pulse shape from the low-Q, short-cavity master
oscillator. The pulse duration (FWHM) was 1.0 ns, and the center wavelength was 289
nm. (b) The temporal pulse shape after the amplifier. The pulse duration (FWHM) was
also 1.0 ns.

no satellite pulse was observed because the pulse duration exceeded the cavity
round trip time (100 ps).
Another 10 mm, 1% doped Brewster-cut Ce: LiCAF crystal was employed
in the power amplifier stage. The amplifier was designed with a confocal double-
pass configuration similar to the double-side, coaxially pumped, confocal
multipass configuration of a Ti:sapphire amplifier [56]. Another 10 ns, Q-sw
Nd :YAG laser for pumping the amplifier operated with variable Q-switch delay.
The 266 nm pulses with 46 mJ and 34 mJ energies from each side were slightly
focused by 70 cm focal-length lens down to ⬃1 mm beam diameter (⬃5 J/cm 2).
The amplifier consisted of a gain medium located at the beam waist of a confocal
lens pair (40 cm focal length) and a roof reflector with dielectric coating for a
small vertical displacement of each pass (Fig. 39). The signal passes coincide
with the pumped region with a small angular separation from the pumping beams
(less than a few degrees). This configuration allowed the signal beams with differ-
ent passes to overlap almost completely in a small pumped region. The output
energy of the amplifier was measured for different pumping delays as shown in
Figure 42. This result indicated that the relative timing between two Nd:YAG
lasers should be controlled within an accuracy of a few nanoseconds. Here the

Copyright © 2002 by Taylor & Francis

Figure 42 Output energy of the amplifier for different delay timings of the probe pulse
from the oscillator and the amplifier-pumping pulse from the Nd :YAG laser.

optimum relative delay timing between the Q-switch trigger signals for the two
Nd :YAG lasers was 15 ns. A single-pass gain of over 10 times was observed.
The double-pass differential small-signal gain reached 100 times, as shown in
Figure 43. The 1 mJ input pulse was amplified up to 14 mJ with 14 MW peak
power at a 2 Hz repetition rate. The amplified pulse duration observed by a streak
camera was 1.0 ns as shown in Figure 41b. There was no noticeable pulse broad-
ening accompanying the amplification process. The output gain dependence on
different input energy of the amplifier is shown in Figure 43. This result clearly
shows that the amplification saturation is reached with sufficient input flux. The
energy extraction efficiency in the amplifier stage exceeded 18%, which was

Figure 43 Output gain dependence on input energy of the amplifier. This result clearly
shows that the amplification saturation is reached with sufficient input flux.

Copyright © 2002 by Taylor & Francis

sufficient for practical use. Even at a 10 Hz operation, 100 mW average power
after amplifier was obtained.

VII. HIGH-PULSE-ENERGY UV LASERS USING LARGE

Ce: FLUORIDE CRYSTALS

A short-pulse (1 ns) Ce :LiCAF MOPA system and an all-solid-state Ce: LLF

laser have been demonstrated. Due to the limited size of the available crystals,
it was difficult to obtain high-energy outputs directly from Ce :LiCAF and Ce:
LLF lasers. Recently, the growth of large Ce: LiCAF and Ce :LLF crystals has
become successful [57], making possible high-energy pulse generation from Ce:
LiCAF and Ce: LLF lasers.

A. High-Energy Pulse Generation from a Ce:LiCAF

Oscillator
The schematic diagram of a Ce :LiCAF laser resonator using the large Ce :LiCAF
crystal is shown in Figure 44. The laser resonator is established by a flat high
reflector and a flat output coupler with 30% reflection for 290 nm and 75% trans-
mission for 266 nm separated by 4 cm [58]. The large Ce :LiCAF crystal (18
mm in diameter, clear aperture 15 mm, length 10 mm) is doped with 1.2 mol%
Ce 3⫹ ions. There is no coating on the parallel end faces of the crystal that are
perpendicular to the optical axis of the resonator. The fourth harmonic of a Q-
sw Nd :YAG laser is used as the pumping source. The quasilongitudinal pumping
method was used to obtain a high-quality laser beam. Because it is difficult to
fabricate an end mirror with high reflection for 290 nm and high transmission

Figure 44 Experimental setup of the Ce : LiCAF laser oscillator pumped by the fourth
harmonic of a Q-switched Nd: YAG laser using a quasilongitudinal pumping scheme.

Copyright © 2002 by Taylor & Francis

for 266 nm pump beams while maintaining a high damage threshold, the Ce:
LiCAF crystal was pumped from the output coupler side that has almost the same
transmission for the pump and output wavelengths. To obtain high-energy output
without damage to the crystal and optics in the cavity, a large pump beam cross
section is necessary. The horizontally polarized pump beam is focused with a 40
cm focal-length lens to produce a 4 mm diameter spot at the surface of the Ce:
LiCAF crystal without any damage to the crystal. To reduce the diffraction effects
and disturbance to the beam uniformity, it is better to choose a ratio of crystal
radius to beam radius of 2 or more. Therefore, a much larger crystal diameter
than the pump beam diameter is preferred. More than 85% of the incident pump
pulse energy is absorbed by the crystal, so the crystal is long enough for practical
use. Figure 45 presents the obtained output energies at 290 nm as a function of
the absorbed 266 nm pump energy. The laser oscillation threshold is 12 mJ, which
corresponds to a threshold fluence of approximately 100 mJ/cm 2. The measured
output energy remained linear with pump fluence with a slope efficiency of 39%.
The efficiency can be improved by using a Brewster-cut crystal with antireflective
coating. The highest pulse energy was 21 mJ at 10 Hz and 290 nm. The satellite-
free single pulse had a pulse duration of 3 ns.
Because the loss of the pumping pulse energy through the output coupler
was still large, a noncollinear pumping scheme was tried, as shown in Figure 46.
The large pumping beam cross section made it possible to achieve efficient pump-
ing. The angle between the pumping beam and the output beam was about 5
degrees. The obtained output energies at 290 nm as a function of the absorbed
266-nm pump energy are shown in Figure 47. In this case, the output pulse energy
was improved up to 30.5 mJ, and the slope efficiency was also 39%.
To demonstrate the tunability of the Ce :LiCAF laser consisting of the large

Figure 45 Laser output energy as a function of absorbed pump energy. The measured
output energy remained linear with pump fluence with a slope efficiency of 39%.

Copyright © 2002 by Taylor & Francis

Figure 46 Experimental setup of the Ce: LiCAF laser oscillator using a noncollinear
pumping scheme.

Ce :LiCAF crystal, a grating was used as the tuning element which also acted as
the end mirror of the oscillator, as shown in Figure 48. The grating worked in
the Littrow condition. The incidence of the grating and its first-order diffraction
overlapped. The grating used here was blazed for 500 nm wavelength. The ob-
tained tuning range was from 284 to 294 nm, as shown in Figure 49, and maxi-
mum output was 6 mJ. The limited tunable range was due to the low diffraction
efficiency of the grating for the 290 nm wavelength. A broader tunable range
can be expected using a grating blazed for 290 nm wavelength. The pulse widths
were approximately 4 ns, and the spectrum widths were about 0.7 nm.
To generate pulses with much higher energy, the fourth harmonics of two
simultaneous Q-sw Nd:YAG lasers were used as the pumping sources (Fig. 50).
The three pump beams were focused with a 40 cm focal-length lens to produce
a spot size of 6 mm in diameter at the surface of the Ce: LiCAF crystal. With

Figure 47 Laser output energy as a function of absorbed pump energy. The output
pulse energy was improved up to 30.5 mJ, and the slope efficiency was also 39%.

Copyright © 2002 by Taylor & Francis

Figure 48 Experimental setup of tunable Ce :LiCAF laser. A grating was used as the
tuning element, which also acted as the end mirror of the oscillator. The grating worked
in the Littrow condition.

Figure 49 Tuning curve for the tunable Ce :LiCAF laser. The limited tunable range
was due to the low diffraction efficiency of the grating for the 290 nm wavelength.

Figure 50 Experimental setup of the Ce :LiCAF laser oscillator pumped by the fourth
harmonics of two Q-switched Nd: YAG lasers. The pulse energy as high as 60 mJ was
achieved at 290 nm at 10 Hz repetition rate.

Copyright © 2002 by Taylor & Francis

the total pumping energy of 230 mJ, the output pulse energy as high as 60 mJ
was achieved at 290 nm at 10 Hz repetition rate, which is the highest performance
reported for a Ce: LiCAF oscillator until now. In this way, high-energy pulses
at 290 nm were generated very easily and efficiently.

B. Ce:LiCAF Oscillator with Noncollinear Brewster Angle

Pumping
To scale the output energy of a Ce :LiCAF laser, a new pumping scheme permit-
ting efficient, high-energy pumping is necessary [59]. A pumping scheme with
pumping at Brewster angles on a disk Ce :LiCAF crystal is proposed, shown
schematically in Figure 51. The disk Ce :LiCAF crystal (20 ⫻ 20 mm 2 in cross
section, 5 mm in length) was placed at the 289 nm wavelength Brewster angle
relative to the oscillator axis. The pumping pulses (the fourth harmonics of two
Q-sw Nd:YAG lasers) were directed to the crystal in the direction near the reflec-
tion of the Ce :LiCAF laser pulse on the crystal surface. In this case, the pumping
pulses at 266 nm wavelength could also be at the Brewster angle. Thus, the
pumping energy of p-polarization pulses could be efficiently coupled into the
crystal from the two sides of the crystal without the risk of possible cavity mirror
damage, as in the collinear pumping scheme [58]. The laser resonator is con-
structed simply, using a flat, high reflector and a flat output coupler with a trans-
mission of 45% (Fig. 51). The pumping pulses with horizontal polarization from
the two Q-sw Nd: YAG lasers operated at 1 Hz were focused softly on the Ce:
LiCAF crystal by two 70 cm focal-length spherical lenses. The delay between
the pumping pulses from the two Nd :YAG lasers was controlled by an electric
delay controller. The pumping pulse energies of the two Nd:YAG lasers were
80 mJ and 120 mJ, and approximately 50% of these were absorbed by the Ce:

Figure 51 Experimental setup for disk Ce :LiCAF laser with noncollinear, near-
Brewster-angle pumping scheme.

Copyright © 2002 by Taylor & Francis

LiCAF crystal. A maximum output energy of 11 mJ was obtained from the Ce:
LiCAF laser with Brewster-angle pumping when the pumping pulses from the
two Nd : YAG lasers arrived on the crystal simultaneously. Even though the
output is presently not very high, this noncollinear Brewster-angle-pumping disk
oscillator scheme will make it possible to scale the output energy significantly
by using larger pumping sources and crystals with higher absorption due to the
higher pumping efficiency. The potential advantages of this scheme are the re-
duced risk of possible mirror damage, ease of pumping beam multiplexing, and
better gain profile compared with the side-pumping scheme.

C. Ce: LLF Laser Using Large Ce: LLF Crystal

A Ce: LLF laser resonator is established by using a flat high reflector and a flat
output coupler [60]. The length of the laser cavity was 6 cm. The layout is shown
schematically in Figure 52. High-quality, large Ce:LLF crystals (φ 18 mm ⫻
10 mm in length) were grown successfully at Tohoku University, Japan by the
Czochralski (CZ) method. The Ce :LLF sample used here was obtained by cutting
the grown Ce: LLF crystal in the middle along its axis (a half-cut cylinder). The
side window and two end surfaces were polished. No dielectric coatings were
deposited on the polished end surfaces and the side window. The pumping pulses
from a randomly polarized KrF excimer laser operated at 1 Hz were focused
softly on the side window of the Ce: LLF crystal using a 50 cm focal-length
spherical lens under a normal-incidence side-pumping condition. Because the
output pulse of the KrF excimer laser has a rectangular shape, it is not difficult
to make a near-line-shaped pumping area on the Ce :LLF crystal through a spheri-
cal lens. Almost all of the pumping pulse energy (maximum: 230 mJ) was ab-
sorbed by the Ce: LLF crystal. To obtain high-output energy from the Ce:LLF

Figure 52 Experimental setup of the high-power Ce: LLF laser pumped by a randomly
polarized KrF excimer laser operated at a repetition of 1 Hz.

Copyright © 2002 by Taylor & Francis

Figure 53 Input–output dependence curve for the Ce: LLF laser. The maximum output
energy reaches 27 mJ at the wavelength of 309 nm.

laser, we tested some output couplers with different transmissions. The best result
was obtained with the coupler with 45% transmission. Figure 53 shows the input–
output dependence curves for the Ce: LLF laser. The maximum output pulse en-
ergy reached 27 mJ with the pumping pulse energy of 230 mJ, and the corre-
sponding pumping fluence was approximately 0.6 J/cm 2. This is the highest-
output pulse energy ever achieved from this laser medium. The free-running Ce :
LLF laser operated at the wavelength of 309 nm. The slope efficiency was approx-
imately 17%.

VIII. CONCLUSIONS AND PROSPECTS

As mentioned above, tunable lasers in the ultraviolet region with tunability cen-
tered around 290 nm are of special interest for applications relating to remote
sensing. The simple, compact, all-solid-state Ce :LiCAF (282–314 nm) laser can
generate coherent radiation in this wavelength. Ce :LLF has a potential longer-
wavelength tuning region of around 305–340 nm, so it is especially attractive
for use in spectroscopy of wide band-gap semiconductors for blue laser diodes,
such as GaN. Their broad gain bandwidth corresponding to a few femtoseconds
is extremely attractive for short-pulse applications.
The new solid-state, tunable, ultraviolet crystals Ce: LLF and Ce :LiCAF
are proving to be very efficient and reliable for realizing UV lasers. Gain for the
Ce :LLF crystal was demonstrated from 323 to 335 nm. The gain bandwidth
makes the generation of 10 fs pulses possible in the transform limit condition.
The saturation fluence of the Ce :LLF crystal was estimated to be 50 mJ/cm 2
through a pump-probe experiment [31]. Subnanosecond ultraviolet coherent

Copyright © 2002 by Taylor & Francis

pulses were generated directly from solid-state lasers simply for the first time
using low-Q, short-cavity Ce :fluoride lasers pumped by KrF laser [39], the fifth
[40] and fourth [53] harmonics of Nd:YAG lasers.
For such new laser materials, a passive self-injection seeding scheme was
proposed for the direct generation of short-pulse trains, which did not require
CW operation capability or an external short-pulse seeding laser. Using this sim-
ple scheme, UV subnanosecond pulse trains are directly and passively generated
from Ce: LLF pumped by a standard 10 ns KrF excimer laser [39] and Ce :LiCAF
pumped by the fourth harmonic of a conventional 10 ns Q-sw Nd:YAG laser
[43].
To prove the tunability of these new laser materials, a tunable all-solid-
state Ce: LLF laser was made with the fifth harmonic of an Nd:YAG laser as
the pumping source [40]. A tunable Ce :LiCAF laser was also demonstrated with
broad tuning region from 282 to 314 nm [53]. Furthermore, tunable UV pulses
around 230 nm were obtained by the sum frequency mixing of Ce: LiCAF and
Nd :YAG lasers [54].
For these new laser materials with large bandwidth, short pulse applications
have been investigated. UV picosecond-pulse amplification with large small
signal gain as high as 17 dB was demonstrated using Ce:LLF in a confocal
four-pass amplifier [31]. The direct generation and efficient amplification of
UV short pulses have been demonstrated from the simplest, all-solid-state, UV
short-pulse, MOPA system composed of Ce :LiCAF crystals and conventional
Q-sw Nd: YAG lasers [55]. In this way, it has been proven that the Ce :LiCAF-
based MOPA system is as effective and practical as other UV short-pulse sys-
tems.
Large Ce: LiCAF crystals with 15 mm diameter were successfully grown
by the Czochralski method recently [57]. Due to the available large Ce: LiCAF
crystal, 60 mJ output energy was obtained from the Ce :LiCAF laser. This is the
highest output directly from a Ce :LiCAF laser reported until now. A much higher
output can be expected by fully utilizing the crystal cross size while using a larger
pumping source. This suggests that Ce :LiCAF is a promising material for high-
energy ultraviolet pulse generation combined with a high-power, Q-sw Nd:YAG
laser [58].
Ce :fluoride laser has been used to environmental sensing [61]. Narrow
lasing linewidth (⬍0.1 nm) operation was demonstrated in distributed-feedback,
tunable Ce 3⫹-doped colquiriite lasers [62]. Ce :LiCAF laser worked efficiently at
20 kHz repetition rate [63]. A Ce: LiCAF laser pumped by the sum-frequency-
mixing (271 nm) of the green and yellow fundamental lines (511 and 578 nm)
from a copper-vapor laser was recently reported [64]. The basic optical properties
of Ce 3⫹ ions in both oxide and fluoride hosts were investigated [65]. The room
temperature fluorescence spectrum of Ce 3⫹-doped LiBaF 3 (Ce: LBF) spans the
spectral region of 300–450 nm [66].

Copyright © 2002 by Taylor & Francis

Figure 54 Tuning curves for Ce 3⫹ ion-activated laser systems.

With the development of the new laser media and nonlinear crystals, solid-
state tunable ultraviolet short-pulse lasers have been realized, as shown in Figure
54 [34,44]. It is reasonable to expect CW solid-state ultraviolet lasers in future
with the improvement of Ce :fluoride crystal qualities and high-power ultraviolet
CW pumping sources.
It is believable that, with improvement in the quality and size of UV laser
crystals and nonlinear crystals (which are important for the CW UV pumping
sources), all-solid-state, compact, ultrashort pulse, ultraviolet, tunable Ce :fluo-
ride lasers will be possible in the near future. Further development of laser sys-
tems using these new laser media will open up new possibilities for simple and
compact tunable UV ultrashort-pulse laser light sources.

REFERENCES

1. GJ Megie, G Ancellet, J Pelon. Lidar measurements of ozone vertical profiles. Appl

Opt 24: 3454–3463, 1985.
2. EV Browell. Applications of lasers in remote sensing. Advanced Solid-State Lasers,
OSA Technical Digest, 1995, pp 2–4.
3. PF Moulton. Spectroscopic and laser characteristics of Ti :Al 2 O 3. J Opt Soc Am B
3: 125–133, 1986.
4. JC Walling, OG Peterson, HP Jenssen, RC Morris, EW O’Dell. Tunable alexandrite
lasers. IEEE J Quantum Electron. QE-16: 1302–1315, 1980.
5. SA Payne, LL Chase, HW Newkirk, LK Smith, WF Krupke. LiCaAlF 6 :Cr 3⫹: a

Copyright © 2002 by Taylor & Francis

 promising new solid-state laser material. IEEE J Quantum Electron. 24: 2243–2252,
1988.
6. SA Payne, LL Chase, LK Smith, WL Kway, HW Newkirk. Laser performance of
LiSrAlF 6 : Cr 3⫹. J Appl Phys 66: 1051–1056, 1989.
7. WL Pryor. Laser Focus World: Buyers’ Guide. Tulsa, OK: PennWell, 1994.
8. JF Pinto, L Esterowitz, GH Rosenblatt. Tunable UV source based on tripled
Cr : LiSAF. Advanced Solid-State Lasers, OSA Technical Digest, 1995, pp 279–
281.
9. D Strickland, G Mourou. Compression of amplified chirped optical pulses. Opt Com-
mun 56: 219–221, 1985.
10. DJ Ehrlich, PF Moulton, RM Osgood, Jr. Ultraviolet solid-state Ce: YLF laser at
325 nm. Opt Lett 4: 184–186, 1979.
11. DJ Ehrlich, PF Moulton, RM Osgood, Jr. Optically pumped Ce: LaF 3 laser at 286
nm. Opt Lett 5: 339–341, 1980.
12. MA Dubinskii, RY Abdulsabirov, SL Korableva, AK Naumov, VV Semashko. New
solid-state active medium for tunable ultraviolet lasers. 18th International Quantum
Electronics Conference, OSA Technical Digest. Washington, DC Optical Society
of America, 1992, pp 548–550.
13. MA Dubinskii, RY Abdulsabirov, SL Korableva, AK Naumov, VV Semashko. A
new active medium for a tunable solid-state UV laser with an excimer pump. Laser
Phys 4: 480–484, 1994.
14. MA Dubinskii, VV Semashko, AK Naumov, RY Abdulsabirov, SL Korableva. Ac-
tive medium for all-solid-state tunable UV laser. In: AA Pinto, TY Fan, eds. OSA
Proceedings on Advanced Solid-State Lasers. Washington, DC: Optical Society of
America, 1993, vol 15, pp 195–198.
15. MA Dubinskii, VV Semashko, AK Naumov, RY Abdulsabirov, SL Korableva.
Spectroscopy of a new active medium of a solid-state UV laser with broadband
single-pass gain. Laser Phys 3: 216–217, 1993.
16. MA Dubinskii, VV Semashko, AK Naumov, RY Abdulsabirov, SL Korableva. Ce 3⫹-
doped colquiriite, a new concept of all-solid-state tunable ultraviolet laser. J Mod
Opt 40: 1–5, 1993.
17. JF Pinto, GH Rosenblatt, L Esterowitz, GJ Quarles. Tunable solid-state laser action
in Ce 3⫹ : LiSrAlF 6. Electron. Lett 30: 240–241, 1994.
18. CD Marshall, SA Payne, JA Speth, WF Krupke, GJ Quarles, V Castillo, BHT Chai.
Ultraviolet laser emission properties of Ce 3⫹-doped LiSrAlF 6 and LiCaAlF 6. J Opt
Soc Am B 11: 2054–2065, 1994.
19. KH Yang, JA Deluca. UV fluorescence of cerium-doped lutetium and lanthanum
trifluorides, potential tunable coherent sources from 2760 to 3220 A. Appl Phys Lett
31: 594–596, 1977.
20. RW Waynant. Vacuum ultraviolet laser emission from Nd ⫹3 : LaF 3. Appl Phys B 28:
205, 1982.
21. WJ Miniscalco, JM Pellegrino, WM Yen. Measurements of excited-state absorption
in Ce 3⫹ : YAG. J Appl Phys 49: 6109–6111, 1978.
22. RR Jacobs, WF Krupke, MJ Weber. Measurement of excited-state absorption loss
for Ce 3⫹ in Y 3 Al 5 O 12 and implications for 5d-4f rare-earth lasers. Appl Phys Lett
33: 410–412, 1978.

Copyright © 2002 by Taylor & Francis

23. DS Hamilton. Trivalent cerium doped crystals as tunable system. Two bad apples.
In: P Hammerling, AB Budgor, A Pinto, eds. Tunable Solid State Lasers. Berlin:
Springer, 1985, pp 80–90.
24. KS Lim, DS Hamilton. Optical gain and loss studies in Ce 3⫹ : YLiF 4. J Opt Soc Am
B 6: 1401–1406, 1989.
25. KS Lim, DS Hamilton. UV-induced loss mechanisms in a Ce 3⫹ : YLiF 4 laser. J Lumi-
nescence 40 & 41: 319–320, 1988.
26. JF Owen, PB Dorain, T Kobayasi. Excited-state absorption in Eu 2⫹ :CaF 2 and Ce 3⫹ :
YAG single crystals at 298 and 77 KJ Appl Phys 52: 1216–1223, 1981.
27. DS Hamilton, SK Gayen, GJ Pogatshnik, RD Ghen. Optical-absorption and photo-
ionization measurements from the excited states of Ce 3⫹ : Y 3 Al 5 O 12. Phys Rev B
39: 8807–8815, 1989.
28. GJ Pogatshnik, DS Hamilton. Excited-state photoionization of Ce 3⫹ions in Ce 3⫹ :
CaF 2. Phys Rev B 36: 8251–8257, 1987.
29. MA Dubinskii, AC Cefalas, E Sarantopoulou, SM Spyrou, CA Nicolaides, RY
Abdulsabirov, SL Korableva, VV Semashko. Efficient LaF 3 : Nd 3⫹-based vacuum-
ultraviolet laser at 172 nm. J Opt Soc Am B 9: 1148–1150, 1992.
30. S Nakamura, M Senoh, S Nagahama, N Iwasa, T Yamada, T Matsushita, Y Sugi-
moto, H Kiyoku. Continuous-wave operation of InGaN multi-quantum-well-struc-
ture laser diodes at 233 K Appl Phys Lett 69: 3034–3036, 1996.
31. N Sarukura, Z Liu, Y Segawa, K Edamatsu, Y Suzuki, T Itoh, VV Semashko, AK
Naumov, SL Korableva, RY Abdulsabirov, MA Dubinskii. Ce 3⫹ : LiLuF 4 as a broad
band ultraviolet amplification medium. Opt Lett 20: 294–296, 1995.
32. RY Abdulsabirov, MA Dubinskii, BN Kazakov. Sov Phys Crystallogr 32: 559, 1987.
33. LM Frantz, JS Nodvick. Theory of pulse propagation in a laser amplifier. J Appl
Phys 34: 2346–2349, 1963.
34. N Sarukura, MA Dubinskii, Z Liu, VV Semashko, AK Naumov, SL Korableva, R
Yu. Abdulsabirov, K Edamatsu, Y Suzuki, T Itoh, Y Segawa. Ce 3⫹ activated fluoride
crystals as prospective active media for widely tunable ultraviolet ultrafast lasers
with direct 10-nsec pumping. IEEE J Select Topics Quant Electron 1: 792–804, 1995.
35. DE Spence, PN Kean, W Sibbett. 60-fsec pulse generation from a self-mode-locked
Ti: sapphire laser. Opt Lett 16: 42–44, 1991.
36. N Sarukura, Y Ishida, and H Nakano. Generation of 50-fsec pulses from a pulse-
compressed, cw, passively mode-locked Ti :sapphire laser. Opt Lett 16: 153–155,
1991.
37. C Lin, CV Shank. Subnanosecond tunable dye laser pulse generation by controlled
resonator transients. Appl Phys Lett 26: 389–391, 1975.
38. D Roess. Giant pulse shortening by resonator transients. J Appl Phys 37: 2004–
2006, 1966.
39. N Sarukura, Z Liu, Y Segawa, VV Semashko, AK Naumov, SL Korableva, RY
Abdulsabirov, MA Dubinskii. Direct and passive sub-nanosecond pulse-train gener-
ation from a self-injection-seeded ultraviolet solid-state laser. Opt Lett 20: 599–601,
1995.
40. N Sarukura, Z Liu, S Izumida, MA Dubinskii, RY Abdulsabirov, SL Korableva.
All-solid-state tunable ultraviolet sub-nanosecond laser with direct pumping by the
fifth harmonic of an Nd: YAG laser. Appl Opt 37: 6446–6448, 1998.

