NMR and Chemistry

www.pdfgrip.com
www.pdfgrip.com
NMR and Chemistry
An introduction to modern
NMR spectroscopy

Fourth Edition
Dr J.W.Akitt
Formerly Senior Research Officer at the
University of Leeds

Professor B.E. Mann

Professor of Chemistry
University of Sheffield

Taylor & Francis

Taylor & Francis Group
Boca Raton London New York

A CRC title, part of the Taylor & Francis imprint, a member of the
Taylor & Francis Group, the academic division of T&F Informa pic.

www.pdfgrip.com
© 2000 J. W. Akitt and B. E. Mann

The right of J. W. Akitt and B. E. Mann to be identified as authors of this

work has been asserted by them in accordance with the Copyright, Designs and
Patents Act 1998.

All rights reserved. No part of this publication may be reproduced or transmitted

in any form or by any means, electronic or mechanical, including photocopying,
recording or any information storage and retrieval system, without permission in
writing from the publisher or under licence from the Copyright Licensing Agency
Limited. Further details of such licences (for reprographic reproduction) may be
obtained from the Copyright Licensing Agency Limited of 90 Tottenham Court
Road, London W1P OLP.

First edition published by Chapman & Hall 1973

Second edition published by Chapman & Hall 1983
Third edition published by Chapman & Hall 1992
Fourth edition published in 2000 by:
Stanley Thornes (Publishers ) Ltd

10 9 8 7 6 5 4

A catalogue record for this book is available from the British Library
ISBN 0 7487 4344 8
Every effort has been made to contact copyright holders of any material reproduced
in this book and we apologise if any have been overlooked.

www.pdfgrip.com
 Contents

Abbreviations ix
Preface to the fourth edition xv

1 The theory of nuclear magnetization 1

1.1 The properties of the nucleus of an atom 1
1.2 The nucleus in a magnetic field 5
1.3 The source of the NMR signal 6
1.4 A basic NMR spectrometer 12
1.5 Questions 14
2 The magnetic field at the nucleus: nuclear screening
and the chemical shift 17
2.1 Effects due to the molecule 17
2.2 Isotope effects 24
2.3 Effects due to unpaired electrons 27
2.4 The chemical shift 32
2.5 Notes on sample preparation, standardization and
solvent and temperature effects 36
2.6 Questions 43
3 Internuclear spin-spin coupling 45
3.1 The mutual effects of nuclear magnets on resonance
positions 45
3.2 The appearance of multiplets arising from spin-spin
coupling 47
3.3 Spin-spin coupling satellites 61
3.4 The description of spin systems 65
3.5 Second-order effects 69
3.6 Questions 86
4 Nuclear magnetic relaxation 101
4.1 Relaxation processes in assemblies of nuclear spins 101
4.2 Dipole-dipole relaxation 104
4.3 Quadrupolar relaxation 109
4.4 Spin rotation relaxation: detailed molecular
motion 117
4.5 Chemical shift anisotropy relaxation 120
4.6 Scalar relaxation 121

www.pdfgrip.com
vi Contents

4.7 Examples of 13C relaxation times 124

4.8 Questions 126
5 The spectrometer 129
5.1 The magnet and field homogeneity 130
5.2 The probe 134
5.3 Field-frequency lock 135
5.4 The transmitter 137
5.5 The detection system 139
5.6 Production of the spectrum 147
5.7 Rapid multiple pulsing 156
5.8 Manipulation of collected data 158
5.9 Questions 166
6 Making the spins dance 167
6.1 Decoupling 167
6.2 Composite pulses 180
6.3 Refocusing pulse 182
6.4 Questions 188
7 NMR spectra of exchanging and reacting systems 189
7.1 Systems at equilibrium 189
7.2 Reaction monitoring of systems not at equilibrium 224
7.3 Questions 230
8 Multiple resonance and one-dimensional pulse sequences 233
8.1 Decoupling difference spectroscopy 234
8.2 The nuclear Overhauser effect 237
8.3 One-dimensional multipulse sequences 253
8.4 Exercises in spectral interpretation 267
9 Two-dimensional NMR spectroscopy 273
9.1 /-resolved two-dimensional NMR spectroscopy 278
9.2 Homonuclear COSY NMR spectroscopy 282
9.3 Heteronuclear COSY NMR spectroscopy 290
9.4 HOHAHA or TOCSY 294
9.5 Two-dimensional INADEQUATE 296
9.6 Overhauser and magnetization transfer based
two-dimensional NMR spectroscopy 296
9.7 Inverse detection 304
9.8 Three-dimensional NMR spectroscopy 312
9.9 Questions 312
10 Magnetic resonance imaging and biomedical NMR 335
10.1 Producing an image 335
10.2 Whole body imaging 337
10.3 Diffusion and flow 341
10.4 Chemical shift imaging 342
10.5 Biological uses of imaging - imaging microscopy 346
10.6 Industrial uses of imaging techniques 348
10.7 Biomedical NMR 350

www.pdfgrip.com
 Contents vii

11 High-resolution solid-state NMR 355

11.1 Magic angle spinning 357
11.2 Spin-1/2 nuclei with low magnetogyric ratios 359
11.3 1= 1/2 nuclei with high magnetogyric ratios 365
11.4 MAS of quadrupolar nuclei 368
11.5 Some applications 376
11.6 Deuterium, an integral-spin nucleus 381
11.7 Questions 383

Bibliography 385
Answers to Questions 391
Index 397

www.pdfgrip.com
www.pdfgrip.com
 Abbreviations

APT Attached Proton Test

ASIS Assisted Solvent-Induced Shifts
ATP Adenosine TriPhosphate
BIRD Bilinear Rotation Decoupling
C222 N(C2H4OC2H4OC2H4)3N
CAMELSPIN a two dimensional NMR experiment now
normally called ROESY
CHESS CHEmical Shift Selection imaging
COSY Correlation SpectroscopY
COSY-45 a COSY spectrum using a 45° pulse
COSY-90 a COSY spectrum using a 90° pulse
CP Cross-Polarization
cpd composite pulse decoupling
CPMAS Cross-Polarization Magic Angle Spinning
CRAMPS Combined Rotation And Multiple-Pulse
Spectroscopy
CSA Chemical Shift Anisotropy
CW Continuous-Wave
CYCLOPS A method of cycling pulse and receiver phases to
minimise artefacts
DANTE Delays Alternating with Nutation for Tailored
Excitation
dB deciBel scale
DEPT Distortionless Enhancement by Polarization
Transfer
DSS Me3SiCH2CH2CH2S03Na
EFG quadrupolar Electric Field Gradient
EXSY EXchange SpectroscopY
FID Free Induction Decay
fod 6,6,7,7,8,8,8-heptafluoro-2,2-dimethyl-3,5-
octanedionate
FT Fourier Transform
GARP Globally optimized Alternating-phase
Rectangular Pulses, a broadband decoupling
pulse sequence
HMBC Heteronuclear Multiple Bond Correlation
HMQC Heteronuclear Multiple Quantum Coherence

www.pdfgrip.com
x Abbreviations

HOESY Heteronuclear Overhauser Effect SpectroscopY

HOHAHA HOmonuclear HArtman HAhn
HSQC Heteronuclear Single Quantum Coherence
Hz Hertz, the unit of frequency in cycles per second
IF Intermediate Frequency
INADEQUATE Incredible Natural Abundance Double Quantum
Transfer Experiment
INEPT Insensitive Nuclei Enhanced by Polarization
Transfer
IR InfraRed
IUPAC International Union of Pure and Applied
Chemistry
JMOD / Modulation
MAS Magic Angle Spinning
MHz mega Hertz, 106 Hz
MLEV-17 A composite pulse sequence which can be used
for spin locking or decoupling
MREV-8 A composite pulse sequence which can be used
for decoupling
NMR Nuclear Magnetic Resonance
NOE Nuclear Overhauser Effect
NOESY Nuclear Overhauser Effect SpectroscopY
PENDANT Polarization ENhancement During Attached
Nucleus Testing
ppm parts per million
PRESS Point RESolved Spectrosopy
PW Pulse Width or duration of the pulse
RF Radio Frequency
ROESY Rotating frame Overhauser Enhancement
SpectroscopY
STEAM STimulated Echo Acquisition Mode
TMS TetraMethylSilane
TOCSY TOtal Correlation SpectroscopY
TSP Me3SiCD2CD2CO2Na
uv Ultraviolet
WALTZ a broad band decoupling pulse sequence
WATERGATE WATER suppression by GrAdient Tailored
Exitation
ZSM-5 a highly siliceous zeolite

Symbols and Abbreviations

A Arrhenius pre-exponential factor

aN nuclear electron hyperfine interaction constant
B0 applied magnetic field
BV,B2 radiofrequency magnetic fields
DW dwell time
E electric field

www.pdfgrip.com
 Abbreviations xi

E0 an energy barrier
Ea Arrhenius activation energy
Ez electric field along a bond
F Nyquist frequency
G field gradient
Gx, Gy, Gz field gradients in the jc, y, and z directions
h reduced Planck's constant, h/2n
I moment of inertia
nuclear spin quantum number
n
j coupling constant through n bonds (in Hz)
n
K reduced coupling constant 47i2/MyAyB, through n bonds
k Boltzmann's constant, 1.3807 x 10~23 J/K
rate constant
K(v) field intensity at a frequency v
m angular momentum quantum number
Mx net magnetization along the x axis
Mxy net magnetization in the xy plane
My net magnetization along the y axis
Mz net magnetization along the z axis
P power
pressure
PA> PE relative populations of sites A and B
Q electric quadrupole moment
q{ electric charge
R relaxation rate, s~!
the gas constant, 8.3145 J Kr1 moH
r a distance
Rl spin-lattice relaxation rate, s'1
^IDD dipole-dipole spin-lattice relaxation rate, s'1
RIQ quadrupolar spin-lattice relaxation rate, s-1
/?1SR spin-rotation spin lattice relaxation rate, s"1
R2 spin-spin relaxation rate, s"1
^200 dipole-dipole spin-spin relaxation rate, s'1
^?2Q quadrupolar spin-spin relaxation rate, s"1
S electron spin quantum number
nuclear spin quantum number of nucleus 5,
T relaxation time, s
T time, s
Tl spin-lattice relaxation time, s
7\CSA chemical shift anisotropy spin-lattice relaxation time, s
riDD dipole-dipole spin-lattice relaxation time, s
7\e electron spin-lattice relaxation time, s
rlQ quadrupole spin-lattice relaxation time, s
T2Q quadrupole spin-spin relaxation time, s
risc scalar coupling spin-lattice relaxation time, s
r2SC scalar coupling spin-spin relaxation time, s
riSR spin-rotation spin-lattice relaxation time, s
T2 spin-spin relaxation time, s

www.pdfgrip.com
xii Abbreviations

T2* apparent spin-spin relaxation time, s

^2DD dipole-dipole spin-spin relaxation time, s
Tp time between pulses
v, u real and imaginary components of the FID
Vl9 V2 voltage
Vxx xx component of the quadrupolar electric field gradient
tensor
Vyy yy component of the quadrupolar electric field gradient
tensor
Vzz zz component of the quadrupolar electric field gradient
tensor
WQ zero quantum transition
Wl single quantum transition
Wm line width at half height
W2 double quantum transition
Z nuclear charge

a, p nuclear spin states

a0 pulse angle
8 chemical shift
AE energy separation
AG* free energy of activation
A//* enthalpy of activation
Av separation of two NMR signals
8N population difference
AS* entropy of activation
AV* volume change of activation
19
<|) F chemical shift scale
y magnetogyric ratio
ye magnetogyric ratio of the electron
r| asymmetry factor
observed NOE
the shear viscosity
viscosity of the liquid
r|max maximum NOE
v frequency
v0 spectrometer frequency
v l5 V2 frequencies of individual signals
G screening constant
Gn, G22, CJ33 components of the screening tensor
Gd diamagnetic term of the screening constant
G£ screening due to electric fields
ap paramagnetic term of the screening constant
G± _L component of the screening constant
G,| || component of the screening constant
T a time in a pulse sequence
an old !H chemical shift scale
T2 orientational correlation time

www.pdfgrip.com
 Abbreviations xiii

TC rotational correlation time

Tex exchange time
tSR angular momentum correlation time
co frequency, rad s-1
3 absolute frequency reference relative to !H of TMS at
100.000000 MHz.
(I magnetic moment
|I0 permeability of a vacuum
|iz the component of JJL in the field direction

www.pdfgrip.com
www.pdfgrip.com
 Preface to the
Fourth Edition

It is some six years since J.W. Akitt wrote the preface to the third
edition and now, in retirement, he welcomes the cooperation of an
old colleague, Professor Brian Mann of the University of Sheffield, to
update the text. The first edition appeared in 1972, over 26 years ago,
and in this time NMR spectroscopy has seen immense changes and
developments. The pace of technical change has perhaps slowed over
the six years since the third edition appeared but the number of appli-
cations continues to increase and it has become time to make big
changes in the presentation of the book while keeping the general
layout of the subjects. High-field spectrometers have become common-
place and their operation is carried out via comprehensive computer
control so that they are easy to use and are now operated even by
undergraduate students as a teaching facility. This has led us to intro-
duce rather more description of the workings and operation of spec-
trometers to help such debutant users to understand better what their
commands actually cause to happen.
One topic that has been finally abandoned in this edition is the old
method of continuous wave NMR spectroscopy. Also, to a large extent,
the old CW NMR spectra, complete with wiggle beats have been
replaced by Fourier transform ones. However, we should remember
that the time-shared lock systems used on the FT equipment are best
understood in CW terms.
Now that multinuclear NMR spectroscopy has become routine with
a wide range of nuclei being easily observable, many examples of their
use have been included, ranging from sensitive nuclei such as 31P to
very insensitive nuclei such as 187Os. The discussion of multipulse and
two-dimensional NMR spectroscopy has also been enlarged to reflect
their increasing importance. Many such experiments have become
routine, with fully automated instruments running them.
Examples and many new problems, mainly from the recent inor-
ganic/organometallic chemistry literature, support the text throughout.
Brief answers to all problems are provided in the text with fuller
answers on the Thornes website at http://www.thorneseducation.com.
We are indebted to many people for their comments and assistance,
especially to Professor K. Elsevier for commenting on early drafts of
parts of the manuscript, Bruker Spectrospin Ltd for insights into
present day electronics, and for permission to reproduce diagrams from

www.pdfgrip.com
xvi Preface

their reports in Figs 5.4, 8.29,10.3,10.13,11.2,11.3,11.4, and 11.10(b),

Varian Associates for permission to reproduce the spectra used in
Exercises 1 and 3 in Chapter 3, Dr A. Romer for the spectra used
in Exercise 6 of Chapter 8, Professor B.L. Shaw for Fig. 7.29, Professor
R.K. Harris for examples of solid state spectroscopy, and to him and
Dr J. Klinowsky for comments about the solid state Chapter 11, to
Drs M. Decorps, A. Ziegler and M. Raybaudi of INSERM, Grenoble,
who showed J.W.A. around their laboratory and provided advice and
several up-to-date examples for the imaging Chapter 10.
Finally, we are indebted to the numerous publishers and more
numerous individuals who have given permission for illustrations from
their publications to be used in this book and whose help is expressly
acknowledged in the figure captions.
J.W.A.
Seez
B.E.M.
Sheffield
January 2000

www.pdfgrip.com
 The theory of nuclear
magnetization 1
1.1 THE PROPERTIES OF THE NUCLEUS OF AN ATOM

The chemist normally thinks of the atomic nucleus as possessing only

mass and charge and is concerned more with the interactions of the
electrons that surround the nucleus, neutralize its charge and give rise
to the chemical properties of the atom. Nuclei, however, possess
several other properties that are of importance to chemistry, and to
understand how we use them it is necessary to know something more
about them.
Nuclei of certain natural isotopes of the majority of the elements
possess intrinsic angular momentum or spin, of total magnitude
h V/(/+ 1). The largest measurable component of this angular moment
is Ih, where / is the nuclear spin quantum number and h is the reduced
Planck's constant, /z/2ir. The spin quantum number / may have inte-
gral or half-integral values (0, 1/2, 1, 3/2, . . .), the actual value
depending upon the isotope. Since / is quantized, several discrete
values of angular momentum may be observable and their magnitudes
are given by fim where the quantum number m can take the values 7,
/ - 1, / - 2, ...,-/. There are thus 21 + 1 equally spaced spin states of
a nucleus with angular momentum quantum number /.
A nucleus with spin also has an associated magnetic moment jx. We
define the components of JJL associated with the different spin states as
rajji//, so that JJL also has 21 + 1 components. In the absence of an exter-
nal magnetic field the spin states all possess the same potential energy,
but they take different energy values if a magnetic field is applied.
The origin of the nuclear magnetic resonance (NMR) technique lies
in these energy differences, though we must defer further discussion of
this until we have defined some other basic nuclear properties.
The magnetic moment and angular momentum behave as if they
were parallel or antiparallel vectors, i.e. pointing in the same or oppo-
site directions. It is convenient to define a ratio between them which
is called the magnetogyric ratio, y:
17 =**± ±
h I =hi (1.1)
^ '
•y has a characteristic value for each magnetically active nucleus and
is positive for parallel and negative for antiparallel vectors. We will

www.pdfgrip.com
2 The theory of nuclear magnetization

see that the sign of 7 influences both spin-spin coupling and the way
energy is exchanged between spins.
If / > 1/2 the nucleus possesses in addition an electric quadrupole
moment, Q. This means that the distribution of charge in the nucleus
is non-spherical and that it can interact with electric field gradients
arising from the electric charge distribution in the molecule. This inter-
action provides a means by which the nucleus can exchange energy
with the molecule in which it is situated and affects certain NMR
spectra profoundly.
Some nuclei have 1 = 0. Important examples are the major iso-
topes 12C and 16O, which are both magnetically inactive - a fact that
leads to considerable simplification of the NMR spectra of organic
molecules. Such nuclei are, of course, free to rotate in the classical
sense, but this must not be confused with the concept of quantum-
mechanical 'spin'. The nucleons, i.e. the particles such as neutrons and
protons which make up the nucleus, possess intrinsic spin in the same
way as do electrons in atoms. Nucleons of opposite spin can pair,
just as do electrons, though they can only pair with nucleons of the
same kind. Thus in a nucleus with even numbers of both protons and
neutrons all the spins are paired and / = 0. If there are odd numbers
of either or of both, then the spin is non-zero, though its actual value
depends upon orbital-type internucleon interactions. Thus we build up
a picture of the nucleus in which the different resolved angular
momenta in a magnetic field imply different nucleon arrangements
within the nucleus, the number of spin states depending upon the
number of possible arrangements. If we add to this picture the concept
that s bonding electrons have finite charge density within the nucleus
and become partly nucleon in character, then we can see that these
spin states might be perturbed by the hybridization of the bonding elec-
trons and that information derived from the nuclear states might lead
indirectly to information about the electronic system and its chemistry.
The most important properties of the elements relevant to NMR
spectroscopy are listed in Table 1.1. This gives the atomic weight of
the nuclear isotope of the element listed, and where there is more
than one magnetically active isotope this is indicated by an atomic
weight in parentheses. The isotope listed is the one most usually used,
though the unlisted ones are in some cases equally usable. The next
column gives the spin quantum number, /, followed by the natural
abundance of the isotope. The receptivity or natural signal strength of
the nucleus is given relative to that of 13C, which itself gives a fairly
weak signal, and this figure is made up of the intrinsic sensitivity of
the nucleus (high if the magnetic moment is high) weighted by its
natural abundance. Some elements are used in enriched forms (particu-
larly 2H and 17O), when the receptivity is, of course, substantially
higher. The data are completed by the quadrupole moment (where / >
1/2) and the resonant frequency in a particular magnetic field. This
can be determined to a much higher precision than shown, for a reso-
nance in an individual compound, and it should be remembered that

www.pdfgrip.com
 The properties of the nucleus of an atom 3

Table 1.1 Nuclear properties of some of the elements

Natural Quadrupole
Atomic Spin abundance Receptivity moment Resonant frequency
Element mass I (%) (13C = LOO) (1030 m2) (MHz) at 2.348 T
Hydrogen 1 1/2 99.985 5670 None 100.00
Deuterium 2 1 0.015 0.0082 0.287 15.35
Tritium 3 1/2 Radioactive — None 106.66
Helium 3 1/2 0.00014 0.0035 None 76.18
Lithium 6 1 7.42 3.58 -0.064 14.72
Lithium 7 3/2 92.58 1540 -3.7 38.87
Beryllium 9 3/2 100 78.8 5.3 15.06
Boron 10 3 19.58 22.1 7.4 10.75
Boron 11 3/2 80.42 754 4.1 32.08
Carbon 13 1/2 1.108 1.00 None 25.15
Nitrogen 14 1 99.63 5.70 1.67 7.23
Nitrogen 15 1/2 0.37 0.022 None 10.14
Oxygen 17 5/2 0.037 0.061 -2.6 13.56
Fluorine 19 1/2 100 4730 None 94.09
Neon 21 3/2 0.257 0.0036 9 7.90
Sodium 23 3/2 100 524 10 26.43
Magnesium 25 5/2 10.13 1.54 22 6.13
Aluminium 27 5/2 100 1170 14 26.08
Silicon 29 1/2 4.7 2.1 None 19.87
Phosphorus 31 1/2 100 377 None 40.48
Sulfur 33 3/2 0.76 0.098 -6.4 7.67
Chlorine 35(37) 3/2 75.53 20.2 -8.2 9.81
Potassium 39 3/2 93.1 2.69 5.5 4.67
Calcium 43 7/2 0.145 0.053 -5 6.74
Scandium 45 7/2 100 1720 -22 24.33
Titanium 49(47) 7/2 5.51 1.18 24 5.64
Vanadium 51(50) 7/2 99.76 2170 -5.2 26.35
Chromium 53 3/2 9.55 0.49 -15 5.64
Manganese 55 5/2 100 1014 40 24.84
Iron 57 1/2 2.19 0.00425 None 3.24
Cobalt 59 7/2 100 1560 42 23.73
Nickel 61 3/2 1.19 0.24 16 8.93
Copper 63(65) 3/2 69.09 368 -22 26.51
Zinc 67 5/2 4.11 0.67 15 6.25
Gallium 71(69) 3/2 39.6 322 11 30.58
Germanium 73 9/2 7.76 0.62 -17 3.48
Arsenic 75 3/2 100 144 29 17.18
Selenium 77 1/2 7.58 3.02 None 19.07
Bromine 81(79) 3/2 49.46 279 27 27.10
Krypton 83 9/2 11.55 1.24 27 3.86
Rubidium 87(85) 3/2 27.85 280 13 32.84
Strontium 87 9/2 7.02 1.08 16 4.35
Yttrium 89 1/2 100 0.676 None 4.92
Zirconium 91 5/2 11.23 6.05 -21 9.34
Niobium 93 9/2 100 2770 -32 24.55
Molybdenum 95(97) 5/2 15.72 2.92 -1.5 6.55
Technetium 99 9/2 Radioactive -0.13 22.51
Ruthenium 99(101) 5/2 12.72 0.815 7.6 4.61
Rhodium 103 1/2 100 0.18 None 3.16
Palladium 105 5/2 22.23 1.43 65 4.58
Silver 109(107) 1/2 48.18 0.28 None 4.65
Cadmium 113(111) 1/2 12.26 7.69 None 22.18

www.pdfgrip.com
 4 The theory of nuclear magnetization

Table 1.1 continued

Natural Quadrupole
Atomic Spin abundance Receptivity moment Resonant frequency
Element mass I (%) (13C = LOO) (1030 m2) (MHz) at 2.348 T
Indium 115(113) 9/2 95.72 1920 86 22.04
Tin 119(115,117) 1/2 8.58 25.7 None 37.29
Antimony 121 (123) 5/2 57.25 530 -33 24.09
Tellurium 125(123) 1/2 6.99 12.8 None 31.55
Iodine 127 5/2 100 541 -79 20.15
Xenon 129(131) 1/2 26.44 32.4 None 27.86
Caesium 133 7/2 100 275 -0.3 13.21
Barium 137(135) 3/2 11.32 45 28 11.19
Lanthanum 139(138) 7/2 99.91 343 22 14.24
Praseodymium 141 5/2 100 1620 -4.1 29.03
Neodymium 145(143) 7/2 8.3 0.393 -25 3.41
Samarium 149(147) 7/2 13.83 0.665 5.6 3.43
Europium 151(153) 5/2 47.82 464 114 24.48
Gadolinium 155(157) 3/2 14.73 0.124 160 3.09
Terbium 159 3/2 100 394 134 24.04
Dysprosium 163(161) 5/2 24.97 1.79 251 4.77
Holmium 165 7/2 100 1160 349 21.34
Erbium 167 7/2 22.94 0.665 283 2.90
Thulium 169 1/2 100 2.89 None 7.99
Ytterbium 171(173) 1/2 14.31 4.5 None 17.70
Lutetium 175(176) 7/2 97.41 173 346 11.43
Hafnium 177(179) 7/2 18.50 1.47 330 4.06
Tantalum 181 7/2 99.988 213 330 12.13
Tungsten 183 1/2 14.28 0.0608 None 4.22
Rhenium 187(185) 5/2 62.93 511 220 23.05
Osmium 187(189) 1/2 1.64 0.0015 None 2.28
Iridium 193(191) 3/2 62.7 0.122 78 1.90
Platinum 195 1/2 33.8 19.9 None 21.50
Gold 197 3/2 100 0.153 55 1.75
Mercury 199(201) 1/2 16.84 5.68 None 17.87
Thallium 205(203) 1/2 70.5 807 None 57.63
Lead 207 1/2 22.6 11.9 None 20.92
Bismuth 209 9/2 100 819 -37 16.36
Uranium 235 7/2 0.72 0.0054 455 1.84
Source: Mason (1987) Multinuclear NMR, Plenum, New York.

each nucleus will have a range of frequencies due to the chemical shift
effect. The resonance frequency is proportional to the magnetogyric
ratio.
The first point to note about this list is that almost all the elements
are represented, the only missing stable ones being argon and cerium,
and that, in principle, virtually the whole of the Periodic Table can be
studied by NMR. Indeed, with modern instrumentation, this is now
realizable, though there are some cases, notably nuclei with very high
quadrupole moments, where study in the liquid state is not rewarding.
The usefulness of a nucleus to the NMR spectroscopist depends in the
first place upon the chemical importance of the atom it characterizes
and then upon its receptivity. Thus the extreme importance of carbon

www.pdfgrip.com
 The nucleus in a magnetic field 5

spectroscopy for understanding the structures of organic molecules has

led to technical developments that have overcome the disadvantages
of its poor receptivity, so that 13C NMR is now commonplace.
Hydrogen, with its very high receptivity, has, of course, been studied
right from the emergence of NMR spectroscopy as a technique useful
to chemists, and proton spectroscopy, as it is often called (proton =
X
H, the term is commonly used by NMR spectroscopists when
discussing the nucleus of neutral hydrogen), has been used for the
identification of the majority of organic compounds and many inor-
ganic ones, and for the physical study of diverse systems. Other much-
studied nuclei are U B in the boron hydrides or carboranes, 19F in the
vast array of fluoro organics and inorganics, 27A1 and 29Si in a wide
range of inorganic and organic compounds, and 31P in its many inor-
ganic and biochemical guises. However, even quite low receptivity
is no longer an insurmountable obstacle and, if a chemical problem is
presented that is capable of solution by NMR spectroscopy, then what-
ever nucleus may require to be observed, the appropriate effort may
well bring the desired rewards.

1.2 THE NUCLEUS IN A MAGNETIC FIELD

If we place a nucleus in a magnetic field U0 it can take up 21 + 1 orien-

tations in the field, each one at a particular angle 9 to the field direc-
tion and associated with a different potential energy. The energy of a
nucleus of magnetic moment |JL in field BQ is -jxz/?0 where jxz is the
component of JJL in the field direction. The energy of the various spin
states is then
j -j r r\

J?0 or individually — fjuB0, — JJLBO , —-— |JUB0 , etc.

The energy separation between the levels is constant and equals

jjJ?0/7. This is shown diagrammatically in Fig. 1.1 for a nucleus with
1=1 and positive magnetogyric ratio. The value of m changes sign as
it is altered from / to -/ and accordingly the contribution of the
magnetic moment to total nuclear energy can be either positive or
negative, the energy being increased when m is positive. The energy
is decreased if the nuclear magnetic vectors have a component aligned
with the applied field in the classical sense. An increase in energy
corresponds to aligning the vectors in opposition to the field. Quantum
mechanics thus predicts a non-classical situation, which can only arise
because of the existence of discrete energy states with the high-energy
states indefinitely stable.
In common with other spectral phenomena, the presence of a series
of states of differing energy in an atomic system provides a situation
where interaction can take place with electromagnetic radiation of the
correct frequency and cause transitions between the energy states. The
frequency is obtained from the Bohr relation, namely

www.pdfgrip.com
 6 The theory of nuclear magnetization

(a)

Figure 1.1 (a) The nuclear spin energy for a single nucleus with 1=1 (e.g. 14N) plotted as a function of
magnetic field B0. The two degenerate transitions are shown for a particular value of B0 (b) The align-
ment of the nuclear vectors relative to BQ that correspond to each value of m. The vector length is
ftV/(7 + 1) and its z component is hm, whence cos 9 = mN 1(1 + 1).

hv = A£
where AE1 is the energy separation. For NMR
hv = fjjJo//
In this case, the transition for any nuclear isotope occurs at a single
frequency since all the energy separations are equal and transitions
are only allowed between adjacent levels (i.e. the selection rule Am =
± 1 operates). The frequency relation is normally written in terms of
the magnetogyric ratio (1.1), giving
v = 7^0/211 (1.2)
Thus the nucleus can interact with radiation whose frequency
depends only on the applied magnetic field and the nature of the
nucleus. Magnetic resonance spectroscopy is unique in that we can
choose our spectrometer frequency at will, though within the limita-
tion of available magnetic fields. The values of y are such that for
practical magnets the frequency for nuclei lies in the frequency range
between a present maximum of 800 megahertz (MHz) and a minimum
of a few kilohertz (kHz) using the earth's magnetic field.

1.3 THE SOURCE OF THE NMR SIGNAL

The low frequency of nuclear magnetic resonance absorption indicates

that the energy separation of the spin states is quite small. Since the
nuclei in each of the states are in equilibrium, this suggests that
the numbers in the different spin states will be similar, though if there

www.pdfgrip.com
 The source of the NMR signal 7

is a Boltzmann distribution among the spin states then we can expect

more nuclei to reside in the lowest energy states. For a system of
spin-1/2 nuclei, a Boltzmann distribution would give
NuINt = exp(-A£/&r)
where Nu and Nf are the numbers of nuclei in the upper and lower
energy states respectively, AE1 is the energy separation, k is the
Boltzmann constant, and T is the absolute temperature. AE is given
above as JJLBO//, which, for / = 1/2, equals 2fjJ?0. Thus
NJNt = exp(-2^B0/kT)
which since Nu ~ Nt can be simplified to
NJNj = 1 - 2[LBJkT
For hydrogen nuclei in a magnetic field of 9.39 T (tesla), where the
resonance frequency is 400 MHz, and at a temperature of 300 K,
the quantity 2^B0/kT has a value of about 6 x 10~5, which means
that the excess population in the lower energy state is one nucleus
per 300 000. This is extremely small. The z components of the magnetic
moments of the nuclei in the upper energy state are all cancelled by
those of the lower energy state, only the small number excess in the
lower energy state being able to give rise to an observable magnetic
effect. This is a weak nuclear paramagnetism, which is only observable
at low temperatures where the population difference is a maximum.
For this reason, we have to resort to a resonance technique in order
to observe a signal.
If / > 1/2 one obtains a similar though more complex picture since
the excess low-energy nuclei do not all have the same value of rajx//,
and, for integral /, one energy level has no magnetic component in
the z direction.
It should be noted that the size of the excess low-energy population
is proportional to 7?0. For this reason the magnetic effect of the nuclei
and therefore their signal intensity increases as the strength of the
magnetic field is increased. Temperature also has an important effect,
and the value of 2^BQ/kT is increased to 7.2xlQ- 5 at 250 K or
decreased to 4.9 x 10~5 at 370 K at 9.39 T. This can produce detectable
changes in signal strength relative to background noise in variable-
temperature experiments.
We have so far built up a picture of the nuclear moments in a
sample polarized with or against the magnetic field and lying at an
angle 6 to it. The total angular momentum (i.e. the length of the vectors
of Fig. 1.1) is V/(7 + 1) and the angle 0 is then given by
m
cos 9 =
7(7+1)
This angle can also be calculated classically by considering the
motion of a magnet of moment JUL in an applied magnetic field. It is
found that the magnet axis becomes inclined to the field axis and

www.pdfgrip.com
 8 The theory of nuclear magnetization

rotates or precesses around it. The magnet thus describes a conical

surface around the field axis. The half-apex angle of the cone is equal
to 0 and the angular velocity around the cone is yBQ, so that the fre-
quency of complete rotations is -yC^ir, the nuclear resonant frequency
(Fig. 1.2(a)). This precession is known as the Larmor precession.
For an assembly of nuclei with 7 = 1/2 there are two such preces-
sion cones, one for nuclei with m = +1/2 and one for ra = -1/2 and
pointing in opposite directions. It is usual, however, to consider only
the precession cone of the excess low-energy nuclei, and this is shown
in Fig. 1.2(b), which represents them as spread evenly over a conical
surface and all rotating with the same angular velocity around the
magnetic field axis, which is made the z axis. Since the excess low-
energy nuclear spins all have components along the z axis pointing
in the same direction, they add to give net magnetization Mz along
the z axis. Individual nuclei also have a component \Lxy transverse
to the field axis in the xy plane. However, because they are arranged
evenly around the z axis, these components all average to zero, i.e.
Mx = My = 0. The magnetism of the system is static and gives rise
to no external effects other than a very small, usually undetectable,
nuclear paramagnetism due to Mr

x My=Mv=0

(a) (b)

Figure 1.2 Freely precessing nuclei in a magnetic field B0. (a) Larmor precession of a single nucleus,
(b) The excess low-energy nuclei in a sample. The nuclear vectors can be regarded as being spread evenly
over a conical surface. They arise from different atoms but are drawn with the same origin.

www.pdfgrip.com
 The source of the NMR signal 9

In order to detect a nuclear resonance we have to perturb the system.

This is done by applying a sinusoidally oscillating magnetic field, Bl9
in the xy plane with frequency yB0/2^. This can be thought of as stim-
ulating both absorption and emission of energy by the spin system
(i.e. as stimulating upward and downward spin transitions), but
resulting in net absorption of energy, since more spins are in the low-
energy state and are available to be promoted to the high-energy state.
The perturbing field is generated by passing a radiofrequency (RF)
alternating current through Helmholtz double coils placed on either
side of the sample.
Classically we can analyse the oscillating magnetic field into a super-
position of two magnetic vectors rotating in opposite directions. These
add at different instants of time to give a zero, positive or negative
resultant (Fig. 1.3). The vector Bl5 which is rotating in the same sense
as the nuclei (Fig. 1.4), is stationary relative to them, since we have
arranged that it should have the same angular velocity. This induces
transitions from one energy level to the other. The result is that the
net nuclear magnetization, M, rotates about Bv There is resultant

Figure 1.3 The RF current in the coil in Fig. 1.4 produces an oscillating magnetic field along the y axis.
This can be equally well regarded as being composed of two rotating magnetic vectors rotating in oppo-
site directions, whose resultant is the oscillating magnetic field. One of these vectors will rotate in the
same sense as the nuclear precession and is conventionally called Bv

www.pdfgrip.com
10 The theory of nuclear magnetization

magnetization Mxy transverse to the main field and rotating at the

nuclear precession frequency. The magnetism of the system is no
longer static and the rotating vector Mxy will induce a radiofrequency
current in the coils placed around the sample, which we can detect,
provided pick-up of Bl can be avoided.
In order to follow the changes in the magnetization in the xy plane
with time, it is convenient to use the rotating frame. In Fig. 1.4, the
magnetic vector, Bl9 is rotating at the Larmor frequency. In the rotating
frame, the x and y axes rotate around the z axis at the Larmor
frequency and Bl is now apparently stationary. In order to differen-
tiate the rotating frame from the laboratory frame used in Fig. 1.4,
the x and y axes are described as x' and y'.
If Bl is of large amplitude, then the magnetization, M, swings very
rapidly around it. The magnetization in the xy plane increases and
reaches a maximum when the nuclear magnets have precessed 90°
around Bl (Fig. 1.5). When Bv is cut off, of course, this precession
stops. In practice, it is possible to cause a 90° precession in very short
times of between 2 and 50 JJLS (1 JJLS = 10~6 s), i.e. with a very short
pulse of Bl (Fig. 1.5).
If the Bl frequency is substantially different from the Larmor
frequency, Bl does not rotate at the same frequency as the nuclei
precess and the nuclear precession around Bl is always changing direc-
tion, so M can never become significant. It is this feature, whereby
the signal is obtained from all the excess low-energy nuclei acting in
concert, and only at a single frequency, that gives to the technique its
name of 'resonance spectroscopy'. However, this applies strictly only
to a Bl pulse which is monochromatic. The fact that we have arranged
for the Bl pulse to be very short means that it is no longer mono-
chromatic, but covers a band of frequencies and resonance occurs with

coil

M y =0
Figure 1.4 If a rotating magnetic vector Bl with the same angular velocity as
the nuclei is now added to the system, the nuclei will also tend to precess
around Bl9 and this causes the cone of vectors to tip and wobble at the nuclear
precession frequency. The resulting rotating vector Mxy in the xy plane can
induce a current in the coils that are wound beside the sample.

www.pdfgrip.com
 The source of the NMR signal 11

Figure 1.5 A sufficiently long or powerful B{ along the x' axis will turn the magnetization into the xy
(x'y') plane and the whole of the nuclear magnetization lies along the / axis. This is in the rotating frame
and My> rotates within the static laboratory frame, contributing to the signal picked up by the coil in Fig.
1.4 and the output is at a maximum. Bl is thus applied in the form of a pulse, and a pulse that has the
effect illustrated is known as a 90° pulse. In the rotating frame, Bl remains pointing in a fixed direction
and the magnetization rotates about this direction. In the laboratory frame, Bl rotates in the xy plane and
the magnetization follows a spiral path away from the BQ axis during the pulse.

all nuclei with precession frequencies situated within this band. The
strength of B\ at frequencies around the Bl frequency varies as sin(x)/x
(Fig. 1.6), though the intensity distribution is reasonably constant near
Br In fact, all nuclei within ±l/4PWHz of the spectrometer frequency
(and therefore of the Bl frequency) are almost equally affected. PW
is the pulse width or duration. There are, in contrast, null points at
±n/PW Hz where the nuclei are not perturbed. The negative intensi-
ties at higher frequency separation indicate a 180° phase change where
the nuclei would swing in the opposite direction.
Evidently, we are interested only in those nuclei in the central region
of the frequency distribution. For these, the magnetization in the labo-
ratory xy plane, M precesses about B0 with the Larmor frequency
and continues to do so after B1 has been switched off. This rotating

Figure 1.6 The effectiveness of a pulse of length, PW, against separation of

the pulse frequency from that of the signal to be excited.

www.pdfgrip.com
 12 The theory of nuclear magnetization

magnetization induces a signal in the coil which the spectrometer

detects with no interference from Bv The signal intensity diminishes
to zero as the system returns to equilibrium with Mxy = 0. An output
will be observable in general for between 1 ms and 10 s. If there is a
single type of nucleus present then the result is a decaying cosine wave
(Fig. 1.7). If there are several types of nuclei giving signals, then the
result is the sum of a number of decaying cosine waves. The detected
signals are subtracted from the spectrometer frequency giving a
resulting low frequency cosine wave which only contains the difference
frequencies between the spectrometer frequency and the nuclear fre-
quencies (Fig. 1.7). This is called a free induction decay, FID. Fourier
transformation of this function of time gives a function of frequency
which is the conventional NMR spectrum with maxima at each nuclear
difference frequency. This is discussed further in Chapter 5.

1.4 A BASIC NMR SPECTROMETER

We are now in a position to understand the principles underlying the

construction of an NMR spectrometer. The object is to measure
the frequency of a nuclear resonance with sufficient accuracy. The
instrument (Fig. 1.8) comprises a strong, highly stable magnet in which
the sample is placed and surrounded by transmitter/receiver coils. The
magnet is normally a superconducting solenoid.
The magnetic field at the sample also inevitably varies throughout
the bulk of the sample (i.e. the field is non-homogeneous), so that the
signal frequency is not well defined. Two further sets of coils, known
as shim coils, are placed around the sample in order to counteract
these variations or field gradients and render the field as perfectly
homogeneous as possible. The shim coils are not shown. Remaining
inhomogeneities are minimized by spinning the sample tube about its
long axis so that the sample molecules experience average fields. Very

Figure 1.7 A free induction decay, FID, from a single type of nucleus. This signal contains the same infor-
mation as a conventional NMR spectrum. The frequency of the signal relative to the spectrometer frequency
is given by the frequency of the cosine wave. The intensity is given by the intensity of the cosine wave at
its beginning, and the linewidth is given by the rate of decay of the cosine wave.

www.pdfgrip.com
 A basic NMR spectrometer 13

-* 1 Magnet I

M vy

|5|is pulse '—,' ' ,— |^?|

• » Probe ' ' i

Power rr—^—i
amplifier | lReceiverl

Puii^l fc] | Phase

timer I I ' sensitive
—7— | detector
• \
Crystal I 1/W^^
controlled Low \ ^
frequency frequency |
synthesizer fc-fn I Output
persists 5 s

[Computer [Data memory |

I

I Printer [/

Figure 1.8 A basic Fourier transform spectrometer. The 5 JJLS radiofrequency

(RF) pulse tips the nuclei in the sample by 90° provided their frequency lies
within the bandwidth of the pulse. (The pulse switching in effect converts the
monochromatic crystal frequency fc into a band of frequencies of width
± 1/4PW, where PW is the pulse length. In this case, the bandwidth is
± 50 000 Hz.) The nuclear output signal will be at a frequency fn close to fc
and the difference frequency fc - fn is obtained at the output of the phase-
sensitive detector. The computer collects the output and then calculates the
frequency of resonance relative to fc. The resonance frequency has limited
definition (i.e. it has width), which is related to the time for which the output
signal persists. Note that a time-dependent output is converted to a frequency-
dependent output for analysis. Note also the very different timescales for the
RF pulse and the output, here given the arbitrary lengths of 5 jxs and 5 s. The
computer and pulser have to be linked in some way to synchronize pulse
timing and data collection.

well defined frequencies, and so excellent resolution of close, narrow

resonances, are obtained in this way. The Bl field is produced by a
gated (switched input) power amplifier driven by a stable, crystal-
controlled continuous oscillator. The nuclear signals following the B1
pulse are then amplified and detected in a device which compares them
with the crystal oscillator output (Bl carrier fc) and gives a low-
frequency, time-dependent output containing frequency, phase and
amplitude information. This output is digitized and collected in a

www.pdfgrip.com
14 The theory of nuclear magnetization

computer memory for frequency analysis using a Fourier transform

program, and the spectrum that results (a spectrum is a function of
frequency) can be output and the resonance frequencies listed.
Not shown is a parallel spectrometer which can detect deuterium in
the sample and use this signal to stabilize the magnetic field. This is
a field-frequency lock, which will be described in detail later.
This type of spectrometer is known as a Fourier transform (FT)
spectrometer. It is also possible to obtain spectra from the more recep-
tive nuclei using a much simpler system, which dispenses with the
pulser and computer and produces the recording of the spectrum
directly. This is the continuous-wave (CW) spectrometer. All early
high-resolution spectrometers were of this type, and indeed the
modern FT instruments were developed from these.

1.5 QUESTIONS

1.1. A spectrometer operating at a fixed magnetic field is set up

to observe the nucleus 13C at a frequency of 25 000 000 Hz
(25 MHz). The sample examined contains two resonances at
25 000 250 Hz and 25 001 000 Hz respectively. What frequencies
will be present at the output to the computer digitizer? What
would these be if the spectrometer frequency were changed to
25 001 250 Hz?
1.2. A particular NMR spectrometer used to obtain 13C spectra at
25 MHz has a 90° pulse length of 100 JJLS. Calculate the bandwidth
(or frequency coverage) of this pulse where uniform excitation
of the nuclei is achieved. It is desired to obtain spectra over a
range of 5000 Hz. Is the bandwidth sufficient for this? If we were
able to increase the magnetic field of the spectrometer so that
the 13C operating frequency became 125 MHz (the frequency
range needed would be 25 000 Hz), would the 100 JJLS pulse still
be of sufficiently wide coverage? What is the maximum pulse
angle that could be tolerated if the whole spectral range were to
be stimulated uniformly?
1.3. A spectrometer observes 17O at 54.256 MHz. A 90° pulse of
length 40 JJLS was used. Explain why a signal 25 000 Hz from the
spectrometer frequency was not observed. Suggest two different
ways to change the spectrometer's operating conditions so that
this signal can be observed.
1.4. In which of the following cases are Mz and/or Mxy zero: (a)
following a 45° pulse; (b) following a 90° pulse; and (c) following
a 180° pulse?
1.5. Which of the following nuclei are NMR inactive: 32P, 32S, 36C1,
40
Ar, 40K, 64Zn, 91Zr, 76As, and 71Ge.
1.6. During a 90° pulse lasting 10 JULS, how many times does the magne-
tization vector M rotate as it spirals down to the xy plane if the
Larmor frequency is 400 MHz?

www.pdfgrip.com
 Questions 15

1.7. Calculate the angle at which a single proton will precess about a
magnetic field.
1.8. What is the receptivity of the lithium nucleus, 6Li, after enrich-
ment to 100%?
1.9. The nucleus, 1H, in water resonates at 400 MHz in a magnetic
field of 9.39 T. The earth's magnetic field is 0.000 05 T. What is
the !H precession frequency in the earth's magnetic field and
what is the excess population of nuclei in the lower energy state
in this field at 300 K?
1.10. A resonance in a given sample and spectrometer has a frequency
of 100 000 000 Hz. A continuous monochromatic Bl of frequency
100 000 002 Hz is applied in the xy plane to the sample. Because
Bl and nuclear frequencies are not the same, Mz will tend to
rotate around the z axis and never leave it to reach the xy plane.
What will be the frequency of rotation of Mz around the z axis
in the rotating frame?

www.pdfgrip.com
www.pdfgrip.com
 The magnetic field at
the nucleus: nuclear
screening and the
2
chemical shift

2.1 EFFECTS DUE TO THE MOLECULE

So far we have shown that a single isotope gives rise to a single nuclear
magnetic resonance in an applied magnetic field. This really would be
of limited interest to the chemist except for the fact that the magnetic
field at the nucleus is never equal to the applied field, but depends in
many ways upon the structure of the molecule in which the atom
carrying the nucleus resides.
The most obvious source of perturbation of the field is that which
occurs directly through space due to nuclear magnets in other atoms
in the molecule. If such nuclei have high magnetic moments, which
means generally !H or 19F, then in the solid state this interaction results
in considerable broadening of the resonance, which obscures much
information. However, in the liquid state, where the molecules rotate
rapidly and randomly, the direct nuclear fields fluctuate wildly in both
intensity and direction and have an average value that is exactly zero.
The resonances are thus narrow and may show much structure. Thus
spectroscopy of the liquid state has been a major preoccupation of
chemists.
Since the magnetic nuclei do not directly perturb the field at the
nucleus, we have therefore to consider the effect that the electrons in
the molecule may have. We will concern ourselves only with diamag-
netic molecules at this stage and will defer till later discussion of para-
magnetic molecules possessing an unpaired electron. When an atom
or molecule is placed in a magnetic field, the field induces motion of
the electron cloud such that a secondary magnetic field is set up. We
can think of the electrons as forming a current loop as in Fig. 2.1
centred on a positively charged atomic nucleus. The secondary field
produced by this current loop opposes the main field at the nucleus
and so reduces the nuclear frequency. The magnitude of the electronic
current is proportional to BQ and we say that the nucleus is screened

www.pdfgrip.com
18 The magnetic field at the nucleus

Figure 2.1 The motion of the electronic cloud E around the nucleus N gives
rise to a magnetic field, shown by dashed lines, which opposes BQ at the
nucleus.

(or shielded) from the applied field by its electrons. This concept is
introduced into equation (1.2) relating field and nuclear frequency by
the inclusion of a screening constant a

»d-,) (2.1)
ZTT
a is a small dimensionless quantity and is usually recorded in parts
per million (ppm). The screening effect is related to the mechanism
that gives rise to the diamagnetism of materials and is called diamag-
netic screening.
The magnitude of the effect also depends upon the density of elec-
trons in the current loop. This is a maximum for a free atom where
the electrons can circulate freely, but in a molecule the free circula-
tion around an individual nucleus is hindered by the bonding and by
the presence of other positive centres, so that the screening is reduced
and the nuclear frequency increased. Since this mechanism reduces
the diamagnetic screening it is known as a paramagnetic effect. This
is unfortunately a misleading term and it must be emphasized that it
does not imply the presence of unpaired electrons. As used here, it
merely indicates that there are two contributions to a, the diamag-
netic term crd and the paramagnetic term <rp and that these are opposite
in sign. Thus
o- - crd + ap (2.2)
Since the magnitude of <rd depends upon the density of circulating
electrons, it is common to find in the literature discussion of the effect
of inductive electron drifts on the screening of nuclei. The screening
of protons in organic molecules, for instance, depends markedly on
the substituents, and good linear correlations have been found between
screening constants and substituent electronegativity, thus supporting

www.pdfgrip.com
 Effects due to the molecule 19

the presence of an inductive effect. Currently, however, it is believed

that most of these variations originate in the long-range effects to be
described below and that the contribution of inductive effects is small,
at least for cr-bonded systems.
The magnitude of the paramagnetic contribution crp is zero for ions
with spherically symmetric S states but is substantial for atoms, particu-
larly the heavier ones, with many electrons in the outer orbitals
involved in chemical bonding. It is determined by several factors.
1. The inverse of the energy separation A£ between ground and
excited electronic states of the molecule. This means that correla-
tions are found between screening constants and the frequency of
absorption lines in the visible and ultraviolet.
2. The relative electron densities in the various p- and d-orbitals
involved in bonding, i.e. upon the degree of asymmetry in electron
distribution near the nucleus.
3. The value of (1/r3), the average inverse cube distance from the
nucleus to the orbitals concerned.
In the case of hydrogen, for which there are few electrons to
contribute to the screening, and for which AE is large, crd and ap are
both small and we observe only a small change in a among its
compounds, most of which fall within a range of 20 x 10~6 or 20 ppm.
In the case of elements of higher atomic number, AE tends to be
smaller and more electrons are present, so that, while both crd and crp
increase, ap increases disproportionately and dominates the screening.
Thus changes in crp probably account for a major part of the screening
changes observed for boron in its compounds (a range of 200 ppm),
and (Tp almost certainly predominates for fluorine (where the range is
1300 ppm) or for thallium (where it is 5500 ppm). We see that the
changes observed for the proton are thus unusually small.
In the case of screening of the fluorine nucleus, the values for fluo-
rine compounds are known relative to the bare, unscreened fluorine
nucleus. In certain of its compounds, fluorine is less screened even
than the bare nucleus, some examples being F2, UF6 or FOOF.
Presumably this arises because of electronic circulation near to, but
not centred upon, the nucleus.
The observable changes in screening of each nucleus do not increase
continuously with atomic number but exhibit a periodicity, increasing
steadily along each period but then falling markedly at the start of the
next. The behaviour down each group is similar, as is shown in
Fig. 2.2. This periodicity follows closely the values of (1/r3) for each
element.
Because we are dealing with the effects of electronic circulation in
a three-dimensional molecule relative to a unidirectional magnetic
field, it is easy to see that the nature of the circulation will change as
the orientation of the molecule changes in the magnetic field. Thus
the screening at any instant is a function of the attitude of the mole-
cule relative to J?0. These orientational effects are fully described by

www.pdfgrip.com
 20 The magnetic field at the nucleus

90001- - v V •y.

8000-
7000
6000
ppm 5000
4000 -
'-
3000
2000
1000H
0 Hi Li , 1 , ,1 t T. iY 1 A
B N F Nal Al p Cll Kl fcdGd As[BrlRbl Ag InSb \ Cd "Adrri
HeBe (: c) NeMg S >i J5 ArCa ZnG eS.e Kr Sr CdSnTeXeBa H gPb

Figure 2.2 Ranges of screening constants for nuclei of main-group and post-transition elements. It should
be remembered that, while the ranges shown reflect the periodicity of the (1/r3) term, they also are influ-
enced by how many compounds of a given element have been measured and, indeed, by the extent of its
chemistry. (After Jameson and Mason (1987) Multinuclear NMR, Plenum, New York, with permission.)

the screening tensor, which has nine components, though only three
influence the observed screening. These, the diagonal components of
the tensor, are called crn, cr22 and cr33 and all three are required
if the nucleus is in a site with no symmetry. If the site is axially
symmetric, then only two values are needed to describe the screening
since an = a22. These components are then called a1 and a33 is denoted
a,,. This is called the screening anisotropy, and the differences between
the values of the components can be substantial.
Screening anisotropy of a nucleus can be observed by taking its
spectrum from a powdered solid where the solid particles all have
different orientations. Two types of spectrum are shown diagram-
matically in Fig. 2.3, and the values of the tensor components are
given by the discontinuities on the curves. The effect is another
source of line broadening in solids but also allows a full description
of the screening mechanism to be obtained. In liquid samples,
the isotropic rotation of the molecules produces an average a where
a = (an + a22 + a33)/3 and the lines become narrow. The high-
resolution spectrum that is thus obtained is generally more useful to
the chemist, but it should not be forgotten that some fundamental
information is lost in the process.

www.pdfgrip.com
 Effects due to the molecule 21

(a) (b)

°11 C?22 <*33 <J|| Si

Figure 2.3 Representations of the powder spectra of two different solid

samples (a) of a nucleus in an asymmetric environment, and (b) of a nucleus
in an axially symmetric environment. Broadening due to through-space
magnetic interaction with other nuclei in the samples is avoided by choosing
samples where protons, for example, are relatively distant.

The variations in screening among the compounds of a given element

depend upon all the factors summarized above, but in a way that is
very difficult to separate into dominance by any particular influence.
For instance, it seems certain that increased charge density upon an
atom results in increased screening of its nucleus but that this occurs
principally because the (1/r3) term is increased; in other words, the
change in orbital radius determines the change in screening. Two
examples serve to emphasize the complexity of the situation. As
expected from simple considerations of charge density, the 14N nucleus
in NH3 is 25 ppm more shielded than in [NH4]+, the influence of the
apparent change in coordination number being slight since NH3 has
an electron lone pair as effective fourth ligand. In contrast, the 14N
nucleus in the pyridinium ion, (2.1), is 100 ppm more shielded than (2.1)
that in pyridine, (2.2). In this case, the nitrogen is part of a TT system
and protonation removes an accessible low-energy transition and so
modifies the A£ term.
Oxidation state also affects screening via electron density changes.
The influence of electron imbalance in the bonds is manifested by
the common effect in which increased screening follows increases
in coordination number; for example 31P screening in PC13 (2.2)
< [PC14]+ < PC15 < [PC16]~, and several similar trends will be observed
in the scales of Fig. 2.14. It must, however, be emphasized that
this effect is not universally true and other considerations operate
in some cases. In all the normal cases, the increase in coordina-
tion number can be regarded as leading to an increase in local
symmetry and so a more balanced structure. Electron imbalance is
also related to bond ionicity, ir-bond order or 5- character and so to
bond angles. Thus the 31P nucleus is 125 ppm more screened in
trimethylphosphine, PMe3, than in the sterically crowded tri-t-
butylphosphine, PBu^, because of the increased C-P-C bond angles
in the latter.

www.pdfgrip.com
 22 The magnetic field at the nucleus

Substituents also exhibit a marked effect upon nuclear screening,

which often correlates with changes in substituent electronegativity.
However, the direction of the correlation is not uniform, and screening
may increase or decrease with increase in electronegativity depending
upon the sequence of ligands chosen. Again, several effects can operate
simultaneously. For instance, the 27A1 nucleus is increasingly screened
in the series of tetrahalo anions [A1C14]~ < [AlBr4]~ < [A1I4]~. This is
known as the normal halogen dependence, since it is in the opposite
sense for certain transition metals. The change in screening between
chloride and bromide (22 ppm) is less than that between bromide and
iodide (47 ppm). The changes are ascribed first to the nephelauxetic
effect of the halogen, the larger halogens expanding the electron
orbitals on the aluminium and so decreasing the (1/r3) term and so ap.
In addition, the heavier halogens produce the heavy-atom screening
effect, which is proportional to Z4, where Z is the nuclear charge of
the halogen, and this enhances the apparent nephelauxetic effect and
explains the bigger increase in screening between the bromide
and iodide. The heavy-atom effect operates via a complex relativistic
spin-orbit coupling mechanism.
Usually the contributions to crd and crp for a nucleus are considered
only for the electrons immediately neighbouring, or local to, that
nucleus. More distant electrons give rise to long-range effects on both
ad and <7p, which are large but cancel to make only a small net contri-
bution to a. It is therefore more convenient to separate the long-range
effects into net contributions from different, quite localized, parts of
the rest of the molecule. Two types of contribution to screening can
be recognized and, though they are small, they are of particular impor-
tance for the proton resonance.

