Full download ebook at ebookgate.

com

The Quantum Physics of Atomic Frequency

Standards Recent Developments 1st Edition
Jacques Vanier (Author)

https://ebookgate.com/product/the-quantum-physics-
of-atomic-frequency-standards-recent-
developments-1st-edition-jacques-vanier-author/

Download more ebook from https://ebookgate.com

More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Quantum Mechanical Signal Processing and Spectral

Analysis Series in Atomic and Molecular Physics Series
in Atomic Molecular Physics 1st Edition Dzevad Belkic

https://ebookgate.com/product/quantum-mechanical-signal-
processing-and-spectral-analysis-series-in-atomic-and-molecular-
physics-series-in-atomic-molecular-physics-1st-edition-dzevad-
belkic/

Coffee Recent Developments 1st Edition Ronald Clarke

https://ebookgate.com/product/coffee-recent-developments-1st-
edition-ronald-clarke/

Physics of Atomic Nuclei 1st Edition Vladimir

Zelevinsky

https://ebookgate.com/product/physics-of-atomic-nuclei-1st-
edition-vladimir-zelevinsky/

Recent Developments In Bridge Engineering 1st Edition

K.M. Mahmoud (Editor)

https://ebookgate.com/product/recent-developments-in-bridge-
engineering-1st-edition-k-m-mahmoud-editor/
Quantum Aspects of Beam Physics Pisin Chen

https://ebookgate.com/product/quantum-aspects-of-beam-physics-
pisin-chen/

Fungal Biomolecules Sources Applications and Recent

Developments 1st Edition Vijai Kumar Gupta

https://ebookgate.com/product/fungal-biomolecules-sources-
applications-and-recent-developments-1st-edition-vijai-kumar-
gupta/

Advances in Atomic Molecular and Optical Physics Ennio

Arimondo

https://ebookgate.com/product/advances-in-atomic-molecular-and-
optical-physics-ennio-arimondo/

Formation and Logic of Quantum Mechanics 3 Volume Set

Vol I The Formation of Atomic Models Vol II The Way to
Quantum Mechanics Vol III The Establishment and Logic
of Quantum Mechanics 1st Edition Mituo Taketani
https://ebookgate.com/product/formation-and-logic-of-quantum-
mechanics-3-volume-set-vol-i-the-formation-of-atomic-models-vol-
ii-the-way-to-quantum-mechanics-vol-iii-the-establishment-and-
logic-of-quantum-mechanics-1st-edition-mitu/

Quantum Physics for Dummies 1st Edition Steven Holzner

https://ebookgate.com/product/quantum-physics-for-dummies-1st-
edition-steven-holzner/
 Physics
Vanier
Tomescu
The Quantum Physics of Atomic Frequency Standards: Recent
Developments covers advances in atomic frequency standards
(atomic clocks) from the last several decades. It explains the use of
various techniques, such as laser optical pumping, coherent population
trapping, laser cooling, and electromagnetic and optical trapping, in the
implementation of classical microwave and optical atomic frequency
standards.

The book first discusses improvements to conventional atomic frequency

standards, highlighting the main limitations of those frequency
standards and the physical basis of the limitations. It then describes
how advances in the theory and applications of atomic physics have
opened new avenues in frequency standards. The authors go on to
explore the research and development of new microwave and optical
frequency standards before presenting the results in frequency
stability and accuracy achieved with these new frequency standards.
They also illustrate the application of atomic clocks in metrology,
telecommunications, navigation, and other areas and give some insight
into future work.

Features
• Describes the basic physics, including quantum mechanics,
behind the operation of atomic clocks
• Explores new frequency standards that provide better stability
and accuracy than older, more conventional standards
• Discusses the importance of the field in the general context
of physics
• Gives an extensive list of the most important references in
the field

Building on the success of the previous two volumes, this up-to-date,

in-depth book examines the vast improvements to atomic clocks that
have occurred in the last 25 years. The improved stability and accuracy
enable the verification of physical concepts used in fundamental
theories, such as relativity, as well as the stability of fundamental
constants intrinsic to those theories.

K16759
ISBN: 978-1-4665-7695-7
90000

9 781466 576957

K16759_Cover_final.indd 1 6/17/15 11:46 AM

The Quantum
Physics of
Atomic
Frequency
Standards
Recent Developments
This page intentionally left blank
The Quantum
Physics of
Atomic
Frequency
Standards
Recent Developments

Jacques Vanier
Université de Montréal, Montréal, Canada
Cipriana Tomescu
Université de Montréal, Montréal, Canada
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2016 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Version Date: 20150617

International Standard Book Number-13: 978-1-4665-7697-1 (eBook - PDF)

This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information stor-
age or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copy-
right.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222
Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that pro-
vides licenses and registration for a variety of users. For organizations that have been granted a photo-
copy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Contents
Preface.................................................................................................................... xiii
Introduction ............................................................................................................xvii
Authors ....................................................................................................................xix

Chapter 1 Microwave Atomic Frequency Standards: Review and Recent

Developments .......................................................................................1
1.1 Classical Atomic Frequency Standards .....................................2
1.1.1 Cs Beam Frequency Standard ......................................2
1.1.1.1 Description of the Approach Using
Magnetic State Selection ..............................3
1.1.1.2 Review of Frequency Shifts and Accuracy .... 7
1.1.1.3 Frequency Stability of the Cs Beam
Standard ...................................................... 15
1.1.1.4 Recent Accomplishments ........................... 16
1.1.2 Hydrogen Maser ......................................................... 33
1.1.2.1 Active Hydrogen Maser .............................. 33
1.1.2.2 Passive Hydrogen Maser ............................. 48
1.1.2.3 Frequency Stability of the Hydrogen
Maser .......................................................... 53
1.1.2.4 State of the Art of Recent
Developments and Realizations .................. 57
1.1.3 Optically Pumped Rb Frequency Standards .............. 69
1.1.3.1 General Description .................................... 69
1.1.3.2 State-of-the-Art Development .................... 71
1.2 Other Atomic Microwave Frequency Standards ..................... 82
1.2.1 199Hg+ Ion Frequency Standard .................................. 83
1.2.1.1 General Description .................................... 83
1.2.1.2 Frequency Shifts ......................................... 85
1.2.1.3 Linear Trap ................................................. 88
1.2.2 Other Ions in a Paul Trap ...........................................90
1.2.2.1 171Yb+ and 173Yb+ Ion Microwave
Frequency Standards .................................. 91
1.2.2.2 201Hg+ Ion Microwave Frequency
Standard ...................................................92
1.3 On the Limits of Classical Microwave Atomic Frequency
Standards ................................................................................. 93
Appendix 1.A: Formula for Second-Order Doppler Shift..................94
Appendix 1.B: Phase Shift between the Arms of Ramsey Cavity ......95

v
vi Contents

Appendix 1.C: Square Wave Frequency Modulation

and Frequency Shifts .......................................................................... 95
Appendix 1.D: Ring Cavity Phase Shift ............................................97
Appendix 1.E: Magnetron Cavity ...................................................... 98

Chapter 2 Recent Advances in Atomic Physics That Have Impact

on Atomic Frequency Standards Development ................................ 101
2.1 Solid-State Diode Laser......................................................... 102
2.1.1 Basic Principle of Operation of a Laser Diode ..........102
2.1.2 Basic Characteristics of the Semiconductor
Laser Diode .............................................................. 105
2.1.3 Types of Laser Diodes .............................................. 106
2.1.4 Other Types of Lasers Used in Special Situations.....108
2.2 Control of Wavelength and Spectral Width
of Laser Diodes .................................................................. 109
2.2.1 Line Width Reduction .............................................. 109
2.2.1.1 Simple Optical Feedback .......................... 109
2.2.1.2 Extended Cavity Approach ....................... 109
2.2.1.3 Feedback from High-Q Optical Cavities ...112
2.2.1.4 Electrical Feedback .................................. 112
2.2.1.5 Other Approaches ..................................... 112
2.2.1.6 Locking the Laser to an Ultra-Stable
Cavity ........................................................ 113
2.2.2 Laser Frequency Stabilization Using an Atomic
Resonance Line ........................................................ 116
2.2.2.1 Locking the Laser Frequency to Linear
Optical Absorption ................................... 116
2.2.2.2 Locking the Laser Frequency
to Saturated Absorption ............................ 117
2.3 Laser Optical Pumping.......................................................... 119
2.3.1 Rate Equations ......................................................... 120
2.3.2 Field Equation and Coherence ................................. 122
2.4 Coherent Population Trapping ............................................... 127
2.4.1 Physics of the CPT Phenomenon ............................. 129
2.4.2 Basic Equations ........................................................ 131
2.5 Laser Cooling of Atoms ........................................................ 136
2.5.1 Atom–Radiation Interaction ..................................... 138
2.5.1.1 Effect of a Photon on Atom External
Properties: Semi-Classical Approach ....... 138
2.5.1.2 Quantum Mechanical Approach............... 143
2.5.2 Effect of Fluctuations in Laser Cooling
and Its Limit ............................................................. 158
2.5.3 Cooling below Doppler Limit: Sisyphus Cooling .... 160
2.5.3.1 Physics of Sisyphus Cooling ..................... 160
2.5.3.2 Capture Velocity ....................................... 164
Contents vii

2.5.3.3 Friction Coefficient ................................... 165

2.5.3.4 Cooling Limit Temperature ...................... 166
2.5.3.5 Recoil Limit .............................................. 166
2.5.3.6 Sub-Recoil Cooling .................................. 167
2.5.4 Magneto-Optical Trap .............................................. 167
2.5.5 Other Experimental Techniques in Laser
Cooling and Trapping............................................... 170
2.5.5.1 Laser Atom-Slowing Using a
Frequency Swept Laser System:
Chirp Laser Slowing ................................. 171
2.5.5.2 Laser Atom-Slowing Using Zeeman
Effect: Zeeman Slower ............................. 173
2.5.5.3 2D Magneto-Optical Trap ........................ 177
2.5.5.4 Isotropic Cooling ......................................180
2.5.5.5 Optical Lattice Approach ......................... 183
Appendix 2.A: Laser Cooling—Energy Considerations.................. 189

Chapter 3 Microwave Frequency Standards Using New Physics ..................... 191

3.1 Cs Beam Frequency Standard ............................................... 192
3.1.1 Optically Pumped Cs Beam Frequency Standard.... 192
3.1.1.1 General Description .................................. 192
3.1.1.2 Frequency Shifts and Accuracy ................ 194
3.1.1.3 Experimental Determination
of Those Shifts .......................................... 197
3.1.1.4 Frequency Stability ................................... 198
3.1.1.5 Field Application ......................................200
3.1.2 CPT Approach in a Beam ......................................200
3.1.2.1 General Description ..................................200
3.1.2.2 Analysis .................................................... 201
3.1.2.3 Experimental Results ................................206
3.1.3 Classical Cs Beam Standard Using Beam
Cooling .............................................................. 208
3.2 Atomic Fountain Approach ................................................... 210
3.2.1 In Search of a Solution ............................................. 210
3.2.2 General Description of the Cs Fountain................... 211
3.2.3 Functioning of the Cs Fountain ................................ 213
3.2.3.1 Formation of the Cooled Atomic
Cloud: Zone A .......................................... 213
3.2.3.2 Preparation of the Atoms: Zone B ............ 217
3.2.3.3 Interrogation Region: Zone C ................... 218
3.2.3.4 Free Motion: Zone D ................................ 218
3.2.3.5 Detection Region: Zone E......................... 218
3.2.4 Physical Construction of the Cs Fountain ................ 219
3.2.4.1 Vacuum Chamber ..................................... 219
3.2.4.2 Microwave Cavity ..................................... 220
viii Contents

3.2.4.3 Magnetic Field .......................................... 221

3.2.4.4 Temperature Control ................................. 221
3.2.4.5 Capture and Selection Zone...................... 221
3.2.4.6 Detection Zone ......................................... 221
3.2.4.7 Supporting Systems .................................. 221
3.2.4.8 Advantages and Disadvantages
of a Pulsed Fountain ................................. 222
3.2.5 Frequency Stability of the Cs Fountain.................... 223
3.2.5.1 Photon Shot Noise.....................................224
3.2.5.2 Quantum Projection Noise........................ 225
3.2.5.3 Electronic Noise........................................ 225
3.2.5.4 Reference Oscillator Noise: Dicke Effect ... 225
3.2.6 Rubidium and Dual Species Fountain Clock ........... 226
3.2.7 Frequency Shifts and Biases Present in the
Fountain.................................................................... 229
3.2.7.1 Second-Order Zeeman Shift ..................... 230
3.2.7.2 Black Body Radiation Shift ...................... 232
3.2.7.3 Collision Shift ........................................... 237
3.2.7.4 Cavity Phase Shift ....................................240
3.2.7.5 Cavity Pulling ........................................... 242
3.2.7.6 Microwave Spectral Purity ....................... 247
3.2.7.7 Microwave Leakage .................................. 247
3.2.7.8 Relativistic Effects ....................................248
3.2.7.9 Other Shifts............................................... 249
3.2.7.10 Conclusion on Frequency Shifts
and Accuracy ............................................ 250
3.2.8 An Alternative Cold Caesium Frequency
Standard: The Continuous Fountain ........................ 251
3.2.8.1 Light Trap ................................................. 252
3.2.8.2 Interrogation Zone, Microwave Cavity..... 253
3.2.8.3 Preliminary Results .................................. 255
3.2.9 Cold Atom PHARAO Cs Space Clock .................... 257
3.3 Isotropic Cooling Approach .................................................. 258
3.3.1 External Cavity Approach: CHARLI ...................... 258
3.3.2 Approach Integrating Reflecting Sphere
and Microwave Cavity: HORACE ...........................260
3.3.3 Different HORACE Approach ................................. 261
3.4 Room Temperature Rb Standard Approach Using Laser
Optical Pumping.................................................................... 262
3.4.1 Contrast, Line Width, and Light Shift ..................... 263
3.4.2 Effect of Laser Radiation Beam Shape .................... 272
3.4.3 Expectations Relative to Short-Term Frequency
Stability .................................................................... 273
3.4.4 Review of Experimental Results on Signal Size,
Line Width, and Frequency Stability ....................... 273
Contents ix

3.4.5 Frequency Shifts....................................................... 278

3.4.5.1 Buffer Gas Shift ........................................ 278
3.4.5.2 Magnetic Field Shift ................................. 279
3.4.5.3 Light Shift ................................................. 279
3.4.5.4 Spin-Exchange Frequency Shift ...............284
3.4.5.5 Microwave Power Shift............................. 285
3.4.5.6 Cavity Pulling ........................................... 286
3.4.6 Impact of Laser Noise and Instability on Clock
Frequency Stability .................................................. 287
3.4.6.1 Spectral Width, Phase Noise, and
Intensity Noise of Laser Diodes ............... 288
3.4.6.2 Impact of Laser Noise on Clock
Short-Term Frequency Stability ................290
3.4.6.3 Medium- and Long-Term Frequency
Stability ..................................................... 295
3.4.7 Other Approaches Using Laser Optical Pumping
with a Sealed Cell..................................................... 297
3.4.7.1 Maser Approach........................................ 297
3.4.7.2 Laser Pulsing Approach............................ 297
3.4.7.3 Wall-Coated Cell Approach ..................... 299
3.5 CPT Approach .......................................................................300
3.5.1 Sealed Cell with a Buffer Gas in Continuous
Mode: Passive Frequency Standard..........................300
3.5.1.1 Signal Amplitude and Line Width ............302
3.5.1.2 Practical Implementation and Its
Characteristics ..........................................307
3.5.2 Active Approach in a Cell: The CPT Maser ............ 315
3.5.2.1 Basic CPT Maser Theory ......................... 315
3.5.2.2 Frequency Stability ................................... 318
3.5.2.3 Frequency Shifts ....................................... 320
3.5.3 Techniques for Improving S/N Ratio in the
Passive IOP and CPT Clock Approach ....................322
3.5.4 CPT in Laser-Cooled Ensemble for Realizing a
Frequency Standard.................................................. 323
3.6 Laser-Cooled Microwave Ion Clocks .................................... 324
3.6.1 9Be+ 303 MHz Radio-Frequency Standard .............. 325
3.6.2 113Cd+ and 111Cd+ Ion Trap ........................................ 327
3.6.3 171Yb+ Laser-Cooled Microwave Frequency
Standard ................................................................... 328
Appendix 3.A: Frequency Stability of an Atomic Fountain ............ 329
3.A.1 Shot Noise ................................................................ 333
3.A.2 Quantum Projection Noise ....................................... 334
Appendix 3.B: Cold Collisions and Scattering Length .................... 337
Appendix 3.C: Optical Absorption of Polarized Laser Radiation
Including Optical Pumping .............................................................. 338
Appendix 3.D: Basic CPT Maser Theory ........................................ 341
x Contents