Copyright © 2002 by Taylor & Francis

41. Y Mori, S Nakajima, A Taguchi, A Miyamoto, M Inakaki, W Zhou, T Sasaki, S
Nakai. Nonlinear optical properties of cesium lithium borate. Jpn J Appl Phys 34:
L296–L298, 1995.
42. R Komatsu, T Sugawara, K Sassa, N Sarukura, Z Liu, S Izumida, S Uda, T Fukuda,
K Yamanouchi. Growth and ultraviolet application of Li 2 B 4 O 7 crystals: generation
of the fourth harmonic and fifth harmonics of Nd: Y 3 Al 5 O 12 lasers. Appl Phys Lett
70: 3492–3494, 1997.
43. N Sarukura, Z Liu, Y Segawa, VV Semashko, AK Naumov, SL Korableva, RY
Abdulsabirov, MA Dubinskii. Ultraviolet subnanosecond pulse train generation from
an all-solid-state Ce: LiCAF laser. Appl Phys Lett 67: 602–604, 1995.
44. Z Liu, N Sarukura, MA Dubinskii, RY Abdulsabirov, SL Korableva. All-solid-state
subnanosecond tunable ultraviolet laser sources based on Ce 3⫹-activated fluoride
crystals. J Nonlinear Opt Physics Mater 8: 41–54, 1999.
45. YS Liu. Nanosecond pulse generation from a self-injected laser-pumped dye laser
using a novel cavity-flipping technique. Opt Lett 3: 167–169, 1978.
46. YS Liu. Generation of high-power nanosecond pulses from a Q-switched Nd : YAG
oscillator using intracavity-injecting technique. Opt Lett 4: 372–374, 1979.
47. P Ewart. Frequency tunable, nanosecond duration pulses from flashlamp pumped
dye lasers by pulsed Q-modulation. Opt Commun 28: 379–382, 1979.
48. CH Brito Cruz, E Palange, F De Martini. High power subnanosecond pulse genera-
tion in Nd-YAG lasers. Opt Commun 39: 331–333, 1981.
49. CH Brito Cruz, F De Martini, HL Fragnito, E Palange. Picosecond pulse generation
by intracavity nonlinear compression in self-injected Nd-YAG laser. Opt Commun
40: 298–301, 1982.
50. EG Erickson, LL Harper, in Digest of Technical Papers, OSA/IEEE Conference on
Laser Engineering and Applications. New York: (Institute of Electrical and Electron-
ics Engineers, 1975, p 16.
51. MJ LaGasse, RW Schoenlein, JG Fujimoto, PA Schulz. Amplification of femtosec-
ond pulses in Ti :Al 2 O 3 using an injection-seeded laser. Opt Lett 14: 1347–1349,
1989.
52. PF Moulton. Tunable paramagnetic-ion lasers. In: M Bass and ML Stitch, eds. Laser
Handbook. Amsterdam: Elsevier Science Publishers BV, 1985, p 284.
53. Z Liu, H Ohtake, N Sarukura, MA Dubinskii, RY Abdulsabirov, SL Korableva, AK
Naumov, VV Semashko. Subnanosecond tunable ultraviolet pulse generation from
a low-Q, short-cavity Ce :LiCAF laser. Jpn J Appl Phys 36: L1384–L1386, 1997.
54. Z Liu, N Sarukura, MA Dubinskii, RY Abdulsabirov, SL Korableva, AK Naumov,
VV Semashko. Tunable ultraviolet short-pulse generation from a Ce: LiCAF laser
amplifier system and its sum-frequency mixing with an Nd: YAG laser. Jpn J Appl
Phys 37: L36–L38, 1998.
55. N Sarukura, Z Liu, H Ohtake, Y Segawa, MA Dubinskii, RY Abdulsabirov, SL
Korableva, AK Naumov, VV Semashko. Ultraviolet short pulses from an all-solid-
state Ce: LiCAF master oscillator and power amplifier system. Opt Lett 22: 994–
996, 1997.
56. N Sarukura, Y Ishida. Ultrashort pulse generation from a passively mode-locked
Ti :sapphire laser based system. IEEE J Quantum Electron. 28: 2134–2141, 1992.
57. K Shimamura, N Mujilatu, K Nakano, SL Baldochi, Z Liu, H Ohtake, N Sarukura,

Copyright © 2002 by Taylor & Francis

 and T Fukuda. Growth and characterization of Ce-doped LiCaAlF 6 single crystals.
J Crystal Growth 197: 896–900, 1999.
58. Z Liu, S Izumida, H Ohtake, N Sarukura, K Shimamura, N Mujilatu, SL Baldochi,
T Fukuda. High-pulse-energy, all-solid-state, ultraviolet laser oscillator using large
Czochralski-grown Ce: LiCAF crystal. Jpn J Appl Phys 37: L1318–L1319, 1998.
59. Z Liu, K Shimamura, K Nakano, T Fukuda, T Kozeki, H Ohtake, N Sarukura. High-
pulse-energy ultraviolet Ce: LiCAF laser oscillator with newly designed pumping
schemes. Jpn J Appl Phys 39: L466–L467, 2000.
60. Z Liu, K Shimamura, K Nakano, N Mujilatu, T Fukuda, T Kozeki, H Ohtake, N
Sarukura. Direct generation of 27-mJ, 309-nm pulses from a Ce : LLF oscillator using
a large-size Ce : LLF crystal. Jpn J Appl Phys 39: L88–L89, 2000.
61. P Rambaldi, M Douard, J-P Wolf. New UV tunable solid-state lasers for lidar appli-
cations. Appl Phys B 61: 117–120, 1995.
62. JF Pinto, L Esterowitz. Distributed-feedback, tunable Ce 3⫹-doped colquiriite lasers.
Appl Phys Lett 71: 205–207, 1997.
63. AB Petersen. All solid-state, 228–240 nm source based on Ce: LiCAF. Advanced
Solid-State Lasers, OSA Technical Digest, 1997, paper PD2.
64. AJS McGonigle, DW Coutts, CE Webb. 530-mW 7-kHz cerium LiCAF laser
pumped by the sum-frequency-mixed output of a copper-vapor laser. Opt Lett 24:
232–234, 1999.
65. DA Hammons, MC Richardson, BHT Chai, M Bass. Spectroscopic properties of
Ce 3⫹ in orthosilicate, garnet, and fluoride crystals. OSA Trends in Optics and Photon-
ics. Washington DC: Optical Society of America, vol 10, Advanced Solid State
Lasers, 1997, p 35.
66. MA Dubinskii, KL Schepler, VV Semashko, R Yu. Abdulsabirov, BM Galjautdinov,
SL Korableva, AK Naumov. Ce 3⫹ : LiBaF 3 as new prospective active material for
tunable UV laser with direct UV pumping. OSA Trends in Optics and Photonics.
Washington, DC: Optical Society of America, vol 10, Advanced Solid State Lasers,
1997, p 30.

Copyright © 2002 by Taylor & Francis

12
Diode-Pumped Picosecond UV
Lasers by Nonlinear Frequency
Conversions

François Balembois, Frédéric Druon, Patrick Georges,

and Alain Brun
Laboratoire Charles Fabry de l’Institut d’Optique, Orsay, France

I. INTRODUCTION

For many applications there is an intense interest in simple, compact, all-solid-

state pulsed ultraviolet (UV) laser sources. For example, fluorescence lifetime
spectroscopy requires UV pulses with a typical pulse duration of 100 ps, which
is shorter than the fluorescence lifetime of a large number of molecules. This
application also requires a high repetition rate, typically between 10 kHz and 1
MHz, to be compatible with photon counting detection chains and to ensure a
fast acquisition process of the fluorescence decay signal. The average power re-
quired is at the milliwatt level. For example, at a 10 kHz repetition rate, the pulse
energy required to reach the milliwatt average power level is 0.1 µJ. To answer
the need for compactness, efficiency, and reliability, the key sources are semicon-
ductors. However, laser diodes cannot directly produce energetic UV short pulses
with high spatial quality. It is necessary to convert the light emitted by laser
diodes with laser crystals that can produce energetic picosecond pulses in the
near infrared and with frequency converters (nonlinear crystals) in order to reach
the UV band.
The laser sources we have developed are based on diode-pumped solid-
state laser crystals emitting in the near infrared. They all consist of an oscillator
producing picosecond pulses, an amplifier to boost the energy, and nonlinear
stages to reach the UV. The sources operate at a high repetition rate between 10

Copyright © 2002 by Taylor & Francis

and 65 kHz, limited either by the storage lifetime of the crystals or by the ampli-
fication process itself.
Three techniques have been used to produce the picosecond pulses: active
mode locking of a laser cavity, gain-switching of a laser diode, and passive Q-
switching of a microlaser with a very short cavity length.
The first UV laser source we developed is based on a tunable diode-pumped
Cr : LiSAF mode-locked oscillator. The output pulses were amplified in an indi-
rectly diode-pumped tunable Ti:sapphire regenerative amplifier. Frequency tri-
pling was achieved by two LBO (LiB 3 O 5) crystals to reach the UV band (273–
285 nm).
The second laser source used a gain-switched laser diode for seeding the
previous Ti: sapphire regenerative amplifier. An LBO crystal and a βB 2 O 4 (BBO)
crystal were used to convert the wavelength to the UV (268–280 nm).
The last source was based on diode-pumped Nd-doped crystals. A Nd :
YAG microchip laser passively Q-switched by a Cr :YAG crystal was amplified
in a geometrical passive multipass Nd:YVO 4 amplifier. KTP or LBO crystals
were used for second harmonic generation. LBO and BBO crystals were used for
frequency tripling to 355 nm and frequency quadrupling to 266 nm, respectively.

II. TUNABLE DIODE-PUMPED UV CHAIN BASED

ON A MODE-LOCKED Cr: LiSAF OSCILLATOR
AND Ti: SAPPHIRE REGENERATIVE AMPLIFIER

This work has been done in collaboration with the University of Limoges in
France (IRCOM) [1].
Tunable UV can be very useful to excite specific molecules in their absorp-
tion bands. Tunable laser crystals with a large fluorescence bandwidth are not
very numerous. In the UV, Ce 3⫹ :LiSAF (Ce 3⫹ : LiSrAlF 6) has proven to be an
efficient crystal [2]. However, it needs to be pumped by a frequency-quadrupled
Nd :YAG laser operating at a low repetition rate [2], so it is not convenient for
our application. In the visible spectrum, praseodymium-doped YLF (Pr: YLiF 4)
is the only efficient laser crystal but it is not tunable [3]. Moreover, it has to be
pumped in the blue by an ionized argon laser. In the near infrared, the Ti: sapphire
crystal has a wide fluorescence bandwidth between 680 and 1080 nm. However,
it has to be pumped in the blue–green wavelength range where no high-power
laser diodes are available. In the same wavelength range, a new crystal from the
colquiriite family was developed in the early 1990s; Cr :LiSAF (Cr 3⫹ : LiSrAlF 6).
This presents an absorption bandwidth in the red where high-power InGaAlP
laser diodes (up to 4 W) are available. We have investigated Cr :LiSAF lasers
in different regimes and under different pumping schemes [4,5]. This crystal
appeared to work well at low pump power levels (typically 1 W maximum) but

Copyright © 2002 by Taylor & Francis

suffered thermal problems at higher pump power level. Moreover, the small sig-
nal gain remains limited in this crystal to 1.1 (typically), which is not enough to
develop an efficient amplifier. These different results have led us to use a Cr:
LiSAF crystal only for a low-power oscillator producing the picosecond pulses.
For that, the only solution is a mode-locked oscillator. Concerning the amplifier,
we decided to use a Ti:sapphire crystal that combines excellent physical proper-
ties (high thermal conductivity, high emission cross section) with an emission in
the fluorescence bandwidth of Cr :LiSAF. We pumped it with a diode-pumped
frequency-doubled Nd :YVO 4 laser emitting pulses at 532 nm in Q-switched op-
eration.

A. Actively Mode-Locked Cr :LiSAF Oscillator

The oscillator is a diode-pumped actively mode-locked Cr :LiSAF laser (Fig. 1).
The cavity was a Z-fold cavity including the gain medium at Brewster angle, an
acousto-optic modulator for mode-locking, and a Lyot filter to tune the laser
wavelength. The Cr : LiSAF was pumped by a red diode emitting 400 mW at 670
nm. The pumping optics comprised a collimator, a cylindrical telescope, and a

Figure 1 Experimental set up for the Cr : LiSAF, Ti :sapphire laser system used for UV
generation.

Copyright © 2002 by Taylor & Francis

doublet focusing the beam inside the Cr :LiSAF crystal. This oscillator produced
tunable pulses from 820 to 880 nm at a repetition rate of 100 MHz and with a
typical pulse duration of 80 ps. At a wavelength of 850 nm, corresponding to
the maximum of the gain curve for Cr :LiSAF, the energy per pulse was 0.15 nJ.
The peak power was only 2 W at this wavelength, making any efficient nonlinear
conversion (doubling and tripling) impossible.

B. Ti:Sapphire Regenerative Amplifier

In order to get higher peak powers, we increased the energy per pulse in the
regenerative amplifier. Because Cr: LiSAF has a long fluorescence lifetime
(67µs), it has a good energy storage capability and seems to be well suited for
an amplifier. In fact, a diode-pumped Cr : LiSAF regenerative amplifier has al-
ready been developed [6]. However, this system required four pumping diodes
and also intracavity elements with very low losses, since the double-pass gain
in this configuration was low (lower than 1.25). We have shown that the small
signal gain in Cr: LiSAF crystals pumped in continuous wave is limited by up-
conversion and thermal quenching of fluorescence [7]. These effects all make
the use of a highly doped crystal unsuitable. Moreover, the use of a low doped
crystal leads to a bad overlap between the pump beam and the cavity beam.
Therefore, despite its good storage capability, we concluded that a diode-pumped
Cr : LiSAF is not suitable as an amplifier, because the energy extraction is limited
by its low gain.
We investigated another possibility with a Ti: sapphire crystal, since it has
a common fluorescence band with the Cr :LiSAF crystal. This crystal is well
suited to amplification since it has an emission cross section eight times higher
than that of Cr: LiSAF. However, its absorption band, located in the blue–green
range, makes direct diode-pumping currently impossible. To overcome this prob-
lem, one solution is to pump the Ti:sapphire crystal in the green using a diode-
pumped frequency-doubled Nd :YVO 4 laser. Another problem is that Ti:sapphire
has a poor storage capability due to its low fluorescence lifetime (3 µs). One
solution is to Q-switch the pump laser. By this means, the pump energy is deliv-
ered to the crystal in a duration much shorter than its lifetime (typically 50 ns).
Moreover, since the Nd:YVO 4 crystal has both good storage capability for pump
energy (lifetime of 115 µs) and a very good small signal gain (emission cross
section of 30 ⫻ 10 ⫺19 cm 2), Q-switched operation under cw diode-pumping is
very efficient. It can be all the more efficient since high-power and high-bright-
ness laser diodes already exist at the pump wavelength (808 nm).
Following these ideas, we developed a Ti:sapphire regenerative amplifier
(Fig. 1). The pump laser used only one laser diode emitting 4 W at 808 nm (SDL
2380). After beam-shaping, the pump light is focused on the Nd:YVO 4 crystal.
The first mirror of the cavity is directly coated on the laser crystal (M 1). The

Copyright © 2002 by Taylor & Francis

others (M 2 and M 3) are concave mirrors with a radius of curvature of 100 mm.
At the waist located between M 2 and M 3 , we placed a KTP crystal for second-
harmonic generation. All the cavity mirrors are highly reflective at 1064 nm; M 3
was also highly reflective at 532 nm while M 2 has a high transmission at 532
nm. An acousto-optic modulator was introduced near M 1 to Q-switch the laser.
This source produced 40 µJ green pulses with a duration of 50 ns at a repetition
rate of 10 kHz.
These pulses were used to pump the Ti:sapphire crystal of the regenerative
amplifier (Fig. 1). It was a three-mirror cavity with two concave mirrors (100
mm radius of curvature) around the crystal and a plane mirror. All these mirrors
supported a high-reflectivity coating between 800 and 900 nm. The Ti:sapphire
crystal was 13 mm long and absorbed 70% of the pump power at 532 nm. We
recycled the transmitted 30% of the pump power using a concave mirror placed
after the second curved mirror of the cavity. For simplicity, this mirror is not
shown in Figure 1. A prism was used to tune the amplifier in order to match the
operating wavelength of the oscillator, which is shifted from the maximum of
the gain curve of Ti:sapphire. A Pockels cell and a polarizer were used to inject
a single pulse prior to the amplification and to dump it when the amplification
process reached its optimum. A two-way switch consisting of a Faraday rotator,
a half-wave plate, and a polarizer was used to separate the ejected beam from
the injected one.
At 820 nm, we measured a small signal gain per double pass higher than
1.6. The build-up time to reach the saturation of the gain was relatively low
(200 ns corresponding to 50 cavity round trips). As a comparison, the build-up
time was 820 ns (corresponding to around 175 round trips) in the case of the
Cr :LiSAF regenerative amplifier developed by Mellish et al. [6]. The energy of
the amplified pulses (vs. wavelength) is shown in Figure 2. The system (oscillator
and amplifier) could be tuned between 820 and 870 nm. The shortest wavelength
was imposed by the Cr :LiSAF crystal, whose gain curve is centered at 850 nm.At
820 nm the oscillator was near the oscillation threshold. The longest wavelength
was fixed by the Ti:sapphire crystal, whose gain curve is centered at 800 nm.
However, the common fluorescence band between these two crystals is sufficient
to ensure output pulses with an energy of more than 3 µJ over a bandwidth of
30 nm from our system.

C. Harmonic Generation
The amplified pulses were frequency-converted to the blue and to the UV by
harmonic generation. Since the peak power in the infrared remained low, tight
focusing of the amplified laser beam was needed to ensure a good conversion
efficiency. The small size of the beams in interaction required us to consider the
strong influence of the walk-off angles for the choice of the most efficient crystal.

Copyright © 2002 by Taylor & Francis

Figure 2 Energy per pulse in the infrared vs. wavelength.

Around 800 nm, three nonlinear crystals provide efficient second-harmonic

generation (SHG): LBO, BBO, and LiIO 3 [8]. We selected an LBO crystal as
described by Nebel and Beigang [9]. Although BBO and LiIO 3 can lead to better
efficiency [9], they have a walk-off angle significantly larger than LBO. Crystals
with a large walk-off produce a harmonic beam with a poor spatial quality, which
is a drawback for subsequent nonlinear mixing processes. We chose type I phase
matching corresponding to a nonlinear coefficient deff ⫽ 1.25 pm/V (at λ ⫽ 830
nm), which is 2.5 times larger than that for type II phase-matching.
For the third-harmonic generation (THG), we selected an LBO crystal with
a type I phase matching, as in [10], for the following reasons: (a) to have a good
spatial profile of the UV beam for efficient coupling into an optical fiber we
rejected the BBO because of its large walk-off (nearly 5 degree) even if a higher
conversion efficiency could be expected with this crystal [9]; (b) LiIO 3 is not
transparent below 300 nm; and (c) type II phase matching is not possible for
THG in LBO in this wavelength range.
The set-up for doubling and tripling is shown in Figure 1. A lens L 1 (200
mm focal length) focused the infrared beam inside the first LBO crystal. This
crystal was cut for type I phase matching at 850 nm (θ ⫽ 90 degrees, φ ⫽ 27
degrees) and its length was 8 mm. We obtained blue pulses with an energy of
more than 0.5 µJ over a bandwidth of 10 nm (Fig. 3). By taking a gaussian
temporal shape for the infrared (IR) pulse, we calculated that the pulse duration
in the blue was 60 ps.

Copyright © 2002 by Taylor & Francis

Figure 3 Energy per pulse in the blue vs. wavelength.

The two beams were then collimated by a 100 mm focal-length lens (L 2)

and separated by a dichroic mirror. A half wave plate rotated the polarization in
the infrared in order to satisfy the type I phase-matching conditions in the second
LBO crystal. This crystal was cut in type I phase matching for sum frequency
generation of 860 nm and 430 nm (θ ⫽ 90 degrees; φ ⫽ 61 degrees). Its length
was 8 mm. Another dichroic mirror recombined the two pulses and a lens L 3 (90
mm focal length) focused the beam inside the second LBO crystal.
The UV pulses have an estimated duration of 50 ps, calculated by assuming
a gaussian temporal shape for the IR pulses. Figure 4 shows the energy per pulse
in the UV as a function of the wavelength. We obtained more than 0.1 µJ over

Figure 4 Energy per pulse in the UV vs. wavelength.

Copyright © 2002 by Taylor & Francis

a bandwidth of 6 nm between 274 and 280 nm. Since the repetition rate was 10
kHz, this corresponded to more than 1 mW of average power. These perfor-
mances could be improved by the use of suitably cut crystals, since the experi-
ments reported here were carried out with crystals cut for SHG of 850 nm radia-
tion and SFG of 860 nm and 430 nm wavelengths. Higher UV powers may be
obtained with a BBO THG crystal at the expense of a highly elliptical beam
shape.

III. GAIN-SWITCHED PICOSECOND LASER DIODE

SEEDING A TI:SAPPHIRE REGENERATIVE AMPLIFIER

This work was has been performed in collaboration with the Institut d’Electro-
nique Fondamentale in Orsay (France) [11].
To simplify the source described in Sec. II, we have investigated the possi-
bility of replacing the mode-locked Cr: LiSAF oscillator with a semiconductor
laser. In this case, tuning can be achieved, for example, by varying the tempera-
ture of the laser diode. For a given semiconductor laser, tunability is on the order
of 10 nm in the infrared, for a temperature excursion range of 30°C, but it would
be possible to have a few low-cost semiconductor lasers with different emission
wavelengths to fulfill the right tuning range.
Compact systems have been realized with a mode-locked external-cavity
laser diode producing femtosecond pulses [12] and used to seed a high-energy
flashlamp-pumped regenerative amplifier. However, the mode-locking technique
has a disadvantage: it produces pulses at a very high repetition rate (100 MHz
typically). This pulse rate is then decreased to the slower repetition rate of the
amplifier (typically 1–250 kHz). This means that almost all the oscillator power
(more than 99.99%) is lost for the amplification process.
Conversely, a laser diode can operate in the gain-switched regime at the
right repetition rate and can produce picosecond pulses (below 100 ps), thanks
to its high gain and low cavity lifetime. Such very simple pulsed sources have
already been used in combination with amplifiers. Seeding an erbium-doped fiber
amplifier with gain-switched laser diode pulses has been reported around 1.5
µm [13]. A Ti: sapphire regenerative amplifier, pumped by a flashlamp pump
frequency-doubled Nd: YAG laser, and a flashlamp-pumped alexandrite regener-
ative amplifier were both injected by a picosecond gain-switched laser diode
[14,15]. Energies reaching the millijoule level at a few Hertz repetition rate were
reported.
In our case, we have used the same indirectly diode-pumped Ti:sapphire
regenerative amplifier as described in Sec. II. We replaced the entire Cr: LiSAF
mode-locked oscillator by a gain-switched laser diode driven by a pulse gener-
ator.

Copyright © 2002 by Taylor & Francis

A. Gain-Switched Laser Diode as Seeder
of the Regenerative Amplifier
The laser diode consisted of a separate-confinement single-quantum-well struc-
ture with a 170 Å GaAs-active layer in an AlGaAs waveguide. The laser stripe
was 5 µm wide and 2 mm long, and lasing occured on a single transverse mode.
To increase the output efficiency, the front face was antireflection coated (reflec-
tion coefficient of 1%) and the back face was high reflection coated (reflection
coefficient of 95%). The gain-switching regime was achieved by pumping the
laser diode with a pulse generator capable of delivering 200 ps current pulses at
repetition rates up to 100 kHz [16]. In these conditions, the laser output spread
over many longitudinal modes. Due to chirping effects on each longitudinal
mode, the pulse spectrum was quasicontinuous and spanned from 800 to 840 nm
with a maximum amplitude at 812 nm (Fig. 5). The pulse duration was 200 ps
and the pulse energy was 22 pJ.
The experimental set up is shown schematically in Figure 6. The picosecond
laser diode output beam was collimated with a high numerical aperture (NA)
objective L 1 (NA ⫽ 0.5, focal length ⫽ 8 mm). Two cylindrical lenses with focal
lengths of ⫹100 mm and ⫺25.4 mm, respectively, were used to reshape the output
beam and match the beam profile to the regenerative amplifier. The amplifier and
its pump laser have been previously described in Section II. To prevent all feed-
back from the amplifier to the laser diode, we added a Faraday isolator. The
overall isolation from the amplifier to the picosecond laser diode was then 60 dB.
All the devices were synchronized with the internal clock of the Pockels
cell driver (MEDOX Electro-optics). A first voltage step was applied to the driver
of the acousto-optic modulator to trigger the pump pulse at 532 nm. A second
voltage step was applied to the pulse generator of the picosecond laser diode.
The delay between the two voltage steps was adjusted to synchronize the picosec-

Figure 5 Spectrum of pulses emitted by the gain-switched laser diode.

Copyright © 2002 by Taylor & Francis

Figure 6 Experimental set up of a Ti: sapphire amplifier seeded by a gain-switched
laser diode. Focal lengths of lenses L 1 , L 2 , L 3 , and L 4 are 8 mm, 200 mm, 100 mm, and
90 mm respectively.

ond laser diode pulse injected into the regenerative amplifier with the green pump
pulse. Two high-voltage steps were applied to the Pockels cell. The first voltage
step was used to trap the picosecond pulse inside the cavity. The second one was
used to eject it, once the amplification process had reached its maximum.

B. Performances of the Ti:Sapphire

Regenerative Amplifier
Because of the presence of a prism in the regenerative amplifier cavity, only a
small window of 5 nm full width at half maximum (FWHM) was selected in the

Copyright © 2002 by Taylor & Francis

spectrum of the picosecond laser diode. The central wavelength could be adjusted
simply by rotating the plane mirror M 3 of the amplifier (Fig. 6). By considering
the transmission of the different optics and the shape of the laser diode spectrum
(Fig. 5), we estimated that the energy actually injected into the amplifier was 1.6
pJ at 812 nm (i.e., at the laser diode spectrum maximum), whereas it was only
0.3 pJ at 803 nm (at the edges of the laser diode spectrum). This energy level
was very low compared to the energy of the pulses coming from the Cr: LiSAF
oscillator used in the previous scheme (150 pJ at 830 nm). Figure 7 illustrates
the amplification process at 812 nm and 803 nm, obtained with a fast photodiode
(risetime 1 ns) monitoring a leakage of one regenerative amplifier mirror. As can
be seen, the amplified spontaneous emission (ASE) cannot be neglected during
this amplification process. To characterize the amplifier seeded by the gain-
switched laser diode, it is important to evaluate the signal-to-background ratio
(SBR). The SBR is given by the ratio of the signal energy to the ASE energy
emitted during the signal pulse. The SBR can be calculated with the data given
in Figure 7, corrected by the the ratio of the FWHM response of the photodiode
to the actual pulse width [17]. The SBR was 314 at 812 nm, of the same order
of magnitude as those previously obtained by Hariharan et al. [17], showing that
the amplifier was correctly seeded. At 803 nm, the actual SBR was 43. Despite
its higher sensitivity to alignment at this wavelength, the amplifier could easily
be injected. Moreover, once the alignment was achieved, the SBR remained the
same during all our experiments (2 hr).
To obtain the injection threshold of our amplifier, we decreased the energy
emitted by the laser diode until the measured SBR was equal to 1. We measured
the average power emitted by the laser diode at the input of the amplifier. By

Figure 7 Pulse amplification at (a) 812 nm and (b) 803 nm obtained from a fast photodi-
ode (1 ns risetime) monitoring leakage of one mirror of the regenerative amplifier.

Copyright © 2002 by Taylor & Francis

taking into account the laser diode spectrum and the spectral window of 5 nm
selected by the amplifier, we estimated that the injection threshold was 0.2 pJ
per pulse at 812 nm.
We next studied the ejected amplified pulse. The amplifier output pulse
consisted of two parts: an amplified picosecond pulse and a pulse of ASE with
a duration of 5 ns that corresponded to one roundtrip in the amplifier cavity.
Figure 8 shows that despite ASE, it was possible to generate picosecond pulses
at a 10 kHz repetition rate with an energy between 1 and 4 µJ and a tuning range
from 803 to 840 nm.
The temporal width of the amplified pulses was measured with a sampling
oscilloscope (30 ps rise time) and a high-speed photodiode (25 ps rise time). A
value of 70 ps (no deconvolution) was found for the amplified pulsewidth at 812
nm (Fig. 9). As we tuned the amplifier, we observed that the pulsewidth remained
between 70 and 100 ps. Note that the amplified pulses were shorter than those
emitted by the laser diode (200 ps). This result actually reflects the spectral selec-
tivity of the amplifier: because the different wavelengths are emitted at different
times during the laser pulse, selecting a wavelength is just equivalent to selecting
a temporal window in the gain-switched pulse.
The sampling oscilloscope was triggered by an electrical signal delivered
by the power supply of the picosecond laser diode. The pulse jitter deduced from
the trace thickness (Fig. 9) was within the measurement accuracy (30 ps). Actu-
ally, timing jitter less than 1 ps can be expected from a gain-switched laser diode
system. Present results show that the amplified optical pulses can be precisely

Figure 8 Tunability of infrared pulses around 820 nm. Circles represent the total energy
extracted per pulse from the amplifier (including ASE). Crosses represent the energy of
the amplified laser pulses. Squares correspond to ASE.

Copyright © 2002 by Taylor & Francis

Figure 9 Temporal shape of the amplified optical pulse measured with a sampling oscil-
loscope.

synchronized with an electrical signal. This feature could be useful in multiple

optoelectronic applications.