2.1.1 Neighbour anisotropy effects

We have already mentioned that in liquid samples, owing to the rapid
and random motion of the molecules, the magnetic fields at each
nucleus due to all other magnetic dipoles average to zero. This is only
true if the magnet (e.g. a nucleus) has the same dipole strength what-
ever the orientation of the molecule relative to the field direction. If
the source of magnetism is anisotropic and the dipole strength varies
with orientation in the applied field, then a finite magnetic field appears
Figure 2.4 Screened and at the nucleus.
descreened volumes of Such anisotropic magnets are formed in the chemical bonds in the
space around a carbonyl
bond. The sign + indicates molecule, since the bonding electrons support different current circu-
that a nucleus in the space lation at different orientations of the bond axis to the field. The result
indicated would be more is that nuclei in some parts of the space near a bond are descreened
highly screened. The mag- while in other parts the screening is increased. Figures 2.4 and 2.5
nitude of the screening show the way screening varies around some bonds.
falls off with increasing
distance from the group A special case of anisotropic screening where the source of the
and is zero in the surface anisotropy is clearly evident occurs in aromatic compounds, which
of the solid figure. exhibit what is called ring current anisotropy. The benzene structure,

www.pdfgrip.com
 Effects due to the molecule 23

for instance, can support a large electronic ring current around the
conjugated ir-bond system when the plane of the ring is transverse to
the field axis but very little when the ring lies parallel to the field axis.
This results in large average descreening of benzene protons since the
average secondary magnetic field, which must oppose the applied field
within the current loop, acts to increase the field outside the loop in
the region of the benzene protons (Fig. 2.6).

2.1.2 Through-space electric field effects

Molecules that contain electric dipoles or point charges possess an Figure 2.5 Screened and
electric field whose direction is fixed relative to the rest of the mol- descreened volumes of
ecule. Such electric fields can perturb the molecular orbitals by causing space around a carbon-
electron drifts at the nuclei in the bond directions and by altering the carbon double bond. The
significance of the signs is
electronic symmetry. It has been shown that the screening &E due to the same as in Fig. 2.4.
such electric fields is given by The cone axis is perpen-
dicular to the plane
<T£ = - AEZ - BE2 (2.3) containing the carbon and
where A and B are constants, A » B, Ez is the electric field along a hydrogen atoms.
bond to the atom whose nuclear screening we require and E is the
maximum electric field at the atom. The first term produces an increase
in screening if the field causes an electron drift from the bond onto
the atom and a decrease if the drift is away from the atom. The second
term leads always to descreening. It is only important for proton
screening in the solvation complexes of highly charged ions where

(a) (b)

Figure 2.6 Ring current descreening in benzene, (a) The plane of the benzene
ring is at right angles to the applied magnetic field B0. A current is induced
in the delocalized Tr-orbitals generating an opposing magnetic field in the
centre of the ring and a reinforcing magnetic field outside the ring. The result
is that protons above the ring move to lower frequency, while those outside
the ring, in the plane of the ring, move to higher frequency, (b) When the
applied magnetic field lies in the plane of the benzene ring, the area of the
ring current loops due to the electrons is much smaller.

www.pdfgrip.com
24 The magnetic field at the nucleus

E can be very large, though it is of greater importance for the nuclei

of the heavier elements.
The electric field effect is, of course, attenuated with increasing
distance. It is an intramolecular effect since the effect of external fields,
for which the BE2 term can be neglected, averages to zero as the mol-
ecule tumbles and E continually reverses direction along the bond.
The descreening of protons that occurs in many organic compounds
containing electronegative substituents X probably occurs because of
the electric field set up by the polar C-X bond. This will also increase
with X electronegativity and produce a similar result to an inductive
electron drift. The effect typically produces proton descreening in
molecules containing keto, ester, or ether groups and in halides, some
values being given below.

Molecule CH3C1 CH3-C(O)O-CH3 (CH 3)2co (CH3)20

ppm less screened 2.83 1.78 3.44 1.86 3.01
than methane

The closer the protons are to the bond generating the electric field,
then the more they are descreened. This also shows up as a fall-off in
descreening along an alkyl chain for the protons further away from
the substituent, and, for instance, in 1-chloropropane (n-propyl chlo-
ride) the comparable descreening figures are a-CH2 3.24 ppm, (3-CH2
1.58 ppm and CH3 0.83 ppm.

2.2 ISOTOPE EFFECTS

Because the bonds in molecules are not rigid fixed entities but have
dimensions determined by vibrational phenomena, the substitution of
an atom by one of its isotopes of different mass alters the vibrational
energies in the bonds to that atom and so alters the electron distrib-
ution about it. This necessarily implies changes in nuclear screening
following the substitution, both near the substitution site and at some
distance from it. Such changes are small but measurable in many cases,
and provide useful spectroscopic data. It is necessary to distinguish
between primary isotope effects and secondary isotope effects. The
former are the effects experienced by the isotopically substituted
nucleus itself. For instance, for 14N/15N substitutions, the change in 14N
screening between 14NH3 and CH314NO2 is not exactly the same as that
for 15N between 15NH3 and CH315NO2, given identical conditions.
Clearly, such changes are not easy to measure and for many elements
are within experimental error, so that primary effects will not be
considered further. The secondary effects are those observed on the
nuclei in the rest of the molecule; in the above example the proton
screening changes, say, between 14NH3 and 15NH3. The trends observed
are as follows.

www.pdfgrip.com
 Isotope effects 25

1. The isotope shift is greatest when the substitution causes the

greatest fractional change in mass; thus ^^H substitution is the
most effective.
2. Screening is greatest for nuclei near the heavier isotope. This is not
always true and in some cases the opposite holds. The nearer the
nucleus observed is to the substituted site, the greater is the effect.
3. The magnitude of the effect reflects the overall range of screening
experienced by the observed nucleus (cf. Fig. 2.2).
4. The effect of multiple isotopic substitutions is additive, or approx-
imately so.
Thus the 15N signal obtained from the nitrite ion, [15NO2]~, in water
and which has had a partial substitution of 16O by 18O, fully random-
ized over the two positions, consists of three signals as shown in
Fig. 2.7. This spectrum shows the additive effect and the greater
screening due to the 18O. It also proves unequivocally that the nitrite
anion contains two oxygen atoms. Similarly, the 31P spectrum of
partially 18O substituted [PO4]3~ is an equally spaced quintet with
spacing of 0.0206 ppm and proves that the orthophosphate contains
four oxygen atoms.
This property of partial isotopic substitution provides an almost
digital technique for measuring numbers of exchangeable atoms in a

[15N1802r

[l5N16o18or

I I I I I I I
0.2 0.0 -0.2 -0.4 -0.6
8/PPM
Figure 2.7 The 15N NMR spectrum of the nitrite ion, [NO2]~, in water. The
ion was enriched to 95% in 15N and 1577% 16
in 1815O. 16The18 three resonances
15 18
arise
from ions of isotopic composition N O2, N 16 O O and N O2 in the
concentration ratios 6:33 :61. Replacement of O by 18O causes a low
frequency shift of 0.138 ppm. (From Van Etten and Risley (1981) /. Am.
Chem. Soc., 103, 5634; copyright (1981) American Chemical Society, reprinted
with permission.)

www.pdfgrip.com
 26 The magnetic field at the nucleus

structure. It is, for instance, now possible to count the number of water
molecules in aqua complexes such as [A1(H2O)]3+. Analytical methods
never give whole numbers, and while the hexaaqua cation is known
to exist in solids from X-ray structural analysis, it is possible to argue
that the hydration number may not be so definite in solution. If
A1(H2O)6(C1O4)3 is dissolved in acetone, it is possible to observe the
water proton signal and note that this is highly descreened relative
to free water. This is a consequence of the strong electric field of
the cation. If a proportion of the water is replaced by D2O, then the
complex will contain H2O, HOD and D2O. The signal of the HOD
protons shows a strong isotope effect, though this is abnormal as they
are less screened even than the H2O. This probably arises because the
deuterium substitution shortens the Al-O bond in that molecule and
so increases the electric field effect.
More importantly, the two proton resonances (HOD and HOH)
show fine structure as in Fig. 2.8. This arises because of long-distance

-20 -40 -60 -80 -100

1 59
10.3 10.2 10.1 H/ppm Co/ppm
(a) (b)

Figure
2
2.8 (a) The 400 MHz 1H NMR spectrum of the water complexed to the cation [A1(H2O)]3+ in
( H6)acetone (deuterioacetone or acetone-d6) taken at -30°C. The complex had been partially deuteriated
so as to contain 35% 2H. The resonance at 10.23 ppm is due to all the HOD molecules in the complex
and that at 10.17 ppm is due to all the HOH molecules. The fine structure arises because different mole-
cules have different total numbers of 2H, each giving a smaller isotope effect due to the more distant
substitution. The stick diagram gives the calculated intensities obtained from the deuterium content and
assumes completely random distribution throughout the sample. (After Akitt et al. (1986) /. Chem. Soc.
Chem. Commun., 1047, with permission.) (b) The 95.7 MHz 59Co NMR spectrum of [Co(NH3)6]3+ in various 3+
mixtures of D2O/H
3+
2O to generate the complete range of nineteen deuterated compounds from [Co(NH3)6]
to [Co(ND3)6] . The introduction of each deuterium moves the signal approximately 6 ppm to low
frequency, (i) 15% D2O/85% H2O, (ii) 50% D2O/50% H2O, (iii) 85% D2O/15% H2O. The number besides
each signal refers to the number of hydrogen atoms that have been replaced by deuterium. (Remeasured,
but based on Russell and Bryant (1983) Anal Chim. Ada, 151, 227, copyright (1994), with permission
from Elsevier Science.)

www.pdfgrip.com
 Effects due to unpaired electrons 27

isotope effects between a given HOD and the other ten replaceable
sites in the complex. A 13-line pattern is theoretically possible, though
some lines are too weak to detect, and one has to calculate an inten-
sity distribution from the known level of deuteration.
The changes can be even more dramatic when a nucleus with a
larger chemical shift range is observed. This is shown in Fig. 2.8(b)
for 59Co in [Co(NH3)6]3+, where replacement of hydrogen by deuterium
produces a shift of about 6 ppm for each replacement.

2.3 EFFECTS DUE TO UNPAIRED ELECTRONS

The electron (spin = 1/2) has a very large magnetic moment and if,
for instance, paramagnetic transition-metal ions are present in the
molecule, large effects are observed. The NMR signal of the nuclei
present may be undetectable, but under certain circumstances, when
the lifetime of the individual electron in each spin state is short, so
that its through-space effect averages to near zero, NMR spectra can
be observed. The screening constants measured in such systems,
however, cover a very much larger range than is normal for the
nucleus, and this arises because the electronic spins can be apparently
delocalized throughout a molecule and appear at, or contact, nuclei.
The large resonance displacements that result are known as contact
shifts and the ligands in certain transition metal-ion complexes
exhibit proton contact shifts indicating several hundred ppm changes
in a. In addition, if the magnetic moment of the ion is anisotropic,
one gets a through-space contribution to the contact interaction similar
to the neighbour anisotropy effect, and this is called a pseudo-contact
shift.
The NMR signals are also broadened by the presence of the
unpaired electron(s). The broadening is proportional to y2/r6, where y
is the gyromagnetic ratio of the nucleus being observed and r is the
distance between the unpaired electron(s) and the observed nucleus.

2.3.1 Paramagnetic transition metal compounds

For the vast majority of paramagnetic transition metal compounds, the
presence of unpaired electrons produce substantial line broadening. In
favourable cases, the line broadening is small compared with the shifts
and well resolved signals are observed. For instance, in Fig 2.9 the 1H
NMR spectrum of [Co(4,6-Me2-phenanthroline)3]2+, (2.3), is shown.
The ion consists of two isomers depending on whether the nitrogen
atoms in the 1 position on the phenanthroline ring are arranged mer
orfac. In the/ac-isomer all the three phenanthroline ligands are equiv-
alent giving rise to one set of signals, but in the raer-isomer they are
all inequivalent. Hence four sets of ligand signals are observed in a
mixture of both isomers. The signals are substantially shifted due
to the presence of the unpaired electrons. This is particularly marked

www.pdfgrip.com
28 The magnetic field at the nucleus

I I I I I I
140 120 100 80 60 40 20 0
1
H/ppm
1
Figure 2.9 The 100 MHz H NMR spectrum of [Co(4,6-Me2-phenanthro-
line)3]2+, (2.3), in CD3OD at -20°C. Note that the compound consists of two
isomers, mer and fac. (Reproduced by permission of the American Chemical
Society from La Mar and Van Hecke (1970) Inorg. Chem., 9, 1546.)

for the H2 and H9 signals which are moved by some HOppm. Note
that despite the presence of the unpaired electron, the signals are rela-
tively sharp.
However, in other cases, the 1H NMR signals may be so broad
that separate signals cannot be resolved. Even then all is not lost.
As the broadening depends on y2, the solution to the problem is to
observe a low y nucleus. 7(2H)2 is only 0.024 that of :H resulting in
much sharper signals, and the result is that deuterated samples of Cr3+
and Cu2+ compounds which fail to give usable !H NMR spectra can
give useful information in their 2H NMR spectra. For example, in
[Cr3(M,-OH)2(ji-02CCR3)4(02CCR3)2(bipy)?][C104], R - H or D, the
resolution of the acetate signals is greatly improved when R = D and
the 2H NMR spectrum is recorded compared with the 1H NMR spec-
trum, when R = H (Fig. 2.10), where all the acetate signals are observ-
able giving five peaks in the ratio 2:1:1:1:1.

www.pdfgrip.com
 Effects due to unpaired electrons 29

3D2HCN

CDHoCN

40 30 20 10
2
H/ppm

50 0 -50
1
(a) H/ppm (b)

Figure 2.10 The 360 MHz 1H, (a), and 76.75 MHz 2H, (b), NMR spectra of [Cr3(fji-OH)2(^-O2CCR3)4
(O2CCR3)2(bipy)2][ClO4], R - H or D, in CD3CN, (a), or CH3CN, (b). In spectrum (b), only the signals
due to the deuteriated acetate ligands and the residual deuterium in the solvent are observed. Note the
better resolution observed for the acetate signals at 16 and 42ppm in the 2H NMR spectrum when
compared with the !H NMR spectrum. The numbered resonances in (a) arise from the bipyridyl ligand.
(Reproduced by permission of the American Chemical Society from Harton et al (1997) Inorg them, 36,
4875.)

As the broadening by the unpaired electrons falls off rapidly with

distance, varying as r6, only the nuclei close to the metal are badly
broadened. The result is that for large molecules the !H NMR spec-
trum can be run for the metal free molecule, and then a metal such
as Mn2+ or Cu2+ added, so that the 1H NMR signals for protons close
to the metal binding site become badly broadened and in effect vanish
from the spectrum. This approach has been used to identify protons
close to metal binding sites of enzymes.

2.3.2 Paramagnetic lanthanide and actinide compounds

Many of the paramagnetic lanthanides and actinides give relatively
sharp !H NMR spectra. For example, many UIV compounds give sharp
signals (Fig. 2.11).
Such effects are particularly marked with the lanthanide cations,
whose complexes are commonly used to simplify the spectra of organic
compounds. These substances are called shift reagents and consist of
a lanthanide element (Pr, Eu, Dy, or Yb) complexed with an organic
ligand chosen, among other things, to make the complex soluble in
organic solvents so that it can be codissolved with the compound to
be investigated. They are octahedral complexes, but the lanthanides
are capable of assuming higher coordination numbers than six, so
that if the organic molecule possesses a suitable donor site, such as
oxygen or nitrogen, it can interact with the shift reagent. This produces
a pseudo-contact shift of all the proton and carbon nuclei in the

www.pdfgrip.com
 30 The magnetic field at the nucleus

1
20 10 0 -10 -20 -30 H/ppm
Figure 2.11 (a) 80MHz *H NMR spectrum of [(Ti5-C5H5)3UC(O)CH2CH2
CH2CH3] in C6D6. (b) 1H NMR spectrum of a mixture of [(T]5-C5H5)3UC
(O)CH2CH2CH2CH3] and [(ti5-C5H5)3UCH2CH2CH2CH3] in C6D6, the latter
giving the primed signals. (Reproduced from Paolucci et al (1984) /.
Organomet. Chem., 272, 363, copyright (1984), with permission from Elsevier
Science.)

molecule, which can represent very large changes in screening. For

instance, the normal proton spectrum of pentanol consists of five
signals with the signals due to the 7-CH2 and 8-CH2 protons being
coincident (Fig. 2.12). If [Eu(fod)3], fod - 6,6,7,7,8,8,8-heptafluoro-2,2-
(2.4) dimethyl-3,5-octanedionate, (2.4), is added, all the signals move to high
frequency. Alternatively, if [Pr(fod)3] is added, all the signals move to
low frequency. The extent of the movement depends on the distance
from the lanthanide to the proton, so the a-CH2 moves most and the
CH3 group moves the least. The OH proton has moved out of the
spectral region shown in Fig. 2.12. In both cases, the overlapped signals
due to the 7- and 8-CH2 protons are resolved. The fine structure is
due to spin-spin coupling, which we will meet in the next chapter.
If the organic molecule is rigid, then the magnitude of the shifts
can be used to calculate its geometry using an interactive method
that optimizes both geometry and position of the lanthanide. This is
normally supported by also using the corresponding gadolinium
complex to broaden the signals, and the signal broadening is propor-
tional to r~6, where r is the distance between the broadened proton
and the metal.
A second use of these reagents is in determining chirality or optical
purity of organic substrates. This is achieved by preparing chiral shift
reagents, i.e. ones in which the ligands are themselves chiral. Such
reagents produce slightly different screening effects in substrates of
different handedness. It is, for instance, possible to obtain separate
signals for the two optical isomers of C5HnCHDOH, which can be
prepared using optically active reducing agents. Figure 2.13 shows the
very small separation between the signals of the two forms of some
0.07 ppm.

www.pdfgrip.com
 Effects due to unpaired electrons 31

<al

(b)

(c)

5 4 31 , 2 1 0
H/ppm

Figure 2.12 Three 400 MHz 1H NMR spectra of 1-pentanol in CDC13. (a)
With added [Eu(fod)3]. (b) No added shift reagent, (c) With added [Pr(fod)3].
Note that fod is 6,6,7,7,8,8,8-heptfluoro-2,2-dimethyl-3,5-octanedionate (2.4).
Signals due to the fod ligand are also observed.

www.pdfgrip.com
32 The magnetic field at the nucleus

(C)

(b)

(a)

Figure 2.13 The 2H NMR signals of the CHD deuterons in various samples
of the optically active hexanol, C5HUCHDOH, made using an optically active
reducing agent and obtained in the presence of a chiral shift reagent, in which
the ligands are hexafluoropropyl camphor ate (see formula). The lower trace
shows the racemic mixture and the two upper traces are from materials made
from reducing agents of opposite chirality.

2.4 THE CHEMICAL SHIFT

So far we have talked only in terms of the changes in the screening

of nuclei in different environments. It is, however, more usual to
describe these changes as 'chemical shifts' although the word
'screening' will still be encountered from time to time. It is usual to
calibrate chemical shifts by using a suitable compound as a marker
resonance. Thus if we have unknown and marker resonances in
different environments with screening constants o^, and cr2, then the
two nuclear frequencies in a given magnetic field BQ are
Vi= (1 CTi) (2 4a)
fr ~ -
"2 = ^ (1 - *2) (2'4b)

whence
V l _ V 2= ^(a2-ai) (2.5)

It is possible to measure a frequency with very high precision but

not a magnetic field strength so this expression is not of great use. We

www.pdfgrip.com
 The chemical shift 33

thus eliminate field from the equation by dividing through by v1 (equa-

tion (2.4a)). This gives us the frequency change as a fraction of v1
V
i ~ v 2 _ q-2-cTi n~
(2 6)
v, "I-*, '
which since a « 1 in many cases reduces to

^ = .,-., (17)
V
l

when the chemical shift range is small, although the approximation

cannot be used for large chemical shift differences as is found for many
of the heavier nuclei.
Thus the fractional frequency change is the same as the difference
in screening in the two nuclear environments. This is called the chem-
ical shift and is given the symbol 8. Its value is expressed in parts per
million (ppm). It can be determined with high accuracy since it is
possible to resolve shifts of 0.001 ppm for spin-1/2 nuclei, or even less
in favourable cases.
In order to establish a chemical shift scale for a given nucleus, it is
necessary to choose some substance as a standard and define its chem-
ical shift as zero. The usual, almost universal, standard for the four
nuclei, the proton (1H), deuterium (2H), carbon (13C) and silicon (29Si),
is tetramethylsilane, (CH3)4Si, usually called TMS. It gives a narrow
singlet resonance for protons, and for the other three nuclei if decou-
pling is used (see Chapter 7 for an explanation of this), which in all
cases is outside the normal range of chemical shifts of the compounds
studied. It is miscible with most organic solvents, it is inert and, being
highly volatile, can easily be removed after measurements have been
made.
In the case of 1H and 13C spectroscopy, the resonances of interest
come predominantly from nuclei that are less screened than is TMS,
the main exceptions occurring where metal atoms are present in the
compounds studied.
It is usual in recording spectra to depict the descreened region to
the left-hand side of the spectrum with TMS to the right. The
frequency then increases to the left of TMS since the magnetic fields
at descreened nuclei are apparently higher and the nuclear precession
frequency is higher. For historical reasons, however, this region is very
commonly referred to as 'low field', the aptness of this name being
apparent from equation (2.1). Both the names 'low field' and 'high
frequency' are met in practice, the latter being the most logical since
it is indeed the frequencies that are measured. The older name, though,
does explain the apparently eccentric way in which the shifts are
displayed increasing to the left. This is summarized in Fig. 2.14.
Because we have chosen our standard arbitrarily, we also find that we
have introduced sign into the scale. Thus in 1H spectroscopy all those
protons low frequency of TMS (high field) have negative shift values.
This scale is called the 8 scale, and it is important to note that the

www.pdfgrip.com
34 The magnetic field at the nucleus

TMS

Less screened than TMS More screened than TMS

Low field High field

High frequency Low frequency

Most organic protons Metal hydrides

—I
510 0 -10
Figure 2.14 Summary of proton chemical shift scales.

symbol 8 may either refer to this scale or simply be used as short-

hand for 'chemical shift'.
In early work, the sign convention that low frequency (high field)
is positive was often used and this led to the 'tau' scale for !H NMR
spectroscopy, where T = 10 - 8. During the 1970s the sign convention
that high frequency (low field) is positive became generally accepted,
and became the rule as a result of decisions by IUPAC in 1972 and
1976. Unfortunately, the literature prior to this ruling can be very
confusing with chemical shifts quoted using either sign convention. The
result is that great caution is necessary concerning the sign of chem-
ical shifts published prior to about 1975. The problem is particularly
severe for 19F and 31P chemical shifts.
It is, of course, possible to use subsidiary standards, which are
referred in turn to TMS. !H spectroscopy in aqueous solution thus
often has recourse to either (CH3)3Si(CH2)3COOH or [(CH3)4N]+.
Calculation of a chemical shift is a simple matter. Modern spec-
trometers give the frequency separation of signals and it is only neces-
sary to divide by the spectrometer operating frequency. If in a given
solution the signal of benzene is 2901 Hz higher in frequency than
TMS and the frequency of TMS is 400.134 394 MHz (i.e. BQ = 9.39 T),
then the chemical shift is

8
- T«ilf394 x W - W S f f m M
Or, more easily remembered, it is the frequency difference in hertz
divided by the frequency of the reference in megahertz. We see also
that the frequency separations will be different in spectrometers oper-
ating at different frequencies. Thus in a 200 MHz spectrometer, the
frequency separation above will be 1450 Hz.
The conventions adopted for other nuclei are less firm. The shifts
are usually large, so that it is not quite so important to be able
to compare different workers' results with high accuracy, and the
standard substance is often chosen according to the dictates of

www.pdfgrip.com
 Sample preparation, standardization, solvent and temperature effects 35

H bonded
e.g. CHC13
'Hppm HF; C 6 H 6 C 2 H 4 (CH3)2O (CH3)2CO CH4 TMS(O) MH
. 1 . 1 _J ^ .
to-20 7.2 5.5 3.2 2.1 0.2 Oto-50

BMe3 BC13 (MeO)3B Et 2 OBF 3 'BH; Bi;

n
Bppm I 1 1 -j » '
84.3 47.6 18.3 0 -43 -127
Boron hydrides
Metal carbenes Acetic acid Acetone
andcarbynes CS2 CO C 6 H 6 CHC13 CH 3 TMS
c
PP™ ^ | , |
to 400 192.8 178.3 128.6 77.2 30.4 0
NO;
NOr NOF MeNO2 RSCN RNC NH^ NH 3
Nppm
, i ' I I - » j
250 100 0 -90 -200 -355 -382

MeOH
MnO; CrO^~ Me2CO Ni(CO)4 SO^- H2O j Me2O
17
0 ppm ^ 1 1 1 1 1 T~^
1230 835 569 362 167 0 -53
Heteropolyanion oxygens span whole range

FOOF F2 B WF6 CFC13 CF3Ph C6F6 F~ MeF GIF

865 422 162 0 -63.9 -162.9 -200 -271.8 _448

Bu l 3 Al Al 2 Me 6 A1C1; Al(OH); A1(H 2 O)^ + Al(MeCN)^ +

27
A1 ppm —I 1 1 1 1 1
255 156 103 80 0 -33
H3P04
PBr3 P(MeO)3 Me3PO PMe3 PBr 5 PH 3 P4
31 __i i i i
Pppm -j '
227 141 36 0 -62 -101 -238 -461

CIO; SO2C12 CC14 PC13 SiCl4 Cl"(aq) (CH2C1)2

35 _ j-1-1-L_-1-1-1_
C1 ppm
1003 760 500 320 174 0 -80

Co(H20)36+ Co(NH 3 ) 3+ Co(CN)^- Co(CO); Co(PF3);

59
Co ppm
15100 8150 0 -3200 -4200

Figure 2.15 Various chemical shift scales and chemical shifts of some compounds.

www.pdfgrip.com
36 The magnetic field at the nucleus

convenience. The standard is assigned 0 ppm and the 8 scales are then
as in Fig. 2.15. An aqueous salt solution is often used as standard for
groups 1, 2, 3, and 17: 7Li+ (aq) for lithium, [27A1(H2O)6]3+ for
aluminium and 35C1~ (aq) for chlorine, for instance. Some much used
references are: (CH3O)3B and (CH3CH2)2O -»BF3 for n B spec-
troscopy; nitromethane, CH3NO2, or nitrate ion, [NO3]~, for 14N and
15
N spectroscopy; H2O for 17O spectroscopy; 85% orthophosphoric
acid, H3PO4, for 31P spectroscopy; and the refrigerant CFC13 is
commonly used as standard for 19F work. This 19F chemical shift scale
is often called the 4> scale. Other standards used for 19F spectroscopy
are hexafluorobenzene, C6F6, or trifluoroacetic acid.
These and some other chemical shift scales are illustrated in Fig.
2.15, which shows the standards used (0 ppm) and the chemical shifts
of some compounds, chosen to cover the full range of shifts for a given
element rather than to pick out any particular trends with composi-
tion. For many of the heavier elements, the shift is very medium sensi-
tive and the spot value given is purely illustrative. For the 19F scale,
the shift marked B is that of the bare nucleus (189 ppm). Note also
the very large chemical shift range for the transition-metal nucleus
59
Co.

2.5 NOTES ON SAMPLE PREPARATION,

STANDARDIZATION AND SOLVENT AND
TEMPERATURE EFFECTS

An NMR experiment involves a highly sophisticated instrument

capable of resolving resonances and making measurements to one part
in 1011 in favourable circumstances. (This is equivalent to comparing
the lengths of two steel rods each 1 km long to within 10 A or three
atoms of iron). One must, therefore, accept the responsibility of
preparing a sample that will not degrade the spectrometer perfor-
mance. However, any sample placed in the magnet will distort the
magnetic field. Fortunately, the distortion occurs externally to a cylin-
drical sample and the field remains homogeneous within it except at
the ends, though its magnitude is changed by an amount that depends
both on the shape of the sample and on the bulk magnetic suscepti-
bility of the tube glass and of the sample itself. Imperfections in the
glass, variations in wall thickness, variations in diameter, or curvature
of the cylinder along its length all lead to degradation of the field
homogeneity within the sample, with consequent line broadening. For
this reason high-precision bore sample tubes are always used for NMR.
Since solid particles distort the field around them, suspended solids
must also be filtered from the liquid sample prior to measurement.
Paramagnetic metal ions and complexes also cause signal broadening
and if present must be removed, often by chromatography.
As the NMR tube distorts the magnetic field, adjustment of the
magnetic field using the shim coils can be kept to a minimum by always

www.pdfgrip.com
 Sample preparation, standardization, solvent and temperature effects 37

setting the NMR tube in the spinner at the same depth, using a depth
gauge. Provided enough solvent is used so that the top of the solution
is well above the coil, there will be only minor changes to the field
homogeneity, and this is the normal procedure for samples with strong
signals. Where sensitivity is low and the quantity of compound is
limiting, it is advisable to have as much compound as possible inside
the detector coils. This is done by adjusting the position of the bottom
of the tube and the top of the solution to match the bottom and top
of the detector coil, but considerable time will be required to reshim
the magnet to obtain best homogeneity.

2.5.1 Standardization
We have shown that chemical shifts are invariably measured relative
to a standard of some sort. There are five ways of standardizing a
resonance, which are now given.

2.5.1.1 Internal standardization

The standard substance is dissolved in the sample solution and its
resonance appears in the spectrum. This method has the advantage
that the magnetic field is exactly the same at sample and standard
molecules. The standard must be chosen so as not to obscure sample
resonances and also must be inert to the sample. Internal standard-
ization is the method normally used for !H, 2H, 13C, 14N, 15N, 19F, and
29
Si. The main disadvantage of the method is that weak interactions
with the solvent produce small chemical shifts, which are difficult to
predict and which reduce the accuracy of the measurements by an
unknown amount. These shifts are said to arise from solvent effects.
TMS is reasonably free from solvent effects and is the reference of
choice. For aqueous solutions, TMS is not soluble and either TSP,
Me3SiCD2CD2CO2Na, or DSS, Me3SiCH2CH2CH2SO3Na, are used as
water soluble references. Generally, the former is now preferred, as
on account of the deuteration only the Me3Si group gives a 1H NMR
signal.
For most other nuclei, the normal reference material cannot be
added to the solution without severe risk of reaction or strong inter-
actions.

2.5.1.2 Reference to an absolute frequency

On modern spectrometers, it is easy to reference one nucleus to the
frequency of another. There are a number of reports of the absolute
frequencies of reference substances. For example, the usual reference
for 199Hg is [HgMe2]. This compound is highly toxic and in practice is
rarely used - recently, a fatality in a laboratory has been attributed
to its use as an NMR reference. However it has been reported that
when TMS is exactly at 100 000 000.00 Hz, the signal of 199Hg in

www.pdfgrip.com
38 The magnetic field at the nucleus

[HgMe2] is at 17 910 841 Hz. This absolute frequency referencing

is given the symbol S, and it and others are quoted as precise
numbers, scaling the frequency to TMS being at exactly 100 MHz.
Referencing to a standard compound presents a major problem for
many nuclei with chemical shifts which are very temperature, solvent
and concentration dependent. For instance, a temperature dependence
of the shift of 1 ppm/°C is common. Hence referencing to 1H of
TMS is far more accurate than referencing to some compound of the
heavier nuclei.
The use of frequency referencing is simple. If a sample containing
TMS and raer-[RhQ3(SMe2)3] is examined, giving the frequency of
1
H in TMS as 400.134 394 MHz and 103Rh in mer-[RhC!3(SMe2)3] as
12.693 503 MHz, then the absolute frequency of 103Rh in mer-[RhC\3
(SMe2)3] can then be calculated as S = 3.172 310 MHz as
r , ™ ,o« MN I2-693 503 x 100.000 000
H (m,r-[RhC!3(SMe2)3]) = 400.134394
- 3.172 310 MHz (2.9)
In practice, the use of the large numbers of the E scale is inconve-
nient. Frequently, these values are used to calculate the spectrometer
reference frequency and then the usual 8 scale is used. For example,
the 1H NMR spectrum of a mercury sample containing TMS is
measured. The frequency of the TMS is determined exactly, say for
example 400.134 394 MHz. The frequency of the [199HgMe2] reference
is then calculated as
199 400 134 394
Hg reference frequency = 100000000 X 17-910 841 MHz

- 71.667 435 MHz (2.10)

and this frequency is put into the spectrometer computer memory as
the 199Hg reference frequency. The chemical shifts are quoted in ppm
relative to this absolute frequency.
In a few cases it has been proposed that an arbitrary absolute
frequency is used in preference to the frequency for a reference
compound. This is done for 103Rh at S = 3.16 MHz and 195Pt at S =
21.4 MHz. This avoids the problems associated with E for a reference
compound being solvent, concentration and temperature dependent,
and in many cases the necessity to correct for magnetic susceptibility
differences when the reference is external, see section 2.5.1.4.

2.5.1.3 Standardization using the 2H NMR signal of the lock

Rather than using the 1H NMR signal of TMS as the frequency refer-
ence, the frequency of the 2H lock is often used. In most solutions the
relationship between the lock frequency and the frequency of the *H
NMR signal of TMS is constant to much better than 0.1 ppm. This is
far better than is achieved in referencing many nuclei. Hence this

www.pdfgrip.com
 Sample preparation, standardization, solvent and temperature effects 39

method is commonly used to reference signals. The computer systems

of modern NMR spectrometers are set up to automatically reference
to the 2H lock and it is then up to the operator to select a more appro-
priate method of referencing. At best, this method is only as good as
the frequencies which have been input into the computer and it is
important that the calibration is checked by running some reference
compounds.

2.5.1.4 External standardization

The standard is sealed into a capillary tube that is placed coaxially
within the sample tube. The main disadvantage of the method is that,
since the volume magnetic susceptibilities of the sample and standard
will differ by several tenths of a ppm, the magnetic fields in each will
be different and a correction will have to be made for this. Since the
volume susceptibilities of solutions are often not known, these must
be measured, so that a single shift determination becomes quite diffi-
cult if accurate work is required. The magnitude of the correction
depends upon the sample shape and is zero for spherical samples, so
that this disadvantage can be minimized by constructing special
concentric spherical sample holders. The method is used with very
reactive samples or with samples where lack of contamination is
important.
The change from the old permanent and electromagnets to the
current superconducting solenoids has resulted in a change of field
direction from being at right angles to the NMR tube to being coaxial
with it. This has resulted in a change in sign of the magnetic suscep-
tibility correction. For a perfect pair of coaxial tubes, the susceptibility
correction in a permanent or electromagnet with the magnetic field
direction at right angles to the sample is

8
int = 8obs - -y (Xref ~ Xsample) (211)

while in a superconducting solenoid with the magnetic field parallel

to the sample is

8
int = 8obs + -y (Xref ~ Xsample) (2'12)

where 8int is the corrected chemical shift, S0ts and Sotjsare the measured
chemical shifts, xref and Xsample are tlie magnetic susceptibilities of the
reference capillary and the sample, respectively. Most of the results
quoted in the literature using an external reference have not been cor-
rected for susceptibility effects, e.g. for 31P, an external reference of
85% H3PO4 is normally used, and the result is a discrepancy of up to
1 ppm between the older work recorded using a permanent or elec-
tromagnet and the more recent work using a superconducting solenoid.
Because the capillary holding the standard distorts the magnetic field
around it, the field homogeneity in the annular outer part of the sample

www.pdfgrip.com
40 The magnetic field at the nucleus

is destroyed. This can be restored by spinning, which is essential with

this type of standardization. Distorted capillaries can even then
degrade the resolution, and if there is any asymmetry in the annular
region this will be averaged by spinning to give a field different in
value from the true average, i.e. a small shift error will result. The
method is nevertheless the only one suitable for measuring solvent
shifts.

2.5.2 Solvent effects

The solvent shift effects mentioned under internal standardization
(section 2.5.1.1), while a nuisance to those interested simply in struc-
ture determination, are of interest in their own right, since they tell
us something about the weak interactions that occur between solvent
and solutes. The effect is particularly large for aromatic solvents or
where specific interactions occur. The chemical shifts of a number
of substances relative to TMS in a series of solvents are given in
Table 2.1.
The variation in 8 between solvents, of course, contains contribu-
tions from the solvent effect on both solute and standard. Table 2.1
nevertheless is useful in indicating the existence of certain interactions
involving the solutes. Thus the low frequency shifts obtained in the
aromatic solvent benzene and for all solutes but chloroform in the
aromatic solvent pyridine are obvious. Even the relatively inert cyclo-
hexane is shifted to low frequency by 0.05 ppm. This arises because
solutes tend to spend a larger amount of time face on to the disc-
shaped aromatic molecules and so on average are shifted low
frequency by the ring current anisotropy. The magnitude of the effect
depends upon molecular shape and is also increased if there is any
tendency for polar groups in the molecule to interact with the aromatic
IT electrons.
In the case of complex solutes, each type of proton in the molecule
suffers a solvent shift, but because the proximity of each to solvent
depends upon the shape of the molecule, each suffers a different

Table 2.1 Internal chemical shifts 8 of solutes in different solvents

Solvent
Solute CDC13 (CD3)2SO Pyridine Benzene CF3COOH
Acetone,
(CH3)2CO 2.17 2.12 2.00 1.62 2.41
Chloroform,
CHC13 7.27 8.35 8.41 6.41 7.25
Dimethyl-
sulfoxide,
(CH3)2SO 2.62 2.52 2.49 1.91 2.98
Cyclohexane,
QH17 1.43 1.42 1.38 1.40 1.47

www.pdfgrip.com
 Sample preparation, standardization, solvent and temperature effects 41

solvent shift. For this reason, a complex solute may have quite different
spectra in, for instance, chloroform and benzene, and a change from
one solvent to the other may remove some degeneracy or avoid the
overlapping of signals that would otherwise be difficult to disentangle.
Figure 2.16 shows the quite large changes that can occur. This tech-
nique is often known by the acronym ASIS, or Assisted Solvent-
Induced Shifts.
Chloroform as a solute suffers considerable solvent shifts. The pure
liquid is self-associated by hydrogen bonding, but upon progressive
dilution in an inert solvent the proportion of hydrogen-bonded mol-
ecules is reduced and its resonance is shifted 0.29 ppm to low
frequency. The shifts noted in Table 2.1 are in excess of this and we
must consider the existence of several other types of interaction. Thus
specific interactions with the Lewis bases, dimethylsulfoxide and
pyridine, result in high-frequency shifts. In the case of pyridine, this
implies a preference for an edge-on approach to the aromatic ring and

(d)-
(c)-

(b)-
(a) =
2.4 2.2 2.0 1.8 1.6 1.4 1.0 0.8 0.6 0.4
1
H/ppm
Figure 2.16 400 MHz 1H NMR spectra of camphor, (a) Using CDC13 as solvent, (b) As (a), but with
increased gain, (c) Using C6D6 as solvent, (d) As (c), but with increased gain. Note that in CDC13, the
signals due to H5a and H6a overlap, while they are separated in C6D6. However, in C6D6, the signals due
to H3a, H4, and H5b now overlap. The positions of the signals are markedly solvent dependent. The fine
structure is due to spin-spin coupling, to be discussed in the next chapter.

www.pdfgrip.com
42 The magnetic field at the nucleus

therefore some ring current deshielding. In benzene, on the other hand,

hydrogen bonding is reduced, and there is probably face-wise inter-
action of the chloroform with the benzene TT electrons. Both processes
tend to increase the screening, so that the chloroform is shifted strongly
to low frequency. In addition, the chloroform molecule is polar and
its dipole electric field will polarize the surrounding solvent by an
amount related to the solvent dielectric constant e. This induced charge
gives rise to an electric field, which is called the reaction field and
which will also produce chemical shifts of the chloroform solute. Thus
some of the variation observed in Table 2.1 will originate from differ-
ences in solvent dielectric constant.
Two further contributions to solvent shifts are also usually consid-
ered. One arises from the van der Waals interactions and is respon-
sible for vapour to liquid shifts of 0.1 to 0.5 ppm. The other arises in
the case of external standardization and is due to bulk diamagnetic
susceptibility differences between solvents. These susceptibility shifts
can be comparable in magnitude with those due to the other effects
and so must be considered in any interpretation of solvent shifts, but
they do not, of course, arise from any chemical interaction.
The various contributions to the solvent shift 8S can be summarized
by a five-term equation
8
S = 8B + 8A + 8E + 8H + 8W (2-13)

where 8B is the bulk susceptibility contribution, 8A is the anisotropy

contribution, 8E is the reaction field contribution, 8H is the contribu-
tion of hydrogen bonding and specific interactions, and 8W is the van
der Waals contribution.
Although for 1H chemical shifts, the solvent shift is usually less than
I ppm, the effect for other nuclei can be very substantial. For example,
the 195Pt reference compound, [PtCl6]2~, changes its chemical shift by
II ppm simply by changing the solvent from H2O to D2O, while a wide
range of solvents produces a variation in chemical shifts over a range
of 400 ppm.

2.5.3 Temperature effects

Chemical shifts are temperature dependent. As noted above, this
makes referencing of nuclei difficult and 1H in TMS is often preferred
as the nucleus to reference all nuclei. Many heavy nuclei change
their chemical shift by more than 1 ppm per degree C. For example,
the temperature dependence of the 195Pt chemical shift of aqueous
[PtCl6]2~ is 1.1 ppm/°C. At 9.4 T, this corresponds to 94 Hz/°C! Where
this effect is most detrimental is in the way it affects the resolution of
the NMR signal. In an NMR experiment, a number of NMR spectra
are added together to produce the final result. If during the experi-
ment the temperature changes or if there is a temperature gradient in
the NMR tube, the signal is broadened. This is a major factor in
producing line broadening for nuclei other than 1H. For example, if a

www.pdfgrip.com
 Questions 43

sample gives a resolution of 0.2 Hz for !H, it should give a resolution

of better than 0.1 Hz for 13C and 31P. However, a resolution of better
than 1 Hz is difficult to achieve due in part to temperature changes
during the course of the experiment. The problem can be alleviated
by using the spectrometer's variable temperature accessory to ther-
mostat the sample and by incorporating a substantial number of
dummy scans where the spectrometer goes through the process of spec-
trum acquisition without actually storing the FID. This enables the
sample to come to thermal equilibrium. In addition to the probe often
being at a different temperature to the room, *H decoupling, see
Chapter 5, is frequently used which also warms the sample.
Normally temperature effects do not have a major effect on the
linewidths of !H spectra. There is one major exception. When D2O is
the solvent and lock, temperature variations can have a disastrous
effect on resolution. Any 1H or 2H NMR signal from hydrogen or
deuterium atoms involved in hydrogen bonding shows a substantial
temperature dependence. Consequently if D2O is used as the lock
substance, any temperature changes cause the lock signal to move and
with it all the other signals in the spectrum due to the consequential
magnetic field changes. This problem is particularly severe away from
room temperature. It is often then necessary to wait a long time for
temperature stability to be achieved, monitoring the stability by
measuring the position and lineshape of a reference compound, e.g.
TSP or DSS.

2.6 QUESTIONS

2.1. Which of the two chemically different types of protons in

CH2C1CHC12 resonate at higher frequency?
2.2. A proton spectrometer operating at 100 MHz was used to
measure the frequency separation of the resonances of chloro-
form, CHC13, and TMS, which was found to be 730 Hz, the CHC13
being to high frequency. What is the chemical shift of chloroform
on the 8 scale? What would the frequency separation and chem-
ical shift be if the sample were measured in a spectrometer oper-
ating at 300 MHz?
2.3. The 100 MHz spectrometer above is used to produce the proton
spectrum of a complex molecule. Would the resolution be better
or worse if the 2H NMR spectrum of the fully deuterated form
of the same molecule were obtained? Assume that the linewidths
of 1H and 2H resonances are the same. The frequency used for
2
H spectra is 15.35 MHz. Assume also that there is no primary
isotope effect and ignore spin-spin coupling.
2.4. The value of A£ for the separation of the highest occupied and
the lowest unoccupied orbitals of the cobalt in [Co(OH2)6]3+,
[Co(NH3)6]3+ and [Co(CN)6]3- are 16 600, 21 000 and 32 400 cnr1
respectively. Show that this is consistent with the chemical shifts

www.pdfgrip.com
 44 The magnetic field at the nucleus

given in Fig. 2.15. Use the data to estimate <rd for 59Co. Assume
that AE1'1 is the only significant term contributing to the ap
variations.
2.5. In annulene,

the inner protons have a chemical shift of 8 -3.0 while the outer
ones come at 8 9.3. Explain why this happens when the XH chem-
ical shift of ethene is 8 5.5.
2.6. Figure. 2.17 shows the 195Pt NMR spectrum of [PtCl6]2-. Explain
why the signal is a multiplet and account for the number of signals
and their intensities.
2.7. Many old reports involving the use of NMR spectrometers with
electromagnets report the 31P chemical shift of PPh3 as being
between 8 -6 and -7. More recent reports using NMR spec-
trometers with superconducting magnets report the 31P chemical
shift of PPh3 as being between 8 -5 and -6. Explain why the
apparent chemical shift has changed.
2.8. E for 125Te in the reference compound Me2Te has been reported
as being 31.549 802 MHz. If TMS in a sample in CDC13 has been
measured as being at 400.134 394 MHz, what frequency should
be used to reference a 125Te signal from this solution with Me2Te
I at zero? The 125Te NMR signal of a sample of Ph2Te was deter-
100 OHz mined as being at 126.365 452 MHz. What is the chemical shift
in ppm relative to Me2Te?
Figure
195
2.17 The 85.6 MHz 2.9. There are two common references for 195Pt, namely E = 21.4 MHz
Pt NMR spectrum of and [PtCl6]2~ which has E = 21.496 700 MHz. The compound
2M Na2[PtCl6] in D2O. [Pt(CN)4]2- has E = 21.395 634 MHz. Calculate the chemical shifts
(Reproduced by permis-
sion of The Royal Society of [PtCl6]2- and [Pt(CN)4]2- using first E = 21.4 MHz and then
of Chemistry from Sadler [PtCl6]2~ as the reference. Explain why when using equation (2.6),
et al (1980) /. Chem. Soc., the chemical shift separation of [PtCl6]2~ and [Pt(CN)4]2~ in ppm
Chem. Commun., 1175.) depends on the chosen reference.
2.10. Using equations (2.11) and (2.12), work out an expression which
allows the magnetic susceptibility differences between two liquids
to be obtained from measurements of the chemical shifts of reso-
nances in the two liquids in both a superconducting magnet and
an electromagnet.

www.pdfgrip.com
 Intei nuclear spin-spin
coupling s
3.1 THE MUTUAL EFFECTS OF NUCLEAR MAGNETS ON
RESONANCE POSITIONS

The Brownian motion in liquid samples averages the through-space

effect of nuclear magnets to zero. However, in solutions of POC12F,
for example, the phosphorus nucleus gives two resonances whose sepa-
ration does not depend upon the magnetic field strength. (The chlo-
rine nuclei (/ = 3/2) have no effect. This is explained in Chapter 4.)
The two resonances correspond to the two spin orientations of the
fluorine nucleus so that the nuclei are nevertheless able to sense one
another's magnetic fields. Theoretical considerations show that the
interaction occurs via the bonding electrons. The contact between one
nucleus and its s electrons perturbs the electronic orbitals around
the atom and so carries information about the nuclear energy to other
nearby nuclei in the molecule and perturbs their nuclear frequency.
The effect is mutual and in POC12F both the fluorine (7 = 1/2) and
the phosphorus (/ = 1/2) resonances are split into doublets of equal
hertz separation. The magnitude of the effect for a particular pair of
nuclei depends on the following factors:
1. The nature of the bonding system, i.e. upon the number and bond
order of the bonds intervening between the nuclei and upon the
angles between the bonds. The interaction is not usually observed
over more than five or six bonds and tends to be attenuated as the
number of bonds increases, though many cases are known where
coupling over two bonds is less than coupling over three bonds.
2. The magnetic moments of the two nuclei. The interaction is directly
proportional to the product yAyE where ^A and yB are the magne-
togyric ratios of the interacting nuclei.
3. The valence s electron density at the nucleus and therefore upon
the s character of the bonding orbitals. This factor also means that
the interaction increases periodically as the atomic number of either
or both nuclei is increased in the same way as does the chemical
shift range.
The magnitude of the coupling interaction is measured in hertz (Hz)
since it is independent of the magnetic field strength. It is called
the coupling constant and is given the symbol /; its magnitude is very

www.pdfgrip.com
46 Internuclear spin-spin coupling

variable and values have been reported from 0.05 Hz up to thousands

of hertz. The value of / gives information about the bonding system
but this is obscured by the contribution of yA and ^B to /. For this
reason correlations between the bonding system and spin-spin
coupling often use the reduced coupling constant, K, which is equal
to 4TT2J/hyAyB.
It is important to understand that coupling constants can be either
positive or negative and that the frequency of one nucleus may be
either increased or decreased by a particular orientation of a coupled
nucleus, the sign depending upon the bonding system and upon the
sign of the product 7A^B.
Considerable data are available upon the magnitudes of interproton
spin coupling constants from the mass of data accumulated for organic
compounds. Interproton coupling is usually (though not always) largest
between geminal protons (H-C-H), and depends upon the angle
between the two carbon-hydrogen bonds. / is typically -12 Hz in
saturated systems. / falls rapidly as the number of intervening bonds
is increased, being 7 to 8 Hz for vicinal protons (H-C-C-H) and near
zero across four or more single bonds. The same rules apply if oxygen
or nitrogen forms part of the coupling path, and methoxy protons
(H3COCHR2) do not usually show resolvable coupling to the rest of
the molecule, though alcoholic or amino protons may do so to vicinal
protons in, for example, HOCH3. On the other hand, coupling may
be enhanced if there is an unsaturated bond in the coupling path, due
to a CT-TT configuration interaction, and may be resolved over up to as
many as nine bonds, e.g.9/(H-H) = 0.4 Hz between the hydrocarbon
protons in H3C(C=C)3CH2OH. In saturated molecules, a planar
zig-zag configuration of the bonds may also lead to resolvable
coupling over four or five single bonds. Note the use of the super-
script 9 in the acetylene example to indicate the number of bonds over
which the interaction occurs.
Karplus has calculated the values of the vicinal interproton coupling
constants and shown that these depend upon the dihedral angle (/>
between the carbon-hydrogen bonds (Fig. 3.1). Unfortunately, the
magnitude of coupling is also influenced by such factors as the nature
of the other substituents on the two carbon atoms, their electronega-
tivity, their orientation, the hybridization at the carbon, the bond
angles other than the dihedral angle and the bond lengths. For this
reason, two curves are shown, which give an idea of the range of
coupling constants that may be encountered. If the geometry of the
molecule is fixed, then the value of the coupling constant enables an
estimate to be made of the value of 4>. If the geometry is not fixed,
then the coupling constant is an average over the possible values of
c)>. If free rotation is possible, then the average is over the whole
of the curves illustrated. Thus in fluoroethane (ethyl fluoride) and
ethyllithium the vicinal coupling is 6.9 and 8.4 Hz respectively, the
value increasing with decreasing substituent electronegativity. If
the geometry is static, then the values of the coupling constants may

www.pdfgrip.com
 The appearance of multiplets arising from spin-spin coupling 47

15-

N
I 10-
if
X

CO 5-

0-
0 45 90 135 180
0
Torsion Angle/

Figure 3.1 Karplus curves, using the equation modified by Altona, relating the dihedral angle 4> in a
HC-CH fragment and the vicinal proton-proton coupling constant. The inset shows a view along the
carbon-carbon bond. Two curves are shown relating to differently substituted molecules, CH3CH3, dashed
line, (a) and CH3CH2F, solid line, (b). The horizontal lines, (c) and (d), show the average value obtained
when a group can rotate freely, giving rise to an averaged /vic.

show quite clearly what the conformation of the molecule studied has
to be. Thus, in Fig. 3.2, part of the proton spectrum of menthol is
given in which various doublet splittings due to spin-spin coupling are
evident. The resonance of H1 is split into a doublet of triplets with a
small / of 4.2 Hz and two large ones of 10.5 Hz. H1 is easily identifi-
able by its chemical shift, being next to the electronegative OH group,
and it is the large couplings J(Hl-H2ax) and /(H^H6) that are due to
these axial-axial pairs with a dihedral angle of near 180° and large
predicted /. The other lesser interaction is /(ff-H2^), H1 and H2 being
an axial-equatorial pair with dihedral angle of near 60° and so small
predicted /.
This vicinal 3J dependence on dihedral or torsion angle seems to be
quite general and Karplus-type curves have been established for the
coupling paths ^C-C-C^H, 31P-C-C-31P and 13OC-C-31P and are
also likely for coupling between 77Se or 125Te and 1H or 13C or for the
more exotic systems such as 3/(199Hg-13C) in alkylmercury compounds.

3.2 THE APPEARANCE OF MULTIPLETS ARISING FROM

SPIN-SPIN COUPLING

The appearance of these multiplets is very characteristic and contains

much information additional to that gained from chemical shift data.

www.pdfgrip.com
 48 Internuclear spin-spin coupling

3.5 3.0 2.5 2.0 1.5 1.0

1
H/ppm

Figure 3.2 The 400 MHz !H NMR spectrum of menthol whose structure is shown in the inset. The numbers
represent hydrogen atoms attached to the ring. The resonances are split by spin coupling with adjacent
hydrogen atoms. In particular, H1, at 8 3.4, is split into a doublet, 4.2 Hz, of triplets, 10.5 Hz, by coupling
with the equatorial H6 and the two axial H2 and H6 nuclei.