Chapter 4 Optical Frequency Standards ........................................................... 345

4.1 Early Approach Using Absorption Cells ............................... 347
4.2 Some Basic Ideas ................................................................... 349
4.3 MOT Approach...................................................................... 351
4.4 Single Ion Optical Clocks ...................................................... 352
4.4.1 The Concept ............................................................. 352
4.4.2 Outline of Particular Implementations with
Individual Ions.......................................................... 357
4.4.2.1 27Al+ (I = 5/2) ............................................ 357
4.4.2.2 40Ca+ (I = 0) and 43Ca+ (I = 7/2) ............... 359
4.4.2.3 87Sr + (I = 9/2) and 88Sr+ (I = 0) ................. 361
4.4.2.4 115In+ (I = 9/2) ........................................... 362
4.4.2.5 137Ba+ (I = 3/2) and 138Ba+ (I = 0) ............. 363
4.4.2.6 171Yb+ (I = 1/2), 172Yb+ (I = 0),
and 173Yb+ (I = 5/2) ...................................364
4.4.2.7 198 Hg+ (I = 0) and 199Hg+ (I = 1/2) ............. 366
4.4.3 Systematic Frequency Shifts in Single Ion Clocks ....366
4.4.3.1 Doppler Effect........................................... 366
4.4.3.2 Zeeman Effect .......................................... 368
4.4.3.3 Biases due to the Presence of Electric
Fields ......................................................... 371
4.5 Optical Lattice Neutral Atoms Clock .................................... 377
4.5.1 The Concept ............................................................. 377
4.5.1.1 Trapping Characteristics ........................... 382
4.5.1.2 Atom Recoil .............................................. 383
4.5.1.3 Atom Localization .................................... 383
4.5.1.4 Magic Wavelength .................................... 384
4.5.1.5 Clock Transition........................................ 385
4.5.2 Type of Atoms Used in Optical Lattice Clocks ....... 386
4.5.2.1 Strontium Atom ........................................ 386
4.5.2.2 Mercury Atom .......................................... 387
4.5.2.3 Ytterbium Atom ........................................ 389
4.5.2.4 Magnesium Atom ..................................... 390
4.5.2.5 Calcium Atom ........................................... 391
4.5.3 Important Frequency Biases..................................... 391
4.5.3.1 Zeeman Effect .......................................... 391
4.5.3.2 BBR Shift.................................................. 392
4.5.3.3 Lattice Light Shift..................................... 393
4.5.3.4 Other Shifts............................................... 394
4.5.4 Frequency Stability of an Optical Lattice Clock...... 395
4.5.5 Practical Realizations ............................................... 395
4.6 Frequency Measurement of Optical Clocks .......................... 397
4.6.1 Optical Comb ........................................................... 398
4.6.2 Clock Frequencies and Frequency Stabilities
Realized .................................................................... 399
Contents xi

Chapter 5 Summary, Conclusion, and Reflections............................................ 401

5.1 Accuracy and Frequency Stability ........................................402
5.2 Selected Applications of Atomic Frequency Standards ........404
5.2.1 The SI: Towards a Redefinition of the Second .........405
5.2.2 Tests of Fundamental Physical Laws .......................407
5.2.2.1 Fundamental Constants ............................407
5.2.2.2 Time Dilation and Gravitational
Red Shift ...................................................408
5.2.2.3 Fundamental Physics in Space..................409
5.2.3 Clocks for Astronomy and Earth Science ................ 410
5.2.3.1 VLBI and Geodesy ................................... 410
5.2.3.2 Deep Space Network ................................ 410
5.2.3.3 Earth Clocks Network .............................. 410
5.2.3.4 Navigation Systems................................... 411
5.3 Last Reflections ..................................................................... 412

References ............................................................................................................. 415

Index ...................................................................................................................... 457
This page intentionally left blank
Preface
Volumes 1 and 2 of The Quantum Physics of Atomic Frequency Standards, henceforth
referred to as QPAFS (1989), were written in the 1980s and were published in 1989.
They covered, in some detail, work done up to 1987 on the development of atomic
frequency standards. The text included a description of their development at that
time, as well as a description of the research on the physics supporting that devel-
opment. Since that time, the field has remained a very active part of the research
program of many national laboratories and institutes. Work has remained intensive
in many sectors connected to the refinement of classical frequency standards based
on atoms such as rubidium (Rb), caesium (Cs), hydrogen (H), and selected ions in the
microwave range, while new projects were started in connection to the realization of
stable and accurate frequency standards in the optical range.
For example, intensive studies were made on the use of lasers in the optical pump-
ing and cooling of Rb and Cs as well as on the development of a new type of standard
based on the quantum-mechanical phenomenon called coherent population trapping
(CPT). Regarding Cs and Rb, laser cooling of atoms has made possible the realiza-
tion of an old dream in which a small blob of atoms, cooled in the microkelvin range,
is projected upward at a slow speed in the gravitational field of the earth and the
atoms fall back like water droplets in a fountain. In their path, the atoms are made
to pass through a microwave cavity, and upon falling back after having spent their
kinetic energy, pass through the same cavity, mimicking, with a single cavity, the
classical double-arm Ramsey cavity approach. The system is called atomic fountain.
Its advantage over the classical approach resides in the reduction of the width of the
resonance hyperfine line by a factor of the order of 100 relative to that observed in
the room temperature approach. The resulting line width is of the order of 1 Hz.
Work has also continued on the development of smaller H masers, in particular in
the development of passive devices and in the use of a new smaller so-called magne-
tron cavity. The advent of the solid-state laser in the form of the conventional edge-
emitting type (GaAs) and vertical structure (VCSEL) has opened the door to a new
approach in optical pumping for implementing smaller and more performing Rb and
Cs cell frequency standards.
Since the 1990s, laser cooling has been studied extensively and aside from pro-
viding a means for realizing the fountain clock mentioned above, it has allowed the
realization of clocks based on microwave transitions in ions such as mercury (Hg+),
barium (Ba+), strontium (Sr+), and ytterbium (Yb+) confined within an electromag-
netic trap.
On the other hand, intense work has been carried out in several laboratories in
extending the work done at microwave frequencies to the optical frequency range. The
gain in that approach relies mainly on the increase in the frequency of the atomic transi-
tions involved, which provides for a line width similar to that obtained in the microwave
range a resonance quality factor millions of times larger. Laser cooling has been applied
successfully to such atoms as mercury (Hg), ytterbium (Yb), and strontium (Sr)

xiii
xiv Preface

stored in optical lattice traps in order to reduce their thermal motion. Laser cooling
has also been used in the mono-ion trap to implement optical frequency standards. In
that case, a single ion, say Sr+ or Yb+, is maintained in a Paul or Penning trap and its
motion within the trap is damped by laser cooling. Clocks at optical frequencies have
been implemented as laboratory units with unsurpassed accuracy and frequency sta-
bility reaching the 10−16 to 10−18 range. In both cases, the clock frequency is derived
from a transition between the ground S state of the atom and an excited metastable
state with a lifetime of the order of 1 second or more leading to a very narrow
resonance line. The clock transition is detected by means of monitoring changes in
the fluorescence level created by the cooling radiation when the clock transition is
excited.
The large gap in frequency between the microwave and the optical range has
always been an roadblock in the use of optical frequencies in various applications
such as frequency standards or still high precision spectroscopy and fundamental
research. The reason is mainly due to the fact that gaps between available optical
frequencies for the realization of clocks are very large. It is extremely difficult to
connect those frequencies to the microwave range. This connection is required
because most of the applications are in the low frequency range of the spectrum and,
furthermore, because the SI (International System of Units) definition of the second
is based on a microwave hyperfine transition in Cs, in the X band. We have given in
Volume 2 of QPAFS examples of the conventional method used to make that connec-
tion. That method comprises frequency- and phase-locking together banks of lasers
with appropriate heterodyning in several steps in order to interconnect various opti-
cal frequencies to reach finally the microwave range. The connection has to be done
over a large number of steps and involves tremendous investment of space and time
to finally measure what very often happens to be just a single frequency. Such a task
has been reduced considerably by the invention of the so-called optical comb, which
comprises locking the repetition rate of a femtosecond laser to a stable atomic fre-
quency standard of high spectral purity, such as an H maser referenced in frequency
to a primary Cs atomic clock. When observed by means of a nonlinear optical fibre,
the resulting laser spectrum consists of a spectrum of sharp lines, themselves called
the teeth of the comb, which covers a frequency range of the order of 1 octave.
Frequencies over a broad range are then measured essentially in a single step on an
optical table, resulting in a considerable reduction in work and size as compared to
the previous heterodyning technique, which required entire rooms filled with lasers.
This volume covers those subjects in some detail. It is divided into five chapters.
Chapter 1 is an introduction, presenting a review of recent developments made on
the improvement of conventional atomic frequency standards described in the two
volumes of QPAFS. It highlights the main limitations of those frequency standards
and the physical basis of those limitations and outlines the progress made during the
last 25 years. Chapter 2 is a description of recent advances in atomic physics, theory
and applications, that opened new avenues. Chapter 3 is concerned with research
and development done in the development of new microwave frequency standards.
Chapter 4 describes research and development done in the optical range to implement
optical frequency standards based on new results in atomic physics as described in
Chapter 2. Chapter 5 summarizes the results in frequency stability and accuracy
Preface xv

achieved with those new frequency standards and outlines selected applications.
A short reflection is included giving some insight into future work.
Such a text cannot be written without significant help from experts in the field.
We wish to recognize the contribution and collaboration of many scientists. In particu-
lar, we wish to recognize the invaluable help of André Clairon, who has read the whole
manuscript and helped in improving its exactness and completeness. We also show our
gratitude to the following scientists who helped us through their encouragement, sup-
plied original figures or material, and contributed by means of comments on various sec-
tions of the text: C. Affolderbach, A. Bauch, S. Bize, J. Camparo, C. Cohen-Tannoudji,
E. De Clercq, A. Godone, D. Goujon, S. Guérandel, P. Laurent, T. Lee, S. Micalizio,
G. Mileti, J. Morel, W.D. Phillips, P. Rochat, P. Thomann, R.F.C. Vessot, and S. Weyers.

Jacques Vanier
and
Cipriana Tomescu
University of Montreal
This page intentionally left blank
Introduction
This book is about recent developments in the field of atomic frequency standards,
developments that took place after the publication in 1989 of the first two volumes
with the same title. Atomic frequency standards are systems providing an electri-
cal signal at a cardinal frequency of, say, 10 MHz, a signal generated usually by
a quartz crystal oscillator locked in phase or in frequency to a quantum transition
inside an atom. The atom is selected for its properties such as easy detection of the
particular quantum transition chosen and relative independence of its frequency of
the environment. In early work, those conditions limited development around hydro-
gen and alkali atoms, which have transitions in the microwave range and could be
manipulated easily as beams or atomic vapour with the techniques available at that
time. Progress in the development of lasers and their stabilization extended that
work to the optical range. A major task encountered in the early development of
microwave standards has always been the elimination of Doppler effect. Atoms at
room temperature travel at speeds of several hundred metres per second and, conse-
quently, Doppler effect causes frequency shifts and line broadening of the resonance
signal. This effect is generally eliminated by means of various storage techniques
based on Dicke effect, or still beam techniques using the Ramsey double-arm cavity
approach. These techniques are not well adapted to optical frequencies because of
the shorter wavelengths involved. However, progress in the understanding of inter-
actions between atoms and electromagnetic interactions has provided new means
of reducing the velocity of atoms and reducing, if not eliminating, the constraints
introduced by Doppler effect.
An atomic frequency standard that is operated continuously becomes an atomic
clock. The operation is essentially a process of integration and the date set as the
constant of integration provides the basis for implementing a timescale. This is the
origin of atomic timescales, in particular the one maintained by the International
Bureau of Weights and Measures. Various systems in operation have their own
timescale, for example, the global positioning system (GPS) of the United States,
the Russian Glonass system, the Chinese Beidou system, and the European Galileo
systems under development, all playing an important role in navigation on or near
the surface of the earth.
Although time is central to physics and is used in our day-to-day life, it is a con-
cept that is difficult to grasp, let alone to define. We use it without questioning its
origin and its exact nature. It is basic in physics for describing the dynamics of sys-
tems and ensembles of systems by means of equations that model the evolution of
objects forming our universe. The concept is used as such without questioning much
its exact nature and origin. In Newtonian mechanics, objects evolve in space and
their behaviour is described by means of differential equations and functions of time
and space. Both space and time are independent and in common language they are
said to be absolute. In that context, time is not a function of space and space is not a
function of time. However, in attempts to relate mechanics and electromagnetism by

xvii
xviii Introduction

space and time transformations, a difficulty arose. This is due to the finiteness and
invariability of the speed of light, made explicit in Maxwell’s equations, whatever
the motion of the frame of reference in which it is generated and measured. In this
context, with Einstein, Poincaré, Lorentz, Minkowski, and others, time and space
become entangled and functions of each other. There is no such thing as an absolute
space in which objects evolve in an absolute time framework, both independent of
each other. Time and space form a single four-dimensional framework and cannot be
treated independently. This concept forms the basis of the theory of relativity. This
theory has been shown to be valid through multiple experiments and verifications to
a level that raises its validity to a high degree. It should be pointed out that the most
accurate verifications were done with atomic clocks, the instruments that are the
content of this book. There is another question also often raised regarding the nature
of time: Could it be discrete? If so what would be the size of its smallest quantity,
the time quantum? Could it be that Planck’s time is the smallest time entity? This
is a totally unknown subject and appears to be a roadblock to in the development of
a sustainable quantum theory that includes the concepts elaborated in the theory of
general relativity.
Although we may feel somewhat uncomfortable in the context of such questions,
time remains the most basic concept in physics, is fundamental, and is the quantity
that is measured with the greatest precision. Current atomic clocks can commonly
keep time to an accuracy of 1 s in a million years, or in other words are stable to
better than 1 ms in a year. For example, the timescale generated by the GPS satellites
for navigation, based on atomic frequency standards on satellites and on ground,
is stable after appropriate processing and filtering to about 1 ns/day. On the other
hand, on the basis of our inability to measure time by astronomical means with such
accuracy, it was decided in 1967 to replace the astronomical definition of the second
by one in terms of a particular atomic hyperfine transition in the Cs atom. The fre-
quency of that transition is set at 9,192,631,770 Hz. Furthermore, since now the speed
of light is defined exactly as 299,792,458 m/s, providing at the same time a definition
of the metre, the mechanical units of the SI become essentially determined by the
basic time unit, the second. The concept of unifying all SI units in terms of a single
quantity goes further due to the Josephson effect phenomenon, which relates voltage
to frequency in a most fundamental expression, 2e/ℏ, involving only fundamental
constants. This is the subject that will be described in Chapter 5.
From this discussion, it is evident that time plays a most important role in physics
and technology and the realization of the highest accuracy and precision of the SI
unit, the second, has remained one of the most active preoccupations of several labo-
ratories and institutes over the past 50 years. Starting with tremendous improvements
in the realization of the second within the microwave range, work has extended to
the optical range with proven increase in frequency stability and accuracy by several
orders of magnitude. These achievements were possible mainly through a better
understanding of the interactions between electromagnetic radiation and atoms, pro-
viding a means of altering the properties of atoms. This book is about those improve-
ments that have taken place mainly during the past 25 years, on the realization of
stable and accurate frequency standards.
Authors
Jacques Vanier completed his undergraduate studies in
physics at the University of Montreal, Québec, Canada,
before moving to McGill University to undertake his gradu-
ate studies. During his career he has worked in various
industries (Varian, Hewlett-Packard); taught physics; and
carried out research at Laval University, Montreal, Québec,
Canada, and has also been an active member of the National
Research Council of Canada, in Ottawa, Ontario, Canada.
His research work is oriented towards the understanding
and the application of the quantum electronics phenomena
and he has been a consultant for several companies engaged in the development of
atomic clocks. Jacques has also been very active on the academic circuit, giving lec-
tures and presenting at numerous conferences in universities, national institutes, and
summer schools around the world. He has written more than 120 journal articles and
proceedings papers and is the author of review articles and books on masers, lasers,
and atomic clocks. His book The Quantum Physics of Atomic Frequency Standards,
written with C. Audoin, is recognized as a main reference in the field. He is the author
of The Universe: A Challenge to the Mind published by Imperial College Press/
World Scientific. Jacques is a fellow of the Royal Society of Canada, the American
Physical Society, and the Institute of Electrical and Electronic Engineers. He has
received several awards for his contributions to the field of measurement science. He
is currently an adjunct professor in the Physics Department, University of Montreal,
Québec, Canada.