C. Frequency Conversion to Blue and UV

The amplified pulses were frequency-converted to the blue and the UV by har-
monic generation (Fig. 6). In contrast to the set up used for the first source (Sec-
tion II), we used a BBO for third-harmonic generation (8 mm length, cut in type
I for sum frequency generation of 780 nm and 390 nm) and we introduced a
delay line to adjust the optical path length of the infrared beam finely to that of
the blue beam. The accuracy of the path length in the harmonic generator needed
to maximize the UV signal was approximately 2 mm.
Spectral widths of 1 nm and 0.3 nm (FWHM) were measured in the blue
and UV, respectively. This may be compared with the 5 nm (FWHM) spectrum
of the infrared pulses. Spectral widths were limited by the spectral acceptance
of the two nonlinear crystals. Although, the LBO and BBO crystals were not
optimized for the operating wavelength range (803–840 nm), we obtained good
conversion efficiency: reaching 30% from IR to blue and 12% from IR to UV.
We achieved a tunability of more than 15 nm in the blue and of more than 10
nm in the UV (Fig. 10). In both cases, the pulse energy was above 0.1 µJ. For
a 10 kHz repetition rate, average powers larger than 11 and 4 mW were achieved
at 406 and 271 nm, respectively. As compared with the frequency conversion
stage used in Sec. II, we obtained a more powerful UV beam but also more
elliptical. This was due to the lower angular acceptance of BBO than LBO and
to a higher nonlinear coefficient.

Copyright © 2002 by Taylor & Francis

Figure 10 Tuning curves for the second- and third-harmonic generation.

To estimate the maximum contribution of ASE to the UV signal, we oper-

ated the amplifier in free running mode (i.e., without seeding) with an output
energy per pulse of 4 µJ. We measured a UV energy per pulse lower than 2 nJ;
that is, more than two orders of magnitude lower than the energy of picosecond
UV pulses. Thus, by means of the nonlinear conversion process, the pulse was
removed from its nanosecond ASE background.
The output level in the UV (at 271 nm) was monitored for 2 hr. Over this
period, the seeding of the amplifier remained unchanged and the SBR remained
at its initial value of 314 (at 812 nm). Peak-to-peak fluctuations were found to
be ⫾2.5%, corresponding to IR fluctuations of less than ⫾0.8%. Such good sta-
bility obtained in spite of a nonlinear conversion process reflects the very good
reproducibility of pulses emitted by the gain-switched laser diode as well as the
stable pumping of the regenerative amplifier by a continuous wave (CW) laser
diode.
The repetition frequency was increased to 25 kHz, which was the maximum
value acceptable for our Pockels cell. Because of the long fluorescence lifetime
of the Nd 3⫹ ion in the yttrium vanadate (YVO 4) matrix (115 µs), the frequency-
doubled Nd:YVO 4 laser then emitted green pulses of lower energy (18 µJ at
25 kHz instead of 40 µJ at 10 kHz). However, we still obtained 1 mW average
power in the UV at 271 nm.

D. Improvement of Spectral Quality

Harmonic conversion efficiency was limited by the wide spectrum of the infrared
pulses (5 nm FWHM). Moreover, it could be useful for spectroscopic applications

Copyright © 2002 by Taylor & Francis

to emit pulses with narrower spectra. For these reasons, we modified the injection
setup shown in Figure 11. We used a 2000 groove/mm grating that spatially
dispersed the beam of the gain-switched laser diode. Because the injection was
highly sensitive to alignment, we could rotate the grating to select finely the
wavelength of the pulses seeded into the amplifier. The optical path length be-
tween the grating and the input of the regenerative amplifier was 140 cm. With
this setup, the amplified spectrum was less than 1 nm FWHM compared with 5
nm FWHM in the former experiment. The pulse duration remained unchanged:
between 70 and 100 ps.
At a wavelength of 812 nm corresponding to the maximum of the seeded
energy, the SBR was equal to 43. It was lower than previously measured (SBR
of 314) because the energy seeded into the amplifier was only 0.3 pJ (compared
with 1.6 pJ). A smaller part of the diode spectrum was selected and the transmis-
sion of the grating was only 80%.
We also compared the frequency conversion in the two cases with and
without the grating. At the same injection level (with an SBR of 43) and at the
same wavelength (271 nm), we obtained a UV average power of 3.4 mW with
the grating and 1 mW without the grating. As expected, the spectral narrowing
of the fundamental wavelength increased the nonlinear conversion efficiencies.
We achieved tunability by simultaneously rotating the grating and the back
mirror of the regenerative amplifier (Fig. 6) in order to match the wavelength of
the injected beam with the free running wavelength of the amplifier. The system
was tunable between 810 nm and 817 nm with SBR between 20 and 43. The
UV average power was between 1 mW and 3.4 mW in the spectral range 270–
272.3 nm. The performance quality was lower than before because more than
the half of the energy ejected from the amplifier was lost in ASE. However,
the average powers obtained would be sufficient for time-resolved spectroscopic
studies such as those performed on biological media.

Figure 11 Set up for the injection seeding with a grating.

Copyright © 2002 by Taylor & Francis

IV. Q-SWITCHED ND:YAG MICROLASER SEEDING
A Nd :YVO4 AMPLIFIER

This work has been performed in collaboration with Nanolase (France) [18].
The third technique investigated was based on the following observation:
widely tunable crystals are generally associated with relatively low emission
cross-sections and therefore with low small-signal gain. In the case of our pre-
vious Ti: sapphire regenerative amplifier, the small signal gain was approximately
1.6 per double pass. More than 50 passes were needed to reach the microjoule
level. The regenerative amplifier with its costly Pockels cell is the only efficient
solution for extracting the energy stored inside the Ti:sapphire crystal.
Conversely, the emission cross section is much higher for crystals emitting
at fixed wavelengths. For example, the peak emission cross section is 30 ⫻ 10 ⫺19
cm 2 for Nd: YVO 4 crystals compared with 4.1 ⫻ 10 ⫺19 cm 2 for Ti: sapphire crys-
tals. This means that the stored energy can be extracted in a few passes in the
case of a Nd: YVO 4 amplifier with a simple geometrical multipass amplification
scheme. For example, a small signal gain of 10 per pass can be easily achieved
in a diode-pumped Nd:YVO 4 crystal: only six passes are needed to increase the
pulse energy from the pJ level to the µJ level. For this case, the amplification
process is totally passive; no active element such as a Pockels cell limits the
repetition rate.
Different seeding sources are available for an Nd-doped amplifier operating
at 1.06 µm. If we exclude the solution of a mode-locked oscillator for the reasons
mentioned in Sec. III, it is possible to use a gain-swiched semiconductor laser
operating at 1.06 µm. However, the energy per pulse is only in the picojoule
range, due to the short fluorescence lifetime in semiconductors. The second possi-
bility is to use a passively Q-switched diode-pumped Nd :doped microlaser [19–
21]. In this case, thanks to the long fluorescence lifetime of the Nd 3⫹ ion (230
µs in YAG), the energy per pulse is six orders of magnitude higher, reaching the
microjoule level. The high energy level of the seeded pulses contributes to a good
extraction of the stored energy inside the amplifier within a few passes. Moreover,
this monolithic device is very simple to use because of its totally passive process
and extremely compact due to the small size of the microchip and the diode-
pumping system.
However, it is difficult to fulfill the source requirements defined in the
introduction (high repetition rate and picosecond pulses) with a passively Q-
switched microchip laser in standard use. To overcome this problem, we used a
high-brightness laser diode to pump a passively Q-switched Nd :YAG microchip
laser. Afterwards, we amplified the output pulses in a Nd :YVO 4 multipass ampli-
fier.

Copyright © 2002 by Taylor & Francis

A. Passively Q-Switched Microchip Laser
Our experiment was performed with a microchip laser composed of a 750 µm
long piece of gain medium (Nd 3⫹ : YAG) on which was grown, by liquid phase
epitaxy, a 60 µm layer of saturable absorber (Cr 4⫹ :YAG) [22]. The concentration
of Nd 3⫹ and Cr 4⫹ ions was 1.2 ⫻ 10 20 cm ⫺3 and 2.18 ⫻ 10 18 cm ⫺3, respectively.
The output–coupler–mirror transmission was 5%.
We measured a pulse duration of 500 ps with a fast photodiode and a fast
sampling oscilloscope. The predicted pulse duration given by the theory [23,24]
for this device was 520 ps, which is in good agreement with our experimental
measurement.
This kind of short-pulse duration microchip laser is usually pumped with
a fiber-coupled CW laser diode and would produce pulses at a relatively low
repetition rate. For example, with a 0.85 W, 100 µm core-diameter fiber-coupled
diode, the repetition rate would be 5.9 kHz and the average power 15 mW. In
fact, for the production of 500 ps pulses, our microchip concept consists in having
a short length of gain medium and a saturable absorber with a high ion concentra-
tion, both of which are incompatible with a high repetition rate.
To increase the repetition rate, the idea would be to modify the microchip
laser structure. To keep the pulse duration short, one should make this modifica-
tion for the Nd concentration, which requires a different technology. An alterna-
tive method of obtaining a higher repetition rate is to optimize the pumping sys-
tem. We found experimentally that the higher the power density on the laser
diode-emitting area, the higher the repetition rate for a given pump power (Table
1). First, we used a standard 1 W, 1 ⫻ 100 µm 2 junction CW laser diode whose
beam was reshaped using cylindrical lenses optimized for the repetition rate. This

Table 1 Influence of Pump System Brightness on Microlaser Performance

0.85 W 1W, 1W, 2W,
LD fiber coupled 1 ⫻ 100 µm 2 1 ⫻ 50 µm 2 1 ⫻ 100 µm 2
Pump at 808 nm 100 µm diameter LD LD LD Ti:Sapphire

Incident pump power 0.85 0.9 0.9 1.8 0.67

(W)
M 2 (pump source) 18 10 5 10 1
Repetition rate (kHz) 5.9 21 41 45 101
Energy per pulse 2.5 0.58 0.37 0.89 0.28
(µJ)
Average power 15 15 15 40 28
(mW)

LD, laser diode.

Copyright © 2002 by Taylor & Francis

technique allowed, with approximately the same incident power, a repetition rate
3.6 times higher than in the case of the fiber-coupled diode. A second step was
to use very high brightness diodes. We used a 1W, 1 ⫻ 50 µm 2 laser diode,
which improved the repetition rate by a factor of 6.9. Finally, we pumped with
a 2W, 1 ⫻ 100 µm 2 CW laser diode from SDL, Inc. Because its power density
was the same as the previous 1W, 1 ⫻ 50 µm 2 diode, the repetition rate was
approximately the same, but its power was twice as high, which allowed a higher
average output power of 40 mW.
In order to estimate an upper limit to the repetition rate, at a given pump
power, we pumped the microchip laser with a diffraction-limited beam at 808
nm from a CW Ti:sapphire laser. The highest repetition rate obtained was 101
kHz at 0.67 W incident pump power. This pump power corresponded to the limit
for single-longitudinal-mode operation with a stable pulse train. With a higher
incident power, multiple longitudinal modes began to appear, and the pulse train
became erratic.

B. Nd:YVO 4 Passive Multipass Amplifier

To increase the energy per pulse, but without reducing the repetition rate, we
used a geometric multipass amplifier configuration. The main advantage of a
multipass amplifier is its totally passive process, which induces no repetition-
rate limitation. For simplicity, efficiency, and compactness we chose a Nd:YVO 4
crystal because of its high-emission cross section at the 1064 nm-emitted wave-
length of the Nd :YAG/Cr: YAG microchip laser, which favored good energy
extraction even in a configuration having few round trips. Moreover, to reduce
the thermal effects, we used a composite crystal [25] made with a 1 mm long
undoped YVO 4 crystal bonded to a 1% doped 3 mm long Nd:YVO 4 crystal.
Since the pulse seeded in the amplifier had significant energy (fraction of µJ),
the energy stored in this crystal could be extracted in a maximum of four passes.
The experimental setup is shown in Figure 12. For the first two passes, the
signal beam was focused into the crystal by a lens (L2), reflected on the back of
the crystal (which was HR coated at 1.064 µm), and recollimated with the same
lens. The beam could be intercepted at this point in order to evaluate the two-
pass-amplifier performance. For the four-pass amplifier, the beam was reinjected
into the gain medium after a small spatial shift. To optimize the third and fourth
passes, and to compensate for the thermal lensing that occurred during the first
two passes, the beam propagated through a slightly-detuned 1: 1 telescope (L3–
L3).
To have a high gain per pass and thus further increase the extraction effi-
ciency of the amplifier, we used a high-brightness 2W, 1 ⫻ 100 µm 2 pump diode.
The use of high-brightness diodes seemed to be optimal for an efficient multipass
amplifier. For example, after two passes, the gain obtained with a more powerful

Copyright © 2002 by Taylor & Francis

Figure 12 Experimental setup for the Nd :YAG microlaser amplified in an Nd :YVO 4
amplifier.

but less bright diode (4W, 1 ⫻ 500-µm 2) was 2.5 times lower than the brighter
diode. Despite a higher amount of deposited energy in the crystal with the 4W
diode, the maximal (after four passes) energy extraction remained almost the
same as with the high brightness diode.
With the same objective of having a simple and efficient amplifier, we
studied the influence of the injection power (Fig. 13). The energy extraction was
naturally higher with a 40 mW injection than the 15 mW one. This explains why
a 2W, 1 ⫻ 100 µm 2 diode was also used to pump the microchip laser to obtain

Figure 13 Output power vs. pump power for two different incident signal powers for
two passes or four passes inside the amplifier.

Copyright © 2002 by Taylor & Francis

Figure 14 IR and UV beam profiles in the far field. (a) Oscillator beam profile. (b,c)
Amplified beam profile (at 1064 nm) after two passes and four passes. (d,e) UV beam
profile (355 nm) in the two configurations (two passes and four passes).

a highly efficient amplifier. The average output powers were 408 mW after two
passes and 510 mW after four passes at 1064 nm.
We recorded the beam profiles and the average output power for the two-
pass and four-pass configurations. We observed that the beam profile shrank in
the horizontal direction during the third and fourth passes (Fig. 14 a–c). This
could be explained by the fact that the signal beam of the third and fourth passes
was amplified on the edges of the gain area. However, the M 2 of the amplified
beam was nearly the same in the two-pass and in the four-pass configuration, on
the order of 1.2.
The choice between the four-pass configuration, which delivered the maxi-
mum output power, and the simpler two-pass configuration, which delivered less
energy but had a improved beam profile, was made based on the efficiency of
harmonic generation by nonlinear processes.

C. Nonlinear Conversion Stage

For the nonlinear processes, we first used a 5 mm long type II KTP (KTiOPO4)
crystal for second-harmonic generation. The output green power (at 532 nm)
was up to 180 mW (4 µJ per pulse, 37% conversion efficiency) in the four-pass
configuration, and 120 mW (2.33 µJ per pulse, 31% conversion efficiency) in
the two-pass scheme. Subsequently, for optimization of the third harmonic, the
IR and the green were separated after the doubling crystal in two different arms

Copyright © 2002 by Taylor & Francis

Figure 15 Output power of the source for fundamental and harmonic waves vs. the
pump power of the amplifier.

and recombined to give a better overlap in the second nonlinear crystal (8 mm

long type II LBO). In order to obtain a circular beam profile in the UV (at 355
nm), we used the angular acceptance and the walk-off of the LBO crystal to
reshape the distorted IR beam profile delivered by the multipass amplifier. The
UV beam profile was circular for the two-pass amplifier (Fig. 14d) and slightly
elliptical for the four-pass amplifier (Fig. 14e), but much less than in the infrared
(Fig. 14c). An output power of 38 mW (10% efficiency) using the two-pass con-
figuration was obtained compared to 35 mW (7.2% efficiency) using the four-
pass design. Because of its elliptical shape, the output beam after four passes
could not allow an efficiency as high as that obtained after two passes. The opti-
mal system for both simplicity and efficiency in the UV was the two-pass ampli-
fier configuration. This UV source delivered 0.84 µJ pulses at 45 kHz correspond-
ing to 40 mW of average power (Fig. 15) with a pulse duration shortened by
nonlinear conversion down to 300 ps. The M 2 of the UV beam was 1.34.
We also replaced the LBO crystal used for frequency tripling by a BBO
crystal phase matched for frequency doubling the green output. We obtained 0.5
µJ at 266 nm.

D. Towards Higher Power in the UV

Figure 13 demonstrated the strong influence of the input signal average power
(for a given repetition rate) on the extraction of the power stored inside the Nd :
YVO 4 . The upper limit for average power can be obtained from the crystal perfor-
mance in a cavity in the CW regime. Optical-to-optical conversion efficiency of
50% can be expected from an optimized Nd :YVO 4 laser operating in CW. In

Copyright © 2002 by Taylor & Francis

our case, the upper limit for the average output power from the amplifier was
on the order of 1 W. This suggests that it is possible to extract a higher-output
power from the amplifier with a higher-input signal.
Following this idea, we used a different microchip laser producing higher
average power than the previous one, at the expense of a slight increase in the
pulse duration. The microchip laser consisted of a 1.5 mm long Nd :YAG crystal
bonded with a 100 µm long Cr :YAG. The pulse duration was 800 ps (compared
to 500 ps for the previous one). The repetition rate was 52 kHz and the average
power was a factor of 4 higher (150 mW).
For this case, the power after two-pass-amplification was 800 mW in the
infrared. To simplify the experimental setup, we replace the KTP crystal cut in
type II by an LBO cut in type I for second-harmonic generation at 532 nm. In
this case, frequency tripling can be achieved without separation between the infra-
red and the green beams. After frequency doubling, the power was 120 mW at
532 nm. The frequency tripling was achieved with the same LBO as used previ-
ously. We obtained 100 mW at 355 nm (corresponding to an pulse energy of 1.9
µJ).

V. CONCLUSION

We have presented several diode-pumped sources emitting picosecond pulses in

the UV. All the sources we have developed are based on diode-pumped laser
crystals emitting in the near infrared and on two nonlinear crystals converting
the light to the UV band.
The first two laser sources are based on an indirectly diode-pumped Ti:
sapphire regenerative amplifier seeded either by an actively mode-locked diode-
pumped Cr :LiSAF picosecond oscillator or a gain-switched picosecond laser
diode. Typical performances were 0.1 µJ at 10 kHz (1 mW of average power)
between 270 and 285 nm.
The last laser source described was based on a passively Q-switched Nd :
YAG laser seeding a two-pass Nd:YVO 4 amplifier. We obtained more than 1 µJ
at 40 kHz (40 mW of average power) at 266 nm after frequency conversion.
The average power obtained at 355 nm reached the 100 mW level. As expected,
operation at a fixed wavelength is more simple and more efficient. We are cur-
rently working on a tunable version of this system based on nonlinear crystals cut
for optical parametric generation and amplification: quasi-phase-matched crystals
(periodically poled lithium niobate crystals) can efficiently produce tunable red
light from pump pulses at 532 nm [26]. Subsequently, a frequency doubler can
be used to reach the UV band.
The total optical pump power for all the systems does not exceed 5 W.

Copyright © 2002 by Taylor & Francis

Considering the excellent electrical-to-optical conversion efficiency of the laser
diodes (30%), only a few tens of watts of electrical power are needed to supply
these sources. These compact systems promise to have many potential applica-
tions in the domains of fluorescence spectroscopy, optical sampling, or the semi-
conductor industry.

REFERENCES

1. F Balembois, M Gaignet, F Louradour, V Couderc, A Barthelemy, P Georges, A

Brun. Tunable picosecond UV source at 10 kHz based on an all solid-state diode-
pumped laser system. Appl Physics B 65: 255–258, 1997.
2. CD Marshall, SA Payne, JA Speth, WF Krupke. Ultraviolet laser emission properties
of Ce 3⫹-doped LiSrAlF 6 . In TY Fan, BHT Chai, eds. OSA Proceedings on Advanced
Solid State Lasers. Washington, DC: Optical Society of America, 1994, pp 389–
394.
3. T Sandrock, T Danger, E Heumann, G Huber. Continuous wave laser action of Pr-
doped fluorides at room temperature. In TY Fan, BHT Chai, eds. OSA Proceedings
on Advanced Solid State Lasers. Washington, DC: Optical Society of America,
1994, pp 357–360.
4. F Falcoz, F Balembois, P Georges, A Brun. Self-starting self mode-locked femtosec-
ond diode pumped Cr :LiSAF laser. Optics Letts 20: 1874–1876, 1995.
5. F Falcoz, F Kerboull, F Druon, F Balembois, P Georges, A Brun. Small signal gain
investigations for a continuous-wave diode-pumped Q-switched Cr: LiSAF laser.
Optics Letts 21: 1253–1255, 1996.
6. R Mellish, SCW Hyde, NP Barry, R Jones, PMW French, JR Taylor, CJ Van der
Poel, A Valster. All-solid-state diode-pumped Cr: LiSAF femtosecond oscillator and
regenerative amplifier. Appl Phys B 65: 221–226, 1997.
7. F Balembois, F Falcoz, F Druon, F Kerboull, P Georges, A Brun. Theoretical and
experimental investigations of small signal gain for a diode-pumped Q-switched Cr:
LiSAF laser. IEEE J Quant Electron 33: 269, 1997.
8. CS Adams, AI Ferguson. Tunable narrow linewidth ultra-violet light generation by
frequency doubling of a ring Ti: sapphire laser using lithium tri-borate in an external
enhancement cavity. Opt Commun 90: 89, 1992.
9. A Nebel, R Beigang. External frequency conversion of cw mode-locked Ti: Al 2 O 3
laser radiation. Opt Lett 16: 1729, 1991.
10. L Goldberg, DAV Kliner. Tunable UV generation at 286 nm by frequency tripling
of a high-power mode-locked semiconductor laser. Opt Lett 20: 1640, 1995.
11. F Balembois, M Gaignet, P Georges, A Brun, N Stelmakh, JM Lourtioz. Tunable
picosecond blue and UV pulses from a diode-pumped laser system seeded by a gain-
switched laser diode. Appl Optics 37: 4876–4880, 1998.
12. PJ Deffyett, A Yusim, S Grantham, S Gee, K Kabel, M Richardson, G Alphonse,
J Connolly. Ultrafast semiconductor laser-diode-seeded Cr :LiSAF regenerative am-
plifier. Appl Opt 36: 3375–3380, 1997.

Copyright © 2002 by Taylor & Francis

13. A Galvanauskas, ME Fermann, P Blixt, JA Tellefsen, D Harter, Hybrid diode-laser
fiber-amplifier source of high-energy ultrashort pulses. Opt Lett 19: 1043–1045,
1994.
14. DJ Harter, O Montoya, J Squier, J Nees, P Bado, G Mourou. Short pulse generation
from Ti: doped materials. In: Technical Digest Series vol. 7, Conference on Lasers
and Electro-Optics. IEEE/Lasers and Electro-Optics Society and Optical Society of
America, Anaheim, 1988, paper PD 6.
15. DJ Harter, P Bado. Wavelength tunable alexandrite regenerative amplifier. Appl Opt,
27: 4392–4395, 1988.
16. N Stelmakh, J-M Lourtioz, G Marquebielle, G Volluet, JP Hirtz. Generation of high-
energy (0.3µJ) short pulses (⬍400 ps) from a gain-switched laser diode stack with
sub-ns electrical pump pulses. IEEE J Sel Topics Quant Electron 3: 245–250, 1997.
17. A Hariharan, DJ Harter, TS Sosnowski, S Kane, D Du, TB Norris, J Squier. Injection
of ultrafast regenerative amplifiers with low energy femtosecond pulses from an Er-
doped fiber laser. Opt Commun 132: 469–473, 1996.
18. F Druon, F Balembois, P Georges, A Brun. High-repetition-rate 300-ps pulsed ultra-
violet source with a passively Q-switch microchip laser and a multipass amplifier.
Optics Letts. 24: 499–501, 1999.
19. JJ Zayhowski, C Dill III. Diode-pumped passively Q-switched picosecond microchip
lasers. Opt Lett 19: 1427, 1994.
20. JJ Zayhowski. Nonlinear frequency conversion with passively Q-switched microchip
lasers. CLEO 96 Tech Dig 9: CWA6, 236, 1996.
21. S Zhou, KK Lee, YC Chen, S Li. Monolithic self-Q-switched Cr,Nd : YAG laser.
Opt Lett 18: 511, 1993.
22. L Fulbert, J Marty, B Ferrand, E Molva. Passively Q-switched monolithic microchip
laser. CLEO 95 Tech Dig 15: CWC5, 176, 1995.
23. JJ Degnan. Optimization of passively Q-switched lasers. IEEE J Quant Electron 31:
1890, 1995.
24. C Larat, M Schwarz, JP Pocholle, G Feugnet, MR Papuchon. High repetition rate
solid-state laser for space communications. Proc SPIE 2381: 256, 1995.
25. M Tsunekane, N Taguchi, H Inaba. High power operation of diode-end pumped
Nd: YVO4 laser using composite rod with undoped end. Electron Letts 1 (32): 40,
1996.
26. U Bäder, JP Meyn, J Bartschke, T Weber, A Borsutszy, R Wallenstein, RG Batchko,
MM Fejer, RL Byer. Visible nanosecond PPLN optical parametric generator pumped
by a passively Q-switched single frequency Nd: YAG laser. CLEO 99 Tech Dig
CThB4, 1999.

Copyright © 2002 by Taylor & Francis

13
Atmospheric Ultraviolet
Spectroscopy

Kelly Chance and Laurence S. Rothman

Harvard–Smithsonian Center for Astrophysics, Cambridge,
Massachusetts

I. INTRODUCTION

Ultraviolet (UV) spectroscopy is a rich source of information about the Earth’s

atmosphere. In this chapter we will survey the field of ultraviolet remote sensing
measurements of the Earth’s atmosphere as currently practiced and as planned
for the next decade, with emphasis on satellite-based measurements. We will
review the underlying physics, particularly as it differs from that contributing to
atmospheric measurements at longer wavelengths, and describe the techniques
of data analysis used to process raw spectra into atmospheric quantities. We will
describe progress in constructing and maintaining the extension of the HITRAN
database into the ultraviolet, to serve as the standard repository for absorption
cross sections for the analysis of atmospheric spectra. Finally, we will review
the heritage of satellite UV spectrometers for atmospheric measurements and the
current and planned instruments.
Ground-based (and balloon-based and aircraft-based) measurements of the
atmosphere using ultraviolet spectroscopy have a rich history and have, naturally,
provided much of the heritage for the satellite measurements discussed here, as
well as substantial validation of the satellite measurements. They are not specifi-
cally discussed in this chapter, but much of the discussion of underlying physics,
measurement techniques, and database consideration applies equally to them.