The coupling patterns and constants enable connectivity to be estab-

lished between atoms in the molecules and hence enable the structure
of the molecule to be deduced. We have already seen one such appli-
cation in Fig. 3.2 and can see the patterns produced by coupling can
be very complex. The simplest case to consider is the effect that a
single chemically unique nucleus of / - 1/2 has on other nuclei in the
molecule that are sufficiently closely bonded. In half the molecules in
the sample, the spin of our nucleus A will be oriented in the same
direction as the field and all the other nuclei in these molecules
will have corresponding resonance positions. In the remainder of
the molecules, the spin of A will be opposed to the field and all the
other nuclei in this half of the sample will resonate at slightly different
frequencies to their fellows in the first half. Thus when observing
the sample as a whole, each of the nuclei coupled to A gives rise to
two lines. The line intensities appear equal since the populations
of A in its two states only differ by < 0.01%, which is not detectable.
We say that A splits the other resonances into 1:1 doublets. Because
the z component of magnetization of A has the same magnitude in
both spin states, the lines are equally displaced from the chemical shift

www.pdfgrip.com
 The appearance of multiplets arising from spin-spin coupling 49

positions of each nucleus, which are therefore at the centres of the

doublets.
Let us in illustration consider the molecule CHC12CH2C1 and its
proton resonance. This contains two sorts of hydrogen, with the CHC12
proton resonating to high frequency of the CH2C1 protons due to the
greater electric field effect of two geminal chlorine-carbon bonds. The
two CH2C1 protons have the same frequency since rotation around
the carbon-carbon bond averages their environments and makes them
chemically equivalent. It also averages the vicinal coupling to the single
proton to be the average of the Karplus curve. These two protons thus
have the same chemical shift and the same coupling constant to all
other magnetically active nuclei in the sample; a trivial condition in
this case since there is only one other proton and the spectra are indif-
ferent to the chlorine nuclei for reasons that we shall see later. We
say that the CH2 nuclei are isochronous and magnetically equivalent.
They are, of course, coupled quite strongly, but because they are
isochronous they resonate as if they were a unit and give a singlet
resonance unless coupled to other nuclei. This is a consequence of the
second-order effects to be considered later.
In the present example, however, the CH2C1 resonance is split
into a 1 :1 doublet because of coupling to the non-isochronous
CHC12 proton. Equally, since the coupling interaction is mutual,
the CHC12 proton is split by the two CH2C1 protons, though the
splitting pattern is more complex. We can discover the shape of the
CHC12 multiplet in several ways.
1. By an arrow diagram (Fig. 3.3). In some molecules, both the CH2C1
spins will oppose the field, in others both spins will lie with the
field, while in the remainder they will be oriented in opposite direc-
tions. The CHC12 protons in the sample can each experience one
of three different perturbations and their resonance will be split
into a triplet. Since the CH2C1 spins can be paired in opposition in
two different ways, there will be twice as many molecules with them
in this state as there are with them in each of the other two. The
CHC12 resonance will therefore appear as a 1 : 2 :1 triplet with the
spacing between the lines the same as that of the CH2C1 1 :1
doublet. An actual spectrum is shown in Fig. 3.4, where it should
be noted that the intensities are not exactly as predicted by the
simple first-order theory but are perturbed by second-order effects.
Note also that the total multiplet intensity is proportional to the
number of protons giving rise to each multiplet. The coupling
pattern thus counts precisely the number of protons in interaction
and the intensity enables us to relate different, non-coupled groups
of lines in the spectrum.
2. We can also work out the multiplicity by considering the possible
values of the total magnetic quantum number Sm of the two CH2C1
protons. We have / = 1/2, and m can be ±1/2.
Therefore for two protons (Fig. 3.3)

www.pdfgrip.com
50 Internuclear spin-spin coupling

Figure 3.3 A simulated signal demonstrating the splitting due to two / = 1/2
nuclei. When Sra = 0 there is no perturbation of the coupled resonance so
that the centre line corresponds to the chemical shift position. This holds for
all multiplets with an odd number of lines. The spacing / between the middle
of the lines corresponds to a change in Sra of unity.

2m = + -i- + -L = +1 (3.1)

or

f +2^ r - 42 - = o
Sm = J
(3.2)
1-1 + 1 = °

^-i-i.-i
or

(3.3)

Therefore we have three lines.

Methods 1 and 2 are equivalent, but method 2 is particularly useful
when considering multiplets due to nuclei with / > 1/2 where arrow
diagrams become rather difficult to write down clearly. Note that when
2m = 0 there is no perturbation of the chemical shift of the coupled
group so that the centre of the spin multiplet corresponds to the chem-
ical shift of the group.
Next let us consider the very commonly encountered pattern given
by the ethyl group CH3CH2-. The isochronous pair of CH2 protons
are usually found to high frequency of the CH3 protons and are spin-
coupled to them. The CH3 protons therefore resonate as a 1 : 2 : 1

www.pdfgrip.com
 The appearance of multiplets arising from spin-spin coupling 51

6.0 5.5 5.0 4.5 4.0

1
H/ppm
Figure 3.4 The 400 MHz 1H NMR spectrum of CH2C1CHC12 in CDC13.

triplet. The splitting of the CH2X resonance caused by the CH3 group
can be found from Fig. 3.5, and is a 1 : 3 : 3 :1 quartet. A typical ethyl
group spectrum is shown in Fig. 3.6.
We have done enough now to formulate a simple rule for splitting
due to groups of / = 1/2 nuclei. Thus the number of lines due to
coupling to n equivalent / = 1/2 nuclei is n + 1. The intensities of the
lines are given by the binomial coefficients of (a + 1)" or by Pascal's
triangle, which can be built up as required. This is shown in Fig. 3.7.
A new line of the triangle is started by writing a 1 under and to the
left of the 1 in the previous line and then continued by adding adja-
cent figures from the old line in pairs and writing down the sum as
shown. The multiplicity enables us to count the number of I = 1/2
nuclei in a group and the intensity rule enables us to check our assign-
ment in complex cases where doubt may exist, since the outer compo-
nents of resonances coupled to large groups of nuclei, e.g. the CH of
(CH3)2CH may be too weak to observe in a given spectrum.
Coupling to nuclei with / > 1/2 leads to different relative intensities
and multiplicities. In the case of a single nucleus the total number of
spin states is equal to 21 + 1 and this equals the multiplicity. If / = 1/2

www.pdfgrip.com
 52 Internuclear spin-spin coupling

we get two lines, 1=1 gives three lines, / = 3/2 gives four lines, and
so on. The spin populations of each state are virtually equal and so
the lines are all of equal intensity and of equal spacing (Fig. 3.8).

Figure 3.5 The splitting due to three / = 1/2 nuclei. There is no line that
corresponds to Sra = 0. The multiple! is, however, arranged symmetrically
about the Sra = 0 position, so that the centre of the multiplet corresponds to
the chemical shift. This rule holds for all multiplets with an even number of
lines.

JL J_
—r — IMS
I 3.5 3A 1.7 I

4.0 3.0 2.0 1.0 0.0

8/PPM
Figure 3.6 The 400 MHz !H NMR spectrum of ethyl bromide CH3CH2Br in CDC13. The CH2 and CH3
signals are expanded.

www.pdfgrip.com
 The appearance of multiplets arising from spin-spin coupling 53

AT
0 1 Singlet
1 1 1 Doublet
+

2 1 2 1 Triplet
+ +
II II
3 1 3 3 1 Quartet
+ + +

4 1 4 6 4 1 Quintet
Figure 3.7 Pascal's triangle can be used to estimate the intensities of the lines
resulting from coupling to different numbers, N, of equivalent / = 1/2 nuclei.
The numbers in each line are obtained by adding adjacent pairs of numbers
in the line above.

/e.g.
\ 3lPMe3
2
J(31P1H) = 30Hz

1 [14NH4]+

1
J(14N1H) = 52.5Hz

f [UBH4]-
1
J(11BH) = 80.5Hz

3 [10BH4]-
1
J(10BH) = 27.2Hz

Figure 3.8 Multiplets observed in the proton spectra of a variety of species

due to coupling to single nuclei with / = 1/2 (31PMe3), 1=1 ([14NH4]+),
/ = 3/2([nBH4]-) and I = 3 ([10BH4]-). In the latter case, the two spectra are
obtained simultaneously in the same spectrum since natural boron contains
both isotopes, in the ratio 80.22 :19.78. Note the smaller coupling constant to
the isotope with the smaller magnetogyric ratio.

Splitting due to multiple combinations of / > 1/2 nuclei is much less

common, but a few examples have been recorded. Figure 3.9 illus-
trates the pattern commonly encountered when obtaining !H spectra
in the solvent (2H6) acetone {deuterioacetone, (CD3)2CO}. In practice,

www.pdfgrip.com
54 Internuclear spin-spin coupling

2.05 2.04 2.03 2.02 2.01

1
8( H)
1
Figure 3.9 The 400 MHz H NMR spectrum of (CD3)C(O)(CHD2). The multi-
plet pattern obtained for a proton coupled to two deuterium nuclei with
7 = 1 . The value of 2J(2H1H) is 2.2 Hz. The value of 2J(1H1H) can be calcu-
lated from this spectrum. This coupling exists but cannot be measured in the
all-hydrogen form.

there will always be a small amount of hydrogen present in these mol-

ecules and some of the methyl groups will be CD2H-groups. The
proton sees two equivalent deuterons with 7 = 1 . The maximum total
Sra is 1 + 1 = 2. Thus Am = ±1 and there are therefore five spin states
and the proton resonance will be a quintet. In order to determine the
line intensities, we have to determine the number of ways each value
of Sra can be obtained. This is shown in Table 3.1 and indicates rela-
tive intensities of 1 : 2 : 3 : 2 : 1 , a distribution that differs from the
simple binomial.

Table 3.1 Line intensities for coupling to two nuclei with 7 =

Possible spin combinations Number of spin combinations
2 (+1, +1) 1
1 (+1, 0), (0, +1) 2
0 (+1, -1), (0, 0), (-1, +1) 3
-1 (-1, 0), (0, -1) 2
-2 (-1, -1) 1

www.pdfgrip.com
 The appearance of multiplets arising from spin-spin coupling 55

Thus the rule given for 7 = 1/2 nuclei can be generalized to include
groups of nuclei of any given /, the number of lines observed for
coupling to n equivalent nuclei of spin / being 2nl + 1. Obtaining the
relative intensities is, however, tedious and it is better to use a 'Pascal's'
triangle type of construction as shown in Fig. 3.10 for when 1=1 and
Fig. 3.11 for when 7 - 3/2.
In the case of / = 1, the numbers in the triangle are placed immedi-
ately under the previous ones, since there is always a line in the centre
of the multiplet. For / = 3/2 (and all half-integral spins), the numbers
are staggered, since only if n is even is there a line at the centre of the
multiplet. The triangles are constructed by moving a box, which can
enclose 27+1 numbers of a line, along the line, enclosing first the 1,
then two numbers, then three, then four, and so on. The numbers
enclosed by the box are added to give a number for the next line. Thus
for 7 = 3/2 the box can enclose four numbers, and if we sweep through
the line for n = 1, we enclose first a 1, then 1 + 1 = 2, then 1 + 1 + 1
= 3, then 1 + 1 + 1 + 1 = 4 and reducing again to 3, 2, 1. The construc-
tion given above for 7 = 1/2 nuclei is the first example of this series.

N 7=1
0 1
1 1 1 1
2 1 2 3 2 1
3 1 3 6 7 6 3 1
1
4 1 4 10 16 19 16 10 4 1
Figure 3.10 A version of 'Pascal's triangle' to determine the intensities of
lines resulting from coupling to different numbers, N, of equivalent 7 = 1
nuclei.

N 7=3/2
0 1
1 1 1 1 1
2 1 2 3 4 3 2 1
3 1 3 6 10 12 12 10 6 3 1
i
4 1 4 10 20 31 40 44 40 31 20 10 4 1
Figure 3.11 A version of 'Pascal's triangle' to determine the intensities of
lines resulting from coupling to different numbers, N, of equivalent / = 3/2
nuclei.

www.pdfgrip.com
 56 Internuclear spin-spin coupling

More complex coupling situations also arise where a nucleus may

be coupled simultaneously to chemically different groups of nuclei of
the same or of different isotopes or species. The patterns are found
by building up spectra, introducing the interactions with each group
of nuclei one at a time. Thus Fig. 3.12 shows how a group M coupled
to two chemically different / = 1/2 nuclei A and X is first split by
/(A-M) into a doublet, and shows that each doublet line is further
split by /(M-X). If /(A-M) - /(M-X), a 1 : 2 : 1 triplet is obtained;
but if /(A-M) */(M-X), then a doublet of doublets with all lines of
equal amplitude arises. This can be distinguished from a n B coupling
because the line separations are irregular, and of course the prepara-
tive chemist is usually aware whether or not there should be boron in
the compound.
The same technique can be used to predict the shapes of multiplets
due to several equivalent nuclei with / > 1/2, i.e. introducing the split-
ting due to one nucleus several times as shown in Fig. 3.13 for two
/ = 3/2 nuclei. Comparison with the 'Pascal's triangle' diagrams will
show that the two methods are exactly equivalent.
When analysing such multiplets, it always has to be borne in mind
that overlap of lines may occur so that fewer than the theoretical
number of lines are observed and the intensities are unusual. Such a
case is illustrated in Fig. 3.14, which shows the 9Be spectrum of the
complex [(Tj5-C5H5)BeBH4]. The 9Be resonance is split by coupling to

(a) Coupling to HA gives a

doublet

Coupling to HX then
gives a doublet of
doublets

Figure 3.12 Splitting of the H¥ protons of Z2CHA-C(HM)2-CHXY2 due to coupling to HA and Hx. (a)
/(A-M) = /(X-M), the centre lines overlap and the multiplet is a 1: 2 :1 triplet just as if HM were coupled
to a CH2 group, (b) /(A-M) * /(M-X) and we get a doublet of doublets from which both /(A-M) and
/(M-X) can be measured.

www.pdfgrip.com
 The appearance of multiplets arising from spin-spin coupling 57

Figure 3.13 An illustration showing how by successively introducing

the 1:1:1:1 quartet splitting due to coupling to two equivalent nuclei with
/ = 3/2 the correct 1 : 2 : 3 : 4 : 3 : 2 : 1 pattern is obtained.

the nuclei of the BH4 ligand only, with /(9Be-H) - 10.2 Hz and /(9Be-
n
B) = 3.6 Hz. The first coupling causes splitting to a 1 : 4 : 6 : 4 : 1
quintet, each line of which is further split into a 1:1 :1 :1 quartet by
the nB. So 20 lines are expected, but because the ratio of the coupling
constants is almost integral (it is 2.8) some lines overlap and only 16
are observed. The overlap is not exact but the close pairs of lines are
not resolvable because of the appreciable natural linewidth of the 9Be
resonance.

3.2.1 Rules for the analysis of any first-order multiplet arising from
7 = 1 / 2 coupling
The analysis of any well resolved first-order multiplet due to spin 1/2
nuclei is actually a trivial exercise. There are some very simple rules
which make the analysis straightforward.
1. Check that the multiplet is centro-symmetric. If it is not, then it
consists of either overlapping multiplets or is second order, see
section 3.7.
2. Assign an intensity to each line in the multiplet, starting with 1 for
the outermost lines and using only integers. The distribution of the
intensities must be centro-symmetric.
3. Add up the assigned intensities. The sum must be 2", where n is
the number of / = 1/2 nuclei coupling. If the sum does not come

www.pdfgrip.com
58 Internuclear spin-spin coupling

Figure 3.14 The 9Be spectrum at 14.06 MHz of [(t]5-C5H5)BeBH4] in C6F6

solution. Coupling is observed to both n B (/ = 3/2) (/ = 3.6 Hz) and *H (/ =
1/2) (/ = 10.2 Hz) of the BH4 group. Some lines overlap and the multiplicity
is less than the maximum given by the basic rules. (From Gaines et al. (1981)
/. Magn. Reson., 44, 84, with permission.)

to 2", then go through rule 2 again paying particular attention to

lines where there could be doubt about the intensity.
4. The separation of the outermost pair of lines to one side of the
multiplet must be a coupling constant. The intensities of the outer-
most pair of lines give the pattern. Thus 1:1 shows that this first
coupling constant arises from coupling to one nucleus. 1:2 shows
that this first coupling constant arises from coupling to two nuclei
with the same coupling constant, and is part of a 1:2:1 pattern.
1:3 shows that this first coupling constant arises from coupling to
three nuclei with the same coupling constant, and is part of a
1 : 3 : 3 : 1 pattern. This argument continues through the 'Pascal
triangle'.
5. Draw the first coupling pattern on the paper as an inverted stick
diagram. Write the intensity of the first line of the multiplet on the
central line of the stick pattern. In the case of coupling to two or
more nuclei, remember that the separation of each adjacent pair
of lines is equal, so that once the first pair of lines and their intensi-
ties have been identified, the rest of that pattern has been identified.

www.pdfgrip.com
 The appearance of multiplets arising from spin-spin coupling 59

6. Subtract the intensities of the first coupling pattern from those

generated in 3.
7. Go to the next line in the original coupling pattern which still has
intensity after instruction 6. Treat this as the outermost line of the
coupling pattern generated in 5. The outermost line may now have
an intensity greater than 1. If so all the lines of the pattern must
be multiplied by this intensity, which will be an integer. Draw the
first coupling pattern on the paper as an inverted stick diagram.
Write the intensity of the first line of the multiplet on the central
line of the stick pattern.
8. Subtract the intensities of this coupling pattern from those gener-
ated in 6.
9. Keep repeating 7 and 8 until all the lines in the original multiplet
have been accounted for. You will have now removed the smallest
coupling constant, and generated a simpler multiplet. Go back to
3 and keep repeating the procedure until a singlet is generated.
The stick diagram will have been completed.
(These rules to analyse multiplets have been reproduced from Mann
(1995) /. Chem. Ed., 72, 614, with permission copyright © 1995,
Division of Chemical Education, Inc.)
The application of these rules is illustrated in Fig. 3.15 for their
application to the =CH- proton of allyl bromide, CH2=CHCH2Br.

\ ' i ' i ' i ' i ' i • r

6.10 6.08 6.06 6.04 6.02 6.00 5.98
1
H/ppm
!
Figure 3.15 The 400 MHz H NMR spectrum of the signal at 5 6.04 of allyl
bromide, CH2=CHCH2Br.

www.pdfgrip.com
60 Internuclear spin-spin coupling

The intensities are put on each line and they sum to 16 = 24. Hence
there are four / = 1/2 nuclei coupling. The separation of the first pair
of lines is a coupling constant, and their intensities are 1 :2. Hence,
from 'Pascal's triangle', this is part of a 1:2:1 triplet. The separation
between this pair of lines must equal that between the central one and
the third line of intensity one, so the first, second, and fourth lines are
selected as making up the triplet. The triplet is drawn above the multi-
plet. We now move to the next unassigned line, the third one and
repeat the same triplet, using the same coupling constant. We keep
repeating this until all the lines are accounted for. A 1:1 :1 :1 quartet
has been generated. This rapidly reduces to a doublet, and then a
singlet and the pattern is analysed.

3.2.2 Rules for the analysis of any first order multiple! arising from
both 7 = 1 / 2 and / > 1/2 coupling
The rules for determining the sum of intensities of the lines need modi-
fying for nuclei with 7 > 1/2.
i
Sum of intensities = JJ (21 j + 1)"'
;=o
where there are n^ nuclei with spin 7;. The appropriate intensity
triangle for the nuclear spin has to be used. We return to the case of
[(^5-C5H5)BeBH4] to illustrate this (Fig. 3.14). The intensity ratio
of the lines is 1:1 :1 : 5 : 4 : 4 :10 : 6 : 6 :10 : 4 : 4 : 5 :1 :1:1. (Note
that the intensities are not obvious from the spectrum in this case
due to the overlap of the lines.) The sum of the intensities is 64. The
coupling is to four protons and one nB. The sum of intensities should
therefore be (2 x 3/2 + 1)(2 x M + I)4 = 4 x 24 = 64.
The separation of the outer pair of lines is a coupling constant. This
could be to either n B or to 1H, but coupling to the four protons gives
a 1 : 4 : 6 : 4 : 1 quintet, and the multiplet starts 1:1:1. Hence the
smallest coupling constant is to nB, giving a 1:1 :1:1 quartet. This
1:1:1:1 pattern is drawn together by drawing a vertical line above
the middle of it and the intensity of the outermost line, 1, is attached.
The 1:1:1:1 intensity is taken away from the first four lines to leave
residual intensities 0 : 0 : 0 : 4 : 4 : 4 :10 : 6 : 6 :10 : 4 : 4 : 5 :1 :1:1. The
same pattern is subtracted, but the intensity starts 4, so the intensities
to be subtracted are 4 : 4 : 4 : 4 , and 4 is written beside the central
vertical line. These intensities are then subtracted to leave residual
intensities 0 : 0 : 0 : 0 : 0 : 0 : 6 : 6 : 6 :10 : 4 : 4 : 5 :1:1:1. The same
pattern is subtracted again, but the intensity starts 6, so the intensi-
ties to be subtracted are 6 : 6 : 6 : 6 , and 6 is written beside the central
vertical line. These intensities are then subtracted to leave residual
intensities 0 : 0 : 0 : 0 : 0 : 0 : 0 : 0 : 0 : 4 : 4 : 4 : 5 : 1 : 1 : 1 . The procedure
is continued reducing the multiplet to a 1:4:6:4:1 quintet, consistent
with four protons coupling.

www.pdfgrip.com
 Spin-spin coupling satellites 61

3.3 SPIN-SPIN COUPLING SATELLITES

A glance at the table of nuclear properties (Table 1.1) will show that
certain elements have as principal isotope a magnetically non-active
species, but that they have also a more or less small proportion of
magnetically active species, some with 7 = 1/2. Examples are platinum
with 33.8% of the active nucleus 195Pt and carbon with just 1.1% of
the active 13C. The spectra of other nuclei in compounds of these
elements will thus arise from differentiable molecular species: those
with the non-active isotope, which will give rise to intense patterns
depending upon the other nuclei present, and those with the active
isotope, which will give a weaker sub-spectrum in which spin-spin
coupling will be seen arising from interaction with this less abundant
isotope. These weak doublets are centred approximately around the
corresponding lines in the spectrum of the main species and are thus
called satellites. They are usually particularly obvious in !H or 31P
spectra of platinum compounds, since their intensity is about a quarter
of that of the central line of the major species. We will give two exam-
ples of compounds containing one and two platinum atoms.
First, we will consider the proton spectrum of the mononuclear
square planar complex ion, [PtMe(PMe2Ph)3]+. The methyl groups
attached to platinum and to phosphorus are both coupled to 195Pt, and
their spectra, in the low frequency region, are shown in Fig. 3.16, with
the effect of the 31P atoms removed by a process of double irradia-
tion, which we will discuss in detail in Chapter 5. There are three
major methyl resonances which originate from the complexes with
magnetically inactive platinum, and each is flanked by two lines of
one quarter intensity, which are the doublets due to coupling to one

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4

1
H/ppm
Figure 3.16 The 400 MHz 1H spectrum of [PtMe(PMe2Ph)3][PF6] in CDC13, with the effect of the 31P
nuclei, / = 1/2, removed by a double irradiation technique. The P-Me and Pt-Me resonances are 1:4:1
triplets, the weaker lines in each triplet being spin coupling satellites due to protons coupled to the minor
platinum isotope 195Pt.

www.pdfgrip.com
 62 Internuclear spin-spin coupling

spin-1/2 nucleus, 195Pt. The shorter-distance coupling path gives an

appreciably stronger coupling than the longer-distance one via phos-
phorus. Evidently, the values of the coupling constants give structural
information and the existence of the 1:4:1 pattern is good proof of
the presence of 195Pt in the complex.
If we now consider what sort of molecular populations we might
expect if we have a complex that contains two platinum atoms, we will
realize that we should have three sub-spectra, one each from those
complexes which contain no 195Pt, those which contain one and the
smaller number which contain two. If the complex is symmetric, such
(3.1) as (3.1), then we get three interlaced spectra as shown in Fig. 3.17,
which gives the 31P spectrum with the couplings to the protons now
suppressed by double irradiation. This complex contains 43.8% of its
molecules with no 195Pt and in which the phosphorus atoms are entirely
equivalent and give a central, strong singlet. Some 44.8% of the mole-
cules contain one 195Pt atom and this is coupled unequally to the two
phosphorus nuclei with lJ(l95Pt3lP) - 2409 Hz and 2/(195Pt31P) = 783 Hz.
This makes the two phosphorus atoms non-equivalent and they also
can couple to one another with 2/(31P31P) = 163 Hz. The resulting sub-
spectrum contains two doublets, one widely spaced due to the 31P
directly bonded to 195Pt and one with a smaller spacing due to the
more distant 31P nucleus. The phosphorus nuclei then split mutually

(a)
40 20 0 -20 6(31P)

(b)

(c)
11 11
(d)
Figure 3.17 The 31P NMR spectrum of [Pt2(CO)2(PPh3)2(fi-MeO2CC>
CCO2Me)], (3.1), with the effect of the !H nuclei removed by a double
irradiation technique, (a) The experimental spectrum with signals due to
coupling to 195Pt, I = 1A, shown above with increased gain, (b) The simulated
spectrum for the molecules containing no 195Pt. (c) The simulated spectrum
for the molecules containing one 195Pt atom, (d) The simulated spectrum for
the molecules containing two 195Pt atoms. (Reproduced with permission from
Kose et al (1981) Inorg. Chem., 20, 4408, copyright (1981) American Chemical
Society.)

www.pdfgrip.com
 Spin-spin coupling satellites 63

into doublets, to give eight lines in all. The remaining 11.4% of the
molecules contain two 195Pt atoms and give a weak spectrum. This sub-
spectrum is a second-order spectrum type, [AX]2, to be discussed in
section 3.5.4. Analysis of this sub-spectrum yields 1/(195Pt195Pt) =
786 Hz. The spectra become much more complex for bigger clusters
and are of considerable use in studying the structures of this inter-
esting class of compound.
Another nucleus that provides interesting examples of spin coupling
satellites is 183W, which has a natural abundance of 14.28%. The 19F
spectrum of the binuclear complex [W2O2F9]~ is shown in Fig. 3.18,
and consists principally of a doublet of intensity 8 and a nonet of inten-
sity 1. The nine fluorine atoms can thus be divided into an isochro-
nous set of eight atoms and one unique atom. The main spectrum
arises from those molecules in which both tungsten atoms are the
magnetically inactive tungsten isotopes but satellite lines are also
observed due to the molecules with one 183W atom in their structure.

Figure 3.18 The 56.4 MHz 19F spectrum of [W2O2F9]-, (3.2), in (MeO)2SO,
showing spin satellites due to 14.28% of 183W. (a) The multiple! due to183the
terminal fluorides, (b) Expansions of the central signals which are the W
satellites, (c) The multiplet of the bridging fluoride recorded at higher gain,
(d) The structure of the molecule showing the various coupling constants in
hertz. The outer lines of the nonet are lost in the baseline noise and the
student should confirm that the intensity ratios of the observed lines corre-
spond to those expected for the inner seven of the nonet rather than to those
expected for a septet. (After McFarlane et al. (1971) /. Chem. Soc. A, 948,
with permission.)

www.pdfgrip.com
 64 Internuclear spin-spin coupling

In principle, lines should also exist due to those molecules with two
183
W atoms, but their proportion is low and their resonances are too
weak to observe. Each of the lines of the intense doublet has two 183W
satellites, each of which is further split into a 1 : 4 : 6 : 4 : 1 quintet.
This pattern must arise from coupling to four fluorine atoms. We can
therefore conclude that we have four of the eight isochronous fluo-
rine atoms associated with the 183W atom and therefore split into a
satellite doublet and then further coupled to the remaining four, which
are equally associated with the NMR inactive tungsten isotopes. This
provides considerable confirmatory evidence that the structure is
[OWF4»F»WF4O]~ with a fluorine atom bridging the tungsten atoms.
Finally we must consider the effect that the 1.1% of naturally occur-
ring 13C has on the proton spectra of organic compounds, which contain
principally the inactive 12C. Some of the molecules will however

JCH'H) °J( H'H)

I /
7.0 6.9 6.8 6.7 6.6
6(1H)
Figure 3.19 The 400 MHz !H NMR spectrum of the olefinic protons of dimethyl fumarate,
MeO2CCH=CHCO2Me in CDC13. In addition to a spectrum at normal gain,13 above there is a spectrum
with 13increased gain to show the 13C satellites. The molecules, MeO 12
2C CH= CHCO2Me show coupling.
The CH proton is split into a large doublet, 1J(13C1H) = 167.9 Hz, which is further split by the 12CH
proton with 3J(1H1H) = 15.8 Hz. The 12CH proton is split into a small doublet, 2J(13C1H) = 5.9 Hz, which
is further split by the 13CH proton with 3J(1H1H) = 15.8 Hz.

www.pdfgrip.com
 The description of spin systems 65

contain one 13C atom, distributed randomly. In fact, the spin satellites
due to this minor isotopic component are not easy to observe among
the intense JH-12C resonances, but have proved to be of considerable
use. In simple compounds such as acetone, (CH3)2CO, for instance,
the proton resonance has two pairs of spin coupling satellites due to
molecules with one 13C atom in the methyl group (1J(1H-13C) = 126 Hz}
and to those with 13C in the carbonyl group [2J(1H-13C) = 5.9 Hz}.
Thus we can measure proton-carbon coupling constants, and with
double-resonance techniques we will see later that we can discover
correlations between the proton and carbon spectra of a molecule, see
Chapter 8.
The reduction of symmetry in a molecule caused by the presence
of 13C can also prove valuable. A simple example is to determine
whether dimethyl fumarate, MeO2CCH=CHCO2Me, has a cis- or
frYws-configuration at the double bond. The compound gives two
singlets in the ratio 1 :3 in the 1H NMR spectrum due to the olefinic
CH and the methyl protons. In the 12C compounds, the olefinic protons
are equivalent and give a singlet. However, the presence of one olefinic
13
C carbon atom removes the equivalence and the coupling between
the two olefinic protons can be measured as 15.8 Hz (Fig. 3.19). The
15.8 Hz coupling is in the range to be expected for a trans-stereo-
chemistry about the double bond in view of the electronegativity of
the substituents.

3.4 THE DESCRIPTION OF SPIN SYSTEMS

3.4.1 The alphabetic description of spin systems

A nomenclature has been adopted for these cases in which the magnet-
ically non-equivalent sets of spins are labelled with letters from the
alphabet, choosing letters that are well separated in the alphabetic
sequence to signify large chemical shift separation. Thus CH2C1CHC12
is an A2X system, CH3CH2R is an A3X2 system, and CH3CH2F is an
A3M2X system. Their spectra are called first order.
If the chemical shift between coupled groups is reduced, then the
rules given above no longer apply, the coupling patterns become
distorted and more complex and the spectra are said to become
second order. To signify this and the fact that the chemical shifts
between the coupled nuclei are relatively small, the spins are labelled
with letters close together in the alphabet. Thus, for example, two
coupled protons resonating close together are given the letters
AB, and an ethyl group in (CH3CH2)3Ga, where the methyl and
methylene protons resonate close together, is described as an A3B2
grouping. Mixed systems are also possible and a commonly encoun-
tered one is the three-spin ABX grouping where two nuclei resonate
close together and a third is well shifted or is of a different nuclear
species.

www.pdfgrip.com
 66 Internuclear spin-spin coupling

3.4.2 Magnetic equivalence

Coupling is only observed between nuclei which are magnetically
inequivalent. What is meant by magnetic equivalence? Nuclei are
magnetically equivalent if they satisfy two conditions:
(3.3) 1. They have the same chemical shift. In most cases this is achieved
as a result of symmetry and/or dynamic behaviour which averages
their properties.
2. They have the same coupling constant to any other magnetically
inequivalent nucleus.
The first condition is easily understood, but the second is quite
subtle. A classic example is 1,1-difluoroethene, (3.3). The two protons
have the same chemical shift as do the two fluorines. In both cases
they are related by both a C2 axis and a mirror plane. However, they
do not fulfil the second condition. /(HA-FA) */(HB-FA). Similarly,
/(HA-FA)*/(HA-FB). Hence the two protons and the two fluorines
(3.4) are magnetically inequivalent. The spectrum is shown in Fig. 3.33.
As the two protons are magnetically inequivalent, they can couple
to each other. As they have the same chemical shift, and are as close
as they can be in chemical shift terms, the spectrum is second order
and complex. This gives rise to an [AX]2 spin system, see section 3.5.4.
The symbolism AA'XX' with primed letters is frequently used as an
alternative description for this spin system. However, this notation is
now being used to differentiate certain slightly different types of
system. The [AX]2 notation requires the two A nuclei and the two X
nuclei each to have the same chemical shift as a result of the symmetry
of the molecule. The AA'XX' notation is used when the two A nuclei
(3.5) and the two X nuclei each have the same chemical shift as a result of
accidental degeneracy. An example of this will be given in section
3.5.3.
[AX]2 spectra are frequently encountered. Some common examples
are 1,2-dichlorobenzene, (3.4), 4-nitrophenol, (3.5), and ButCH2CH2
SiMe3. In the latter case, the bulk of the Bul and SiMe3 groups results
in the drafts-configuration predominating (Fig. 3.20).
The resulting [AX]2 spectra are second order and cannot be analysed
using simple first-order rules given above (Fig. 3.21). A procedure to
analyse [AX]2 NMR spectra is given in section 3.5.4.

3.4.3 Diastereotopic atoms and groups

Figure 3.20 A Newman Apparently chemically equivalent atoms and groups can be inequiva-
projection of Ph3SiCH2 lent due to diastereotopism. The simplest example is when a CH2R
CH2SiMe3, showing the group is attached to a chiral centre, -CXYZ. There are three rotamers
trans-anangement of the possible, (3.6), (3.7), and (3.8). If one rotamer predominates, say (3.6),
Ph3Si and SiMe3 groups. then it is easy to see that HA and HB are in different environments,
The A result is that
/(H -H X ) * 7(H A -H X ') with HA spending most of its time between groups X and Z, while HB
and the spin system is spends most of its time between groups X and Y. Even if there are
[AX]2.

www.pdfgrip.com
 The description of spin systems 67

50 Hz

(b)

XX
(a).
Figure 3.21 A partial 250 MHz !H NMR spectrum of Ph3SiCH2CH2SiMe3 in CDC13, showing the CH2CH2
protons, (a) Experimental spectrum, x = impurity, (b) Calculated spectrum. (Reproduced with permission,
V.E. McGrath, PhD thesis, Sheffield, 1993.)

equal populations of all three rotamers, the chemical shifts of HA and

HB can still be different as the X-C-C-H8 torsion angle in (3.6) does
not have to equal the X-C-C-HA torsion angle in (3.8).

www.pdfgrip.com
 68 Internuclear spin-spin coupling

Diastereotopism is not restricted to -CH2R groups, and is observed

in any group of the type -MA2B, where M is a tetravalent atom. The
methyl resonances of -SiMe2But and -PMe2Ph groups are commonly
used to probe the symmetry of molecules using this effect. The
diastereotopic centre can be several bonds removed in the molecule,
and the only test that need be applied is whether or not there is a
plane of symmetry through the bond joining the -MA2B moiety to the
(3.9) rest of the molecule. Hence in (3.9) the PMe2Ph methyl groups are
equivalent, while in (3.10) they are inequivalent.
Molecules need not be chiral to produce diastereotopic atoms or
groups. This is the case in (3.10). Another common example is
where two identical groups are attached to one carbon atom, e.g.
MeCH(OCH2CH3)2. From the viewpoint of one -OCH2 group, there
are three different groups attached to the tertiary carbon, and X = H,
(3.10) Y = Me, and Z = OEt in the rotamers (3.6) to (3.8). Consequently,
the CH2 protons are diasterotopic, the spectrum being shown in
Fig. 3.22.
The common presence of diastereotopism in molecules leads to
coupling between apparently equivalent CH2 protons being frequently
observed.

5.0 4.0 3.0 2.0 1.0 0.0

1
H/ppm
Figure 3.22 The 400 MHz 'H NMR spectrum of MeCH(OCH2CH3)2 in CDC13. The methyl protons are
seen between 1.2 and 1.4 ppm, the CHMe being a doublet and the methyl groups of the ethoxys a triplet.
The CH proton gives a quartet at 4.69 ppm. The geminal CH2 protons give individual resonances at
3.65 and 3.49 ppm which are split into multiplets by coupling to each other and to the vicinal methyl
protons.

www.pdfgrip.com
 Second-order effects 69

3.5 SECOND-ORDER EFFECTS

The rules so far discussed apply to spectra of nuclei of the same species
where the separation between multiplets in hertz (i.e. the chemical
shift) is large compared to the value of the coupling constant between
them, or to coupling between nuclei of different elements or isotopes
where the differences in NMR frequency are invariably large.
Second-order spectra arise when the frequency separation between
multiplets due to different magnetically equivalent sets of nuclei is
similar in magnitude to the coupling constant between them; under
these circumstances, the effects due to spin coupling and chemical shift
have similar energy and become intermingled, leading to alterations
in relative line intensities and in line positions. Because it is the ratio
between / and the frequency separation that is important, chemical
shifts are always expressed in hertz (Hz) and not in parts per million
(ppm) when discussing second-order spectra. The hertz separation is
obtained by multiplying the chemical shift 8 by the spectrometer oper-
ating frequency and is written v08. The perturbation of the spectra
from the first-order appearance is then a function of the ratio //v08
and is different for spectrometers operating at different frequencies.
If a high enough frequency is used, many second-order spectra
approach their first-order limit in appearance and are then much more
easily interpreted. This is one of the advantages of high-field instru-
mentation. We will describe first the simplest possible system consisting
of two 7 = 1/2 nuclei.

3.5.1 The AB second-order system

We consider a system consisting of two isolated but mutually spin-
coupled / = 1/2 nuclei with different chemical shifts. When the
chemical shift between them is large, much larger than the coupling,
then we see two doublets. A typical arrangement with / = 10 Hz and
v08 = 200 Hz is shown in Fig. 3.23(a). If we reduce the chemical shift
progressively to zero, we can imagine the two doublets approaching
one another until they coincide. However, we know that an isolated
pair of equivalent nuclei give rise to a singlet, and the problem is how
can two doublets collapse to give a singlet. If the coupling constant
remains at 10 Hz, why do we not get a doublet A2 spectrum?
The behaviour of the multiplets can be predicted using a quantum-
mechanical argument. The system can have any one of four energy
states which are characterized by the spin orientations of the two
nuclei. The wave functions of the two orientations are normally written
a for / = +1/2 and p for / = -1/2. There are four such spin states
with the wave functions aa, a(3, (3a, and (3(3, where the first symbol
in each pair refers to the state of the A nucleus and the second symbol
to the state of the B nucleus. The energies of the states aa and (3(3
can be calculated straightforwardly, but the two states a(3 and (3a have
the same total spin angular momentum and it is found that the

www.pdfgrip.com
 70 Internuclear spin-spin coupling

=00
J- =
v08
(h)

- - =2.5
v08
(9)

1
vV
08
(f) JUUUl

-i = 0.866

<•>

- = 0.4
(d)

-4 =0 - 2
v06
(0

=0
v—s -1
08
(b)

-: =0.05
v08
(a)
1 1 1 1 1 1 1 1 1

100 50 0 -50 -100

Hz
Figure 3.23 AB quartets for several values of the ratio 7/v08. In each case, the coupling constant is kept
at 10 Hz, and the chemical shift separation reduced in the sequence 200, 100, 50, 25, 17.32, 10, 4, and 0 Hz.
The result is the inner lines become more dominant and the outer lines become weaker.

www.pdfgrip.com
 Second-order effects 71

quantum-mechanical equations can only be solved for two linear

combinations of ap with pa. This is described as a mixing of the states
and means that none of the observed transitions corresponds to a pure
A or a pure B transition. The form of the wave functions and the
energy levels derived are shown in Table 3.2. Cl and C2 are constants,
VA and VB are the A and B nuclear frequencies in the absence of
coupling and v0S = I VA - VB I . The transition energies are the differ-
ences between four pairs of energy states, 3-4, 2-4, 1-3 and 1-2, each
transition involving a mixed energy level. Because of the mixing, the
transition probabilities are no longer equal as in the first-order case
and intensity is transferred from lines in the outer parts of the total
multiplet into the central region. The transition energies relative
to the centre of the multiplet, i.e. to the mean frequency M(vA - VB),
and the intensities are shown in Table 3.3.
There are thus four lines as in the AX spectrum but with perturbed
intensities and resonance frequencies. The resulting spectral pattern
and the corresponding energy level diagram are shown in Fig. 3.24.

Table 3.2 Wave functions and energy levels for the AB second-order spin
system

State Wave function Energy /eve/ (7/z)

1
1 cm ^
- (VA + vB) + 4^
.< 1
2
2 C^ap) + C2(pa) ~V(v 0 &) 2 + / - i /
-i ,.

3 -C2(ap) + Q(pa) - -A/(v0S)i2 + 7 2 - ^ /

1 1
4
2 4

Table 3.3 Transition energies relative to the centre of the multiplet and
relative intensities for the AB second-order spin system.

Transition Energy (Hz) Relative intensity

1 1/ l J
a 3-^1 >)2 + J2 V(v0§)2 -h /
2

1 1
2/+2 y)2 + J2
1 ~\-
b 4-*2
°' V(v08)2 -h / 2

tf + J2
1 ~\-
c 2->l 2
V(v08) - h /
2

1 1 i ! J
d 4^3 J ))2 + J2 ± |
h/2
-2 ~2^

www.pdfgrip.com
 72 Internuclear spin-spin coupling

(a) (b)
Figure 3.24 (a) The energy level diagram for a system of two spins related to the resulting AB quartet.
Note that the vertical scale is misleading. The energy gaps represented by the arrows are in MHz, while
the energy gap between energy levels 2 and 3 is in Hz. (b) The spectrum was calculated for / = 10 Hz,
v08 - 25 Hz, i.e. //v08 = 0.4.

The energy levels are marked with the appropriate spin state and the
transitions, which in the first-order case can be regarded as arising
from transitions of the A or B nucleus, form opposite sides of the
figure. The three line separations are a-b = c-d = J and b - c =
V(v08)2 + J2 - \J\. The separation a - c or b - d, which in the first-
order case is the same as the separation between the doublet
centres, and is therefore the chemical shift in hertz, v08, is now simply
V(v08)2 + J2 and is larger than the true chemical shift. In other words,
although v08 is reduced to zero, the doublet centres never coincide
and are separated by / Hz. The outer lines, however, have intensity
zero at this point, while the inner lines are coincident, i.e. we predict
a singlet spectrum as is observed (cf. Fig. 3.23(h)). Thus arises our
rule, 'isochronous coupled protons resonate as a unit'.
We can calculate some simple rules for analysing an AB spectrum:
1. The spectrum contains two intervals equal to /, a - b and c - d.
2. The true AB chemical shift v08 is found as
(a - d)(b -c) ={ V(v08)2 +72 + / } { VO^+T 2 - / }
= (v0S)2 + P-J2
= (v08)2
so
v08 = V ( a - < * ) ( f r - c ) (3.4)
where a - d is the separation between the outermost lines and
b - c is the separation between the innermost pair of lines.
3. The assignment can be checked against the intensity ratios of the
larger and smaller lines. The intensity ratio, stronger/weaker, is

www.pdfgrip.com
 Second-order effects 73

1+
V(v08)2 + /2
J
1- -
V(v08)2 + ^2
which gives

V(v08)2 + /2 + /
V(v0s)2 + /2 - y
for the ratio of the line separations (a - d):(b - c). Note that
changing the sign of / does not alter the pattern.
Figure 3.23 shows the form of the AB quartet for several values of
//v08. It is important to remember that if a multiplet shows signs
of being highly second order then both intensities and resonance posi-
tions are perturbed from their first-order values. A spacing corre-
sponding to /AB remains in AB-type spectra since only one coupling
interaction exists, but in more complex systems the spacings are combi-
nations of coupling constants. On the other hand, if the intensity
perturbation is only slight {Fig. 3.23(a)} then the line positions are not
detectably perturbed.
Examination of the spectra in Fig. 3.23 show several features. The
first visible onset of second-order behaviour is the inner lines becoming
more intense. The two doublets are said to 'lean' towards each other.
This feature can be very valuable in deciding which multiplets are
coupled but care is necessary as the heights of signals can be distorted
by insufficient digitisation, see section 5.6.3. There are two features
which are traps waiting for the unwary and have led to the misinter-
pretation of spectra.
1. When the separation of the two signals is / s , the spectrum has
the appearance of a 1 : 3 : 3 : 1 quartet as if it arises from coupling
to a methyl group (Fig. 3.23(e)).
2. When //v08 is large, the outer lines are weak and can be lost in the
noise leaving an apparent doublet.
Four examples of actual spectra that contain an AB multiplet are
given in Fig. 3.25. We have however reached a point in the history of
NMR where, in many cases, the analysis of second-order spectra can
be avoided by operating at a high enough frequency. In those cases
where this will not work, computer programs are available that permit
the full analysis of virtually any system.

www.pdfgrip.com
 74 Internuclear spin-spin coupling

7.9 7.8 7.7 7.6 7.5 7.4 7.3 7.2 7.1 7.0 6.9
1
(a) H/ppm

3.6 3.4 3.2 3.0 2.8 2.6 2.4

1
(b) H/ppm
Figure 3.25 The 400 MHz !H NMR spectra of some compounds in CDC13 containing AB groupings of
protons, in order of decreasing v08//. (a) (PhCH=CH)2CO. Note the AB pattern arising from the CH=CH
group at 8 7.08 and 7.74. The signal at 8 7.61 is due to the orr/zohydrogen atoms of the phenyl groups,
while the meta-and /?0ra-hydrogen atoms are at 8 7.4. (b) The 400 MHz !H NMR spectrum of
C(CH2OCH2CHCH2O)4 in CDC13. The inequivalence of the CH2 protons is caused by the presence of a
chiral CH group in1 the epoxide ring. This produces 5 a substantial chemical shift difference for the CH2
protons on both C and C3, but is also observed at C H2, where a tightly coupled AB pattern is produced
centred at 83.4. (c) The 101.2MHz 31P NMR spectrum with the {H coupling removed of [PtCn2-
CH2=CHCO2Me)(PPh3)2] in ^-toluene. Note the three AB patterns centred at 13, 32, and 51 ppm due to
3ip_3ip coupling. The AB patterns at 13 and 51 ppm arise from coupling to 195Pt, I = 1/2, 34% abundant,
and the AB pattern at 32 ppm rises from the 31P attached to NMR inactive platinum. (Reproduced from
Chaloner et al (1997) /. Organomet. Chem., 527, 145, copyright (1997), with permission from Elsevier
Science.) (d) 109 MHz 31P NMR spectrum with TH coupling removed of [Pt2Cl2(CO)2{P(naphthyl)Ph2}2] in
CDC13. The signal at 30 ppm is due to molecules containing no NMR active platinum nuclei. There are
two AB patterns, one at 32.2 and 40 ppm, and the other195at 20 and 27.6 ppm, due to [Pt195PtCl2(CO)2{P(naph-
thyl)Ph2}2]. There are additional weak signals due to [ Pt2Cl2(CO)2{P(naphthyl)Ph2}2]. (Reproduced from
Mingos et al. (1997) /. Organomet. Chem., 528, 163, copyright (1997), with permission from Elsevier
Science.)

www.pdfgrip.com
 Second-order effects 75

I I I I

55 50 45 40 35 30 25 20 15 10 5
31
(c) P/ppm

40 30 20
31
(d) P/ppm

www.pdfgrip.com
76 Internuclear spin-spin coupling

3.5.2 The AB2 second-order system

The complete analysis of the AB2 spin system will not be performed
here, but an examination of the spin system is instructive, as on going
from the AX2 spin system to the AB2 one, not only do the positions
of lines change, but lines which were degenerate singlets become non-
degenerate doublets. This is a common feature of second-order
spectra, with, in some cases, what would be sharp multiplets in a first-
order spectrum becoming broad due to the overlap of many lines,
degenerate in the first-order case but moving apart in the second-order
case, though not enough to be resolved.
A calculated AB2 spectrum is given in Fig. 3.26. When 7/v08 is small,
a first-order spectrum is observed with a 1 : 2 : 1 triplet for A and a
1 :1 doublet for B. When //v08 is significant, the central line of the
triplet and both lines of the doublet split, resulting in eight lines. This
can be misleading causing the erroneous interpretation that two
different coupling constants are operating in the A and B parts of the
spectrum. The derivation of the chemical shifts and coupling constant
is trivial. If the lines are lettered a to h from the weaker A to the
stronger B signals, then
VA = C

"B=^

J
_ \a - d + f-h\
AB ~ ~

-]-r-J i I ' I ' I ' I ' I ' I ' I ' I ' I

10 0 -10 Hz

Figure 3.26 A calculated AB2 spectrum with VA = 5 Hz and v# = -10 Hz, and
/AB = 4 Hz. Dotted lines are included to mark the chemical shifts of A and B.

www.pdfgrip.com
 Second-order effects 77

3.5.3 The ABX second-order system

It is to be expected that the positions of the lines of nuclei involved
in second-order spectra are moved from their ideal or first-order posi-
tion, but less evident is the fact that nuclei distant in frequency which
couple with the second-order group of nuclei are also affected. The
simplest example of this occurs for the ABX spin system. In addition,
the form of the spectrum depends on the relative signs of the coupling
constants /(AX) and /(BX). This behaviour is frequently encountered
in second-order spectra.
The ABX spin system also introduces the concept of sub-spectral
analysis. As the frequency difference between the two AB and the X
nuclei is large compared with /(AX) and /(BX), the spin states of X
are not perturbed, and it is possible to analyse the AB part of the
system as two separate problems for each of the two possible signs of
the magnetic quantum number, m, of X. Hence the AB part of the
spectrum consists of two AB sub-spectra. See Fig. 3.27, which has been
calculated by computer simulation using input data supposing that
A, B, and X are protons. The two AB sub-spectra can readily be

1020 1010 1000 990 980 220 210 200 190 180
Hz Hz

Figure 3.27 Simulated ABX spectrum using 8(A) = 190 Hz, 8(B) = 210 Hz, 5(X) = 1000 Hz, /(AB)
10 Hz, /(AX) = 4 Hz and /(BX) = 13 Hz.

www.pdfgrip.com
78 Internuclear spin-spin coupling

identified as they are centro-symmetric and have the same /(AB)

(10.0 Hz in this case). Examination of Fig. 3.27 shows the AB groups
to be g, h, k, m and i, j, 1, n.
Taking the first sub-spectrum, its centre of gravity is 204.23 Hz, half
way between lines h, k, or better, a quarter the sum of the shifts of
all four lines. From equation (3.1) in section 3.5.1 we have:
Separation = V(g - m)(h - k} = 24.46 Hz
between the A and B resonances in this multiplet.
On the scale of the spectrum, these points are at 204.23 ± 12.43 Hz
or 216.46 Hz and 192.00 Hz.
For the second AB sub-spectrum, we find that the centre of gravity
is 195.72 Hz , whence the AB shift separation is 15.42 Hz, placing them
on the spectral scale at 195.72 ± 7.41 Hz or 204.43 Hz and 188.01 Hz.
These four points at 216.46, 204.43, 192.00 and 188.01 Hz are the
frequencies of the A and B signals in the absence of AB coupling but
still with coupling to X. There are two solutions which give the same
AB sub-spectra and we need to find both then decide which is correct.
The two solutions are:
1. A frequencies 192.00, 188.01; difference 4 Hz; average A shift
190 Hz, B frequencies 216.46, 203.43; difference 13 Hz; average B
shift 210 Hz;
2. A frequencies 203.43, 192.00; difference 11.4 Hz; average A shift
197.72 Hz, B frequencies 216.46,188.01; difference 28.5 Hz; average
B shift 202.24 Hz.
The differences are the coupling constants of A or B to X and the
averages of the doublet A or B frequencies is then the actual A and
B shifts. The difference between the two solutions is that for 1, /(AX)
and /(BX) have the same sign whereas in 2 they have opposite signs.
It is evident that the first solution is the correct one from the constants
fed into the simulation. In practice, however, we would not know this
and would have to make a logical choice, for which there are two bits
of information to help us in the present case. One is that our experi-
ence should tell us that 28.5 Hz is unreasonably large for a proton-
proton coupling constant so that 1 is the correct solution. The second
is to realize that the X part of the spectrum depends on the relative
signs of/(AX) and /(BX), even though the AB part does not. A simu-
lated spectrum will show which is the correct solution, as is demon-
strated by Figs 3.27 and 3.28, which latter shows a simulated ABX
spectrum using the alternative parameters. Comparison shows that all
the lines are in the same positions, but there are marked changes in
the intensities of lines a, c, d, and f. It is generally easiest to compare
the two solutions by simulating the spectra.
It is very tempting to analyse the X part of the spectrum to obtain
/(AX) and /(BX). In many cases it has the appearance of a doublet
of doublets, especially when the outer weak lines, a and f, are lost in
the noise. The separation of lines b and e does give I/(AX) +/(BX)I,

www.pdfgrip.com
 Second-order effects 79

1020 1010 1000 990 980 220 210 200 190 180
Hz Hz
Figure 3.28 Simulated ABX spectra using 8(A) - 197.72 Hz, 8(B) - 202.24 Hz, 8(X) - 1000 Hz, /(AB)
10 Hz, /(AX) = 11.4 Hz and /(BX) - -28.5 Hz.

but the lines c and d move together as a result of the second-order

nature of the AB part of the spectrum and true values of /(AX) and
/(BX) are not derived from a simple first-order analysis. In the
extreme, the lines c and d can merge to give a deceptively simple
triplet, as is discussed later.
To emphasize the sensitivity of the spectra to the signs of /(AX)
and /(BX), we calculate a spectrum in Fig. 3.29 using the data for
Fig. 3.27 in which only /(BX) has been changed from +13 to -13 Hz.
Note that the spectrum differs substantially from that given in
Fig. 3.27 as a result of the change of sign of /(BX). The sensitivity of
second-order coupling patterns to sign matters. For example, in mol-
ecules containing sp3 carbon atoms 2J(lHCllrL) is negative, while
3 1
J( HCC1H) is positive and this influences the nature of the second-
order spectra obtained from such groups.
Note that there are four states of the AB nuclei (aa, a(3, Pa, (3(3)
and there should, on a simple view, be four X lines, but six may be
observed. This is due to combined transitions of the type a(3p-(3aa
for which Am = 1 even though all three spins flip.
The ABX spin system is often observed in 13C NMR spectra where
the AB nuclei are abundant / = 1/2 nuclei, such as 19F, 31P, or 103Rh.
This occurs because 13C is only 1.1% abundant so that there is effec-
tively only one such atom present per molecule. An example is shown
in Fig. 3.30, for [(Ti5-C5Me5)2Rh2(fjL-CH2)2{fju-CH2CH(CH2CH=CH2)
CH2}], (3.11). The 13C NMR signal shown arises from those molecules
with a 13CH2 group in the hexylene bridging ligand. The chemical shifts
of the two 103Rh nuclei in this isotopomer are different due to a
secondary isotope shift of 0.36 ppm due to one 103Rh nucleus being

www.pdfgrip.com
 80 Internuclear spin-spin coupling

\J

1020 1010 1000 990 980 220 210 200 190 180
Hz Hz
Figure 3.29 Simulated ABX spectra using 8(A) - 190 Hz, 8(B) = 210 Hz, 8(X) - 1000 Hz, /(AB) - 10 Hz,
/(AX) - 4 Hz and /(BX) - -13 Hz.

(3.11)

Figure 3.30 The 100.62 MHz 13C NMR spectrum of the 13CH2 group of
[(Ti5-C5Me5)2Rh2(fjL-CH2)2{fju-CH2CH(CH2CH-CH2)CH2}], (3.11). The spec-
trum is ABX with the two 103Rh nuclei giving the AB part of the spectrum,
not shown, and the 13CH2 giving the X part. This spectrum has 1/(103Rh103Rh)
- 13.5 Hz, y^Rh^C) - 30.0 Hz, 2/(103Rh13C) - 0.0 Hz, and A8(103Rh) -
0.36 ppm or v08 = 4.5 Hz. (Reproduced with permission from Mann et al
(1985) /. Chem. Soc., Dalton Trans., 1555.)

attached to 12C and the other to 13C. A double irradiation process is

used to eliminate any effect of the hydrogen nuclei on the spectra.
When /(AB) is much larger than the chemical shift difference
between the A and B nuclei, the X part of the spin system collapses
to a 1:2:1 triplet with the two outer lines separated by

www.pdfgrip.com
 Second-order effects 81

I/(AX)+/(BX)I and the rest of the intensity concentrated in the

centre of the multiplet. This is often observed in the 13C NMR spectra
of ligands of metal complexes with two identical phosphorus ligands.
Frequently these spectra are referred to as AA'X rather than as ABX,
to indicate that no 12C/13C isotope shift of the phosphorus has been
detected. An example of this occurs with the meta-carbon in the phenyl
groups of Ph2PCH2CH2PPh2. The spectrum is a 1 : 2 : 1 triplet (Fig.
3.31). It is tempting, but wrong, to attribute the occurrence of the
triplet to equal coupling between the meta-carbon and both 31P nuclei.
It arises because 3/(31P31P) is large compared with 2/(31P13C) and the
spectrum is AA'X. Such multiplets are referred as being 'deceptively
simple'. The only coupling information that can be derived from such
a spectrum is I3/(31P13C) + 6/(31P13C)l from the separation of the outer 128.4 128.3ppm
lines. Figure 3.31 The 100.62
MHz 13C NMR spec-
trum of the meta carbon
3.5.4 A four-spin [AX]2 system of the phenyl groups
of Ph2PCH2CH2PPh2 in
As discussed in sections 3.4.1 and 3.4.2, the [AX]2 spin system CDC13 with coupling to
frequently arises for molecules such as (3.3) to (3.5). Both the A and hydrogen removed by dou-
X parts of the spectrum are identical in appearance. The spectra can ble resonance.
be readily analysed. 1,1-difluoroethene, (3.3), will be taken as an
example. A simulated spectrum of either the 1H or 19F NMR signal is
shown in Fig. 3.32. Sub-spectral analysis indicates that when both spins
of one nucleus are in the same state (aa or (3(3), the other two nuclei
will resonate as a unit and produce a singlet. There are two such
singlets separated by the sum \J(A1X1)+J(A1X?)\. The analysis is

50 40 30 20 10 0 -10 -20 -30 -40 -50

Hz
Figure 3.32 A simulated spectrum for either the :H or 19F NMR spectrum of
1,1-difluoroethene, (3.3).

www.pdfgrip.com
 82 Internuclear spin-spin coupling

straightforward. It is usual to define some terms.