Cipriana Tomescu completed her studies in physics at

the University of Bucharest, Romania, where she obtained
her PhD degree.
From 1982 to 2004, she was a researcher at the National
Institute of Laser Physics, Plasma and Radiation, Bucharest,
Romania. In the early years of her employment, she par-
ticipated in the construction of H masers used by the
Bucharest Observatory, the Institute of Metrology, and
the Faculty of Physics. During the period 1996–2004, she
was laboratory director. During the period 1992–2006, she
also worked in various national laboratories, in particular,
Paris Observatory, LNE-SYRTE, France; Neuchâtel Observatory, Switzerland; and
Communication Research Laboratory, Japan. At those locations, she contributed to
the development of advanced state-of-the-art atomic frequency standards, such as Rb
and Cs fountains using atom trapping techniques and laser atom cooling. From 2008
to 2012, she worked at the University of Liege, IPNAS, and at Gillam-Fei. She was
responsible for the implementation of the first H maser realized in Belgium under

xix
xx Authors

Plan Marshall: SKYWIN-TELECOM. She is the author of numerous publications

in scientific journals and conference proceedings and she has been invited to make
presentations at numerous symposia, universities, and national institutes. In 1985, she
received the D. Hurmuzescu Prize of the Romanian Academy for work on the physics
of the H maser. She is currently an invited researcher in the Physics Department of the
University of Montreal, Québec, Canada.
 1 Microwave Atomic
Frequency Standards
Review and Recent
Developments

At the end of the 1980s, atomic frequency standards reached a level of refinement
that made it the envy of many other fields of physics. The accuracy of primary cae-
sium (Cs) standards maintained in operation at national institutes reached a level
better that 10−13 and the frequency stability of the hydrogen (H) maser in the medium
term was better than 10−14. These characteristics made possible the verification to
great accuracy of basic physics predictions such as those resulting from the theory of
relativity and made possible the maintenance of a timescale to an unsurpassed stabil-
ity. It also opened the use of such devices in many applications. The time unit, the
second, became the most accurate unit of the International System of Units (SI), with
consequences for the implementation of other units such as the metre, the volt, and
the ohm. On the other hand, Rb standards had reached a level of development that
made them an excellent support of digital communication systems with improved
reliability and also made them appropriate for navigation systems using satellites
requiring medium frequency stability and small size.
There has been extensive research on the possibility of using other atoms as the basis
for new types of frequency standards. However, those systems are still under study in lab-
oratories; Cs, H, and Rb therefore remain the atoms at the heart of atomic frequency stan-
dards used at large either as references in basic research or in practical systems requiring
precise and accurate timing. Although the Cs standard in its original beam implemen-
tation using magnetic state selection has been dethroned as the most accurate primary
standard with the introduction of optical pumping and laser cooling, it still remains in
many laboratories the work horse for implementing a local timescale, for confirming the
accuracy of other standards, and, to a limited extent, for reliable reporting to the BIPM
(Bureau International des Poids et Mesures) in the maintenance of the second.
In this chapter, we recall the physical construction and the characteristics of
those frequency standards based on Cs, H, and Rb, as well as of some selected other
types of microwave frequency standards, which still show promise regarding pos-
sible specific applications. We examine the physics at the heart of the operation of
those standards and behind their limitations relative to size, accuracy, and frequency
stability. We also see that those limitations were overcome to some extent, showing
that, with some imagination, improvements could still be made on instruments that
had already attained a very high level of maturity.

1
2 The Quantum Physics of Atomic Frequency Standards

1.1 CLASSICAL ATOMIC FREQUENCY STANDARDS

We usually group Cs beam frequency standards, H masers, and optically pumped
Rb standards under the terminology “classical atomic frequency standards.” In the
following paragraphs, we review their physical construction and recall the essen-
tial theoretical results that were developed in parallel with their implementation.
Theoretical investigations were required for an understanding of the various phe-
nomena causing biases observed when evaluated relative to accuracy and frequency
stability. The reader will find in Volumes 1 and 2 of The Quantum Physics of Atomic
Frequency Standards (QPAFS) a detailed description of the operation of such stan-
dards and a description of the basic physics involved. In the following sections, we
recall the main concepts behind their operation in order to simplify reading of sub-
sequent sections, in which we discuss recent progress in understanding the physics
involved. We then present new analysis and realizations that have resulted in better
understanding of their operation, greater accuracy, better frequency stability, and in
some instances reduction in size and weight.

1.1.1 CS BEAM FREQUENCY STANDARD

A frequency standard using Cs and the separate oscillatory field approach proposed
by Ramsey (Ramsey 1950) was implemented as early as 1955 (Essen and Parry
1955). Intense laboratory and industrial development followed (see, e.g., McCoubrey
1996). Development showed great success and soon after the construction of Cs
primary standards in several laboratories, the frequency of the Cs ground state
hyperfine transition was adopted for the definition of the second by the Conférence
Générale des Poids et Mesures (CGPM, 1967–1968). The frequency adopted was
9,192,631,770 Hz. It was the best number obtained by means of precise astronomi-
cal measurements by which the Cs hyperfine frequency was determined relative to
the second, whose formal definition at the time of measurement was the ephemeris
second based on astronomical observation (Markowitz et al. 1958). That choice has
remained till date (2015).
Why was Cs selected for providing the basis of the definition of the second? First,
the choice of the Cs atom in the implementation of a frequency standard has resulted
from the considerable accumulation of knowledge on that atom over the years and
from the several advantages that it provides over other candidates. In particular, Cs
has a single stable isotope, 133Cs, and is relatively abundant in nature. Its melting
point is 28.4°C. Its vapour pressure is such that it is possible to implement a rather
intense atomic beam from an oven at a relatively low temperature of the order of 425
to 500 K. Its ionization energy is low, 3.9 eV, making it easy to detect by conventional
procedures such as ionization with a hot wire detector and ion counting. Finally, Cs
has a ground state hyperfine frequency falling in the X band, a microwave region
that has known extensive development, which makes possible atom–microwave
interaction by means of small structures such as cavities whose dimensions are in
the centimetre range.
The Cs atom has a nuclear spin I = 7/2 and has a single s electron outside closed
electronic shells. Its ground state consists of two hyperfine levels F = 3 and F = 4
Microwave Atomic Frequency Standards 3

2.0 4
3
E/hνhf 2
1
1.5 0
−1
−2
−3
1.0

F=4 0.5

0.0

F = 3 −0.5

−1.0
−4
−3
−2
−1.5 −1
0
1
2
−2.0 3
0.0 0.2 0.4 0.6 0.8 1.0
B (T )

FIGURE 1.1 Ground state energy level manifold of the caesium atom as a function of the
magnetic induction B in tesla.

and in a low magnetic field the structure consists of two manifolds of 7 and 9 energy
levels, respectively. This ground state is shown in Figure 1.1 as a function of the
magnetic induction B.

1.1.1.1 Description of the Approach Using Magnetic State Selection

A conceptual diagram of the classical Cs beam frequency standard using magnetic
state selection is shown in Figure 1.2 (Vanier and Audoin 2005).
A beam of Cs atoms is formed by proper collimation from an oven heated at a
temperature of the order of 50–100°C depending on the intensity required. This
beam is directed as to pass through a so-called Ramsey cavity that provides a region
of electromagnetic interaction and excite transitions between the two ground state
hyperfine levels mF = 0 of the atoms. Magnets A and B are generally dipole magnets
and create an intense inhomogeneous field in which atomic trajectories are deflected.
They are called Stern–Gerlach selector magnets or filters. The deflection is caused
by the interaction of the atom magnetic moment with the magnetic field gradient
and by the tendency of atoms to seek states of low potential energy. Consequently,
according to Figure 1.1, atoms having higher energy in high magnetic fields are
deflected to regions of low magnetic field in order to lower their potential energy.
Similarly, those atoms having lower energies at high magnetic fields seek regions
of high magnetic field for the same reason. Selection is accomplished by means of
magnet A whose orientation is such as to force atoms in level F = 4, mF = 0 to pass
through the Ramsey microwave cavity, and reach the second deflecting magnet B.
Atoms in the other F = 3, mF = 0 level, being deflected away from the F = 4, mF = 0
atoms, are eliminated from the beam by appropriate collimation. The analysis of the
4 The Quantum Physics of Atomic Frequency Standards

Modulation,
synchronous detector,
Microwave generator
and frequency lock
system

Hot wire
detector Electron
multiplier
Ramsey cavity
Cs oven
Signal out
Bo
Signal out
Magnet A Magnet B
state selector Magnetic shields analyzer

FIGURE 1.2 Simplified conceptual diagram of the Cs beam frequency standard using magnetic
state selection. The inset shows the shape of the resonant signal observed when the frequency-
lock loop is open and the microwave frequency is scanned slowly over the atomic hyperfine
resonance. Although in the figure the magnetic induction is shown parallel to the beam direc-
tion, in practice it is very often made perpendicular to the beam. (Data from Vanier, J. and
Audoin, C., Metrologia, 42, S31, 2005. Copyright Bureau International des Poids et Mesures.
Reproduced by kind permission of IOP Publishing. All rights reserved.)

beam composition is done by the combination of magnet B, called the analyzer, and
a hot wire ionizer followed by a counter usually assisted by an electron multiplier. In
their transit through the Ramsey cavity, the atoms are submitted to an electromag-
netic field of angular frequency ω in the two arms of the cavity called the interaction
regions. In the first arm of the cavity, atoms are excited into a Rabi oscillation that
puts them into a quantum superposition of the two hyperfine levels F = 4, mF = 0
and F = 3, mF = 0 of the ground state. We define τ, the time of transit of an atom at
speed v inside that first arm of length l. The power fed into the cavity is adjusted to
such a value as to make the electromagnetic radiation appear as a π/2 pulse, that is to
say a microwave pulse that puts the atoms in an exact superposition state of the two
hyperfine levels when they exit that first arm at the most probable speed. The atoms
are subsequently left to drift unperturbed in the space within the double arm cavity.
A uniform magnetic induction Bo provides an axis of quantization and the atoms
remain in the same state. They then penetrate inside the second arm at distance L
from the first arm. We call T the time of transit between the two arms of the cavity.
If v is the speed of a given atom, then T is simply L/v and is affected the spread in v.
In that second arm, the atoms are again submitted to a field of the same intensity and
same frequency as in the first arm. Atoms having the same speed as in the first arm
are, thus, submitted again to a π/2 pulse. The atoms at the exit of the second arm find
themselves in the lower state F = 3, mF = 0 and the transition is complete as if they
had been submitted to a π pulse. If the frequency applied to the cavity is not exactly
the resonance frequency of the atoms, the phase of the field in the second arm is not
Microwave Atomic Frequency Standards 5

coherent with the phase of the magnetic moment of the entering atoms and interference
takes place. The total effect of the radiation inside the two arms is smaller than that
of a π pulse and the transition is not complete. The probability of the transition is
not a maximum and upon detection some atoms appear as having made the transi-
tion to the lower state and some not. The actual state of affairs is analyzed by means
of a second Stern–Gerlach filter that selects only atoms having made the transition
and orients them towards the detector. That detector is made of a hot wire of tung-
sten or other material that ionizes the atoms reaching it. The resulting ion current is
measured either directly (usual in laboratory standards) or by means of an electron
multiplier (usual in commercial standards). If the frequency of the radiation fed to
the cavity is swept over a certain range, a kind of interference pattern is observed at
the output of the detector as shown in the inset of Figure 1.2. It is worth mentioning
that the role of atoms in level F = 4, mF = 0 and of atoms in level F = 3, mF = 0 can
be inverted without affecting the operation of the system.
This type of signal is called Ramsey fringes (Ramsey 1956). A complete calcula-
tion of the shape of that signal is done in QPAFS, Volume 2 (QPAFS 1989).
Since the central peak, being the resonance peak, is the signal of interest we con-
centrate on it. For a given atomic velocity v leading to a transit time τ in each arm
of the cavity, the shape of the central fringe signal can be represented approximately
by the following equation (for ease in reading, we import into the present text a few
important equations from QPAFS 1989):

1 2
P (τ) = sin bτ 1 + cos ( ΩoT + φ )  | Ωo| << b (1.1)
2
where:
τ is the time of interaction of the atom with the microwave field in each interac-
tion region
T is the time spent by the atom between the interaction regions
ϕ is the phase difference that exists between the microwave fields in the two arms
of the cavity, including the effect of asymmetries and cavity losses
Ωο is the difference between ω, the angular frequency of the microwave radia-
tion in the Ramsey cavity, and ωo, the resonance angular frequency of the
atom:

Ωo = ω − ωo (1.2)

The parameter b is the Rabi angular frequency in the interaction region and is a
measure of the amplitude of the microwave induction Bmw. It is defined by the equa-
tion (our definition of b is different from that used by Ramsey (1956) by a factor
of 2, consistent with the notation used in QPAFS 1989):
µB
b= Bmw (1.3)

where:
μB is the Bohr magneton
ℏ is the Planck’s constant h over 2π
6 The Quantum Physics of Atomic Frequency Standards

Assuming ϕ = 0 or π, the full width at half maximum of the central fringe as given
by Equation 1.1 is readily calculated as:
π
W= (1.4)
T
If the resonance signal was observed with a single cavity (Rabi resonance), its line
width would be π/τ. With the double arm cavity the width of the signal is reduced by
the factor, L/l = T/τ, which could be very large in laboratory units designed with a
large distance L between the two arms of the Ramsey cavity of individual lengths l.
In those units, L/l may be in practice of the order of 100 or more.
We note that Equation 1.1 is valid under the assumption that | Ωo | << b. A better
approximation to the central fringe is obtained by means of a first-order expansion of
the full Ramsey fringe equation as developed in QPAFS, Volume 2 (QPAFS 1989):

1 2   2Ω o  1 
P (τ) = sin bτ 1 + cos ( ΩoT + Φ ) −   tan 2 bτ sin ( Ω0T + Φ )  (1.5)
2   b  
Equation 1.1, however, is an excellent approximation of the central fringe shape and
is used in most calculations concerned with frequency shifts introduced by various
phenomena. The third term in Equation 1.5 introduces a correction that amounts in
some cases to a few percent of the biases calculated and is sometimes used for better
precision in the evaluation of various effects (Makdissi and de Clercq 2001).
The above calculation was done under the assumption that the atoms in the beam
have all the same velocity and spend the same time in the two arms of the cavity. In
practice, the beam is composed of atoms travelling at thermal velocities. In a gas,
atomic velocities are spread according to a Maxwell distribution. However, in the
case of a collimated beam and state selection by magnets, this distribution is greatly
altered. If f(τ) is the resulting distribution of interaction times τ in each arm of the
cavity, then an average of the probability P over this distribution must be made:
∞

P=
∫ f (τ)P(τ)d τ
0
(1.6)

where the following relations between speed v and interaction time distributions hold:
∞ ∞

∫ f (τ)d τ = ∫ p(v)dv =1
0 0
(1.7)

l l
f (τ) = p (1.8)
τ2  v 

The fringe pattern is smeared out to some extent by the velocity spread. It turns out,
however, that the central fringe is not much affected by the averaging, if the velocity
distribution is made sufficiently narrow as is done in some implementation (Becker
1976). A typical experimental result is shown in Figure 1.3 for two scan widths.
Microwave Atomic Frequency Standards 7

1.6 kHz 40 Hz

(a) (b)

FIGURE 1.3 Ramsey fringes observed experimentally in the NRC Cs VI standard. (Data
from Mungall, A.G. et al., Metrologia, 17, 123, 1981. Copyright Bureau International des
Poids et Mesures. Reproduced by permission of IOP Publishing. All rights reserved.)

A practical implementation of a Cs standard consists of frequency locking to the

atomic resonance the frequency of the quartz crystal oscillator that is used to gener-
ate the microwave radiation that induces the atomic transitions in the Ramsey cavity
as illustrated in Figure 1.2. In that system, the frequency of the crystal oscillator is
modulated at a low frequency with a modulation depth less than the line width of
the central Ramsey fringe. The resulting signal obtained by means of synchronous
detection is a discriminator pattern that is used in a feedback loop to lock the fre-
quency of the crystal oscillator to the hyperfine resonance line.
As is evident from our earlier discussions, the atoms are relatively free in the
beam. There are still present, however, some physical phenomena that cause small
frequency shifts or biases. One of the main tasks in the implementation of a primary
frequency standard of the type just described is the precise evaluation of those shifts
or biases. It is only after such an evaluation that a given standard may be accepted as
a representation of the SI unit, the second, which is the main goal in national stan-
dards laboratory implementations of such units.
Those shifts may be separated into three main groups: those intrinsic to atomic
properties, those introduced in the detection of the resonance signal, and those intro-
duced in the locking of the microwave generator to the resonance line. We outline
the nature of these shifts. In Section 1.1.1.4 we show how progress was made in
recent years in the accuracy of their evaluation and sometimes in their reduction.