Copyright © 2002 by Taylor & Francis

II. PHYSICS CONSIDERATIONS

The physical characteristics of the observing environment (temperature, pressure,

interfering absorption, and scattering properties) are particularly important to the
viability of UV atmospheric measurements.
The average temperature at the Earth’s surface (the bottom of the ‘‘tropo-
sphere’’) is about 285 K, at a pressure of 1 atm. At increasing altitude the pressure
and temperature fall, with the pressure decreasing at an exponential scale height
of about 8 km and the temperature decreasing at roughly the adiabatic lapse rate,
that of cooling by adiabatic thermodynamic expansion (the dry adiabatic lapse
rate is ⬃10 K km ⫺1), until one reaches the ‘‘tropopause,’’ which is the boundary
between the troposphere and stratosphere. The tropopause is at ⬃15 km, 0.1 atm,
and 215 K (although this is highly variable with latitude, season, and dynamical
conditions). Above the tropopause the pressure continues to fall, but the tempera-
ture rises, due to photochemical heating induced by the absorption of UV radia-
tion by O 2 in the Herzberg continuum, producing oxygen atoms, the associated
photochemistry of ‘‘odd oxygen’’ (oxygen atoms and ozone), and the further
absorption of UV radiation by ozone. In very simplified terms, the oxygen photo-
chemistry (the ‘‘Chapman’’ chemistry [Chapman, 1930; Brasseur and Solomon,
1986]) can be summarized as:

O 2 ⫹ hν(λ ⱕ 242nm) → 2O (1)

O ⫹ O ⫹ M → O2 ⫹ M (2)
O ⫹ O2 ⫹ M → O3 ⫹ M (3)
O ⫹ O 3 → 2O 2 (4)
O 3 ⫹ hν(λ ⱕ 310nm) → O 2 ⫹ O (5)

where M is any third body (usually an N 2 or O 2 molecule) needed in order to

provide momentum conservation for binary recombination. Most of the strato-
spheric heating is from absorption of O 3 in the Hartley band, from 242 to 310
nm, which also is responsible for much of the atmospheric opacity to UV radia-
tion. Some additional contribution comes from absorption, with weak dissocia-
tion, in the O 3 Huggins bands, from 310 to 400 nm. (The long-wavelength limit
is arbitrarily selected by intensity; there is no physical band cutoff.) The depar-
tures from this simplified scheme, particularly those affected by anthropogenic
actions, occupy much of the effort of stratospheric ozone chemistry studies. For
present purposes we are concerned with the physical background conditions for
measurements, and these are well described by Eqs. (1)–(5). The relative heating
reaches a maximum at the ‘‘stratopause’’ (⬃50 km, 0.0008 atm, 270 K). Above

Copyright © 2002 by Taylor & Francis

this, in the ‘‘mesosphere’’ (which reaches an altitude of 80 km) polyatomic con-
stituents, including ozone, become increasingly rarer, and the atmosphere again
cools.
Atmospheric UV measurements are dominated by absorptions in the strato-
sphere, with significant contributions from the troposphere and, to a lesser extent,
the mesosphere. Conditions of local thermodynamic equilibrium (LTE), where
populations are adequately described by a single electronic, vibrational, rota-
tional, and translational temperature, normally apply to the physical processes
encountered in UV atmospheric spectroscopy. Non-LTE conditions are occasion-
ally encountered, particularly at higher altitudes, in the mesosphere and above,
but they will not be addressed in this chapter, where we concentrate on the UV
spectroscopy applicable to the troposphere and stratosphere.
Molecular absorption in the ultraviolet wavelength range is thus dominated
by absorption of ozone. Much of the additional contribution to spectral absorption
is due to small, often chemically reactive species. Many of the more complex
species that enrich the infrared atmospheric spectrum lack electronic transitions
in the ultraviolet, or have transitions that are too weak to be (as yet) useful diag-
nostics of the atmospheric state (NO 2 is a notable exception). The strong presence
of absorptions due to reactive species is in part the result of the physics that
dictates that, for example, small oxygen-containing free radicals have transitions
in the portion of the UV atmospheric spectrum readily accessible to measurement.
It is also due to the highly oxidized, photochemically driven state of the upper
portions of the atmosphere.
Rayleigh scattering is the other major contributor to the UV atmospheric
spectrum. Rayleigh scattering (also known as ‘‘molecular scattering’’) is dipole
scattering in which electric dipole moments are induced in the air molecules
(primarily N 2 and O 2) by the interaction of their polarizabilities with the electric
field of the radiation. Rayleigh scattering has been thoroughly reviewed by Bates
[1984]. The well-known approximate λ ⫺4 intensity of the Rayleigh scattering
cross section is described more exactly by the relationship

3.9993 ⫻ 10 ⫺4 σ 4
Q R ⫻ 10 24 (cm 2) ⫽ (6)
1 ⫺ 1.069 ⫻ 10 ⫺2 σ 2 ⫺ 6.681 ⫻ 10 ⫺5 σ 4

where σ (µm ⫺1) ⫽ 1/λ (µm), as derived by Chance and Spurr [1997] from an
analysis of the data of Bates [1984]. This expression is valid from 0.2–1 µm
(200–1000 nm). In addition to being a significant competitor to absorption in
the modulation of transmitted UV solar radiation, Rayleigh scattering is the major
source of light for measurements in limb scattering in the UV and visible (to
about 1000 nm) and for backscattered light measurements below about 310 nm.
It provides the backscatter signal for the SBUV measurements of ozone profiles

Copyright © 2002 by Taylor & Francis

in the stratosphere by differential penetration into the O 3 Hartley band versus
wavelength (see sections on BUV, SBUV, and SSBUV instruments, below).
Atmospheric UV spectroscopy is thus largely the measurement of O 3 plus
the measurement of small absorptions in the presence of O 3 absorption and Ray-
leigh scattering. A further complicating effect must be considered, particularly
for small absorptions (absorption optical depths as small as several ⫻10 ⫺4 may
provide useful measurements, as discussed in the section on GOME, below):
Since N 2 and O 2 are not spherically symmetrical, their molecular polarizabilities
are anisotropic. The polarizability anisotropies introduce an inelastic component
to the Rayleigh scattering, where a portion of the Rayleigh-scattered light (3.7%
of Rayleigh scattering at 350 nm, for example) is rotationally Raman-scattered
(vibrational Raman scattering also contributes, but to a much lesser extent). This
effect was first described by Grainger and Ring [1962], who observed the filling
in (broadening and reduction in depth) of solar Fraunhofer lines when they were
viewed from the ground in scattered sunlight. The effect was firmly established
as being primarily due to rotational Raman scattering by Kattawar et al. [1981]
and has since been fully quantified by Chance and Spurr [1997] using the data
provided by Bates [1984].
Since the Ring effect spreads out several percent of the Fraunhofer spec-
trum in wavelength, by convolving it with the rotational Raman spectra of N 2
and O 2 , it constitutes a major interference source for the analysis of spectra: a
‘‘non-Beer’s law’’ contribution. It can usually be included in spectrum fitting as
a small additional pseudoabsorption source, where methods of determining the
correction include measuring the polarization of scattered sunlight [Solomon et
al., 1987] and modeling the effect directly from molecular scattering processes
[Fish and Jones, 1995; Joiner et al., 1995; Chance and Spurr, 1997].
Contributors to the UV atmospheric spectrum include transitions with dis-
crete line spectra as well as broader spectra exhibiting significant dissociation
broadening. The cataloging of spectra is generally organized into line parameters
(i.e., for line-by-line description of spectra) and absorption cross sections, as
detailed in the review of HITRAN database, below. For the range of temperatures
and pressures encountered in atmospheric UV spectroscopy, pressure broadening
of discrete lines is usually negligible in comparison to Doppler broadening. The
narrowest discrete lines will have widths (full-width at 1/e intensity) of less than
0.001 nm, setting a natural limit to the resolution for atmospheric spectrometers
that is rarely approached in practice. A more practical limit, as deduced from
GOME studies for a variety of molecules, is that spectral resolution capable of
fully resolving the highly temperature-dependent and weakly predissociating O 3
Huggins bands (0.2–0.3 nm) is sufficient for practically all detailed spectroscopic
measurements of the atmosphere in the UV. (In contrast, higher resolution in
the visible and near infrared is capable of producing extra information on cloud

Copyright © 2002 by Taylor & Francis

properties [Kuze and Chance, 1994] and on the vertical distribution of H 2 O
[Maurellis et al., 2000].)

A. Measurement Geometries
1. Backscatter Measurements
These measurements are performed by looking from the satellite in the direction
of the Earth’s surface. The surface itself is often masked: by ozone absorption
in the wavelength range of the Hartley band, by Rayleigh scattering that limits
penetration, or by clouds, so that spectral measurements sample volumes repre-
sentative of the wavelength-dependent penetration. The BUV method, described
below, uses wavelength-dependent penetration into the atmosphere in the ozone
Hartley band to produce profiles of stratospheric ozone. This method may be
combined with measurements in the temperature-dependent Huggins bands at
sufficient spectral resolution to resolve the vibrational hot-band structure, taking
advantage of temperature decrease with altitude in the troposphere, to produce
altitude profiles down to the ground (or cloud-top scattering layer) [Chance et
al., 1991, 1997; Munro et al., 1992, 1998]. In general, because of the strength
of Rayleigh scattering, UV backscatter measurements are more sensitive to strato-
spheric gases than tropospheric gases on a per-molecule basis. Backscatter mea-
surements are useful for a number of gas total-column measurements in the UV,
including BrO, OClO, NO 2 , SO 2 , and HCHO.
Spectral Analysis. The GOME satellite has fulfilled initial expectations
of being able to measure absorption optical depths as small as several ⫻10 ⫺4
(31/2 orders of magnitude below the full-scale backscattered radiance). Several
methods have been used for analyzing the spectra obtained from GOME backscat-
ter measurements and proposed for use in later instruments. These generally fall
into two categories: those in which line-of-sight column abundances (slant col-
umns) are first fitted, and then adjusted to give vertical column abundances; and
those in which vertical abundances, and sometimes altitude profile information,
are fitted directly. For occultation and limb-scattering measurements, series of
spectra measured at different elevation angles must be analyzed together to give
vertical profile information.
The most fundamental method of spectral analysis for slant columns con-
sists of synthesis of the satellite radiance measurements, beginning with a
satellite-measured irradiance, and adjusting the parameters describing the synthe-
sis by iteration until they agree with the measurements in the least squares sense.
This is the long-standing technique used in generations of ground-based, airborne,
and satellite (e.g., infrared) measurements. Parameters for the analysis of GOME
spectra are typical of what will be needed for other planned satellite instruments:

Copyright © 2002 by Taylor & Francis

absorption cross sections, Ring effect correction, albedo, wavelength calibration
correction, a correction for undersampling of the atmospheric spectrum by the
GOME instrument (radiance and irradiance measurements are taken at slightly
different Doppler shifts; resampling to construct a synthetic radiance spectrum
introduces aliasing since the GOME instrument does not fully Nyquist sample
the spectrum [Goldman, 1953; Chance, 1998; Slijkhuis et al., 1999]), and low-
frequency closure terms to account for albedo variation and Mie scattering.
Differential optical absorption spectroscopy (DOAS) is a commonly used
variant of the procedure for determination of slant column abundances [Platt,
1994]. In the DOAS technique, the radiance is divided by the irradiance and the
logarithm taken, to linearize Beer’s law contributions. The data are then high-
pass filtered to give resulting variations about zero. The result is fitted using
linear least-squares analysis to a sum of high-pass filtered cross sections times
abundances, a high-pass-filtered version of the Ring effect spectrum divided by
the Fraunhofer spectrum, and low-frequency closure terms. If wavelength adjust-
ment is needed, this is accomplished by alternating the linear fitting steps with
nonlinear steps in wavelength scale adjustments.
Although DOAS was principally designed for situations where the irradi-
ance is unstable (i.e., a varying lamp source) or cannot be precisely measured
(as in ground-based atmospheric measurements), it has proved quite useful in
the analysis of GOME satellite data, and is used in the current operational pro-
cessing for O 3 and NO 2 column abundance measurements [Loyola et al., 1997].
However, because it involves discarding part of the spectral information and in-
troducing an approximation in the treatment of the Ring effect, it is inherently
less accurate than the basic fitting procedure.
For either method of slant column determination, it is necessary to divide
the measured slant column abundance by an air mass factor (AMF), given by

slant column τ
AMF ⫽ ⫽ s (7)
vertical column τ

where τs and τ are the optical depths for the slant and vertical columns. In practice,
AMFs are determined from radiative transfer calculation with assumption about
the shape of the species’ vertical distribution usually obtained from modeling.
In UV measurements, it is chiefly Rayleigh scattering that affects the atmospheric
sampling so that determinations of AMFs are not simply geometric calculations
(as noted earlier, UV backscatter measurements are usually more sensitive to
stratospheric gases than tropospheric gases on a per-molecule basis).
GOME has successfully demonstrated the technique of determining ozone
profiles and tropospheric ozone abundances by combining the backscatter infor-
mation on the stratospheric profile [Singer and Wentworth, 1957] with spectrally
resolved measurements in the temperature-dependent Huggins bands. Because of

Copyright © 2002 by Taylor & Francis

the altitude-dependent temperature structure of the atmosphere, the Huggins
bands provide spectral fingerprints for the tropospheric ozone distribution
[Chance et al., 1991, 1997; Munro et al., 1992, 1998].
Optical absorption coefficient spectroscopy (OACS) is a novel technique
for applying the opacity coefficient method (OCM) to fit GOME spectra in re-
gions where the absorptions are discrete and highly structured [Maurellis et al.,
2000]. OCM is a refinement over exponential sum-fitting techniques [Kato et al.,
1999] for sampling average transmittance over detector pixels. OACS is success-
ful in determining H 2 O column abundances from GOME spectra in the visible,
and is a promising technique for determining height-resolved (profile) informa-
tion from backscatter measurements of highly structured spectra of H 2 O and other
species.
Cloud Parameter Determination. Information on cloud parameters pres-
ent in backscatter data includes absolute radiances levels and their spatial varia-
tion; absorption depth in the O 2 A band and the O 2 –O 2 collision-induced absorp-
tion; and the depth of the Ring effect, particularly for the Ca II H and K lines
near 395 nm. The GOME operational data processor employs an initial cloud-
fitting algorithm (ICFA) based on cloud-top height climatology from the ISCCP
[Rossow et al., 1996] and absorption in the visible O 2 A band [Kuze and Chance,
1994; Loyola et al., 1997]. This method is being upgraded to the cloud retrieval
algorithm for GOME (CRAG), which employs spatial variations of the albedo
from the GOME sensor in combination with O 2 A band absorption depth and
absolute radiance level to determine cloud height, fractional coverage, and optical
depth simultaneously [Kurosu et al., 1997, 1998, 1999]. Note that the ODUS
instrument will be flown simultaneously with an O 2 A band camera for cloud
coverage measurements on the GCOM-1A satellite. Joiner and Bhartia [1995]
have proposed the use of the Ring effect for cloud-top pressure measurements,
and tested the method on Nimbus 7 TOMS and SBUV data. They indicate that
for an instrument with 0.2 nm spectrum resolution (e.g., GOME) this method
should be capable of cloud-top pressure determination to 30 mbar precision for
fully cloudy scenes. The use of absorption depth due to the O 2 –O 2 collision com-
plex as an indicator of cloud coverage has also been proposed as an indicator of
cloud coverage and used with some success with GOME data [Chance et al.,
1991; Wagner and Platt, 1998].

2. Occultation Measurements
Solar occultation is the most common of this type of measurement, although lunar
occultations are planned (for example, by SAGE III) to permit measurements of
species, including NO 3 and OClO, that are easily photolyzed and are observable
principally at night. Occultations have the inherent advantages of high vertical
resolution and high signal due to the direct beam source, and simplified restitution

Copyright © 2002 by Taylor & Francis

of measurements to location on the limb in comparison to limb-scattering and
limb-emission measurements. (Limb pointing is, in general, quite difficult; the
availability of ephemeris data for occultation measurements greatly simplifies the
geolocation, although it may still be complicated by refraction.) A disadvantage
of occultation measurements is that occultation opportunities occur infrequently
(e.g. one sunrise and one sunset per orbit) and at locations that are not freely
selectable. The use of stellar occultations, to be employed by GOMOS, over-
comes this by making use of a number of stars of sufficient magnitude. An addi-
tional advantage of stellar occultations is that many of the important atmospheric
species have concentrations with strong diurnal variation. Thus a sunrise occulta-
tion is not equivalent to a sunset occultation and the different altitudes encoun-
tered during a solar occultation are observed at different solar times, requiring
photochemical modeling to correct the data [Roscoe and Pyle, 1987]. Solar oc-
cultation measurements have by far the greatest heritage in satellite as well as
balloon-borne measurements. In general, they are capable of measuring to below
the tropopause, to altitudes determined by cloud and aerosol interference. In the
longer-wave UV and into the visible, they are capable of measuring to an altitude
as low as 5 km a significant fraction of the time.
The spectral analysis of occultation spectra will normally be done either
by fitting to obtain line-of-sight column abundances and then determining profile
information sequentially, using spectra at successively lower measurement tan-
gent heights (the ‘‘onion-peeling’’ method) or by simultaneous fitting of a set of
occultation spectra to a selected altitude grid, with appropriate constraints on
interpolation (the ‘‘global fitting’’ method [Carlotti, 1988]).

3. Limb-Scatter Measurements
Measurements in limb-scattered radiation use Rayleigh scattering as the light
source. They trade the advantages of occultation measurements for the ability to
measure at any geolocation (except for the direct illumination of an occultation
source). Because of the weaker line-of-sight signal, limb-scatter measurements
suffer from more interference by multiply scattered light arising from lower in the
atmosphere than the tangent height of observation, requiring more sophisticated
retrieval algorithms. This interference and the weaker source may make limb-
scatter measurements more difficult at lower altitudes in the troposphere than
for occultations. Altitude registration is more difficult than for occultations, but
significant progress has been made in using the intensity of Rayleigh scattering
outside of molecular absorption for this purpose, for example by the Rayleigh
scattering attitude sensor (RSAS) [Janz et al., 1996], which flew on Space Shuttle
flight STS-72 in 1996. Limb-scatter measurements have substantial balloon-borne
heritage [McElroy, 1988]. The technique is planned for SCIAMACHY, SAGE
III, OSIRIS, and OMPS.

Copyright © 2002 by Taylor & Francis

 Spectral analysis is similar to that required for occultation measurements,
except that radiative transfer modeling is generally more important to correct for
multiply scattered radiation from below the tangent height, and limb registration
is more difficult due to the lack of ephemeris data.

B. HITRAN Database
The HITRAN spectroscopic database was created as a compilation of spectro-
scopic parameters that could serve to provide the basic information to computer
programs modeling atmospheric attenuation due to molecular gaseous absorption
in the infrared. The first edition [McClatchey et al., 1973] was made public in
1973. The work had evolved from several monographs produced in the 1960s at
the National Bureau of Standards [Gates et al., 1964; Calfee and Benedict, 1966;
Benedict and Calfee, 1967].
HITRAN is the international standard molecular spectroscopic database
and has developed over time with respect to several key aspects:

The spectral range now spans millimeter through ultraviolet transitions (the
database employs the units of wavenumber in cm ⫺1 throughout, which
is proportional to energy and very convenient to most modeling codes).
Most of the transitions included in HITRAN are due to the electric dipole
moment of the molecules, although there are also magnetic dipole and
electric quadrupole transitions (which, although weak, are of interest in
the case of very abundant species such as O 2 and N 2).
The HITRAN compilation includes absorption cross sections in addition
to the traditional line-by-line spectroscopic parameters. The cross sec-
tions are used where the density of spectral features is very crowded (as
is the case for ‘‘heavy’’ molecular species in the infrared such as the
chlorofluorocarbons and especially for most bands in the UV). The
cross sections are given at different pressure–temperature pairs, which
facilitates modeling of transmission or radiance in nonisothermal paths.
The file structure and format for the cross section files are summarized
in Table 1.
The number of gases covered has expanded from the original seven that
were the principal atmospheric absorbers in the infrared to about 60 dif-
ferent molecules along with their isotopomers when significant to terres-
trial atmospheric remote-sensing problems.
The quantity of parameters for individual transitions in the line-by-line por-
tion of the HITRAN compilation has substantially increased. The origi-
nal database contained information for the line position (transition fre-
quency in vacuum wavenumbers), the intensity of the transition, the
Lorentzian halfwidth (broadening of the spectral line due to collisions

Copyright © 2002 by Taylor & Francis

Table 1 File Structure and Format (in FORTRAN) for Cross-Section Files

ν1, init ν1, final # of Reference

Molecule cm ⫺1 cm ⫺1 points T1 K P 1 torr σ max cm 2 Resolution, etc.
A10 F10.4 F10.4 I10 F10.4 F10.4 E10.3 3A10

Cross sections (10 per line) in cm 2 /molecule T 1, P1: 10E10.3

ν 1, init ν 1, final # of Reference

Molecule cm⫺1 cm ⫺1 points T2 K P 2 torr σ max cm 2 Resolution, etc.
A10 F10.4 F10.4 I10 F10.4 F10.4 E10.3 3A10

Cross sections (10 per line) in cm 2 /molecule T 2 , P 2: 10E10.3

ν 2, init ν 2, final # of Reference

Molecule cm ⫺1 cm ⫺1 points T1 K P 1 torr σ max cm 2 Resolution, etc.
A10 F10.4 F10.4 I10 F10.4 F10.4 E10.3 3A10

Cross sections (10 per line) in cm 2 /molecule T 1 , P 1: 10E10.3

ν 2, init ν 2, final # of Reference

Molecule cm ⫺1 cm ⫺1 points T2 K P 2 torr σ max cm 2 Resolution, etc.
A10 F10.4 F10.4 I10 F10.4 F10.4 E10.3 3A10

Cross sections (10 per line) in cm 2 /molecule T 2 , P 2: 10E10.3

of the molecule with air), the lower-state energy of the transition, and
the unique quantum number identification of each transition. These pa-
rameters were considered minimal for the calculation of molecular atten-
uation via the Lambert–Beers law. With the increased sophistication of
high-resolution instrumentation, and the advances in retrieval algorithms,
more fundamental parameters have been added, including transition
probabilities, self-broadened halfwidths, temperature dependence coef-
ficients for the halfwidth, and pressure–shift coefficients.
The media and software for HITRAN have also evolved. Originally re-
leased on magnetic tape, it is currently available on CD-ROM and more
and more data are becoming accessible on the Internet. The CD-ROM
has certainly emphasized the archival nature of the database. HITRAN
is not a program (although the MODTRAN band-model code has been
supplied with the 1996 edition of HITRAN). However, software has been
included that allows various database functions, such as filtering, sorting,
plotting the data, and access to relevant references.

Copyright © 2002 by Taylor & Francis

Current information concerning HITRAN, as well as links to related spectro-
scopic databases, can be found at the HITRAN web-site (http:/ /CfA—
www.Harvard.edu/HITRAN).

1. UV Data in HITRAN
The edition of the HITRAN compilation of 1996 [Rothman et al., 1998] extended
into the ultraviolet region for the first time with the inclusion of UV cross sections
for N 2 O at 296 K in the wavelength region 170–222 nm, and SO 2 at 213 K in
the wavelength region 172–240 nm. The cross-section data employ the identical
format as adopted for the IR cross sections, namely a table of cross sections at
equal wavenumber interval. These sets of cross-sections are introduced with a
standardized header providing essential information about each pressure–temper-
ature set (see Table 1). High-resolution data for the O 2 Schumann–Runge system
were also added to the compilation with line-transition parameters similar in for-
mat to the main body of HITRAN. The UV data are kept in a separate directory
in the HITRAN compilation and further separated by two subdirectories, one for
the cross-section data and the other for line parameter data.
For the upcoming edition of the database, HITRAN2001, we anticipate
adding high-resolution cross sections for several molecular band systems: O 3 ,
OClO, BrO, H 2 CO, ClO, and NO 2 . In addition, there will be line-by-line data
for O 2 , NO, and OH. The spectra of these gases have applications to a number
of programs, including GOME and the NASA Earth observing system (EOS)
program [King et al., 1995]. While the choices of cross sections used in analysis
of field measurements are currently evolving, HITRAN2001 includes spectra that
have been demonstrated to provide reasonable fitting for GOME spectra. Tables
2 and 3 summarize the cross-section and line-by-line UV data of HITRAN2001.
UV Line-by-Line Parameters. The HITRAN line list for the O 2 Schu-
mann-Runge system contains 11,020 lines covering a spectral range of
44,606.676 to 57,027.590 cm ⫺1 (224–175 nm). This system represents transitions
between the B 3 ⌺ u⫺ electronic state and the X 3 ⌺ g⫺ ground electronic state. Nearly
all of the line parameters are based on high-resolution absorption measurements
from two groups: one from the Australian National University [Lewis et al., 1986]
and the other from the Harvard–Smithsonian Center for Astrophysics [Yoshino
et al., 1983; Yoshino et al., 1992]. A description of this line compilation is given
in Minschwaner et al. [1992], which also contains relevant citations for measured
quantities. The file in HITRAN lists all principal branch triplets over the range
ν′ ⫽ 0–19, ν″ ⫽ 0–2, N″ ⫽ 1–51. Satellite branches are included for ν′ ⫽
0–19, ν″ ⫽ 0, N″ ⫽ 1–15. Principal branches for the 16 O 18 O isotopomer are
listed for ν′ ⫽ 2–16, ν″ ⫽ 0, N″ ⫽ 1–24. (A single prime (′) designates the
upper state of a transition, and double primes (″) denote the lower state.)

Copyright © 2002 by Taylor & Francis

Table 2 UV Cross Sections in HITRAN

Spectral range Temperatures

Molecule Band system (cm ⫺1) (K)

N 2O (1 A″ 1 ⌺ ⫺, 1 ∆)? ← X̃ 1 ⌺ ⫺ 44925–58956 296

SO 2 C̃ 1 B 2 ← X̃ 1 A 1 41691–58452 213
45455–50505 295
O3 1
B 2 ← X̃ 1 A 1 (Hartley-Huggins) 24570–33311 203–293
1
B 1 ← X̃ 1 A 1 (Chappuis)
(Wulf )
HCHO 1
A″ ← X̃ 1 A 1 27391–33311 223, 293
BrO A 2Π3/2 ← X 2Π3/2 25756–32013 228, 298
OClO Ã 2 A 2 ← X̃ 2B1 20992–41228 228
ClO A 2Π3/2 ← X 2Π3/2 32000–37700 220
NO 2 A 2 B 1 ← X̃ 2 A 1 17540–32260 213, 298

There are two caveats for the user associated with these data. First, line
positions of high N″ are not reliable. Second, there is an error in the indexing
of the lower ‘‘global’’ quantum state, in that the B3 ⌺ u⫺ electronic state was re-
ferred to rather than the X 3 ⌺ g⫺.
Line positions were calculated from energy level differences based on mea-
sured molecular constants. Line intensities were obtained from measured band
oscillator strengths normalized according to Hönl-London factors, assuming a
Boltzmann distribution of energies at 296 K. The tabulated widths are measured
predissociation widths at zero pressure. Pressure-broadened widths are not listed
because this effect is comparable to predissociation, broadening only for pres-
sures larger than 1 atm. (The self-broadening coefficient is on the order of 0.20
cm ⫺1 /atm [Lewis et al., 1988].) Use of a Voigt line shape, composed of a thermal-
broadened Doppler profile and a predissociation-broadened Lorentz profile, is

Table 3 UV Line-by-Line Data in HITRAN

Spectral range
Molecule Band system (cm ⫺1)

O2 B 1⌺ u⫺ ← X 3⌺ g⫺ (Schumann-Runge) 44604–57028
A 3⌺ u⫹ ← X 3⌺ g⫺ (Herzberg I) 37000–41600
c 1⌺ u⫺ ← X 3⌺ g⫺ (Herzberg II) 36400–41600
A′ 3∆ u ← X 3⌺ g⫺ (Herzberg III) 36400–41600
OH A 2 ⌺ ⫹ ← X 2Π i 31000–36000
NO A 2 ⌺ ⫹, B 2 Π r, C 2 Π, D 2 ⌺ ⫹ ← X 2 Π i (γ, β, δ, ⑀ bands) 44300–62500

Copyright © 2002 by Taylor & Francis

adequate for most atmospheric applications. Temperature-dependent parameters
listed in the line-by-line portion of the HITRAN compilation are given at a stan-
dard reference temperature of 296 K. The appropriate scaling of these quantities
is given in the Appendix of Rothman et al. [1998]. Line intensities at temperatures
other than 296 K can be obtained using the energy of the lower quantum state
in conjunction with the temperature dependence of the total internal partition
sum. The latter has been calculated over the range 70–400 K, and is reproduced
to within 1% from the following expressions [Gamache et al., 1990]:
16
O 16 O: Q(T) ⫽ 0.604 ⫹ 0.731 T ⫺ 4.00 ⫻ 10 ⫺5 T 2 ⫹ 8.71 ⫻ 10 ⫺8 T 3
O O: Q(T) ⫽ ⫺1.04 ⫹ 1.51 T ⫺ 3.44 ⫻ 10 ⫺4 T 2 ⫺ 1.09 ⫻ 10 ⫺6 T 3
16 16

The summation for 16 O 16 O covers sufficient range of vibrational and rotational

energy levels to ensure accuracy over the temperature range 70–400 K. For
16 18
O O, however, a small error near 400 K may be expected as a result of neglect-
ing the higher vibrational and rotational states.
UV Cross Sections. Cross sections for two species, nitrous oxide and sul-
fur dioxide, had been placed in the 1996 edition of HITRAN. These cross sections
have been cast into the same format as the IR cross-section data described in
Rothman et al. [1998]; transforming the UV data into the equal-interval wave-
number scale required interpolation from measured wavelength scales employed
in the UV.
High-resolution cross-section measurements of N 2 O at 295–299 K have
been performed in the wavelength region 170–222 nm with a 6.65 m scanning
spectrometer of sufficient resolution to yield cross sections independent of the
instrumental function [Yoshino et al., 1984]. The measured cross sections are
available throughout the region 44,925–58,955 cm ⫺1 at intervals of 0.1–0.2 cm ⫺1.
Previously unresolved details of the banded structure, which is superimposed on
the continuous absorption in the region 174–190 nm, are observed.
Laboratory measurements at high resolution of the absorption cross section
of SO 2 at the temperature 213 K have been performed in the wavelength region
172–240 nm with a 6.65 m scanning spectrometer operated at an instrumental
width of 0.002 nm [Freeman et al., 1984]. The measured cross sections are avail-
able throughout the region 172–240 nm at wavenumber intervals of 0.4–0.1
cm ⫺1. The measured cross sections, which are relevant to the photochemistry of
planetary atmospheres, possess significantly more spectroscopic structure, and
are more accurate than previous measurements made at lower resolution. How-
ever, values at peak cross sections could be affected by the instrumental widths
even at 0.002 nm, which is larger than the Doppler widths.
Since these cross-section data in HITRAN are only given at a single temper-
ature (296K in the case of N 2 O and 213K in the case of SO 2), they are insufficient
for simulation of optical paths that are not isothermal.