K = J
A1A2 + JX1X2

L = JAW-JAW
M= J
A1 A2 ~ J X1 X2

N = / A i x i + /AiX2
Each multiplet consists of an intense doublet, separation N, and two
AB patterns, apparent coupling constants K and M. The inner lines
of each AB pattern are separated by V^2 + L2 - K and VM2 + L2 -
M respectively.
TV can be easily determined, 34.6 Hz. K and M are 31.6 and 41.2 Hz.
The corresponding values of A/X2 + L2 - K and VM2 + L2 - M are 14.2
and 11.6 Hz. Substituting values of K and M yields L = 33.2 Hz. The
coupling constants are then obtained from combinations of K, L, M,
and N as follows.
7 K + M on „ TT
<W = —2— = 39.4 Hz
K
/Xlx2 = ~M = 4.8 Hz

A—I I iV
33
/-»/-% f\ f-w-
/ A i x i = —2— = -9 Hz

'A'X>=^^- = 0 - 7 Hz

Typically, in inorganic chemistry, [AX]2 systems are found in metal

phosphine complexes such as cis- and rransi-[PdI2{PF(OPh)2}2] where
the nuclei A are the 31P and the nuclei X are the 19F (Fig. 3.33). The
remaining nuclei either have no influence on the 31P or 19F spectra, or
their effect is removed by double irradiation. There need be no
coupling between the X nuclei, though in fact there will be a small,
not necessarily observable, X-X coupling. There will be strong
coupling between the A nuclei via the metal atom. It should be noted
that the metal atom need not be present from the point of view of
the NMR analysis of the system.
Were we to investigate more complex spin systems, for example bis
complexes of methyl phosphines, we would find the same general
features. An example is shown in Fig. 3.34. This is a 1H NMR spec-
trum of [PdCl(PMe3)3]+, (3.12). The PMe3 trans to Cl is a doublet of
(3.12) triplets. The doublet coupling is 2/(31P1H), i.e. proton-phosphorus
coupling in the same ligand, while the triplet coupling, 4/(31P1H) is to
the phosphorus in the two equivalent PMe3 ligands. These latter two
PMe3 ligands give rise to an [AX9]2 spin system giving the triplet to
low frequency. The structure of the spectrum is very similar to that

www.pdfgrip.com
 Second-order effects 83

Figure 3.33 An example of magnetic inequivalence in a chemically equiva-

lent system. The 31P spectrum, with the effect of the protons removed by
double irradiation, of a mixture of d5i-[PdCl2{PF(OPh)2}2], marked with a c
and rrarcs-[PdCl2{PF(OPh)2}2], marked with a t. Both complexes give rise to
[AX]2 spin systems, the two phosphorus nuclei being inequivalent because the
two P - F couplings are unequal. The spectra of the two complexes are dras-
tically different from the simple 1 : 2 : 1 triplet that would have been observed
for an A2X2 spin system. (Reproduced from Crocker and Goodfellow (1981)
J. Chem Res(M), 742 with permission.)

Figure 3.34 The 100 MHz 1H NMR spectrum of [PdCl(PMe3)3]+ in CH2C12.

(Reproduced with permission from Goggin et al. (1973) /. Chem. Soc. Dalton
Trans., 2220.)

www.pdfgrip.com
 84 Internuclear spin-spin coupling

of the [AX]2 spin system, with the intense TV doublet and a series of
AJB patterns, but on account of the large value of 2/(31P31P), which
means that K, M, etc. are large, the central lines merge into a broad
central singlet and the outer lines vanish into the noise. The result is
a deceptively simple triplet. On the other hand, when two phosphorus
ligands are equivalent and are mutually cis9 2/(31P31P) is small for Pd11
and other complexes, K, M, etc. are small and now the lines crowd
ground the N doublet lines and only a doublet is observed, see Fig.
3.33, complex c. This behaviour has proved to be a valuable tool in
determining the stereochemistry of phosphorus complexes of Os11,
Ru11, Rh1, Ir1, Rhm, Irm, Pd11, Pt11, PdIV, and Pdiv.
Figure 3.35 shows the proton spectrum of [PtMe(PMe2Ph)3]+, already
encountered in Fig. 3.16, where the effect of the 31P atoms was
removed by double irradiation. The full spectrum contains a triplet at
8 1.68 due to the mutually trans PMe2 protons, a doublet at 8 1.29 due
to the PMe2 protons trans to the PtMe group, and a doublet of triplets
due to the Pt-Me group at 8 0.54. In each case, the signals are flanked
by 195Pt satellites.The methyl groups of the phosphine ligands cannot
in any way couple equally to the two phosphorus atoms and their
triplet is the deceptively simple triplet described above. Such triplets
are taken to indicate trans structures for such complexes, whereas
doublets arise from cis configurations.
An example of an organic molecule that has a four-spin [AX]2 spec-
trum is furan, whose proton spectrum is shown in Fig. 3.36 together

Ik. •>_/v/^-<sAA/LA_A_/v*^
\ I I I I I I I I I I I I \
1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4
1
H/ppm
Figure 3.35 The 400 MHz !H NMR spectrum of the complex [PtMe(PMe2Ph)3][PF6] in CDC13 showing
the three methyl resonances in the intensity ratio 4:2:1. The flanking resonances are 195Pt satellites. The
small triplet splittings of the signal at 8 1.68 are due to coupling to the two magnetically inequivalent but
chemically equivalent 31P atoms giving rise to a deceptively simple triplet. The Pt-Me at 8 0.55 is a doublet
of triplets due to coupling to the 31P atoms.

www.pdfgrip.com
 Second-order effects 85

with a formula provided with the various proton spin-spin coupling

constants. The resonances of the a and p protons appear at first glance
to be triplets, but close inspection at high resolution shows the central
line to be composite. A full analysis of the system gives a doublet
spectrum plus a pair of AB sub-spectra with most intensity in the
central resonances. Interestingly, the 13C spin satellites of this mol-
ecule are first order in appearance. The presence of a 13C atom in a
molecule splits the resonance of the attached proton into a widely
separated doublet, so that it is no longer isochronous with its twin.
This introduces an effective chemical shift, which permits these
hydrogen atoms to couple with each of the other three and give simpli-
fied spectra that can aid interpretation of the main spectrum.

8.0 7.5 7.0 6.5

Figure 3.36 The 400 MHz 1H NMR spectrum of furan in CDC13. Coupling constants and splittings are
given in hertz. 1J(1H - 13C) = 201.4 Hz and '/OH - 13C) = 175.3 Hz respectively.

www.pdfgrip.com
86 Internuclear spin-spin coupling

3.6 QUESTIONS

3.1. Figure 3.6 shows the 1H spectrum of an ethyl group. Measure the
chemical shift of the two multiplets and their coupling constant.
3.2. Verify that the ratio between the coupling constants to 1H of the
nuclei 10B and n B in the [BH4]~ are in the ratio of the magnet-
ogyric ratios of the two boron isotopes (Fig. 3.8). What do you
expect the n B NMR spectrum of [BH4]~ to look like?
3.3. Figure 3.9 shows the proton spectrum of a CD2H group in iso-
topically substituted acetone. 2J(2H1W) =2.2 Hz. Calculate the
coupling constant 2/(1H1H) between the protons in the CH3
groups of normal acetone, ignoring isotope effects on /.
1
3.4. Doublet satellite signals are observed in the H spectrum of
acetone, close to the main singlet signal, with a spacing of 5.9 Hz,
which is the value of the coupling 2J(1H-13C). What will be the
pattern in the 13C spectrum of this carbonyl group, i.e. number
of lines and their relative intensities?
3.5. The 60 MHz !H spectrum of ascaridole features a tightly coupled
AB quartet at about 8 = 6.45 ppm. The line positions, starting at
the highest frequency one, are 395.5, 386.9, 385.5 and 376.9 Hz
from TMS. Calculate 3/(Ha-Hb) and the chemical shift between
Ha and Hb in Hz and ppm. Then calculate the relative line inten-
sities and positions expected when the same sample is observed
at 500 MHz. What is the exact chemical shift of the centre of the
quartet?
3.6. Explain why, in Fig. 3.25(a), the two halves of the AB quartet
have different intensities and so different linewidths.
3.7. Assign the spin systems, e.g. AB, to the following compounds,
a. ethyl chloride
b. CH2F2
H

c.

d.

e.

www.pdfgrip.com
 Questions 87

f. CH2=CHBr
g. [PdCl2(PMe3)2]
h. /0c-[IrH3(PR3)3] (ignore any spins in the R groups).
3.8. Figure 3.37 shows the 19F resonance of the CF2H group of
1,1,1,2,2,3,3-heptafluoropropane, CF3CF2CF2H. The two fluorines
are equivalent and are coupled to all the other magnetically active
nuclei in the sample. Pick out the various multiplet patterns and
measure the coupling constants 2/(F-H), 3/(F-F) and 4/(F-F).

1
i ' i ' i '
40 20 0 -20 -40
Hz

Figure 3.37 The 19F resonance of the CF2H group of CF3CF2CF2H. The spec-
trum has been simulated using experimental coupling constants.

3.9. When IrCl3(PEt2Ph)3 is treated with LiAlH4, then a hydride of

formulation IrHm(PEt2Ph)w is formed. The hydride 1H NMR spec-
trum is shown in Fig. 3.38(a). Use the pattern produced by 31P
coupling to the hydride to deduce n. The 31P NMR spectrum,
with coupling from the ethyl and phenyl protons removed, is
shown in Fig 3.38(b). Deduce the value of m from the coupling
of the 31P to hydride. Also deduce the stereochemistry of the
compound.

(b) 8(31P)

Figure 3.38 (a) 90 MHz J H NMR spectrum31 of the hydride 31signal of

IrHm(PEt2Ph)n in C6D6. The coupling is with P. (b) 36.43 MHz P NMR
spectrum of IrHm(PEt2Ph)n in C6D6. The coupling is with *H of the hydride.
Coupling due to the ethyl and phenyl protons has been removed by decou-
pling. (Reproduced with permission from Mann et al (1971) /. Inorg. Nucl.
Chem., 33, 2195, copyright (1971) Elsevier Science.)

www.pdfgrip.com
Internuclear spin-spin coupling

When IrHm(PEt2Ph)n is warmed in benzene with AsMe2Ph,

then one mole of H2 is evolved per mole of compound and
IrH;c(PEt2Ph)3;(AsMe2Ph) is formed, which has the hydride !H
spectrum in Fig. 3.39. Deduce the values of x and y.
You are given one additional piece of information: it has been
shown empirically for indium complexes that, when a hydride is
trans to a tertiary phosphine, 2J(lH-Ir-3lP) is large (about
120 Hz), but, when a hydride is cis to a tertiary phosphine,
is small (about 15 Hz).

100 Hz

-11 -12 -13 -14 -15 -16

5(1H)

Figure 3.39 90 MHz 1H NMR spectrum of the hydride region of

IrH;c(PEt2Ph));(AsMe2Ph) in C6D6. Note that the relative intensities of the two
signals are 2:1. (Reproduced with permission from Mann et al (1971) /.
Inorg. Nucl Chem., 33, 2195, copyright (1971) Elsevier Science.)

3.10. In d8-THF, the 13C NMR spectrum with the effect of the protons
removed by double irradiation, of the lithiated carbon atom of
[(CH2=13CH)6Li] shows two sets of signals. Fig. 3.40 shows the
13
C NMR spectrum of the lithiated carbon atom of the major
isomer. The coupling is to 6Li. How many 6Li atoms are attached
to each carbon atom?

20 Hz

191.4 191.0 190.6 190.2

6(13C)
Figure 3.40 The 100.6 MHz 13C NMR spectrum of the lithiated carbon atom of
the major isomer of [(CH2=13CH)6Li]n in d8-THF at -100°C (Reproduced with
permission from Brauer and Griesinger (1993) /. Am. Chem. Soc., 115,10871,
copyright (1993) American Chemical Society.)

www.pdfgrip.com
 Questions 89

The signal due to the lithiated carbon atom of the minor isomer
is shown in Fig. 3.41 at both -90°C when the molecule is rigid
and at -60°C when the atoms of the molecule are mobile and
the lithium moves between the vinylic CH carbon atoms in the
molecule. Use the coupling pattern to determine how many
lithium atoms are attached to a lithiated carbon atom in the rigid
structure, and how many lithium atoms visit each vinyl CH carbon
atom.

(a)

183.4 183.0 182.6

Figure 3.41 The 100.6 MHz 13C NMR spectrum of the lithiated carbon atom
of the minor isomer of [(CH2=13CH)6Li]n in d8-THF at (a) -90°c and (b)
-60°C. (Reproduced with permission from Brauer and Griesinger (1993) /.
Am. Chem. Soc., 115, 10871, copyright (1993) American Chemical Society.)

3.11. Figure 3.42 shows the 119Sn NMR spectrum of the mixture of iso-
topomers, [SnHnD3_J~, n = 0 to 3, although there is only a trace
of [SnD3]~. Identify all the signals present. Account for the multi-
plicity of each signal. Analyse the coupling patterns and indicate
which splittings are due to 1/(119Sn1H) and 1/(119Sn2H). Derive
the secondary isotope shift which occurs on the replacement of
1
H by 2H.

\ \ i i i i i i i
0 -5 -10
119
Sn/p.p.m.

Figure 3.42 The 134 MHz 119Sn NMR spectrum of the mixture of isotopomers,
[SnHnD3_J~, n = 0 to 3, in liquid ammonia at 20°C. (Reproduced with permis-
sion from Wasylishen and Burford (1987) /. Chem. Soc., Chem. Commun.,
1414.)

www.pdfgrip.com
90 Internuclear spin-spin coupling

3.12. Figure 3.43 shows the 100.62 MHz 13C NMR spectrum of the
mixture of isotopomers, CHnD3_nI, n = 0 to 3. Identify all the
signals present. Analyse the coupling patterns and derive
1 1
J( *C1H) and 1J(13C2H). Derive the secondary isotope shift which
occurs on the replacement of !H by 2H.

2 1 0 - 1 - 2
p.p.m.

Figure 3.43 The 100.62 MHz 13C NMR spectrum of the mixture of
isotopomers, CHnD3_nI, n = 0 to 3. (Reproduced with permission from
Sergeyev et al (1994) Chem.Phys.Lett., 221,385, copyright (1994), with permis-
sion from Elsevier Science.)

3.13. Figure 3.44 shows the hydride signals from [Ru(CO)H2(PPh3)3].

The multiplet structure arises from coupling between the hydrides
and 31P nuclei. Analyse the multiplets to derive the coupling
constants. Identify which coupling constant is for 2/(1H1H) and
which coupling constants are for 2/(31P1H). Given that 2/(31P1H)
is large when the stereochemistry is trans and smaller when it is
cis, propose a structure for the compound.

-6.4 -6.5 -6.6 -6.7 -6.8

»1H>

-8.3 -8.4 -8.5 -8.6 -8.7

5(1H)

Figure 3.44 Expansions of the 400 MHz *H NMR signals from the hydride
atoms of [Ru(CO)H2(PPh3)3] in CD3C6D5.

www.pdfgrip.com
 Questions 91

3.14. Figure 3.45 shows the 1H NMR signals for the CH2 protons of
frans-[PdCl2{P(CH2Ph)2Ph}2]. Account for the appearance of the
spectrum. Predict what signal would be observed for the CH2
protons of fr0ws-[PdCl2{P(CH2Ph)Ph2}2].

3.8 3.7 3.6 3.5 3.4

6(1H)

Figure 3.45 The 220 MHz 1H NMR spectrum of the CH2 protons of trans-
[PdCl2{P(CH2Ph)2Ph}2]. (Reproduced with permission from Nelson and
Redfield (1973) Inorg. Nucl Chem. Letts., 9, 807, copyright (1973) Elsevier
Science.)

3.15. Figure 3.46 shows the 6Li and 15N NMR spectra of the product
obtained when Me215NCH2CH215NMe2 is added to a solution of
6
LiBun in d8-toluene. Suggest a structure for the product. Account
for the multiplicity of each signal.

(a)

2.5 2.0 15.5 15.0 14.5 14.0

61
D
Li/ppm 15
N/ppm

Figure 3.46 44.15 MHz 6Li and 30.408 MHz 15N NMR spectra of the product
obtained when Me215NCH2CH215NMe2 is added to a solution of 6LiBun in d8-
toluene. (Reproduced with permission from Nicholls et al. (1997) /. Am. Chem.
Soc., 119, 5479, copyright (1997) American Chemical Society.)

www.pdfgrip.com
92 Internuclear spin-spin coupling

3.16. Figure 3.47 shows the 235.36 MHz 19F NMR spectrum of cis-
[Ag(CF3)2Cl(CN)]- in THF. Account for the multiplicity of each
signal.

-16.5 -16.9 -27.6 -28.0

19
F/ppm
Figure 3.47 The 235.36 MHz 19F NMR spectrum of cw-[Ag(CF3)2Cl(CN)]- in
THF. Note that silver has two NMR active nuclei, 107Ag, 51.82% abundant,
y = -1.0828 x 107 rad T'1 s'1, and 109Ag, 48.18% abundant, y = -1.2449 x 107
rad T-1 s"1. Both nuclei have / = 1/2. (Reproduced with permission from Eujen
et al (1997) Inorg. Chem., 36, 1464, copyright (1997) American Chemical
Society.)

3.17. Assign the following proton signals from menthol, Fig. 3.2, using
the multiplicities and coupling constants.
8 2.18, septet (7 Hz) of doublets (3 Hz).
8 1.12, doublet (12 Hz) of doublets (10 Hz) of triplets (3 Hz).

SPECTRAL INTERPRETATION

The proton spectra of simple molecules are often sufficient to provide

a full structural description of the molecule. The task is made even
easier if an infrared (IR) spectrum is also to hand, since the two often
give complementary information. Since we are, however, concentrating
on NMR spectra here, we will ignore the IR and attempt to obtain
the maximum from the NMR spectra.
An NMR spectrum contains several pieces of usable information.
First of all is the chemical shift. Thus the position of a resonance indi-
cates the type of group in which the protons reside, sometimes with
remarkable clarity, though often there will be ambiguities. The chem-
ical shift ranges within which several types of proton are found are
given in the accompanying chemical shift chart (Fig. 3.48). It is only
necessary to add that hydrogen-bonded protons are found to high

www.pdfgrip.com
 Questions 93

12 9 6 , 3 0 - 3
1
H/ppm

Figure 3.48 Chart of approximate :H chemical shift ranges of different types

of protons in organic compounds. X = halogen, R = organic substituent, Ar
= aryl. (From Akitt (1987) in Multinuclear NMR, (ed. Mason), Plenum, New
York, with permission.)

frequency, often below 10 ppm, and that metal alkyls are found to low
frequency around and above TMS. Hydrogen bonded to metals occurs
in a wide range also at low frequency of TMS and varies from
about -3 ppm, e.g. [HRe(PPh3)3(CO)2], to -50 ppm, e.g. [HIrCl2(phos-
phine)2], though there are one or two exceptions with shifts to high
frequency of TMS. The second piece of information is the integral
trace of the spectrum, which gives the area under each resonance and

www.pdfgrip.com
94 Internuclear spin-spin coupling

so the relative numbers of protons contributing to each resonance.

This, coupled with the empirical formula, will enable the hydrogen in
the molecule to be split up into chemically different subgroups.
Spin-spin coupling patterns also give this type of quantitative infor-
mation, with the difference that an ethyl quartet-triplet pattern is
diagnostic but does not tell us how many ethyl groups are present.
The existence of coupling also tells us that the coupled groups are
proximate.
The interpretation of spectra is thus a deductive process in which
one attempts to account for all the spectral features in terms of a single
molecular structure. We will work through some examples and will
start with a few simple spectra where the only information is the chem-
ical shift and the formula:
1. The compound has a singlet at 7.27 ppm and formula C6H6.
Evidently we have a sample of benzene.
2. The compound produces a singlet at 2.09 ppm and has a formula
C3H6O. The chemical shift is typical of methyl or methylene in
unsaturated molecules or of CH3CO. Again, it is a short step to
acetone, (CH3)2CO.
3. The compound gives two singlets of equal intensities at 2.01 and
3.67 ppm and has a formula C3H6O2. Clearly, the two singlets arise
from three protons each, which form two equivalent sets and which
are not coupled. There are thus two CH3 and by difference we have
CO2. The chemical shifts are typical of CH3O and CH3CO. Thus
we have the ester CH3COOCH3, with four bonds between the
hydrogen atoms in the two groups and so no detectable coupling.

4.5 4.4 1.8 1.7

1 2 1 11 0 9 8 7 6 5 4 3 2 10
1
H/ppm
Figure 3.49 The 400 MHz 1H NMR spectrum of C3H5O2Br in CDC13.

www.pdfgrip.com
 Questions 95

Now we will consider an example where there is spin-spin coupling.

The spectrum shown in Fig. 3.49 is that of the substance C3H5O2Br.
It consists of nine resonances, of which we can discount two as arising
from the reference TMS (0 ppm) and the remnant CHC13 in the deuter-
ated CDC13 solvent (7.25 ppm). Two of the resonances are split into
regular multiplets with identical splitting of 7.3 Hz. This informs us
that there is spin-spin coupling between two of the groups of protons
and the value of the coupling constant is typical of a vicinal three-
bond coupling pathway, though not exclusively so. The doublet-
quartet pattern must arise from a H-H3 interaction. The integrals are
in the ratio 3(doublet): 1 (quartet): 1 (singlet), and since there are five
protons in the molecule the hydrogen atoms must also be split up in
this ratio. The chemical shift of the doublet is typical of CH3 alkyl;
that of the quartet is rather high frequency for CHX but we need also
to keep in mind the influence of the oxygen-containing part of the
molecule, which is also to high frequency. Finally, the singlet to high
frequency is typical for acidic protons or aldehydes. We can now
attempt to work out the structure, and it helps to do this if we settle
upon one structural unit at a time and subtract this from the formula.
Thus we have C3H5O2Br. The doublet is evidently a CH3 unit. This
leaves C2H2O2Br. It is coupled to an alkyl proton, i.e. a CH, to give
CH3CH, leaving CHO2Br. Two valencies on the CH carbon need to
be filled. Bromide must take one, leaving CHO2, and this carboxyl

8 7 6 5 4 3 2 1 0
1
H/ppm
Exercise 1 The 60 MHz !H NMR spectrum of C3H4SO2 in CDC13. Deduce a
structure which is consistent with this spectrum. (Reproduced with permis-
sion of Varian, Inc.)
Possible answers: CH3SO2OCH, HOCH2SOC=CH, H H
C=C
C-SO 2
H2

www.pdfgrip.com
 96 Internuclear spin-spin coupling

group must then take the other. The molecule is as shown in the
margin.
Now apply the same approach to Exercises 1 to 3. Bear in mind the
following: (a) The fluorine resonances of the compounds containing
fluorine are not visible but the proton resonances may be coupled to
the 19F and so split into multiplets. The same comments apply also
to the phosphorus compound, (b) Alcoholic protons are variable in
position because of differing hydrogen bonding effects. The numbers
on the spectra give the relative numbers of protons contributing to
each resonance and have been obtained from integral traces. Exercises
1 and 2 show expanded regions of the spectra so as to allow fine struc-
ture to be distinguished. The answers are given below, though several
structures are given, only one being correct. All the features of a spec-
trum should be explicable from the structure. It is instructive to also
consider what differences in the spectrum would be obtained from the
incorrect structures given.
A spectrum is shown in Fig. 3.50 with numbers corresponding to
the integrals of each group of resonances. The empirical formula is
C6H14O2. Evidently there are five types of hydrogen and all show
spin-spin coupling. The doublet and triplet at 1.37 and 1.25 ppm are
in the chemical shift region typical of CH3R and the integrals indicate
that they are due to a CH3 and two identical CH3 groups. The former
are coupled to a single proton which should have a quartet resonance.
This is seen at 4.7 ppm, the line spacings being equal, and the shift

1 4.0 3.5
LA.
8 7 6 5 4 3 2 1 0
1
H/ppm
Exercise 2 The 400 MHz J H NMR spectrum of C2H3F3O in CDC13. Draw a structure which is consistent
with this NMR spectrum.
Possible answers: CF2HCHFOH, CF3CH2OH, CH2FCF2OH.

www.pdfgrip.com
 Questions 97

C3F4H40

^*v**~~jJ*»~Jt^^ L/
8 7 6 5 4 3 2 1 0
1
H/ppm
Exercise 3 The 60 MHz *H NMR spectrum of C3F4H4O in CDC13. Deduce a
structure which is consistent with this spectrum. (Reproduced with permis-
sion of Varian, Inc.)
Possible answers: CF2HCH2CF2OH, CF3CHFCH2OH, CF2HCF2CH2OH,
CH3CF2CF2OH.

being typical of a RCH2O type of situation. We thus have a fragment

CH3CH with oxygen close by. The other CH3 groups are coupled to
two hydrogen atoms and are exactly equivalent. The conclusion is that
we have two equivalent ethyl groups but we do not see any other
quartets in the spectrum though the four hydrogen atoms to which
they must be coupled give a slightly second-order group of lines around
3.6 ppm. We have to conclude provisionally that we have two equiv-

TMS

1.4 1.3 1.2

5.0 4.0 3.0 2.0 1.0 0.0

1
H/DDm
Figure 3.50 The 400 MHz ]H NMR spectrum of MeCH(OCH2CH3)2 in CDC13. The intensities of the
signals are given above each one.

www.pdfgrip.com
 98 Internuclear spin-spin coupling

IMS

II Tb ill
CeHeIHA,
I III
Jil
8 7 6
7.5

5 4
2
A
7.0

3
6.6

2 1 0
1
H/ppm
Figure 3.51 The 400 MHz !H NMR spectrum of 2-nitroaniline in CDC13. The intensities of the signals are
given above each one.

alent ethyl groups CH3CH2-. This accounts for all the atoms in the
molecule except the two oxygen atoms. Because the methyl groups
are identical, the oxygen atoms have to be distributed symmetrically
and this leads inevitably to the structure CH3CH(OCH2CH3)2. The
CH2 pairs of hydrogen atoms in this molecule are diastereotopic and
so non-equivalent which accounts for their complex spectrum.
In Fig. 3.51, we see four groups of resonances in the aromatic/alkenic
region of shift which exhibit a larger and a smaller coupling. There
are also two uncoupled protons in the region typical of RNH and, in
view of the empirical formula given on the figure, this suggests an NH2
group which leaves us with a C6H4 aromatic nucleus and an NO2 group.
We thus have a nitro-aniline and need only to deduce which isomer
we have. This can be done using the coupling patterns. Were it the
para isomer, the two pairs of protons would be equivalent and an
[AB]2 pattern with two groups of resonances would result. Were it the
meta isomer then there would be an isolated hydrogen (H^ which
would show only the small coupling. This is plainly not the case.
However, in the ortho isomer all protons have near neighbours and

7.6 7.4 7.2 7.0 6.8 6.6 6.4 6.2 6.0 5.8 5.6
Figure 3.52 A 400 MHz !H NMR spectrum and its integral of 2,6-dichlorophenol in CDC13. The intensi-
ties of the signals are given above each one.

www.pdfgrip.com
 Questions 99

so the larger coupling interaction. Hj and H4 are close to one neigh-

bour and are doublets while the other two are near to two neighbours
and appear to be doublets of doublets though the spectra are compli-
cated by the small longer range couplings. A very similar example will
be found in Fig. 8.7 for the R groups of a phosphine PR3.
In Fig. 3.52 we see the 1H NMR spectrum of C6H4OC12. There are
three resonances with a triplet, intensity 1 and a doublet, intensity 2
in the aromatic region and a singlet at 5.9 ppm still on the edge of the
aromatic region. The coupled protons have the typical pattern for two
chemically equivalent protons coupled equally to one proton, which is
in accord with the intensities. The fourth proton is isolated from the
others. With 6 carbon atoms we then have a benzene nucleus and 7
substituent atoms and we need to decide how they are arranged. The
coupling constant is quite large at 8.5 Hz so the coupled protons are
likely to be on adjacent carbon atoms. For two such protons to be
equivalent, they have to be the outer ones and the molecule has to
be symmetrical around an axis passing through the central one. The
two chlorine substituents have then to be placed next to the two outer
hydrogens which leaves us an OH group as sixth substituent on the
ring to give the formula 2,6-dichlorophenol.

www.pdfgrip.com
www.pdfgrip.com
 Nuclear magnetic
relaxation 4
4.1 RELAXATION PROCESSES IN ASSEMBLIES OF
NUCLEAR SPINS

If we perturb a physical system from its equilibrium condition and

then remove the perturbing influence, the system will return to its orig-
inal equilibrium condition. It does not return instantaneously, however,
but takes a finite time to readjust to the changed conditions. The
system is said to relax. Relaxation to equilibrium usually occurs expo-
nentially, following a law of the form
(rt-"e)r = ("-"e)oexP(- t/T
)
where (n —ne)f, is the displacement from the equilibrium value nQ at
time t and (n —ne)0 that at time zero. The relaxation can be charac-
terized by a characteristic time T. If T is small, relaxation is fast;
whereas if T is long, relaxation is slow. An alternative description is
to use rates of relaxation, which are given the symbol R:
1
K-
R
-T
R has the advantages that its value increases as the relaxation becomes
faster and that, if a system is subject to several parallel relaxa-
tion processes, then the overall rate is the simple sum of the rates of
all the processes, i.e.
R = R» + /?h + R c
The relaxation behaviour of assemblies of nuclear spins shows up
directly in their NMR spectra and is related to the molecular dynamics
of the system. For these reasons, it is of considerable importance to
NMR spectroscopists, on the one hand allowing them to optimize
experimental conditions, even to eliminate some undesirable spectral
feature, or on the other hand allowing close study of the physical and
chemical properties of the motion of a system.
We have already seen that in an assembly of / = 1/2 nuclei immersed
in a strong magnetic field the spins are polarized into two populations
with opposite senses and with a small excess number in the lower energy
state. The nuclei precess around the magnetic field direction with a
net magnetization Mz and no detectable transverse magnetization in

www.pdfgrip.com
102 Nuclear magnetic relaxation

the xy plane. It turns out that this system can be perturbed in two ways,
and that we have to expect that there may be two relaxation processes
with different relaxation times, which are called Tl and T2, or rates of
relaxation ^ and R2. We have already seen in Chapter 1 (Fig. 1.6) that
a Bl radiofrequency pulse can swing the total nuclear magnetization
away from its equilibrium position in the z direction, and this is essen-
tially a perturbation of the system. If we apply a rather long pulse, we
can swing the magnetization back into the z direction but pointing in
the opposite direction. Such a pulse is called a 180° pulse, the reason
for this name being evident in Fig. 4.1. The magnetization has been
inverted and, immediately following the end of the pulse, relaxation
processes start to return the magnetization to its normal state. Thus the
magnetization decays from -Mz via zero to Mz via first-order rate
processes. The characteristic time for this process is 7\. The process is
also called longitudinal relaxation since it takes place in the direction
of U0, and it is also called spin-lattice relaxation. In all cases it must be
emphasized that the inverted magnetization has higher energy than the
normal magnetization and that the return to equilibrium involves an
exchange of magnetic energy with the surroundings.
If instead we use a 90° pulse to perturb the spin system, we move
the magnetization into the xy plane as in Fig. 4.2. Now the magneti-
zation in the z (BQ) direction is zero, and this returns to its normal

MzA
// 180° pulse //

~-^ ^-Mz

relaxation relaxation

BO BO

i \^ relaxation /^
-^ ' ^^
Figure 4.1 The Tl relaxation process. If the magnetization is inverted, then
it has to return to its equilibrium state and does so by decaying to zero and
then increasing again in the normal B0 direction. This process involves an
exchange of energy between the spins and their environment.

www.pdfgrip.com
 Relaxation processes in assemblies of nuclear spins 103

Figure 4.2 The T2 relaxation process. A 90° pulse swings the magnetization
into the xy plane around the Bl vector. This is shown as stationary in the
figure as if the observer were rotating in the same direction at the Larmor
frequency, thus giving a static picture. There is a spread of nuclear frequen-
cies, which causes the spins to fan out and reduce the resultant Mxy. Mz
increases at the same time due to the 7\ process.

value by the mechanism just discussed. However, we have also created

transverse magnetization in the xy plane, which rotates at the nuclear
Larmor frequency. This has to decay to zero at equilibrium and does
so because the frequency of each spin differs slightly from that of its
companions, each varying randomly around the mean precession
frequency. This means that, on average, some spins are slower and
some are faster than the mean, so that the xy magnetization starts to
fan out, to lose coherence and the resultant to become less in magni-
tude. Eventually the spins take all directions in the xy plane and Mxy
is zero. The characteristic time of this process is T2. This is also called
transverse relaxation, and in solid materials it is known as spin-spin
relaxation. Since the process is related to the spread of frequencies of
a nuclear resonance, it is evident in the spectral linewidth. 7\ and
T2 may be equal or they may differ by orders of magnitude. The T2
process involves no energy change. Evidently Mz increases as Mxy
decreases, so that 7\ and T2 can be equal, but T2 cannot be longer
than 7\. On the other hand, T2 can be less than 7\, and in this case
M decays more rapidly than Mz is re-established and the signal
(derived from Mxy) disappears well before equilibrium of the spins
is attained.

www.pdfgrip.com
104 Nuclear magnetic relaxation

T2 is most commonly encountered in the linewidth of a signal. The

linewidth at half height, W1/2, is given by

Wm - -^ (4.1)
This is true for a singlet in a perfectly homogeneous magnetic field,
but in practice there are often extra contributions to the linewidth due
to inhomogeneity of the magnetic field and unresolved coupling.
The nuclei of atoms are extremely well isolated from their surround-
ings and, because the energy of NMR transitions is small, the chance
of a spin transition occurring spontaneously is negligibly small. The
fact that relaxation times can be quite short indicates that transitions
are stimulated, and we must thus consider the various ways that this
can happen.

4.2 DIPOLE-DIPOLE RELAXATION

We have already noted in section 2.1 that the magnetic field at a given
nucleus due to the magnetic moment of a near-neighbour nucleus is
very high but is averaged to zero by the random rotational diffusion
of the molecule in which the nuclei reside. The magnitude of this field
is such that the instantaneous chemical shift displacement of one 1H
nucleus due to the other in a methylene group can be as high as
150 000 Hz. As the group rotates, this field varies by such an amount
on each side of zero. Thus the nuclei have instantaneously different
precession frequencies since all possible orientations of the molecules
will exist at any one instant. Randomization of the frequencies means
not that all will have the same frequency in the long term but rather
that once out of step a nucleus is just as probable to move further
away from its companion's frequency as to reconverge to it. This
dipole-dipole fluctuating field then is the cause of the loss of coher-
ence between spins and so the source of the T2 relaxation process.
The chaotic random motion of a solute in a solvent is called
Brownian motion. This has a timescale that depends upon a number
of factors such as mass of solute, solution viscosity and temperature.
Because the motion is random, this timescale is characterized by a
somewhat loosely defined term, the rotational correlation time TC. This
is the time taken on average for a solute molecule to rotate by one
radian or, more precisely, the time interval after which the molecular
motion contains no vestige of its earlier angular momentum, i.e. has
lost all memory of its previous behaviour. Not only the overall rota-
tion of the molecule contributes to TC, but also local motion, such as
the rotation of a methyl group. The time TC is typically 10"11 s for small
molecules in liquids of low viscosity, which converts on the frequency
scale to 1011 rad s'1 or 15 920 MHz. This is around the maximum rate
of motion in the system, and all slower rates of motion can exist. The
frequency spectrum of such random motion and associated magnetic

www.pdfgrip.com
 Dipole-dipole relaxation 105

fields is simple and is essentially white noise at all frequencies less

than I/TC. Since the maximum NMR frequency that is likely to be
encountered is at present 800 MHz, it is clear that there is a compo-
nent at all possible NMR frequencies. The relaxation field thus
provides a Bl component that varies in intensity and direction and
causes random precession of the nuclei also: hence the dephasing of
spins and also the possibility of energy transfer needed for the Tl
mechanism to operate.
The field intensity at any frequency, K(v), is given by
<4 2)
«" • rrl^v '
This function is plotted in Fig. 4.3, where it will be seen that, when
TC = 10~n s, then over the NMR frequency band the intensity of the
relaxation field is constant. However, if TC is rather longer, this may not
be true. The Debye theory of electric dispersion shows that, for a spher-
ical molecule rotating in a liquid, the correlation time is given by

K (v) (note vertical scale differs for each plot)

d

i i i i i i
1 10 100 1000 10000 100000 1000000
MHz

Figure 4.3 Intensity of fluctuations in magnetic fields in a liquid sample due to Brownian motion, as a
function of frequency, (a) TC - lO'11 s-1. (b) TC - 10;10 s-1. (c) TC = 10'9 s'1. (d) TC - 10'8 s-1. Note that a
different vertical scale is used for each plot as the intensity of fluctuations9 in1 the flat region of the plot
is proportional to TC. TC ~ 10~n s-1 is found for very small molecules, TC ~ 10~ s' is commonly encountered
with molecules of molecular mass, 1000 to 3000 D, while TC > 10~9 s'1 is normally found for molecules of
molecular mass > 5000 D.

www.pdfgrip.com
106 Nuclear magnetic relaxation

4-TTfl3 Tl
Tc =
~3F"r
where T\ is the viscosity of the liquid, T is the temperature and a is
the radius of the molecule. Thus if we vary the viscosity of a sample,
or its temperature, or the mass of the solute molecule, we will change
TC. Examination of equation (4.2) shows that, when 4TT2v2Tc2« 1, the
intensity of the relaxation field is proportional to TC. For the higher
NMR frequencies and the longer correlation times, we start to leave
the flat portion of the noise spectrum and the relaxation field starts
to decrease. Provided we limit our range to the portion of these curves
where 4ir2v2Tc2 « 1, then the relaxation field and so the relaxation rate
increases with TC. Long relaxation times thus occur for low viscosity,
high temperature and small molecular mass.
It will be seen from equation (4.1) that the relaxation field intensity
has its flat frequency response when the quantity 4ir2v2Tc2 is very much
less than unity, i.e.
4^2V2<< _1_

This is known as the region of extreme narrowing, where the corre-

lation time is much shorter than one Larmor period of the nuclei.
We can now make some qualitative predictions about how the relax-
ation times will vary with TC. In the extreme narrowing limit, 7\ and
T2 are determined by the same relaxation field and so are equal.
Increasing TC reduces the relaxation times until we reach the point at
which the frequency spectrum of the field is no longer flat. Its inten-
sity begins to decrease at the higher NMR frequencies, and so the
chance for energy exchange is decreased and 7^ increases with further
increase in TC. There is thus a minimum in 7\. The behaviour of T2 is
quite different since the correlation time is now similar in length to a
Larmor period, and superimposed on the random field we see for short
periods the rotating vector of the neighbouring nuclear magnetic
moment. This is at the nuclear frequency and provides a second means
for loss of coherence in the xy plane. T2 continues to decrease as TC
increases. In the limit of infinite TC we have the solid state. Here there
is no dipole-dipole relaxation field and 7\ is very long and is deter-
mined by the presence of ferromagnetic impurities in the lattice. Hence
the name spin-lattice relaxation. T2 is determined by the now very
strong interaction between spins via the rotating field generated by
the Larmor precession of the spins. Rapid exchange of spin states is
stimulated, the lifetime of the individual spin state is short and the
uncertainty principle dictates a short T2. Hence also the name spin-spin
relaxation.
A full analysis of the spectral density of the relaxation field and the
way this influences the spins gives the following equations for 7\ and
T2, expressed as the rates. We will also introduce the notation RIDD
and R2DD or TIDD and T2DD to indicate that the mechanism discussed
is dipole-dipole interaction.

www.pdfgrip.com
 Dipole-dipole relaxation 107

For / = 1/2 nuclei of the same isotope situated in the same mol-
ecule, the intramolecular dipole-dipole relaxation rates for the yth type
of nuclei, being relaxed by n different types of nuclei of the same
isotope are
1 4Tc
-J—-I? -2WM ^ i ^ V !
R +
T1DD - ™ - Ml + co V IT^oV ) J£ jf

£-^-^rfVr£v)^Ji
where a = 3|m,02/j2/320'Tr2, |JLO is the permeability of a vacuum, r is the
distance between the nuclei, y is their magnetogyric ratio, to is the
NMR frequency in rad s'1, and h is (Planck's constant)/27T. The way
these rates of relaxation vary with TC is shown in Fig. 4.4 for two
protons at two different spectrometer frequencies, to. The main feature
to note is the Tl minimum, which marks the limit of the extreme
narrowing region and the way this moves with spectrometer frequency.
The higher this frequency, the shorter becomes the maximum
permitted TC, the result being that, for large complex molecules where
the highest frequencies may be needed to give the necessary degree
of resolution, the increased correlation times of such molecules may
result in reduced T2 and so increased linewidths. The expression for
riDD and r2DD is much simpler in the extreme narrowing region

TIDD.
10-i

1-
extreme
narrowing
0.1 -

0.01 -

0.001 -

0.0001 -

0.00001
10-12 10-11 10-10 10-9 10-8 10-7 10-6
TC (s)

Figure 4.4 Variation of T1DD and T2DD with TC for two different spectrometer
frequencies. The figures given apply to two protons separated by 160 pm.

www.pdfgrip.com
108 Nuclear magnetic relaxation

1 1 ft 1
T =T =tflDD=tf2DD= 10«74Tc 2 n (4.3)
^ 1DD ^ 2DD /= 1 (/ =£ /)r//

Two further points should be emphasized. In the first place, the rate
of dipole-dipole relaxation depends upon the fourth power of the
magnetogyric ratio, and nuclei with large magnetic moments will be
most strongly subject to such relaxation, for example, 1H or 19F. The
mechanism will be of lesser importance for nuclei with smaller magne-
togyric ratios. Secondly, the efficiency of relaxation depends strongly
upon the inverse distance between the spins. The effect of more distant
spins will be almost negligible.
For an / = 1/2 nucleus in the same molecule as n different / = 1/2
nuclei with magnetogyric ratios yl and -ys and frequencies o)j and oos
rad s"1, respectively, the intramolecular dipole-dipole relaxation rates
of one (species I) due to interaction with other (species S) are
2 2Tc 6Tc
T-
*1DD
= RmD = ^i
*
rfi. f ,2 2 +
[1 + (&>!-0) )^T 2
S C l+Wl2T(2

1?T 1 " 1
+ ^5 ^ J_
l + (coI + cos)VL=1^,)'"/

7
2DD ^
^ = *2DD ^ +3f^sf
\ 1 + <0S Tc2y.=^^..6
c +I ^) $ ^
These equations, though more complex, give plots of form very
similar to those depicted in Fig. 4.4. In the extreme narrowing limit

*1DD = *2DD = ^I2^c 2 —6 (4-4)

3 i=\(i*j}riJ

The most common example of relaxation of one spin by a different

species occurs in 13C spectroscopy, where the carbon nucleus is relaxed
by the moments of directly bonded hydrogen atoms. In addition,
because the relaxation field is stronger if more hydrogen atoms are
attached to the 13C, the rates of relaxation are proportional to the
number of directly bonded hydrogens. Thus for relaxation of 13C in a
CH3 group, the above expression has to be multiplied by three.
Provided that relaxation occurs solely by the dipole-dipole mecha-
nism, and the C-H bond length is known, then the correlation time
may be calculated from the measured relaxation time. Relaxation of
tertiary or carbonyl 13C is much slower than for CHn groups since the
C-H distance is much greater, and in this case other relaxation mech-
anisms may be important. The relaxation of protons on 13C is domi-
nated by interaction with other protons in the molecule.
An unpaired electron also generates a magnetic field. It influences
relaxation in exactly the same way as a nucleus with / = 1/2 relaxes a
second nucleus, but *Ye = 658yH. The result is that unpaired electrons
can be very effective in relaxing other nuclei. The effectiveness

www.pdfgrip.com
 Quadrupolar relaxation 109

depends on Tf, the spin-lattice relaxation time of the electron. When

Tf is long, the unpaired electron is very effective at causing relax-
ation, and equation (4.3) applies, with ye replacing ys. When Tf is
short, the unpaired electron is ineffective at causing relaxation. The
result is that metal ions such as Crm, where Tf is long, cause substan-
tial broadening of 1H NMR signals (Fig. 2.10(a)), while Prm and Eum,
where 7\e is short, cause little broadening (Fig. 2.12). When 2H is
observed, the line is sharper as yD = 0.15357H (Fig. 2.10(b)).
In the case of ligands attached to metals, the unpaired electron is
frequently delocalized onto the ligand. It then becomes meaningless
to use equation (4.4) as the distance r is ill-defined. 7\ and T2 are then
given by
1 1 ^2ys2aN2S(S + !)TC
7\ ~ :T2 ~ 24TT2 ^ }
where aN is the nuclear electron hyperfine interaction constant.

4.3 QUADRUPOLAR RELAXATION

Relaxation times for protons in organic molecules, as will be seen

from Fig. 4.4, are of the order of seconds. For / = 1/2 nuclei with
small magnetogyric ratios, the times are in general much longer, par-
ticularly in the absence of neighbouring protons. If the nucleus in ques-
tion, however, has / > 1/2, it has a quadrupole moment, and this
introduces a second and very efficient relaxation mechanism, which
results in a generally much reduced relaxation time, a factor of as high
as 108 over the expected dipole-dipole relaxation time being possible.
The quadrupole moment of a nucleus arises because the distribu-
tion of charge is not spherical, as is the case for / = 1/2 nuclei, but is
ellipsoidal, i.e. the charge distribution within the nucleus is either
slightly flattened (oblate, like the Earth at its poles) or slightly elon-
gated (prolate, like a rugby ball or an American football). Cross-
sections of the charge of two such nuclei are shown in Fig. 4.5 with
the departure from spheroidal form much exaggerated. If the electric
fields due to external charges vary across the nucleus, then the torque
on each dipole component of the quadrupole is different, and a net
torque is exerted on the nucleus by the electric field as well as by the
magnetic fields present. Quadrupolar electric field gradients, that is
field gradients which are shaped like d-orbitals, exist at atomic nuclei
due to asymmetries in the spatial arrangement of the bonding elec-
trons. The Brownian motion of the molecule causes the direction of
the resulting electric quadrupole torque to vary randomly around the
nucleus in exactly the same way as does the torque due to the magnetic
relaxation field. Rotating electric torque components thus exist at the
nuclear resonant frequency, which can also cause interchange of energy
between the nucleus and the rest of the system (Tl mechanism) and
randomization of nuclear phase (T2 mechanism).

www.pdfgrip.com
110 Nuclear magnetic relaxation

oblate prolate

Figure 4.5 A representation of the charge distributions in quadrupolar nuclei.

Note that there is a higher concentration of positive charge as the curvature
of the surface increases. The magnetic axis is also marked.

The quadrupolar electric field gradient at a point in a molecule arises

from the effect of the surrounding electric charges. The quadrupolar
electric field can be described by three components in a Cartesian
system of axes, but the field magnitude will change as one moves along
an axis and in general all three of its components will change. Further,
the changes may be different depending on which axis one chooses.
The quadrupolar electric field gradient (EFG) thus requires nine
numbers in order to describe it fully, and is a tensor quantity.
Fortunately, by choosing an appropriate system of axes relative to the
molecular geometry, this number can be reduced to three, Vxx, Vyy and
Vzz, the diagonal components of the tensor, which have the further
property that their sum is zero. This permits a further simplification
by introducing the asymmetry factor, T), such that
Vyy - Vx
7] =

Vzz is chosen to be the largest component of the EFG tensor, and this
can then be described by two quantities, Vzz and i\. If the system is
axially symmetric so that Vyy = Vxx then TI = 0. The Vxx, etc. are calcu-
lated as the sum of contributions from all charges i
VKX = Sfcr-W - r,2) (4.6)
where ri is the distance of charge qi from the nucleus and xt is the
coordinate of the charge in the axis system chosen. Similar expres-
sions give V and Vzz.
The equation describing the quadrupole relaxation times (TIQ, T2Q)
of a quadrupolar nucleus with spin / situated in a molecule with an
isotropic correlation time TC and in the extreme narrowing region is

www.pdfgrip.com
 Quadrupolar relaxation 111

tt^-zmftM+i^
where Q is the quadrupole moment of the nucleus of spin, /, e is the
(4.7)

electronic charge and h is Planck's constant. Outside the region of

extreme narrowing, the behaviour is reminiscent of that of dipole-
dipole relaxation but is complicated by the fact that several nuclear
transitions are possible. Thus for a nucleus with spin 3/2 there are
three transitions, 3/2 <-> 1/2, 1/2 <-> -1/2 and -1/2 <-> -3/2, and the relax-
ation rate for nuclei undergoing the first and last transitions differs
from that of those undergoing the 1/2 <-> -1/2 transition. This produces
non-exponential relaxation.
Quadrupolar relaxation depends strongly upon both nuclear prop-
erties (Q, /) and molecular properties (Vzz, TI, TC). Its effectiveness
increases rapidly as Q is increased, though this is to some extent offset
by the fact that large Q tends to be associated with large / and the
function of / in the expression for T1Q decreases rapidly as 7 increases
(Table 4.1). As a result, almost all quadrupolar nuclei are capable of
being detected, only a few having such short relaxation times that they
are unobservable. Of the molecular properties, the correlation time
has a particularly strong influence in that changes in TC brought about
by changes in temperature alter the relaxation times considerably.
Increasing temperature reduces TC and so increases the relaxation time
and reduces the resonance linewidth. Since the resonances of
quadrupolar nuclei are usually quite broad, these changes are evident.
The same effect occurs with dipole-dipole relaxation of / = 1/2 nuclei,
but here the relaxation times are long, the linewidths are small, and
the linewidth changes are not so immediately obvious. However, it is
frequently possible to differentiate the narrow signals from a small
impurity molecule from the broader signals from a higher molecular
weight compound by the different linewidths in their [H NMR spectra.
For a given nuclear species, the quadrupolar relaxation time is deter-
mined mainly by the EFG and this can vary from almost zero to very
large values, so that relaxation times can vary by several orders of
magnitude depending upon the situation of the nucleus. In principle,
the EFG can be calculated at any point in a molecule so that one
could obtain TC, or if TC were known (e.g. from the 13C relaxation time
of a CH moiety in the molecule) the EFG could be determined and
compared with that calculated so as to verify the accuracy of the calcu-
lation. Unfortunately, it proves in practice difficult to calculate the

Table 4.1 The dependence of F(I) = (21 + 3)/{/2(27 - 1)} on /

3 5 7 9
/ 1 2 2 3 2 2
4 _8 1 _2Q 2
F(I) 5 3 25 5 147 27

www.pdfgrip.com
112 Nuclear magnetic relaxation

EFG with sufficient accuracy. Equation (4.4) shows that the EFG is
proportional to the inverse cube of the distance of the charge from
the nucleus. Thus charge close to the nucleus has the predominating
effect, and it is here that the calculations are least accurate. In fact,
there is some confusion in the literature on this point. In the solid
state, the EFG arises from quite distant charges as well as those close
by, in the same way that a Madelung constant is calculated. In a liquid,
the movement of the molecules reduces the distant effects to zero and
the EFG arises quite locally around the nucleus. For instance, in the
anion [A1C13(NCS)]- both the quadrupolar nuclei 27A1 and 14N have
long relaxation times and show spin-spin coupling. There are two
points of low EFG in the molecule, and this can only arise if the EFG
arises in the region quite close to the nucleus. The question is, of
course, how close has the charge to be, to be effective?
The cases of greatest interest are those in which the relaxation times
are relatively long, since this gives the best resolution of resonances
and the possibility of seeing coupling effects. This means that we
should understand the requirements for obtaining a low EFG at a
nucleus. We can do this easily if we remember that for a traceless
EFG tensor the sum of the diagonal elements is zero

If the system is axially symmetric (i.e. T] = 0), Vyy = Vxx, and it follows
that if Vzz is zero then all three terms must be zero and the EFG
vanishes. Thus in an axially symmetric system we need only calculate
Vzz. We take for the model in Fig. 4.6 a tripod of bonds to a nucleus
N disposed regularly around the z axis with effective, equal charges q
called ql9 q2 and q3 at a distance r from N.