1.1.1.2 Review of Frequency Shifts and Accuracy

1.1.1.2.1 Frequency Shifts (Biases) Associated with Physical Atomic Properties
1.1.1.2.1.1 Magnetic Field Shift In a low magnetic field, the shift of the resonant
line of interest with the applied magnetic field is given by the equation (QPAFS 1989,
Volume 1, Table 1.1.7):
ν = ν hf + 427.45 × 108 Bo2 (1.9)
where:
νhf is the unperturbed hyperfine frequency defined as 9,192,631,770 Hz
Bo is the value of the applied magnetic induction in Tesla
8 The Quantum Physics of Atomic Frequency Standards

The field applied may be of the order of 50 to 100 × 10−7 Tesla (50 to 100 mG) and
the displacement of the resonance peak is several parts in 1010. This is the most
important shift in the frequency standard and must be determined with an accuracy
compatible with the accuracy desired in the final evaluation. It is obvious that field
fluctuations must be minimized for reasons of frequency stability. This constraint
forces the use of a rather stable current source supplying the device that creates the
field. This device may be rods or a solenoid. On the other hand, efficient magnetic
shielding is used to prevent environment field fluctuations from reaching the region
of interaction, the Ramsey cavity. This is generally done by means of several lay-
ers of mu-metal or high permeability metallic cylinders surrounding the region of
interaction.

1.1.1.2.1.2 Second-Order Doppler Effect This shift originates from the time
dilation phenomenon of special relativity. For an atom at speed v, the second-order
Doppler effect frequency shift ΔνD2 is given by the equation (QPAFS 1989, Volume 1)

∆ν D2 v2
=− 2 (1.10)
ν hf 2c

where:
c is the speed of light

In the beam, the velocities are spread over a relatively large range and this shift must
be averaged over the velocity distribution. The average transition probability is given
by Equation 1.6. Since the various shifts to be considered are expected to be small,
we may consider them as independent of each other. When the second-order Doppler
shift alone is taken into account, then the transition probability at the exit of the sec-
ond selector magnet is given by:
∞
1    v2   
P=
2 ∫0
f (τ)sin 2 bτ 1 + cos ω − ωo  1 − 2   T  d τ
   2c   
(1.11)

This expression may be used directly by proper adjustment and normalization to obtain
the intensity of the beam reaching the detector. We may, thus, write the beam intensity as:
∞
1    v2   
I = Ib +
2 ∫
I o f (τ)sin 2 bτ 1 + cos ω − ωo  1 − 2
0    2c
  T  dτ
  
(1.12)

where:
Ib is the background atomic flux reaching the detector, which, for example, is
composed of atoms in a wrong state and do not contribute to the signal

The resonance frequency of the standard is identified with the maximum of the
central Ramsey fringe signal, which is simply the maximum of the beam intensity
I reaching the detector. The value of that frequency is obtained by differentiating
Equation 1.12 with respect to ω (see Appendix 1.A). Hence, we obtain:
Microwave Atomic Frequency Standards 9

ωD − ωo
=−
∫0
v2T 2 f (τ)sin 2 b τ d τ
∞
(1.13)
ωo
2c 2
∫0
T 2 f (τ)sin 2 b τ d τ

where:
ωD is the frequency of the maximum of the central Ramsey fringe

In order to evaluate the resulting shift, one needs to know the interaction time dis-
tribution function f(τ) or the speed distribution. In the 1960s, in the field of atomic
beam resonance spectroscopy, a Maxwellian distribution of atomic velocities in the
beam was assumed (Harrach 1966, 1967). With magnets as state selectors, the distri-
bution may be very different from a Maxwell’s distribution. In early developments of
Cs primary standards, it was assumed that the distribution was Maxwellian, but with
low and high speeds cutoff (Mungall 1971). It was then verified by numerical analy-
sis that the velocity spectrum with its chosen width and cutoff frequencies, using
Equation 1.6, reproduces the observed Ramsey fringe pattern to a good approxima-
tion. However, such an approach is rather empirical. We now outline more advanced
methods for evaluating f(τ) and recent developments that have allowed the evaluation
of velocity-dependent shifts to a very satisfying accuracy.

1.1.1.2.1.3 Black Body Radiation This effect is caused by an interaction of

the atoms with the oscillating electric field of the ambient thermal radiation. At
an operating temperature of 300 K the shift using polarizability values reported in
Table 1.1.8 of QPAFS (1989) is calculated to be −1.69 × 10−14 and varies as the fourth
power of the absolute temperature (Itano et al. 1981). The effect was not important
in the evaluation of Cs standards implemented before 1990, due to their limited accu-
racy and frequency stability. However, the shift was observed at room temperature in
a Cs beam standard with greatly improved stability (Bauch and Schröder 1997). The
measurement was made as a function of temperature over a range of more than 150°.
The value found was 1.66 × 10−14 and was in good agreement with the theoretical
prediction. The shift became important in the context of the accuracy reached with
the Cs fountain to be described later. Furthermore, the polarizability value used in its
theoretical evaluation was questionable in the light of new measurements (Micalizio
et al. 2004). We return to this point in Chapter 3 dealing with the Cs fountain.

1.1.1.2.1.4 Spin–Exchange Frequency Shift Collisions between atoms travelling

at different velocities within the beam and with atoms in the background vapour
pressure may cause an exchange of their electrons. This is called spin exchange and
it creates a frequency shift proportional to the collision rate (QPAFS 1989). The colli-
sion rate is proportional to the relative velocities of the atoms and their collision cross
section. The value of this cross section is not known for Cs at room temperature and the
effect, although expected to be small, still needs to be evaluated. The effect is small in
Cs beam standards operating at room temperature but as will be seen in the case of the
Cs fountain it becomes important particularly when accuracy reaches the 10−16 level.
10 The Quantum Physics of Atomic Frequency Standards

1.1.1.2.2 Shifts Introduced by the Resonance Detection System

1.1.1.2.2.1 Phase Shift between the Two Cavities If the phase ϕ between the
fields in the two arms forming the Ramsey cavity differs from either 0 or π by a small
amount ϕ, then the central fringe maximum or minimum is displaced. This phase
shift may be caused by an asymmetry in the cavity construction or electrical losses
in the waveguide creating a travelling wave within the structure.
For atoms at speed v travelling the distance L between the arms in time T = L/v, it
is readily shown from Equation 1.1 that a phase shift ϕ between the fields of the two
arms creates a frequency shift of the resonance line by the amount:

φ
∆ν φ = − (1.14)
2πT
For example, an asymmetry to the extent of 10−4 m between the lengths of the two
arms of the Ramsey cavity may cause a frequency shift of the order of 10−13 depend-
ing on the electrical losses of the waveguide used. The frequency shift changes sign
upon reversal of the velocity. It can thus be determined experimentally by reversing
the direction of the beam. However, due to the presence of a phase shift that varies
with position in the cavity, called generally distributed phase shift, and due to the fact
that the beam does not necessarily retrace the same path upon reversal, the accuracy of
determination of this shift is limited. In short commercial instruments, the frequency
shift is larger since T is smaller than in laboratory standards. It may reach 1 × 10−12.
We have provided in QPAFS, Volume 2 (QPAFS 1989), a detailed analysis of the
effect of this shift. Let us recall the main points of that analysis. The equations to be
used are Equations 1.1 and 1.6. Converting to beam intensity, we obtain:
∞
1
2 ∫ { }
I = Ib + Io f ( τ) sin 2 bτ 1 + cos ( ω − ωo ) T + φ  d τ (1.15)
0

We emphasize the difference between this type of bias and that introduced by the
second-order Doppler effect. The second-order Doppler shift affects directly the fre-
quency of the atoms while the phase shift is introduced through the cavity and is con-
sidered as a step in the time evolution of the atoms. To obtain the frequency where
beam intensity I is maximum, we need to do as in the case of second-order Doppler
effect, that is, differentiate with respect to ω and set ∂I/∂ω = 0 (see Appendix 1.B).
Using the relation T = L/v, the result is:
∞

∫ (1 v) f (τ)sin bτ d τ
2

φ
ωφ − ωo = − ∞
0
(1.16)
L
∫ (1 v ) f (τ)sin bτ d τ
2 2

An important consideration follows from that analysis: the shift as measured from
the maximum of the Ramsey pattern is a function of the velocity distribution and
depends on the value of b, thus on the power fed to the cavity. This is the same thing
Microwave Atomic Frequency Standards 11

as in the case of second order Doppler effect, as calculated through Equation 1.13.
We now return to this point and see how recent advances in understanding the phase
shift effect has resulted in its reduction by means of a new type of cavity.

1.1.1.2.2.2 Cavity Pulling The cavity tuning influences the position of the reso-
nance maximum. The effect is small due to the fact that the cavity Q is low resulting
in weak stimulated emission in the cavity since the number of atoms in interaction is
small. In short, Cs beam frequency standards where the resonance is less selective,
the effect may be significant. However, in laboratory standards where the cavity Q
is intentionally made small and the atomic gain is low, this shift is generally under
control and does not cause a problem. For example, for a cavity Q of 500, a line Q
of 1.5 × 108 corresponding to a line width of 60 Hz as realized in the best standards
shown in tables provided hereunder, a detuning of the cavity by 1 MHz would pro-
duce a fractional frequency shift of 6 × 10−15. Full details of the calculation are given
in QPAFS, Volume 2 (1989). On the other hand, it may be mentioned that cavity
detuning may introduce another frequency shift when square wave modulation of the
microwave interrogating frequency is used. The effect appears because the fields at
the two frequencies resulting from the modulation may not have the same amplitude
if the cavity is not tuned exactly to atomic resonance.

1.1.1.2.2.3 Bloch–Siegert Effect The microwave magnetic induction in the cav-

ity may be thought of as linearly polarized radiation. A linearly polarized field may
be decomposed into two counter-rotating fields. In the rotating frame approach, one
component is seen as resonant with the atomic ensemble and the other rotating in
counter direction is seen as having twice the resonance frequency. An elementary
analysis shows that a frequency shift is introduced in the detection of the resonance
frequency by this off resonance component. This shift is called the Bloch–Siegert
effect (Bloch and Siegert 1940). It is proportional to the ratio l/L of the beam tube and
in laboratory standards of large size it is of the order of 5 × 10−15.

1.1.1.2.2.4 Majorana Transitions If the constant magnetic field along the atomic
beam is inhomogeneous, transitions of random nature can be caused between mF
sublevels of the two manifolds F = 3 and F = 4. These are called Majorana transi-
tions (Majorana 1932). It has been shown that these transitions can cause a shift of
the resonance frequency of the central ΔmF = 0 transition (Ramsey 1956). Since in
the classical approach permanent magnets are used for state selection and detec-
tion, it is possible that stray inhomogeneous fields created by those magnets excite
Majorana transitions with a resultant frequency shift. This effect is absent in opti-
cally pumped beam tubes in which no selector magnets are used and where the mag-
netic field can be made very homogeneous all along the beam path.

1.1.1.2.2.5 Rabi and Ramsey Frequency Pulling This is an effect that is partly
inherent to the atoms and partly introduced by the technique of detection of the reso-
nance. A shift is introduced by the overlapping of the symmetrically situated field
dependent Rabi pedestals with the central fringe of the ΔF = 1, ΔmF = 0 resonance
line (De Marchi et al. 1984; De Marchi 1987). When these pedestals have different
amplitudes, which is the case for magnetic state selection, a small distortion of the
12 The Quantum Physics of Atomic Frequency Standards

central fringe is created by the tails of the field dependent Rabi pedestals, causing a
frequency shift of the central fringe. Furthermore, the microwave field in the cavity
may contain a small perpendicular component causing transitions ΔF = 1, ΔmF = ±1
that are connected to the resonant transition of interest (the ΔmF = 0 transition) by a
common energy level. These transitions may also distort the central fringe and cause
a small frequency shift. This is called Ramsey pulling (Cutler et al. 1991). These
shifts are a function of beam design and depend to some extent on the microwave
power applied to detect the resonance. The effect is function of the applied mag-
netic field and is small when the resonance line is narrow. Consequently, these effects
are much reduced in laboratory standards and in general are small. A considerable
amount of theoretical analysis has been made on these effects (Shirley et al. 1995;
Lee et al. 2003).

1.1.1.2.2.6 Microwave Leakage An undesired microwave field may be pres-

ent all around the microwave cavity when microwave leakage occurs. The spuri-
ous field may originate from the cavity holes that let the atomic beam through or
from unwanted small spacing between different parts of the cavity assembly causing
microwave leaks. It may also originate from electrical feedthroughs. The atomic
beam may then be subjected to a travelling wave in a place where no microwave field
should be present. A Doppler frequency shift then occurs. A model of this effect
has been presented, which accounts for the frequency shifts that may be observed
(Boussert et al. 1998). Since the details of the field configuration are generally not
known, it is rather difficult to evaluate the shift introduced and only careful design
of the system avoiding leakage can minimize the effect.

1.1.1.2.2.7 Gravitational Effect According to the general theory of relativity, a

clock rate is a function of the gravitational potential at the location of the clock.
Consequently, the atomic Cs standard frequency is a function of its altitude in the
earth’s field. Since the laboratory Cs beam frequency standards used in the determi-
nation of the SI unit second are located at different altitudes, it is thus important to
precisely determine the actual altitudes of these standards relative to the geoid and
make the appropriate correction. This shift is small. In the earth’s field, it varies with
altitude h relative to the geoid as:

∆ν gr gh
≅ 2 (1.17)
ν c
where:
g is the acceleration due to gravity at the location of the clock
c is the speed of light (Ashby et al. 2007)
The altitude h is assumed to be small relative to the earth’s radius

The fractional effect on the frequency is about 10−16 m–1. The height above the geoid
is difficult to determine to an accuracy of the order of 10 cm. This corresponds to an
accuracy in clock frequency to the order of 10−17. Consequently, since, as we will see in
Chapter 4, it is possible that optical clock accuracies could reach a level of 10−18 soon, it
should be possible to find a very useful application of atomic clocks in precision geodesy.
Microwave Atomic Frequency Standards 13

1.1.1.2.3 Offsets Introduced by the Electronic Servo System

1.1.1.2.3.1 Spectrum of the Microwave Radiation Imperfection in the spectrum
of the microwave radiation and in its modulation can cause frequency shifts (Audoin
et al. 1978). The microwave radiation at 9.2 GHz is normally obtained by synthesis
from a quartz crystal oscillator at a nominal frequency, say 10 MHz. The process
generally creates sidebands at various frequencies and furthermore amplifies any
spurious spectral components present in the spectrum of the quartz crystal oscillator.
These sidebands create virtual transitions and cause small frequency shifts.

1.1.1.2.3.2 Frequency Shifts Introduced by the Modulation When square

wave frequency modulation of amplitude ωm is used in the servo system to lock the
microwave frequency to the resonance line, the system’s role, by means of synchro-
nous detection and feedback, is to make equal the amplitude of the signal detected on
each side of the line at +ωm and −ωm. If the central Ramsey fringe is asymmetrical,
the servo system may lock to a frequency different from that corresponding to the
maximum of the resonance line as given, for example, by Equation 1.13 including
the Doppler shift (Mungall 1971). The difference in frequency depends on the dis-
tortion of the line and on the amplitude of the frequency modulation ωm used. This
is illustrated in Figure 1.4, where it is clearly seen that for a symmetrical line the
central frequency detected by that means is the same as the frequency for maximum
signal. This is not the case for an asymmetrical line for which the servo frequency
ω′o varies with amplitude of modulation ωm and is not the same as the frequency ωo
for maximum signal.
The asymmetry of the Ramsey fringes may originate, for example, from fre-
quency shifts that depend on velocity. The second-order Doppler effect causes such a

Servo frequency,
Amplitude of symmetrical line A
fringe
Servo frequency,
asymmetrical line B

A
B

−ωm ωo +ωm
−ω′m ω′o +ω′m

FIGURE 1.4 Illustration of the effect on the actual measured frequency of using square
wave modulation for detecting an asymmetrical fringe. A is a symmetrical fringe while B is a
fringe that is made asymmetrical by a bias such as second-order Doppler effect that depends
on the velocity of the atoms.
14 The Quantum Physics of Atomic Frequency Standards

velocity-dependent shift. In that case it is shown that the measured frequency called
ω′D is different from ωD and is given by (see Appendix 1.C) (Audoin et al. 1974):

ω′D − ωo
=
∫ 0
∞
v f (τ)sin 2 bτ sin ωmTd τ
(1.18)
ωo
∫ (1 v) f (τ)sin 2 bτ sin ωmTd τ
2
2c
0

A similar expression is obtained for the shift introduced by a residual phase shift
between the two arms of the Ramsey cavity:

∫ f (τ)sin bτ sin ω Td τ
2
φ m
ωφ − ωo = − 0
∞ (1.19)
L
2c
∫ (1 v) f (τ)sin bτ sin ω Td τ
2

0
2
m

Consequently, it is clearly seen from these expressions that the locking frequency
of the servo system depends on both the amplitude of the microwave field, b, in the
interaction regions and the amplitude of the frequency modulation, ωm. In conclu-
sion, extreme care needs to be taken in order to evaluate the appropriate corrections
for the bias effects mentioned above that are velocity sensitive and distort the central
Ramsey fringe. The reader is referred to QPAFS, Volume 2 (QPAFS 1989), for more
details on this subject.