Copyright © 2002 by Taylor & Francis

III. SATELLITE SPECTROMETERS
A. Heritage
A number of instruments have provided valuable atmospheric measurements us-
ing discrete photometer bands, while demonstrating the utility and providing the
heritage for the more comprehensive spectroscopic measurements of the atmo-
sphere that are now in progress and planned for the future. Most of the previous
measurements are, not surprisingly, of ozone (O 3), with nitrogen dioxide (NO 2)
accounting for most of the remaining gas measurements. A comprehensive review
of satellite ozone measurements is given by Krueger et al. [1990]. Several excel-
lent review articles on the instruments mentioned here and their measurements
have been published in the special issue of Planetary and Space Science, particu-
larly by Miller [1989], Krueger [1989], and McCormick et al. [1989]. Of particu-
lar note among photometer instruments are the following.

1. BUV, SBUV, and SSBUV Instruments

This series of instruments, which began with the Nimbus 4 BUV (1970 launch)
measures solar radiation backscattered in the nadir viewing of the Earth’s atmo-
sphere at a series of discrete wavelengths from 255 to 340 nm, with 1 nm band-
pass (resolution). These are actually scanning monochromator (i.e., spectrometer)
instruments, but the taking of data from specific bands, and subsequent analysis,
mimics the function of photometer measurements. In addition to total ozone (de-
termined by discrete band measurements between 312 and 340 nm in the highly
structured Huggins bands), they measure stratospheric ozone profiles, following
on the technique first proposed by Singer and Wentworth [1957]. This technique
determines profile information for stratospheric ozone from the differential pene-
tration depth of solar radiation into the strong ultraviolet Hartley band (242–310
nm). (See Heath et al. [1973] for further information on the Nimbus 4 BUV
instrument, Fleig et al. [1990] for details of the Nimbus 7 SBUV instrument, and
Hilsenrath et al. [1995] for details of the SSBUV instrument.)

2. Total Ozone Mapping Spectrometer Instruments

The TOMS series of instruments, which began with the Nimbus 7 TOMS (1978
launch, Heath et al. [1975]) measures total ozone using six 1 nm bandpass wave-
length bands from 312 to 380 nm (308–360 nm in later versions). Ozone is deter-
mined from differences in absorption in the Huggins bands.

3. Stratospheric Gas and Aerosol Experiment Instrument

This series (SAGE I and II) includes sun photometers that measure solar occulta-
tions in selected bands in the ultraviolet, visible, and infrared [McCormick et al.,

Copyright © 2002 by Taylor & Francis

1989]. SAGE I, launched in 1979 aboard the AEM-2 satellite, measured O 3 , NO 2 ,
and aerosols using bands at 380, 450, 600, and 1000 nm. SAGE II, launched in
1984 aboard the ERBS satellite (and still operating in 2001), measures O 3 , NO 2 ,
H 2 O, and aerosols using bands at 385, 453, 448, 525, 600, 940, and 1020 nm.
A linear CCD array spectrometer version, SAGE III (described below), is to be
launched in the near future.

4. Polar Ozone and Aerosol Measurement

The POAM II instrument is a sun photometer that measures solar occultations
at nine spectral bands between 350 and 1060 nm [Glaccum et al., 1996]. POAM
II was launched in 1993 on board the SPOT-3 satellite, and measures atmospheric
profiles of O 3 , NO 2 , H 2 O, aerosols, and temperature.

B. Current and Planned Instruments

The Global Ozone Monitoring Experiment [GOME; European Space Agency,
1995] is the first space-based spectrometer to utilize a broad range of the ultravio-
let spectrum at continuous wavelength coverage and moderate spectral resolu-
tion for the purpose of measuring the composition of the Earth’s atmosphere.
It was not, however, the first planned, since it is a scaled-back version of the
SCIAMACHY instrument, which was first proposed in 1985 [Burrows and
Chance, 1991; Bovensmann et al., 1999; Noël et al., 1999]. The utility of mea-
surements in a wide range of the ultraviolet, visible, and near-infrared for de-
termining abundances of a large fraction of species that are of importance in
current atmospheric science is now widely recognized. Table 4 gives an overview
of instruments that are planned over the next decade involving this region. Mea-
surements include various viewing geometries, including nadir, limb, solar (and
lunar) occultation, each with its own advantages for focusing on the various spa-
tial and temporal aspects of atmospheric composition and variability. They also
utilize varying spectral and spatial resolutions and wavelength coverages, gener-
ally optimized for various measurement aspects in tradeoffs that compromise
among global coverage, cost, complexity, and availability of satellite and support
resources (e.g., telemetry bandwidth).

1. GOME
The first satellite instrument to measure UV spectra of the Earth’s atmosphere,
GOME was launched in polar, sun-synchronous orbit on the ERS-2 satellite in
April 1995 [European Space Agency, 1995], and is still performing normally as
of September, 2001. GOME obtains about 30,000 Earth radiance spectra each
day, covering the ultraviolet (237–405 nm at 0.2 nm resolution) and the visible
(407–794 nm at 0.4 nm resolution), measured with silicon diode-array detectors.

Copyright © 2002 by Taylor & Francis

Table 4 Selected UV/Visible Satellite Spectrometers

Wavelengths
Instrument (nm) Geometry Gases a

GOME (GOME-2) 240–790 Nadir O 3, NO 2 , BrO, OClO, SO 2 ,

HCHO, H 2O
SCIAMACHY b 240–2400 Nadir, limb, O 3 , NO 2 , BrO, OClO, SO 2 ,
occultation HCHO, H 2 O
OMI 270–500 Nadir O 3 , NO 2 , BrO, OClO, SO 2 ,
HCHO
OMPS 290–1000 Nadir, limb O 3 , NO 2 , BrO, OClO, SO 2 ,
HCHO, H 2 O
SAGE III 280–1040 Occultation, O 3 , NO 2 , BrO, OClO,
limb H 2 O, NO 3
GOMOS 250–952 Stellar O 3 , NO 2 , NO 3 , OClO
occultation
Odin/OSIRIS 280–800 Limb O 3 , NO 2 , BrO, OClO, SO 2 ,
HCHO, H 2 O
ACE/MAESTRO 285–1030 Occultation, O 3 , NO 2 , BrO, OClO, SO 2 ,
nadir HCHO, H 2 O
ODUS 306–420 Nadir O 3 , NO 2 , BrO, OClO, SO 2 ,
HCHO
a
H 2 O is measured in the visible and infrared.
b
Additional species are measured in the infrared.

GOME also employs a polarization measurement device (PMD) that measures

two polarizations of backscattered light at low spectral resolution, but at 16⫻
higher spatial resolution than the spectral channels. The PMD is primarily for
correction of instrument polarization sensitivity, but it also serves as a higher
resolution sensor for albedo variations. GOME observes the entire Earth with 3
day coverage at the equator. Current operational data products available for
GOME include total column abundances of O 3 and NO 2 , and cloud fractional
coverage. It was demonstrated that GOME should be able to measure a large
set of atmospheric gases, including ozone profiles and tropospheric ozone, and
atmospheric columns of NO 2 , H 2 O, BrO, OClO, SO 2 (e.g., from volcanic erup-
tions), and HCHO, to scientifically useful sensitivity [Chance et al., 1991]. This
has now been shown to be the case, and numerous publications on these mole-
cules and the geophysical processes they illuminate are being produced. Among
the highlights of GOME results are the first direct measurements of tropospheric
ozone from space [Munro et al., 1998]; the global measurement of BrO, including
tropospheric enhancements in the polar springtime [Richter et al., 1998; Chance,
1998; Wagner and Platt, 1998]; and measurement of tropospheric formaldehyde

Copyright © 2002 by Taylor & Francis

from biogenic sources including biomass burning and isoprene production by
forests [Thomas et al., 1998; Chance et al., 2000; Palmer et al., 2001]. Analysis
of GOME data has shown that fitting of spectra to absorption optical depths of
several ⫻10 ⫺4 of the full-scale radiance values is achievable in practice in favor-
able cases, given proper attention to the underlying spectroscopy, the fitting
method, and the instrument characterization. In addition to gaseous components
of the Earth’s atmosphere, GOME measures aerosols using the TOMS-type ab-
sorbing aerosol index [Hsu et al., 1996; Herman et al., 1997; Hsu et al., 1999;
Torres et al., 1999; van Oss, 1999] and cloud parameters, using a combination of
information from the visible O 2 A band, the PMD, and the radiance measurements
[Loyola et al., 1997; Kurosu et al., 1997, 1998, 1999]. GOME instruments (the
‘‘GOME-2’’ series) have been selected as the operational ozone monitors on the
Eumetsat Metop polar platforms beginning in 2003.

2. Scanning Imaging Absorption Spectrometer

for Atmospheric Chartography
The SCIAMACHY is scheduled for launch into polar, sun-synchronous orbit on
the ESA Envisat satellite in 2001. SCIAMACHY is the original concept from
which the GOME instrument was derived, as a smaller instrument with particular
emphasis on atmospheric ozone [Burrows and Chance, 1991; Chance et al., 1991;
Noël et al., 1999; Bovensmann et al., 1999]. SCIAMACHY will observe the
Earth’s atmosphere in nadir viewing, in limb-scattered radiation, and in solar
occultation. The wavelength range 240–1700 nm is measured fully; there are
additional wavelength regions in the infrared at 1940–1040 nm and 2265–2380
nm, principally for the measurement of CO, N 2 O, CH 4 , and CO 2 . The ultraviolet
species measurements from SCIAMACHY will be the same as those for GOME.
Addition of the limb and occultation measurement geometries will permit the
determination of altitude profiles for most species, down to the limit of limb
observation. Sequential limb and nadir viewing may help to determine budgets
of tropospheric components that are difficult to obtain from nadir-only (e.g.,
GOME) measurements because of the overlying stratospheric loading (this is
particularly relevant for O 3 and NO 2).

3. Ozone Monitoring Instrument

The OMI is a contribution of the Netherlands and Finland to the NASA EOS
AURA (formerly CHEM) mission, scheduled for launch in 2003 in polar, sun-
synchronous orbit. OMI measures in the nadir at higher spatial resolution than
GOME (13 ⫻ 24 km 2 nominal instantaneous field of view) and with a wider
orbital swath: OMI provides full global coverage each day. To accommodate the
higher spatial sampling, OMI measures at lower spectral resolution than GOME.
OMI measures 270–500 nm with resolution of 0.45–0.64 nm. OMI will provide

Copyright © 2002 by Taylor & Francis

measurements of O 3 (columns, plus vertical profile information at vertical resolu-
tion between that of SBUV and GOME); column abundances of the gases NO 2 ,
BrO, OClO, SO 2 , and HCHO; cloud height and fractional coverage; and a TOMS-
type absorbing aerosol index.

4. Ozone Mapping and Profiler Suite

The OMPS has been selected as the operational ozone instrument for the US
National Polar Orbiting Environmental Satellite System (NPOESS) satellites.
The first of these sun-synchronous satellites will be launched in 2008. OMPS
replaces the SBUV instruments as the US operational ozone-monitoring instru-
ments. OMPS provides nadir measurements from 250 to 380 nm at 1 nm resolu-
tion and limb scattering measurements, employing a prism-based monochroma-
tor, from 290–1000 nm in 16 total bands, with bandpass resolutions selected
from 1.5–40 nm. (These specifications are subject to minor changes as the de-
tailed design is accomplished.) OMPS UV measurements include ozone total
columns and profiles; the gases NO 2 , BrO, OClO, SO 2 , HCHO; and cloud and
aerosol information. OMPS provides full global coverage each day at spatial
resolutions from 46 ⫻ 50 km 2 (nadir) to 215 ⫻ 50 km 2 (limit of across-track
swath).

5. SAGE III
The SAGE III instruments are planned for three missions, the first scheduled for
launch aboard the Russian METEOR 3-M satellite in 2001. SAGE III is the fourth
generation in this series of instruments, after the Stratospheric Aerosol Measure-
ment II (SAM II), SAGE I, and SAGE II. In addition to the solar occultation
measurements of the past instruments, SAGE III will measure lunar occultations
(primarily for NO 3 and OClO) and make measurements in limb-scattered radia-
tion. The linear array CCD detector provides continuous coverage from 280 to
1030 nm at 1 nm resolution. An additional (discrete photodiode) detector provides
1550 nm measurements to extend aerosol diagnostic capability. SAGE III mea-
sures O 3 , NO 2 , BrO, OClO, H 2 O, NO 3 , and aerosols.

6. Global Ozone Monitoring by Occultation of Stars

The GOMOS instrument will fly on the ESA Envisat satellite in 2001. It is de-
signed to combine the high vertical resolution inherent in occultation measure-
ments (better than 1.7 km in this case) with day- and night-side capability by
performing stellar occultation on a selection of several hundred stars. With the
Envisat orbital coverage, this will provide typically more than 600 profile mea-
surements each day. GOMOS measures from 250 to 952 nm, with 0.7–0.9 nm
resolution in the UV and visible. Additionally, two photometers measure from

Copyright © 2002 by Taylor & Francis

466 to 528 nm and 644 to 705 nm. The GOMOS spectrometer measures profiles
of ozone, NO 2 , NO 3 , OClO, H 2 O, and temperature.

7. Odin/Optical Spectrograph and InfraRed Imaging System

Odin is a Swedish-led joint astronomy/aeronomy mission that includes participa-
tion from Canada, Finland, and France. Odin includes a UV/visible/IR spectrom-
eter, the OSIRIS, and a microwave radiometer. The spectrometer will measure
from 280 to 800 nm at 1 nm resolution, with an additional 10 nm wide radiometer
channel at 1270 nm. It was launched in 2001 into a sun-synchronous orbit. The
UV/visible/IR atmospheric measurements are made by limb scanning in scattered
sunlight. OSIRIS measures O 3 , NO 2 , BrO, OClO, SO 2 , HCHO, and H 2 O.

8. Atmospheric Chemistry Experiment/Measurements

of Aerosol Extinction in the Stratosphere
and Troposphere Retrieved by Occultation
The ACE is a Canadian project to be launched aboard the SCISAT-2 satellite in
2002. It includes an infrared Fourier transform spectrometer, operating in solar
occultation, and the UV/visible/IR MAESTRO instrument. MAESTRO operates
from 285 to 1030 nm with a resolution of 0.6 nm in the UV and 1.0 nm in the
visible/IR. It is primarily an occultation instrument, but will make nadir, back-
scatter measurements subject to available satellite resources. In addition to aero-
sols, MAESTRO will measure O 3 , NO 2 , BrO, OClO, SO 2 , HCHO, and H 2 O.

9. Ozone Dynamics Ultraviolet Spectrometer

The ODUS is scheduled for launch on the Japanese GCOM-A1 satellite in 2005,
in a 70 degree inclination orbit. An additional instrument will fly on the GCOM-
A2 satellite in 2008. ODUS measures backscattered radiation from 306 to 420
nm at 0.5 nm resolution. The nadir field-of-view has 20 km spatial resolution,
degraded off axis in the 120 degree total field-of-view. ODUS will measure O 3 ,
NO 2 , BrO, OClO, SO 2 , and HCHO.

REFERENCES

Bates DR. Rayleigh scattering by air. Planet Space Sci 32: 785–790, 1984.
Benedict WS, RF Calfee. Line Parameters for the 1.9 and 6.3 Micron Water Vapor Bands.
U.S. Dept of Commerce, ESSA paper 2, 1967.
Bovensmann H, JP Burrows, M Buchwitz, J Frerick, S Noël, VV Rozanov, KV Chance,
APH Goed. SCIAMACHY: mission objectives and measurement modes. J Atmos
Sci 56: 127–150, 1999.

Copyright © 2002 by Taylor & Francis

Brasseur G, S Solomon. Aeronomy of the Middle Atmosphere. Dordrecht, The Nether-
lands: D Reidel, 1986.
Burrows JP, KV Chance. Scanning imaging absorption spectrometer for atmospheric char-
tography. Proc SPIE, Future European and Japanese Remote Sensing Sensors and
Programs, 1490: 146–154, 1991.
Calfee RF, WS Benedict. Carbon dioxide spectral line positions and intensities calculated
for the 2.05 and 2.7 micron regions. NBS Technical Note 332, 1966.
Carlotti M. Global-fit approach to the analysis of limb-scanning atmospheric measure-
ments. Appl Opt 27: 3250–3254, 1988.
Chance KV, JP Burrows, W Schneider. Retrieval and molecule sensitivity studies for the
global ozone monitoring experiment and the scanning imaging absorption spectrom-
eter for atmospheric chartography. Proc SPIE, Remote Sensing of Atmospheric
Chemistry, 1491: 151–165, 1991.
Chance KV, JP Burrows, D Perner, W Schneider. Satellite measurements of atmospheric
ozone profiles, including tropospheric ozone, from UV/visible measurements in the
nadir geometry: a potential method to retrieve tropospheric ozone. J Quant Spec-
trosc Radiat Transfer 57: 467–476, 1997.
Chance K, RJD Spurr. Ring effect studies: Rayleigh scattering, including molecular pa-
rameters for rotational Raman scattering, and the Fraunhofer spectrum. Appl Opt
36: 5224–5230, 1997.
Chance K. Analysis of BrO measurements from the global ozone monitoring experiment
Geophys Res Lett 25: 3335–3338, 1998.
Chance K, PI Palmer, RJD Spurr, RV Martin, TP Kurosu, DJ Jacob. Satellite observations
of formaldehyde over North America from GOME. Geophys Res Lett 27: 3461–
3464, 2000.
Chapman S. On ozone and atomic oxygen in the upper atmosphere. Phil Mag 10: 369,
1930.
European Space Agency, The GOME Users Manual, F. Bednarz, ed. European Space
Agency Publication SP-1182, Noordwijk, The Netherlands: ESA Publications Divi-
sion, ESTEC, 1995.
Fish DF, RL Jones. Rotational Raman scattering and the Ring effect in zenith-sky spectra.
Geophys Res Lett 22: 811–814, 1995.
Fleig AJ, RD McPeters, PK Bhartia, BM Schlesinger, RP Cebula, KF Klenk, SL Taylor,
DF Heath. Nimbus 7 Solar Backscatter Ultraviolet (SBUV) Ozone Products User’s
Guide. National Aeronautics and Space Administration, NASA RP-1234, Washing-
ton, D.C., 1990.
Freeman DE, K Yoshino, JR Esmond, WH Parkinson. High resolution absorption cross
section measurements of SO 2 at 213°K in the wavelength region 172–240 nm.
Planet Space Sci 32: 1125–1134, 1984.
Gamache RR, RL Hawkins, LS Rothman. Total internal partition sums in the temperature
range 70–3000K: atmospheric linear molecules. J Mol Spectrosc 142: 205–219, 1990.
Gates DM, RF Calfee, DW Hansen, WS Benedict. Line parameters and computed spectra
for water vapor bands at 2.7 µm. NBS Monograph 71, 1964.
Glaccum W, R Lucke, RM Bevilacqua, EP Shettle, JS Hornstein, DT Chen, JD Lumpe,
SS Krigman, DJ Debrestian, MD Fromm, F Dalaudier, E Chassefiere, C Deniel,

Copyright © 2002 by Taylor & Francis

 CE Randall, DW Rusch, JJ Olivero, C Brogniez, J Lenoble, R Kremer. The polar
ozone and aerosol measurement (POAM II) instrument. J Geophys Res 101:
14,479–14,487, 1996.
Goldman S, Information Theory. New York: Prentice-Hall, 1953.
Grainger JF, J Ring. Anomalous Fraunhofer line profiles. Nature 193: 762, 1962.
Heath DF, CL Mateer, AJ Krueger. The nimbus-4 backscatter ultraviolet (BUV) atmo-
spheric ozone experiment—two years’ operation. Pure Appl Geophys 106/108:
1238–1253, 1973.
Heath DF, AJ Krueger, HA Roeder, BD Henderson. The solar backscatter ultraviolet and
total ozone mapping spectrometer (SBUV/TOMS) for Nimbus G. Opt Eng 14: 323–
331, 1975.
Herman JR, PK Bhartia, O Torres, C Hsu, C Seftor, E Celarier. Global distribution of
UV-absorbing aerosols from Nimbus 7/TOMS data. J Geophys Res 102: 16,911–
16,922, 1997.
Hilsenrath E, RP Cebula, MT Deland, K Laamann, S Taylor, C Wellemeyer, PK Bhartia.
Calibration of the NOAA-11 solar backscatter ultraviolet (SBUV/2) ozone data set
from 1989 to 1993 using in-flight calibration data and SSBUV. J Geophys Res 100:
1351–1366, 1995.
Hsu NC, JR Herman, PK Bhartia, CJ Seftor, O Torres, AM Thompson, JF Gleason, TF
Eck, BN Holben. Detection of biomass burning smoke from TOMS measurements.
Geophys Res Lett 23: 745–748, 1996.
Hsu NC, JR Herman, O Torres, BN Hoben, D Tanre, TF Eck, A Smirnov, B Chatenet, F
Lavenu. Comparisons of TOMS aerosol index with sun-photometer aerosol optical
thickness: Results and applications. J Geophys Res 104: 6269–6279, 1999.
Janz SJ, E Hilsenrath, D Flittner, D Heath. Rayleigh scattering attitude sensor. Proc SPIE
2831: 146–153, 1996.
Joiner J, PK Bhartia, RP Cebula, E Hilsenrath, RD McPeters, H Park. Rotational Raman
scattering (Ring effect) in satellite backscatter ultraviolet measurements. Appl Opt
34: 4513–4525, 1995.
Joiner J, PK Bhartia. The determination of cloud pressures from rotational Raman scatter-
ing in satellite backscatter ultraviolet measurements. J Geophys Res 100: 23,019–
23,026, 1995.
Kato S, TP Ackerman, JH Mather, EE Clothiaux. The k-distribution method and corre-
lated-k approximation for a shortwave radiative transfer model. J Quant Spectrosc
Radiat Transfer 62: 109–121, 1999.
Kattawar GW, AT Young, TJ Humphreys. Inelastic scattering in planetary atmospheres.
I. The Ring effect, without aerosols. Astrophys J 243: 1049–1057, 1981.
King MD, DD Hering, DJ Diner. The Earth observing system (EOS): a space-based pro-
gram for assessing mankind’s impact on the global environment. Opt Photon News
6: 34–39, 1995.
Krueger AJ, B Guenther, AJ Fleig, DF Heath, E Hilsenrath, R McPeters, C Prabhakara.
Satellite ozone measurements. Phil Trans R Soc Lond A 296: 191–204, 1990.
Krueger AJ. The global distribution of total ozone: TOMS satellite measurements. Planet
Space Sci 37: 1555–1565, 1989.
Kurosu T, VV Rozanov, JP Burrows. Parameterization schemes for terrestrial water clouds

Copyright © 2002 by Taylor & Francis

 in the radiative transfer model GOMETRAN. J Geophys Res 102: 21,809–21,823,
1997.
Kurosu TP, KV Chance, RJD Spurr. Cloud retrieval algorithm for the European Space
Agency’s Global Ozone Monitoring Experiment. Proc SPIE, Satellite Remote Sens-
ing of Clouds and the Atmosphere III 3495: 17–26, 1998.
Kurosu TP, KV Chance, RJD Spurr. CRAG—cloud retrieval algorithm for ESA’s global
ozone monitoring experiment. Proc 1999 European Symposium on Atmospheric
Measurements from Space, ESA WPP-161: 513–521, 1999.
Kuze A, KV Chance. Analysis of cloud-top height and cloud coverage from satellites
using the O 2 A and B bands. J Geophys Res 99: 14,481–14,491, 1994.
Lewis BR, L Berzins, JH Carver. Oscillator strengths for the Schumann-Runge bands of
O 2 . J Quant Spectrosc Radiat Transfer 36: 209–232, 1986.
Lewis BR, L Berzins, CJ Dedman, TT Scholz, JH Carver. Pressure broadening in the
Schumann-Runge system of molecular oxygen. J Quant Spectrosc Radiat Transfer
39: 271–282, 1988.
Loyola D, W Balzer, B Aberle, M Bittner, K Kretschel, E Mikusch, H Mühle, T Ruppert,
C Schmid, S Slijkhuis, R Spurr, W Thomas, T Wieland, M Wolfmüller, Ground
segment for ERS-2 GOME sensor at the German D-PAF. Proc 3 rd ERS Symposium,
Space at the Service of Our Environment, ESA SP-414: 591–596, 1997.
Maurellis AN, R Lange, W van der Zande, I Aben, W Ubachs. Precipitable water column
retrieval from GOME data. Geophys Res Lett 27: 903–906, 2000.
McClatchey RA, WS Benedict, SA Clough, DE Burch, RF Calfee, K Fox, LS Rothman,
JS Garing. AFCRL atmospheric absorption line parameters compilation. AFCRL-
TR-0096, 1973.
McCormick MP, JM Zawodny, RE Veiga, JC Larsen, PH Wang. An overview of SAGE
I and SAGE II ozone measurements. Planet Space Sci 37: 1567–1586, 1989.
McElroy CT. Stratospheric nitrogen dioxide concentrations as determined from limb
brightness measurements made on June 17, 1983. J Geophys Res 93: 7075–7083,
1988.
Miller AJ, A review of satellite observations of atmospheric ozone. Planet Space Sci 37:
1539–1554, 1989.
Minschwaner K, GP Anderson, LA Hall, K Yoshino. Polynomial coefficients for calculat-
ing O 2 Schumann-Runge cross sections at 0.5 cm ⫺1 resolution. J Geophys Res 97:
10,103–10,108, 1992.
Munro R, BJ Kerridge, JP Burrows, K Chance. Ozone profile retrievals from the ESA
GOME instrument. 1992 Quadrennial Ozone Symposium, 1992.
Munro R, R Siddans, WJ Reburn, BJ Kerridge. Direct measurements of tropospheric ozone
distributions from space. Nature 392: 168–171, 1998.
Noël S, H Bovensmann, JP Burrows, J Frerick, KV Chance, APH Goede. Global atmo-
spheric monitoring with SCIAMACHY. Physics Chem Earth 24: 427–434, 1999.
Palmer PI, DJ Jacob, K Chance, RV Martin, RJD Spurr, TP Kurosu, I Bey, R Yantosca,
A Fiore, Q Li. Air-mass factor formulation for spectroscopic measurements from
satellites: application to formaldehyde retrievals from the global ozone monitoring
experiment. J Geophys Res 106: 14,539–14,550, 2001.
Platt U, Differential optical absorption spectroscopy (DOAS). In: MW Sigrist, ed. Air

Copyright © 2002 by Taylor & Francis

 Monitoring by Spectroscopic Techniques. New York: John Wiley & Sons, 1994,
pp 27–84.
Richter A, F Wittrock, M Eisinger, J Burrows. GOME observations of tropospheric BrO
in northern hemisphere spring and summer 1997. Geophys Res Lett 25: 2683–2686,
1998.
Roscoe HK, JA Pyle. Measurements of solar occultation: the error in a naive retrieval if
the constituent’s concentration changes. J Atmos Chem 5: 323–341, 1987.
Rossow WB, AW Walker, DE Beuschel, MD Roiter. International satellite cloud climatol-
ogy project (ISCCP) documentation of new cloud datasets. WMO/TD-No. 737,
World Meteorological Organization, 1996.
Rothman LS, CP Rinsland, A Goldman, ST Massie, DP Edwards, J-M Flaud, A Perrin,
C Camy-Peyret, V Dana, J-Y Mandin, J Schroeder, A McCann, RR Gamache, RB
Wattson, K Yoshino, KV Chance, KW Jucks, LR Brown, V Nemtchinov, P Vara-
nasi. The HITRAN Molecular Spectroscopic Database and HAWKS (HITRAN At-
mospheric Workstation), 1996 edition. J Quant Spectrosc Radiat Transfer 60: 665–
710, 1998.
Singer SF, RC Wentworth. A method for the determination of the vertical ozone distribu-
tion from a satellite. J Geophys Res 62: 299–308, 1957.
Slijkhuis S, A von Bargen, W Thomas, K Chance. Calculation of undersampling correction
spectra for DOAS spectral fitting. Proc ESAMS’99—European Symposium on At-
mospheric Measurements from Space, 1999, pp 563–569.
Solomon S, AL Schmeltekopf, RW Sanders. On the interpretation of zenith sky absorption
measurements. J Geophys Res 92: 8311–8319, 1987.
Thomas W, E Hegels, S Slijkhuis, R Spurr, K Chance. Detection of biomass burning
combustion products in Southeast Asia from backscatter data taken by the GOME
spectrometer. Geophys Res Lett 25: 1317–1320, 1998.
Torres O, PK Bhartia, JR Herman, Z Ahmad, J Gleason. Derivation of aerosol properties
from satellite measurements of backscattered ultraviolet radiation: theoretical basis.
J Geophys Res 103: 17,099–17,110, 1999.
van Oss R. Aerosol retrieval with GOME. Proc 1999 European Symposium on Atmo-
spheric Measurements from Space, ESA WPP-161, 1999, pp 581–585.
Wagner T, U Platt. Satellite mapping of enhanced BrO concentrations in the troposphere.
Nature 395: 486–490, 1998.
Yoshino K, DE Freeman, JR Esmond, WH Parkinson. High resolution absorption cross
section measurements and band oscillator strengths of the (1,0)–(12,0) Schumann-
Runge bands of O 2 . Planet Space Sci 31: 339–353, 1983.
Yoshino K, DE Freeman, WH Parkinson. High resolution absorption cross section mea-
surements of N 2 O at 295–299K in the wavelength region 170–222 nm. Planet
Space Sci 32: 1219–1222, 1984.
Yoshino K, JR Esmond, AS-C Cheung, DE Freeman, WH Parkinson. High resolution
absorption cross sections in the transmission window region of the Schumann-
Runge bands and Herzberg continuum of O 2 . Planet Space Sci 40: 185–192, 1992.