Figure 4.6 Calculation of the EFG at N due to three equal charges q.

www.pdfgrip.com
 Quadrupolar relaxation 113

The z coordinate of each charge is rcosO. According to equation

(4.6) then
3g(3r2cos2e - r2) 3g(3cos26 - 1)
r5 r3
In order that Vzz be zero, (3cos20 -1) must be zero, which requires
that
6 - 54.7356°
As we shall see, the quantity (3cos26 -1) appears in many expres-
sions describing NMR phenomena, and the value of 6 at which it
becomes zero has become known as 'the magic angle'. It may seem
astonishing that such a one-sided arrangement of charge should give
zero EFG. If, however, we remember that the electric field is zero
where the z axis intersects the plane defined by the three charges and
is also zero at -<*>, we will see that it has to have a maximum at some
point on the z axis, a point where its gradient is zero. We should also
note that Vzz is zero for any number greater than three of equal regu-
larly disposed charges that form the magic angle with the axis.
It has been customary to state as the necessary condition for low
EFG a cubic arrangement of equal charges (or identical ligands)
around a nucleus. This is correct, and indeed a nucleus at the centre
of a regular tetrahedron, octahedron or cube has near-zero EFG. It
is, however, too limiting a statement. The octahedron is in fact made
up of two back-to-back tripods as depicted in Fig. 4.6 with all six
charges equal and pairs of charges collinear with N. Equally, an octa-
hedral complex in which there are two tripods of differing charge will
still have zero EFG, i.e. an all-cis or fac structure MA3B3 where A
and B are ligands. Such structures, while non-cubic, still have long
relaxation times, and we can begin to see that long relaxation times
for quadrupolar nuclei are a source of structural information.
A specific example is found with the molybdenum tricarbonyl arene
complexes, see formula in margin.
The nucleus 95Mo has a significant quadrupole moment and its relax-
ation times vary typically from about 1 s in the regular tetrahedron
[MoO4]2~ up to about 0.15 ms in less symmetric compounds. In the
arene carbonyl complexes, the times are near 0.07 s and are relatively
long for a non-cubic symmetry. A to-scale sketch of the molecule is
given in Fig. 4.7. The crystal structure of the mesitylene complex has
been obtained and this shows that the three carbonyl ligands are
orthogonal, which means that they each make the magic angle with
the symmetry axis of the molecule. Following Fig. 4.6 then, the
carbonyl ligands give zero net contribution to the EFG at the molyb-
denum. This means that the remaining bonding electrons to the arene
ring must also produce a very small EFG at the metal. So the
metal-arene bonds must lie on a conical surface with a half-apex angle
equal to the magic angle also. This argument does not depend upon
our knowing how many such bonds there may be; they have simply

www.pdfgrip.com
114 Nuclear magnetic relaxation

Figure 4.7 A drawing taken from the crystal structure of [(TQ6-l,3,5-Me3C6H3)

Mo(CO)3] with the methyl groups and hydrogen atoms omitted. The carbonyl
ligands are approximately orthogonal and form half a regular octahedron. The
arene ring carbon atoms give an average angle of 36.4° to the axis.

to be disposed regularly on the conical surface. It is clear that such

bonding orbitals do not intersect the arene ring, that the conventional
picture of overlap of metal and arene orbitals is correct, and that the
bonds are bent. It is also clear that the electron density on the arene
cannot influence the EFG and that this is determined by the electron
density nearest to the metal.
In general, however, many molecules either have some distortion of
their symmetry or their symmetry is non-cubic around the nucleus to
be observed, and the relaxation times and linewidths observed are very
variable. For octahedral complexes, the point-charge model makes
some predictions which are useful in distinguishing between cis- and
trans- and between mer- and /ac-isomers, see Table 4.2.
Where terminal atoms are concerned, the lines are often very broad
and may not even be detectable. This missing intensity problem has
to be kept in mind when working with quadrupolar nuclei, and it is

Table
59
4.2 The predicted quadrupolar electric field gradients and observed
Co linewidths for some cobalt(III) amine complexes
Structure Relative magnitude of Complex Linewidth (Hi)
quadrupolar electric
field gradient
MA6 0 [Co(NH3)6]3+ 2 87
MA5B 2 [Co(NH3)5(N3)] + 350
d5'-MA4B2 2 ds-[Co(NH3)4(N3)2]++ 370
trans-MA4B2 4 frans-[Co(NH3)4(N3)2] 520
/flc-MA3B3 0 /ac-[Co(NH3)3(N3)3] 300
raer-MA3B3 3 mer-[Co(NH3)3(N3)3] 1240

www.pdfgrip.com
 Quadrupolar relaxation 115

often necessary to check the resonance intensity against that of a stan-

dard sample in order to ensure that something is not being missed.
Even if the resonance is visible, its intensity will be incorrect if its
width is more than one-sixth that of the Fourier transform spectral
width. On the other hand, if the resonance is broad, but is resolved
from other resonances, then structural information is present in the
spectrum. A few examples of quadrupolar nuclei in different envi-
ronments are given in Table 4.3 to illustrate the possibilities.
Linewidths are quoted rather than relaxation times, since it is the
linewidths that have been measured in most cases and these are
proportional to R2Q or 1/T2Q.
Solid-state NMR spectroscopy of quadrupolar nuclei or the zero-
field technique of nuclear quadrupole resonance can give values for
the interaction between the nuclear quadrupole and the EFG, and
this is called the quadrupole coupling constant, x> where in terms of
equation (4.4)
eQVzz e2qQ
X
-~7T~ = h
If one can be certain that the EFG in solid and liquid are equal,
and this is likely to be so where the EFG is large and determined

Table 4.3 Linewidths of the resonances of some compounds of quadrupolar

nuclei
Nucleus Molecule Linewidth (Hz) Comments
14N [Me4N]+ 0.1 Regular tetrahedron,
small /OWH) resolvable
14
14
N [PhNH3]+ 100 Non-regular tetrahedron
N MeCN 80 Terminal position
14
N MeNC 0.26 Apparently linear and low EFG
not possible. Requires annular
electronic distribution around
CN bond. See Fig. 4.6.
27
A1 [A1(OH2)6]3+ 2 Regular octahedron, linewidth
27
influenced by proton exchange.
Al 3990 Planar trigonal monomer with
high EFG. Linewidth at <*>
dilution.
14N [N03]- 3.7 Note contrast with previous
example. The difference is
explained by there being
electron density above and
below the plane of this trigonal
ion.
35
C1 [C104 1.2 Regular tetrahedron, aqueous
35
solution.
C1 PCL 7600 Terminal atom. Figure based on
relaxation time measurement
and is typical of many chlorides,
e.g. CH2C12, SiCl4.

www.pdfgrip.com
116 Nuclear magnetic relaxation

primarily by the bonding electrons rather than by long-distance effects

in the solid, then it is possible to obtain correlation times from the
relaxation times of quadrupolar nuclei in molecules in solution.

4.3.1 Spin-spin coupling to quadrupolar relaxed nuclei

Longitudinal relaxation involves changes in spin orientations relative
to BQ and so we might expect relaxation processes to modify the spin
coupling patterns that we observe. In fact, the relaxation times of 7 =
1/2 nuclei are sufficiently long that relaxation normally has no effect
on the analysis of the spectra, but the much shorter relaxation times
of the quadrupolar nuclei do lead to considerable modification of the
observed patterns.
If r1Q is long, then the normal spin coupling effects are observed,
such as the coupling of protons to both 10B and U B shown in Fig. 3.8.
At the other extreme, if TIQ is very short, the nuclear spins inter-
change energy and change orientation so rapidly that a coupled nucleus
interacts with all possible spin states in a short time. It can distinguish
only an average value of the interaction and a singlet resonance results.
This explains, for instance, why the chlorinated hydrocarbons show no
evidence of proton spin coupling to the chlorine nuclei 35C1 and 37C1,
both with I = 3/2. At intermediate relaxation rates, the coupling inter-
action is indeterminate and a broad line is observed. The line shapes
calculated for the resonance of a / = 1/2 nucleus coupled to a quadru-
pole relaxed 7 = 1 nucleus such as 14N are shown in Fig. 4.8. The shape
of the spectrum observed depends upon the product 7^7 where 7\ is
the relaxation time of the quadrupole nucleus, since if the frequency
defined by 1/2717^ is comparable with the coupling constant in hertz,
then the coupled nucleus cannot distinguish the separate spin states.
The situation is equivalent to attempting to measure the frequency of
a periodic wave by observing only a fraction of a cycle. The resonance
of the quadrupolar nucleus will, of course, be split into a multiplet by
the 7 = 1/2 nuclei, but each component line will be broadened by its
relaxation.
A common example of lines broadened by coupling to quadrupolar
nuclei is found in amino compounds. The protons on the nitrogen are
usually observed as a broad singlet, a good example appearing in Fig.
3.51. It is important to remember in this case that the spin-lattice
relaxation time of the amino proton is unaffected and can cause normal
splitting in vicinal protons bonded to carbon. In contrast, the protons
in the highly symmetric ammonium ion give narrow resonances
because the nitrogen quadrupole relaxation is slow (Fig. 3.8).
Since quadrupole relaxation is sensitive to temperature and viscosity,
the line shapes observed for coupled nuclei are altered by viscosity
and temperature changes, an increase in temperature leading to
slower relaxation and a better resolved multiplet. This fact is stressed,
since on a first encounter it seems contrary to one's expectation.
Alternatively, lowering the temperature increases the rate of quadru-

www.pdfgrip.com
 Spin rotation: detailed molecular motion 117

(a)-2 - 1 0 1 2 (e)-2 -1 0 1 2

i
(b)-2 - 1 0 1 2 (f) -2 -1 0 1 2

(c)-2 - 1 0 1 2 (g)-2 -1 0 1 2

(d)-2 -1 0 1 2 (h)-2 -1 0 1 2
Figure 4.8 The resonance line shape of a spin-1/2 nucleus coupled to a spin-
1 nucleus having various rates of quadrupole relaxation, (a) Tl ~ 0.05/Tr/. (b)
T; - 0.1/7T/. (c) TI « 0.2/ir/. (d) 7\ « 0.4/7T/. (e) 7\ - 0.8/ir/. (f) Tl « 2.0/ir/. (g)
Tl ~ 4.0/ir/. (h) 7\~ 8.0/TT/. The intensities are not to scale. The horizontal
axis is in units of Av//.

pole relaxation and, if the resonance of the coupled I = 1/2 nucleus

was already broad, this may well narrow as the relaxation becomes
too fast for any coupling to be exhibited. This technique can some-
times be used to advantage to simplify the spectra of / = 1/2 nuclei
coupled to quadrupolar nuclei.

4.4 SPIN ROTATION RELAXATION: DETAILED

MOLECULAR MOTION

Theories of molecular motion differentiate two linked types of motion,

which are given two correlation times T2 and TSR. The time T2 is the
orientational correlation time or the time between significant changes

www.pdfgrip.com
118 Nuclear magnetic relaxation

in orientation. This is clearly equivalent to the correlation time TC used

here, which can be obtained from measurements of riDD or TIQ, both
of which measure the reorientation of internuclear vectors, which may
or may not correspond to bonds. The time TSR is the angular
momentum correlation time or the time between significant changes
in angular momentum. TC and TSR are not independent since, if the
angular velocity is maintained for a longer time and T^ is long, the
orientation must change more rapidly and TC is short. Alternatively,
rapid random changes in angular velocity leave the orientation more
or less unchanged. Thus TC and TSR are related by Hubbard's equation
TcTsR =
6fcr
where T is the absolute temperature and 7 is the moment of inertia.
This molecular rotation has a further effect upon nuclear relaxation,
which can be detected if the dipole-dipole mechanism is weak. If the
rotation is particularly fast, the system of bonding electrons is subject
to some displacement relative to the atomic nuclei, which gives rise
to a small magnetic moment proportional to the angular momentum
of the molecule. Changes in the direction of the magnetic moment
provoke relaxation in the same way as with other processes, but with
the difference, which is unique to spin rotation relaxation, that the
faster the rotation, the longer the corresponding correlation time TSR.
The time T1SR is given by
2IkTC2 SR
T
1
- ^ISR -
1SR
TSR can be replaced with TC by using Hubbard's equation to give
1 2I2kC2
rp - 1SR ~~ 15/C
-I Qkfc2_
1
1SR " Tc

Thus increasing the temperature reduces TC and increases the efficiency

of this relaxation process so reducing 7\SR. A study of the tempera-
ture dependence of the nuclear relaxation time can thus distinguish
the spin rotation mechanism from the other mechanisms. T1SR is signif-
icant for small molecules, or rapidly rotating side groups in larger
molecules. The mechanism is particularly efficient in the gas phase,
and thus resonances of gases are much broader than observed in solu-
tions of the same molecules.
Two examples will illustrate these points. In the first, the relaxation
time of the nitrogen nuclei is measured in liquid nitrogen under pres-
sure and over a wide range of temperature. The two types of motion
were separated by using 14N2, which relaxes by the quadrupolar mech-
anism and gives T1Q and so TC and nitrogen enriched in the / = 1/2
nucleus 15N, which, because of its small magnetogyric ratio, relaxes by
the spin rotation mechanism and gives TISR and so TSR. The results are
shown in Fig. 4.9, where the opposite temperature dependences are
obvious.

www.pdfgrip.com
 Spin rotation: detailed molecular motion 119

The second example concerns the relaxation behaviour of

the quadrupolar nucleus 9Be in the hydrated cation [Be(H2O)4]2+. The
quadrupole moment of this nucleus is small, so that the quadrupole
relaxation mechanism does not necessarily dominate its relaxation.
Interaction with the protons of the water ligands causes dipole-dipole
relaxation, but this can be eliminated by using deuteriated water as
solvent. The results are shown in Fig. 4.10, where it will be seen that
the relaxation shows three different types of behaviour with change
in temperature: Tl increases with temperature at low temperatures,
then has a maximum where there is little change and then decreases
with further increase in temperature. Replacing H2O by D2O increases
the relaxation time at low temperature but has no effect at high
temperature. In the low-temperature region, then, for [Be(H2O)4]2+ we
have both quadrupolar and dipole-dipole relaxation, and for
[Be(D2O)4]2+ we have effectively only quadrupole relaxation, so that
the measurements in the two solvents allow us to separate the two
processes. Evidently, even for this tetrahedral molecule with low EFG,
the quadrupolar mechanism is the most effective. At high tempera-
ture the changed temperature dependence indicates that the spin

Temperature /K
80 90 100 120 140
30
20

CD
.i 10
4-»

I 6
X
05 4
A
o
a:

.J

12 10 8
3
10 /Temperature (K)
Figure 4.9 Measured values of the spin-lattice relaxation time, 7\, for liquid
14
N2 (in ms), o, and liquid 15N2 (in s), x, on the liquid vapour coexistence line
versus temperature (on a scale of 103 K/71, where T is the absolute tempera-
ture); c.p. indicates the critical point, and b.p. the boiling point at 1 aim.
(From Powles et al (1975) Mol Phys. 29, 539, reprinted with permission,
copyright (1975) Taylor & Francis.)

www.pdfgrip.com
120 Nuclear magnetic relaxation

rotation mechanism now predominates. The various relaxation rates

at 80°C are approximately RIDD= 7.7 x 10~3, RIQ = 4.3 x 10~2 and #1SR
= 0.24 s-1 respectively, the first two processes accounting for less than
18% of the overall relaxation rate.

4.5 CHEMICAL SHIFT ANISOTROPY RELAXATION

We have already discussed in Chapter 2 how the chemical shift is a

tensor quantity and varies as a molecule tumbles relative to the mag-
netic field direction. This averages to an isotropic value but neverthe-
less means that there is effectively a fluctuation in magnetic field
strength at the nucleus that can also cause relaxation. The mechanism
is not very efficient and, in the extreme narrowing region, depends upon

Temperature/°C
10080 60 40 20 0

100
80
60
40

20
.0)

10
8
6
4

3.0 3.5
3
10
-y- /K

Figure 4.10 The 9Be spin-lattice relaxation time in 1 M aqueous [Be(NO3)2]

as a function of temperature: (O) H2O solution; (•) D2O solution; (D) rlDD
calculated from the difference between the rates of relaxation in both solvents.
(From Wehrli (1976) /. Magn. Reson., 23, 181, with permission.)

www.pdfgrip.com
 Scalar relaxation 121

1 _2 7I 2 /? 0 2 (CT||-C7 J .) 2 T C
T
1
1CSA
1S
LJ

where yl is the magnetogyric ratio of the nucleus whose relaxation

time is required and the a are defined in Chapter 2. This mechanism
can be distinguished from others by the fact that it depends upon BQ2.
It is important for nuclei with a rather high screening anisotropy and
high chemical shift ranges and at high magnetic fields. In some cases,
it is an advantage since it reduces otherwise long relaxation times at
high field, making easier the observation of 13C signals from non-
proton bearing carbon atoms which are sp2 or sp hybridized, with
°"ll ~ °± °f the order of 200 and 400 ppm, respectively. In other cases
such as for 199Hg in [Me2Hg] where <T|| - a± is of the order of 7500 ppm,
signals can be substantially broadened.
An example has recently been reported that contrasts the relaxation
behaviour of the / = 1/2 nucleus 195Pt in the two anions [Pt(CN)4]2-
and [Pt(CN)6]2~. The former has square planar geometry and so a large
chemical shift anisotropy (CSA), which has been measured to be a
a|| - a± = 2500 ppm in the solid state. The octahedral anion has no
such anisotropy. The relaxation times were measured as functions of
both temperature and of magnetic field. Two spectrometers were used,
one operating at 4.70 T (195Pt at 42.8 MHz) and one at 8.48 T (195Pt at
77.04 MHz). The results are plotted in Fig. 4.11. Only one set is shown
for [Pt(CN)6]2~ since the relaxation rates were the same at both
magnetic fields, the rate of motion being fast enough to ensure that
the extreme narrowing condition was met. However, the rate of relax-
ation increases with temperature and so the mechanism of relaxation
must be that of spin rotation, any dipolar contribution from the 14N
nuclei or from the solvent (D2O) being small. The behaviour of the
relaxation of [Pt(CN)4]2~ is quite different; the rate is increased at the
higher magnetic field and the plots are curved, indicating the presence
of two competing relaxation mechanisms, spin rotation at the highest
temperatures and CSA at the lowest, with opposite temperature
dependences.
It is not only the nucleus itself that is affected, but nuclei to which
it is coupled. This is illustrated in Fig. 4.12 for 3J(l95PilH). At 90 MHz,
the 195Pt satellites on either side of the proton signal for H2 are sharp,
but as the magnetic field is increased, so that !H resonates at 250 and
then 400 MHz, the 195Pt satellites progressively broaden, reflecting the
shortening of T^Pi).

4.6 SCALAR RELAXATION

We have already mentioned that in many cases quadrupolar nuclei

relax so rapidly that any effects of spin-spin coupling to other nuclei
are completely lost. If, however, the rate of relaxation of the quadrupo-
lar nucleus is very fast indeed, and if the coupling constant in the

www.pdfgrip.com
122 Nuclear magnetic relaxation

8.48 T
1.5- [Pt(CN)4f-

1.0-

0.5-
4.70 T
[Pt(CN)4f-

0.0-

-0.5-
[Pt(CN)6f-

2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6

103/T (K~1)

Figure 4.11 The 195Pt relaxation rates for aqueous solutions of K2Pt(CN)4 and
K2?t(CN)6 at B0 = 4.70 and 8.48 T as functions of inverse temperature. The
i95pt relaxation rates of K2Pt(CN)6 are independent of BQ. (From Wasylishen
and Britten (1988) Magn. Reson. Chem., 26, 1075, copyright John Wiley and
Sons Ltd, New York, reprinted with permission.)

absence of the quadrupolar relaxation is significant, then the relaxation

time of the coupled nucleus may be influenced by the quadrupolar
nucleus. For Tl9 the effect of scalar coupling is rarely significant,
especially at high field. The contribution of scalar coupling to Tl of one
nucleus, /, by a second rapidly relaxing nucleus, S, is given by
1 8v2J2S(S + 1)7^
r lsc ~3{l + (o)7-o)5)2(r/)2}
where / is the coupling constant between / and S, S is the nuclear spin
quantum number of nucleus 5, T^ is the spin-lattice relaxation time
of nucleus 5, co7 and o>5 are the frequencies in rad s'1 of the nuclei
I and S. Clearly this term is only significant when (a)7 - o)5)2 is small
and / is large. This condition is rarely met. T1SC has been reported as
being significant for 13C attached to 79Br and to 185>187Re.
In contrast, r2SC can be dominant. It follows the equation

www.pdfgrip.com
 Scalar relaxation 123

400 MHz

250 MHz

90 MHz

Figure 4.12 1H NMR spectra for H2 in [PtCl(COC9H6N)(PEt 1

3)] at magnetic
3 l95 l
field strengths corresponding
195
to 90, 250, and 400 MHz for H. J( Pt H) =
21.1 Hz. Note that the Pt satellites are sharpest at 90 MHz. (Reproduced with
permission from Anklin and Pregosin (1985) Magn. Reson. Chem.,23,671, John
Wiley and Sons Ltd, reprinted with permission.)

-±- = \^ + 4-^S(S + l}T,s

2 Z J
2SC ^ISC

where / is the coupling constant between the observed nucleus and

the quadrupolar one, and 7\5 is the relaxation time of the quadrupolar
nucleus.
There can be substantial contributions to T2 and hence the linewidth
when / is comparable with 1/Tf. This condition is commonly met for
nuclei attached to nuclei such as UB, 14N, 27A1, 51V, 55Mn, 59Co, 63Cu,
and 65Cu, often resulting in substantial line broadening. This is illus-
trated in Fig. 4.13 for [Mn(CH2Ph)(CO)5]. At 30°C, the carbonyl
signals are very broad due to broadening by the 55Mn via T"2SC, as a
result of the large U^Mn^C) and moderate T/^Mn). On cooling to
-87°C, the 71! for the 55Mn (/ = 5/2, 100% abundant) shortens due
to the lengthening of TC, and the broadening of the 13CO signals
due to the 55Mn is considerably reduced because 1/(55Mn13C) has not
changed in value, but T15(55Mn) is much shorter producing a smaller
1/T2SC and sharper 13C signals.

www.pdfgrip.com
124 Nuclear magnetic relaxation

(b)
Figure 4.13 The 13C NMR spectra of [Mn(CH2Ph)(CO)5] in CH2C12 at (a)
30°C, and (b) -87°C. Note the improvement in both the linewidth and
signal:noise at low temperature. (Reproduced from Todd and Wilkinson
(1974) /. Organomet. Chem., 80, C31, copyright (1974), with permission from
Elsevier Science.)

In the case of [195Pt(CN)6]2-, for instance, if the T2 value of the 195Pt

equalled 7\, the linewidth would be around 1 Hz. In fact, it is 25 to
60 Hz, depending upon temperature, and this is due to coupling with
the rapidly relaxing 14N nuclei in the anion. In this case the frequency
difference 195Pt-14N is too great for there to be any influence on Tv

13.5 14.8 4.7 EXAMPLES OF 13C RELAXATION TIMES

HC—CH
when
11 9 HC C—C17 we are setting up an NMR experiment, it is important that we
\ / \\ have some idea of Tl for the observed nuclei so that an appropriate
HC==CH CH2 pulse angle and relaxation delay can be selected. In small molecules
7 8
- Tl is generally long. For example, in styrene, 4.1, the Tl values for 13C
range from 7.8 to 75 s. When it is remembered that 13C relaxation is
(4-l) dominated by dipole-dipole relaxation, the variations in Tl can be
explained in terms of the C-H distance and the correlation time. The
ipso, non-proton bearing carbon has the longest Tl as there are no
close protons to relax it. The CH2 group has the shortest as the carbon
has two attached protons. The variation in Tl for the remaining CH
carbons is due to differences in molecular motion. The molecule can
tumble anisotropically, but it is simplest to attribute most of the differ-
ences to the differences in TC due to rotation about the vinyl-phenyl
bond. The para-CH group lies on this axis and is not moved by the

www.pdfgrip.com
 Examples of 13C relaxation times 125

rotation and has the shortest 7\ value of the CH carbon atoms. The
ortho- and meta-CH and the vinylic CH groups lie off the axis and
the hydrogen atoms move with respect to the carbon atoms due to the
rotation. The result is that TC is shortened and Tl becomes longer.
Thiamine hydrochloride provides a more representative example.
The 7\ values are given in Fig 4.14 for the compound in CH3OH and
CD3OD. Note that certain non-proton bearing carbon atoms have
much longer 7\ values than the proton bearing ones, but the Tl is
reduced for these non-proton bearing carbons in the protio solvent
due to the proximity of the solvent protons. The Tl of the methyl
group is lengthened by rotation relative to the CH carbon atoms. By

(a) (b)
13
Figure 4.14 The C relaxation times of thiamine hydrochloride in (a) CH3OH and (b) CD3OD, measured
at 25.15 MHz.

extrapolation, one might expect the relaxation time to be about 0.1 s

in the absence of CH3 rotation.
The final example is a compound with a long chain, phytol, 4.2. The
central CH2 carbon atoms have 7\ values around 0.3 s, while the CH
carbon atoms have Tl values around 0.5 s as they only have one proton
to provide the relaxation. The carbon atoms at the ends of the chain
have longer Tl values. This is particularly noticeable at the CHMe2
end of the chain. This arises because rotation around the C-C bonds
provides an extra mode of motion, increasing Tv The effect is not so
marked at the HOCH2 end of the chain as hydrogen bonding to the
OH group tends to anchor this end of the chain. Despite having three
protons to provide relaxation, the methyl carbon atoms have consid-
erably longer 7\ values due to rotation of the methyl groups.

(4.2)

www.pdfgrip.com
126 Nuclear magnetic relaxation

4.8 QUESTIONS

4.1. A / = 1/2 nucleus is relaxed by three different mechanisms and

has a measured relaxation time of 1 s. The three contributions
are spin rotation, whose characteristic time is 2.5 s, chemical shift
anisotropy, with characteristic time 1.8 s and a long-range
dipole-dipole interaction. Calculate the characteristic time of this
latter contribution.
4.2. Given an isolated 13C1H2 fragment in which the C-H distance
is 109 pm and the H-H distance is 177 pm, use equations (4.2) and
(4.3) to calculate the ratio of the rates of dipolar relaxation of the
two nuclear species. The magnetogyric ratios should be taken as
proportional to the nuclear frequencies given in Chapter 1.
4.3. It is pointed out in Table 4.1 that the electric field gradient at
14
N in the nitrate ion, [NO3]~, is almost zero, and that this is
surprising. Use equation (4.4) to obtain an expression for the
EFG at 14N given that there are three charges q producing the
field gradient and that these are situated in the N-O bonds at a
distance r from the nitrogen atom. It will be found most conve-
nient to choose the z axis as that perpendicular to the NO3 plane
and passing through the nitrogen atom, and to calculate Vzz. How
far do we have to displace the charges from the NO3 plane
keeping r constant in order that the EFG will be zero at N?
Does it make any difference as to which side we make the
displacement?
4.4. In metal carbonyls, Tl of 13CO is normally dominated by chem-
ical shift anisotropy. Take cr|| - crj_ = 400 ppm and TC = 10~n s.
Calculate 7\CSA for 13C in such a carbonyl at 4.7 and 9.4 T. Note
that yc = 6.7263 x 107 rad T'1 s-1.
4.5. In the molecule, shown in the margin, use equation (4.2) to calcu-
late the relative 7\ values of the CH and CH2 protons, assuming
that only dipole-dipole relaxation contributes to Tl and the same
TC applies to both types of protons. Take the HC=CH and CH2
interproton distances as 270 pm and 185 pm respectively. Ignore
any dipole-dipole relaxation arising between the HC=CH and
CH2 groups.
4.6. For a C-D bond, the 2H nuclear quadrupole coupling constant,
X, is usually 180 ± 20 kHz. Given that 7\(2H) of CD3I is 6 s, use
equation (4.5) to calculate TC for the molecule. Assume that TI is
negligible.
4.7. The 13C NMR signal of C2'6 of pyridine, with the effect of the
protons removed by double irradiation, is much broader than the
signals due to C3'5 and C4. Explain why this occurs. On cooling,
the signal sharpens. Explain why this happens.
4.8. The 25.2MHz 13C Tl values for PhC^CC^CPh are given
overleaf.
(a) Account for why the 7\ of the para-carbons are much
shorter than for the ortho- and mem-carbons.

www.pdfgrip.com
 Questions 127

5.2s 5.5s
H H H H
C—C C—C
// \\ 82s J 36s / \
a p
1.1s HC C —C =C -C=C—C x , CH
\ / \\ //
c=c c—c
H H H H

(b) When the 13C Tl values are determined at 63.1 MHz, values
of 15 s and 30 s were found for Ca and Cp. Explain why the
Tl values have decreased by so much.
4.9. The 13C Tl values have been determined at 25.15 MHz for the
ipso carbon in bromobenzene as 3.6 s for C6H579Br and 16 s for
C6H581Br. Explain why the two values are different. What is going
to be the effect on these Tl values of increasing the magnetic
field so that the 13C frequency becomes 50.3 MHz?
4.10. Although the relaxation of proton bearing 13C nuclei is normally
dominated by the dipole-dipole relaxation mechanism, no signif-
icant dipole-dipole relaxation has been found for 103Rh in
rhodium hydrides. Explain this observation.
4.11. In Fig. 4.11, it is shown that for J?0 - 8.48 T and Iffi/T = 2.85, the
195
Pt relaxation rate in [Pt(CN)4]2- is given by In R± = 0.8.
Calculate the 195Pt linewidth of this anion under these conditions
assuming Tl = T2 and a perfectly homogeneous magnetic field.
The measured value of the linewidth is, in fact, 90 Hz. Calculate
T2 and estimate the contribution of scalar relaxation to this value.

www.pdfgrip.com
www.pdfgrip.com
 The spectrometer 5
We have seen in the last three chapters that there are three principal
parameters in an NMR spectrum: the chemical shift, the coupling
and, less evidently, the relaxation. The present chapter is devoted to
showing in more detail how the spectrum is obtained and how these
parameters may be derived from the data collected.
Chapter 1 discussed how a 90° B1 pulse at the nuclear frequency
can swing the magnetization from the direction of the magnetic field
into the xy plane, and showed that this magnetization, which is now
rotating at the nuclear frequency, can be detected in a suitable coil.
In Chapter 4 we discussed how this magnetization is affected by a
variety of relaxation processes, in particular, how T2 leads to an expo-
nential reduction with time of the xy magnetization. The spectrometer
output thus consists of a signal at the nuclear frequency, which decays
with time until it is no longer detectable. We need to know how
this can be translated into typical spectral form. The equations
governing the behaviour of the transverse and longitudinal magneti-
zation M and Mz and their return to equilibrium following a 90° Bl
pulse are

(M,)t = (M z )Jl-exp(-^-)|
L \ M/J

i.e. Mz increases from zero to its equilibrium value, and

(Jlf ) r =(Jlf )0exp - —

L
\ 2I

i.e. the transverse magnetization falls from its maximum value (equal
to Mz) to zero after sufficient time has elapsed. This behaviour is
summarized diagrammatically in Figs 4.1 and 4.2.
In general, the rate of decay of the spectrometer output is faster
than predicted from the value of T2 in a given system. This occurs
because of inhomogeneities in the magnetic field throughout the
sample due principally to imperfections in the magnet system, which
means that the nuclear frequencies are slightly different in different
parts of the sample volume and this increases the rate at which the
spins throughout the sample lose phase and coherence. We write that
the apparent relaxation time, T2*, is equal to

www.pdfgrip.com
130 The spectrometer

J___JL 1
T
1 # ~~ 1T LT
2 2 inhomo

where rinhomo is the decay in intensity due to the field inhomogeneities

alone. This decaying output is known as the free induction decay or
FID. We now need to know how the FID is produced in practice, how
it is collected and how it is processed to produce a spectrum. A block
diagram of an NMR spectrometer is shown in Fig. 5.1 and should be
referred to in the discussion which follows.

5.1 THE MAGNET AND FIELD HOMOGENEITY

A modern NMR spectrometer is built around a superconducting solen-

oid magnet which, indeed, is the heart of the system. The magnet
consists of a coil of superconducting niobium-tin alloy sitting in a bath
of liquid helium at 4.2 K. In order to reduce liquid helium loss by
boiling off, the liquid helium is protected by a vacuum jacket, which

iField correction! H H^ lmagnet

*—7! LJ: ' |500 MHz |

r "^ probe • 1
F
•"Vf^ijrJ. jji^->F
,t M t , Powf [|22MHz|
Shim Coil [Amplifier | f
Current f j||f I Quadrature
Supply I pui-- I—*— ' ^ Detector
I Timer - - T |22 MHz| ' I I
I Auto I | Shaper] Low *f f\
' Shim |iiiA Frequency I Analogue-
—i— I T T T T T — . _ - Qjgjtai
KS^ 1°^pl
r
\ ' i

! TT
^--[Digital Filters]

.u..r -] Computer
i | [Data memory I
! i i
Interactive! [Printer] |Disk storage]
display [

Figure 5.1 A block diagram of an NMR spectrometer.

www.pdfgrip.com
 The magnet and field homogeneity 131

1 Ports for liquid N2

2 Ports for liquid He
3 Superinsulation and high vacuum
4 Main magnet coils and liquid helium
5 Sample lift and spinner assembly
6 NMR tube
7 Shim assembly
8 Probe

Figure 5.2 A representative cross-section through a superconducting magnet.

(Reproduced with permission from Braun et al (1996) 100 and More Basic
NMR Experiments, VCH, Weinheim.)

itself is then cooled with liquid nitrogen, which in turn is also protected
by the vacuum jacket (Fig. 5.2). Modern systems are very efficient with
25 1 of liquid helium lasting for at least three months. The liquid
nitrogen needs topping up regularly, typically once a week. Once the
current has been established in the superconducting coil, it will
continue to give an acceptable magnetic field for at least ten years,
before the current has decayed sufficiently to need to be topped up
again. The current in the superconducting coil is quantized, as is the
magnetic field. If an NMR tube containing CHC13 is placed in the
magnet, and a single pulse spectrum taken every hour, then it will be
found that the position of the signal remains constant for a while, and
then jumps one or two hertz to lower frequency, and again remains
constant for a while. This experiment monitors the quantized jumps
of the magnetic field, which of course have to be counteracted to give
a permanently stable field.
Although there are no reasons to believe that a static magnetic field
can cause harm to personnel, it is usual to mark a 5 gauss (0.0005 T)
contour line around the magnet, and to try to stay outside this region
as far as is possible. The major recognized hazard associated with the
stray magnetic field is its ability to capture large metallic objects such
as tools and gas cylinders which accelerate rather rapidly towards the
magnet! Nor should those with heart pacemakers go anywhere near.
Anything, including floppy discs and credit cards, containing magnetic
records can be damaged by the stray magnetic field.

www.pdfgrip.com
132 The spectrometer

The sample sits in the middle of the solenoid. It is vital to obtain a

uniform magnetic field over the sample. A modern superconducting
magnet is capable of producing a magnetic field which gives a sample
linewidth of better than 0.1 Hz in, say, 400 MHz. The magnetic field
then has to be uniform to better than 1 part in 4 x 109. This is achieved
by attention to three techniques:
1. shimming;
2. sample spinning;
3. the use of very uniform samples.

5.1.1 Sample spinning

For high resolution NMR spectroscopy, the solution of the sample is
placed in an NMR tube, frequently of 5 mm outside diameter. In order
to partially average inhomogeneity at right angles to the main magnetic
field direction, the sample is rotated around its axis, typically at 20 Hz,
using an air turbine which consists of a cylindrical spinner in which
the sample tube fits and a static outer fixed in the centre of the solen-
oid and to which the spinning air is fed. The spinner/tube combination
is removed from, or put into, the outer by means of an air lift, which
gently manipulates the spinner up or down.

5.1.2 Shimming
Two sets of coils are wound to produce magnetic field gradients to
cancel those inherent in the main magnetic field. The first set consists
of superconducting coils, which are wound within the liquid helium
bath and are adjusted as part of the magnet installation. The currents
in these coils are not subsequently changed by the operator, but the
operator has to be aware of their existence as they can quench, i.e.
lose their current, and the resolution deteriorate. The second set of
shim coils is mounted around the probe and the currents in them are
adjusted to cancel any remaining field gradients and give optimum
resolution. This second set of coils comprises:
1. Zero-order coil. There is a single zero-order coil with its axis
along the direction of the superconducting solenoid and this is
used to adjust the main magnetic field strength within narrow
limits.
2. First-order coils. There are three first-order coils called XI, Yl, and
Zl, which produce magnetic fields as the functions x, y, and z. They
produce field gradients shaped like p-orbitals. The x and y coils are
aligned at right angles to the main magnetic field, while the z coil
is coaxial with the main field.
3. Second-order coils. There are five second-order coils called XY,
X1-Y1, ZX. ZY, and Z2, which produce magnetic fields as the func-
tions Jty, x2 - y2, zx. zy, and z2. They produce field gradients shaped
like d-orbitals.

www.pdfgrip.com
 The magnet and field homogeneity 133

4. Third-order coils. There are seven third-order coils called Jt3, Y3,
Z2X. Z 2 Y, ZXY, Z(X2-Y2), and Z3, which produce magnetic
fields as the functions x3, y3, z2x. z2y, zxy, z(x2-y2), and z3. They
produce field gradients shaped like /-orbitals.
5. The magnets designed for the highest magnetic fields are often also
equipped with fourth- and fifth-order z-shims.
It has to be emphasized that the quality of the spectra depends
absolutely on minimizing the field gradients with the shims. Adjusting
all the shims, however, is a highly skilled operation and beyond the
scope of this book, but some excellent texts are available, and listed
in the Bibliography. The inexperienced user is best advised to restrict
changes in the shim currents to those in the Z and Z2 coils when the
sample is spinning. The shim currents are adjusted by optimizing
the height of the lock signal for maximum, see section 5.3. This may
need to be accompanied by adjustment of the lock phase for maximum J
height. If the X and/or Y shim currents are mis-set, then the magnetic
field at a given point in the sample fluctuates as the sample spins, the 20 0 -20 Hz
nuclear frequencies are modulated at the spinning frequency and the
NMR signals will be flanked by spinning side-bands (Fig. 5.3). These Figure 5.3 A 1H NMR
shim currents then have to be adjusted with a non-spinning sample. spectrum of CHC13 show-
Shimming is critically important in obtaining a high-quality NMR spec- ing spinning side bands
due to mis-setting of the
trum. A good quality control of the shim settings is to routinely current in the X shim
examine the signal from the TMS reference. It should be a singlet with coil. The sample tube was
well-resolved 29Si satellites, 2J(29SilH) = 6.6 Hz. It is also useful to spun at 20 Hz and the
watch the FID as it accumulates. It is often possible to identify a spinning side bands are
symmetrically situated at
sample that is giving poor resolution after only one FID has been ± 20 Hz of the main signal.
acquired. For 1H, the FID should last for several seconds, and the (Reproduced with per-
presence of a beat after one or two seconds while it may be due to mission from Braun et al
coupling, often indicates poor resolution. (1996) 100 and More Basic
The position of the bottom of the NMR tube and the top of the NMR Experiments, VCH,
Weinheim.)
solution in the tube distort the magnetic field and so contribute to
the field gradient. Provided a gauge is used to position the spinner
on the NMR tube, so that the bottom of the NMR tube is always in
the same position, and a relatively long sample length is used, then,
on change of sample, it should only be necessary to adjust the shim
currents in the Z and Z2 coils when the sample is spinning.
Manual shimming is becoming replaced by computer shimming. For
many years, it has been possible to use iterative routines to adjust the
shims automatically, but these routines are slow, and are normally only
used to compensate for drift during long data accumulations. It is
becoming common for probes to be equipped with a z-magnetic field
gradient coil, see section 5.8.6. This coil can be used to map magnetic
field inhomogeneity in the z-direction, and the Z, Z2, Z3, Z4 and Z5
corrections are then calculated and applied by the computer. This only
takes a few minutes and uses the same methods as will be described
in Chapter 11 for magnetic resonance imaging. Less common is for
probes to be equipped with #-, y- and z-magnetic field gradient coils,

www.pdfgrip.com
134 The spectrometer

where the complete magnetic field inhomogeneity can be mapped and

corrected, but this is a lengthy procedure.

5.1.3 Sample preparation for high resolution NMR spectroscopy

Sample preparation is extremely important to obtain well-resolved
spectra. The sample must be dissolved, usually in a deuteriated solvent.
The resulting solution must be homogeneous and free from particles
and paramagnetic impurities. This is usually achieved by making up
the solution in a sample tube or flask and then filtering the solution
or passing it through a short alumina plug in a Pasteur pipette into
the NMR tube. If it is then necessary to add more solvent to the tube,
the resulting solution must be well mixed to avoid concentration gradi-
ents which imply changes in magnetic susceptibility and which destroy
the resolution. In order to obtain good quality NMR spectra, the NMR
tube must be accurately cylindrical and of uniform wall thickness to
avoid distortion of the field by irregularities in the glass, which is of
different magnetic susceptibility to the sample. Thus, good quality
NMR tubes must be used, with the specification becoming tighter for
increasing magnetic field strength.

5.2 THE PROBE

A cross-section through a typical multinuclear NMR probehead is

shown in Fig. 5.4. There are two coils. In a conventional probe, the
inner coil provides X-nucleus observation, usually covering the range
of frequencies from 109Ag to 31P. This coil is double-tuned so that it
is also used for the 2H lock channel, see section 5.3. The outer coil
is used for ]H observation and decoupling. The sensitivity of such a
probe is optimized for the X-nucleus, and the 90° pulse for the
X-nucleus will be short, but the 1H 90° pulse will be long. An alterna-
tive configuration is found in the inverse probe, where 1H and the 2H
lock are on the inner coil and the X-nucleus is on the outer coil. This
type of probe is ideal for XH sensitivity and inverse detection of X
using techniques such as HMQC, HMBC, and HSQC, see Chapter 9.
There are many variations on this design.
There are several glass tubes present which support the coils and
direct the variable temperature gas flow. Immediately below the
sample is a thermocouple. Around the whole assembly is a Dewar that
stabilizes the temperature. Warmed or cooled gases can be passed
through the probe in order to vary the temperature of the sample,
which is held constant at a given value by a control circuit based on
the thermocouple. Temperature stability is very important since
temperature gradients in the sample cause changes in magnetic suscep-
tibility and chemical shifts which can degrade the resolution. If the
spectrum is temperature sensitive (see Chapter 6) then temperature
gradients can also broaden the lines. The presence of glass tubes in

www.pdfgrip.com
 Field-frequency lock 135

gasket —
outer inner glass
glass tube" tube
decoupling receiver
coil coil

Figure 5.4 A cross-section through a typical NMR probehead. Note all the
glass in the probehead which makes it fragile. The coils are wound as saddles
on the glass tubes (Helmholtz double coils) to give the RF field normal to
B0. (Reproduced with permission of Bruker Spectrospin.)

the probehead means that it is fragile and failure to insert the NMR
tube on a cushion of air can result in breakage.

5.3 FIELD-FREQUENCY LOCK

These devices, which are relatively simple, are used in conjunction

with FT spectrometers to provide a final stage of stabilization using a
nuclear resonance in the sample other than the one to be observed,
often the strong deuterium signal from a deuteriated solvent. They are
based on a time sharing principle with transmitter and receiver on
alternately (Fig. 5.5). The response which is detected during receiver
on-periods is strong enough to give an analogue output. In fact, the
modulation of the transmitter gives sidebands at the pulse modulation
frequency which are continuous and which provoke slight precession
of the magnetization of the lock nucleus which therefore gives a contin-
uous output which can be detected as either absorption or dispersion
signal, see section 5.6. The device is used in two ways. A small trian-
gular repetitive sweep is applied temporarily to the magnetic field and
the deuterium is observed on the monitor in the absorption mode as
its resonance is traversed repeatedly. We observe a signal, and can use
this, maximizing its height with shim adjustment, in order to optimize
the homogeneity of the magnetic field. We then switch off the field
sweep and turn our attention to the dispersion signal produced by the
solvent. This has the property that, at resonance, the output is zero,

www.pdfgrip.com
136 The spectrometer

Figure 5.5 Receiver/transmitter time sharing of the field-frequency lock. The

receiver is switched off just before the transmitter is turned on, and remains
off for a while after the transmitter is turned off.

but if drift of field or frequency occurs, either a negative or a positive

output is obtained, the sign depending upon the direction of drift, and
we can use this output to provide a correction voltage to alter the
magnetic field until the output is again zero (Fig. 5.6). Thus the
frequency and field are locked together indefinitely and we can in prin-
ciple record as many FIDs as we wish and collect them in memory,
100 000 being quite feasible, though the number is always kept as small
as possible since there is very often a queue of people waiting to use
the spectrometer. It is, of course, necessary that the frequency of the
lock device is derived from the FT drive frequency, otherwise it could
vary independently. The lock then keeps the ratio BQ /v0 constant and
is known as a field-frequency lock. The whole system is shown in the
block diagram of Fig. 5.7.

Stabilizer
output

Figure 5.6 A resonance in the dispersion mode can be used to stabilize an

NMR spectrometer. At exact resonance, the output is zero, but if drift occurs,
it becomes either positive or negative. The output can be used to provide a
correction to the magnetic field, which reverses the drift until the exact reso-
nance position is regained.