1.1.1.2.3.3 Frequency Shifts Related to Imperfections in Modulation and

Demodulation These shifts are related to distortion of the modulation and demodu-
lation signals that are used in the creation of the error signal by the synchronous detec-
tion process (QPAFS 1989, Volume 2; Audoin 1992). Even harmonics in the spectrum
cause frequency shifts. Distortion ratios less than 10−6 are preferable in order to make
the effects negligible at the level of accuracy encountered in state-of-the-art standards.

1.1.1.2.3.4 Frequency Control Loop Finite dc gain in the control loop and volt-
age offsets can cause frequency offsets in the frequency lock loop. In state-of-the-art
designs, digital servo loops are used and such offsets are eliminated (Garvey 1982;
Nakadan and Koga 1985; Rabian and Rochat 1988; Sing et al. 1990).
The size of the various offsets described above is summarized in Table 1.1 along
with the present state-of-the-art accuracy in the determination of these offsets. The
table is given without reference to particular systems implemented and is given
solely as a guide to the reader making explicit the relative importance of a given
shift and how accurately it can be determined in the best experimental conditions.
At present, it appears that the biggest shift is the magnetic field offset. However, it is
felt that the accuracy with which it is determined does not cause a major problem if
care is taken in the design of the magnetic environment around the clock. The great-
est cause of inaccuracy is probably still the cavity distributed phase shift limiting
the accuracy to which phase asymmetry in the Ramsey cavity can be determined.
Microwave Atomic Frequency Standards 15

TABLE 1.1
Approximate Size of Biases or Offsets Present in Laboratory Cs Beam
Frequency Standarda
Typical Smallest
Typical Size in Evaluation Uncertainty
Laboratory Standards Achieved
(parts in 1015) (parts in 1015)
Magnetic field >100,000 0.1
Second-order Doppler effect Depends on construction > −50 1
Black Body radiation ~20 0.3
Spin–exchange interactions Unknown Expected ≤1
Cavity phase shift Depends on construction >100 1 to 10
Cavity pulling ~5 to 10 0.6
Bloch–Siegert effect ~1 Expected ≤0.3
Majorana transitions ~2 <1.3
Rabi and Ramsey pulling <2 0.02
Microwave spectrum <1 0.1
Electronics, modulation, 1 1
demodulation, etc.
Microwave leakage Depends on construction <1
Gravitation Depends on location <0.1
Microwave spectrum <1 0.1
Fluorescence light shift in <2 <0.5
optically pumped standards

Source: Vanier, J. and Audoin, C., Metrologia, 42, S31, 2005. Copyright Bureau International des Poids
et Mesures. Reproduced by kind permission of IOP Publishing. All rights reserved.
a The uncertainty given is that achieved in best circumstances and is given as a reference to the accuracy

that may be achieved in practice.

1.1.1.3 Frequency Stability of the Cs Beam Standard

The frequency stability of the Cs beam frequency standard depends on the averag-
ing time, on factors such as modulation and frequency locking scheme and on the
constancy of all the shifts enumerated above. In the so-called short-term region,
where shot noise at the beam detection is important, the frequency stability is given
approximately by (QPAFS 1989, Volume 2):

k′
σ(τ) = (1.20)
Ql (S N )τ1/ 2

where:
Ql is the atomic line Q
S/N is the signal-to-noise ratio essentially limited by shot noise at the detector
k′ is a factor close to unity
16 The Quantum Physics of Atomic Frequency Standards

The range of application of this equation depends on the servo loop, integrating
filter type and bandwidth. As an example, in some well-designed laboratory stan-
dards using magnetic state selection, a frequency stability of 5 × 10−12 τ −1/2 over a
range extending to 40 days has been measured, in general agreement with the above
expression (Bauch et al. 1999). In the case of optically pumped standards, a better
signal-to-noise ratio (S/N) can be obtained and a frequency stability better than that
just mentioned by an order of magnitude can be realized (3.5 × 10−13 τ −1/2) (Makdissi
and de Clercq 2001).
The long-term frequency stability of the Cs beam frequency standards depends
on the stability of the various frequency shifts and offsets enumerated above. Con-
sequently, the frequency of a unit is dependent to a certain extent on its environment.
Depending on construction type, temperature, humidity, atmospheric pressure, and
magnetic field play a role to various degrees in determining long-term frequency
stability. Temperature fluctuations appear to have the most important effect acting
through some of the shifts enumerated above. In general, best results are obtained in
a temperature controlled environment.
Fluctuations of unknown origins generally limit the frequency stability in the
very long term. When the averaging time τ is increased, frequency stability, as given
by Equation 1.20, improves and reaches a plateau called the flicker floor. The level
of this flicker floor is generally a function of unknown parameters. In practice it is
found that better quality in construction and design lowers this flicker floor to nearly
undetectable levels.
Several national institutes and laboratories have been very active during the
period 1970–1990, in developing Cs beam frequency standards using the classical
approach. Those standards reached a high level of maturity. That stage was attained
through intensive research and development, sophistication of the units, better under-
standing of the fundamental phenomena taking place and collaboration between the
institutions. Table 1.2 is a compilation of the main characteristics of several selected
laboratory units that have been developed during that period. They have played and,
in some cases, still play an important role in the accuracy of TAI (Temps Atomique
International) maintained by the BIPM. Most of them have been influential in the
design of classical primary Cs standards implemented later.

1.1.1.4 Recent Accomplishments

During the years following 1990, there appeared to be a radical change regarding the
study and development of laboratory atomic frequency standards, in particular, labo-
ratory primary standards using the Cs atom. That state of affair occurred because of
the refinement of stable solid-state laser diodes that became available with the proper
wavelength and spectrum for efficient optical pumping of alkali atoms such as Cs.
It was then possible to replace the selector and detector magnets in the classical Cs
beam standard by using the technique of optical pumping providing extremely ver-
satile approaches to state selection. That approach also avoided the problems caused
by inhomogeneous magnetic fields that may be created by the state selector magnets
as well as the problems introduced by the alteration of the velocity distribution of
the atoms by the same selector magnets. With optical pumping for state selection,
the velocity distribution is known analytically and the evaluation of frequency biases
 Microwave Atomic Frequency Standards
TABLE 1.2
Characteristics of Selected Laboratory Primary Cs Frequency Standards Developed during the Period 1970–1987
NRC PTB GOSSTRDT GOSSTRDT NIST NIM
NRC (Canada) (Canada) Cs (Germany) (USSR) (USSR) MCs (USA) CRL (Japan) NRLM (Japan) (China)
Cs V VI A & C CS2 MCs R 101 R 102 NBS 6 Cs1 NRLM II Cs2
Distance between 2.1 1 0.8 0.65 1 3.7 0.55 1 3.68
Ramsey cavities (m)
Microwave-magnetic ⊥ ⊥ = ⊥ = ⊥ ⊥ ⊥
field direction relative
to the Cs beam
State selector 2 poles 2 poles Tandem: 2 poles Hexapole 2 poles Hexapole 2 poles 2 poles
analyzers hexapole
– quadrupole
Mean atom velocity 250 200 93 170–220 220 195 110 300
(m/s)
Line width (Hz) 60 100 60 130–200 110 26 100 150
σy(τ) τ−1/2 3 × 10−12 3 × 10−12 2.7 × 10−12 3 × 10−12 5 × 10−12 2 × 10−12 5 × 10−12 <8 × 10−12 1.8 × 10−11
Accuracy 1 × 10−13 1 × 10−13 2.2 × 10−14 1 × 10−13 5 × 10−14 9 × 10−14 1.1 × 10−13 2.2 × 10−13 4.1 × 10−13
References Mungall et al. Mungall et al. Bauch et al. Abashev et al. Abashev et al. Lewis Nagakiri et al. Koga et al. 1981; Xiaoren
1973; Mungall 1981 1987 1983, 1987 1983, 1987 et al. 1981, 1987 Nakadan and 1981
and Costain, 1981 Koga 1982
1977

Source: Vanier, J. and Audoin, C., Metrologia, 42, S31, 2005. Copyright Bureau International des Poids et Mesures. Reproduced by kind permission of IOP Publishing.
All rights reserved.

17
18 The Quantum Physics of Atomic Frequency Standards

that are functions of velocity can be computed more easily. Furthermore, those same
laser diodes allowed the mechanical manipulation of atoms such as reducing their
speed to such limits that it became possible to create small atomic ensembles of
atoms, essentially small balls of the order of a cm or so in diameter, characterized
by a very low temperature. The old dream of the atomic fountain clock, called in the
early days as Fallotron, proposed by Zacharias in the 1950s, could be realized (see
Forman 1985).
Nevertheless, some institutions continued the refinement of their classic Cs beam
standards, sometimes even in parallel with developments in the new avenues, opti-
cal pumping state selection and atomic fountain just mentioned. Those refined fre-
quency standards using the classical approach reached a level of accuracy such that
it proved the high degree of understanding reached in the physics involved by those
who persisted in improving them. It left no doubts on the quality of the work that was
performed. Some of those standards, at the time of writing of this book, are still used
in some cases in the implementation of the atomic timescale (TAI). They are also
used as reference to check, within their own limits, the reliability and absolute accu-
racy of the standards of the new wave. However, it is clear that their contribution as
a primary standard is limited by their accuracy. An examination of all the work pub-
lished during that period makes evident that the goal was that of breaking the 10−14
barrier as far as accuracy is concerned. It was only achieved after extensive work.
The Cs and Rb fountains are now the workhorse in many primary standards labo-
ratories and have reached a level of accuracy in the 10−15 and better. However, it is
worth examining the physics behind the work that was done in the improvement of
the classical Cs frequency standards using magnetic state selection since their under-
standing is essential to the success of the optical pumping and fountain approaches.
We hence review in the following paragraphs some of those refinements that were
accomplished during the last two decades and have provided the scientific commu-
nity with some of the most reliable frequency standards.
As readily observed from Table 1.1, the largest frequency biases, which are observed
in classical Cs beam standards using magnetic state selection and which need to be
evaluated as accurately as possible, are the magnetic field and its homogeneity, the
second-order Doppler effect, and the cavity phase shifts, either nominal between
the two interaction regions or distributed within each region. Black Body radiation,
cavity pulling, Majorana transitions, and Rabi–Ramsey pulling cause shifts much
smaller, less than 10−14, but have also been carefully re-examined in order to certify
that they do not introduce inadvertently any important bias in the measurements.

1.1.1.4.1 Magnetic Field Generation

The magnetic field is applied to provide a quantization axis to the system. It is also
required to separate the field-dependent transitions from the clock transition F = 4,
mF = 0 – F = 3, mF = 0 in order to prevent overlapping of the field-dependent lines as
much as possible. We mentioned above the effect to the tail of the neighbouring Rabi
pedestal on the clock transition. If the amplitude of these transitions on each side
of the clock transition is not identical, a distortion of the central fringe takes place
and the peak of the signal is displaced. The size of the magnetic field required, then,
depends on the accuracy desired. In general, the field is set to a value of the order of
Microwave Atomic Frequency Standards 19

50 to 100 × 10−7 T (50–100 mG). This gives a Zeeman frequency of the order of 25 to
50 kHz for transitions between field-dependent levels. It provides enough separation
between the field dependent Rabi pedestals to guarantee little effect from overlap.
Thus, the applied field is quite large. The displacement ΔνB of the central fringe of
the Ramsey pattern is obtained from Equation 1.9 as:

∆ν B = 427.45 × 108 B02 (1.21)

where:
B is in Tesla

For 50 × 10−7 T (50 mG), this corresponds to a displacement of 1 Hz or 1 × 10−10.

Consequently, in order to reach fractional frequency stability of the order of 10−15,
the field must be stable to something like 10−5 in fractional value. This is very
demanding on several aspects of the construction of such a device. This includes
the stability of the current source driving the circuit generating the field. It also
sets requirements on homogeneity of the field within the system that includes the
Ramsey cavity, which is between 1 and 2 m long in laboratory standards. Finally,
magnetic shielding of the region where the interaction takes place must be well-
designed in order to avoid external field fluctuations that can affect the atomic beam
resonance frequency. These problems have been addressed in various ways. In one
type of design, the field is created by four metallic rods situated along the length of
the standards (Mungall et al. 1973). In that case, the rods are spaced very accurately
by means of a number of pyrex spacers interspaced with coils whose purpose is to
excite Δm = +/−1 transitions and whose resonance frequencies νz are in the kHz
range. Such transitions, when excited, affect the amplitude of the clock transition, an
effect seen on the amplitude of the detected signal. The field can then be determined
exactly through the relation:

ν z = 349.86 × 107 B (1.22)

That relation applies within the upper states manifold F = 4. In another design, the
field is created by means of a long solenoid enveloping the whole Ramsey cavity
with small end coils used for trimming the field at both ends (Bauch et al. 1996). The
whole structure is enveloped in all cases in multilayer high permeability material
such as mu-metal or moly-permalloy. In that approach, as mentioned in the article
cited, one main cause of instability is temperature affecting the length of the sole-
noid support, causing a change in actual solenoid dimensions with a resulting fluc-
tuation of the magnetic field. Temperature regulation is thus required.
Nevertheless, as reported in Table 1.1, it appears that the task of creating a field of
sufficient homogeneity and stability for implementing a clock with accuracy in the
10−15 range is not insurmountable. Actually, even at this high magnetic field, an accu-
racy in magnetic field bias determination of 10−15 has been realized in Physikalisch-
Technische Bundesanstalt (PTB) CS1 (Bauch et al. 2000a, 2000b). The question
of the magnetic field bias is thus not a major problem in the construction of such
a device although it must be implemented with great care, particularly regarding
homogeneity. The main reason of this requirement is that what is measured by means
20 The Quantum Physics of Atomic Frequency Standards

of the Zeeman frequency of ΔmF transitions is the actual value of the field along the
beam while the bias frequency of the clock transition is proportional to the average
of the field squared. In general, <B2> is not equal to <B>2. In practice, the effect
may be evaluated from the position of the Ramsey resonance on top of the Rabi
pedestal (Bauch et al. 1996).
It should, however, be recalled that the Ramsey cavity design approach intro-
duces a desirable property regarding frequency shifts introduced by magnetic field
inhomogeneities. For example, the magnetic field may be different in the two inter-
action regions and may be different from the field in the drift region. The resulting
frequency shift of the central Ramsey fringe, for a drift region L long compared to
the interaction region l, is the sum of the shifts multiplied by the ratio l/L. The actual
shift is given by the equation (QPAFS 1989):

ω − ωo l  ω′o − ωo ω′′o − ωo 
=  +
∫0
τ f (τ)(1 − cos bτ)sin bτ d τ
(1.23)
ωo L  ωo ωo  ∞

∫ 0
bτ2 f (τ)sin 2 bτ d τ

where:
ω ο′ and ωο″ are the resonance frequencies in the two interacting regions, respectively
ωο is the resonance frequency in the drift region

For a laboratory standard with a 1 m long cavity, a frequency shift of the order of
10−4 Hz may thus be reduced by a factor of 100 and made negligible.