Copyright © 2002 by Taylor & Francis

14
Ultraviolet Spectroscopy
in Astronomy

George R. Carruthers
Naval Research Laboratory, Washington, D.C.

I. ULTRAVIOLET ASTRONOMY: AN INTRODUCTION

A. Why Ultraviolet Astronomy?
Astronomy, to a much greater degree than most other fields of science, is depen-
dent on remote sensing measurements, because most of the objects under study
are inaccessible to direct, in situ measurements or sample returns. Until the advent
of space-based observations, astronomical observations (including planetary as-
tronomy and remote-sensing measurements of Earth’s upper atmosphere) were
limited to the wavelength ranges of the electromagnetic spectrum that can pene-
trate Earth’s lower atmosphere, which is opaque to radiations having wavelengths
below about 300 nm ⫽ 3000 Å (see Fig. 1). This precluded observations of nearly
all of the ultraviolet, x-ray, and gamma ray regions of the spectrum.
Spectroscopy is the most detailed and quantitative remote sensing tech-
nique for determination of the compositions and temperatures of the objects under
study. By observing radiation emitted or absorbed at specific wavelengths of the
spectrum, one can determine, to varying degrees (depending on the wavelength
and spectral resolution selected), both the composition and the temperature of
an astronomical object. In some cases, one can also determine the velocity (in
the line of sight) of the object or gas cloud, using the Doppler effect.
The most sensitive and quantitative measurements of the compositions of
gaseous objects are obtained from the study of resonance transitions: spectral
transitions involving the ground state of an atom or molecule. This is due to the
fact that in most cases, even at relatively high temperatures, nearly all of the
atoms or molecules of common substances reside in the ground electronic state.

Copyright © 2002 by Taylor & Francis

Figure 1 Graph of the altitude at which unit optical depth (attenuation to 1/e, or 37%)
for radiation vertically incoming to Earth’s atmosphere is reached, versus wavelength in
the ultraviolet spectral range. The atmospheric constituents contributing to this absorption,
and significant ionization limits, are shown. (From R Meier Naval Research Laboratory
[NRL].)

Therefore, particularly for the most common atoms and molecules present in
astronomical objects or their atmospheres, observations of resonance spectral
transitions provide the highest sensitivity of detection, and (especially for reso-
nance absorption spectroscopy) the greatest quantitative accuracy.
The resonance transitions of the most abundant atoms and molecules in
astronomical objects, such as those of H, H 2 , He, C, CO, O, O 2 , N, N 2 , and
Ar, all lie in the far-ultraviolet region of the electromagnetic spectrum, which is
inaccessible to ground-based observatories (except for very high-redshift objects,
such as quasars).
The temperatures of objects can be determined from observations of both
spectral line transitions (of gases) and the continuous spectrum (of stars and
cooler, high-density objects such as planets). As is well known, the peak of the
continuous spectrum of a ‘‘black body’’ radiator shifts toward shorter wave-
lengths in inverse proportion to temperature, and the total emitted radiation in-
creases as the fourth power of temperature. Although stellar spectra show only
crude resemblance to the spectra of black body radiators, it is nevertheless true
that the peak of the continuous spectrum shifts toward shorter wavelengths as

Copyright © 2002 by Taylor & Francis

the ‘‘effective’’ temperature of the star (the temperature of a perfect black-body
radiator having the same total luminosity and surface area as the star) increases.
For stars much hotter than the Sun, most of the emitted radiation falls in
the ultraviolet (UV) wavelength range below the transmission limit of Earth’s
atmosphere (about 300 nm), and hence it is difficult to make accurate determina-
tions of the total stellar luminosity and effective temperature from ground-based
observations alone. Figure 2 is a graph of theoretical model atmosphere flux vs.
wavelength curves, for stars with a range of effective temperatures, showing that

Figure 2 Theoretical model atmosphere predictions of surface radiation flux vs. wave-
length, for a range of effective temperatures. Only the portion of the spectra longward or
300 nm are accessible to ground-based observatories. (From Kurucz, 1979.)

Copyright © 2002 by Taylor & Francis

the sensitivity of emitted flux to temperature is much greater near or shortward
of the peak of maximum emission than far longward of that peak (Kurucz, 1979).
Prior to the advent of space-based far-UV observations of hot stars, the
only directly observable evidence concerning their high temperatures (besides
their visible-light spectra) was the fluorescence of interstellar gas (emission nebu-
las, also known as ionized hydrogen, or H II, regions) induced by the far-UV
emissions from these stars. Far-UV radiation of wavelengths less than 91.2 nm,
the series limit of the Lyman (resonance) line series of atomic hydrogen, is capa-
ble of ionizing hydrogen. When the hydrogen ions recombine with electrons,
spectral line emission in the visible (Balmer) series, as well as in other line series,
of atomic hydrogen is produced. The integrated intensity of the nebular emission,
when ratioed to the observed visible brightness of the star, provides an indication
(but not an accurate measure) of the temperature of the star.

B. Difficulties of Ultraviolet Astronomy

The primary difficulty, which prevented ultraviolet astronomy until the early
1950s, is that atmospheric absorption prevents observations in the UV wave-
length range below 300 nm from ground-based observatories. Therefore, the ad-
vent of UV astronomy had to await the advent of the ‘‘space age’’ to allow space-
based observations. Figure 1 shows the altitudes to which the instrumentation
must be carried to observe 1/e (36.8%) of the total incoming radiation from an
(overhead) astronomical object, vs. wavelength. As seen, although significant
improvements can be made in observations of infrared and microwave wave-
lengths, at altitudes accessible by aircraft and balloons, only rocket vehicles can
reach altitudes high enough to enable observations in the UV and x-ray wave-
length ranges.
The first UV spectra of an astronomical object, the Sun, were obtained by
Naval Research Laboratory scientists using instruments flown on V-2 rockets
brought to the United States from Germany after World War II (see Fig. 3).
Similar observations of the (much fainter) other stars were made significantly
later, beginning in the early 1960s, in part because of the requirement for accurate
pointing systems on the rocket vehicles used for the observations.
As is true of most other scientific investigations requiring space flight, the
cost and required reliability of the instrumentation are much higher than for
equivalent instrumentation in ground-based observatories. Also, opportunities for
making space-based observations (except from spacecraft already in orbit) are
much fewer and more limited in observing flexibility.
Another difficulty is that the nature of instrumentation usable for ultraviolet
measurements is somewhat different than for equivalent instrumentation op-
erating in visible and near-visible wavelengths (see Samson, 1967). In particular,
the reflectivities of commonly used mirror coatings and the transmissions of re-

Copyright © 2002 by Taylor & Francis

Figure 3 Spectra of the Sun obtained with a spectrograph flown on a V-2 rocket in
1946, vs. altitude, showing (at the higher altitudes) the UV wavelength range below 300
nm (3000 A) for the first time. (From R Tousey, NRL.)

fractive optics made of commonly used glasses are very low in the ultraviolet,
particularly at the far-UV wavelengths (below 200 nm). Also, window and mirror
coating materials that are usable in far-UV wavelengths may be degraded by
exposure to moisture and other forms of contamination to a greater degree than
materials used at visible and near-visible wavelengths. These considerations re-
sult in a much more limited choice of refractive materials and mirror coatings,
and may require minimizing, to the extent possible, the total number of reflective
and/or refractive elements in the optical system.
At the very shortest (extreme-UV; EUV) wavelengths (below 100 nm), the
limitations on reflectivities and transmissions of optical materials are even more
severe than in the far-UV (100–200 nm) wavelength range. This requires, in
general, use of grazing incidence reflective optics, and/or limiting the instrument
to a single reflection. No known materials usable for windows or refractive optics
transmit below about 105 nm, which is the transmission limit of lithium fluoride
(LiF). However, very thin films of metals such as Al and In can be used to
transmit selected wavelength ranges in the EUV, and recently developed
multilayer interference coatings can provide high normal-incidence reflectivities
in selected, relatively narrow wavelength ranges.
As is true in other wavelengths of the electromagnetic spectrum, there is
a tradeoff between spectral resolution and sensitivity: in general, the higher the
resolution, the lower the sensitivity, because the same (or similar) amount of
radiation is spread over a larger number of detector elements. Therefore, to an

Copyright © 2002 by Taylor & Francis

even greater extent than in visible wavelengths, the ultraviolet spectrograph must
be optimized for the primary scientific objective it is to address, and it is difficult
to build a practical ‘‘multipurpose’’ instrument.

II. ULTRAVIOLET SPECTROSCOPIC INSTRUMENTATION

A. Characteristics of Materials for Transmission Optics
Most glasses used in ground-based astronomical instruments do not transmit, or
transmit poorly, in UV wavelengths shortward of about 300 nm. The most com-
monly used UV transmissive materials for refractive optics and windows, in order
of decreasing short-wavelength transmission cutoff, are as follows:

UV-grade fused silica (SiO 2) 165 nm

Sapphire (Al 2 O 3) 145 nm
Barium fluoride (BaF 2) 135 nm
Strontium fluoride (SrF 2) 128 nm
Calcium fluoride (CaF 2) 123 nm
Magnesium fluoride (MgF 2) 115 nm
Lithium fluoride (LiF) 105 nm

It is noteworthy that the fluorides have short-wavelength limits that are

temperature sensitive: they shift to longer wavelengths when heated above room
temperature, and to shorter wavelengths when cooled below room temperature.
This temperature dependence can be used to ‘‘fine tune’’ the short-wave-
length cutoffs. For example, we have used LiF windows cooled to liquid nitrogen
temperatures to detect hydrogen Lyman-β (102.5 nm) radiation (Carruthers,
1971), and SrF 2 heated to 100°C to transmit atomic oxygen 135.6 nm radiation
while mostly excluding atomic oxygen 130.4 nm radiation (Carruthers and
Seeley, 1996).
Since the fluorides are crystalline materials, they are more difficult to work
with in generating optical surfaces than are fused silica and most glasses. Also,
they have relatively high coefficients of thermal expansion and hence are suscep-
tible to thermal shock. They are also (to varying degrees) water soluble, and their
transmission characteristics may be degraded by exposure to moisture or humid
air.
Another difficulty in using these materials for refractive optics is that the
index of refraction varies with wavelength more rapidly toward shorter wave-
lengths (and the curve is very steep near the short-wavelength transmission limit).
This means that refractive elements such as lenses and Schmidt correctors must
be designed for specific wavelengths, or relatively narrow-wavelength ranges,
of interest. On the other hand, the high indices of refraction near the short-

Copyright © 2002 by Taylor & Francis

wavelength cutoffs mean that flatter prisms and/or less steeply curved lens sur-
faces can be used, which minimizes transmission losses in these wavelength
ranges.

B. Characteristics of Coatings for Reflective Optics

In the visible and near-UV spectral regions, aluminum is usually the coating of
choice for mirrors and reflection gratings. However, at shorter wavelengths the
naturally occurring oxidation of bare aluminum coatings causes the reflectivity
to fall off markedly in the far-UV range.
This problem can be circumvented by deposition of a thin film of a UV-
transmissive material, such as MgF 2 , on the mirror surface immediately after
aluminization, while still in vacuum. As seen in Figure 4, optimal coating thick-
nesses of MgF 2 can provide reflectivities as high as 70% down to about 115 nm
wavelength. With the use of LiF coatings, the useful range of reflectivity can be
extended to below 102 nm (although with the disadvantages of somewhat lower
maximum reflectivity, and of higher susceptivity to degradation by atmospheric
humidity) (see Bradford et al., 1969).
For wavelengths below 100 nm, no known materials can provide high re-
flectivities (greater than 50%) over significant wavelength ranges. The best mate-
rials to use include the heavy metals platinum, osmium, and tungsten, and also
silicon carbide (SiC). Figure 5 shows typical reflectance vs. wavelength curves
for these materials.

Figure 4 Reflectance vs. wavelength for mirror coatings useful in the far-UV spectral
range. (From Bradford et al., 1969.)

Copyright © 2002 by Taylor & Francis

Figure 5 Reflectance vs. wavelength for mirror coatings useful in the wavelength range
below 100 nm. (From W Hunter, NRL; Keski-Kuha et al., 1988.)

In recent years, multilayer interference coatings have been developed that

can provide very high reflectances in selected, narrow-wavelength ranges in the
extreme ultraviolet (EUV). These have been particularly useful for monochro-
matic imaging of the sun in selected EUV and soft x-ray wavelengths, usually
corresponding to known emission-line wavelengths.

C. Minimizing the Number of Optical Elements

to Maximize Throughput
The throughput of spectrographs can be maximized, although sometimes with
some compromise in the optical quality, by using a single reflective or transmis-
sive element for more than one purpose. The best example is the use of a concave
grating, which both disperses the radiation and brings it to a focus, instead of
using a plane reflectance grating plus a focusing mirror or lens. Likewise, a tele-
scope with a single, long-focal-length primary mirror may be used to feed a
spectrograph instead of a more compact, two-mirror Cassegrain or Ritchey–
Chretien telescope of the same collecting aperture.

Copyright © 2002 by Taylor & Francis

 As mentioned, in the EUV spectral region, reflectivities of broad-band
normal-incidence mirror coatings are quite low in comparison to shorter wave-
lengths, and grazing-incidence optics may be required to obtain acceptable
throughputs. This may cause the optical system to be considerably longer (and
hence more difficult to accommodate in a spacecraft) than an equivalent-aperture
normal-incidence-reflection optical system.

D. Typical Spectrograph Configurations

1. Objective Grating Spectrographs
The objective grating, or so-called slitless, spectrograph is the simplest type of
spectrograph, consisting of only a grating and a camera (or only the grating, if
it is a concave grating), plus detector. Instruments of this type were among the
first ones used in ultraviolet astronomy, largely because of (a) the relatively poor
pointing accuracies achievable in early sounding-rocket experiments; and (b) the
desire to survey large regions of the sky, as distinct from targeting only a few
selected objects. Since the field of view can be large compared to that achievable
with a slit-type spectrograph, the spectra of several objects can be recorded simul-
taneously (using a two-dimensional imaging detector or film).
A disadvantage of this technique is that there is no intrinsic wavelength
reference, since the wavelength recorded at a particular point on the detector
format depends on both the actual wavelength and the position of the object in
the camera field of view. However, it is often the case that the positions of particu-
lar spectral lines or features known or expected to be present in the spectrum
can be used to calibrate a wavelength scale for each detected object.
Another disadvantage is that diffuse sources of large angular size cannot
be studied with as high spectral resolution as pointlike sources, such as stars.
Also, any diffuse sky glow (due to upper atmospheric or near-Earth space emis-
sions) will not be spectrally resolved, but is also recorded (as a diffuse back-
ground), and hence may degrade the detectivity or signal-to-noise ratio of the
spectra of objects of interest.
Objective Plane Grating Spectrographs. Figure 6 shows diagrams of two
types of plane grating objective spectrographs. This type of instrument was used
in sounding rocket flights, beginning in the early 1960s, by groups at Princeton
University and the Naval Research Laboratory, to observe the spectra of hot stars
extending down to wavelengths as short as 100 nm (see Morton, 1967; Carruthers,
1970, Carruthers et al., 1984). Among the results of these experiments were the
first detections of interstellar atomic hydrogen, via its Lyman-α (121.6 nm) reso-
nance absorption line (Princeton), and of interstellar molecular hydrogen via its
Lyman band resonance absorption in the 100–115 nm wavelength range (NRL).

Copyright © 2002 by Taylor & Francis

Copyright © 2002 by Taylor & Francis
 Inverse Wadsworth Concave Grating. This is the simplest of all spectro-
graphs (see Fig. 7), consisting of only a concave grating and a detector. Parallel
light rays are incident on the concave grating, which then projects a focused
spectral image on the detector. A collimator is usually necessary on the input
side to avoid contamination of the spectrum by off-axis radiation.
Because of the typically larger focal ratios (f/numbers) of concave gratings
used in this mode than with the use of a plane grating with a Schmidt or other
small-focal-ratio camera, the field of view is usually smaller, but the useful spec-
tral resolution is usually higher. This type of instrument has been used in sounding
rocket flights in which a single star was the target, rather than a wide field of
stars, but where the pointing uncertainties were too large to make use of a slit
spectrograph, and/or where the number of optical surfaces needed to be mini-
mized in order to maximize sensitivity (see Smith, 1969, 1972).

2. Rowland Circle Spectrographs

The Rowland-circle spectrograph (see Fig. 8) is the simplest slit spectrograph
design based on the use of a concave grating. Unlike in the case of objective
grating spectrographs, the spectral resolution is independent of the angular size
of the source. It can be used alone to observe the spectra of diffuse sources whose
angular extent is larger than the field of view of the grating, as projected through
the slit. However, for observations of stars and other discrete objects, an external
optical element or telescope must be used to focus an image of the object onto
the slit, and a very accurate and stable pointing system is needed to keep the
object fixed on the slit.
The spectral resolution is determined by (a) the radius of curvature of the
grating, (b) the number of lines per millimeter of the grating ruling, and (c) the
width of the entrance slit. Resolution increases in direct proportion to variables
(a) and (b), and in inverse proportion to (c). As one might guess, sensitivity varies
in the opposite direction from resolution for all three variables.
Another important consideration in spectrograph design is the wavelength
range to be covered in a single observation. Spherical concave gratings are subject
to increasing aberrations with increasing angle from the grating normal (as pro-
jected on the detector), although special figuring of the grating (e.g., toroidal

Figure 6 (a) An objective grating spectrograph, using a Schmidt camera and photo-
graphic film, flown on sounding rockets by Princeton University to obtain far-ultraviolet
spectra of hot stars. (b) An objective grating spectrograph, using a two-mirror, all-
reflecting camera with electrographic recording on film, flown on sounding rockets by
the Naval Research Laboratory to observe the far-UV spectra of hot stars.

Copyright © 2002 by Taylor & Francis

Figure 7 An inverse-Wadsworth spectrograph, a type of which was used in sounding
rocket flights by NASA Goddard Space Flight Center to observe the far-UV spectra of
hot stars.

figuring) can give significantly improved useful wavelength ranges. Therefore,

it is usually easier to increase spectral resolution (over a given wavelength range
of simultaneous observation) by increasing the focal length of the grating, rather
than increasing the number of lines per millimeter of its ruling; however, this
also makes the total instrument larger (which is a disadvantage for space flight

Copyright © 2002 by Taylor & Francis

Figure 8 Diagram of a Rowland-circle spectrograph, using a concave diffraction
grating.

experiments) and increases the effective focal ratio (which decreases the sensitiv-
ity to diffuse sources).

3. Other Spectrograph Designs

There have been numerous other types of spectrographs and spectrometers
used in UV space astronomy, as is true also in ground-based astronomy and
laboratory spectroscopy. Among these are devices using combinations of prisms
and diffraction gratings, and ones incorporating two (rather than one) diffraction
gratings.
For very high spectral resolution studies, a combination of two diffraction
gratings whose rulings are oriented at right angles to each other (i.e., so-called
cross-dispersed arrangements) can be used. A very high dispersion (i.e., finely
ruled) grating produces high spectral resolution, but only a small portion of the
spectral range can be recorded in the direction of dispersion. However, if a more
coarsely ruled echelle grating is placed ahead of this high-dispersion grating with
its dispersion in the transverse direction, multiple segments of the spectrum can

Copyright © 2002 by Taylor & Francis

be recorded simultaneously, on a two-dimensional format detector (or film). This
type of spectrograph has been used in several UV astronomy space instruments,
initially for observations of the Sun, and later for stellar observations. Instruments
of the latter type include the International Ultraviolet Explorer (IUE) satellite,
launched in 1978 and remaining in operation for more than 10 years (see Kondo
et al., 1989); spectrographs used in the Hubble Space Telescope (HST) (see
Harms et al., 1982; Brandt et al., 1982); and the Interstellar Medium Absorption
Profile Spectrometer (IMAPS) (Jenkins et al., 1988). The latter three instruments
are discussed in more detail below.

E. Ultraviolet Detectors for Astronomical

UV Spectroscopy
The earliest ultraviolet astronomy experiments, through the mid-1960s, used one
of two sensor types, similar to those then used in ground-based astronomy:
(a) photographic film, or (b) photomultiplier tubes (or other equivalent single-
element detectors). Although adequate for many first-look investigations in UV
astronomy, the limitations of these detectors are such that they have now been
almost totally replaced by two-dimensional electronic imaging detectors.

1. Photographic Film in Ultraviolet Astronomy

Photographic film was the first detection technology used in UV space astronomy,
such as the pioneering observations of the Sun in the far-UV obtained by NRL
in V-2 rocket flights beginning in the late 1940s (see Fig. 3), and objective-
grating spectroscopy of stars in the UV obtained from sounding rockets and a
Gemini manned space mission in the mid-1960s. At the present time, however,
photography has been almost totally replaced by electronic media, particularly
for long-duration space missions.
The most important disadvantages of photographic film in UV astronomy
are: (a) the film has to be specially prepared to provide useful sensitivity in the
far UV, because the absorption of UV by the gelatin in photographic emulsions
greatly reduces the amount of UV that can activate the silver halide grains that
are the active ingredients of the film; (b) the film cannot be made sensitive only
in the UV and insensitive to visible light, which is a problem particularly for
observations of the Sun and other objects that are much brighter in visible light
than in the UV; and (c) it is difficult to obtain quantitative photometric informa-
tion from film images, because of the nonlinear, highly variable, and exposure-
time-dependent responses of photographic emulsions. Other potential problems
include degradation of the photographic emulsion with time, in long-duration
space missions, due to the high vacuum and/or energetic charged particle envi-
ronments in space.

Copyright © 2002 by Taylor & Francis

2. Photomultipliers and Channel Electron Multipliers
Photomultipliers, and the closely related channel electron multipliers, are the sim-
plest of the purely electronic detectors. Unlike film, they have linear and easily
calibrated responses to electromagnetic radiation and can be made to have high
sensitivity in the UV, and little or no visible light sensitivity, by appropriate
choice of the photocathode and window materials (including, in space, totally
windowless operation). They also can have much higher detection efficiencies
than film (counts per incident photon, or quantum efficiency). Their output elec-
tronic signals can be sent back to Earth by radio telemetry, eliminating the need
for film return and hence making possible long-duration and/or deep-space mis-
sions.
Their major disadvantage is that they can record only one wavelength or
spatial resolution element at one time. To record a full spectrum of an astronomi-
cal object, the detector must be scanned over the desired wavelength range, taking
a reading at each spectral resolution increment along the spectrum. Not only does
this greatly increase the time required to record a spectrum, but it also does not
differentiate time variations of spectral intensity (e.g., due to pointing oscillations
or changes in atmospheric absorption) from changes of spectral intensity with
wavelength.
Both of these devices make use of photoelectric emission, from a photosen-
sitive surface (photocathode). By the use of appropriate photocathode materials
and window materials, the sensitivity of the device can be tailored to the wave-
length range of interest (Sommer, 1968). For example, photocathode materials
that are highly sensitive in the middle- and far-UV ranges but are totally blind
to visible light can be used. Windowless photomultipliers and channel multipliers
can be made sensitive in the EUV, as well as far-UV, wavelength ranges.
Both devices make use of the process of secondary emission to amplify
the charge corresponding to a single photoelectron to produce output signals of
10 6 –10 8 output electrons, which is adequate to produce a readily detectable sig-
nal, or count, in the output electronic signal. A device that can count the individ-
ual photoelectrons released from the front-end photosensitive surface provides
the maximum possible signal-to-noise ratio. This ratio is limited by statistics:
S/N ⫽ N/√N ⫽ √N, where N is the total number of photoelectron counts.
In a photomultiplier, the secondary amplification is achieved using a series
of secondary-emitting dynodes (see Fig. 9). In the channel electron multiplier, a
continuous tube of semiconducting glass is used to amplify photoelectrons pro-
duced at the front end of the device, which makes possible a simpler and more
compact device than the discrete-dynode photomultiplier (see Fig. 10).
A major UV astronomy mission that used photomultiplier detectors was
the Princeton University’s instrument on the third Orbiting Astronomical Obser-
vatory (OAO-3, re-named Copernicus) launched by the National Aeronautics and

Copyright © 2002 by Taylor & Francis

Figure 9 Diagrams of two photomultiplier tubes, using different types of dynodes for
amplification (by secondary emission) of electrons produced by photoemission from semi-
transparent photocathodes.

Space Administration (NASA) in 1972. The primary scientific objective of Co-

pernicus was to study the composition and spatial distribution of interstellar gas,
via its far-UV absorption spectra. This instrument (see Fig. 11) used a concave
grating spectrograph and an 80 cm aperture telescope mirror. Windowless photo-
multipliers were physically scanned over the Rowland circle of the spectrograph,
providing moderate (0.02 nm) and high (0.005 nm) spectral resolutions in the
95–160 nm wavelength range.

3. Electronic Imaging Devices

Television Camera Tubes. Television camera tubes, of the type called
secondary electron conduction (SEC) vidicons, were used in two major UV as-
tronomy missions. The first was for far-UV direct imagery of stars, in the second
Orbiting Astronomical Observatory (OAO-2) launched in 1968. The second was
for the spectrographs on board the International Ultraviolet Explorer (IUE)
launched in 1978. These devices circumvented most of the disadvantages of film;

Copyright © 2002 by Taylor & Francis

Figure 10 Diagram of a channel electron multiplier, which uses a continuous second-
ary-emitting tube surface to amplify photoelectron input to, or produced at, the front end
of the tube.

Figure 11 Diagram of the instrumentation of the third Orbiting Astronomical Observa-

tory (OAO-3), named Copernicus, launched into orbit in 1972. The instrument, provided
by Princeton University, used a reflecting telescope and concave-grating spectrograph,
and windowless photomultiplier tubes which were scanned along the Rowland circle, to
record the far-UV spectra of stars. (Courtesy of Princeton University Observatory.)