www.pdfgrip.com
 The transmitter 137

1 Dispersion
— PSD >-
I 1 to magnet
field control
, , I Receiver | I 90° Phase
~~~1^-^" switch I Chan9er
Probe ^T '
I ^^^--^ Transmitter |
1
switch I 1 Absorption
' I PSD ^~
pulse J^ck 1 ^ to monitor
frequency] [frequency screen
Frequencyl L Phaqp I
Synthesizer Mhase
i— 1 ' [adjustment

Computer

Figure 5.7 A block diagram of the field-frequency lock system. The frequency
synthesizer provides both the pulsing frequency and the lock frequency. The
pulsing produces sidebands of the lock frequency and the lock operates on
one of these. The continuous output of the nuclear lock signal is fed to two
phase sensitive detectors (PSD, to be described in section 5.5.1) operating in
quadrature which produce the dispersion and absorption signals to correct the
magnetic field and enable the signal to be monitored. Means are provided to
adjust the phase to ensure that the spectrometer is operating at its optimum
position.

It is possible to use a number of deuteriated solvents for the lock,

such as CDC13, C6D6, (CD3)2CO, etc., and these all have different
chemical shifts and so lock frequencies. It is arranged that each lock
substance has its own frequency stored in memory and that the
magnetic field shall be identical whatever lock substance is used. This
means that all standards are referenced to an absolute value.

5.4 THE TRANSMITTER

The transmitter provides the pulses to stimulate the FID and also to
manipulate the nuclear spins in a variety of ways. The centre of the
transmitter system is a quartz controlled frequency synthesizer which
generates the frequencies needed for a given experiment and also that
required for the lock system described above. The frequencies
produced are locked together so that they always have the same rela-
tionship and are continuous and coherent, which is to say that they
are pure, continuous sine waves with no irregularities. Some of the
outputs are also used in the receiver, as we shall see. The transmitter
frequencies are used to drive the power amplifiers which produce the
pulses but have first to be processed. Pulses are produced by a switch

www.pdfgrip.com
138 The spectrometer

which applies the signal to the amplifier for the required short time.
Such pulses may be rectangular or they may be shaped by a computer
controlled device to give pulses with either a broad bandwidth or a
narrow bandwidth. These pulses are effectively bits of the input contin-
uous wave. Means are also provided to shift the pulse oscillations rela-
tive to the drive signal, in other words to phase shift them. The
processed signals are then amplified in the power amplifiers, one for
each frequency, and provision is made for the output to be varied in
intensity, the degree of attenuation usually being expressed in a dB
(decibel) scale, the more dB, the less the intensity. The decibel scale
is a log scale, being related to power, P, by

dB = 101og]0(^) = 201og10(^)
\r2' \v 2f

where Pl and P2 are the powers and Vl andV2 are the applied voltages
being compared. Hence increasing the dB value by 3 units approxi-
mately halves the power being used. However, as the radiofrequency
field strength is proportional to voltage, to halve it, an approximate
increase in the dB value of 6 units is required.
It is particularly important to be able to control the relative phases
of the output signals. This can be understood by referring to Fig. 5.8.
The pulse gives the field Bl in the xy plane, rotating with the nuclei
on, say, the x' axis in the rotating plane and the nuclear magnetiza-
tion rotates around B^ into the yf axis. If we use a 90° pulse then this
is called a (90°)^ pulse. If we phase shift the transmitter output by 90°
(do not confuse pulse lengths with phase shifts even though both are
expressed in the same units), this moves Bl 90° in the x'y' plane to
the y' axis and the nuclear magnetization now rotates into the -x' axis.
This is known as a (90°)^ pulse. It is equally possible to place the
magnetization along the jc' or -y' axes using a (90°)^ or (90°)_r phase
shifted pulse. We shall find this facility extremely useful below.

(90°)* (90°)y (90°)_x (90°)_y

Figure 5.8 The effect of the phase of the 90° pulse on the magnetization with
respect to the x'y' rotating frame. The direction of the arrow within the circle
represents the direction of the magnetization in the x'y' plane after applying the
90° pulse about the jc, y, -Jt, or -y axes as indicated by the subscripts after the
bracket.

www.pdfgrip.com
 The detection system 139

5.5 THE DETECTION SYSTEM

For further processing the FID needs to be detected and stored in a

computer memory. The nuclear frequency is high, tens or hundreds
of megahertz (MHz), and so is difficult to handle with normal data
storage devices. For this reason it is necessary to reduce its frequency,
and this is done in two stages. First, the superheterodyne principle is
used and the nuclear signal is mixed with another a little different in
frequency and locked to the spectrometer frequency, to give the much
lower difference frequency which is called the intermediate frequency
or IF. Then this is detected using a device called a phase-sensitive
detector, in which the modified nuclear signal is compared with the
frequency source used to generate Bl9 also modified by mixing to
reduce its frequency. The frequency of the nuclear signal thus is
reduced to the difference between the two, which is zero to a few kilo-
hertz. The intensity of each signal present is preserved and, in addi-
tion, the phase of the signal is determined relative to the phase of Bl9
the reference signal. Knowledge of all three factors is essential in order
that a spectrum can be extracted from the data, and knowledge of
phase is particularly useful in some of the more complex experiments
that we will encounter. It is useful, perhaps essential, to understand
what the phase-sensitive detector does.

5.5.1 The phase-sensitive detector

This device is essentially a switch driven by the Bl signal and which
reverses its polarity in step with each half cycle of B^ A simplified
circuit is shown in Fig. 5.9. The input nuclear signal is thus directed
in one direction during the positive Bl half-cycle and in the opposite
sense during the other half-cycle. If the Bl and nuclear frequencies

Figure 5.9 Schematic diagram of a phase-sensitive detector. The reference

signal operates the switches, one way for the positive half cycle and the other
way for the negative part of the wave. This regularly reverses the output
polarity relative to the input polarity. The input sine wave is thus rectified
and the fast fluctuations in voltage removed by the filter to give a smoothed,
constant or slowly varying output. The polarity of the output depends on the
relative phases of signal and reference and is zero if these are shifted by 90°.
The switching is done digitally using fast gallium arsenide transistors.

www.pdfgrip.com
140 The spectrometer

are exactly the same, then the nuclear signal is rectified with, say, the
negative half cycles reversed in polarity, and a direct current output
is obtained after removal of the high frequency components by a suit-
able low pass filter. If the two signals are in phase, then the output is
positive; if they are 180° out of phase, the output is negative; and if
they are of intermediate phase, then the output is reduced and is zero
when the phase difference is 90°. The intensity of the output thus
depends on both the intensity and phase of the nuclear signal input.
If the nuclear and Bl frequencies are not the same, then their relative
phase fluctuates with time and the output becomes a wave oscillating
at the difference frequency. The phase information is still retained
since the output will have some phase angle relative to Bl immedi-
ately after the 90° pulse, i.e. at the start of the decaying output. The
spectrometer output after phase-sensitive detection is thus oscillatory,
referred to Bl and with a much reduced frequency. It decays with time
characterized by T2* (Fig. 5.10). We should also note that this result
is obtained if the spectral frequencies are all higher than the B1
frequency or all lower, though the order of the spectral lines is reversed
between the two cases. Evidently, we cannot have some lower and
some higher as this would mix the resonance positions up so as to
prevent sensible analysis of the spectrum.

5.5.2 Analogue filters

The output of the phase-sensitive detector consists of wanted, low
frequency signals, electronic noise of all frequencies and the higher

(a) (b)

(c) (d)

Figure 5.10 The possible range of outputs when the nuclei are exactly on
resonance with Bl and are in-phase (a) or 90° out-of-phase (b). If the nuclear
and Bl frequencies are different, then the output is at the difference frequency,
but starts on a maximum if they start in phase (c) or a minimum if they start
90° out-of-phase. Any intermediate phase is possible.

www.pdfgrip.com
 The detection system 141

frequency IF inputs. To separate the signal we want from the rest, we

use an analogue filter, based on a simple resistance-capacitor network.
Such filters are designed to pass the low frequencies and attenuate the
high frequencies. Thus there is no attenuation up to what is called
the cut off frequency and then the attenuation increases steadily with
frequency. Such filters remove the IF satisfactorily but leave much noise
at frequencies above the pass band and so reduce the signal to noise
ratio in the eventual spectrum, thus reducing our ability to detect weak
signals. Unwanted NMR signals outside the pass band also get through,
and as we shall see this gives rise to problems. Unfortunately, there is
a stability limit to the sharpness of cut off of an analogue filter and we
can do no more than remove IF and some of the noise intensity outside
the pass band. There are, fortunately, ways around these problems but
before we describe them it is necessary to discuss the needs of the com-
puter and how this handles the input and what restraints it introduces.

5.5.3 Collection of data in the time domain

The output of the detector is a decaying wave which is entirely a func-
tion of time, and is said to exist in the time domain. In order to carry
out numerical processing of the output of an NMR spectrometer, we
need to convert the analogue electrical output into digital information
using an analogue-digital converter. This is characterized by the
maximum number of bits by which it can represent a voltage. Typically
this is 16 bits (though this can be varied at will) and the system gain
is arranged to avoid overflow of the converter output with the
maximum voltage present at the start of the FID. Sixteen bits gives
an enormous dynamic range for the input so that weak signals are also
well digitized. The signal is sampled at regular intervals, the voltage
registered converted into a binary number, and this number then
stored sequentially in one of a series of computer memory locations,
one location being a word of 32 bits (Fig. 5.11). The magnitude of the
sampling interval is very important, and is often known as the dwell
time, DW. The maximum frequency or minimum period that sampled
information stored in a computer memory can represent is limited to
that where alternate numbers are positive and negative, assuming a
cosine wave input varying symmetrically about zero. The period of
this wave is then 2(DW) and its frequency is 1/2(DW) = F. A lower
frequency, F-f, or a higher frequency, F + f , give the same pattern
in memory, so that there is an ambiguity in the digital representation
of the signals. This is avoided by ensuring that all the nuclear frequen-
cies lie in the range 0 to F Hz for single phase detection or ±F Hz for
quadrature detection. This is often known as the sweep width or as
the spectral width. The frequency F is known generally as the Nyquist
frequency and the low pass filter should cut off above this frequency.
Given that the spectrometer is highly stable, a series of isolated 90°
pulses will each give an almost identical nuclear response, so that these
can be added together in the computer memory to give a much

www.pdfgrip.com
142 The spectrometer

(a) Intensities
in memory
F
Nyquist
N, -A/, A/, -A/,
Frequency

(b)

F±F/2 A/,0,-A/,0,A/,0,-/V

-N r
Memory Location
Figure 5.11 (a) How a waveform at the Nyquist frequency, F, gives alternate
positive and negative values of the number N corresponding to peak voltage.
This assumes that the wave and the computer memory sweep are in a certain
timing relationship, (b) A waveform of lower frequency, F-F/2, gives the
same pattern as a waveform of frequency higher than the Nyquist frequency
by the same amount, F + F/2.

stronger total response. Each signal is, however, accompanied by

unwanted random noise; indeed, weak signals may not be visible
because of the noise. While the noise intensity also increases as more
responses are added, the noise is incoherent, i.e. its intensity at a given
memory location is sometimes + and sometimes -, and so it adds up
relatively slowly. A series of TV FIDs when added together in this way
have a signal-to-noise ratio that is vN times better than that of a single
FID. It is this feature that renders Fourier transform spectroscopy so
useful for the less receptive nuclei, allowing a FID to emerge from an
apparently hopeless morass of noise (Fig. 5.12).
We should also remember that the FID appears immediately fol-
lowing the transmitter pulse. In practice, the pulse breaks through with
the FID and distorts the first point, which gives considerable distortion
of the final spectrum. This is avoided by leaving a short delay between
the end of the pulse and the start of data collection so that the pulse
break-through has time to die out. This has to be allowed for in sub-
sequent treatment of the spectrum. It is a great disadvantage when the
FID is very short as important initial intensity is lost and in this case
special electronic circuitry has to be used to eliminate break-through.

5.5.3.1 Folding
There are two sorts of folding, or aliasing, of signals and noise. One
is due to the ambiguity described above where the computer cannot
distinguish between signals higher or lower than the Nyquist frequency.
This means that the spectrometer has to be set up with all nuclear

www.pdfgrip.com
 The detection system 143

1 acquisition

(a)

(b) **»,»•*. ^^LL~» I.... ..J .^J.,.-..i 11.-.,.^.

200 150 100 50 0

13
C/ppm

Sum of 200 acquisitions

(c)

(d)
200 150 100 50 6
13
C/ppm

Figure 5.12 (a) An 100.62 MHz 13C NMR FID obtained from a single pulse
applied to carvone in CDC13. (b) The resulting Fourier transform, (c) The
resulting FID after adding together 200 FIDs. (d) The resulting Fourier trans-
form of the sum of 200 FIDs.

signals appearing between 0 and the Nyquist limit, otherwise signals

above this limit appear out of their proper place. The high frequency
noise, of course, is always folded in and is only partially limited by
the analogue filters. The other type of folding occurs with the simple
phase-sensitive detector described above which cannot discriminate
between signals higher or lower in frequency than the reference
frequency. All nuclear signals have to be placed on one side or the
other of reference but noise is always collected from both sides, the
effective noise bandwidth being twice that of the filter. Fortunately,
both these types of folding can be avoided in a modern spectrometer.

5.5.3.2 Quadrature detection

Signals higher or lower in frequency than the Bl frequency fold into
the same spectral space with the phase-sensitive detector described

www.pdfgrip.com
144 The spectrometer

above. There is, however, a difference between them even if they are
equally separated from Bl in that the higher frequency component
reaches its first null beat with Bl earlier than the lower frequency
signal. In other words, they are phase shifted and it is possible to
discriminate between them using two phase-sensitive detectors (Fig.
5.13). This quadrature detector as it is called, then gives two outputs
with phase in quadrature which are collected separately in memory
and then Fourier transformed together so that all the signals appear
in their correct place relative to Bl and without phase anomalies. It
will be clear from Fig. 5.13 that this eliminates one kind of folding
and also allows us to place Bl in the centre of the spectral range. The
maximum frequency which we need now is only half that needed for
a single phase-sensitive detector for the same spectral range. This
means that the cut off frequency of the low pass filter can be halved
which in turn means that less noise is collected, halving the bandwidth
giving an increase in signal to noise ratio of \2 for an equal number
of FIDs. This is the normal method of operation.

I Reference] ^
| 90 ° Phase | low pass
I S™er I filters

»- PSD *-

(a) | Signal in

-CON
Single PSD noise Spectral range-y—[noise
(b)

,-''' 2 0 2 ^*,
i I I 1^
Quad. PSD noise]—»F Spectral range-y—[noise |
(c)

Figure 5.13 A quadrature detection system. The two boxes marked PSD are phase-sensitive detectors of
the type described in Fig. 5.9. The same reference is applied to both but one is made to be 90° out of
phase with the other. The two low frequency outputs are filtered and then converted to numbers and fed
sequentially to the computer memory. The lower part of the figure demonstrates the difference between
operating with a single or quadrature phase-sensitive detector. In the first case the spectrum range is 0 to
the Nyquist frequency -CON Noise folds in from both ends of the spectrum but is particularly troublesome
on the low frequency side because the low pass filter allows noise through without attenuation from
frequencies of 0 to +CON Hz. With the quadrature system the filter can cut off at half the frequency and
both the high and low frequency noise is attenuated. The dotted lines represent the fraction of signal with
a given frequency which pass through the filter into the receiver.

www.pdfgrip.com
 The detection system 145

A problem that occurs with the quadrature detector is that of quad-

rature images in which an intense signal to one side of the reference
frequency is reflected through to appear, albeit attenuated, on the other
side of reference. These artefacts are eliminated by phase cycling, a
process which involves changing the receiver phases and switching the
two outputs in concert between the two different computer locations
where the two FIDs are collected. Different phase cycling sequences
are used for different NMR experiments and it is useful to know what
is going on during a spectral accumulation even if it is all set up by the
computer. A commonly used sequence is called CYCLOPS.
We should perhaps note at this point that NMR spectroscopists put
considerable effort into creating these acronyms which help to give
the subject a human face. We will meet many such as we delve deeper!

5.5.3.3 Digital filtering

This still leaves the problem of folding around the Nyquist frequency
because of the slow cut off of the analogue filters. This situation is
improved by adding devices called digital filters which can be designed
to have nearly infinitely sharp cut off. The first step necessary to achiev-
ing this is to digitize the detector outputs not once every dwell time but
many times for each dwell period. This does not affect the Nyquist fre-
quency since this is determined by the intervals at which the computer
accepts data. If there are twenty points produced for each dwell period
then these are averaged using an appropriate weighting and a single
data point produced which forms part of the FID to be processed. This
operation can be carried out either in the computer memory, i.e. all
points collected then averaged or in a separate digital averaging device
before being transmitted to the computer. This technique first of all
cuts out the folded noise but also the averaging process reduces the
remaining noise further, the gain in signal to noise ratio being V (samples
per dwell period). A similar gain in dynamic ratio is also realized.
These digital filters are indeed so effective that they can be used to
select only a part of a spectrum and study it in isolation without any
folding from the rest of the non-observed resonances present. This
multiple sampling technique is known as over sampling.
Modern spectrometers use digital technology to the maximum, both
for the convenience of interaction with the computer and for the
increased stability of digital circuits which are much less disturbed by
drift in component values than are the equvalent analogue circuits.
Thus throughout, analogue signals are converted to digital signals by
an analogue to digital converter at the earliest possible moment, are
processed and then converted back to analogue form.

5.5.3.4 Phases of the pulses and receiver

The radiofrequency used in an NMR experiment is coherent. That
means that the sine wave oscillation of the electromagnetic radiation

www.pdfgrip.com
146 The spectrometer

continues without any discontinuity for ever. This is the basis of many
NMR experiments. This means that it is possible to phase shift the
frequency. In other words, shift the sine waves by any angle, frequently
90°, 180°, or 270°. In terms of the rotating frame, the maximum can
then be at y\ -Jt', or -yf instead of jc, see Fig. 5.8.
It is not only the transmitter that has variable phases. As a direct
consequence of quadrature detection, the receiver has phase encoding
also. The quadrature detector produces two outputs, 1 and 2. These
outputs are stored at two locations in the computer, A and B. The
way that the outputs are transferred to the computer provides a mech-
anism for receiver phase cycling. This is summarized in Table 5.1. For
example a y receiver phase means that output 2 is subtracted from
location A and output 1 is added to location B.
Extensive use is made of phase changing in producing the required
changes in magnetization. We will meet many examples of this in
Chapters 7 and 8. Phase is also cycled, i.e. changed systematically over
a group of 4, 8, 16, or 32 spectra in order to minimize artefacts. Even
the routine observation of an NMR spectrum involves the use of the
CYCLOPS phase cycle to remove images due to errors in the quad-
rature detection by averaging the response from imperfectly balanced
receivers.
The CYCLOPS phase cycle is given in Table 5.2. The phase of the
transmitter pulse is cycled as is that of the receiver. The result is to
considerably reduce the size of the quadrature image.

5.5.4 The analogue-digital converter and dynamic range

Sixteen bit analogue to digital converters or digitizers are typical but
they can be bigger or smaller. The speed at which they operate is
inversely proportional to the number of bits and a 16 bit device can
cope with a digitization rate of some 200 kHz. A sixteen bit digitizer

Table 5.1 The switching of the quadrature receiver outputs, 1 and 2,

between the two computer memory locations A and B

Receiver phase Location A Location B

x +1 +2
y -2 +1
-x -1 -2
-y +2 -1

Table 5.2 The CYCLOPS phase cycle for reducing quadrature images.

Acquisition Pulse phase Receiver phase

1 x x
2 y y
3 -x -x
4 -y -y

www.pdfgrip.com
 Production of the spectrum 147

produces a number within the range ±(215 -1) or ±65 535 with the
sixteenth bit being used to indicate the sign. This gives a very large
dynamic range, permitting weak signals to be detected in the presence
of very strong signals. However, this can be compromised by incom-
petent spectrometer operation. The receiver gain has to be set to use
most of this digitization range, but not so high that the signal ever lies
even slightly outside the range. For example, an analogue signal of
strength +65 536 turns on the sign bit as well and produces a number
which the computer reads as -65 535. The resulting distortion of the
FID produces a wave in the baseline in mild cases, and extra signals
in more extreme cases (Fig. 5.14).

5.6 PRODUCTION OF THE SPECTRUM

5.6.1 Exponential relaxation and line shape

The T2 relaxation process means that the nuclear frequency is not
precisely defined and that the spins can be thought of as having a
frequency distribution around their resonance frequency co0. A plot of
this frequency distribution against frequency is the line shape f(aj)
(Fig. 5.15). Individual nuclei have quite random frequencies within this
range, but over a short enough time interval we can regard them as
having this quite regular behaviour.
If we have an assembly of N nuclei, we can imagine that a small
proportion nk will have a particular angular velocity, cok. Each n is
given by the line shape function
n a
k f(«>k ~ wo)
which we can normalize by writing
nk= Nf(a>k-a>o)
i.e. we have chosen n0 so that f /(o)fc-a)0) = 1.

(b)

(a)
5.0 4.0 3.0 2.0 1.0 0.0
1
H/ppm

Figure 5.14 A 400 MHz !H NMR spectrum of ethanol in CDC13 recorded

with the receiver gain set too high, (a) Normal gain, (b) Increased amplitude
to show artefacts.

www.pdfgrip.com
148 The spectrometer

f («,( -CflQ )

Figure 5.15 A plot of the number of nuclei with angular frequency a)k different
from the resonant frequency o>0 for a small time interval t. The function nk =
f(a)k-a)^) is the line shape function.

At a small time t after the Bl pulse, those nuclei with angular velocity
a)k will have moved an angle 0^ from those nuclei exactly at resonance,
O)Q (Fig. 5.16). We can resolve the magnetic moments of the nk nuclei
along the WQ direction and at right angles to it, and, following stan-
dard alternating current theory, we shall differentiate the normal
component by the prefix i, where i = V-l. We have that
§k = ((Ok - 0)Q)t

and the two components are

nkcos[(a>k - o>0)£] and mksin[(a)k - c«>0)/]

cos 0k

sin 6k

Figure 5.16 How the total magnetic moment of the group of nuclei nk with
angular velocity a)k is displaced an angle 0^ from the on-resonance nuclei time
t after a Bl pulse has produced magnetization in the xy plane.

www.pdfgrip.com
 Production of the spectrum 149

The next step is to replace the nk by the line shape function.

However, Fig. 5.10 indicates that these may not be the same for the
two components, and so we shall denote them as two separate func-
tions by the letters v and u. The total intensity of the two components
is then proportional to
Nv(<x)k - (D0)cos[(a)k - a)o)t] and \Nu(a)k - w0)sin[(a>A: - o>0)(|
Summing over all possible values of k gives us the total intensity
of each component. We normally would wish to compare this with
the initial intensity at t = 0. This is proportional to the sum of all the
nuclei, N. Thus N cancels from the formulae. The intensity of the
normal component is, of course, only significant if o>0 * O)BI. If N is
large we can use the integral form of the equations, which, remem-
bering that a) is the variable and t is constant, gives
-00

v(o)£ - co0)cos[(<i>£ - a>0)?]dco = e~'/r2 (5.1)

o
and
f°° u(u> - a> )sin[(a>£ - o> )f]da> = e~r/r2 (5.2)
k 0 0
o
It remains to find the form of the functions v and u. Inspection of
tables of definite integrals will show that these are
M K-^o)^ 2 (53)
u
- i + r22K-co0)2 <"'
v=
i + r22K-<o0)2 (5 4)
'
These two functions are plotted in Fig. 5.17; u is the dispersion mode
and v is the absorption mode of the Lorentzian line shape. The absorp-
tion mode has its maximum at o>0 and its intensity is proportional to
the number of nuclei producing the signal. The dispersion mode is of
zero intensity at resonance and of different sign above and below reso-
nance. Spectra are displayed in the absorption mode, though, as we
have already seen, the dispersion mode has an important use in spec-
trometer locking systems.
The problem remains of how to obtain the plots of Fig. 5.17 from
the actual spectrometer output of Fig. 5.10. This latter is known as the
free induction decay (FID) and is stored digitally in computer memory
where it can easily be mathematically processed. Time and frequency
domains are related through the Fourier relationship

F(w) = f/(0e-io>'d/
—00

This can be written as

00

F(o)) = J /(r)[cos(wO-ism(<of)]df (5.5)

www.pdfgrip.com
150 The spectrometer

v mode absorption signal

C00

u mode dispersion signal

Figure 5.17 The absorption and dispersion mode signals available at the
output of the phase-sensitive detector

The Fourier transform of the output data thus contains two compo-
nents, often called real (R) and imaginary (I), which correspond to
the u and v components. The inverse transform is also possible, i.e.

/(0 = T-r^He-ia"d/
^IT -oo

and this should be compared with the relations (5.1) and (5.2) derived
above.

5.6.2 Time and frequency domains

The Fourier transform (FT) is a mathematical relationship that relates
a function of time to one of frequency, i.e. the time and frequency
domains. The output of an NMR spectrometer is a sinusoidal wave
that decays with time, varies entirely as a function of time and so exists
in the time domain. Its initial intensity is proportional to Mz and so
to the number of nuclei giving rise to the signal. Its frequency is a
measure of its chemical shift, and its rate of decay is related to T2 and
the quality of the magnetic field. Fourier transformation of this FID
gives a function whose intensity varies as a function of frequency and
is said to exist in the frequency domain. The parameters of the absorp-
tion curve contain all those of the FID: the position reproduces the
frequency and so the chemical shift. Equation (5.4) can be used to
predict the linewidth, which we define as the width at half-intensity.
The half-intensity points occur at frequency a) above and below reso-
nance co0, in other words when

www.pdfgrip.com
 Production of the spectrum

l + r 2 2 (o> 0 -a>) 2 = 2
i.e. when T22(a)0 - a))2 = 1
or
0)0-0) = ± —-
1
2

The separation of the half-intensity points is then twice this in

radians. It is, however, more usual to express linewidths in hertz
(frequency = c&ir), giving
<5 6)
"•«=A -
where v1/2 is the frequency separation of the half-height points. The
linewidth thus normally gives us l/TrT2*. The intensity of the absorp-
tion curve depends upon linewidth and, in fact, the product of
linewidth and peak height is proportional to the area under the curve,
which is proportional to the initial FID intensity and so to the nuclear
concentration. These relationships are summarized in Fig. 5.18.
In general, any function of time will have an equivalent represen-
tation in the frequency domain, and it is of interest to consider some
examples, which are illustrated in Figs 5.19, 5.20 and 5.21. We can see
that:
1. An infinitely long wave of constant intensity is represented by a
single monochromatic frequency. The precision with which we may
measure this frequency depends upon how long we are willing to
spend in its measurement.
2. Any distortion of this wave in the time domain results in the
production of extra frequency components in the frequency
domain. If the wave persists only for a limited time, then a packet
of frequencies is produced, which gives the line definite width in
the frequency domain. The line shape depends upon how the wave
decays in the time domain. If the decay follows the law
/, = /oexp{-(/iry} (5.7)
then if p = 1, we have an exponential decay as illustrated in Figs
5.19 and 5.20(a). If p = 2, we have a Gaussian decay; and if p - 4,
we have a super-Gaussian decay. These and their frequency-domain
equivalents are shown in Fig. 5.19.
3. If several frequency components are present in the time-domain
signal, then these interfere and produce a beat pattern such as
shown in Fig. 5.21. Each component gives a line in the frequency
domain with width determined by the rate of decay of the time-
domain signal. Note that when there are many, regularly spaced
components present, the time-domain signal starts to resemble a
series of short pulses.
This introduction of new frequencies with distortion of the wave is
a general phenomenon, and one way of stating the Fourier theorem

www.pdfgrip.com
152 The spectrometer

Figure 5.18 The relationships between time and frequency domains. The
period P of the FID gives the position of the line. The rate of decay T2* gives
the linewidth and the initial amplitude A gives the line its area and therefore
its intensity proportional to

is to say that any periodic function of time, however complex, may be

synthesized from a suitable combination of pure cosine waves.
One waveform that we will use consistently in producing spectra is
the rectangular pulse generating B^. Usually a chain of pulses is used
to produce a series of responses, which are added together. A single
rectangular pulse (Fig. 5.21) has a very rapid decay, which can be
represented as following the law of equation (5.7) with the exponent
p being very large. The flanking waves in the frequency domain are
now much more pronounced, and the intensity distribution follows a
sine curve, where sinc(;t) = sin(x)/x. The width of the central part of
the response is the inverse of the length of the pulse and, since NMR
spectrometers are designed to have 90° pulses of length generally less
than 100 JJLS, the width of the frequency coverage of the B1 field is at

www.pdfgrip.com
 Production of the spectrum 153

Figure 5.19 Equivalent time- and frequency-domain plots, the latter being
shown on the right of the figure. The lowest trace is a cosine wave of period
P, which gives a single infinitely narrow line at frequency IIP. (a) An expo-
nential decay It = /Oe~m, which gives a (m)2
Lorentzian line with sharp top but wide
skirts, (b) A Gaussian decay 7, = /Oe~ , giving a line with narrow skirts but
a thickened top. (c) A super-Gaussian decay /, = /Oe~(m)4, which introduces
flanking waves at the base of the peak. (From Akitt (1978) /. Magn. Reson.,
32, 311, with permission.)

least 10 kHz. Thus the use of short Bl pulses ensures that all the nuclei
in a sample, whatever their chemical shift, are swung around Bl by
the same angle. Intensity distortion only occurs if the chemical shift
range is very large. The same comments apply to a train of pulses,
except that the frequency coverage is now discontinuous. This has little
effect since the time between pulses will normally be longer than T2*
and the spacing between the peaks of energy will be less than a
linewidth. In all cases we will have present simultaneously in the
output, signals due to all the chemically different nuclei in a sample,
and the output will be complex and so require mathematical analysis
to obtain a spectrum.
It is also possible to produce a frequency-selective pulse. The spec-
trometer output has to be much reduced so that a 90° pulse has to be
very long. In such a case, the frequency spread of the pulse will be
only a few hertz and one can choose nuclei of a particular chemical
shift to precess around Bl without affecting any other nuclei in the

www.pdfgrip.com
154 The spectrometer

CO

ffl

CO

CO
Figure 5.20 Time- and frequency-domain equivalents of multiple-frequency
responses containing three, two, four and six components. The time-domain
signals for the lower three are relatively simple because the components have
been chosen to be of equal amplitude and regularly spaced. If the spacing
and amplitudes are irregular, the time-domain signal becomes much more
complex.

Figure 5.21 The frequency-domain equivalents of short, rectangular pulses.

The upper pair of traces show a single pulse of length t seconds whose
frequency is spread out over a range of lit Hz. The lower pair of traces show
a train of such pulses separated by an interval r seconds. In this case the
frequency-domain signal is not continuous, but consists of peaks of energy
separated by I/T Hz with the same spectral envelope as the single pulse.

www.pdfgrip.com
 Production of the spectrum 155

sample. Alternately, the pulse can be shaped to reduce its frequency

spread.
It should be clear from all this that there is an inverse relationship
between time intervals in the time domain and spread of frequency in
the frequency domain. The degree of resolution in the spectrum
(frequency domain) is thus determined by the time for which the
response is collected following the Bl pulse. The minimum linewidth
is determined by T2*, but if the time of collection is shorter than T2*
then this maximum degree of resolution cannot be attained.

5.6.3 Free induction decay, data size and zero filling

The time during which a single FID may be collected is limited by the
finite size of the computer memory. If there are M locations, then they
will all have been traversed after M(DW} seconds. Since, strictly, we
should wait until 5Tl seconds have elapsed and the system is again at
equilibrium before we apply the next pulse, we may also have to intro-
duce a waiting time. Thus a standard FT experiment can be summa-
rized by Fig. 5.22. The size of the memory does not affect the spectral
width. After transformation of the accumulated FID, the dispersion
and absorption parts of the spectrum each occupy Af/2 locations of
the same memory (this gives maximum utilization of memory space),
and the smallest frequency interval that can be detected is 2F/M Hz.
Memory size therefore determines the resolution. If lines are closer
than 2F/M they can never be separated. We note that 2FIM is equal
to 1/M(Z)W), the time for which we observe an individual response,
and that it is a fundamental fact that if we observe a response for only
t seconds then our resolution cannot be better than 1/t Hz. Any line
narrower than this limit will be represented by a single point in the
transform and its absorption intensity may be zero if its frequency falls
between those defined by two locations, though the broad wings of
the dispersion part will still be visible. Adequate computer memory
resolution is therefore essential for full definition of the spectrum. Each
line must be represented by several points and so the memory size
should be as large as necessary, which is not a limitation in view of
the low cost of computer memory.

Figure 5.22 The ideal Fourier transform experiment, which allows the spins
to relax to equilibrium before successive pulses are applied. The data are
collected until the memory is completely traversed and all memory locations
contain data.

www.pdfgrip.com
156 The spectrometer

In practice, the data size for the FID should be chosen so that
the FID is undetectable between 40 and 80% of the way through the
dataset being collected. If the FID finishes earlier, then the noise in
the remainder of the dataset will be converted into extra noise in the
final spectrum. Also if Tl = T2, time is being wasted during the noise
collection. Alternatively, if the FID continues beyond the end of the
collected data, resolution is being lost and care is necessary if zero
filling is going to be used, as described in the next paragraph, since
this will produce a step in the data.
The quality of the spectrum is affected by the number of data
points used for both acquisition and Fourier transformation. These
do not need to be the same as it is possible to increase the data size
after the end of accumulation of the FID by adding a block of zeros.
This is called zero filling. If an inadequate number of datapoints is
used for the acquisition of the FID, then resolution is lost; compare
Fig. 5.23(a) with 5.23(d). In the case shown in Fig. 5.23(a), the FID is
non-zero at the end of the data set and attempts to improve the
resolution by adding zero filled datapoints does result in a sharpening
of the signal, but also in ringing on either side of the signal (Fig.
5.23(b)). The result of increasing the number of datapoints used for
acquisition results in the FID being fully contained in the dataset, and
the resolution gives what at first sight is an acceptable spectrum
(Fig. 5.23(c)). However, after zero-filling, the peak shape is better
mapped, and more importantly the relative heights of the two
signals is now correct (Fig. 5.23(d)). A further increase in data size
does not improve resolution, but can improve the appearance of
spectra as peaks which were previously defined by a few data points
will now be defined by more and will have a smoother appearance,
the technique being equivalent to interpolating points in the frequency
domain.

5.7 RAPID MULTIPLE PULSING

The waiting time needed to allow the spin system to come to equi-
librium is a waste of our equipment since it is quiescent while waiting.
It was quickly found that the signal-to-noise ratio can be maximized
in a given time if we adopt the procedure: pulse - acquire data - pulse
- acquire data, and omit the waiting time. Thus we allow the computer
acquisition time to determine the pulse repetition rate. This time will
usually be much shorter than the ideal of Fig. 5.22, so that some trans-
verse magnetization will still exist at the time of each pulse after the
first. In fact, the later 90° pulses will tip the magnetization through
the xy plane and the intensity will be reduced. The signal strength
is then optimized by reducing the length of the pulse, and so the
pulse angle, until a maximum steady-state output is achieved.
The pulse lengths commonly found correct for this type of data
collection are in the range of 40° to 30°, depending on nuclear isotope

www.pdfgrip.com
 Rapid multiple pulsing 157

(a) (b)

(c) (d)

Figure 5.23 Examples of the effect of zero-filling. In each case only one doublet from a 162 MHz 31P AX
NMR spectrum is shown. A spectral width of 10 204 Hz was used, (a) A 16K FID transformed into a 8K
real spectrum. The angular appearance of the spectrum is due to the digitization interval being 1.3 Hz,
while the separation of the doublet is 5 Hz. (b) A 16K FID, increased to 128K by adding zeros, trans-
formed into a 64K real spectrum. The extra digitization has resulted in a reduction of the linewidth, but
the FID had not gone to zero by the end of the dataset and the result is that there is ringing on either
side of the signals, (c) A 64K FID transformed into a 32K real spectrum. This is much better, but due to
the lack of digitization, the relative heights of the two signals are wrong, (d) A 64K FID, increased to
128K by adding zeros, transformed into a 64K real spectrum. This spectrum finally gives true heights and
linewidths to the signal.

observed, chemical shift range, relaxation time Tl and memory size.

These parameters are related by the Ernst equation
f T\
cos(a0) = exp --f
\ M/

where a0 is the pulse angle (determined by its length) and Tp is the

time between pulses. Now, the relaxation times of nuclei situated in
different parts of the molecule will usually be different, so that the

www.pdfgrip.com
158 The spectrometer

optimum pulse length will be different for each. This means that there
will be a distortion of the intensity of the signal from each type of
nucleus, which fortunately is not too grave a disadvantage in many
cases since we know a molecule is made up of integral numbers of
each type of atom. If precise quantitative data are essential, then the
pulse sequence of Fig. 5.22 must be used or a means found of reducing
Tl in our sample.

5.8 MANIPULATION OF COLLECTED DATA

FT NMR involves the collection of data in a computer memory. Now

a computer is an infinitely variable instrument; it will do anything that
you wish to the data that it contains, provided that you possess a suit-
able program. The computers used for NMR spectroscopy have a
section of memory containing programs as well as memory used simply
for data accumulation, together with a backing store in the form of a
disc or discs. Thus a set of data that may have taken several hours
to acquire can be stored in permanent form and recalled to allow a
variety of mathematical processes to be carried out with the object of
improving the final spectrum.

5.8.1 Line-shape manipulation

5.8.1.1 Exponential multiplication

The technique most commonly used in the time domain is to multiply
the data progressively by an exponentially decaying function,
k = e-'/7V
where t is the time corresponding to each point in the FID dimension,
and T^ is a constant which behaves the same as the spin-spin relax-
ation time. Hence if a signal has a T2 of 1 s, then this is a good value
to use for T^ in exponential multiplication. In practice, it is usual to
express T^ as a line broadening in Hz, which is related to Tj as in
equation (4.1). This treatment increases the rate of decay of the FID
and so broadens the lines obtained after Fourier transformation. This
loss of resolution is normally acceptable because the signal-to-noise
ratio is significantly improved. This comes about because the signal
and noise have different shapes in the time domain; the signal is intense
at first but decreases with time and may be undetectable towards the
end of the FID, whereas the noise has no decay in intensity with time
and may be the only signal present at the end of the decay. The expo-
nential multiplication thus decreases the noise most where there is
no signal to be attenuated. The effect of the process when applied to
a noisy spectrum is shown in Fig. 5.24. It allows the signal-to-noise
ratio to be increased in a time that is very much shorter than that
needed to accumulate sufficient extra data. The amount of noise that

www.pdfgrip.com
 Manipulation of collected data

(d).

(0.

(b)

(a)

60 50 40 30 20
13
C/ppm

Figure 5.24 The effect of the exponential multiplication process on the 100.62 MHz 13C NMR signal of
bromocyclohexane in (CD3)2CO at -16°C. (a) A transformed spectrum where no window function has
been applied, (b). A transformed spectrum which has been broadened by multiplying by a decaying
exponential with a time constant corresponding to a broadening of 1 Hz. (c) As b, but 4 Hz broadening,
(d) As b, but 10 Hz broadening. Note that as the broadening is increased, the broad signals at 8 26.5, 38,
and 53 become more clearly seen, but the coupling on the CD3 signal of the solvent at 8 20.4 becomes
more and more obscured.

www.pdfgrip.com
160 The spectrometer

is accumulated with a spectrum is determined by the rate at which

data are accumulated, and there is relatively little flexibility in spec-
trometer setting possible. Thus, after a given time of accumulation,
one has a certain signal-to-noise ratio, and this can be improved either
by continuing with accumulation of data or by exponential multipli-
cation if the loss in resolution is acceptable. If it is decided to continue
accumulation of data, it is necessary to continue for at least as long
as has already been done to obtain a significant improvement. This is
a simple rule of thumb and is reasonable in view of the square-root
law for signal-to-noise improvement, which predicts a 40% improve-
ment if the number of accumulations is doubled. In contrast, expo-
nential multiplication is almost instantaneous.

5.8.1.2 Lorentzian-Gaussian transformation

It would often be useful to be able to sharpen the lines and artificially
improve the resolution. The use of multiplication by the rising expo-
nential k = e~'/r*2 with a negative value for T*2 does work, but the
signal-to-noise ratio is badly degraded. The term becomes very large
at large t values, i.e. at the right hand side of the FID where there
is predominantly noise. The generally preferred function is the
Lorentzian-Gaussian transformation equation
k = e-'/r*2e-*'2 (5.7)
The function initially increases if a negative value is chosen for T2, but
if a positive value is chosen for b, the function then decreases to
give a bell shape. FIDs before and after multiplication are shown in
Figs 5.25(a)-(d). The multiplication transforms the lineshape from that
of the ideal Lorentzian NMR signal to a Gaussian lineshape and
has the effect of sharpening the signal and reducing the linewidth.
This can be seen by comparing the multiplets in Figs 5.25(a) and (c)
or (d).
A more extreme example is the determination of the long-range
coupling between the protons in 2-chlorobenzaldehyde, a study of
which permits the study of the barriers to rotation of the formyl group.
Figure 5.26 shows two of the lines due to the proton H4 obtained with
an instrumental resolution better than 0.05 Hz. Structure is visible in
one of the lines, and after resolution enhancement the small coupling
6
/(F-H4) is seen with a value of about 0.024 Hz. The variation of this
very small coupling between the formyl proton and ring proton H4
with temperature enables the formyl group rotation to be studied. The
wiggles in the baseline in both these spectra arise because of the rapid
truncation of the data that occurs in the tail of the FID with the appli-
cation of the noise-reducing function and which introduces a super-
Gaussian element into the spectrum (Fig. 5.7(c)). For this reason,
resolution enhancement is also known as a Lorentzian-Gaussian trans-
formation. The technique thus has to be used with care, since these
wiggles can in extreme cases be mistaken for hidden peaks.

www.pdfgrip.com
 Manipulation of collected data 161

(d).

(0-

(b)_

(a)-
2.70 2.60 2.50 2.40 2.30
1
H/ppm

Figure 5.25 The influence of using Gaussian resolution enhancement on a *H NMR spectrum using equa-
tion (5.6). (a) No enhancement, (b) A T2 of -0.637 s, and b = 0.157. (c) A T2 of -0.318 s, and b = 0.076.
(d) A T2 of -0.59 s, and b = 0.039.

5.8.1.3 The sine-bell weighting functions

Although the exponential and Lorentzian-Gaussian transformation
weighting functions are commonly used in one-dimensional NMR spec-
troscopy, in two-dimensional NMR spectroscopy, see Chapter 9, the
sine-bell function is frequently used. It is common in two-dimensional
NMR spectroscopy to have truncated FIDs. With the simple sine-bell
function, the FID is multiplied by a sine wave going from sin(0°) to
sin(180°). There are variations, where the sine bell is shifted, and, for
instance, a shift of 30° simply means that the multiplying sine-bell starts
at sin(30°) and runs to sin(180°). Also a sine-bell squared function is
used where the multiplying sine function is squared.

www.pdfgrip.com
162 The spectrometer

Figure 5.26 The !H spectrum of a fragment of the spectrum of 2-chloroben-

zaldehyde showing how resolution enhancement can reveal the very small
coupling between the formyl proton and ring proton H4. (From Laatikainen
et al (1990) Magn. Reson. Chem., 28, 939-46; copyright (1990) John Wiley
and Sons Ltd, reprinted with permission.)

5.8.2 The Fourier transform

The production of the frequency-domain spectrum from the time
domain FID is done by Fourier transformation. Equation (5.5) for the
relationship between time and frequency domains is
f°°
F(o>) = I /(f)[cos(o>f) - i sin((oO]d£

It can be shown that it is valid to replace the continuous functions

by discrete ones such as we collect in a computer, i.e. we can replace
the integral symbols by summations and can carry out valid transforms
in a computer. The process of Fourier transformation can be thought
of as carrying out a series of multiplications of the time-domain data
by sine and cosine functions that are stepped along the data. The
results of this series of multiplications are summed and, if there is
present in the data a wave of the same period, then there is an output;
otherwise there is none. The use of both sine and cosine functions
means that intensity and phase information are retained. The process
is repeated for all possible frequencies up to the Nyquist frequency,
so that all the signals present in the data may be detected. This would
obviously be a very time-consuming process, even in a modern
computer, and would require a vast amount of extra computer memory
in order to store all the intermediate data needed. Fortunately, Cooley

www.pdfgrip.com
 Manipulation of collected data 163

and Tukey have shown that, provided all the data are contained in a
memory with number of locations equal to a power of 2 (e.g. 1024,
2048, 4096,..., etc., or IK, 2K, 4K,..., etc., locations), then the process
can be factorized in such a way as to remove many redundant multi-
plications and can be carried out in the memory containing the data,
with only a few extra locations being needed to store numbers
temporarily. The process is thus fast and economical of memory space.
The existence of the Cooley-Tukey algorithm and of minicomputers
were both necessary before FT NMR could become a commercial
possibility. In fact, the current generation of dedicated computers are
so advanced that they can perform a transform on accumulated data
in another part of memory in order to observe how the accumulation
is proceeding while at the same time more data are being added to
the FID. The sine and cosine multiplications produce, in effect, the
two components of the nuclear magnetization in phase with and
normal to Bl9 and these are kept separate in the memory. The final
result of the transform is thus two spectra, which should in principle
be the absorption and dispersion spectra. This assumes that the Bl
signal fed to the phase-sensitive detector is exactly in phase with the
nuclear signal, a condition that it is very hard to realize in practice.
Thus the two sets of spectra contain a mixture of dispersion and
absorption. In addition, we cannot be sure that we have started to
collect data immediately after the Bl pulse has ended; indeed, to avoid
breakthrough of the strong pulse signal, we may have purposely
delayed the collection of data, commonly by one dwell time. This
means that the starting point for each frequency component differs,
since each type of nuclear magnet will have precessed by a different
amount by the time data collection starts, their phases are different
and the degree of admixture of the two spectra varies across the spec-
tral width. The two components thus have to be separated, a proce-
dure known as phase correction, which is an essential part of FT
spectroscopy. The two halves of the sets of spectra are weighted and
mixed according to the formulae below.
new disp = (old disp) cos0 - (old abs) sinO
new abs = (old abs) cosO + (old disp) sinO
where 'abs' means one half of the data and 'disp' the other. The multi-
plication is carried out point by point through the data and 0 is adjusted
until one line is correctly phased, i.e. purely absorption in one half of
memory and purely dispersion in the other half. If not all the lines
are then correctly phased, it is necessary to repeat the correction but
allow 0 to vary linearly as a function of position in memory. It is best
to arrange that 0 = 0 for the correctly phased line and allow it to vary
to either side. This second process is then continued until all the lines
are correctly phased. Phase correction can be done automatically by
the computer, though the operator may intervene if needed. The
process is illustrated in Fig. 5.27.

www.pdfgrip.com
164 The spectrometer

real imaginary

before
phasing

after
phasing

Figure 5.27 Phase correction. The upper pair of spectra are the two results
of the transform operation with scrambled phase of the lines of the AB2 spec-
trum. A two-parameter phase correction gives the purely absorption spectrum
in one half and the purely dispersion spectrum in the other.

Sometimes it is not possible to make a perfect phase correction. This

may arise because the behaviour of the spin system is affected by the
way the spectrum has been obtained; we have already mentioned that
too rapid pulsing if spin-spin coupling is present may result in distor-
tion of line intensities, and in such a case phase anomalies are also
inevitable.

5.8.3 Manipulations in the frequency domain

Once we have the fully phase-corrected spectrum in memory, we now
have to extract from it the information that we require. We will also
need to make some sort of permanent record. Again, the computer
allows us to carry out a variety of processes.

5.8.3.1 Visualization and plotting data

The contents of the data memory, whether the data are time or
frequency domain, can be displayed continuously on a monitor screen

www.pdfgrip.com
 Manipulation of collected data 165

connected to the computer. The X axis is swept through the data

memory and the numbers stored there are transferred sequentially to
the Y axis. The result is a steady display of FID or spectrum, which
allows the various manipulations to be monitored. A more permanent
record is provided by plotting the data using some sort of printer which
can print what is on the screen and much other data stored in the
computer for each spectrum. Different sets of spectral data can be
drawn in different colours, together with expansions or integrals (see
below), and they can also reproduce text and so calibrate the paper,
give spot chemical shifts for individual lines and numerical integral
values.

5.8.3.2 Integration
The initial intensity of a FID is proportional to the number of nuclei
contributing to that signal and this transforms as the area of the
Lorentzian absorption. An integral of the spectrum (we always, of
course, refer to the absorption spectrum) then will tell us how many
nuclei contribute to a given line and can give us invaluable quantitive
data about a molecule - in the absence of relaxation effects discussed
in Chapter 4. The integration is carried out simply by adding the
numbers in successive memory locations. Where there is no resonance,
this sum will remain constant, but will increase in the region of any
resonance. The integral then forms a series of steps, rising at inter-
vals, with each rise corresponding to a resonance, see Fig. 5.28. For
the plot to be vertical and horizontal as shown, it is necessary that the
baseline is at zero intensity. Means are thus provided to correct the
baseline to give the most acceptable integral.

7.6 7.4 7.2 7.0 6.8 6.6 6.4 6.2 6.0 5.8 5.6
1
H/ppm

Figure 5.28 A 400 MHz *H NMR spectrum and its integral of 2,6-
dichlorophenol in CDC13. The ratio of the heights of the steps is approxi-
mately 2 : 1 : 1 as required for H3,H5:H4:OH.

www.pdfgrip.com
166 The spectrometer

5.8.3.3 Expansion
The typical monitor screen is rather small to be able to observe much
detail, and means are provided to enable a part of the spectrum to be
selected and expanded to fill the screen to assist in the various oper-
ations needed to improve a spectrum and, for instance, show up small
splittings.

5.8.3.4 Storing data

Data can also be recorded on some form of magnetic storage system,
usually a hard disc, at least for a limited period. It is usual to store
the FID, which can then later be recalled and reprocessed to, say,
show up some unexpected feature. Archiving is now possible using
that contradiction in terms a writable CD ROM or magnetic tape.

5.9 QUESTIONS

5.1. It is required to run a series of spectral accumulations for a

nucleus and a group of its compounds where the range of chem-
ical shifts is likely to be 150 ppm. In order to accommodate all
resonances without folding, a spectral width is chosen of 200 ppm.
The spectrometer frequency is 90 MHz. What dwell time should
be used? The longest relaxation time that is likely to be encoun-
tered is 20 s. A resolution of about 1 Hz is required. Calculate
the appropriate memory size to be used to accumulate data
(IK, 2K, 4K, 8K, 16K, 32K, 64K, ..., where IK = 1024 are the
permitted memory sizes) and the time required to traverse
the memory and collect each FID following the read pulse. Will
the nuclei be fully relaxed when the memory has been traversed
and should the pulse length of the read pulse be 90° or less?
What is the maximum permissible length of the stimulating read
pulse if the nuclei are to be turned towards the xy plane by an
equal amount whatever their chemical shifts?
5.2. At time 0, the z-magnetization is Mz of a nucleus with 7\ = 1 s.
You apply a 90° pulse. What is now the value of Mxy in terms of
Mz? You now wait 1 s. What is now the value of Mz?
5.3. The lock operates on a very precise frequency to ensure exact
calibration of the spectra. What happens if the phase is set incor-
rectly so that the dispersion signal is lop-sided?
5.4. What is the gain in signal-to-noise ratio for a quadrature detector
operating with a 16 sample digital filter compared to a simple
phase sensitive detector operating with analogue filters?
5.5. Use the Ernst equation to calculate the ideal pulse angle for
maximum signal: noise ratio for a nucleus with 7\ = 1 s and a
time of 1 s between pulses.

www.pdfgrip.com
 Making the spins dance ®
We are now going to discuss how we can predetermine the behaviour
of the nuclear magnetization using pulses of various lengths and phase
and carry out a variety of NMR experiments.

6.1 DECOUPLING

If we consider a nuclear system AnXm in which all nuclei A are equiv-

alent as are all nuclei X, and the two types of nuclei are spin-coupled,
then the resonances of X and A will both be split into multiplets whose
multiplicity depends upon the values of n and m and the spin quantum
numbers of A and X. A and X may be the same isotope and so chem-
ically shifted, preferably by a substantial amount, or they may be
different isotopes and so with markedly different NMR frequencies.
We will observe A while applying a second strong radiofrequency field
B2 to X, remembering that in the homonuclear case this may require
some technical modifications of the experiment. B2 has to have the
same frequency as the centre of the X multiplet and so is stationary,
or nearly so, in the rotating frame relative to these nuclei. The X nuclei
then precess around B2 at a frequency that depends upon the magni-
tude of B2, and, if this precession is fast enough, rapidly and repeat-
edly reverse their direction so that their z magnetization effectively
disappears at the coupled A nucleus whose multiplicity disappears and
so becomes a singlet. A is said to be decoupled from X. For decou-
pling to be complete, it is necessary that
^»/(AX)
ZTT

An example of complete decoupling is shown in Fig. 6.1. In fact,

there will always be a little residual coupling and a decoupled singlet
is thus slightly broadened because of this. If B2 is small, then the A
spectra can become more complex and exhibit extra splittings.
Signals close to the B2 irradiation frequency are displaced from their
true frequencies. This is called the Bloch-Siegert shift. The Bloch-
Siegert shift obeys the equation

fM2(v A -v 2 )(v obs -v A )

www.pdfgrip.com
168 Making the spins dance

1 I I I I I
3.,30 3.25 3.30 3 .25 3.,30 3.25
1H/ppm 1H/ppm 1H/ppm

Figure 6.1 The 1H NMR spectrum31 of P(OCH3)3. (a) Normal, with 31P
coupling, (b) With irradiation at the 31 P frequency to completely decouple the
phosphorus, (c) With low-powered P decoupling.

where B2 is decoupler field strength, VA is the position of the signal

in the absence of decoupling, vobs is the position of the signal in the
presence of decoupling, and v2 is the decoupler frequency. This equa-
tion applies provided that (yB2)2 « (VA - v2)2. This equation can there-
fore be used to calibrate the strength of the homonuclear B2 by
carrying out irradiation near to a resonance.

6.1.1 Homonuclear decoupling

For homonuclear decoupling, A and X are the same nuclear isotope.
We give an example of the decoupling of two proton resonances
(Fig. 6.2). The chemical shift between the two coupled resonances of
2,6-dichlorophenol is some ten times larger than the coupling constant.
It is thus easy to apply a B2 strong enough to perturb the nuclei of
one resonance without directly affecting the other. Thus by irradiating

www.pdfgrip.com
 Decoupling 169

(0-

(b)

(a):
7.6 7.4 7.2 7.0
16.8 6.6 6.4 6.2 6.0 5.8 5.6

Figure 6.2 An example of homonuclear double irradiation, (a) The normal 400 MHz !H NMR spectrum
of 2,6-dichlorophenol in CDC13. (b) As 4a, but with B2 placed on the doublet due to H3 and H5 at 8 7.24.
Note that the triplet at 8 6.80 due to H has been reduced to a slightly broadened singlet, (c) As a, but
with B2 placed on the triplet due to H4 at 8 6.80. Note that the doublet at 8 7.24 due to H3 and H5 has
been reduced to a slightly broadened singlet. There is also a decoupling spike at 8 6.80 due to leakage of
the decoupling frequency into the receiver. The OH proton at 8 5.89 is not affected.

one, the other becomes a singlet. Decoupling is established in such an

experiment very quickly since precession around B2 starts as soon as
B2 is applied.
At the same time as the precession of individual nuclei takes place
under the influence of B2, there is saturation of the X spin system.
The word 'saturation' derives from usage in the old continuous wave
NMR spectroscopy, where it is found that, if the irradiating power is
too large, the intensity of the signal decreases. The spin system absorbs
energy from the irradiating field so that the low-energy excess of
spins becomes depleted, a process that is in competition with the
Tl relaxation processes. This also happens for B2, and if this is
large enough then the spin populations become equalized and the total
magnetization of the irradiated resonances become zero. This is a
non-equilibrium state and all the relaxation mechanisms present will
work towards the re-establishment of equilibrium. If the predominant
mechanism of relaxation is dipole-dipole, then the energy exchange
required by this relaxation causes mutual spin flips of A and X,
and the spin population of the A nuclei is disturbed, so that decou-
pling between / = 1/2 nuclei is usually accompanied by intensity

www.pdfgrip.com
170 Making the spins dance

modifications of the signal of the decoupled nucleus. This is the nuclear

Overhauser effect (NOE). Because relaxation processes are involved
in determining the NOE, this is established much more slowly than
the decoupling.
In order to decouple, the decoupler must be on during the acquisi-
tion of the FID. The introduction of the B2 homonuclear decoupling
field is likely to cause interference with the detection of the FID. This
can be avoided by using time-shared decoupling in which B2 is only
switched on while the receiver is resting between analogue-to-digital
conversions. This mode of operation is made possible because the
receiver is only required to operate for sufficient time to activate the
analogue-digital converter at the end of each dwell time. The inter-
ruptions in B2 need then be only quite a small proportion of the total
time, see Fig. 6.3. This is the case if the sampling is made every dwell
time though if over sampling is used and the digitizer is working at
its maximum rate then the transmitter can be on for only about 10%
of the time.

6.1.1.1 Single frequency heteronuclear decoupling