1.1.1.4.2 Second-Order Doppler Effect

As is made explicit in Equation 1.18, the second-order Doppler shift is a function of
the distribution of velocities in the detected beam. Because the second-order Doppler
effect is velocity dependent, it makes the line asymmetrical and shifts the fringe line
to the centre. According to the explanation given in Figure 1.4, if square wave modu-
lation is used in the servo system, as is generally the case, the frequency measured is
dependent on the modulation amplitude. This is shown by Equation 1.18 above and
reviewed in some detail in QPAFS, Volume 2 (1989).
An important step in the evaluation of the actual frequency shift is, thus, the eval-
uation of the integrals included in that equation. For this, the velocity or interaction
time distribution must be known. A major effort has been spent by several laborato-
ries on the evaluation of such a distribution. It is well known that the velocity distri-
bution at the exit of magnetic dipole selectors is not Maxwellian as was assumed by
Harrach (Harrach 1966, 1967). In one type of approximation, cut-off velocities were
introduced in the velocity spectrum assumed to be Maxwellian (Mungall 1971). This
is a gross approximation for selector dipole magnets. Nevertheless, the approxima-
tion was checked by calculating numerically the actual Ramsey fringe shape with
the assumed velocity spectrum, and approximate agreement was obtained with the
experimental data. However, in such a procedure the cut-off velocities are chosen
rather arbitrarily and are not a priori based on experimental data. This did not appear
Microwave Atomic Frequency Standards 21

to be a proper avenue for calculating a bias affecting the accuracy of a primary stan-
dard. We now outline regarding how the question of determining the velocity distri-
bution to evaluate carefully the frequency bias caused by the second-order Doppler
effect as well as by cavity phase shift bias was addressed.

1.1.1.4.3 Cavity Phase Shift

According to Table 1.1, and as discussed above, the second most important bias
is the phase shift that can exist between the Ramsey cavity arms. In general, this
phase shift is evaluated by means of beam reversal. This can be done because of
the unique property that the phase shift between the two arms of the cavity changes
sign upon inversion of the beam direction. The measurement may be done by inter-
changing the Cs source and the detector. This is a major operation and depending on
design may lead to opening the system to atmospheric pressure and re-evacuating
the system after the interchange has been effectuated. In some cases, the instrument
is constructed with both, source and detector, at both ends (Mungall et al. 1973;
Bauch et al. 1996). Their position is then switched directly under vacuum by means
of a rotating or sliding mechanism. In all cases, the question remains as to the exact
retrace of the Cs beam within the cavity and consequently as to the reproducibility
of the phase shift with that retrace. This question was examined carefully in QPAFS
(1989). Let us recall the main conclusions.
The Ramsey cavity is normally implemented with an X band waveguide in a
U shape. Two ways for the orientation of the waveguide relative to the direction
of the beam are represented in Figure 1.5, which, for convenience, is reproduced
from QPAFS (1989). The relative orientation dictates the direction of the static mag-
netic field applied in order to satisfy the quantum transition probability condition
for ΔmF = 0 transitions. The radiation H-field in the cavity must be parallel to the
static induction Bo applied. In Figure 1.5a, the static field must be perpendicular to
the beam direction and in Figure 1.5b it must be parallel. The ends of the U cavity
are short circuits. If the material of the waveguide were a perfect conductor, the elec-
tromagnetic radiation within the cavity would be represented by a perfect standing
wave. There would be no travelling waves in the waveguide required to feed energy
losses in the walls of the guide. The position of the antinodes of the magnetic field
would be well determined. In that case, if the Cs beam passes, say close to the short
circuit (Figure 1.5a) or at the λg/2 antinode point from that short circuit (Figure 1.5b),
the atoms would see the same phase of the field in both arms of the cavity. That phase
would change sharply by π from one antinode to the other, but would be constant
in the region of each antinode. All atoms in a narrow beam, a few mm in dimen-
sion, would see the same phase independently of their position within the beam and
furthermore, they would see the same phase whatever is the actual position of the
beam in traversing the cavity around an antinode. This is schematically represented
in Figures 1.6a and b.
This is an ideal situation. In practice, the cavity is made of a material with electri-
cal losses and the wave is attenuated all along its path. Furthermore, upon reflection
at the short circuit its amplitude is smaller than the incoming wave because the
reflection coefficient is less than unity. The standing wave ratio varies along the path
traversed by the wave and its phase is continuously changing with distance travelled.
22 The Quantum Physics of Atomic Frequency Standards

Atomic beam axis (ABA)

(ABA) (ABA)

(a)

ABA

ABA

(b)

FIGURE 1.5 Schematic representation of the two usual implementations of the Ramsey
cavity. In (a) ABA stands for atomic beam axis. The coupling to the cavity is made through an
E-plane T-junction and the cavity is bent in the E-plane; (b) the coupling is done through an
H-plane T-junction. The magnetic field is represented by dotted lines. In the implementations
chosen, in (a) the atomic beam passes close to the end while in (b), it passes at a so-called anti-
node (1/2)λg from the short circuit end. (Data from Bauch, A. et al., IEEE T. Instrum. Meas.,
IM-34, 136, 1985; Mungall, A.G. et al., Metrologia, 9, 113, 1973.)

This is illustrated in Figures 1.6c and d. Let us examine the phase of the microwave
standing wave along the direction travelled called the zg direction. If we call z′g as the
distance within the guide from the nearest antinode point identified by the number
p = 0, 1, 2 …, then the phase is given by:

1  2π z′g
φ ( z′g ) =  pα g λ g + α g z′g + rm  tan (1.24)
2  λg

The full calculation is given in QPAFS, Volume 2 (1989).

In order to understand the following discussion and development, it is important
to know well the effect of each term in this expression and their size. The first term
Microwave Atomic Frequency Standards 23

Short circuit
Short circuit
H H

zg zg
0 λg/2 0 λg/2
(a) (c)

Short circuit Short circuit

ϕ ϕ

zg zg
0 λg/2 0 λg/2
(b) (d)

FIGURE 1.6 Schematic representations of the amplitude of the magnetic field H and the
phase of H at the first antinode(s) near the end of the arms of the Ramsey cavity. (a) and
(b): case where the walls of the waveguide are perfect conductors. (c) and (d): case where the
walls of the waveguide have finite electrical conductivity. (Data from Bauch, A. et al., IEEE
T. Instrum. Meas., IM-34, 136, 1985.)

on the right represents absorption along the waveguide characterized by the absorption
coefficient αg due to the extra length travelled by the wave upon its reflection to reach
the antinode point (1/2) pλg. The second term is of the same nature but represents
the same effect as a function of distance close to an antinode with z′g being a small
distance from the antinode centre. The last term, rm, represents the effect of the finite
conductivity of the cavity termination on the actual reflection of the wave at that
short circuit. The parameter rm is the real part of the surface impedance of the short-
circuit material normalized to the waveguide impedance. Hereunder, we provide the
value of the various parameters for a copper waveguide WR 90

αg = 1.33 × 10−2 m–1

rm = 4.6 × 10−5
λg = 4.65 cm (9,192,631,770 Hz)

We recall that p is an integer with value 0 at the short circuit and 1 at the first antinode
close to the short circuit, which in some implementation is the point where the beam is
oriented as in Figure 1.5b. Because of the electrical losses, two effects result. If the arms
do not have exactly the same length, that is, the T feeding the cavity is not well centred,
the phase is different in the two arms. As a first approximation, neglecting the small
effect of reflection less than unity at the short circuit, rm can be neglected relative to the
other terms in Equation 1.24 and the phase shift ϕ between the two arms is calculated as:

2πα g Lo ∆Lo
φ= (1.25)
λg
24 The Quantum Physics of Atomic Frequency Standards

where:
L o is the mean value of the length of each of the two arms
ΔL o is the difference between their lengths caused by an inexact centring of the T
feeding the cavity

For example, for L o = 15 λg and an error in construction of ΔL o = 10−4 m between the

two arms length, we obtain ϕ = 1.25 × 10−4 rad. The guide wavelength is 4.65 cm,
and for an average speed of 200 m/s, using Equation 1.19, one obtains a fractional
frequency shift of the Ramsey central fringe of the order of 6 × 10−13, which is very
large in the present context. In order to reach the 10−15 range accuracy, the phase
shift should be known to the order of one microradian. This sets relatively rigid
constraints on the fabrication of the cavity, the centring of the coupling T, and the
measurement process itself.
On the other hand, as made evident by Equation 1.24, the phase of the magnetic
field varies with distance within the arms. Thus, the phase seen by the atoms travers-
ing the cavity is different depending on their exact transverse position within the
beam. If the beam is large, several mm to cm depending on the collimation and the
hole made in the cavity, different atoms will see a different phase shift.
According to Equation 1.19 the resulting frequency shift is a function of the sign
of the difference in phase between the two arms of the cavity. In practice, this prop-
erty is used to determine the value of the phase shift by measuring the frequency
for reverse orientations of the atomic beam. As we have already discussed, a basic
requirement is thus that the beam retraces exactly the same path in both directions.
This is made evident from actual values of phase shift with distance as calculated
from Equation 1.24. For small values of z′g, we have approximately:

πα g pLo z′g 2πα g 2πrm z′g

φ1 ( z′g ) = ( z′g ) +
2
+ (1.26)
λg λg λg

For convenience, the results are shown in Figure 1.7, partly replicated from QPAFS
(1989) and for the value of the parameters given above for a WR 90 copper wave-
guide. The variation of the phase along the other axis xg′ perpendicular to the beam
direction for the case of Figure 1.5b, that is in the direction of the larger transverse
dimension, a, of the waveguide, is calculated in the same manner. Those results are
also shown in Figure 1.7.
In practice, the measured gradient of the distributed phase shift may be as large as
10−4 rad/mm, slightly larger than that calculated above (as reported for PTB’s CS2,
Bauch et al. 1987). On the other hand, in a particular design, the precision of the
retrace upon beam reversal and after great experimental care was found to be of the
order of 0.13 mm (Bauch et al. 1993). This would thus correspond to a possible error
of 1.3 × 10−5 rad in the evaluation of the phase shift. Using Equation 1.9, for a cavity
structure 1 m long and a mean atomic beam speed of 100 m/s, this would corre-
spond to a fractional frequency shift of the order of 2 × 10−14. As can be seen, it thus
appears that the phase shift, although great care is taken in evaluating it by means of
beam reversal, limits the accuracy of the clock. Furthermore, the beam trajectory is
affected by the earth’s gravitational field (QPAFS 1989). A simple calculation shows
Microwave Atomic Frequency Standards 25

ϕ1 (10−4 rad.)

6 b

2 a

c
Distance (cm)

−0.6 −0.4 −0.2 0.2 0.4 0.6

−2

−4

FIGURE 1.7 Variation of the phase as a function of the transverse direction across the
beam. Curve b is for a beam traversing the waveguide at an antinode point situated at λg/2
from the short circuit while curve a is for the beam traversing the waveguide close to the short
circuit. Curve c is for the ring cavity studied below. The dotted line curve is for direction x,
perpendicular to the direction of propagation as explained in the text.

that for a 1 m long structure and even with an average atomic beam speed of 100 m/s,
the atoms fall by 0.5 mm during the traversal of the cavity. This number varies with
the speed of the individual atoms forming the beam. This effect of course creates a
real challenge in mechanical design and adjustment to reproduce reliably the beam
path upon reversal. With such an error in path reproducibility, the corresponding shift
could be nearly 10−13 for the case above. For this reason, some systems using slow
atoms have been constructed vertically. However, the particular approach used was
abandoned due to other difficulties encountered in connection to several other detri-
mental effects in particular with magnetic shielding (A. Bauch 2012, pers. comm.).
As is readily observed in Equation 1.26, if the odd terms could be eliminated by
means of a different cavity configuration, the phase shift could be reduced. This can
be accomplished by means of a so-called ring cavity (De Marchi 1986). Such a cav-
ity is shown in Figure 1.8.
In that configuration, the cavity resonant at 9.192 GHz is excited by means of
a rectangular waveguide, and waves are generated symmetrically around the ring.
A standing wave is thus excited in the structure. In the configuration shown, at
the entrance through the tee, the wave separates into two waves travelling to the
right and to the left with amplitudes b1 and b 2, or since we are interested in the
magnetic field components of the wave as H10 and H20. We assume that the struc-
ture is characterized by a propagation vector γ = α + iβ, where α is the absorption
26 The Quantum Physics of Atomic Frequency Standards

T-junction
b3 a3

a1 a2

b1 b2

Cs beam
Cs beam entrance hole z′
entrance hole
(a) (b)

FIGURE 1.8 Ring cavity used in some of the recently implemented Cs atomic beam
frequency standards. (a) 3D view, (b) Identification of the various field components.

coefficient and β the propagation constant 2π/λg, characteristic of the waveguide.

At a small distance z′ from the central point where the beam enters the interaction
region, we have:

− iω t − γ ( l 2+ z ′ )
H1( z′) = H1e
(1.27)
− iω t − γ ( l 2 − z ′ )
H 2 ( z′) = H 2e

With a ring having a circumference l = (n + 1/2) λg, where n is an integer, and

assuming equal amplitude of the field (b1 = b2) at the entrance, we may write at the
antinode where the beam passes through the cavity:

(
H ( z′) = H oeiω t − γ l 2 e − γ z′ + e + γ z′ ) (1.28)

By using trigonometric relations (see Appendix 1.D), it is then shown that the phase
of H close to the beam axis is given approximately by:

φ = αβ z′2 (1.29)

What are the advantages of such a cavity in comparison to the standard one using
short circuits? This question can be addressed by comparing the result just obtained
to that of Equation 1.26 for the standard approach using a cavity terminated with
short circuits. Its first advantage is that the value of phase shift is independent of
the antinode chosen for the interaction. The second advantage is that, since there is
no short circuit used to terminate the cavity, there is no reflected wave and thus the
phase shift does not depend on the attenuation caused by partial reflection at such
a short circuit. For example, for a displacement of the beam along z′ we have the
following result for the phase shift, assuming the same characteristics as those used
previously for the waveguide:
Microwave Atomic Frequency Standards 27

• Cavity with short circuits:

• Beam passing at λg/2 from short circuit: ϕ = 2.09 × 10−5 rad/mm
• Beam passing close to the short circuit: ϕ = 4.87 × 10−6 rad/mm
• Ring cavity: ϕ = 1.8 × 10−6 rad/mm

The distributed phase shift thus appears to be reduced somewhat in the ring cavity
compared to the standard short-circuit approach. However, asymmetries in the ring
may have an effect on the phase shift and position of the antinode. Asymmetries can
be of two types. A phase shift may be introduced in construction through a mechani-
cal asymmetry, a tilt angle of the T-junction, for example. The resulting effect is an
imbalance of the two counter-propagating waves excited in the cavity (De Marchi
et al. 1988). In such a case, the asymmetry may cause a small displacement of the
antinode relative to its position when the structure is symmetrical. It is evaluated
that a tilt angle of 50 mrad can produce a frequency shift of the order of 3 × 10−15 .
A similar effect may result from an asymmetry of the propagation constant γ in the
two halves of the cavity. It is calculated that if the asymmetry, in terms of Δγ/γ, is
less than 10−2, the displacement of the antinode is then less than 0.2 mm, which is
negligible in the present context.
The ring cavity was used in a few recent implementations of laboratory primary
standards (Bauch et al. 1998, 1999a, 1999b). It was also used in some units where
optical pumping was used for state selection (Shirley et al. 2001; Hasegawa et al.
2004). It was found that the distributed phase shift was reduced to some extent when
compared to the short-circuit approach. In the case of PTB’s CS1 refurbished with
ring cavities, it was concluded that the phase gradient could be as large as 20 μrad/
mm although using the analysis presented above the phase over the beam diameter
of 3 mm should not vary more than approximately 4 μrad. It is concluded, however,
that this is much better than the 94 μrad/mm expected in their case for the standard
cavity using short-circuit terminations (Bauch et al. 1998). We note that in order
to evaluate correctly the effect of the phase shift, we must evaluate it by means of
Equation 1.19 and that the velocity distribution or interaction time distribution must
be established for the particular device used. This is what we examine next.