Copyright © 2002 by Taylor & Francis

in particular, since they generated electronic signals that could be radioed to
ground stations, they were not subject to the limitations on observing time set
by an initial supply of film, or to degradation of film in the space environment. In
particular, the IUE’s ability to take long time exposures with its two-dimensional,
integrating SEC vidicon detectors enabled it to observe much fainter stars than
was possible with Copernicus, which had a much larger telescope aperture (al-
though with lower spectral resolution, and limited to wavelengths longward of
115 nm).
Microchannel Plate-Based Detectors. The microchannel plate (MCP) is
an extension of the channel electron multiplier technology to provide a two-
dimensional array of individual, miniaturized channel multipliers that can operate
simultaneously, but with independently recorded signal outputs (see Fig. 12).
This provides, as a result, the major advantages of photographic film (simultane-
ous recording of a two-dimensional field of view) with the much-improved sensi-
tivity, quantitative accuracy, and wavelength range selectability of a photoelectric

Figure 12 Diagram of a microchannel plate, which is based on the principle of the

channel electron multiplier (Fig. 10). Here, many channel multipliers are fused together
into a continuous plate, which has the ability to detect simultaneously the input radiation
at all points over a two-dimensional image.

Copyright © 2002 by Taylor & Francis

detector. When the MCP is used with various photocathode materials, deposited
either on the front face of the MCP or on a separate photocathode substrate, the
wavelength range of response can be varied to meet the investigation’s objectives.
In comparison to UV-sensitive television camera tubes, the MCP detectors
have the advantage of being much more compact and lighter in weight, and can
count individual photoelectron events rather than producing an analog video sig-
nal. This results in both better sensitivity and better quantitative accuracy.
Charge-Coupled Device-Based Detectors. The charge-coupled device
(CCD) is a solid-state electronic imaging device that can provide large-format,
two-dimensional images in visible and near-infrared radiation by the principle
of photoconduction. However, CCDs can also be used for detection of ultraviolet
radiation by one of three techniques: (a) the CCD can be made sensitive to UV
radiation, as well as visible and UV radiation, by special processing; (b) the CCD
can be coupled to an image-intensifier tube with a UV-sensitive photocathode,
to provide UV images without contamination by visible and near-IR radiation;
and (c) photoelectrons emitted by a UV-sensitive photocathode can be made to
impact the CCD directly, thereby producing an electronic output signal (electron-
bombarded CCD). All three of these approaches have been used in UV astronomi-
cal investigations.
Film Recording Electronic Imaging Devices. Many of the disadvantages
of film can be circumvented by using it in conjunction with electronic imaging
devices. Image intensifiers or image converters are devices with photoelectric
emission surfaces (photocathodes) of the same types usable with photomultipliers
and channel electron multipliers, but that form so-called electron images on a
phosphor screen, which then gives off visible light (as in a cathode-ray tube).
These devices may also include microchannel plates to amplify the electron im-
age before it reaches the output phosphor screen. The light from the phosphor
screen can then be transferred to the recording photographic film, using either
lenses or (more efficiently) a fiber-optic rear faceplate (with the phosphor screen
deposited on one side and the film contacting the other side).
Another variation on this theme is the electrographic or electronographic
detector (see Fig. 6b), in which film is used to record the electron image of the
image tube directly. This has the advantage of greater simplicity, for detectors
operating in the vacuum ultraviolet, provided that photocathode materials (such
as the alkali halides) relatively insensitive to degradation by outgassing of the
recording emulsion are used (Carruthers, 1965). The electrographic technique is
more quantitative than direct or image-intensified photography, because energetic
electrons striking the emulsion directly leave discrete spots, or tracks, which can
be individually counted, and (unlike ordinary photography) the optical density
of the processed emulsion is directly proportional to the total number of photo-
electron hits per unit of area.

Copyright © 2002 by Taylor & Francis

Figure 13 An objective-grating spectrogram of the region of the Belt of Orion, obtained
by NRL in a 1982 sounding rocket flight, using an all-reflecting objective-grating spectro-
graph and electrographic recording. The field of view is 10 degrees in diameter, and the
spectral range is 90–160 nm (shorter wavelengths to the right in each spectrum). The
spectral resolution is about 0.3 nm.

The electrographic recording technique has been used by NRL in several

sounding rocket flights of objective-grating spectrographs to record the spectra
of hot stars, in some cases down to the Lyman limit of 91.2 nm (see Fig. 13).

III. EXAMPLES OF CURRENT UV SPECTROGRAPHIC

INSTRUMENTS AND APPLICATIONS TO UV
SPACE ASTRONOMY
A. Solar Ultraviolet Spectroscopy Space Missions
As mentioned previously, the Sun was the first target of UV space astronomy,
both because of its practical importance to us here on Earth and because it is
many orders of magnitude brighter than any other star, even in the far UV. Most
of the spectrograph types and detector technologies described in the preceding

Copyright © 2002 by Taylor & Francis

section were used first in solar physics studies, before they were used to study
objects in deep space. Solar UV astronomy began with the first V-2 rocket flights
from White Sands Missile Range in the late 1940s, and followed with other
sounding rocket flights (continuing to the present time). The first dedicated solar
ultraviolet astronomy satellite missions were the Orbiting Solar Observatories
(OSOs), beginning in the 1960s. Later missions included the Skylab space station
launched in 1973, the Solar Maximum Mission (NASA SMM, 1987), launched
on a Delta rocket in 1980 (and later repaired by a space shuttle crew in 1984),
and the Spacelab-2 mission launched in 1985.
Because the Sun is many orders of magnitude brighter in the visible than
in the far UV, spectrograph and detector technologies must include special pre-
cautions to avoid contamination of the far-UV spectra by stray visible light. These
include use of cross dispersion (see Fig. 14), in which a low-dispersion diffraction

Figure 14 Diagram of a solar ultraviolet spectrograph, using a predisperser grating to

minimize the input of visible solar radiation into the main spectrograph.

Copyright © 2002 by Taylor & Francis

grating, ahead of the main spectrograph entrance slit and oriented perpendicularly
to the primary spectrograph grating, predisperses the radiation before it enters
the main spectrograph slit. This effectively prevents most of the visible light from
entering the main spectrograph slit. Other methods include the use of solar-blind
photoelectric detectors, which are sensitive in the far UV but insensitive to visible
and infrared radiation.
An example of a high-resolution solar UV spectroscopy instrument is the
Naval Research Laboratory’s High-Resolution Telescope and Spectrometer
(HRTS) (Bartoe and Brueckner, 1975), which was flown in several sounding
rocket flights and as part of the Spacelab-2 Shuttle mission in 1985 (see Fig. 15).
This long-slit imaging spectrograph was used to measure both the spatial and
spectral distributions of solar far-UV radiation at various locations on the solar
surface (including, for example, sunspots and other active regions) and variations
with solar activity (by multiple sounding rocket reflights).
Another complementary instrument developed by NRL was the Solar Ultra-
violet Spectral Irradiance Monitor (SUSIM) (Brueckner et al., (1991). This (see
Fig. 16) was intended to monitor the total solar UV radiation output, with lower
spectral resolution than provided by HRTS, but emphasizing high photometric
accuracy and time stability. This instrument flew, along with HRTS, on the
Spacelab-2 shuttle mission. SUSIM also was flown on three later Shuttle-based
Atmospheric Laboratory for Applications and Science (ATLAS) missions, the
first launched in 1992. It also flew on a much longer duration unmanned satellite
mission, the Upper Atmosphere Research Satellite (UARS), launched in 1991
(and still in operation at this writing). Because solar activity varies in an 11
year cycle, long-duration monitoring with highly stable and accurately calibrated
instrumentation is essential.
The overlap between the ATLAS and UARS missions allowed direct, si-
multaneous comparisons of the results of two SUSIM instruments. The Shuttle-
based instrument had the advantage of being calibrated immediately before and
after its relatively short-duration missions, so that it could track any changes in
the calibration of the long-duration UARS instrument.

B. Hubble Space Telescope

The Hubble Space Telescope (HST) is the most powerful and best known (to
the general public, as well as the scientific community) of all nonsolar space
astronomy projects (NASA NP-126) (see Fig. 17). Although its primary objective
is not specifically UV space astronomy, this is nevertheless an important part
of the HST ’s total suite of observation programs and complement of scientific
instruments. HST’s large aperture, high spatial resolving power, and long dura-
tion on orbit make it by far the best single example illustrating the instrumental
technologies, and their evolution in time, of UV space astronomy.

Copyright © 2002 by Taylor & Francis

Figure 15 Diagram of the Naval Research Laboratory’s High Resolution Telescope and
Spectrograph (HRTS), which was flown on several sounding rockets and the Spacelab-2
Shuttle mission to obtain high-resolution, spatially resolved far-ultraviolet spectra of the
Sun.

Copyright © 2002 by Taylor & Francis

Figure 16 (a) Diagram of the Naval Research Laboratory’s SUSIM, which was intended
for highly accurate measurements of the intensity and time variations of the solar UV
spectrum. (b) Diagram of the use of two identical spectrometers and a calibration lamp,
which ensure the highest possible accuracy of measurements by SUSIM of the solar UV
spectrum and its potential time variations.

Copyright © 2002 by Taylor & Francis

Figure 17 Diagram of the Hubble Space Telescope (HST) shows the main telescope
optics and the scientific instrument modules. (From NASA NP-126.)

When the HST and its first-generation suite of scientific instruments were
approved for development and flight, in the latter 1970s (NASA CP-2244, 1982),
the state of the art in UV astronomy instrumentation, and specifically detector
technology, was far from what it is at present. However, since HST was designed
for servicing by astronauts in space shuttle refurbishment missions, this has al-
lowed the initially installed instruments to be upgraded or replaced with newer
ones over the duration of HST’s mission.

1. Faint Object Spectrograph and High-Resolution

Spectrograph
When HST was launched into orbit in 1990, it contained (among several other
instruments) a faint object spectrograph (FOS; Harms et al., 1982) and a high-
resolution spectrograph (HRS; Brandt et al., 1982). Both had sensitivities ex-
tending into the far UV, down to wavelengths as short as 115 nm. As their names
imply, the two instruments had different tradeoffs between spectral resolution

Copyright © 2002 by Taylor & Francis

and sensitivity, although both instruments individually provided wide ranges of
selectable spectral resolutions.
The FOS (see Fig. 18) had two separate but similar spectrographic chan-
nels, each incorporating its own detector, emphasizing different but overlapping
wavelength ranges. The ‘‘red’’ detector was sensitive in the 180–850 nm spectral
range, whereas the ‘‘blue’’ detector covered the 115–500 nm spectral range. The
FOS also had a selection of gratings (and a prism) providing resolving powers
(λ/∆λ) in the range 100–1200.
The HRS (see Fig. 19) also had two separate detector channels, covering
the wavelength ranges 110–170 nm and 115–320 nm, and selections of gratings
providing spectral resolving powers in the range 2000–100,000. The latter, high-
resolution modes utilized cross dispersion gratings in conjunction with first-order
gratings to provide this capability.
Both the FOS and HRS utilized one-dimensional electronic imaging detec-
tors, called digicons, which were similar in principle to the current electron-bom-
barded CCDs described previously, except that they had only one-dimensional
arrays of 512 detector elements each. This meant that they could simultaneously
image 512 spectral elements, but only one spatial element (transverse to the dis-

Figure 18 Diagram of the FOS, which was one of the first set of scientific instruments
on HST when it was first launched in 1990. As shown, two separate optical systems and
detectors were used to observe in the UV and visible spectral ranges. (From NASA CP-
2244.)

Copyright © 2002 by Taylor & Francis

Figure 19 Diagram of the HRS, one of the original complement of HST scientific in-
struments. The HRS used two detectors and various combinations of mirrors and gratings,
to cover the UV spectral range with various spectral resolutions and wavelength ranges.
(From NASA CP-2244.)

persion direction), at a time. Nevertheless, the capabilities of these instruments

for UV astronomy (in the wavelength ranges to which they were sensitive) far
exceeded those of previous space experiments, particularly following the installa-
tion of the spherical-aberration-correcting optics in the first Shuttle HST servicing
mission in 1993.
However, NASA also solicited proposals for more advanced scientific in-
struments, to be installed in HST during later servicing missions. The first of
these with far-UV spectroscopic capabilities was the Space Telescope Imaging
Spectrograph (STIS), which replaced both the FOS and HRS in the 1997 shuttle-
servicing mission. Planned for a near-future servicing mission is another far-UV
spectrograph, the Cosmic Origins Spectrograph (COS).

2. Space Telescope Imaging Spectrograph

The STIS (Woodgate et al., 1986) was one of several second-generation scientific
instruments replacing or supplementing originally installed scientific instruments
in space shuttle servicing missions. STIS was designed to replace both the FOS

Copyright © 2002 by Taylor & Francis

Figure 20 Diagram of the STIS, which replaced both the FOS and HRS during the
1996 Shuttle refurbishment mission to HST. By the use of two-dimensional imaging detec-
tors, STIS was able to increase greatly the efficiency of observations, especially those of
extended sources. (From B Woodgate, NASA Goddard Space Flight Center.)

and the HRS, and was installed in the second space shuttle servicing mission in
1997.
The major improvement in the STIS (see Fig. 20) over the previous FOS
and HRS was provided by its use of two-dimensional-format electronic imaging
devices, in place of the one-dimensional digicons used in the first-generation
instruments. This made possible the concept of an imaging spectrograph, in which
spectral imaging takes place along one dimension of the two-dimensional array,
and spatial imaging (in the along-slit direction) takes place in the transverse di-
mension. In effect, this allows the variation in intensity of incoming radiation at
each wavelength along the spectrum to be measured in the direction perpendicular
to that of spectral dispersion. For ‘‘extended’’ objects (i.e., other than stars or
other pointlike sources), such as planets, nebulae, and galaxies, this provides
much higher sensitivity and/or provides a much more efficient capability for
‘‘mapping’’ the brightness distribution over the extended object (see Table 1).

Copyright © 2002 by Taylor & Francis

This has proven very effective in, for example, verifying the presence of massive
black holes in the centers of active galaxies.
The STIS uses two types of detectors: (a) a microchannel-plate-based detec-
tor, known as the multianode microchannel array (MAMA), used for the middle-
and far-UV spectral ranges (110–300 nm); and (b) large-format CCD arrays for
the near-UV, visible, and near-infrared spectral range.

3. Cosmic Origins Spectrograph

The COS is a new, third-generation instrument to be installed in HST in a future
Shuttle servicing mission (Green et al., 1999). Its scientific objectives require
studies of some of the faintest and most distant objects yet observed in the far-
UV. This requires it to have the maximum possible sensitivity in the far-UV
spectral range, which is achieved by minimizing the number of reflections in the
optical system and by other technological improvements vs. the current multipur-
pose STIS instrument.

C. Hopkins Ultraviolet Telescope

The Hopkins Ultraviolet Telescope (HUT) (Davidsen et al., 1993) was one of
three UV astronomy instruments which was part of the Astro space shuttle pay-
load, flown on two shuttle Spacelab missions, in 1990 and 1995 (see Fig. 21).
It was optimized for the very far ultraviolet wavelength range, below 110 nm,
which is inaccessible to HST and most previous UV astronomy space missions
(except Copernicus). This is achieved by using iridium coatings on the telescope
mirror and concave grating, and using a windowless microchannel-plate detector.
The HUT made the first high-sensitivity, moderate-resolution spectro-
graphic studies in the 91–110 nm wavelength range, which included studies of
quasars and other extragalactic objects, as well as stars in our galaxy, planets in
our solar system, and our Moon.

D. Interstellar Medium-Absorption Profile Spectrometer

The IMAPS instrument is a small, special-purpose instrument developed by
Princeton University, which was first flown on sounding rockets and then on two
Shuttle AstroSPAS missions (Jenkins et al., 1988) (see Fig. 22). It was specifically
designed to provide extremely high spectral resolution (λ/∆λ ⫽ 2 ⫻ 10 5) at very
short far-UV wavelengths (95–115 nm), which are not accessible with HST due
to the low reflectance of its mirror coatings in this wavelength range. Species of
the interstellar medium of special interest in this wavelength range include H 2 ,
HD, D, and O ⫹5 (O VI). As shown in Figure 22, IMAPS uses a cross-dispersed
optical system, in which the echelle grating is operated as an objective grating.

Copyright © 2002 by Taylor & Francis

Table 1 STIS Modes Summary

MAMA MAMA CCD

(Csl)/MgF 2 (Cs 2 Te)/MgF 2 MgF 2 window
Detector (photocathode)/window 115–170 165–310 305–555 550–1000
Wavelength (nm)
Spectral band 1 2 3 4

Pixel angular size (arc sec) 0.025 0.025 0.050 0.050

Low-resolution spectral Mode number 1.1 a 2.1 a 3.1 b 4.1 b

imaging (first order) Resolving power 960–1400 525–985 560–1010 505–971
Slit length (X, arc sec) 25 25 50 50
Effective area (cm 2) 500 940 6500 5200
Medium-resolution spectral Mode number 1.2 2.2 3.2 4.2
imaging (scanning) Resolving power (⫻10 4) 1.00–1.51 0.93–1.8 0.54–1.0 0.49–0.90
Slit length (X, arc sec) 30 30 50 50
Scan angle (degree) 5.92 9.55 9.06 8.10
Effective area (cm 2) 360 740 7100 5900
No. exposures/band 11 18 10 9
Medium-resolution echelle Mode number 1.3 2.3 N/A N/A
Resolving power (⫻10 4) 2.3 2.3
Slit length (X, arc sec) 0.43 0.27
Effective area (cm 2) 320 660
No. exposures/band 1 2

Copyright © 2002 by Taylor & Francis

High-resolution echelle Mode number 1.4 2.4 N/A N/A
Resolving power (⫻10 5) 1.05 1.05
Slit length (X, arc sec) 0.12 0.1
Effective area (cm 2) 230 480
No. exposures/band 3 5
Objective spectroscopy Mode number N/A 2.5 N/A N/A
Dispersing element Prism
Resolving power 26–200
Field height (Y, arc sec) 25
Field width (X, arc sec) 25
Effective area (cm 2) 1200
Camera Mode number 1.6 a 2.6 a 3.6 b 4.6 b
Field height (Y, arc sec) 25 25 50 50
Field width (X, arc sec) 25 25 50 50
Effective area (cm 2) 830 1600 9400 7200

All six-reflection configurations, except where noted.

Effective spectral resolution element: 50 µm for MAMA detectors (2 pixels); 42 µm for CCD detector (2 pixels).
Resolving power values use 0.05 arc sec wide slit for bands 1 and 2 and 0.10 arc sec wide slit for bands 3 and 4.
a
Four-reflection configuration.
b
Lyot stop with grating wheel element.

Copyright © 2002 by Taylor & Francis

Figure 21 Diagram of the Hopkins Ultraviolet Telescope (HUT), which was flown on
two Shuttle Astro Spacelab missions, in 1990 and 1995. HUT was a telescope/spectro-
graph combination optimized for very short UV wavelengths (down to or below 90 nm).
(From Davidsen, 1993.)

E. Far-Ultraviolet Spectroscopic Explorer

The objectives of FUSE, like Copernicus, HUT, and IMAPS, are to study astro-
nomical objects in the very far UV wavelength range (below the 115 nm limit
of HST ), and to do so with high sensitivity (to reach faint objects) with both
high and moderate spectral resolutions (Sahnow et al., 1996). However, it has
the advantage of being a long-duration, dedicated mission (rather than a limited-
duration space shuttle investigation). FUSE, which was launched in June, 1999,
actually consists of four separate but coaligned telescope/spectrograph combina-
tions, with combinations of two wavelength ranges and two spectral resolution
capabilities (see Fig. 23). Two of the instruments use silicon carbide mirror coat-
ings, which provide the best reflectivity in the 90–100 nm wavelength range,
whereas the other two use LiF-overcoated aluminum coatings, which provide the
best reflectance in the 100–110 nm range. The actual wavelength ranges provide
overlapping coverage of the 90.5–118.7 nm wavelength range, with about 0.005
nm spectral resolution.
The use of two-dimensional, photon-counting microchannel plate detectors
and other technical improvements makes FUSE far more sensitive than the Co-
pernicus instrument, which had similar wavelength range and spectral resolution
capabilities.

Copyright © 2002 by Taylor & Francis

Figure 22 Diagram of Princeton University’s IMAPS and its detector system, flown
on sounding rockets and on two AstroSPAS Shuttle missions (in 1993 and 1996). (From
Jenkins et al., 1988.)

Copyright © 2002 by Taylor & Francis

Figure 23 Diagram of the FUSE satellite instrument, launched in June, 1999. As shown,
FUSE consists of four nearly identical telescope/spectrometer assemblies, with two pairs
having different mirror/grating coatings to optimize different portions of the 90–187 nm
spectral range. (Adapted from Moos et al. Ap J Lett 538(1): L1, 2000.)

Copyright © 2002 by Taylor & Francis

F. Extreme Ultraviolet Spectroscopy Missions
The extreme-UV spectral range (below about 90 nm wavelength) requires the
use of different optical technologies than do the longer UV wavelength ranges.
This is due to the extremely low normal-incidence reflectivities of most mirror
coatings. Therefore, as is also true for the x-ray spectral range, EUV systems
make use of grazing incidence reflective optical systems (see Fig. 24). However,
relatively recently, multilayer interference mirror coatings have been developed
that provide relatively high reflectivities with normal-incidence reflecting optics,
but only for very narrow, specifically selected wavelength ranges.
There have been numerous missions with instruments devoted, primarily
or in part, to EUV spectroscopy of the Sun. These have been primarily to observe
the very-high-temperature regions of the solar atmosphere; the corona, and active
regions (associated with solar flares). Temperature diagnostics can be obtained
from comparison of the spectral emissions due to different stages of ionization
of a single element (such as Fe or Si; see Fig. 25) but it is also important to have
imaging spectrographic capability, to map out variations with position on the Sun
or in its atmosphere. Observations in the EUV spectral range complement those
in the far-UV and x-ray wavelength ranges, by covering an intermediate range
of temperature.
The EUV is also characterized, for observations of objects outside the solar
system, by the fact that atomic hydrogen in the interstellar medium had been
thought to absorb nearly all of the radiation from celestial objects below the

Figure 24 Diagram of an extreme-ultraviolet (below 90 nm) telescope and spectrometer

flown on the EUVE satellite, launched in 1991.

Copyright © 2002 by Taylor & Francis

Figure 25 Variation of the relative degree of ionization with temperature, for (a) silicon
and (b) iron.

Copyright © 2002 by Taylor & Francis

Figure 26 Diagram of the expected opacity of the interstellar medium vs. wavelength,
in the far- and extreme-UV spectral ranges, with contributions of the predominant atoms
or ions indicated. (From R Cruddace, NRL.)

hydrogen ionization threshold wavelength (91.2 nm) (see Fig. 26). For this rea-
son, it was originally thought that there would be little to be seen in this wave-
length range from objects beyond our solar system.
However, early exploratory investigations, including an EUV instrument
carried on the Apollo-Soyuz mission in 1975 (Lampton et al., 1976), revealed
that in directions in space with exceptionally low hydrogen column densities,
very hot objects could be detected at EUV wavelengths. This led to the develop-
ment and launch of the Extreme Ultraviolet Explorer (EUVE) in 1992 (Bowyer,
1994). This satellite had instruments for both direct imaging and spectroscopy
of objects during the mission, which had an all-sky EUV survey as a major objec-
tive.

Copyright © 2002 by Taylor & Francis

 The EUVE sky survey verified that the distribution of absorbing interstellar
material was far less uniform than thought previously, and showed that observa-
tions in some directions could extend to very great distances, including some
extragalactic objects. Objects detected included not only very hot, white dwarf
stars, but also many late-type stars (similar to our Sun in surface temperature,
or even cooler) from which chromospheric and coronal EUV emissions (also
similar to those of our Sun) were observed. EUVE also observed solar system
planetary atmospheric emissions, and the Moon’s reflectance of solar EUV radia-
tion.

IV. EXAMPLES OF SCIENTIFIC RESULTS FROM UV

SPACE ASTRONOMY MISSIONS

We summarize here some of the new and unique results of UV space astronomy,
categorized by the type of object or measurement (rather than by mission or
instrument type). Although UV astronomical measurements have only been pos-
sible in the most recent half-century, compared to the four centuries since the
invention of the astronomical telescope, it is clear that the contributions to astro-
nomical knowledge from observations in this part of the electromagnetic spec-
trum have been significant.

A. Measurements of the Solar Atmosphere

and Solar Activity
The benefits of the availability of far- and extreme-ultraviolet spectroscopy in
solar physics include the ability to detect and measure the very-high-temperature
regions of the solar atmosphere, without need for a solar eclipse (which, in any
case, would allow observations of only those portions of the solar atmosphere
not covered by the Moon or a coronagraph disk).
In general, the characteristic temperatures of observable ionic species in-
crease as one goes toward shorter wavelengths, since the photons are more ener-
getic. However, there is also the advantage that emissions from the outer regions
of the solar atmosphere (the chromosphere and corona) are not obscured by the
(in the visible) far more intense radiation from the solar photosphere. The relative
intensities of different ionization stages of the same element (e.g., Fe ⫹, Fe 2⫹, and
Fe 3⫹) can be used to determine the temperature, since the degree of ionization
increases with temperature in a known fashion, as shown in Figure 25. In addition,
even in cooler regions (such as sunspots), far-UV spectroscopy can detect species
not observable at longer wavelengths; for example, the NRL HRTS instrument
first detected H 2 emission lines in sunspots in a 1975 sounding rocket flight.

Copyright © 2002 by Taylor & Francis

 The Doppler shifts of emission lines of known rest wavelengths can be
used to determine the line-of-sight velocities of gases being ejected from (or
falling back into) active regions on the solar surface. This, in combination with
methods for measuring the direction and magnitude of the local magnetic field,
has verified that motions of ionized gases near sunspots and other regions of
solar activity are guided by magnetic field lines in the solar atmosphere. Hence,
variations in the far-UV and EUV radiation from the Sun can serve to trace
variations in the magnetic field strength and configuration.
Although the far-UV and x-ray regions of the solar spectrum constitute
only a small percentage of the Sun’s total energy output, this portion (and its
time variation) has practical importance because it is responsible for the creation
of the ionosphere, the ionized component of Earth’s upper atmosphere that facili-
tates long-distance radio communication. It also is responsible for heating of
Earth’s upper neutral atmosphere. Variations in the upper atmospheric tempera-
ture produce corresponding changes in the atmospheric density at high altitudes
(which, in turn, affects atmospheric drag on satellites in low Earth orbit). There-
fore, it is of high practical importance to have instruments in space to monitor
the solar far-UV and x-ray outputs, and accurately measure any variations on
both short- and long-term time scales.
Figure 27 shows examples of solar far-UV spectra obtained with HRTS.
Since this is an imaging spectrograph with high spectral resolution, it can mea-
sure, simultaneously, variations in spectral intensity with wavelength and with
position on the solar surface. It also can measure velocities of emitting gas clouds,
by the Doppler shifts in their spectral images.
Figure 28 is a graph of the middle- and far-UV spectrum of the Sun ob-
tained with SUSIM. As mentioned previously, the short-wavelength parts of the
solar spectrum are much less bright than the near-UV and visible portions, but
also are much more variable with solar activity.

B. Measurements of the Temperatures and Luminosities

of Hot Stars
Before the advent of UV space astronomy, the temperatures of stars much hotter
than our Sun could be determined only by (a) comparing the spectral distribution
of the star’s (ground accessible) continuous spectrum with those of classical
‘‘black body’’ radiators or theoretical model atmosphere flux distributions (see
Fig. 2), after correcting for ‘‘reddening’’ of the starlight due to interstellar dust
extinction (if known); or (b) comparing the visible brightnesses of the star and
of any associated emission nebula (ionized hydrogen, or H II region) associated
with the star (the ionization of the nebula being attributed to extreme-UV emis-
sion from the star, at wavelengths shortward of the atomic hydrogen ionization

Copyright © 2002 by Taylor & Francis

(a)

(b)

Figure 27 (a) A far-UV spectral image (and direct image, in Hα, of the slit field of
view) of the Sun, obtained by NRL’s HRTS in a sounding rocket flight. (b) A more detailed
view of the region of the C IV doublet (154.8–155 nm) shows redshift (downflow) in a
sunspot region, near vertical center.