Heteronuclear decoupling occurs when A and X are different elements
or isotopes (Fig. 6.4). The NH2 group of formamide, H2NCHO, consists
of two signals due to restricted rotation about the N-C bond. This is
just observable in the simple :H NMR spectrum but there is exten-
sive broadening of the NH2 signals by coupling with the quadrupolar
14
N. The CH proton is a doublet due to coupling to one of the NH
protons. When the 14N is decoupled, the two separate NH signals are
well resolved at 8 6.85 and 7.1 and the coupling to the CH proton is
also evident, as is the much smaller coupling to the other NH and the
small coupling between the two NH protons.
In more complex molecules, there may be two or more types of
X nuclei coupled to the A nuclei, and, provided the chemical shift

Decoupler On
RXon
.A/D
one dwell time interval i -Time
Figure 6.3 Time-shared homonuclear decoupling. Three parts of the spec-
trometer have to be switched on and off independently though in a particu-
lar order. The receiver alone is switched on for sufficient time to establish an
output. This output is fed to the analogue-to-digital (A/D) converter, which
starts to convert the voltage into a number at a precise time, set by the dwell
time in use, and takes a finite time to make the conversion, 5 to 20 JJLS. The
decoupler is switched on during the A/D conversion process and remains on
until a little while before the receiver is switched on again. The effect of the
decoupler may be modified by reducing the length of each transmitter pulse
(decoupler on) as required.

www.pdfgrip.com
 Decoupling 171

(b)

(a).

8.50 8.25 8.00 7.75 7.50 7.25 7.00 6.75 6.50 6.25
1
H/ppm

Figure 6.4 The 400 MHz 1H spectrum of formamide, H2NCHO, in CD3CN.

(a) The proton resonances of the NH2 group are badly broadened by the 14N
nucleus and the CH resonance is slightly broadened, (b) The signals sharpen
on irradiation of the 14N nucleus.

between the types of X nuclei is larger than any coupling between

them, then it is possible to carry out selective decoupling experiments
in which the effects of each group of X nuclei can be examined in
turn. A heteronuclear example is that of the 31P spectrum of triethyl
phosphite, (CH3CH2O)3P, whose spectra are shown in Fig. 6.5. The
phosphorus nucleus is coupled to both types of proton, that to the
methylene protons giving a septet and that to the methyl protons giving
a decet. The coupling to the methyl protons is, however, quite small,
and the normal spectrum is a septet of broad lines that contain the
decet structure due to the coupling to the methyl group. This broad-
ening reduces the intensity of the resonances, to the extent that it is
difficult to observe the two weak outer lines of the septet (Fig. 6.5(a)).
Because the proton resonances of the ethyl and ethylene groups are
separated by much more than the interproton coupling constant, it is
possible to irradiate one with B2 sufficient to eliminate the coupling
to the 31P without affecting the coupling of the other type of protons.
One effectively eliminates one of the coupling interactions, observes
what the other produces on its own and then irradiates the other group
to observe the effect of the first alone. The 1H NMR spectrum of the
P(OCCH3)3 fragment is in Fig. 6.5(c) and the well resolved methylene
coupling in Fig. 6.5(b). It is, of course, possible in principle to apply
sufficient B2 power that both types of proton are decoupled from the

www.pdfgrip.com
172 Making the spins dance

(C)-

31
Figure 6.5 The 162 MHz :
P spectrum of triethyl phosphite, (CH3CH2O)3P.
(a) Obtained without H irradiation, showing coupling to the methylene
protons broadened by the coupling to the methyl protons, (b) The methyl
protons irradiated, which gives a well resolved septet due to coupling to the
methylene protons, (c) The methylene protons irradiated, which selectively
removes their influence from the spectrum.

phosphorus, whose resonance then becomes a singlet. In practice, it

proves technically difficult to provide sufficient power to do this, and
means have to be found to overcome this difficulty. The use of two
separate B2 frequencies at the same time is possible, but it has been
found that the best method is to ensure a broad band of frequencies
covering, in this case, the proton chemical shift range. This used to be
achieved by frequency modulation of the B2 frequency with a randomly
varying waveform but now sequences of short B2 pulses, such as
WALTZ, see section 6.1.1.4, are used to swing the spins around. This
gives such a spread of frequencies, that the power requirements are
much less stringent. The technique is called broad-band decoupling
and is used extensively, indeed almost invariably, when observing the
nucleus 13C in hydrogen-containing compounds. The hydrogen is
broad-band irradiated and this removes all spin coupling to the carbon
atoms, so simplifying their resonances to narrow singlets, increasing
the intensity of the resonance of a given type of carbon atom and so
reducing markedly the time necessary to obtain a given signal-to-noise
ratio with this insensitive nucleus. In addition, there is also a valuable
intensity enhancement due to the nuclear Overhauser effect, which
combined with the simplification of the spectra gives a time reduction
factor of the order of 250 times compared with a spectrum obtained
without double irradiation.

www.pdfgrip.com
 Decoupling 173

6.1.1.2 Off-resonance decoupling

Single frequency decoupling works well when it can be placed on the
frequency of the nucleus to be irradiated, but it does not produce satis-
factory results when placed off resonance. For example, the 100 MHz
13
C NMR spectrum of carvone is shown in Fig. 6.6(c) and with the 1H
decoupling frequency placed at 8 10, Fig. 6.6(b). This has the effect of
reducing the small 2J(13C1H) and 3/(13C?H) to a negligible size, while
the multiplicity due to the larger 1J(13C1H) can still be resolved. This
technique used to be used to decide how many protons are attached
directly to a given carbon atom and worked well on low field instru-
ments, where the 1H spectral width was less than 1000 Hz, but on high
field instruments with spectral widths of up to 8000 Hz, the !H decou-
pling frequency is too far from many of the protons to produce a
substantial reduction in 2'3/(13C1H). The consequence is that this tech-
nique is now rarely used. It has been replaced by other techniques
such as /-modulation, APT, PENDANT, INEPT and DEPT, see
Chapter 8.
The shrinkage of the apparent /(X!H) can be used to calibrate the
B2 for !H decoupling of an X nucleus. Provided that yB2/2TT»
I VA - v2 I the equation (overleaf)

(c)-

(b)-

(a)
i
200 100 0
8/ppm

Figure 6.6 The 100.62 MHz 13C NMR spectrum of carvone in CDC13.
(a) Obtained with WALTZ decoupling,
13
(b) Obtained with the !H decoupling
frequency placed at 8 10. The C nuclei attached to high frequency protons
near 810 show big reductions in 1J(13C1H). (c) Obtained without *H
decoupling.

www.pdfgrip.com
174 Making the spins dance

•y/? 2 _/(X 1 H)lv A -v 2 l

2ir /(X1H)R
applies, where /(X!H) is the coupling in the absence of :H decoupling,
/(X1!!^ is the coupling in the presence of !H decoupling, VA is the
position of the 1H NMR signal in the absence of decoupling, and v2
is the decoupler frequency.

6.1.1.3 Broadband heteronuclear decoupling

In order to be able to decouple all the protons from an X-nucleus,
the single 1H frequency was replaced by a band of frequencies. This
technique works, but suffers from problems. The power required is
relatively high and typically 10 watts were required to produce mod-
erate decoupling. This produced considerable warming of the sample.
As chemical shifts are temperature dependent, the signals were broad-
ened by the temperature gradients and the signal-to-noise ratio was
reduced by the poorer Boltzmann population difference at the higher
temperatures. Secondly, as magnetic fields increased, and the !H
frequency range increased, it became more difficult to decouple uni-
formly across the whole 1H spectral width. This technique has now been
abandoned in favour of the more efficient composite pulse decoupling
techniques such as WALTZ and GARP, which we describe below.

6.1.1.4 WALTZ
In order to have some appreciation of how composite pulse decou-
pling works, let us examine the effect of applying a sequence of 180°
pulses to X H while observing 13C of a compound such as CHC13. In
the absence of !H pulses or decoupling, the 13C NMR signal will consist
of a doublet, due to coupling to 1H. The doublet arises because half
the 13C nuclei are attached to 1H nuclei with mI = 1/2 and the other
half are attached to 1H nuclei with ml = -1/2. If we now apply a 180°
pulse to XH, then we reverse the ml values of the !H nuclei. The 13C
nuclei which were attached to *H nuclei with ml = 1/2 are now attached
to !H nuclei with ml = -1/2 and vice versa. If we were to apply a
sequence of 180° pulses, each 13C nucleus would see a !H nucleus
which was rapidly reversing its spin. The result is XH decoupling.
The experiment described above is inefficient and composite pulses
are preferred, see section 6.2. Generally, the WALTZ-16 pulse
sequence is used. It is built up of composite pulses based on the
sequence (90°)JC(180°)_JC(270°);c. This is given the shorthand notation
1 "2T 3, with 4 being used for a (360),, pulse. The numbers represent
multiples of 90° and the bar over the 2 represents a reversal of phase
to produce a 180° phase shift. The shorthand notation leads to the
name WALTZ as the dance is based on a 1-2-3 step sequence. There
are several variations depending on the length of the sequence used,
but it is generally the WALTZ-16 sequence which is preferred.

www.pdfgrip.com
 Decoupling 175

WALTZ-16 - 3 4_2 3_1 2_4 2_3 3_4 2_3 1_2 4 2 3 3 4 2 3

1 2 4 2 3 3 4 2 3 1 2 4 3 2
The WALTZ pulse sequence is repeated continuously to produce
the decoupling. It decouples efficiently over a width of ± 4 kHz
reducing a 150 Hz /(^CH) down to 0.15 Hz. This bandwidth is suffi-
cient to produce a lOppm decoupling bandwidth even at 800MHz.
Typically the !H decoupler is attenuated so that the 90° pulse is around
100 JJLS. The power used is much lower than the older broadband
decoupling, and consequently problems associated with sample heating
have been considerably reduced.
There are circumstances where the bandwidth of the WALTZ pulse
sequence is insufficient. The most common example is where 13C is
decoupled from 1H in inverse 13C detection, see section 9.7. Under
such circumstances, the GARP composite pulse decoupling sequence
is preferred.

6.1.1.5 GARP
In order to increase the bandwidth of the decoupling, it is necessary
to use flip angles which are not multiples of 90°. This has the effect
of broadening the bandwidth, but makes the decoupling efficiency less.
The bandwidth doubles, but the residual splitting increases to around
0.3 Hz. The GARP-1 pulse sequence is based on the pulse sequence
RR R 7F, where in terms of pulse angles
R= 30.5 352 257.8 2683 69.3 622 85.0 ~9L8 134.5 25O
66.4 ~45^ 25.5 72J 119.5 1382 258.4 "649 70.9 772
98.2 133.6 255.9 65.5 53.4
Typically the X decoupler is attenuated so that the 90° pulse is
around 70 JJLS. The acronym GARP comes from Globally optimized
Alternating-phase Rectangular Pulses.

6.1.2 Measurement of 7\ and T2

Different experiments are needed to measure Tl and T2 accurately,
though, if they are likely to be equal, then measurement of 7\ alone
usually suffices. However, in many systems T2 < 7\ and both need
to be measured. In addition, an understanding of the method used to
measure T2 will prove helpful when we come to discuss two-dimen-
sional NMR.

6.1.2.1 Measurement of 7\ by population inversion

This is the most popular of several available methods. We have already
seen how a maximum NMR signal is obtained after a 90° Bl pulse.
The same is obtained after a 270° pulse, except that the spins have
been inverted relative to the 90° pulse and the output is 180° out of

www.pdfgrip.com
176 Making the spins dance

phase with that after the shorter pulse. If we arrange our computer
to give us a positive-going absorption peak from the FID following
a 90° pulse, then after a 270° pulse we will get a negative peak. If,
instead, we use a 180° pulse, we create no Mxy but turn the excess
low-energy spins into the high-energy state, -Mr They will relax to
their normal state with characteristic time Tl and the magnetization
will change from - Mz through zero to + Mz (Fig. 5.35). This, of course,
produces no detectable effects. However, if, at some time r (do not
confuse this with rc) after the 180° pulse, we apply a second pulse of
90°, we will create magnetization Mxy equal in magnitude to Mz at that
instant. The spectrum that results from processing the FID will be
negative-going if Mz is still negative (as if part of the magnetization
had undergone a 270° pulse) and positive-going if Mz has passed
through zero (Fig. 6.7). A series of spectra are obtained in this way
for a number of different values of r and the intensity of the resulting
peaks plotted as a function of r, so allowing us to extract a value for
7\. The equations governing the behaviour of the transverse and longi-
tudinal magnetization Mxy and Mz and their return to equilibrium
following a 180° Bl pulse are

(M,X = (M,)Jl-2exp(^)|
L \M/J

FT

FT

Mxy remains zero, no nuclear signal

-Mz relaxes to Mz directly

Figure 6.7 (a) Production of a nuclear response using a (90°),, pulse. The FID and its transform are repre-
sented on the right of the figure, (b) A similar response is obtained after a (270°),, pulse, but the FID is
180° out of phase, giving an inverted transform, (c) A 180° pulse gives no xy magnetization but places Mz
in a non-equilibrium position opposing the field. This magnetization relaxes back to its equilibrium value
and no transverse magnetization is produced at any time throughout this process.

www.pdfgrip.com
 Decoupling 177

Sufficient time must elapse between each 180°/90° pair of pulses to

allow Mz to relax fully to its equilibrium value or incorrect results will
be obtained. Usually a waiting period of five to ten times 7\ is used.
If Tl is very long, then the experiment can be very time-consuming,
but other pulse sequences have been worked out that will allow the
total time to be reduced (Figs 6.8, 6.9).
If there are several resonances in our spectrum, each relaxes at its
own rate and it is possible with this method to measure the relaxation
times of all the nuclei of the same isotope in a molecule and to obtain
detailed information about molecular motion.
If the nuclei are spin-coupled then the apparent relaxation times
may not be simply related to the real ones. Carbon-13 relaxation
times are thus measured with the protons decoupled by double irra-
diation and this gives satisfactory values of 7\.

Figure 6.8 How the pulse sequence 180°-wait T-90° can be used to perturb Mz and follow what then
happens to Mr

www.pdfgrip.com
178 Making the spins dance

Figure 6.9 The full population inversion experiment. A series of spectra are
obtained at different T. A plot of their intensities as a function of time gives
the rate of relaxation, from which Tl can be derived. (After Martin et al
(1980) Practical NMR Spectroscopy, John Wiley and Sons Ltd, reprinted with
permission.)

6.12.2 Measurement of T2
The contribution of magnetic field inhomogeneity to T2 will usually
be of the order of 3.0 to 0.3 s for a high-resolution magnet. Thus,
provided T2 is less than about 0.003 s, it can be measured directly
either from the line width or from the rate of decay of the FID. In the
latter case we have to suppose that there is only a single resonance.
Accurate measurement of longer T2 is made using a spin-echo exper-
iment. This depends upon the rate at which the FID decays being
faster than the rate of longitudinal relaxation R{. The decay is due to
the T2* mechanism, which has two components. The loss of phase
coherence between the spins in the xy plane is caused both by the
random relaxation field and by the fact that the homogeneity of the
magnetic field within the sample is not perfect, so that nuclei in
different parts of the sample have different precession frequencies.
There is an important difference between these two relaxation
processes: the first is entirely random, and so unpredictable; whereas

www.pdfgrip.com
 Decoupling 179

the second acts continuously and is constant at each part of the sample,
so that we can in principle correct for this relaxation contribution. We
assume that the random contribution to relaxation is negligible. If we
apply a 90° pulse, all the magnetization Mxy is in the xy plane, but
spins in different parts of the sample have different angular velocities
due to the inhomogeneities in the magnetic field, and so some move
ahead of the average and some lag behind. We wait a period of r
seconds for the spin distribution to evolve; the value of Mxy will
decrease, and may even become zero if the field inhomogeneity is
large. We then apply a 180° pulse, and all spins precess around Bl to
the other side of the xy plane. They do not take up their mirror posi-
tions. We have thus put the faster spins at the rear and the slower
spins in front. They continue to precess, but after further time r
seconds they come into phase again and the magnetization Mxy is again
a maximum. The 180° pulse is a refocusing pulse and the intensity of
the spectrometer output rises following this pulse to a maximum r
seconds after and then decays again. Another 180° pulse will refocus
the magnetization, which indeed can be refocused indefinitely, given
perfect pulses. The refocusing does not work for the random relax-
ation process, and so the duration of the experiment is limited by the
intrinsic transverse relaxation. Indeed, the train of echoes decays in
intensity at a rate determined by the real T29 which can be measured
from a plot of intensity versus time. The experiment is summarized in
Fig. 6.10. Each echo is effectively two back-to-back FIDs. It is possible
to split them and Fourier transform each so that, if the sample contains
several resonances, then T2 can be measured separately for each from
the resulting absorption spectra. If spin-spin coupling is present,
however, then this does not work and modulation of the echoes is
produced, which, as we shall see, proves useful in multidimensional
spectroscopy.

6.1.2.3 Measurement of T]p and spin-locking

The Carr-Purcell pulse sequence can be turned into a spin-lock pulse
sequence. All that is required is to make T in Fig. 6.10 very short, so
that the pulse sequence becomes (90°),,- {(180°),^, where n is a large
number, see Fig. 11.5. B^ is now continuous and the magnetization is
now locked in the / direction, and the effective magnetic field is now
no longer BQ but B^. The nuclei now precess about this field with the
appropriate Larmor frequency. This is typically in the region of a few
tens of kHz. Molecules which previously were tumbling slowly
compared with an NMR frequency of yBQ/2^ are now tumbling rapidly
when compared with yB^ir. This is exploited in a number of exper-
iments. For example in the experiment ROESY, see section 9.6.3, it
is possible to obtain NOE correlations between protons in large mol-
ecules where the conventional NOE measurements fail. The relaxation
time, Tlp, the relaxation time in the rotating frame, is obtained from
the diminution in signal strength with time of spin locking. In the

www.pdfgrip.com
180 Making the spins dance

(a)

(b)
Figure 6.10 Illustrating the Carr-Purcell pulse sequence for measuring T2. The behaviour of the spins is
shown relative to the rotating frame, as if they were stationary. The (90°)^. pulse produces magnetization
Mxy, which then decreases as the spins move apart, s = slow, f = fast. The (180°)^ pulse alters the relative
positions of the slower and faster spins, which now close up again and Mxy increases, reaching a maximum.
The spins are said to be refocused. The output then decreases again. Fig. 6.10(b) shows diagrammatically
how the first and second (the echo) FIDs appear.

extreme narrowing region it is the same as T2, but as the relaxation

is occurring in a magnetic field strength of Bl9 it is the frequency
oo = yBl which applies, greatly extending the extreme narrowing region
of relaxation.

6.2 COMPOSITE PULSES

Modern NMR spectroscopy relies on the use of a computer to apply

accurately timed pulses to samples. We have earlier talked in terms
of 90° and 180° pulses as if they were exactly that, but in practice it
is difficult to apply an accurately set 90° or 180° pulse over the obser-
vation volume of the sample. At the edges of the coil, the effective-
ness of the pulse can deviate markedly from that set. This is easily

www.pdfgrip.com
 Composite pulses 181

observed if the length of the 180° pulse is measured by gradually

increasing the pulse width. The FID intensity increases up to the 90°
point and then decreases, theoretically to zero at 180°. However, a
perfect null is never achieved (Fig. 6.11) (unless a short sample is used
and the loss in resolution can be tolerated).
The exact length of a 90° or 180° pulse depends on the sample and
the exact tuning of a probe. It is therefore possible that the pulse
length set differs significantly from the ideal value. This can lead to a
loss in signal to noise ratio and to artefacts in spectra. A second
problem arises from spectral width. A typical 180° pulse has a length
of 20 JJLS. Such a pulse only produces acceptable excitation over
± 12.5 kHz. When it is remembered that the full spectral width of 19F
is 300 kHz and that of 31P is less than 100 kHz at 9.4 T, it can be appre-
ciated that such a spectral width is inadequate. The problem can be
considerably reduced by using composite pulses. Only a few of the
simplest ones are illustrated here, but much more complex and more
efficient ones can be constructed.

6.2.1 Composite 180° pulse

The composite pulse (90°)x(lSO°)y(90°)x represents one of the simplest
composite pulses. The 180° pulse is broken into two 90° pulses and
an additional pulse is applied in the middle. Figure 6.12 illustrates the
path of the magnetization vector if an error is made and the actual

Figure 6.11 The determination of the length of a 180° pulse by gradually

increasing the pulse length. The numbers given above or below each signal
are the pulse lengths in microseconds used to obtain the particular signal. The
180° pulse is at 16 JJLS.

www.pdfgrip.com
182 Making the spins dance

Figure 6.12 The path followed by the tip of the magnetization vector during
a (800)JC(160°)y(800).c pulse sequence is shown by the solid line. The pulse
sequence has converted the z component of Mz to -0.998MZ compared with
-0.940M., produced by a single 160° pulse.

180° pulse corresponds to a 160° pulse. The first pulse rotates the
magnetization vector 80° around the x axis towards the xy plane. The
second pulse rotates the magnetization vector 160° around the y axis.
The third pulse rotates the magnetization vector 80° around the x axis
and places it close to where it would have been if a true 180° pulse
had been applied.
In practice, it has been found that a slightly different sequence, the
(9Q0)x(24Q°)y(9Q°)x composite pulse sequence works best.

6.2.1.1 Composite 90° pulse

The pulse sequence (9Q°)x(9Q°)y provides a simple way to minimize the
effects of being off-resonance. If instead of placing the 90° pulse at
the frequency of the resonance, it is placed 5 kHz off-resonance, and
if the 90° pulse lasts 10 JJLS, then during this time the nuclear magneti-
zation will have moved 0.000 01 x 5000 x 360° = 18° relative to the
rotating frame set at the spectrometer frequency. This could produce
significant errors if a pulse sequence were being used with a series of
pulses at the spectrometer frequency. A subsequent 90° pulse along the
y axis brings the magnetization close to the y axis, see Fig. 6.13.

6.3 REFOCUSING PULSE

Pulse sequences are frequently used with delays between pulses. This
can cause very bad phasing problems. Consider what happens if a

www.pdfgrip.com
 Refocusing pulse 183

Figure 6.13 The path followed by the tip of the magnetization vector during
a (800)^(800)y pulse sequence is shown by the solid line.

(90°)^ pulse is applied to a collection of nuclei in different chemical

environments, say, the *H decoupled 13C NMR spectrum of a typical
organic compound, where the chemical shift range is large. Once the
magnetization is in the rotating frame, the separate magnetization
vectors for each type of 13C fan out and there are phase shifts with
respect to the instrument frequency. This is illustrated in Fig. 6.14.
The problem is easily avoided by placing a (180°)^ pulse halfway
between the (90°)x pulse and the beginning of the acquisition. In later
chapters, you will find many pulse sequences where this technique is
used. There is an additional advantage in doing this. If there is signif-
icant line broadening due to magnetic field inhomogeneity, then effects

200 180 160 140 120 100 80 60 40 20 0

13
C/ppm
Figure 6.14 A 13C NMR spectrum of cholesteryl acetate which has been
measured using the pulse sequence ST^-QOMXOOls-acquire FID.

www.pdfgrip.com
184 Making the spins dance

due to this will also be refocused (Fig. 6.15). Comparison of this pulse
sequence with that used for T2 determination shows that it is the
Carr-Purcell pulse sequence.

6.3.1 Selective pulses

It is often necessary to apply a pulse to a single resonance or even
a line of a multiplet. When it is remembered that the bandwidth of a
pulse depends on its length (Fig. 1.6), it can be seen that a selective
pulse which has a length of 0.1 s, will uniformly excite nuclei within
± 2.5 Hz but will have effects several 10s of Hz further away from it.
Hence, in order to apply a selective pulse, there are two problems to
overcome. Firstly, the length of a 180° pulse must be increased from
typically 20 JJLS to say 0.1 s. Secondly, the pulse needs to be made more
selective.

6.3.1.1 DANTE pulses

Many of the older NMR spectrometers do not have the facility which
enables the operator to change the power of a pulse. We are there-
fore forced to use short powerful pulses. The selectivity is then
achieved using the DANTE pulse sequence. DANTE stands for Delays
Alternating with Nutation for Tailored Excitation. Instead of using a
single 180° pulse, the pulse is broken up into n short 180°//7 pulses
separated by a time T. This is illustrated in Fig. 6.16. By a suitable
choice of n and T, the pulse can be made to last whatever time is
required. Hence 11 short 16.4° pulses separated by 0.01 s produces a
180° pulse with a selectivity of ± 2.5 Hz. There are problems associ-
ated with the technique. One is that not only is a single excitation
frequency produced at the spectrometer frequency, but additional ones
(sidebands) are produced at ± W/T, where m is any integer, on either
side of the spectrometer frequency. Great care is necessary in the
choice of T to make sure that these sidebands do not fall on any other
signal in the spectrum. Secondly, the pulse amplifier does not generate

Figure 6.15 The effect of the (180°) refocusing pulse on a collection of nuclei, a, b, c, d, and e, with
different resonance frequencies, (a) The alignment of the nuclear spins immediately following a (90°)^
pulse, (b) After a time T, the nuclei have fanned out with nuclei a and b rotating faster than the rotating
frame, and nuclei d and e rotating slower. Nuclei c are on resonance, (c) The effect of the (180°)v refo-
cusing pulse, (d) After a time T, the nuclei have refocused.

www.pdfgrip.com
 Refocusing pulse 185

p1 d1 p1
(a) (b)

Figure 6.16 (a) The DANTE pulse sequence, (b) The frequency distribution generated by the DANTE
pulse sequence.

a perfect square pulse. There is a rise and fall time. The 180° DANTE
pulse has to be calibrated, and each component can be significantly
longer than 180°/n when the required length is less than 1 JJLS.

6.3.1.2 Shaped pulses

On modern spectrometers it is possible to attenuate the pulse and
hence lengthen the 180° pulse to that required by selectivity.
Examination of Fig. 6.17(a) shows that a rectangular pulse is not partic-
ularly selective with significant perturbation of signals extending well
outside the range of uniform excitation. This problem is considerably
reduced by using shaped pulses. The relationship between the square
pulse and the siruc/x shape in Fig. 6.17(b) is a Fourier transform. It
is therefore of no surprise that if the power of the pulse is modulated
following the sirut/Jt relationship, the result is highly selective. In
practice, this can cause problems as the pulse becomes very long.
Frequently a compromise is used, such as a pulse with a Gaussian
shape, which produces a Gaussian frequency distribution (Fig. 6.17(c)).

Pulse shape Excitation profile

(a)

(c)

Figure 6.17 (a) A rectangular pulse and its Fourier transform, (b) A sin xlx
pulse and its Fourier transform, (c) A Gaussian pulse and its Fourier trans-
form.

www.pdfgrip.com
186 Making the spins dance

6.3.2 The BIRD pulse sequence

The BIRD pulse sequence permits the selection of nuclei attached to
a second coupled nucleus in preference to those not so coupled. For
example, 1H attached to 13C or 15N can be observed while the signals
from the 1H attached to 12C or 14N in the same sample are suppressed.
The acronym, BIRD, stands for Bilinear Rotation Decoupling. The
experiment is essentially a 180°-T-90° pulse sequence for the protons
attached to the NMR inactive nuclei, e.g. 12C, with T chosen to give
zero signal. The pulse sequence is given in Fig. 6.18. The first !H (90°)^.
pulse, pi, brings all the !H magnetization into the xy plane, pointing
along the y' axis in the rotating frame. The two components of the !H
doublet due to coupling to 13C rotate with respect to the *'/ rotating
frame. One component, coupled to 13C with an a spin, rotates more
slowly and the other, coupled to 13C with a £ spin, rotates more rapidly.
Each arm of the doublet is separated from the middle of the doublet
by JI2. Hence one arm will be rotating at + JI2 Hz with respect to the
rotating frame and the other at -JI2 Hz. After a time, 1/27 s, the two
arms of the doublet will have rotated 90° and now point in line along
the dbt' axes of the rotating frame (Fig. 6.19). (180°)^. pulses are applied
to 1H and 13C. In this case, the (180°), !H pulse will have no effect,
but if the !H NMR signal was off-resonance, it would act as a refo-
cusing pulse, see section 6.3. The 180° 13C pulse swops the spins of
the 13C nuclei. The result is that the protons which were attached to
13
C with a-spin are now attached to 13C nuclei with p-spin and vice
versa. As the sign of the 13C spin determines the direction of rotation
of the *H magnetization of the coupled protons with respect to the

-x
1
H

d1 p1 d2 p2 d2 p3 d3 P5

x
13C

p4
Figure 6.18 The BIRD pulse sequence, dl 1is a relaxation delay, 57\ being ideal. d2 = 1/27 s. d3 is a delay
for the -z magnetization generated in the H nuclei attached to 12C to decay to zero, pi, p3, and p5 are
90° pulses, while p2 and p4 are 180° pulses. The phases of the pulses are given above them. The actual
BIRD pulse sequence comprises pi, d2, p2, p3, and p4. d3 and p5 have been added here to demonstrate
the consequences of applying the pulse sequence.

www.pdfgrip.com
 Refocusing pulse 187

rotating frame, the *H magnetizations now reverse their direction of

rotation and refocus after 1/27 s. The final (90°)^ pulse returns the X H
magnetization to the z direction. Compare this with what happens to
the magnetization of the protons attached to 12C (Fig. 6.19(b)). If the
1
H nuclei are on resonance, nothing happens during the 1/27 s waiting
period. The (180°)^ !H pulse, not only refocuses, but rotates the 1H
magnetization to point along the -y direction. As the 1H is attached
to 12C, not 13C, the 180° 13C pulse does nothing. The final (90°)., pulse
causes the 1H magnetization to rotate into the -z direction. We now
have the situation where the magnetization of XH attached to 12C is
pointing in the -z direction, while that of 1H attached to 13C is pointing
in the +z direction. During the following waiting period, d3, the magne-
tization of the !H attached to 12C gradually recovers by the 7\ mech-
anism. d3 is chosen so that the magnetization of the 1H attached to
12
C has recovered to approximately zero at the end of d3 (Fig. 6.18),
though due to the spread of 7\ values in molecules, a compromise
value of d3 has to be chosen. The final *H pulse gives an output for
the ^-^C protons but very little for the ^-^C protons. Normally
the BIRD pulse sequence is used as part of a longer pulse sequence
combined with phase cycling to suppress more effectively the signals
of 1H attached to 12C.

6.3.3 Magnetic field gradients

It may initially appear to be perverse to apply magnetic field gradi-
ents after having spent time shimming the magnetic field to obtain
optimum resolution but this turns out to be beneficial. We will discuss
simply in later chapters the application of pulses of field gradient
during RF pulse sequences. They permit the selection of certain
types of interaction with the elimination of others. This can be done

(a)

180°(13C)

(b)

Figure 6.19 Part of the BIRD pulse sequence to show the behaviour of the nuclear spins in the xy plane,
(a) The behaviour of the 1H13C doublet, (b) The behaviour of the 1H12C singlet.

www.pdfgrip.com
188 Making the spins dance

using phase cycling of the type described above, with the difference
that phase cycling requires many acquisitions to complete a cycle
whereas field gradient pulses do the job in one acquisition. They are
used particularly in biochemical spectroscopy.

6.4 QUESTIONS

6.1. Why do we use time shared decoupling for homonuclear decou-

pling but continuous decoupling for heteronuclear decoupling?
6.2. Where does the nuclear magnetization point after the three basic
pulses used to make up the WALTZ decoupling sequence, i.e.
(90°)x(180°)_x(270°)x assuming that the magnetization was in the
z direction initially?
6.3. You have placed the 1H decoupling frequency 1000 Hz from a
proton which is coupled to a 13C nucleus with 1J = 125 Hz. You
measure the 13C NMR spectrum and find an apparent 1J of
100 Hz. What is the strength of the decoupling field? Note that
yc = 6.7263 x 107 rad T'1 s'1.
6.4. In the inversion-recovery experiment used to measure 7\, the
signal intensity is negative for short recovery times but positive
for long recovery times. Calculate, in terms of 7\, at what
recovery time the signal intensity is zero.

www.pdfgrip.com
 NMR spectra of
exchanging and
reacting systems
7
7.1 SYSTEMS AT EQUILIBRIUM

One of the most important contributions that NMR has made to chem-
istry is the insight it has given into the dynamic, time-dependent nature
of many systems, particularly those which are at equilibrium or where
simply intramolecular motion is involved. Spectroscopy based on
higher-frequency radiation, such as classical infrared (IR) or ultra-
violet (UV) Spectroscopy, has given mostly a static picture because the
timescale of many processes is slow relative to the frequency used.
However, the lower frequencies used for NMR and the smaller line
separations involved, coupled with the small natural linewidths
obtained, means that many time-dependent processes affect the spectra
profoundly. As an example, we consider the spectroscopic behaviour
of ethanol in Fig. 7.1. The proton spectrum of 50% ethanol,
HOCH2CH3, in CDC13 is a methyl triplet due to coupling to the CH2,
an OH triplet for the same reason and a doublet of quartets for the
methylene protons. Any acidic impurity catalyses interchange of OH
protons between molecules:
H+
EtOH + EtOH* ^ ^ EtOH* + EtOH
The exchange of the protons results in a short break in the CH2OH
coupling path and, since the total spin of the CH2 protons in the two
molecules between which the proton jumps may not be identical, then
some of the OH protons will suffer an abrupt change in frequency.
The result is to introduce uncertainty into their nuclear frequency and
thus line broadening. In the spectrum of Fig. 7.1(b), which is of 50%
ethanol, coupling of the OH and CH2 proton signals is evident.
Addition of acid causes acceleration of the rate of proton exchange
to the extent that the OH-methylene coupling is completely lost and
only an average frequency can be detected (Fig. 7.1(c)). The lines are
now sharp. This is called the fast exchange region. When the lines
are broadened (Fig. 7.1(a)) it is called the region of intermediate
exchange rates and where coupling is fully developed is called the slow
exchange region.

www.pdfgrip.com
190 NMR spectra of exchanging and reacting systems

CH3

CH2 OH
(a)
CH3

OH

4.954.90 3.7 3.6

(b)
CH,

OH CH2

(c)
5 4 3 2 1 0
1
H/ppm

Figure 7.1 The 400 MHz ]H NMR spectra of ethanol. (a) 2% EtOH in CDC13. (b) 50% EtOH in CDC13.
(c) As (b), but with a drop of acid added.

Ethanol also allows us to demonstrate another aspect of fast

exchange. In the concentrated solution, there is extensive hydrogen
bonding between OH oxygen in one molecule and OH hydrogen in
another. This interaction causes a high frequency shift of the OH
proton resonance. If the alcohol is dilute (Fig. 7.1 (a)) then the
hydrogen bonds become dissociated to an extent depending upon
the dilution, and there is a low-frequency shift of the OH resonance.
The solution now can be regarded as containing two types of ethanol,
hydrogen-bonded and non-hydrogen-bonded. These two species have
different OH proton chemical shifts but are not observed separately
because of the fast exchange between the two types of ethanol, which
results in a signal of the average frequency being observed weighted
by the concentrations of the two species. Thus chemical shift-dilution
plots give information about the hydrogen bond dissociation. The lines
in this solution are also broadened, presumably because of some acid
impurity in the solvent.
Finally, we should note that in compounds such as ethanol there is
very rapid rotation around the C-C bonds. At any instant, two of the

www.pdfgrip.com
 Systems at equilibrium 191

methyl protons are close to the OH group and one is pointing away;
see the Newman projection in Fig. 7.2. These differences are not
observed in the spectrum, however, because each proton has an
average chemical shift due to the rapid internal rotation. In certain
molecules, such conformational changes can be quite slow and then
the effects of the motion can be detected in the NMR spectra.
Following the Karplus relationship (Fig. 3.1), the interproton coupling
constants will also be different in an ethyl group; these are also aver- Figure 7.2 A Newman
aged by the rotation to the value of the average of the Karplus curve. projection about the C-C
bond of ethanol, showing
the two environments for
7.1.1 The effects of exchange on the lineshape of NMR spectra the methyl protons.
We have described exchange by such words as 'fast' and 'slow'. It is
now necessary to determine the timescale within which we can apply
these terms correctly. A set of theoretical spectra for a two-site
exchange with equal population in the two sites and no spin-spin
coupling is shown in Fig. 7.3.
We start from the situation where the nuclei spend a long time in
a given location in the molecule, Fig. 7.3(a), as shown in section 4.1,
the linewidth is given by

^1/2 = —
vT2
When chemical exchange occurs, the nuclei change their chemical
shift as a consequence, and spend a time, TC, in a given location in the
molecule. There is an extra contribution to the linewidth due to
exchange and the linewidth of the exchange broadened line at half
height, (W1/2)ex, becomes

W/2)ex = J- + J-
TtT2 TTTex

This is a direct result of the Heisenberg uncertainty principle. As the

lifetime in a given site gets shorter, the energy, and hence the linewidth
becomes less well defined. TC is simply related to the rate of exchange,
k, by

k=±
T
ex

Hence

W/2L = ^ + T
and k can be derived from the linewidth of the broadened line, (W1/2)ex,
by the simple calculation

k = or {(W1/2)ex - ^r} = ^{(W1/2)ex - (W1/2)0) (7.1)

www.pdfgrip.com
192 NMR spectra of exchanging and reacting systems

(a),

(b).

(c).

(d)-

(e).

(f)-

(g)-

(h).

(0-

(j)-

Figure 7.3 Calculated spectra for exchange between two equally populated
sites separated by 40 Hz.1 T2 values of both sites are 1 s. 1(a) k = 0.1 s~l. (b)
k = 5s- . (c) k = 10s- . (d) k = 20s- . (e) k = 40s' . (f) k = 88.8s-1.
1 1

(g) k = 200 s-1. (h) k = 400 s'1. (i) k = 800 s~l. (j) k = 10 000 s'1, where k is
the rate of exchange between the two sites, and is normally varied by varying
the temperature.

where (W1/2)o is the linewidth in the absence of exchange. Equation

(7.1) is very simple and is general for any exchange problem, provided
exchange is slow enough to result in separate signals for the exchanging
sites. It does not matter if the populations are unequal, but caution is
necessary when there is coupling, and the equation should only be
applied if separate signals are observed for each line of the multiplet.
Normally, (W1/2)o is estimated from a signal not involved in the
exchange, and provided that (W1/2)ex is substantially larger than (W1/2)p>
an accurate value is obtained for k. The result of equation (7.1) is
that for most compounds all signals involved in exchange initially
broaden equally.

www.pdfgrip.com
 Systems at equilibrium 193

As the rate of exchange increases, the lines broaden, and the trough
between the lines gradually fills. Once the lines are no longer well
resolved, e.g. Fig. 7.3(d), the use of equation (7.1) becomes subject to
error. The temperature where k is such that the trough between the
two lines rises to be level with the peaks, Fig. 7.3(e), is known as
the 'coalescence temperature'. At the coalescence temperature
TT|VA-VB|
k- j- (7.2)

where VA and VB are the frequencies of the two sites in hertz at the
coalescence temperature. This equation only applies to exchange
between two equally populated uncoupled sites and (Wl/2)o has to be
negligible when compared with I VA - VB I. In principle, very accurate
rates can be obtained at the coalescence temperature, provided
that (VA - VB) is known. As chemical shifts are temperature dependent,
(V A -V B ) must be determined by extrapolation from lower tempera-
tures, or, better, by complete lineshape fitting. Caution is necessary
when extrapolating from low temperature, as when the signals begin
to overlap and the trough between them fills, the maxima move
together (Fig. 7.4).
It must be remembered that there is not a single coalescence temper-
ature for a compound. (VA - VB) depends on the magnetic field strength,
the pair of nuclei exchanging, and the solvent. It is therefore mean-
ingless to quote the coalescence temperature for a compound.
Above coalescence, the signals of the two lines are averaged, initially
to a non-Lorentzian lineshape, but as the rate of exchange increases
further, a Lorentzian lineshape is achieved (Fig. 7.3(g)-(j)). Then the
equation
k= ^(VA ~ ^B)2 (73)
2{(W1/2)ex - (W1/2)0)
applies. This equation only applies to exchange between two equally
populated uncoupled sites. The equation must be used with caution,
as it is frequently difficult to obtain a reliable estimate of (VA - VB) at
the experimental temperature. The dependence of equation (7.3) on
(VA - vB)2 means that the line broadening of signals above the coales-
cence temperature can differ markedly for each pair of exchanging
signals.
Frequently exchange processes affect several signals from a
compound. This is illustrated here for W-methylaniline (Fig. 7.5). At
-133°C, separate 13C NMR signals are observed from all the six
aromatic carbon atoms, showing that there is slow rotation about the
Ph-N bond. At -126°C, the rotation produces a noticeable broadening
of the signals due to C2, C3, C5, and C6 as can be observed by comparing
the heights of these signals with the height of the signal due to C4. On
warming to -115°C, the signals due to C3 and C5 have averaged to
give one signal, while one broadened signal is observed for C2 and C6.

www.pdfgrip.com
194 NMR spectra of exchanging and reacting systems

62°C

40°C
36°C
32°C

26°C

10°C

-2°C

-27°C
6.0 5.5 5.0
1
H/ppm

Figure 7.4 The 100 MHz 1H NMR spectrum of [Ti(V-C5H5)2(7]5-C5H5)2] in

toluene at different temperatures. At -27°C the two types of cyclopentadi-
enyl group are observed. For the r^-group, there are rapid shifts of the tita-
nium around the ring averaging all the protons of this ring. At higher
temperatures, the two types of cyclopentadienyl ring exchange and an aver-
aged signal is observed at 62°C. The spectrum at 36°C, where the doublet
structure is just lost, is said to be at coalescence. (Reproduced with permis-
sion from Calderon et al (1971) /. Am. Chem. Soc., 93, 3587 copyright, (1971)
American Chemical Society.)

This is readily explained when it is remembered that the coalescence

temperature depends on the signal separation. For C3 and C5, the signal
separation is small, while for C2 and C6 the separation is large.
Coalescence of the signals due to C2 and C6 in fact occurs at -115°C,
while at this temperature, the averaged signal due to C3 and C5 is rela-
tively sharp. The broadness of the signal due to C2 and C6 is substan-
tial and such a signal can easily be missed. At -75°C, sharp averaged
signals are observed for all the carbon atoms.
Exchange between unequally populated uncoupled sites results in
differential broadening of the lines at exchange rates below the coales-
cence point (Fig. 7.6). The weaker signal from site A broadens more

www.pdfgrip.com
 Systems at equilibrium 195

3,5
2,6

-75°C

-115°C

-126°C J Jl

-133°C

Figure 7.5 The variable temperature 25.16 MHz ^C^H} NMR spectrum of
W-methylaniline in Me2O. (Reproduced from Lunazzi et al (1979) Tetrahedron
Letts., 3031, copyright (1979), with permission from Elsevier Science.)

than the stronger one from site B. This is a direct consequence of the
law of microscopic reversibility. This predicts that
^ = * B or *A = *B£*
PE PA PA
where A:A and kE are the rates of leaving sites A and B, respectively,
and pA and pE are the populations of sites A and B, respectively.
Hence if pA = 3/?B, then according to equation (7.1), the line broad-
ening is proportional to k, the signal of B will broaden twice as much
as the signal due to A. This is apparent in Fig. 7.6(c).
Above coalescence an average signal is observed at the weighted
average chemical shift of the two signals, vav, where
_ VAPA + VBPB
Vav
PA+PB

www.pdfgrip.com
196 NMR spectra of exchanging and reacting systems

(a)

(b)

(c)

(d)

(e)

(f)

(9)

(h)

(i)

(j)

Figure 7.6 Calculated spectra for exchange between two sites with popula-
tions in the ratio 1: 3 separated by 40 Hz. T2 values of both sites are 1 s. (a)
k = 0.11 s-1. (b) k = 5 s- 1
. (c) k = 10 s'11. (d) k = 20 s-11. (e) k = 40 s'1. (f) k=
88.8 s- . (g) k = 200 s- . (h) k = 400 s' . (i) k = 800 s' . (j) k = 10 000 s'1.
1

Bromocyclohexane provides an example of a two-site exchange

problem involving exchange between two molecular species of unequal
populations, namely the conformers with axial and equatorial
bromides.

equatorial axial
bromide bromide
The variable temperature 13C NMR spectra are shown in Fig. 7.7.
The determination of the exchange rates and ratio of isomers where
separate signals are observed is easy and accurate up to about -44°C.
At higher temperatures, where only average signals are observed, the
determination of accurate values of (VA - VB) and ratio of concentration

www.pdfgrip.com
 Systems at equilibrium 197

2e 3e
6e 5e

1e 2a 4e 3a
6a 5a
1a 4a
-94 °C_ J_

-60 °C-

-44 °C-

-30 °C.

-16 °C.

-2°C-

11 °C. Al i
20 °C
55 50 45 40 35 30 25 20
13
C/ppm
Figure 7.7 The variable temperature 100.62 MHz 13C NMR spectrum of
bromocyclohexane in CD3C6D5. The 3a and 5a resonance is partially obscured
at -44°C and below by the CD3 solvent resonance. The a and e refer to the
position of the bromine atom, axial or equatorial.

www.pdfgrip.com
198 NMR spectra of exchanging and reacting systems

of isomers are subject to considerable error. Hence, exchange rates

above coalescence have to be treated with considerable caution.
Examination of Fig 7.7 is instructive. Between 20 and -2°C, the
signal due to C4 is relatively sharp, but as the temperature is lowered
through the range, there is differential broadening of the signals due
to C1, C2'6, and C3'5. Although for all four carbon nuclei the rate of
exchange is the same, the broadening depends on the chemical shift
separation between the axial and equatorial isomers. This is greatest
for C3'5 and least for C4. At -44°C, separate signals are observed for
the two isomers, and now the linewidths of all the signals for the major
isomer are equal, as are those for the minor isomer. The relative
linewidths of the major and minor isomers are 0.28 :1, being inversely
proportional to their relative concentrations. On further cooling, all
the signals sharpen as the rate of exchange slows.
The analysis of spin-coupled systems usually requires computer
fitting of the spectra. This can lead to accurate rate constants and valu-
able mechanistic information. For instance, two different mechanisms
of exchange have been proposed for the trigonal bipyramidal struc-
ture shown in Fig. 7.8. Figure 7.9 shows the 31P{1H} NMR spectrum
of [Rh{P(OMe)3}5]+ at low temperature, along with spectra calculated
on the basis of two mechanisms of exchange. Examination of the
spectra shows that the better agreement is between the spectra
calculated based on the Berry pseudo-rotation mechanism. The differ-
ences are only observed due to the second-order nature of the
spectrum, which is A2B3X, where the X group is 103Rh, 100% abun-
dant, I = 1/2.

7.1.2 The use of magnetization transfer to study exchange

The difficulty of lineshape analysis is that the only information avail-
able is the rate of leaving each site. No information is directly avail-
able on the destination. In the case of a two-site exchange problem,
the destination is normally obvious, but for multisite exchange prob-
lems, the destination is often far from obvious. There are a number
of different magnetization transfer experiments used to examine such
situations and they involve disturbing the Boltzmann population distri-
bution in one or more sites.

p1 p2 p1 p2

2
IP3 1
IP3 2
IP3 5
IP1
P -R< 4A ^=^P -R< 4A P -R< 4, ^=^P -RI< 4.
l>
p5
l>
p5
l>
po
l>
p3
K
(a) (b)

Figure 7.8 The two possible mechanisms of exchange in [Rh{P(OMe)3}5]+,

with the P(OMe)3 ligands represented by individual phosphorus atoms, (a)
The pairwise exchange mechanism which exchanges one axial phosphorus
ligand with an equatorial one. (b) The Berry pseudo-rotation mechanism
which exchanges the two axial phosphorus ligands with two equatorial ones.

www.pdfgrip.com
 Systems at equilibrium 199

Berry
pseudo-rotation
calculated

Observed
-114 °C

Pairwise
exchange
calculated

Figure 7.9 Observed and calculated 36.43 MHz 31P NMR spectra of
[Rh{P(OMe)3}5]+ in CHC1F2/CH2C12, 9:1 at -114°C, with the effect of the
protons removed by double irradiation. Arrows are used to draw attention
to the regions of the spectrum where differences are most marked.
(Reproduced with permission from Meakin and lesson (1973) /. Am. Chem.
Soc., 95, 7272, copyright (1973) American Chemical Society.)

The simplest experiment is saturation transfer. In this experiment,

the decoupler is turned on to one site for several seconds. Then the
decoupler is turned off and a general 90° pulse is applied to examine
the effect. A second reference spectrum is taken where the decoupler
is set on a region of the spectrum where there is no signal. A differ-
ence spectrum is taken to show what changes have been induced in
the spectrum as a result of the irradiation. Let us analyse what is
happening for exchange between two sites A and B. Before the decou-
pler was turned on, the spin system was at equilibrium with equilib-
rium magnetization, Mz(0), at both sites. The decoupler, applied to
site A, destroys the magnetization and MZA = 0. If there is exchange

www.pdfgrip.com
200 NMR spectra of exchanging and reacting systems

to the second site, nuclei from site A with no magnetization move to

site B, while nuclei from site B, move to site A, where their magne-
tization is destroyed. The result is that the magnetization of site B is
decreased by the exchange. The magnetization recovers by the Tl
processes. The experiment can be done dynamically, where the decou-
pler is turned on for various times, but here we will only analyse the
case where the decoupler is turned on until equilibrium is achieved,
typically 57^. In a complicated molecule, it is easier to see the results
of magnetization transfer in a difference spectrum, where the spec-
trum in the presence of magnetization transfer is subtracted from a
reference spectrum so only changes are shown.
The time dependence of the magnetization at site B, MZB(Y), is given

"y d^)^Ao)-BM,°(,)_,BM;E(t) (74)

at i^
The first term gives the 7\ recovery of the magnetization in site B and
the second term is due to the loss of magnetization from site B by
exchange. At equilibrium when t = <*>, the magnetization does not
change, so
M B(0)
' -BM-V)-M*,V) = °
1
l

or
1 M 7 B (0)-M 7 B H
kB
-T» M/H ^
If kE » I/TV, then M,BH -0. If kE « 1/7\B, then M,B(o>) ~MZB(0).
Hence if kE » IIT^, magnetization transfer causes the signal due to
B to vanish as well as the signal due to A. If kB «1/7\B, the signal
due to B is unaffected by magnetization transfer by exchange. If
1/5Tj 8 «k E «5/7"1B, then kB can be determined by measuring
MzB(oo), MZB(0), and T*. This works well if 7\A ~ Tf. Unfortunately
if 7\A * 7\B, it is very difficult to measure T^ and T^ as exchange
occurs during the waiting period of the IT - T - ir/2 pulse sequence,
resulting in a partial averaging of 7\A and T-f.
Considerable use is made of magnetization transfer measurements
by decoupling in order to determine which signals are exchanging,
and it can be seen accidentally during NOE measurements, see section
8.2. An example of its use is to determine how the osmium atom
moves around the cyclooctatetraene ring in [Os(iri6-C8H8)(Ti4-C8H12)]
(Fig. 7.10).
Figure 7.10(a) shows the normal 1H NMR spectrum of the cyclooc-
tatetraene protons of [Os(in6-C8H8)(V-C8H12)]. The spectra in Fig.
7.10(b) and (c) are presented as difference spectra. A reference spec-
trum was also taken where the decoupling frequency was placed well
away from any signal. This was then subtracted from the spectra where
the decoupler was placed on H2 in Fig. 7.10(b) and H4 in Fig. 7.10(c).

www.pdfgrip.com
 Systems at equilibrium 201

's(cod)

5.0 4.0
1
H/ppm

Figure 7.10 A partial 400.13 MHz 1H NMR spectrum of [Os(t]6-C8H8)

(T]4-C8H12)] in CD3C6D5 at 22°C. (a) The 1H NMR spectrum
1 4
of the T]6-C8H8
protons. The signals at 8 5.132 and 5.30 are due to H and H , while those at
8 3.94 and 5.05 are due to H and H3 respectively, (b) A difference spectrum
obtained by recording the spectrum with the decoupling frequency at 8 3.94
and then subtracting a second spectrum with the decoupling frequency set well
away from any signal, (c) As (b), but with the signal at 8 5.30 irradiated.
(Reproduced with permission from Mann (1988), Adv. Organomet. Chem., 28,
397.)
The resulting spectra show only the changes that occur as a result of
the presaturation of H2 or H1. Examination of spectrum in Fig. 7.10(b)
shows that when the magnetization at H2 at 8 3.94 is destroyed a large
negative signal results. There is magnetization transfer to H3 at 8 5.05,
which also gives a negative difference signal. The other signals have
not changed so do not appear in the difference spectrum. Similarly,
examination of spectrum in Fig. 7.10(c) shows that when the magne-
tization at H4 at 8 5.30 is destroyed a large negative signal results.
There is magnetization transfer to H1 at 8 5.13. These results show
that the dominant mechanism is a [1,5] metal shift (Fig. 7.11).

www.pdfgrip.com
202 NMR spectra of exchanging and reacting systems

Os ^ Os
(r|4-C8H12) (i!4-C8H12)

Figure 7.11 The [1,5] shift of the cyclooctatetraene ring in [Os(Ti6-C8H8)

(T!4-C8H12)].

The magnetization of one site can also be changed by a selective

TT-pulse. In this experiment, the pulse sequence, relaxation time -
Ti(selective) - T - ir/2 is used. A series of measurements are performed
with different values of T, chosen to map the exchange and the subse-
quent spin-lattice relaxation. The resulting data are analysed using the
family of equations for n magnetically nonequivalent sites, /,

^.
ai
£ MM+ t W<> +L P1
^} (7.6)
/=!(/*;) y=i('>y) ^i *

The technique is particularly powerful, when applied to a multisite

exchange problem. Typical experiments are shown in sections 7.1.2.1,
7.1.5.3 and 7.1.5.4.

7.1.2.1 Magnetization transfer in f(rf-C5H5)2(H)Nb=CHOZr(H)

(7f-C5Me5)J
If the rate of exchange is slow, so that linewidths are not significantly
perturbed, then magnetization transfer is the ideal technique to use to
examine exchange. This is applied here to [(T|5-C5H5)2(H)Nb=CHOZr
(H)(ti5-C5Me5)] (Fig. 7.12). The formyl signal at 8 11.63 was inverted
using a selective 180° pulse. Ideally, the inverted signal which is
obtained after a short delay time of 0.002 s should be equal to that
obtained after 3.0 s though of opposite sign. Due to imperfections, this
was not achieved, but this is of no consequence as the magnetization
has been substantially perturbed from equilibrium, and the return to
equilibrium can be monitored. The formyl proton signal recovers to
its normal intensity by two processes, spin-lattice relaxation and chem-
ical exchange from the niobium hydride. The niobium hydride signal
at 8 -3.14 initially decreases in intensity because of exchange of the
negative magnetization from the formyl proton. Both signals return to
their normal sizes by spin-lattice relaxation. By fitting the evolution
of the signal intensities with time to the differential equations, the two
unknown quantities Tl and the rate of exchange, k, can be determined.
Values obtained at 32.5°C are k = 15.7 s'1 and Tl = 0.835 s. Notice
that only one 7\ value has been obtained although the two hydrogen
environments will almost certainly have different 7\ values. This is
because during one 7\ period of 0.835 s, the proton visits both sites a

www.pdfgrip.com
 Systems at equilibrium 203

O—ZrH O—2

Mb:
/
H H delay time
0.002s

61H/PPM 611.65 8-3.14

Figure 7.12 The 90 MHz 1H spectrum of [(^5-C5H5)2(H)Nb=CHOZr(H)
(Ti5-C5Me5)] in C6D6 at 32.5°C. A selective 180° pulse has been applied to the
formyl proton (left set of spectra) and exchange transfers to the niobium
hydride (right set of spectra). The delay time is that between the selective
180° and observing 90° pulse. (Reproduced with permission from Threlkel
and Bercaw (1981) /. Am. Chem. Soc., 103, 2650, copyright (1981) American
Chemical Society.)

number of times and 7\ is averaged over both sites. This makes the
analysis of the data easier, but it also removes a possible extra compli-
cation. There is another mechanism by which one proton can perturb
the intensity of another, namely the nuclear Overhauser effect, see
section 8.2. This effect operates on the same time scale as Tl and by
choosing a temperature where k is fast compared with Rl9 it can be
neglected.

7.1.3 Temperature measurement

A factor that weighs heavily on the accuracy is, of course, the temper-
ature of the sample. This is controlled by constructing a Dewar shield

www.pdfgrip.com
204 NMR spectra of exchanging and reacting systems

around the sample holder and passing over the sample cold nitrogen,
boiled-off from liquid nitrogen, or air which have been adjusted to the
required temperature by using a controlled heater. The heater is regu-
lated by a thermocouple placed just below the sample tube. The
method is prone to temperature gradients and so inaccuracies, which
are minimized by calibrating the temperatures. This is done in two
major ways. In the first method a capillary tube, containing MeOH
for below room temperature and HOCH2CH2OH for above room
temperature, is placed inside the NMR tube. The separation between
the OH and CH3 or CH2 groups, Av ppm, is measured. The temper-
ature is related to Av by the equations
for methanol
r°C - 403.0-29.46Av - 23.8Av2
for ethylene glycol
T°C = 466.0 -101.64Av
In the second method, the NMR tube is replaced by one containing
a thermocouple in the same solvent.

7.1.4 The determination of activation energies from NMR data

We can relate T to temperature very accurately at coalescence and
can, for example, obtain a value for the free energy of activation at
that temperature using the Eyring theory

AG* = RT J23.759 + In fj\\ (7.7)

where the rate constant k is obtained from k= l/Tex.

Equation (7.7) permits the calculation of AG* values from NMR
data and the values obtained are usually accurate to better than 1 kJ
moH, especially at and below coalescence. It is then very tempting to
use the relationship
AG* = A//* - TAS*
Combining this equation with equation (7.7) permits the derivation of
AH* and AS*. A plot of In (kIT) against IIT gives a straight line with
gradient A//*/r and intercept (AS*AR + 23.759). Unfortunately this
approach is subject to hidden errors, substantially greater than given
by regression analysis, and many of the values found in the literature
are grossly in error. For example, although the published values of
AG* for rotation about the N-CHO in Me2NCHO are in the range
93 ± 6 kJ moH, the values of A//* range from 25 to 115 kJ moH. The
errors in A//* arise from many causes, some of which are temperature
dependent chemical shifts, mis-estimates of the linewidth in the
absence of exchange, poor temperature calibration, and signal broad-
ening due to unresolved coupling.

www.pdfgrip.com
 Systems at equilibrium 205

There are several approaches to obtaining reliable values of A//*

and AS1*, two of which are described here. Firstly, we can obtain data
over a wide range of temperature if we can monitor coalescence points
for a system at different spectrometer frequencies (or magnetic field
strengths) and for different nuclei or groups with different chemical
shifts that are affected equally by the exchange. This approach has
been used to study hindered rotation of the TV-ethyl groups of the iron
complex [(CH3CH2)2N-C(S)SFe(CO)2(Ti5-C5H5)]. Both the methyl and
methylene protons of the ethyl groups are non-equivalent at low
temperatures, but the chemical shift between methyl signals is less than
that between methylene signals, so that we can monitor two coales-
cence points. In addition, we can obtain two more coalescence points
from the 13C NMR spectra of the two groups, and by using two
different spectrometers we can observe a total of eight coalescences.
The results are shown in Fig. 7.13, which gives the Arrhenius para-
meters Ea = 66 kJ moH and logA = 13.1.
Alternatively, two independent methods can be used to determine
the rate at substantially different temperatures. Lineshape analysis is
reliable from the point when the rate of exchange is sufficient to
produce line broadening which is at least ten times greater than the
natural linewidth until coalescence. This can be complemented
by magnetization transfer which is accurate from a rate of around
\ until the linewidth due to exchange is around 10% of the signal

2.0

13,

.1.5
O)
o

1.0

3.00 3.50 4.00

1/7x10'3
Figure 7.13 Determination of the activation parameters characterizing the
hindered rotation around the C-N bond of [(CH3CH2)2NC(S)SFe(CO)2
(tl5-C5H5)]. (From Martin et al (1980) Practical NMR Spectroscopy, copyright
(1980) John Wiley and Sons Ltd, reprinted with permission.)

www.pdfgrip.com
206 NMR spectra of exchanging and reacting systems

separation. This procedure was applied to Me2NCHO to give A/f* =

84.8 ± 1.3 kJ moH and AS* = -6.2 +3.6 J Kr1 moH (Fig. 7.14).

7.1.5 Some further examples of chemical exchange

7.1.5.1 Cis and trans isomers of a vinyl diamide

A more complex example of the hindered rotation in amides is shown
in Fig. 7.15 for the cis and trans isomers of a vinyl diamide. Only the
methylene proton resonances are shown in the figure, and they indi-
cate that several different rotation processes take place. If we take the
trans isomer first, we see that at the highest temperature recorded
there is a well resolved methylene quartet overlying a broadened reso-
nance. This latter broadens and splits at lower temperatures with a
coalescence temperature of around 345K. Thus one of the amides is
rotating more slowly than the other, which only shows coalescence
between 273 and 295K. Exchange is no longer evident in the spectra
at 228K. Only three quartets are observed because of overlap of the
two high frequency quartets. In order to understand the spectra
completely, it is necessary to identify which amide resonance is
which, and this is done using nuclear Overhauser and two-dimensional
proton carbon correlation spectroscopy, which we will describe later.
It is sufficient to record here that it is the amide on the CMe carbon

u
ln(£)
data from
-2- lineshape analysis

-4-

-6-

-8-
data from
magnetization transfer
-10
0.0022 0.0024 0.0026 0.0028 0.003 0.0032

Figure 7.14 The Eyring plot as applied to exchange data for Me2NCHO in
DMSO/d6-DMSO. (From Mann et al (1977) /. Magn. Reson., 25, 91, reprinted
with permission.)

www.pdfgrip.com
 Systems at equilibrium 207

Figure 7.15 The 300 MHz *H NMR spectra of the vinyl diamides shown at a series of temperatures. The
solvent is d8-toluene. (From Szalontai et al. (1989) Magn. Reson. Chem., 27, 216-22, copyright (1989) John
Wiley and Sons Ltd, reprinted with permission.)

that is rotating faster. The cis isomer shows much more complex behav-
iour, though at the higher temperatures there are two corresponding
coalescence points at 355K and 333K. The quartets, however, split
further at lower temperatures, and each resonance is transformed into
a doublet of quartets due to the introduction of extra spin-spin
coupling between geminal protons on the same carbon atom. The
methylene protons have thus been rendered non-equivalent by some
further restriction in the motion of the molecule. The activation para-
meters were obtained by calculating T at the coalescence points.
To do this accurately, it is of course necessary to know the chemical
shift between the two coalescing signals. This may vary with temper-
ature, and for the present example it was found necessary to measure