1.1.1.4.4 Velocity Distribution Evaluation and Control

The evaluation of some of the biases just mentioned, as shown above by means of
Equations 1.13 and 1.16 relies heavily on the exact determination of the velocity dis-
tribution at the exit of the state selector magnet. It is interesting to follow the evolu-
tion of the assumptions and technique of determination of this distribution over the
years. As briefly described above, this evolution started from the use of a Maxwell
distribution (Harrach 1967) and went to the use of a distribution altered by cut-offs
at both low and high speeds made necessary by the use of selector magnets (Mungall
1971). This was followed by an experimental approach that uses a numerical analy-
sis of the response of the beam standard to pulsed radiofrequency (RF) excitation
(Hellwig et al. 1973). Another approach comprised analyzing the shape of the observed
Ramsey fringes directly, which depends on the velocity distribution of the atoms
in the beam. This was done numerically and the so-called moments of the velocity
distribution were evaluated (Audoin et al. 1974). Finally, a most powerful approach
28 The Quantum Physics of Atomic Frequency Standards

using Fourier transform techniques to analyze the observed Ramsey fringes was
introduced early in the development of laboratory standards (Kramer 1973; A.
Bauch 2012, pers. comm.). The technique was used in various ways and has been
an important factor in improving accuracy. It was outlined in QPAFS, Volume 2
(1989). Let us recall that technique and address the question in the more gen-
eral context of the control of atomic velocities in laboratory Cs beam frequency
standards.
The need of knowing accurately the velocity distribution, or the interaction
time distribution, arises in the evaluation of both the second-order Doppler effect
and the cavity phase shift. These are given by Equations 1.13 and 1.16. In those
equations, it is observed that although the shifts are functions of velocity, they are
rather small, being of the order of parts in 1013. Consequently, although producing
a measurable frequency shift, the effect on the shape of the Ramsey fringes them-
selves arising from those shifts is very small. The shape of the signal detected, as
given by Equations 1.12 and 1.15 with the term v2/2c2 and ϕ neglected can then
be used as an excellent representation of the Ramsey line shape, which remains a
function of velocity through the function f(τ). In that context, those expressions
become:
∞
1
2 ∫
I = Ib + Io f ( τ) sin 2 bτ 1 + cos ( ω − ωo ) T  d τ
0
(1.30)

However, this equation can be rearranged. We remove the constant term Ib not con-
tributing to the useful signal and hence obtain:
∞ ∞
1 1
I=
2 ∫
0
2 ∫
Io f ( τ) sin 2 bτ d τ + Io f ( τ ) sin 2 bτ cos ( ω − ωo ) d τ
0
(1.31)

This equation relies on the approximation made earlier that (ω − ωο) is much smaller
than bτ. The first term is independent of ω and is simply the maximum of the Rabi
pedestal. The second term is the Ramsey modulation of that pedestal. This expres-
sion contains the effect of the atoms' velocity distribution on the Ramsey fringes.
The Rabi pedestal being broad with a flat top, a simple approach consists in looking
only at the second term as a good representation of the effect of the velocity spec-
trum on the signal observed. It is readily seen that the expression is a cosine trans-
form of the term f(τ) sin2 bτ. Consequently, with a knowledge of the Rabi frequency
b in the cavity, f(τ) can in principle be obtained through an inverse Fourier transform
of the measured Ramsey pattern (Kramer 1973; Daams 1974).
A particular approach in applying such a technique consists in setting ω = ωο,
and measuring the signal amplitude as a function of the field intensity b in the cavity
(Boulanger 1986). Let us now recall the general idea. For ω = ωο, the expression for
the Ramsey fringe amplitude becomes:
∞

∫
I = Io f ( τ) sin 2 bτ d τ
0
(1.32)
Microwave Atomic Frequency Standards 29

This can also be written as

∞
1 1
2 2 ∫
I = Io − Io f ( τ) cos 2bτ d τ
0
(1.33)

In that case, the second term is the cosine transform of f(τ) that we call F(b). An
experimental plot of I against b gives F(b) and its inverse transform gives:
∞
4
f ( τ) =
π ∫
F( b) cos 2bτ db
0
(1.34)

This expression is evaluated by means of numerical analysis of the experimental

results obtained for I. The difficulty in the techniques is the evaluation of b, a mea-
sure of the microwave field in the cavity. It requires a good knowledge of the cav-
ity Q, its dimension, and the power fed into it. This is a difficult and not a precise
exercise. In order to initiate the calculation, it is best to use an approximately known
value of b, such as a value for which the signal is optimal. This value is given by
bopt = (π 2)(ν l ). The value of b is then altered by changing the power fed into the
cavity and normalized to that value. A graph of the maximum of the central fringe is
obtained as a function of b and the calculation of the Fourier transform of this result
can be done. The subsequent exercise then consists in calculating the shape of the
Ramsey patterns with the results obtained and in comparing the result to the shape
obtained experimentally. The width of the patterns obtained is then an important
parameter in concluding about the agreement of theoretical and experimental results.
The exercise can be repeated until the agreement is satisfactory. The value of f(τ) can
then be obtained through normalization by means of Equation 1.7. The technique
was used successfully and provided clear information of the velocity spectrum. It
was claimed for example that the second-order Doppler shift could be evaluated with
accuracy better than a few parts in 1015.
Another approach consists in using Equation 1.31, considering, as mentioned
above, that the Ramsey fringes given by the second term are the cosine transform
R(Ωo) of f (τ)sin 2bτ through the equation (Daams 1974):
∞

R(Ωo) =
∫ f (τ) sin bτ cos ( Ω τ) d τ
2
o (1.35)
0

The function f (τ)sin 2 bτ, called F(τ), can then be obtained from the data recorded as
a function of Ωo by means of an inverse transform (Shirley 1997):
∞
4
F( τ) = f ( τ) sin 2 bτ =
π ∫
R ( Ωo ) cos ( Ωo τ ) d Ωo
0
(1.36)

As in the previous approach, b needs to be determined in order to isolate f(τ). This

may be done as in the previous case using b at its optimal value, by making several
measurements at various values of b relative to that value and averaging the results.
Another random document with
no related content on Scribd:
 CHAPTER II
THE FRAME, AND ITS ACCESSORIES

Under this title should be included the frame, axles, springs,

wheels, steering gear and brakes.
From the beginning it was recognized that the different strains and
stresses set up by the passing of the wheels over uneven ground
and by the motor and driving mechanism, must be taken care of
before reaching the body of the automobile, which otherwise would
soon go to pieces.

Fig. 1. Views of Plain Frame.

The Frame.—Therefore, not only springs had to be interposed

between the body and the wheel axles, but also a substructure for
the body, called the frame, which must be rigid enough to prevent
any destructive strains from reaching the body.
In Fig. 1, A shows a top view of a frame made up of channel bars
and B shows a side view to illustrate how the torsion or twist takes
place. It will be understood that the frame thus made is not
designed to lend itself to the entire inequalities of the road, as the
springs are interposed for that purpose.
Experience in the construction and use of tubular frames, as first
employed in bicycles, proved too expensive for assembling, when
used in automobiles. The tubular form of construction was very soon
displaced by frames consisting of metal parts bolted or riveted
together. The main or side members are now usually made of
channel steel which gives great rigidity and strength, compared with
its weight.

Fig. 2. Quarter Elliptic.

How the Frame is Suspended.—The important feature is to mount this

frame on the axle. The frame, carrying a body and all the load of the
vehicle, has to permit three distinct movements.
First. That due to the inequalities of the road, which produces a
torsional twist.
Second. A lateral swing, caused by traveling alongside a hill, or
due to centrifugal force when making a turn rapidly.
Third. A fore and aft movement, as when traveling over undulating
surfaces, or in suddenly stopping and starting.

Fig. 2a. Half Elliptic.

For these reasons springs must be made to compensate for such

motions, and to absorb the jar as much as possible.
 Fig. 3. Three-quarter Elliptic.

The Springs.—Many forms of spring mountings have been devised,

but the following illustrations show the types which set forth the
principles involved. Outside of coiled springs which are used in some
forms of delivery cars, the standard springs are leaf springs, built up
from a number of steel leaves.
There are four distinct forms of springs used, as follows:
1. The quarter elliptic, used on Ford, and similar cars, as
illustrated in Fig. 2.
2. The half elliptic, Fig. 2a, which is the most widely-used form.
These springs are usually attached with their front end directly to
the frame, and with the rear end by means of a shackle; the center
is fastened by spring clips to the axle.

Fig. 4. Full Elliptic.

Where a distance rod is used, as on the rear axle, both ends are
attached by shackles.
3. The three quarter elliptic, Fig. 3, always used as a suspension
for the rear axle. This form gives more flexibility than a half elliptic,
and is still stiffer so far as side motion is concerned, than the
following type.
4. The full elliptic, Fig. 4, was formerly used much more than at
the present time.
There are also in use springs comprising a combination of half
elliptic, or three quarter elliptic, on each axle, in which the front end
is shackled to the frame, and the rear ends connected by shackles to
another half elliptic spring, the center of which is fastened to the
frame.

Fig. 4a. Cantilever Spring.

Fore and Aft Motion. Provision must be made, in all cases, for the
fore and aft movement of the car body which takes place in stopping
or starting, and, particularly when the wheels strike an obstruction.

Fig. 5. Fore and Aft Motion.

Flues. Fig. 5 shows a side view of a car, in which the dotted lines
indicate the position of the body, relative to the normal, when the
wheels strike an obstacle.
Lateral Motion. In like manner when the car swings around a
corner, or is traveling along a hill-side, the springs must hold the
body from swinging too far. Fig. 6 illustrates, by means of the dotted
lines, the side movement. It is obvious, therefore, that the springs
have a duty to perform in addition to that of merely giving flexibility
to the body.

Fig. 6. Lateral
Motion.

Cantilever Spring.—A special form of half elliptic springs, lately

developed, and of increasing use, is the cantilever spring, where the
axle is attached to one end, the center of the spring being pivoted to
the frame, and the other end shackled to or sliding in the frame.
Shock Absorbers.—Shock absorbers are mechanical means placed
between the frame and the axles for the purpose of dampening the
sudden recoil of the springs after being compressed, when meeting
a road obstacle. In the absence of such a device the recoil is likely to
suddenly throw up the frame, body and passengers, or produce an
unpleasant shock.
Originally, simple leather straps were used, reaching from the
body to the axle, which only limited, but did not dampen or
gradually absorb the shock. Now different forms of frictional resisting
toggle-levers are used, which not only absorb the shocks, but also
prevent the bumping of the axle against the frame, and eliminate
breaking of springs.
The Axle.—Axles are of two kinds, generally designated as “live,”
when they turn the wheels; and “dead” when they do not turn the
wheels, but simply support the weight of the frame and of the body.
Dead axles are used with double chain drive, as, in that case, the
sprocket wheels are attached directly to the sides of the wheels and
the wheels turn on the studs, or ends of the dead axle.
 Live Axles.—1. Plain live axles originally consisted of a shaft
without differential gearing, having one wheel fast on it, the other
turning. Modern construction shows two axle shafts in a housing, the
weight of the car, and the tooth pressure of the differential being
carried by the axle shafts.
2. Semi-floating axles have the weight of the car carried by the
axle shafts, whereas the tooth pressure of the differential is
supported by the housing, and only the turning effect or torsion is
transmitted by the axle shafts.

Fig. 7. Floating Axle.

Fig. 8. Semi-floating Axle.

3. Full floating axles carry the full weight of the car, and the
differential bevel gear teeth pressure with the housing, so that the
axle shafts carry no load but only the torsional stress.
 Both full and semi-floating constructions are applied to rear axles
only. The front wheels are now universally applied to knuckles, which
swing on vertical pivot pins at the ends of the dead axles.
Wheels.—Wheels are now in a transition state. The ultimate wheel
has not yet appeared; but whatever its form or construction, certain
things are essential.
Flexibility.—In the ordinary wagon or carriage wheel, there is but
little, if any, flexibility; but in automobiles, where speed is a
consideration, elasticity, either in the rim, or in some other part of
the wheel, is necessary.
One of the reasons for this is, that on account of tire expense,
motor wheels are smaller than carriage wheels. Making them
smaller, however, produces certain disadvantages. One is that in
going over the inequalities of the road, the axle on the small wheel
has a greater vertical movement than on a large wheel, and the jar
on striking an obstruction is more pronounced, also. These
disadvantages, however, are more than counterbalanced by the
elasticity of the invention.
Large vs. Small Wheels.—Fig. 9 shows a large wheel A, passing over
a depression B. The large arc of the wheel does not permit the rim
to go to the bottom. On the other hand, the small wheel C goes to
the bottom of the depression, and the vertical distance which the
axle of this wheel must travel, is three times as far as in the case of
the wheel A.
In Fig. 10, where the large wheel strikes an obstruction D, the
angle of its upward movement, as designated by the line E, is much
less than the impact force of the small wheel, as shown by the
greater slope or incline of the line F.
 Fig. 9. Crossing Fig. 10. Striking
Depression. Obstruction.

Minimizing Shocks.—It is obvious, therefore, that if part of this

shock can be taken up by the tire, the difference due to the smaller
diameter of the wheel, will not be so apparent.
The thickness, or widths of the tires also minimizes the impact and
distribute the jars while running, so that with these advantages a
small wheel has been found to be more practical than a large one.
Resiliency.—Most wheels are now made with wooden spokes,
secured by means of a pair of metal-flanged hub plates, bolted
together so as to clamp the radiating spokes, but wire wheels are
now coming more into favor, whereas cast or pressed solid steel
wheels are used on some heavy trucks.
 CHAPTER III
TIRES, TUBES, AND RIMS

Tires.—Three kinds of tires are now used, namely: Solid, cushion,

and pneumatic. These forms all use rubber, or some compound with
the qualities and characteristics of rubber, so as to afford a good
tractive surface, as well as resiliency.

Fig. 11. Solid Tire.

The solid tires are used on heavy trucks, where weight and not
speed must be provided for.
Cushion tires are sometimes employed on cars and trucks of
medium weight.
Pneumatic tires, in which air is used, are universally used in
automobiles for all other purposes.
 Fig. 12. Single
Tube.

The air is confined in two ways:

First, by what is known as the “single tube.” (Fig. 12.)
Second, by the “double,” or inner tube system. (Fig. 13.)
The single tube is well adapted for light vehicles, or where great
speed or weight are not considered, and this type is now confined to
bicycles. But it has certain disadvantages, namely: That of creeping,
due to the impossibility of properly securing it to the rim of the
wheel. Sand and grit are also liable to creep in between the tire and
rim, and wear the material, thereby ruining it.
The outer casing, or shoe, is split on its inner side, and usually
provided with an annular flange on each side of the split, which rests
against the rim of the wheel, and is adapted to receive a rim which
securely fastens the annular flange of the shoe, to the rim of the
wheel.
 Fig. 13. Double
Tube.

Various ways are provided for holding the shoe to the rim of the
wheel; but in the different types shown by the illustrations, Figs. 13
and 14, the shoe has a flange which is held within channels on the
rim, or by some form of fastening device.
The inner tube is usually of thin elastic rubber, so made that when
properly inflated it will fit the outer tube or casing. The outer part,
which can be made of a different rubber compound, and is better
adapted to stand wear, whereas, the inner tube, which is made of
the best, and more costly material is protected.
Advantage of Double Tube.—The great advantage of the double tube
is due to the positive means of fastening it to the rim of the wheel,
so as to prevent creeping.
In the single tire construction the latter is liable to roll out of its
bed where quick turns are made, but with the double tube this is not
possible.
 Fig. 14. Illustrating Tire-removing
Tool.

Putting On and Taking Off Double Tubes.—To do this properly with

clincher tires is quite an art. A pair of blunt, round-ended levers is
best for the purpose.
The practice is to use cold chisels, screw drivers and like sharp or
pointed tools. This is bad practice. A pair of levers, as shown in Fig.
14, can be made by any one, and you may be sure that their use will
not be liable to jag a hole in the inner tube during the removal
process.
When the inner tube is put into the outer casing, or tire, as it is
called, powdered talcum should be liberally applied, to the tube and
also placed within the casing. The tube is then put in and carefully
distributed and straightened out before the clinchers are put on.
A little air blown into the tube will prevent it from being pinched
under the flanges of the casing. The spare tubes should be inclosed
in a receptacle of some kind which will exclude light, and protect
them from heat. With the advent of the quick detachable rims of
different forms these troubles have happily disappeared in the
modern automobile.
Damages to Tires.—Many things must be provided for in the matter
of tire keep. The thing most necessary to guard against is punctures,
caused either by sharp stones, or nails. When a casing has a heavy
protective tread the inner tube may not be effected, but it frequently
happens that the outer casing is slitted for some distance, and the
great pressure forces the thin wall of the inner tube into the slitted
opening, and it is thus ruptured, not on account of its being
punctured, but because the outer tire did not afford protection
against the pressure.
 Repairs to Tires.—It is not a difficult job to repair tires, and the
apparatus for doing it is very simple. Rubber, in its natural state, is a
white, thick, milky juice, which after several heating and refining
processes becomes dark and sticky.
Vulcanizing.—When in this condition and properly mixed with
sulphur, it may be vulcanized, which destroys the stickiness, and
makes it firm and elastic. Vulcanizing is a kind of baking process, the
maximum heat being about 275 degrees, but generally less. The
time required is from 12 to 15 minutes, dependent on the thickness
of the mass to be vulcanized.

Fig. 15.
Vulcanizer.