Copyright © 2002 by Taylor & Francis

Figure 28 A full-disk solar ultraviolet spectrum, obtained with NRL’s Solar Ultraviolet
Spectral Irradiance Monitor, during the Spacelab-2 Shuttle mission.

threshold of 91.2 nm). Such estimates were fraught with many uncertainties,
which could be circumvented only by more complete observations, particularly
of the ground-inaccessible far-UV photospheric spectrum.
Accurate determination of the total energy outputs of hot stars is important
because this determines the lifetimes of these stars, which are initially much more
massive than our Sun. This is also important for a better understanding of the
total galactic radiation environment, and its effects on the interstellar medium.
Ultraviolet spectroscopy from space has provided much new information
about the atmospheres and total luminosities of stars of all types (including cool
stars, with temperatures similar to or less than that of our Sun). The OAO-2,
Copernicus, and IUE satellites, with their long observing times, built upon early
sounding rocket investigations to increase greatly our knowledge of stellar atmo-
spheres and stellar evolution.
One complication in such measurements is that for stars lying behind sig-
nificant amounts of interstellar dust, the extinction due to this dust, and its wave-
length dependence, has to be taken into account before the UV brightnesses of

Copyright © 2002 by Taylor & Francis

he stars can be accurately determined. As discussed in the next section, UV
measurement of dust extinction is an astrophysical research objective in its own
right. The technique used is to compare two or more stars of spectral types as
nearly the same as possible, one of which is known (from ground-based observa-
tions) to lie behind significant amounts of interstellar dust, and the other behind
a negligible or very small column density, of dust.
One of the early results from rocket flights (Morton 1967; Carruthers et
al., 1984) (see Fig. 29), which was confirmed in more detail by the following
long-duration satellite missions, is that very hot and luminous stars have intense
‘‘stellar winds’’ in which matter is being ejected at much higher rates than the
‘‘solar wind’’—in some cases, at rates exceeding 10 ⫺6 solar masses per year.
This is revealed by the so-called ‘‘P-Cygni profile’’ of some of the strong spectral
lines (named for the star in which this type of line profile was first observed in
the ground-accessible wavelength range). It consists of ‘‘blue-shifted’’ absorption

Figure 29 Microdensitometer tracings of some of the stellar far-ultraviolet spectra ob-

tained by NRL in a sounding rocket flight (Fig. 13), showing absorption lines, and ‘‘P
Cygni’’ profiles (O VI, 1035 A; N V, 1235 A; Si IV, 1400 A, and C IV, 1550 A). Note
that the latter are prominent in the supergiant star ζ Ori, less prominent in the giant stars
δ Ori and ι Ori, and absent (replaced by pure absorption lines) in the main-sequence star
σ Ori. The Lyman-α (121.6 nm) interstellar absorption line due to atomic hydrogen is
prominent in all of these stellar spectra.

Copyright © 2002 by Taylor & Francis

features in combination with ‘‘red-shifted’’ emission lines. The interpretation of
this profile, as shown in Figure 30, is that the star is ejecting mass at high velocity
in all directions. Gas in the line of sight between us and the star is moving toward
us while also absorbing starlight, whereas gas moving away or at right angles to
the line of sight is responsible for the emission features. It is found that, for a
given spectral type of star, the mass loss is more extreme for highly luminous
giant or supergiant stars than for main-sequence (‘‘dwarf ’’) stars and, for a given
luminosity classification, increases with surface temperature. Resonance features
that most prominently exhibit this P-Cygni profile, in order of increasing tempera-
ture of the ejected gas, are C IV (154.8–155.1 nm), Si IV (139.4–140.3 nm),
N V (123.9–124.3 nm), and O VI (103.2–103.8 nm). These indicate that even
for these very hot stars (30,000–40,000 K effective temperatures), their outer
atmospheres must be very much hotter still (150,000–300,000 K). Also, it is clear
that these intense stellar winds significantly affect the evolution and total lifetimes
of these initially very massive stars (10 to more than 40 times the Sun’s mass).
Figure 31 shows Copernicus spectra of two hot stars, ζ Ophiuchi (spectral
type O9.5) and ζ Puppis (spectral type O5) (Morton, 1976) showing different

Figure 30 Diagram illustrates the geometry of ‘‘stellar winds’’ that produce the P-
Cygni profiles of spectral features. The gas moving toward the observer produces the
‘‘blue-shifted’’ absorption line component; the gas moving away from the observer (but
not hidden by the star itself ) produces the ‘‘red-shifted’’ emission line component.

Copyright © 2002 by Taylor & Francis

Figure 31 Copernicus spectra, with 0.2 Å (0.02 nm) resolution, of two stars in the
1200–1260 Å (120–126 nm) spectral range show in detail the ‘‘P Cygni’’ profile of the
N V spectral feature. The star ζ Puppis, of spectral type O5, shows a more prominent
outflow than the star ζ Ophiuchi, of spectral type O9.5. However, the interstellar absorp-
tion line of atomic hydrogen at 1216 Å (121.6 nm) is stronger in the latter star, which is
observed through a larger column density of interstellar gas. (From Morton, 1976.)

degrees of mass loss in the ‘‘P-Cygni profile’’ of the four-times-ionized nitrogen

(N V) spectral feature. The ‘‘rest’’ wavelengths of the two components of this
doublet resonance feature are indicated.

C. Measurements of the Composition and Properties

of the Interstellar Medium
Measurements of the interstellar medium, including both gaseous and solid (dust)
components, have benefitted greatly from the advent of UV astronomical mea-
surements. In particular, the resonance absorption lines of most of the common
gaseous components of interstellar gas, including atomic and molecular hydro-

Copyright © 2002 by Taylor & Francis

gen, helium, oxygen, nitrogen, and carbon, lie in the far UV. As mentioned previ-
ously, since the vast majority of these atoms or molecules reside in the ground
states (even in relatively ‘‘hot’’ regions, such as in emission nebulae), resonance
absorption measurements provide much more accurate measurements of the rela-
tive abundances and total column densities of these constituents of the interstellar
gas in the lines of sight to UV-bright stars. Also, access to shorter wavelengths
in the UV allows access to resonance lines of a given element in higher stages
of ionization.
Figure 32 shows some of the spectral line properties used in quantitative
measurements of interstellar gas column abundances. As shown, for very weak
absorption lines (low optical depth at line center), the spectral line equivalent
width is proportional to the column abundance. At the other extreme, where the
starlight is totally absorbed within the Doppler width of the line, the equivalent
width of the absorption line (including wings) is proportional to the square root
of the column density. For intermediate cases, the analysis is more complex, and
often involves comparing two or more spectral lines of the same specie having
different intrinsic absorption strengths.
The Doppler half-intensity width of an ‘‘optically thin’’ spectral line is
given by

√
T
∆λ 1/2 ⫽ 7.16 ⫻ 10 ⫺8 λ
M

Figure 32 Some of the considerations used in quantitative measurements of column

densities of gas, from absorption line spectroscopy.

Copyright © 2002 by Taylor & Francis

where T is the temperature (in kelvins (K )), M is the atomic or molecular mass,
and the wavelength λ is in nanometers.
For the case of atomic hydrogen, by far the most abundant constituent of
the interstellar gas, the square root proportionality of equivalent width to column
density, N, usually applies with high accuracy. The line equivalent width vs.
column density relationship (Morton, 1967) is given by:

√冢 冣
2πe 2 2g 1
Wλ ⫽ λf N ⫽ 7.31 ⫻ 10 ⫺11 √N nm
me c2 g2

where W λ is the equivalent width, f is the oscillator strength of the transition,

and g 1,2 are the statistical weights of the lower and upper levels.
In addition to measurements of the column abundances of various atoms,
molecules, and ions, high-resolution UV spectroscopy can also be used to deter-
mine the temperatures and velocities of interstellar gas clouds, by measuring the
Doppler broadening and Doppler shifts of the spectral lines (see Fig. 33).

Figure 33 Illustration of how high-resolution spectroscopy can be used to determine

the line-of-sight velocities, as well as column densities, of interstellar gas clouds.

Copyright © 2002 by Taylor & Francis

 The first far-UV measurements of atomic hydrogen, by Princeton Univer-
sity (Morton, 1967), and of molecular hydrogen, by NRL (Carruthers, 1970),
were obtained in sounding rocket flights. Among the early results of rocket UV
astronomy were the findings that, in many regions of space, the column densities
of atomic hydrogen were much lower than estimates from ground-based measure-
ments, such as atomic hydrogen 21 cm radio radiation.
These early results were followed up in much more detail with the launch
of the Copernicus satellite in 1972. Figure 31, the Copernicus spectra of 0.02
nm resolution, shows that the Lyman-α absorption line of atomic hydrogen at
121.6 nm in the spectrum of ζ Ophiuchi (which is known to be behind a signifi-
cant amount of interstellar material) is considerably stronger than that in the spec-
trum of the star ζ Puppis.
The Copernicus observations of interstellar hydrogen Lyman-α absorption
could also be extended to some much cooler stars (in which cases the interstellar
Lyman-α was seen superimposed on the stellar chromospheric Lyman-α emis-
sion line (DuPree et al., 1977; see Fig. 34)). Observations of these relatively
nearby stars indicate that the distribution of atomic hydrogen in the near-solar
neighborhood is unexpectedly nonuniform.
Figure 35 shows a higher-resolution Copernicus spectrum of ζ Ophiuchi,
at somewhat shorter wavelengths (Spitzer and Jenkins, 1975). The spectrum, with
a resolution of 0.005 nm, reveals interstellar absorption lines of molecular hydro-

Figure 34 Observations with the Copernicus satellite of the Lyman-α emission line of
the star α Aurigae (Capella). The spectra, with resolutions of 0.02 and 0.005 nm, show
the absorptions due to interstellar atomic hydrogen and deuterium superimposed on the
stellar line. (From DuPree et al., 1977.)

Copyright © 2002 by Taylor & Francis

Figure 35 A high-resolution (0.005 nm) Copernicus spectrum of the star ζ Ophiuchi
shows absorption features due to interstellar molecular hydrogen. Also present are absorp-
tions due to interstellar HD and Ar. (From Spitzer and Jenkins, 1975.)

gen. Analysis of these spectra indicates that the amounts of atomic and molecular
hydrogen in the line of sight to this star are very similar. By intercomparing the
individual H 2 lines (which indicate the degree of rotational excitation), one can
also determine the temperature of the gas, which in this case was found to be
about 80 K.
The Copernicus satellite made the first observations of deuterium (‘‘heavy
hydrogen’’) in the interstellar gas, in both the atomic form (Rogerson and York,
1976, DuPree et al., 1977; see Figs. 34, 36), and in the form of the molecule HD
(Fig. 35). The ratio of D to H in interstellar space is important to theories of the
origin and evolution of the universe, since D can be destroyed, but not created,
by thermonuclear reactions in stars.
Another unanticipated discovery was that five-times-ionized oxygen (O
VI), with resonance lines at 103.2 and 103.6 nm, is a significant component of
the low-density regions of the interstellar medium (Spitzer and Jenkins, 1975).
With a characteristic temperature of more than 2 ⫻ 10 6 K, this ‘‘coronal’’ gas
is thought to be heated by supernova explosions. The IUE satellite also observed
this ‘‘hot’’ phase of the interstellar medium (Kondo et al., 1989), detecting ab-
sorption lines of C IV, Si IV, and N V, not only toward stars in our galaxy but
also in the Magellanic Clouds. Intercomparisons of the strengths of these lines
in the same line of sight can provide more accurate measurements of the gas
temperature than measurements of only one line.

Copyright © 2002 by Taylor & Francis

Figure 36 High-resolution Copernicus spectra of interstellar atomic hydrogen Lyman-
β, γ, δ, and ε absorption lines toward two stars show the corresponding features due to
interstellar deuterium. (From York and J. Rogerson, 1976.)

Copyright © 2002 by Taylor & Francis

 The HST STIS instrument has extended studies of O VI absorption to inter-
galactic space; at large distances from our galaxy, the red-shift of spectral lines
due to expansion of the universe is sufficient to shift the O VI lines into the
wavelength range accessible by HST (Tripp et al., 2000). It was found that the
strengths of the O VI lines were not well correlated with atomic hydrogen
(Lyman-α or Lyman-β) absorptions at the same redshifts. Since the O VI can
only be excited by collisions with other hot gas particles (mostly ionized hydro-
gen), this shows that there is a significant amount of (otherwise invisible) ionized
hydrogen in intergalactic space.
Currently, the FUSE satellite is extending the observations shortward of
115 nm made by Copernicus to include much fainter and more distant stars (and
external galaxies), and observations made by HST to shorter wavelengths, for
studies of the interstellar and intergalactic media (see FUSE, 2000). Figure 37
is a spectrum of the central star of a planetary nebula, observed by FUSE, in the
91–99 nm wavelength range, with about 0.005 nm resolution. Among other early
results of FUSE are measurements of O VI absorption in the galactic halo (using
distant active galactic nuclei and quasistellar objects as light sources), and much
more extensive mapping of the distributions of H 2 and HD in our galaxy and in
the Magellanic Clouds.
The IMAPS instrument, although it has a relatively small collecting aper-
ture, has the highest spectral resolution (0.001 nm) of any instrument yet used
in the 95–110 nm spectral range. This has given it the ability to measure the
profiles and central wavelengths of interstellar absorption lines toward bright stars

Figure 37 A FUSE spectrum of the central star of a planetary nebula, K1-16, in the
91.2–99.2 nm spectral range. Numerous stellar and interstellar absorption lines, as well
as the P-Cygni O VI profile, are apparent. (From HW Moos et al., Ap J Lett 538 (1): L1,
2000.)

Copyright © 2002 by Taylor & Francis

Figure 38 IMAPS spectra of interstellar molecular hydrogen absorption lines toward
the star ζ Orionis, obtained in the AstroSPAS-1 shuttle mission. Three of the four marked
features are accompanied by stronger features due to the same transition, but redshifted
by 25 km/s. This indicates that two separate and discrete clouds with different tempera-
tures are responsible for the two component features. (From Jenkins and Peimbert, 1997.)

with high accuracy, from which temperature and velocity information, as well
as abundance information, on interstellar clouds in the line of sight can be ob-
tained (see Fig. 33, Fig. 38). It has been used to measure, with high accuracy,
the ratio of D to H in the interstellar medium (from higher members of the Lyman
line series), and has indicated that this ratio is variable (Jenkins et al., 1999).

Copyright © 2002 by Taylor & Francis

 Ultraviolet spectroscopy also gave new insights into the compositions and
properties of interstellar dust particles. Ground-based observations had shown
that interstellar dust ‘‘reddens’’ the light of stars seen through it: its effectiveness
in absorbing or scattering starlight increases toward shorter wavelengths. How-
ever, no spectral features gave clues to the compositions of the dust particles.
The advent of space-based UV measurements, beginning with a NASA Goddard
Space Flight Center sounding rocket flight (Stecher, 1965), revealed that interstel-
lar dust extinction vs. wavelength reached a maximum near 220 nm, beyond
which the extinction decreased with decreasing wavelength, down to about 160
nm, below which the extinction began to increase again toward shorter wave-
lengths (see Fig. 39).
Observations of interstellar dust extinction, beginning with the OAO-2 sat-
ellite (Code, 1972) and greatly extended by the IUE satellite, allowed many more
stars (including some in the Magellanic Clouds; see Koorneef and Code, 1981)
to be observed, and showed that the extinction vs. wavelength curve was highly
variable (Witt et al., 1984). It was also found that the variability of the extinction
curve with direction in space was much more marked at the shorter wavelengths
than in the wavelength range longward of the ‘‘220 nm peak,’’ as also indicated
in Figure 39. Theoretical and laboratory studies initially led to the ‘‘220 nm
bump’’ being attributed to graphite particles, but later studies indicated that this
is unlikely.
Observations of ultraviolet starlight reflected or scattered by interstellar
dust complement the observations of extinction (which has both true absorption
and scattering components) in providing more information about the nature, and
potentially the composition, of interstellar dust particles. It has been found that,
in at least some cases, the scattering efficiency (albedo) of dust grains has a
minimum near 220 nm and then rises toward shorter wavelengths (Lillie and Witt
1976; see Fig. 40).
Currently, there is still no consensus regarding the identity of the material
responsible for the 220 nm extinction peak, but it is generally presumed to be
carbon (or carbonaceous compounds) in some form. Both carbon and silicates
may be responsible, in varying proportions, for the far-UV extinction below 200
nm.
Measurements of the gaseous component of the interstellar medium also
shed light on the likely compositions of dust particles. It has been observed,
beginning with Copernicus observations and extended with HST HRS and STIS
observations, that in regions where there are significant concentrations of dust
and molecular hydrogen, the amounts of some of the gaseous components of the
interstellar medium (relative to the total hydrogen column density and relative
abundances in the solar atmosphere) are reduced. The elements most depleted
from the gas are those with high ‘‘condensation temperatures,’’ such as iron,
silicon, calcium, and titanium. That is, those most likely to form solid particles

Copyright © 2002 by Taylor & Francis

Figure 39 Typical extinction curves due to interstellar dust, covering the infrared
through ultraviolet spectral ranges, normalized to a ‘‘reddening’’ or color excess,
E(B-V), of one magnitude. Note the prominent peak in the middle-UV near 2200 A (220
nm), and the marked variations for different lines of sight in the far-UV range (below
200 nm).

at the low temperatures of the interstellar clouds (Savage and Sembach 1996;
see Fig. 41).

D. Studies of Extragalactic Objects and Cosmology

Observations of external galaxies in the ultraviolet allow comparisons of the dis-
tributions of hot stars with those of the much more common (and, in the visible,
in total much brighter) cooler stars. They also allow more accurate measurements
of the effective temperatures and total luminosities of individual objects, includ-

Copyright © 2002 by Taylor & Francis

Figure 40 Measurements of the reflectance, or albedo, and scattering phase function
asymmetry g, of interstellar dust particles, obtained with the Orbiting Astronomical Obser-
vatory-2 (OAO-2) narrow-band photometers. (From Lillie and Witt, 1976.)

ing active galactic nuclei and quasistellar objects (quasars). For sufficiently lumi-
nous individual stars or star groupings, the compositions, column densities, and
spatial distributions of interstellar gas (within the external galaxy) and of any
intergalactic gas in the line of sight to the galaxy can also be determined.
Since galaxies and quasars at very large distances from our galaxy display
redshifts of their radiation due to the expansion of the universe, it is possible to
observe radiation (with any given instrument) to shorter rest wavelengths in more
distant objects. This allows, for example, determinations of the distributions in
space of intergalactic gas clouds (for example, atomic hydrogen clouds in interga-
lactic space give rise to a ‘‘Lyman-α forest’’ of absorption lines, each localized
in redshift by its apparent wavelength). With a far-UV spectrograph (as on HUT
and FUSE ), it is possible to observe neutral helium (resonance line at 58.4 nm)
and even ionized helium (30.4 nm resonance line rest wavelength) in clouds of
sufficiently high redshift.

Copyright © 2002 by Taylor & Francis

Figure 41 Ratio of the abundances of elements in gaseous form in the interstellar me-
dium to that of hydrogen, and relative to the same ratio in the atmosphere of the Sun, as
observed with the HST HRS. Note that elements expected to be major constituents of
interstellar dust particles are under-represented in the interstellar gas, as they have been
depleted in the process of dust particle production. (From Savage and Sembach, 1996.)

The STIS instrument on HST, normally limited to wavelengths longward

of about 115 nm, has observed strong, redshifted O VI absorption lines (rest
wavelengths 103.2–103.7 nm) toward distant quasars (Tripp et al., 2000; see Fig.
42). In comparing these with hydrogen Lyman-α and Lyman-β absorption fea-
tures at these same redshifts, it was found that much of the O VI absorption was
occurring in regions of space with little corresponding H absorption. From this
it is inferred that (invisible) ionized hydrogen must be of higher than previously
known abundance in these very distant regions of intergalactic space, assuming
the O/H abundance ratio is uniform. Also, collisions with energetic protons are
the only practical means for creating the observed highly ionized oxygen.
Current and future observations with STIS and COS are likely to prove
valuable complements of observations with large ground-based telescopes, and
space-based infrared telescopes, in furthering our understanding of the very dis-
tant regions of our universe.

Copyright © 2002 by Taylor & Francis

Figure 42 HST STIS high-resolution spectra of redshifted O VI absorption lines, toward
the quasistellar object H1821 ⫹ 643. The two pairs of O VI lines are produced in inter-
galactic space, in clouds with redshifts less than that of the QSO. The amount of O VI
in this line of sight, not correlated with hydrogen Lyman α or Lyman β absorption, implies
a large amount of unseen, ionized hydrogen in intergalactic space. (From Tripp et al.,
2000.)

REFERENCES

Bartoe JDF, Brueckner GE. New stigmatic, coma-free, concave-grating spectrograph. J

Opt Soc Am 65: 13, 1975.
Bohlin RC. Copernicus observations of interstellar absorption in Lyman-Alpha. Astrophys
J 200: 402, 1975.
Bowyer CS. Astronomy and the Extreme Ultraviolet Explorer Satellite. Science 263: 55,
1994.
Bradford AP, Hass G, Osantowski JF, Toft AR. Preparation of mirror coatings for the
vacuum ultraviolet in a 2-m evaporator. Appl Optics 8: 1183, 1969.
Brandt JC, HRS Team. The high resolution spectrograph for the Space Telescope. In: The
Space Telescope Observatory, NASA CP-2244, 1982, p 76.
Brueckner GE, Bartoe JDF, Sandlin GD, VanHoosier ME. Lines of H 2 in extreme-
ultraviolet solar spectra. Nature 270: 326, 1977.
Brueckner G, Lean J, VanHoosier M. SUSIM—Solar Ultraviolet Spectral Irradiance Mon-
itor. Naval Research Laboratory report, 1991.
Carruthers GR. Magnetically focused image converters with internal reflecting optics. Rep
NRL Progr July: 7, 1965.
Carruthers GR. Rocket observation of interstellar molecular hydrogen. Astrophys J Lett,
161: L81, 1970.
Carruthers GR. Narrow-band filter for the Lyman-β wavelength region. Appl Optics 10:
1461, 1971.
Carruthers GR, Heckathorn HM, Opal CB. Stellar ultraviolet flux distributions in the 912–

Copyright © 2002 by Taylor & Francis

 1200 A wavelength range. In: Future of Ultraviolet Astronomy Based on Six Years
of IUE Research. NASA Conference Publication 2349, 1984, p 534.
Carruthers GR, Seeley TD. Global Imaging Monitor of the Ionosphere (GIMI): a far-
ultraviolet imaging experiment on ARGOS. In: Proc SPIE, Ultraviolet Atmospheric
and Space Remote Sensing: Methods and Instrumentation 2831: 65, 1996.
Code AD, ed. The Scientific Results From the Orbiting Astronomical Observatory (OAO-
2). NASA SP-310, 1972.
Davidsen AF, ed. The Hopkins Ultraviolet Telescope: Collected Scientific Papers, 1991–
1992. Baltimore: Johns Hopkins University, 1993.
Dupree AK, Baliunas SL, Shipman HL. Deuterium and hydrogen in the local interstellar
medium. Astrophys J 218: 361, 1977.
Far Ultraviolet Spectroscopic Explorer (FUSE) Special Issue. Astrophys J Lett, July 20:
538, 2000.
Green JC, Morse JA, Andrews JP, Wilkinson E, Siegmund OHW, Ebbetts, D. Performance
of the cosmic origins spectrograph for the Hubble Space Telescope. In Ultraviolet-
Optical Space Astronomy Beyond HST, Astronomical Society of the Pacific Con-
ference Series, 164: 176, 1999.
Harms RJ, FOS Teams. Astronomical capabilities of the Faint Object Spectrograph on
Space Telescope. In: The Space Telescope Observatory, NASA CP-2244, 1982,
p 55.
Jenkins EB, Joseph CL, Long D, Zucchino PM, Carruthers GR, Bottema M, Delamere
WA. IMAPS: high resolution echelle spectrograph to record far-UV spectra of stars
from sounding rockets. In: Proc SPIE, Ultraviolet Technology II. 932: 213, 1988.
Jenkins EB, Peimbert A. Molecular hydrogen in the direction of ζ Orionis A. Astrophys
J 477: 265, 1997.
Jenkins EB, Tripp TM, Wozniak PR, Sofia UJ, Sonneborn G. Spatial variability in the
ratio of interstellar atomic deuterium to hydrogen. I. Observations toward δ Orionis
by the Interstellar Medium Absorption Profile Spectrograph. Astrophys J 520: 182,
1999.
Keski-Kuha RAM, Osantowski JF, Herzig H, Gum JF, Toft AR. Normal incidence reflec-
tance of ion beam deposited SiC films in the EUV. Appl Optics 27: 2815, 1988.
Kondo Y, Boggess A, Maran SP. Astrophysical contributions of the International Ultravio-
let Explorer. Annu Rev Astron Astrophys 27: 397, 1989.
Koorneef J, Code AD. Ultraviolet interstellar extinction in the large Magellanic Cloud
using observations with the International Ultraviolet Explorer satellite. Astrophys
J, 247: 860, 1981.
Kurucz RL. Model atmospheres for G, F, A, B, and O stars. Astrophys J Suppl 40: 1,
1979.
Lampton M, Margon B, Paresce F, Stern R, Bowyer S. Discovery of a nonsolar extreme-
ultraviolet source. Astrophys J Lett 203: L71, 1976.
Lillie CF, Witt AN. Ultraviolet photometry from the Orbiting Astronomical Observatory.
XXV. Diffuse galactic light in the 1500–4200 A region and the scattering properties
of interstellar dust grains. Astrophys J 208: 64, 1976.
Morton DC. The far-ultraviolet spectra of six stars in Orion. Astrophys J, 147: 1017, 1967.
Morton DC. P Cygni profiles in Zeta Ophiuchi and Zeta Puppis. Astrophys J, 203: 386,
1976.

Copyright © 2002 by Taylor & Francis

NASA NP-126. Exploring the Universe with the Hubble Space Telescope. NASA–US
Government Printing Office.
Sahnow DJ, Friedman SD, Oegerle WR, Moos HW, Green JC, Siegmund OHW. Design
and predicted performance of the Far Ultraviolet Spectroscopic Explorer (FUSE).
Proc SPIE 2807, 1996.
Samson JAR. Techniques of Vacuum Ultraviolet Spectroscopy. New York: John Wiley &
Sons, 1967.
Savage BD, Sembach KR. Interstellar abundances from absorption-line observations with
the Hubble Space Telescope. Annu Rev Astron Astrophys 34: 279, 1996.
Shull JM, Beckwith S. Interstellar molecular hydrogen. Annu Rev Astron Astrophys 20:
163, 1982.
Smith AM. Rocket spectrographic observations of α Virginis. Astrophys J 156: 93, 1969.
Smith, AM. Rocket spectroscopy of ζ Orionis. Astrophys J 172: 129, 1972.
Sommer AH. Photoemissive Materials: Preparation, Properties, and Uses. New York: John
Wiley & Sons, 1968.
Spitzer L, Jenkins EB. Ultraviolet studies of the interstellar gas. Annu Rev Astron
Astrophys 13: 133, 1975.
Stecher TP. Interstellar extinction in the ultraviolet. Astrophys J 142: 1683, 1965.
Tripp TM, Savage BD, Jenkins EB. Intervening O VI quasar absorption systems at low
redshift: a significant baryon reservoir. Astrophys J Lett 534: L1, 2000.
Witt AN, Bohlin RC, Stecher TP. The variation of galactic interstellar extinction in the
ultraviolet. Astrophys J 279: 698, 1984.
Woodgate BE, Boggess A, Gull TR, Heap SR, Krueger VL, Maran SP, Melcher RW,
Rebar FJ, Vitagliano HD, Green RF, Wolff SC, Hutchings JB, Jenkins EB, Linsky
JL, Moos HW, Roesler F, Shine RA, Timothy JG, Weistrop DE, Bottema M, Meyer
W. Second generation spectrograph for the Hubble Space Telescope. In: Proc SPIE,
Instrumentation in Astronomy VI. 627: 350, 1986.
York DG, Rogerson JB. The abundance of deuterium relative to hydrogen in interstellar
space. Astrophys J 203: 378, 1976.

Copyright © 2002 by Taylor & Francis