When the torn or cut portion of the tube or tire is carefully

cleaned, it is filled with the plastic rubber, and the heater is applied.
The heater, one form of which is shown in Fig. 15, is merely a shell
with a heater connection, and this being partly filled with water,
generates steam, the temperature of the shell being, of course,
dependent on the pressure of the steam developed.
To repair the inner tube, it should be first rubbed with sand paper,
and liquid rubber cement applied. When this becomes tacky apply
the patch and dry. It is then ready to be vulcanized.
Oil as an Enemy of Tires.—All literature on the subject of tires give
warnings as to the insidious character of oil, which deteriorates the
rubber. Most manufacturers now make an oil proof quality, but the
cheaper grades are not to be depended on.
The action of oil shows itself in several ways, but principally
because it dissolves the rubber.
 Non-Skidding Tires.—Various means are provided in the shape of
tire treads to prevent skidding, the most important being vacuum
cups, the herring-bone formation, and various ribbed or ridged
surfaces. Nevertheless, for smooth asphalt pavements, chains or
similar substitutes are found most satisfactory.
Sudden application of the brakes, or the sliding of wheels on
hillsides or the skidding of the car in making short turns at too great
speeds, are the most destructive things for tires, however good they
may be.
Tires For City Use.—A tire which may be of good service for
country roads, might not be available for city work. The tendency of
many drivers is to hug the curb too closely, and the result is a wear
on the side, which is its weakest point. It is like the side of a shoe,
the upper of which can be readily worn through, whereas the sole
will stand hard usage.
In country use the great danger is in the winter months, where
the wheels must pass over or along frozen ruts. There the same
difficulties of side wear are liable to destroy the best material.
Side Slipping.—The same remarks apply to the weakness of tires
due to side slipping. The fibers of the fabric are ruptured at the
weak point and the least external abrasion assists in destroying it.

Fig. 16. Turning Action on Front

Wheel.

Faulty Alinement.—Another cause of ruptured tires is attributable to

improper alinement of wheels, due to the wheel being not exactly
true, through a bent axle, or improper adjustment. This is more
frequently the case with front than with rear wheels.
 It will be readily understood that while the rear wheels have the
traction applied to them, the front wheels, fixed as they are, to the
short turning knuckles, are affected by a movement diagonally
across the tire, at every turn which is made.
This is shown by reference to Fig. 16. The movement of the car is
in the direction of the arrow A, consequently, when the wheels are
turned, the momentum of its forward end is in the direction of the
arrows B B.
When the turn is to the right, the strain is on the inside of one tire
and on the outside of the other, and when the movement is to the
left the conditions are reversed in the stress, and this explains why
the tires of front wheels are so liable to yield, in all cases where
turns are made at high speeds.
Broken Fabric.—The fabric of a tire may be ruptured without giving
any indications on its outer side. When there is a strong impact
force, like a transverse ridge, which will force in the tire, several
things occur. First, the body of the tire is flattened out so that it has
a bulging cheek on each side; and, second, a strain is produced on
the longitudinal fibers.
Bruises.—The result of such a severe bruise is to cause a break,
not transversely, or longitudinally, but usually, obliquely, for the
following reason. The fabric has one set of its threads running across
the tire, and the other set around the perimeter. This arrangement
of the fabric usually prevents a straight break in either direction, and
the weakest part of the fabric is across the diagonal direction.
 Fig. 17. Illustrating the Strain on
Fabric. Fig. 18.

Try the experiment with a handkerchief, as shown in Fig. 17 by

stretching it in the direction of the threads; and then look at Fig. 18,
in which case the tension is diagonally, or across the corners. This
will be sufficient, probably, to suggest to your mind the reason for
the break on diagonal lines.
The rubber material is not sufficient to prevent the stretch which
the fabric permits, hence the break follows.
Under Inflation.—To permit a wheel to run flat causes a tire to
stretch more on the tread than along the clinch line.
Stretched Tires.—A good illustration of this is shown in Fig. 19,
where the tread is a succession of irregular wavy surfaces, whereas
the sides remain round and full.
Many attribute this to poor or defective tires. The best tire in the
market will show symptoms of this kind, if allowed to run when
deflated. In such cases the flatness produces a continual pouching
out of the sides, which follow the wheel around, and tend to
produce a creeping of the fabric.

Fig. 19. Effect of Flat-Tire.

In time the rubber works away, or along on the fabric, until it

becomes stretched at the tread, and all the pressure in the tire will
not again restore it to the proper condition.
Blistered Tires.—A blister is a plain case of the rubber being
separated from the fabric. At first the injury may be a small cut
down to the fabric, which, after being neglected for a time, permits
sand to enter, and a grinding takes place, each movement of the
parts causing a further separation, and pressure expands the rubber,
until, finally, it bulges out and gives an unsightly appearance, as well
as starts the tire on its road to destruction.
Such defects can be cured, if taken in time, as many compounds
are on the market for this purpose.
Rim Cutting.—This is caused by sand or sharp particles being
forced in between the tire and edges of the rim, which causes a
wearing out at the contact points. Insufficient air is another cause.
The tires flatten and are then cut by the metal.
Frequently the tire is too small for the rim, and this is always bad
for it. Heavy loads will cause cutting, because the tire will be
flattened out, although inflated to the proper tension.
It is good practice to turn a tire, when one side wears more than
the other. This wearing on one edge excessively, shows some defect
in the wheel alinement, which needs correcting. Possibly the wheels
may not be parallel. This is a frequent trouble with front wheels, on
account of the bending of the arm which runs from the knuckle.
Inflation Pressures.—Manufacturers of tires furnish data with
respect to the proper pressures for their products, and these vary
somewhat, and it is wise to observe the pressures which they
indicate for the different sizes.
Expansion of Heated Air.—There is another cause of tire expansion,
not generally considered, which is due to the expansion of heated
air. It is not infrequently the case that a tire will, in running, heat up
fifty or sixty degrees, which means an expansion of one-eighth the
volume of air within the tube. If, therefore, there is any weakness in
the walls of the tire, a blowout follows.
As this heating is liable to take place to a greater extent in the
summer than in winter, it is obvious that it is better to under inflate
during that period, than to have an over pressure, particularly with
old, or considerably worn, or injured tires.
 CHAPTER IV
THE STEERING GEAR AND BRAKES

The Steering Column.—This is a very important mechanical element

of the car. Its direct useful functions are to carry or hold the
mechanism for steering the machine, and for the motor control,
controlling the air supply for the fuel, as well as for regulating the
sparking mechanism.
Motor Control.—Some machines are provided with a foot lever
mechanism (accelerator) as well as the throttle lever on the steering
wheel. This is advantageous, because in moving through crowded
streets, where frequent and quick changes are necessary, the foot is
the most convenient for controlling purposes.
Throttle Movement.—A downward pressure of the foot opens the
throttle, and a spring returns it to its normal position. The foot
throttle is also convenient when shifting the transmission gear, as
both hands are otherwise engaged, one to operate the gear-shifting
levers, and the other for steering.
The hand throttle on the steering wheel, however, is most
convenient for long runs, when little change is required, and it can
then be set so as to avoid the use of the foot lever.
The levers are so arranged that they do not entirely close the
throttle, but, when fully thrown to a closed position, will still provide
a sufficient opening to keep the engine running light.
 Fig. 20. Steering Wheel.

Steering Wheel Type.—The drawing, Fig. 20, shows a type of

steering wheel, which has a segment A. The long lever B is for
throttling purposes, as above described, and the short lever C for
operating the sparking device.
These levers are differently disposed and arranged on the wheel,
or on the column supporting the wheel shaft, but the illustration is
sufficient to show the principle of construction, and we are
interested only in the types and not in the modifications which are
available, and are constantly being made to meet certain conditions.
 Fig. 21. Steering
Gear.

Steering Gear.—Fig. 21 shows an approved form of construction for

the gear, which converts the rotating motion to a direct line
movement. In this the hollow supporting column A, is firmly fixed to
a base B.
The shaft C which passes through the column, has a worm D at its
lower end, and is journaled in a base E, which carries a cross shaft F,
in which is mounted the worm wheel G. One end of the shaft F has
an arm H for moving the arms of the wheel knuckles.
Within the tubular shaft C, is a tubular shaft I, for the throttle
lever to operate, the lower end of which has an arm J, and within
the shaft I, is a shaft K for the sparking lever, the lower end having
an arm L.
In the best cars all these parts are made adjustable, so as to
provide for wear. In examining or selecting a car, this is one of the
points to note.
 Fig. 22. Type of Front Axle.

Front Axle.—Fig. 22 shows a common form of front axle, with

knuckles and cross connecting rod A, the latter providing means, by
the nuts B C, for alineing the wheels.
The Brakes.—These are made in two types, one which is usually in
the form of a contracting band, and the other which expands.
All cars are now equipped with two braking systems, one being
the service, or running brake, and the other the emergency brake.
These brakes are all of the drum type, and are either expanding, or
contracting bands tightening against the drums.

Fig. 23.
Contracting Fig. 24.
Brake. Expanding Brake.

Running Brake.—The running brake is operated by the foot pedal,

whereas the emergency brake is generally connected up with the
lever at the side of the seat.
 The foot pedal is on some cars connected with the clutch in such a
way that when pedal is pressed to set the brake, the clutch is
released. This prevents an inexperienced or confused driver from
applying the brake when he forgets to release the clutch.
Double-Acting Contracting Brake.—Fig. 23 shows the manner in
which a double-acting contracting brake operates. As the band A,
has a tension on each end, when the rod B, is drawn forwardly, it is
immaterial which way the brake drum C travels.
In Fig. 24 the drum C has a pair of oppositely-disposed shoes D,
which are held in such a position that they are not revoluble, and
may be moved outwardly by the lever E and links F.
These figures, of course, show merely the simple forms of the two
types, and do not go into the refinements of construction which
make them so effective in service.
It is obvious, however, that the power exerted through either type
of brake, depends on the leverage afforded by the relative lengths of
the limbs of the bell-crank lever E, to each other.
Contracting Brake.—Fig. 25 shows a well-known type of contraction
brake, in which the cylinder A, has thereon two brake bands B C,
hinged together at their rear ends. At their front ends they are
connected with a bell-crank lever D, the forward movement of the
upper end of the lever being such as to cause the bands to pinch the
drum A.
A contractile spring E draws back the lever when the foot releases
the pedal, and the link F, between the bell-crank lever and the upper
band C, has a turnbuckle arrangement to provide for taking up in
case of wear.
The brake bands have means for automatically holding them clear
of the wheels when not in use.
 Fig. 25. Contract Mechanism.

Equalizers.—Sometimes the brake is placed on the propeller shaft;

but when one of the brakes is placed on each wheel, an equalizing
bar, or other means, must be used. One form of this is shown in Fig.
26, in which A is the bar, B the rod which goes to the brake lever,
and C C, the rods that run back to the brakes on the wheels.
Naturally, the equalizer will not act with the same effect on both
wheels, unless they are in the same condition. Frequently one of the
brake cylinders will be dry and the other coated with grease, or
accumulate moisture from some source. It is, therefore, a necessary
part of inspection and care to keep them in serviceable condition.

Fig. 26. Equalizer Bar.

The Emergency Brake.—The emergency brake has a pawl which acts

in the teeth of a segment alongside of the lever, so it may be held in
any position to which the lever may be thrown. This lever has no
provision whereby the clutch is disengaged when the brake is
applied, for the reason that should it become necessary to stop a car
going up hill, and when the emergency brake is required, the brakes
would have to be released before the clutch could be thrown in, so
that the car would be likely to start down hill before this could be
done. On this account the emergency brake has no connection with
the clutch.

Fig. 27. Rear axle.

Service and Emergency
Brake.

Combined Service and Emergency Brake.—Fig. 27 represents a

standard type of service and emergency brake, each of the internal
expanding type. As both are inclosed in a drum they are absolutely
free from dirt and dust, and the construction shown eliminates
rattling of the parts.
The wheel bearing is also represented by the annular ball-bearing
type of construction, in which the balls are unusually large, and
therefore, capable of taking great weight and high speed without
undue wear.
 CHAPTER V
THE DIFFERENTIAL

The Meaning of Differential.—This is a term used to designate the

difference in the turning movement of two wheels on opposite ends
of an axle. For various reasons they do not turn at the same rate of
speed, particularly in turning corners, where the outer wheel must
travel a greater distance than the inner wheel.
If both wheels are fixed to the shaft the latter would be submitted
to a torque, or one of the wheels would slip, and thus be destructive
of tires.
On the other hand, if one wheel should be loose, then, as power is
applied to the shaft, the tractive action would be on one wheel only,
and this would be bad practice, and frequently cause the wheel to
slip, and thus unduly increase the wear of the tire.
The differential is made up of a system of gears, which are so
arranged that one wheel may turn independently of the other, and at
the same time the effective driving power is utilized by each.
Various forms of this mechanism have been developed. While the
differential is an exceedingly simple piece of mechanism, it is not
such an easy matter to describe its operation, so that the principle
will be explained by a series of illustrations.
Equalizer Bar.—Examine Fig. 28. Let A be an equalizer bar,
mounted on the end of a thrust bar B, by a pivot C, so the ends will
swing back and forth freely. A horizontal bar D is hinged at each end
of the equalizer, which bars project forwardly parallel with each other
and these are provided with right-angled bends E E, simply for
convenience in describing the operation.
 Fig. 28. Equalizing
Mechanism.

Fig. 29. Resistance in

Equalization.

While differential gears are very simple structurally, it is not an

easy matter to explain the principle on which a faster motion is
transmitted to one wheel than another, and under conditions where
the speed is constantly changing.
 Fig. 30. Equalizer and Differential
Movements.

For instance, in Fig. 30, a cord A, over a pulley B, has weights C,

D, at its ends. If the pivot or fulcrum E, of the wheel, is stationary,
as in sketch 1, and the wheel is turned, say a quarter of the way
around, one weight will move down below the line X the same
distance that the other weight moves above it, as shown in 2.
Thus far we have an equalizer, pure and simple. But a differential
requires something more. It is necessary, under certain conditions,
for the weight D to move a greater distance in the same time than
C, or the reverse. Or, as sometimes happens, one of the weights, as
for instance, in 3, remains fixed while the other moves.
In this case, with the pivot pin E fixed, such a thing would be
impossible, hence, in order to make such a relative movement
between the two weights, the pin must move, and this motion is
shown in 3, where it moves down from the line F. That movement,
or change of position of the pivot E, is what takes place in the small
intermediate gears in a train of differential gearing.
Transmission Wheel.—In Fig. 32 is shown a section of the
differential housing, 1, in which, for convenience, all refinements of
construction are eliminated. This shows the divided axle shafts 5, 6.
In Fig. 33 is shown a side view of the same housing. This may be
connected with the motor shaft by means of bevel gears, or driven
by a sprocket chain. In either case the housing 1 is the substitute for
the thrust bar B, in Fig. 28, and the bevel pinions 2, which are
mounted within the wheel 1, represent the equalizer bar of that
figure.

Fig. 31. Differential in

Housing.

The gears which make up the train are usually put into a suitable
casing, as illustrated in Fig. 31, which gives a good example of the
construction. The housing A is fixed to the side of a large bevel gear
B, this gear being designed to receive power from the motor through
a bevel pinion C. One part of the axle D passes through the gear B,
and is fixed to a bevel gear E within the housing, and the other part
of the axle F passes through the housing and is fixed to a bevel gear
G, the same size as gear E.
Intermediate the two gears is a pair of bevel pinions H, H, and
these latter are mounted on pivots I, I, projecting inwardly from the
housing.
The fact that the pinions are attached to housing has the effect of
complicating the matter, so that it may be well to show the relative
arrangement of the gears without the housing.

Fig. 32. Section of Differential.

Fig. 33. Side View of Differential

Wheel.

In Fig. 34 we have added to Fig. 33, two bevel gears 3, 4, which

are mounted on the axles 5, 6, these representing the rear drive
axles of the car.
Action of Transmission Gearing.—From the foregoing it will be seen
that the axles abut each other, within the hub of the large gear 1,
within which they are journaled. We might, therefore, call these
pinions the counterparts of the bars E E.
 Fig. 34. Top View of Differential
Wheel.

As long as the resistance to the turning movements of the pinions

3, 4 is the same, the housing through pinions 2, 2, will simply carry
the bevel gears 3, 4 around with it, without turning them, just the
same as the equalizer bar B was moved forward without either end
swinging back or forth; but the moment the wheel of the shaft 5, for
instance, is compelled to travel at a higher rate of speed, or the
wheel on shaft 6 meets with a greater resistance, the small
equalizing gears 2 will turn, and the revoluble motion of the housing
1, while transmitting the power, and also carrying the gears, will act,
in effect, the same as the push bar shown in the previous
illustration.
Like the equalizing bar, the effect is to turn one wheel, say 3, with
less, and the other wheel 4 with more than the normal power or
speed.
Fig. 28 shows the principle on which all differential automobile
gearing is based, that is, that both wheels receive half of the driving
power even if one wheel should turn faster, as shown at Fig. 29,
which is the case when turning a corner. This is what causes the
power to drive both wheels at all times, whether going straight or on
a turn.