the chemical shifts in the slow exchange limit over a wide range of
temperatures to ensure that the correct frequency separation had
been used.

www.pdfgrip.com
208 NMR spectra of exchanging and reacting systems

7.1.5.2 [Co4(CO)12J
In the case of systems exhibiting fast exchange, it may simply be neces-
sary to reduce the temperature sufficiently to slow down the exchange
process and cause the slow exchange limit spectrum to appear, since
this will be much more informative than the fast exchange spectrum,
which may be just a singlet. A typical example is the 17O spectrum of
the cobalt carbonyl compound, [Co4(CO)12]. In this case, the 17O NMR
spectrum was recorded rather than the 13C NMR spectrum due to
extreme broadening of the 13C NMR signals by scalar relaxation by
59
Co, see section 4.6. The carbonyl groups move around the cobalt
cluster so that they all experience the full range of chemical environ-
ments and shifts in the molecule. Spectra at several temperatures are
shown in Fig. 7.16, and it will be evident that the ambient-tempera-
ture trace is not very informative but that at -25°C all four types of
carbonyl group can be distinguished. Note that the 59Co spectrum of
this compound contains two resonances in the intensity ratio 3 :1 so
that the carbonyl exchange takes place on a cluster in which the posi-
tions and bonding of the cobalt atoms are invariant. The nucleus 59Co
is quadrupolar and the asymmetry of the environment around the
metal atoms means that the relaxation time is very short. Linewidths
are of the order of 7500 Hz but 59Co chemical shifts are very large and
so the resonances are resolved.

7.1.5.3 A multisite lineshape exchange problem, [(rf-C8H8)Ru(CO)3]

A final example of fast exchange where the structure can be deduced
only at low temperature is that of the intramolecular transfer of the
Ru(CO)3 group around the ring of the r|4-cyclooctatetraene complex,
[(T]4-C8H8)Ru(CO)3]. This compound, whose structure is shown in
Fig. 7.17, contains four distinguishable types of hydrogen atom, and
its proton spectrum should consist of four chemically shifted lines with
intensity ratios 2 : 2 : 2 : 2 . In fact, only a singlet is observed at 25°C,
as shown in Fig. 7.18. Cooling causes broadening, until, by -107°C, the
expected structure starts to emerge as the exchange is slowed down.
Interestingly, there is an obvious asymmetry in the spectrum of the
outer triplets between -107 and -115°C, and this allows us to deduce
the mechanism of the transfer of the Ru(CO)3 group.
There are five possibilities for exchange in complexes of this type:
(i) 1,2-shifts; (ii) 1,3-shifts; (iii) 1,4-shifts; (iv) 1,5-shifts; and (v) a
random shift process, where all types of shift occur with equal prob-
ability. This is shown in Fig. 7.17, where the chemically shifted
protons/carbons are numbered 1 to 4. Table 7.1 shows how the spin
chemical shifts change for each process. Examination of Table 7.1
shows that when a 1,2-shift occurs, atom 1 and I' move to positions
r or 2', i.e. for half the moves it remains in the same chemical
shift position. Atom 4 and 4' behave similarly, moving to positions
3 or 4. Contrast this with atoms 2, 2', 3, and 3' which move from their

www.pdfgrip.com
 Systems at equilibrium 209

+85 °C

+57 °C

+21 °C

-25 °C

-55 °C
O

500 400 300 6(170)

(a) (b)

Figure 7.16 The structure, (a), and the 54.25 MHz 17O NMR spectrum, (b), of [Co4(CO)12] in CDC13. The
carbonyl groups exchange positions, but this can be slowed down sufficiently by cooling to enable the four
types of carbonyl group to be seen. The reference is water. (From Aime et al. (1981) /. Am. Chem. Soc.,
103, 5920, copyright (1981) American Chemical Society, reprinted with permission.)

original position each time. Remembering that the line broadening

below coalescence is proportional to the rate of leaving the site, clearly
if the mechanism is a 1,2-shift then the signals associated with atoms
2 and 3 will be broadened twice as much as those associated with atoms
1 and 4. A similar analysis can be performed for the other mechanisms.
For a 1,4-shift the reverse occurs with the signals associated with atoms
1 and 4 which will be broadened twice as much as those associated
with atoms 2 and 3. For 1,3-shifts, 1,5-shifts and random shifts, all the

www.pdfgrip.com
210 NMR spectra of exchanging and reacting systems

Figure 7.17 The possible mechanisms of fluxionality of the Ru(CO)3 group

in [(V-C8H8)Ru(CO)3].

signals broaden equally. In principle, the 1,5-shift can be distinguished

as atom 1 exchanges exclusively with atom 4 and atom 2 with atom 3,
so the high temperature limiting spectrum should consist of two signals
in the ratio 4 :4. However this presupposes that all the other possible
mechanisms are of such high energy as not to be significant at 25°C.
Examination of Fig. 7.18 shows that between -115 and -107°C there
is clearly differential line broadening. It is the triplets which broaden
most. Strictly, this is an [ABCD]2 spin system, but at the resolution
achieved at low temperature, a simple analysis works. The doublets
arise from H1 and H4, while the triplets arise from H2 and H3. It is
therefore H2 and H3 that broaden most and the mechanism is a 1,2-
shift. The derived AG* is 32 kJ moH.
When [(7i6-C8H8)Cr(CO)3] was examined, it was found that the four
different J H NMR signals from the C8H8 ring broadened approximately
equally. We have the same fluxional processes possible as for [(r\4-
C8H8)Ru(CO)3], including random shifts. Examination of the table
above shows that the rates of movement of a specific :H or 13C from
their original position to another is equal, hence equal broadening, for
a 1,3-, a 1,5-, or a random shift. The 1,5-shift mechanism was dismissed

www.pdfgrip.com
 Systems at equilibrium 211

25°

-84°
-87° - 0.00048s

-89° 0.00105s

-93° 0.0022s
-95° 0.00275 s
-100° 0.0032s
0.0070 s
-107°
0.0115s
-108°
0.015s
-109°
0.022 s
-115°
0.12s

-120
0.20s

-128°
0.25s
-147 0 100 200
Hz
0 100 200
Hz

Figure 7.18 Variable temperature 100 MHz 1H NMR spectrum of

[(Ti4-C8H8)Ru(CO)3] in CFHC12/CF2C12. The left hand set of signals are the
experimental, while the right hand set are calculated on the basis of a 1,2-
shift mechanism. (From Cotton et al (1969) /. Am. Chem. Soc., 91, 6598,
copyright (1969) American Chemical Society, reprinted with permission.)

as it only exchanges 1 and 4 and 2 and 3 and will give rise to two
signals at the high temperature limit, rather than the observed singlet.
This left the 1,3- and random shift mechanisms, but line shape analysis
cannot differentiate between these mechanisms as both mechanisms
lead to equal broadening of the lines. This led to a disagreement
between Cotton who favoured the 1,3-shift mechanism and Whitesides
who favoured the random shift mechanism.
The argument was resolved by using saturation transfer (Fig. 7.19).
The most informative spectrum is (b) where irradiation at site 4
produces a greater reduction in the intensity at sites 2 and 3 rather

www.pdfgrip.com
212 NMR spectra of exchanging and reacting systems

Table 7.1 The predicted movements for the CH groups of a cycloocta-

tetraene undergoing different shifts of the coordinated metal group
Starting Fluxional mechanism
position 1,2-shift 1,3-shift 1,4-shift 1,5-shift Random
shift
1 r 2' 3' 4' Any
2 i V £ 3' Any
3 2 1 r 2' Any
4 3 2 i r Any
4' 4 3 2 i Any
3' 4' 4 3 2 Any
2' 3' 4' 4 3 Any
V 2' 3' 4' 4 Any

than 1. This is in agreement with a 1,3-shift which moves carbon 4 to

sites 2 and 3. The intensity at site 1 also goes down, but this is from
a double movement, e.g. 4 to 3 and then 3 to 1. A quantitative analysis
of the spectra showed that the dominant mechanism is a 1,3-shift, but
a 1,2-shift was also detected at approximately a quarter of the rate of
the 1,3-shift. This clearly demonstrates the power of the magnetiza-
tion transfer method to unravel complex exchange pathways.
Note that the intensity does not go to zero on irradiation of one
site. Once the saturated nuclei move to another site, they are no longer
being irradiated, and can recover their intensity, by re-equilibration
losing the excess of energy to the lattice, i.e. through spin lattice relax-
ation. The size of the residual signals depends on a competition
between exchange and relaxation.

7.1.5.4 A multisite magnetization transfer problem [Ir4(CO)n(PEt3)]

One of the largest multisite problems ever to have been examined is
that of carbonyl exchange in [Ir4(CO)n(PEt3)]. This compound exists
in solution as a mixture of two isomers, the major one with the PEt3
ligand orthogonal to the carbonyl bridged Ir3 face and the minor
one with the PEt3 ligand in the plane of the carbonyl bridged face
(Fig. 7.20). There are 14 different carbonyl sites. The use of selective
180° pulses demonstrated the mechanism of exchange and the rates
were determined. Fig. 7.21 shows one of many experiments. The
carbonyl region of the 13C NMR spectrum is shown in Fig. 7.21(a),
obtained with a long T so that the selectively inverted nuclei have
regained their normal magnetization. This is then a normal 13C NMR
spectrum. A second spectrum is shown in Fig. 7.21 (b) where a selec-
tive pulse was then applied to carbonyl a in the major isomer followed
immediately, 3jxs, by a general 90° pulse. The result is that the signal
of carbonyl a is inverted. A third spectrum was recorded with a wait
of 0.01 s between the 180° selective pulse and the 90° general pulse.
The spectrum in Fig. 7.21 (b) was subtracted from this third spectrum

www.pdfgrip.com
 Systems at equilibrium 213

(c)

(b)

21

(a)

140 120 100 8(13C)

Figure 7.19 The application of the saturation transfer method to the

25.15 MHz 13C NMR spectrum of [(T]6-C8H8)Cr(CO)3] in CD2C12/CH2C12 at
-15°C. (a) Normal spectrum, (b) and (c) with irradiation as shown by arrows.
(Reproduced with permission from B.E. Mann (1977) /. Chem. Soc., Chem.
Commun., 626.)

to give the spectrum in Fig. 7.21(c). Examination of Fig. 7.21(c) shows

that the magnetization of CaO has partially recovered. This is due to
two factors, exchange bringing nuclei with equilibrium magnetization
into this site, and Tl relaxation. The signals due to CAO, CdO, and
CfO have decreased in intensity due to the exchange of CaO into these
sites, bringing negative magnetization with them. In the full study, a
series of experiments were performed varying T and the site of the
selective 180° pulse, in order to map out the exchange network. It was
concluded that two exchange mechanisms are operating, a mechanism
involving a bridge-opened intermediate (Fig. 7.20(a)), and a mechan-
ism involving a jx3-CO intermediate (Fig. 7.20(b)) which interconverts
the major and minor isomers and rates were determined for each
mechanism.

www.pdfgrip.com
214 NMR spectra of exchanging and reacting systems

(a)

(b)

(c)
Figure 7.20 The two mechanisms, (a) and (b), of exchange identified for the
fluxionality of the carbonyls and interchange of isomers of [Ir4(CO)n(PEt3)].
Letters are used to denote the carbonyls and PEt3 is abbreviated to P. (c).
The lettering used to identify the carbonyls in the minor isomer in Fig. 7.21.

7.1.5.5 Solvent exchange on cations

One very wide field of interest is that of ligand exchange on complex
cations. These processes have been studied in a variety of ways, and
the rates measured for ligand exchange vary over many orders of
magnitude and depend upon the nature of the cation chosen for study.
In many instances, the rates fall, or can be made to fall, by changing
the temperature, in the range accessible to NMR studies, so that this
relatively new technique has been used to extend our knowledge of
such systems. In addition, since NMR provides the means to study
systems at equilibrium, our studies can be of the unperturbed system.
A particularly well studied system is that of solvent exchange on the
cation [A1(H2O)6]3+ in aqueous solution. This cation contains the
NMR-active nuclei 1H, 27A1 and, if isotopically enriched, 17O, all of
which have been used in its study. Such a highly charged ion is subject
to hydrolysis and so solvent exchange can be quite a complex process:
ex
[AI(H20)6]3+ + H20* ^ [AI(H20)5(H20*)]3+ + H2O

+ H+ -H+ +H + -H+

[AI(H2O)5(OH)]3+ + H2O* ~ [AI(H20)4(H2O*)(OH)]3+ + H2O

www.pdfgrip.com
 Systems at equilibrium 215

(C)

(a)
13
210 200 190 180 170 160 C/ppm

Figure 7.21 The effect of applying a selective 180° pulse in the pulse sequence,
relaxation time-'TT(selective) - T - ir/2, to the bridging carbonyl in the major
isomer of [Ir4(CO)n(PEt3)] in CD2C12. (a) The 100.62 MHz 13C NMR spec-
trum of the carbonyl region of the spectrum acquired with T > 57\. (b) The
100.62MHz 13C NMR spectrum of the carbonyl region of the spectrum
acquired with T = 3 JJLS. (c) The difference 100.62 MHz 13C NMR spectrum of
the carbonyl region of the spectrum after subtracting spectrum (b) from a
spectrum with T = 0.01 s. (Note that this is at higher gain.) (Reproduced with
permission from Mann et al (1989) /. Chem. Soc., Dalton., 889.)

The hydrolysis reaction causes exchange of hydrogen atoms, which

in the present case is very fast, although the backward reaction is much
faster than the forward reaction and the hydrolysis constant is only
10~5 (pK = 5). The oxygen exchange is much slower, though two path-
ways are possible via either the hydrolysed or the non-hydrolysed ions.
We would like to know the rates of exchange of both hydrogen and
oxygen, whether the latter is influenced by the acidity of the solution
and so whether the hydrolysed species contributes to the exchange,

www.pdfgrip.com
216 NMR spectra of exchanging and reacting systems

and, if possible, the mechanism by which a water molecule replaces a

complexed water molecule.
The rate of proton exchange can be studied using 1H NMR spec-
troscopy. If a solution of an aluminium salt is cooled, this slows down
the rates of exchange. If the temperature is low enough, and this can
be achieved either by adding an antifreeze such as acetone or by using
a very concentrated salt solution, then two proton resonances are
observed due to solvent and complexed water, as shown in Fig. 7.22.
Integration of the resonances together with knowledge of the
aluminium concentration allows the hydration number (the number of
water ligands attached to the cation) to be determined, and this is
approximately equal to six, the method being less precise than the
isotope fine-structure method depicted in Fig. 2.8. The chemical shift
between the two resonances is 4.2 ppm, and we can calculate that
coalescence will occur when the lifetime of the protons is about
1.8xlO~ 3 s. We can therefore study the rate of exchange in dilute
solution by measuring the proton relaxation times as a function
of temperature and acidity, and this gives a protolysis rate constant of
0.79 x 105 s'1 at 298K. The 27A1 resonance is also influenced by this
process since the hydrolysed cation has a much broader resonance
than that of the non-hydrolysed species. The latter has a width of some
2 Hz, but the fast exchange mixes in the rapid quadrupolar relaxation
of the hydrolysed cation and gives lines that are generally in the range
10-20 Hz wide. Note that, in this case, the exchange of one atom (H+)
results in apparent exchange of a second (27A1).
The rate of oxygen exchange is best studied using the 17O resonance.
Here, the rate of exchange is known to be slower, so that resonances
should be observed for bound and bulk water molecules. In fact, the
chemical shift between the two is very small and, because of the
quadrupolar broadening of the lines, they are not resolved. Separating
the two resonances has been done in a variety of ways, but the one
used for the most comprehensive studies was to add a paramagnetic
cation, Mn2+, for which the whole water molecule exchange rate is

HoO

[AI(OH2)n]3+

Figure 7.22 The 60 MHz *H NMR spectrum of 3 mol H A1C13 at -47°C The
water complexed by the cation is seen 4.2 ppm to high frequency of free, bulk
water. (After Schuster and Fratiello (1976) /. Chem. Phys., 47, 1554, with
permission.)

www.pdfgrip.com
 Systems at equilibrium 217

very fast. The water not attached to aluminium thus is subjected

rapidly to the very large magnetic field of the electron spins on the
Mn2+, and has its T2 very much reduced and so its linewidth increased,
to the extent that its signal disappears in the baseline of the spectrum.
Only the oxygen bound to A13+ is observed and, provided the exchange
on A13+ is slow enough, its relaxation is determined entirely by its own
natural relaxation time and exchange lifetime (equation (7.1)). The
problem in making such measurements with quadrupolar nuclei is that
the relaxation time in the absence of exchange is not known and that
it is very temperature-dependent. The relaxation time, or linewidth,
thus has to be measured over a range of temperature wide enough
that some data are acquired in the region where the exchange broad-
ening is negligible. A set of measurements is shown in Fig. 7.23, where
it will be seen that the temperature dependence of relaxation falls into
two regions: lower temperatures, where the linewidth is dominated by
the quadrupolar mechanism, and higher temperatures, where it is
dominated by exchange. Extrapolation of the quadrupolar influence
in principle allows the exchange rates to be determined as a function
of temperature. In fact, because of the curvature, it is difficult to esti-
mate the slope of the line giving kex with the required precision, and
a separate experiment was needed in order to increase the range of
values of &ex observed. This was done using an injection or stopped-
flow technique in which an aluminium salt solution in H2O was injected
rapidly into acidified H2O enriched in 17O, both solutions being at low
temperature around 256K. Mn2+ was also present in the 17O water.
Thus, as the water exchanged with the aluminium solvation sphere,
the 17O signal of the mixture increased in intensity and the exchange
rate could be obtained by plotting this as a function of time. This
isotope exchange type of experiment is a common way of investigating
systems that are at equilibrium. One is effectively perturbing the
system but the perturbation is slight. The combined linewidth and
stopped-flow results are shown in Fig. 7.23(b). In the case of aluminium
it is possible to suppress hydrolysis completely by the addition of suffi-
cient acid. Thus the data gave thermodynamic parameters for oxygen
exchange on the non-hydrolysed cation: k = 1.29 s~\ A//* = 84.7 kJ
moH, AS* = +41.6 J K^moH. In addition, by measuring the exchange
rates as a function of pressure at constant temperature, it was possible
to obtain an activation volume for the oxygen exchange: AV* =
+5.7 cm3 moH. We will discuss such experiments in a little more detail
later, but the inference is that the transition state during the exchange
involves a short-term increase in total volume. This occurs if the ex-
changing ligand leaves before the entering solvent molecule takes its
place; in other words, it is a dissociative mechanism of exchange. In
fact, the A V* term is a little smaller than theory would demand if there
were full dissociation, and the mechanism probably involves inter-
change during the dissociation, commonly called the Id mechanism.
In the case of the gallium cation, [Ga(H2O)6]3+, the hydrolysis of the
ion is too strong to be suppressed completely by the addition of acid,

www.pdfgrip.com
218 NMR spectra of exchanging and reacting systems

10.0

9.0

tf 8.0

^
7.0

6.0
2.5 3.0 3.5 4.0
(a)

10.0

5.0
£
.0^

^ 0.0

-5.0

2.5 3.0 3.5 4.0

103
(b)
Figure 7.23 Plots of natural logarithms of the 17O relaxation rates 1/T2 of 17OH2 bound to the aluminium
cation, as a function of the reciprocal temperature. The T2 values were obtained from the 17O linewidths.
(a) How the linewidth decreases as the temperature is increased (from the right-hand side of the plot)
and how these data can be extrapolated satisfactorily into the exchange perturbed region. The onset of
detectable exchange causes the plot to curve upwards as the temperature is increased beyond 333 K. (b)
This contains the same data but includes some fast injection data also, (•). (From Merbach et al. (1985)
Helv. Chim. Acta, 68, 545, with permission.)

www.pdfgrip.com
 Systems at equilibrium 219

and in this case oxygen exchange on both hydrolysed and non-hydrol-

ysed species has to be measured. It is found that
ir - / irC l + k2
/Cex
~ [H+]
where ^ = 403 s'1 and k2 = 14 mol s'1, both at 298K.

7.1.5.6 Effect of exchange on coupling patterns

The behaviour of t-butyllithium demonstrates the averaging effects of
exchange upon coupling constants and coupling multiplicities. The
compound was studied using 13C spectroscopy and the spin-spin
coupling of this nucleus to the lithium nuclei. The common isotope of
lithium is 7Li, which has a quadrupole moment sufficiently large that
its quadrupole relaxation times are quite short, and so any coupling
is not well resolved. On the other hand, the less abundant isotope 6Li
has a quadrupole moment that is some 50 times smaller, which is
indeed the smallest among all the elements, and which thus behaves
much more like al =1/2 nucleus than like a quadrupolar one. Further,
6
Li is easily available via the nuclear industry and so is proving to be
a useful tool for the study of organolithium compounds by NMR
methods. The structure of t-butyllithium is shown in Fig. 7.24, and the
13
C spectrum of the a-carbon atoms of a sample highly enriched in 6Li
is shown in Fig. 7.25 as a function of temperature. This compound
forms tetramers in cyclopentane solution in which the lithium atoms
are arranged in a tetrahedral cluster with the alkyl groups bonded to Figure 7.24 The structure
three lithium atoms by a multicentre bond and so situated over each of tetrameric t-butyl-
lithium. The lithium atoms
face of the tetrahedron. The tetramer is, however, fluxional and the are arranged in a tetrahe-
expected spectrum is only observed at low temperatures. The a-carbon dron with the alkyl groups
atom of the t-butyl group is coupled equally to its three nearest-neigh- bonded equally to three
bour 6Li atoms with a coupling constant of 5.4 Hz. For 6Li, / = 1 so lithium atoms and so posi-
that a seven-line multiplet results, with intensities in the ratio tioned above a face of the
tetrahedron.
1 : 3 : 6 : 7 : 6 : 3 : 1 , and this is the spectrum observed at -88°C in
Fig. 7.25. The coupling to the distant lithium nucleus is undetectably
small. As the temperature is increased, fluxion of the structure occurs,
and at 26°C this is fast enough for the a-carbon atoms to be influ-
enced equally by all four 6Li nuclei, which produces a nine-line multi-
plet with relative intensities of 1: 4 :10 :16 :19 :16 :10 : 4 :1. In
addition, the close and long-distance coupling constants are mixed and
the coupling constant of 4.1 Hz has three-quarters of its low-temper-
ature value. Note that there can be no intermolecular exchange since
this would involve breaking the coupling path and would collapse the
multiplet to a singlet.

7.1.5.7 High-pressure NMR spectroscopy

While it is normal to study the dynamics of a system as a function of
temperature in order to obtain the thermodynamic properties of the

www.pdfgrip.com
220 NMR spectra of exchanging and reacting systems

experimental calculated

o^ °r*
£.0 u 10,000s

-7°C 70s 1

-9°C 48s 1

-16°C 13s1

-19°C 7s-1

pp or*
-OO U OS'1

I I I T I I I
20 0 -20 3J 0 -3J
/Hz

Figure 7.25 The 22.6 MHz 13C NMR spectrum of [(CH3)3C6Li]4 with the effect
of the protons removed by double irradiation at different temperatures in
cyclopentane solvent. Only the resonances of the a-carbon atom are shown.
(From Thomas et al (1986) Organometallics, 5, 1851, copyright (1986)
American Chemical Society, reprinted with permission.)

changes observed, it is equally valid to use pressure as the variable,

as this may give further insight into the mechanisms of the reactions
observed. Often, if a full study is required, both temperature and pres-
sure have to be varied.

www.pdfgrip.com
 Systems at equilibrium 221

In order to work at high pressure, a special probe has to be

constructed. There are several variations on the structures of such
probes, depending upon exactly what type of experiment it is required
to undertake. A typical assembly consists of a pressure containment
vessel of a size such that it can be placed in the probe space of an
NMR spectrometer and with electrical leads through to the sample
coil and with an inlet for the pressure fluid. This vessel contains
a sample space surrounded by a detector coil and a temperature
measuring device such as a platinum resistance thermometer. The
sample is enclosed in a small glass tube with a close-fitting plastic
piston to transmit pressure to the sample. Resolution of such probes
can be good, even without sample spinning, and temperature control
is excellent since the heat transfer medium is the pressurizing fluid.
Not surprisingly, the effect of pressure on the hindered rotation of
amides has been quite deeply studied. 1H spectra of the compound NJN-
dimethyltrichloroacetamide are shown in Fig. 7.26 as a function of pres-
sure at constant temperature. The expected doublet is coalesced near
ambient pressure but splits as the pressure is increased. Thus the pres-
sure decreases the rate of rotation, though there is relatively little
change above pressures of 200 MPa. Now, the rate of hindered rotation
in such compounds can be expressed in the usual transition-state form
k oc Q-EofRT
where EQ is some energy barrier. The pressure then has no direct effect
on the amide rotation but operates via a change in the viscosity of the
solvent, which increases with pressure. We thus have to write
k = F(-r])e-Ev/RT
where T] is the shear viscosity and F(j]) is a function of j] that does
not vary linearly with either pressure or viscosity. It is possible to
calculate F(t]) by making a variety of assumptions, and the results
suggest that there is some form of coupling between the solvent motion
and the rotation of the Af,7V-dimethyl groups.
If the exchange is very slow, so that the resonances of the various
species present are well resolved, it may not be possible to obtain
exchange rates, but the intensity of the resonances may vary, and this
indicates a displacement of the equilibrium between the components,
which can be caused by changes in either temperature or pressure. In
the latter case, change occurs if there is a difference in volume between
the species in equilibrium. As an example of such a study, we take
the equilibrium:

www.pdfgrip.com
222 NMR spectra of exchanging and reacting systems

100 Hz
M W

150MPa 400 MPa

100MPa 300 MPa

SOMPa 250 MPa

0.1 MPa 200 MPa

Figure 7.26 The !H spectra of the TV-methyl resonances of Me2NC(O)CCl3

dissolved in pentane at 282.3 K as a function of applied pressure. (Jonas et
al (1990) /. Chem. Phys., 92, 3736; reprinted with permission.)

The value of the equilibrium constant K increases with pressure, so

that the p-diketone is favoured at high pressures and its volume thus
must be smaller than that of the enol. The difference in partial molar
volumes is found to be -4.5 cm3 moH and is believed to arise because
of the extra volume occupied by the hydrogen-bonded ring formed by
the enol.
The exchange of ligands on cations is also studied by high-pressure
NMR. Here we will consider the two complexes of beryllium,

www.pdfgrip.com
 Systems at equilibrium 223

[Be(OSMe2)4]2+ and [Be{OC(NMe2)2}4]2+. Both ligands bond via the

oxygen. Exchange is studied in the presence of excess ligand (c.f. the
A13+ water system described above), but with organic ligands it is
possible to dilute the system with an inert solvent such as deuteriated
acetonitrile, CD3CN, and vary the ratio between the concentrations of
free ligand and complex and so obtain data about the order of reac-
tion. By 'inert' we mean here a solvent that will not compete with the
ligand for sites on the cation but which may nevertheless form
complexes in an even more inert solvent. Such systems are most easily
studied using 1H spectroscopy of the methyl groups. Two signals are
observed due to bound and free ligand, and the exchange rates can
be found from the shapes of the spectra, which for much of this work
demands the calculation of full spectral envelopes. In the case of the
(Me2N)2CO complex, it was found that the rate of exchange was inde-
pendent of the concentration of free ligand, so that the reaction is first
order and this implies that the ligand must dissociate from the cation
before exchange can take place. In the case of the OSMe2 complex,
the exchange rate did depend upon the concentration of ligand, so
that the reaction is second order, which implies that a ligand must
associate with the [Be(OSMe2)4]2+ before another ligand will leave.
Variable-temperature determinations then permit the thermodynamic
parameters A//* and AS* to be obtained as well as the rate constants,
and these are given in Table 7.2. The different signs of the entropy
values are also in accord with the different reaction mechanisms in
the present case, though there are many examples where such data
are ambiguous. Determination of the exchange rates at different pres-
sures, however, produces a very convincing demonstration of the
reality of this difference. Some spectra are shown in Fig. 7.27, together
with calculated envelopes, from which the exchange rates were
obtained. All the spectra are on the slow exchange side of coalescence,
but it will be evident that the spectra of (Me2N)2CO are nearest coales-
cence at the lowest pressure, whereas the opposite is true for the
Me2SO. This difference occurs because a dissociation produces a
temporary increase in volume, which is discouraged by an increase in
pressure, whereas association produces a temporary decrease in total
system volume. The variation of the rate constant with pressure allows
a volume change of activation to be obtained from

Ay-J»r(«ln£) T

Table 7.2 Thermodynamic data for solvent S exchange on [BeS4]2+

Rate at AW AS* AY*
S 298 K (kJ mol-1) (J K-1 mol-1) (cm3 mot1)
Me2SO 213 moH s-1 35.0 -83.0 -2.5
(Me2N)2CO 1.0 s-1 79.6 +22.3 +10.5

www.pdfgrip.com
224 NMR spectra of exchanging and reacting systems

(Me2N)2CO
Observed Calculated Observed Calculated
P/MPa /c/s 1 P/MPa

202 359.0

106 323.5

298.3

3.4 2.2 3.4 2.2 3.3 2.1 3.3 2.1

1
8( H)/PPM

Figure 7.27 The 200 MHz 1H spectra of acetonitrile-d6 solutions containing Be(II) and (NMe2)2CO (left)
or Me2SO (right), obtained as a function of pressure. The actual spectra are on the left and the calculated
envelopes are on the right. (From Merbach (1987) Pure Appl Chem., 59, 161, with permission.)

These values are also given in Table 7.2. AV* is large and positive for
(Me2N)2CO, and this is in accord with a fully dissociative mechanism
for this ligand exchange. AV* is negative for Me2SO but its value is
appreciably smaller, and this allows us to conclude that, while the mech-
anism is associative, there must be some tendency for the bound ligand
to move away from the cation and that the mechanism is, more pre-
cisely, association and interchange of ligands near the cation, which lim-
its the extent of the possible volume contraction were full association
to occur. Note also the difference between the meaning of volume
changes in this and the previous example. Here, the total volume before
and after reaction is unchanged, and we observe only the adjustment
necessary to allow reaction to proceed. There is thus no change in the
position of equilibrium with pressure, in contrast to what happens when
reaction produces a permanent volume change.

7.2 REACTION MONITORING OF SYSTEMS NOT AT

EQUILIBRIUM

The previous example was just such an experiment, in which we were

monitoring a physical change in a system, albeit under somewhat

www.pdfgrip.com
 Reaction monitoring of systems not at equilibrium 225

extreme conditions. In fact, NMR is used extensively to monitor

what is happening during chemical preparations and is particularly
useful because starting materials, intermediate, side products and main
products may have resolved signals and the progress of a reaction may
well be fruitfully examined without any previous purification.

7.2.1 Inter ion alkyl exchange

The slow equilibration of mixtures is easily followed by NMR. For
instance, if the aluminate salts NaAlMe4 and NaAlEt4 are dissolved
together in solution, it is found that the ligands scramble to form the
mixed salts NaAlMe/lEt4^I. If a coordinating solvent is used, the solutes
ionize to a greater or lesser extent to give Na+ and the regularly tetra-
hedral [A1R4]~, in which the 27A1 relaxation time can be sufficiently
long to permit spin-spin coupling to the protons of the alkyl groups.
In order to measure the rate of alkyl interchange, the two salts were
dissolved in the non-coordinating solvent benzene and the coordinating
solvent (Me2N)3PO, added in varying amounts. Immediately following
dissolution, a sample of the solution was placed in an NMR spec-
trometer and the proton spectrum of the Al-Me groups measured as
a function of time (Fig. 7.28). Any effect of the 27A1 was removed by
double irradiation. The Al-Me signal then appears as a sharp singlet
initially, but with time new signals are seen to emerge to low frequency
corresponding to anions with one, two or three ethyl groups replacing
the methyl substituents. By measuring the intensities of the resonances
as a function of time, the rate of ligand scrambling is obtained.
Interestingly, this is found to be a function of the concentration ratio
of (Me2N)3PO/Na+, and it is believed that this indicates that the scram-
bling takes place in aggregates containing two anions and sodium
cations solvated by the (Me2N)3PO.
Reactions taking as little as 60 s can be studied in this way, and even
faster processes can be followed using stopped-flow NMR spec-
troscopy. Two solutions containing the species it is proposed to react
are held in reservoirs in the magnetic field so that the magnetization
of the nuclei to be studied can reach its equilibrium value. Two power-
driven syringes are then used to force measured amounts of the solu-
tions into the NMR sample tube within a time of a few milliseconds.
The collection of FID signals is initiated at the same time and each
response stored separately in memory for later transformation into
spectra. Reactions taking as little as 5 s can be studied in this
way, one example having already been given for the exchange of water
on A13+.

7.2.2 Oxidative addition to platinum(II)

Our next examples illustrate two reaction processes and how NMR
can provide details about reactions not easily obtainable in other ways.
The first example comes from the realm of organometallic chemistry.

www.pdfgrip.com
226 NMR spectra of exchanging and reacting systems

— ^-—~ -rf
40 min 6h 22 h 46 h 54 h 78 h 102h

Figure 7.28 27
The 100 MHz ^-NMR spectra of the Al-Me protons with the
effect of the A1 removed by double irradiation of a 1:1 mixture of NaAlMe4
with NaAlEt4 in benzene with added (Me2N)3PO, as a function of time. (From
Ahmad et al (1984) Organometallics, 3, 389, copyright (1984) American
Chemical Society, reprinted with permission.)

An oxidative addition was carried out on the platinum complex

[PtMe2{Me2PC6H3(OMe)2}2] using PhCH2Br, with the objective of
forming the octahedral complex [PtMe2{Me2PC6H3(OMe)2}2(PhCH2)
Br], where Ph is a phenyl group. The proton-decoupled 31P spectrum
of the starting material is shown in Fig. 7.29 and is the typical 1 : 4 : 1
triplet. The initial product has a similar spectrum, though with different
shift and coupling constant but, even after 1.5 h reaction, there is
already a significant amount of another product present. After three
days this becomes a predominant spectral feature and is seen to consist
of a 1 : 4 : 1 set of AB sub-spectra. The initial product has isomer-
ized to give non-equivalent phosphines, which have a small cis P-P
coupling. The full structure can then be deduced from the methyl
resonance pattern in the !H spectrum, where one large and one small
coupling to 31P are observed, so indicating the presence of a trans and
a cis methyl group. The reaction that takes place is

where R = C6H3(OMe)2 and Ph = phenyl.

www.pdfgrip.com
 Reaction monitoring of systems not at equilibrium 227

(a)-

(b)

(c)

Figure 7.29 Three 40.5 MHz 31P spectra obtained with double irradiation of
the protons so as to simplify the spectra. The upper spectrum (a) is that of
the starting material and is distinguished by squares in all three spectra. The
central trace (b) was obtained 1.5 h later and shows the presence of another
substance in which the two phosphorus nuclei are still equivalent and which
is distinguished by triangles. After three days of reaction (c) both these spec-
tral patterns are much diminished and a third, more complex, pattern (x) has
taken their place. There is now a triplet of AB patterns, which shows that
the two phosphorus nuclei in the complex are no longer equivalent. (Example
supplied by Professor B.L. Shaw.)

7.2.3 The forced hydrolysis of aluminium salts

The second example examines the nature of the species contained in
highly hydrolysed aluminium salt solutions. Aluminium salt solutions
are highly acidic and will, for instance, dissolve many metals with the
evolution of hydrogen. If we add sodium carbonate solution slowly
so as to ensure that there is no precipitate formed, we find that we
can add up to 1.25 moles per mole of A13+ and form a solution whose

www.pdfgrip.com
228 NMR spectra of exchanging and reacting systems

stoichiometric composition is [A1(OH)25X05], where X is the anion

present. The rest of X forms the sodium salt and CO2 is evolved. The
problem has been to discover what exactly is present in such solutions,
and 27A1 NMR has been able to provide useful new information about
the nature of the molecular species formed. It was known that a cation
[A1O4A112(OH)24(H2O)12]7+ could be crystallized as the sulfate salt
from the solutions to which the maximum amount of carbonate had
been added. The structure of this cation has been obtained. It has a
central aluminium atom tetrahedrally coordinated by four oxygen
atoms and surrounded by 12 aluminium atoms, which are octahedrally
coordinated and are linked by OH bridges and the oxygen atoms
coordinated to the central Al. They also carry a terminal water ligand
(Fig. 7.30). The four-coordinate aluminium atom is in a regular envi-
ronment and so gives a narrow 27A1 resonance, which is diagnostic of
the presence of this molecular species. The octahedral aluminium
atoms are in a highly distorted environment, and so have a short relax-
ation time and a very broad resonance some 8000 Hz wide, which can
only be observed under rather specialized conditions and so is normally
not detectable. The system has been much investigated by pH titra-
tion methods but does not seem to be very amenable to these methods,
although different workers had proposed the presence of some 24 total
species in order to explain their results. Later workers, however,
seemed to be agreed that only the tridecameric cation was present.
Unfortunately, in partly neutralized solutions, the 27A1 NMR gives a

7+

Figure 7.30 Structure of the tridecameric hydrolysed aluminium cation,

[A1O4A112(OH)24(H2O)12]7+, with the hydrogen atoms omitted. This consists
of four groups of three A1O6 octahedra, which share an apical oxygen atom
and three edges to form a triangular structure. The apical oxygen is also coor-
dinated to the central Al atom, which is thus in a tetrahedral environment
since there are four A13 groups. Each A13 group is attached to the three others
via double OH bridges. There are also three OH bridges in each triangular
cluster. The remaining oxygen coordinated to the Al is fully protonated and
is a water ligand.

www.pdfgrip.com
 Reaction monitoring of systems not at equilibrium 229

spectrum containing three resonances, none very broad. These can be

seen in the traces of Fig. 7.31, where the highest frequency line is
assigned to the tridecamer, the lowest frequency line to unconverted
A13+ or [A1(H2O)6]3+ and a rather broader line just to high frequency
of the latter which originates from an oligomer. There is no doubt that
this species exists but it is not detected in the pH titrations. A major
difference between the two techniques is that pH titrations are carried
out in dilute solutions whereas, because NMR is a rather insensitive
technique, it requires much higher concentrations. (This difference is
currently less important because of the existence of the very high-field
spectrometers coupled with Fourier transform techniques.) The
obvious question was asked: Was this difference in interpretation a
concentration effect? It was found that, if a solution that had been
hydrolysed so as to contain very little tridecamer and much oligomer
was diluted, then the relative concentration of the tridecamer increased
and that of the oligomer decreased, and that, with sufficient dilution,
only the tridecamer could be detected. It was also possible to follow
how the change occurred by simply diluting a solution with water and

[AI(OH)4]

Al) / ppm

Figure 7.31 The 104.2 MHz 27A1 NMR spectra of a partially hydrolysed aluminium salt solution as a func-
tion of time immediately following dilution with water. The 3+
aluminium concentration before dilution was
1.0 M and this was reduced to 0.1 M by dilution. [A1(OH2)6] is the chemical shift reference. A capillary
containing [A1(OH)4]~ was used to provide a concentration reference. (From Akitt and Elders (1988) /.
Chem. Soc., Dalton Trans., 1347, with permission.)

www.pdfgrip.com
230 NMR spectra of exchanging and reacting systems

observing the 27A1 resonance as a function of time. A set of spectra

are shown in Fig. 7.31. The oligomer resonance is reduced in inten-
sity immediately upon dilution; a change that cannot be followed by
this technique and which requires comparison of the spectra before
and immediately following dilution. The tridecamer concentration then
increases regularly over about 19 h. Simple dilution of these solutions,
then, causes profound changes in composition and this explains some
of the problems of the earlier workers. It is also remarkable that such
a weak driving force as a quite moderate change in concentration
should give rise to such a complex molecule as the tridecamer. It is
suggested that the oligomer is, in fact, a mixture of species that have
structures related to those of fragments of the tridecamer, one possi-
bility being a fused A13O13H18+ unit such as forms one of the trian-
gular, flat faces of the tridecamer.

7.2.4 A technological application: crystallization of amorphous

polyethylene
The proton spin relaxation time in amorphous, solid polyethylene,
while shorter than that in liquids, is nevertheless long enough to obtain
spectra by normal methods. When the material forms extended-chain
crystals, the relaxation time becomes much shorter because the chain
motion becomes more restricted, and this difference can be used to
separate the signals of the two forms of the polymer. The technique
used to do this was to apply the pulse sequence used to determine
T2, i.e. a 90° - T - 180° - T - echo sequence in which T is made equal
to 300 fjus, too long to observe a response from the crystalline material
but short enough to see an echo from the amorphous component. The
echo intensity decreases with the time of crystallization as the amount
of amorphous materials decreases. There is an induction period when
no change in echo intensity is detected, and then the intensity falls
regularly as crystallization proceeds. The rate increases as the temper-
ature is decreased, and increased pressure allows crystallization to
occur both at higher temperatures and at higher rates.

7.3 QUESTIONS

7.1. The 1H decoupled 31P NMR spectrum of [Rh{P(OMe)3}5]+ is an

A3B2X pattern at very low temperature and A5X at room temper-
ature. Is the mechanism of exchange intra- or inter-molecular?
7.2. You have two isomers in equilibrium. Their concentrations are
in the ratio 10:1. There is a contribution to the linewidth of a
1
H NMR signal due to the major isomer of 1 Hz from exchange.
What is the exchange contribution to the linewidth of the minor
isomer in the 1H and in the 13C NMR spectra, assuming that
the signals are below coalescence? Accepting that the height of
a signal is inversely proportional to its linewidth and ignoring

www.pdfgrip.com
 Questions 231

non-exchange contributions to linewidth, what is the ratio of

heights of the NMR signals due to the major and minor isomers?
7.3. Given the following rates as a function of temperature for the
inversion on ds-decalin, calculate A//* and AS* for the dynamic
process using the Eyring equation.

k (s~>) T (K) T(K) T(K)

0.128 203 25.8 246 1000 282
0.57 214 60 253 1890 292
2.57 226 120 261.5 3450 300
11.1 237 350 271 7300 312

7.4 At -60°C, the 13C signals of the a-C5H5 group of [Fe(7i5-C5H5)

(a-C5H5)(CO)2] (see margin) are broadened by intramolecular
chemical exchange. The linewidths of C2 and C5 are 11 Hz and
of C3 and C4 are 6 Hz, while the linewidth in the absence of
exchange is 1 Hz. Deduce the mechanism of exchange.
Calculate AG* for the carbon leaving the C2 and C5 positions.
7.5. Bullvalene undergoes the degenerate Cope rearrangement. At
-80°C, the 13C NMR spectrum shows the expected four signals
at 8 21.01, 31.01, 128.37, and 128.54, in intensity ratio 3 : 1 : 3 : 3 .
Above -50°C, the signals broaden due to the degenerate Cope
rearrangement. The signals at 8 21.01, 31.01, and 128.54 broaden
equally, but the one at 8 128.37 broadens by only one third of
the amount. Use this information to assign all the 13C NMR
signals.

etc

7.6. In the 13C NMR spectrum of ds-decalin (see margin), at 226 K

there is slow inversion of the two six-membered rings producing
exchange C1'5 <-> C4'8 and C2'6 <-> C3'7. If the signal due to C3'7 is
presaturated for long enough so that the signal due to C2'6 reaches
an equilibrium value, its height is now 37% of its original value.
Given that Tl of C2'6 and C3'7 is 0.66 s, calculate the rate of inver-
sion of c/s-decalin and AG* at 226K.
13
7.7. There are two C NMR signals of equal intensity separated by
1 ppm, which exchange. The separation of the signals is temper-
ature independent and the natural linewidth is negligible. At
25.2 MHz, the coalescence temperature was determined as 25°C,
while at 100.62 MHz, the coalescence temperature was deter-
mined as 42°C. Explain why the coalescence temperature is
magnetic field dependent and determine AG*. Assuming that
AS* = 0, calculate the linewidth of the signal(s) at 0°C and 60°C
at both magnetic fields.

www.pdfgrip.com
www.pdfgrip.com
 Multiple resonance
and one-dimensional
pulse sequences
8
Until the advent of computer controlled NMR spectrometers, there
was a limited selection of NMR experiments available. However, with
computer control, it is possible to carry out a much wider selection of
experiments.
Many NMR experiments may be summarized by the sequence:
Relaxation - Preparation - Evolution - Mixing - Acquisition. During
the relaxation period, the nuclear spins are allowed to return towards
their Boltzmann equilibrium. In principle, the relaxation period should
be in excess of 57\ for the slowest relaxing signal. In practice, we are
impatient, and normally wait between Tl and 27\. It is during the
preparation period that the spins of the nuclei are tuned to give us
the information required. There is then an evolution period, which is
often related to a coupling constant. It is usually necessary to apply a
refocusing 180° pulse in the middle of the mixing period to make sure
that the resulting spectrum can be phased. One or more other pulses
are applied during the mixing period. At the end of the mixing period,
the FID is acquired, frequently with decoupling. In the more compli-
cated experiments, a string of pulses are applied, often to two or more
different nuclei. A single such experiment is a one-dimensional NMR
experiment. The data are collected in a single time dimension as the
FID and are then transformed into the frequency dimension to give
a conventional NMR spectrum. (Yes, strictly speaking there are two
dimensions, intensity and time or frequency, but such spectra are
described as one-dimensional, referring to the time dimension.) A
second time dimension is introduced by varying a time during the
mixing period. A series of FIDs are collected, each with a different
value of time, a technique which gives rise to two-dimensional NMR
spectroscopy. We will see in Chapter 9 that this provides a valuable
tool for assigning signals.
In the present chapter we will be examining a range of one-dimen-
sional NMR experiments. This is not intended to be an exhaustive
coverage, but many of the important techniques will be examined. In
all cases, the techniques give information on the connectivity between
nuclei in compounds, and hence on the structure of compounds.

www.pdfgrip.com
234 One-dimensional NMR spectroscopy

It is useful at this stage to introduce a commonly used nomencla-

ture to indicate when decoupling is used. It is common to use irradi-
ation to remove coupling or perturb a second nucleus. This is normally
indicated by using curly brackets. Thus a 13C{!H} NMR spectrum
is one where the 13C NMR spectrum is observed with irradiation
of protons. In this case, there will be broadband :H decoupling. In
Fig. 8.1(b) we have a 1H{1H] NMR spectrum, where the 1H NMR spec-
trum is recorded with irradiation at one of the hydrogen sites. Equally,
Fig. 8.2(b) shows a ^p1?} NMR spectrum, where an individual 31P
site is being irradiated.

8.1 DECOUPLING DIFFERENCE SPECTROSCOPY

Difference spectroscopy is a powerful technique used to show which

signals change during a given experiment. A simple example is decou-
pling difference spectroscopy. In a typical experiment, the computer
is instructed to take a series of NMR spectra with the decoupler set
at different frequencies for each spectrum corresponding to different
resonances. A reference is included with the decoupler set well away
from any signal (Fig. 8.1(a)), which gives the single resonance spec-
trum. This reference spectrum is then subtracted from a spectrum
obtained where some signal is decoupled (Fig. 8.1(b)). In the differ-
ence spectrum (Fig. 8.1(c)) the multiplets coupled to the irradiated
signal show up as a mixture of fully coupled (negative) and decoupled
(positive). The signals due to protons which are unchanged by the irra-
diation should subtract completely. In practice this does not happen
perfectly. In part this is due to instrumental instability, but there
are two other major contributions. Firstly, there will be nuclear
Overhauser effects, see section 8.2. Secondly, there will be Bloch-
Siegert shifts, see section 6.1. As a result of the Bloch-Siegert shifts,
it is common to carry out the experiment with a relatively small value
of B2 so that signals are not fully decoupled, and the difference spec-
trum consists of the coupled spectrum sitting on a broad hump.
Decoupling difference NMR spectroscopy is valuable as a means of
dis-entangling overlapped signals, where it is important to derive
coupling constants. If the required information is connectivity and
assignment, then it is normal to use the two-dimensional technique,
COSY, see section 9.2.
Decoupling difference NMR spectroscopy is not restricted to ^pH}
NMR spectroscopy, but can be applied to any pair of nuclei. An
example of ^f31?} decoupling difference NMR spectroscopy is given
in Fig. 8.2. The compound is [Fe(Ti5-C5H5)(PPh3){l,2-(PMePh)2C6H4}]
[B{3,5-(F3C)2C6H3}4]. Due to the crowding of the molecule, rotation
about the Fe-PPh3 bond is slow on the NMR timescale. Due to the
crowding in the molecule, the near neighbour anisotropy of the phenyl
rings has spread these signals over a l.Sppm range, making the reso-
lution of separate signals easier. The result is that all five PPh groups

www.pdfgrip.com
 Decoupling difference spectroscopy 235

(C)

(D)

(a) =
2.7 2.6 2.5 2.4 2.3 2.2
1
8( H)/ppm

Figure 8.1 A partial 400 MHz JH NMR spectrum of carvone in CDC13. (a)
The reference spectrum, where the decoupler was set at 8 10. (b) The spec-
trum where the decoupler was set on the signal at 8 6.77. (c) The difference
spectrum, where spectrum (a) is subtracted from spectrum (b). Responses are
observed only for the multiplets at 8 2.45 and 2.28 which have changed on
decoupling. The unchanged coupled spectrum gives the negative signals, while
the decoupled spectrum gives positive signals.

give separate !H NMR signals. Selective 31P decoupling permits the

identification of the PPh3 phosphorus signal and the three phenyl
groups attached to it. This is illustrated in Fig. 8.2. Fig. 8.2(a) shows
the normal XH NMR spectrum of the aromatic hydrogen atoms. When
the PPh3 31P nucleus is decoupled (Fig. 8.2(b)), it is obvious that a
large coupling is removed from the signals at 8 6.01 and 6.22 leaving
these signals as doublets, and hence these are ortho-protons of two of
the three phenyl groups, but the location of the third set of ortho-
protons is difficult to discern. There is also the removal of a small
coupling from the signals at 8 6.77 and 7.04, leaving triplets of inten-
sity 2, making these meta-protons. The identification of the remaining
signals is more readily made in the difference spectrum in Fig. 8.2(c).
The missing ortho-protons are at 8 7.27 and can be recognized by
comparison of the appearence of the difference signal with those
assigned at 8 6.01 and 6.22. The other signals are more difficult to
assign, but responses are detected at 8 7.54, 7.48, and 7.24.

www.pdfgrip.com
236 One-dimensional NMR spectroscopy

(c>-

(b)-

(a)--'
7.8 7.6 7.4 7.2 7.0 6.8 6.6 6.4 6.2 6.0
1
H/ppm

Figure 8.2 A partial 400 MHz *H NMR spectrum of [Fe('n5-C5H5)(PPh3){l,2-(PMePh)2C6H4}][B{3,5-(F3C) 2

C6H3}4] in CD2C12. (a) The reference spectrum, (b) The spectrum where the decoupler was set on the 31P
NMR signal of PPh3. (c) The difference spectrum, where spectrum (b) is subtracted from spectrum (a).
The coupled spectrum gives the positive signals, while the decoupled spectrum gives negative signals.

8.1.1 Selective population transfer

As artefacts occur in decoupling difference spectra due to Bloch-
Siegert shifts, it would be advantageous to minimize these shifts by
using very low powered decoupling. This can be achieved by selective
population transfer. We illustrate the principle of this technique using
an AX spin system (Fig. 8.3). The relative populations of the energy
levels follow from the Boltzmann distribution. Relative to the pp
energy level, the a(3 and PCX energy levels have an excess population
of 8N, while the cm energy level has an excess population of 28N.
When the decoupler is turned on at the frequency of the X, acx-cxp
transition, the populations of these two energy levels are made equal.
Note what this does to the population differences associated with the
A-transitions. For the aa-(3a transition the population difference
decreases from SN to 0.58N, i.e. the intensity of the signal halves. For
the ap-pp transition, the population difference increases from 8N to
1.58N, i.e. the intensity of the signal increases by 50%. The result is
that a doublet changes from being a 1:1 doublet to a 0.5 :1.5 doublet,
and the changes are easily detected in a difference spectrum.

www.pdfgrip.com
 The nuclear Overhauser effect 237

(y-^PP

A transition X transition

6N ap—^ ^^pa8N

X transition A transition

(a)
^ aa 25N

A transition X transition

1.55N ap—^ \-pa5N

* X transition A transition
irradiate \, /
(b) ^ 'aa 1.56N

Figure 8.3 The energy level diagram for a homonuclear AX spin system, (a)
In the absence of irradiation, (b) In the presence of irradiation at the aa-a(3
X transition.

8.2 THE NUCLEAR OVERHAUSER EFFECT

The nuclear Overhauser effect (NOE) provides information on con-

nectivity through space. This contrasts with connectivity information
from coupling constants, which normally only operate through a few
bonds. The technique is valuable in determining the stereochemistry
and conformation of molecules. For example, the stereochemistry of a
highly substituted alkene can be found or the cis-trans relationship
between ligands in square-planar metal complexes can be determined.
In outline the experiment is simple. When one nucleus is irradiated,
the intensities of the signals due to other nuclei which are close in
space change. In the case of 1H with 1H irradiation, the neighbouring
protons can increase in intensity by 50%. In the case of 13C with 1H
irradiation, the 13C signals can increase in intensity by 199%.

www.pdfgrip.com
238 One-dimensional NMR spectroscopy

The NOE arises directly from dipole-dipole relaxation, see section

4.2. The mechanism of operation can be explained for a nucleus /
being relaxed by a nucleus S by dipole-dipole relaxation (Fig. 8.4).
There are four energy levels for this spin system, aa, a£, (3a, and (5(5.
The transition probabilities between each energy level are given by

w mfriV*2 5(5 + I K
W
°~ 120nV 1 + K - cos)2Tc2 ( }

w MtfriV*2 s(s + IK
Wl ( }
= SOnV 1 + o^V

w MD^iSs^2 5(5 +1)T C

^ " 20^r6 1 + (co, + o>s)2Tc2
2 l J

where r is the distance between the two nuclei and a>T and oos are the
frequencies of the nuclei. The subscripts to W refer to the change in
the total spin quantum number when the transition described by W
occurs. The transition associated with W0 is a zero quantum transition,
that associated with W1 is a single quantum transition and that asso-
ciated with W2 is a double quantum transition.
The theoretical maximum NOE, iimax, when nuclei, /, are observed
and nuclei, /, are irradiated is given by
= . TfcC^-Wq)

^max y,(W0 + 2W, + W2) ^ *

The signal intensity is then 1 + Timax. In order for 7]max to be obtained,
the relaxation pathway has to be exclusively dipole-dipole. If any other
relaxation mechanism operates, then the observed iq is related to inmax,
Tl and TIDD by
^max^l
^1=-^—
MOD

Figure 8.4 The relaxation pathways for two nuclei.

www.pdfgrip.com
 The nuclear Overhauser effect 239

This equation is a very valuable tool to enable us to derive TIDD for

nuclei such as 13C. The treatment becomes more complex for !H, as
in most molecules there are many protons and there are many path-
ways for relaxation.
In the extreme narrowing limit, equation (8.4) simplifies to
= JYs. (8.5)
^Imax j^

When the extreme narrowing limit does not apply, r\max is dependent
on both frequency and TC (Fig. 8.5).
A qualitative understanding of the origin of the NOE can be under-
stood by examining the AX spin system with one of the protons
undergoing double irradiation (Fig. 8.6). The relative populations of
the energy levels follows from the Boltzmann distribution. Relative
to the pp energy level, the ap and Pa energy levels have an excess
population of 8N, while the aa energy level has an excess population
of 28N. When the decoupler is turned on at the X frequency, irradi-
ating both lines, transitions are induced between X spin states and the
populations of the connected energy levels are equalized. Hence
the populations of the pa and pp energy levels become M(0 + SN) =
M8N. Similarly, the populations of the ap and aa energy levels become

13
C{1H}
2 -,

1 -

o-

-1 --
3
l
-2-

-3-

-4-
1 5 1
N{H}
-5- i i i i i i i
-12 -11 -10 -9 -8 - 7 - 6 -5 -4
log(Tc)
Figure 8.5 The variation of Timax with correlation time for J, and
pH). The values have been calculated for B0 = 9.4 T.

www.pdfgrip.com
240 One-dimensional NMR spectroscopy

6N pa8N

(a)

irradiate

1.58N pa 0.58N

irradiate
(b)

Figure 8.6 The energy levels for a homonuclear AX spin system, (a) The
relative populations are given for each energy level at equilibrium, (b) The rela-
tive populations are given immediately after applying a radiofrequency
decoupling field at the X nucleus.

M(SN + 28N) = %SN. This is in the absence of relaxation. If we now

take the case of the extreme narrowing limit, and examine equations
(8.1) - (8.3), it can be seen that W2 will be the largest term. The W2
process will try to return the population difference of the energy levels
aa and pp from 8N towards 28N. In practice, at equilibrium, in the
presence of irradiation of X, the population difference becomes %SN
as WQ and Wl do make a contribution. The effect of the radiation plus
relaxation is to increase the population difference between the aa and
Pa and between the ap and pp energy levels from 8N to %SN with a
corresponding increase of 50% in the intensity of the A transitions.
In contrast, when COTC is large, Wl and W2 become unimportant and
WG is dominant. The result is to equalize the population difference
between the ap and pa energy levels in the AX spin system with X

www.pdfgrip.com
 The nuclear Overhauser effect 241

irradiation. The consequence of the X irradiation is then to equalize

the populations of all the energy levels, with irimax - -1 and all the
signals vanish!

8.2.1 The experimental aspects of the ^l^Hj nuclear Overhauser

effect
The NOE is very valuable in giving through-space connectivity infor-
mation. Although theoretically we can expect enhancements of up to
50%, in practice the enhancements are frequently <5%. This makes
them difficult to detect. The low value arises from a number of causes.
1. The NOE only arises through dipole-dipole relaxation. Any other
relaxation pathway reduces the magnitude of the NOE. For XH, the
most common competing pathway is paramagnetic relaxation due
to the presence of paramagnetic transition metal ions or O2. This
is a particular problem with small molecules, < 200 D.
2. Examination of Fig. 8.5 shows that at 9.4 T, when the correlation
time falls below 10~10 s, the value of Timax falls, becoming negative
at 8.8 x 10~10 s. This can happen because of larger molecules, viscous
solvents, or the presence of solids.
3. The NOE builds up by dipole-dipole relaxation. In a small mole-
cule, riDD can be in excess of 10 s. In most molecules, rlDD arises
from a number of H-H dipole-dipole interactions, not a single one.
In the NOE experiment, we are using individual H-H dipole-dipole
interactions, and each one will have a TIDD contribution with a time
constant considerably larger than the measured T1DD. As a conse-
quence, the NOE can take many tens of seconds to build up to its
maximum value. This is particularly a problem when the distance
between the nuclei is substantial.
There are a number of procedures which are employed to facilitate
the detection of these weak changes in intensity.
1. If the decoupler were to be left on during acquisition, it would be
very difficult to distinguish between changes due to decoupling and
those due to the NOE. The answer is simple. NOE builds up
and decays over a period of many seconds. Decoupling is instant.
The pulse sequence given in Fig. 8.7 is used. The signal is irradi-
ated for the required period of time, the decoupler is switched off,
an observing pulse is applied and the FID acquired.

0
CO
D
Q.

decouple

Figure 8.7 The basic pulse sequence for the measurement of an NOE.

www.pdfgrip.com
242 One-dimensional NMR spectroscopy

2. The magnitude required for B2 for irradiating the proton for an

NOE experiment is very small. This raises a problem. How do we
irradiate a multiplet, say a doublet? If the decoupling frequency is
placed in the middle of the doublet, the power is so low, that neither
line is perturbed. If the decoupling frequency is placed on one line
of the doublet, selective population transfer occurs, see section
8.1.1. The result is that the signal which is coupled to the doublet
undergoes substantial changes in the intensity of the lines of its
multiplet, and this can obscure an NOE. The solution is very simple.
Rather than irradiating at a single frequency, the irradiation
frequency is cycled through the multiplet, pausing briefly at each
line of the multiplet. This technique is now the normal method of
irradiation in an NOE measurement.
3. By using the first two procedures, NOEs of 0.1% can be detected
in favourable cases. The problem is that artefacts still occur. These
can be considerably reduced by using field gradient pulses.

8.2.2 Some applications of the ^t1!!} nuclear Overhauser effect

So far we have only discussed the NOE in the presence of strong,
broad-band irradiation or the generalized NOE that occurs if one tran-
sition is weakly irradiated. If the whole resonance of a proton is irra-
diated, whether it be a singlet or a multiplet due to spin-spin coupling,
this will cause intensity changes to the resonances of other nuclei that
are spatially close. The enhancements possible are quite small: ^H/^H
means that T] = 1/2 only. It is usual then to obtain such spectra by
using the difference mode, i.e. spectra are obtained with and without
double irradiation, the FIDs subtracted and the difference Fourier
transformed to give a spectrum that ideally contains responses only
from those resonances perturbed in intensity by the double irradiation.
The NOE experiment is excellent for assigning protons. An example
is given in Fig. 8.8. The 1H NMR spectrum of tris-(2-carboxalde-
hyde)triphenylphosphine can readily be partially assigned as the signal
of the aldehydic proton, H7, at 8 10.56, H3 and H6 are at 8 8.02 and
6.90, though we do not know which is which, and H4 and H5 at 8 7.56
and 7.44, though again, we do not know which is which. The aromatic
signals are assigned on the basis that the signals due to H3 and H6
show only one 3/HH of around 7 Hz, while H4 and H5 show two such
couplings. The problem is to assign H3 or H6 and then H4 and H5 can
be assigned by decoupling experiments. H3 was assigned by use of the
NOE from the formyl proton, H7. On irradiation of H7, the signal at
8 8.02 shows an 11% enhancement, and must be H3.
An example is shown in Fig. 8.9 of a molybdenum complex that
exists in solution as a mixture of two isomers, one with the phenyl
group of the CHPh group pointing towards the cyclopentadienyl ring
and the other with it pointing away. NOE difference spectroscopy
enables not only the signal of the two isomers to be assigned but also
those of the methyl signals between 8 2.5 and 3.0. In Fig. 8.9(b), the

www.pdfgrip.com
 The nuclear Overhauser effect 243

(a)

10.0 9.0 8.0 7.0

1
(b) H/ppm

Figure 8.8 (a) The 400 MHz !H NMR spectrum of tris(2-carboxaldehyde)

phenyl phosphine in CDC13. (b) A difference spectrum showing the enhance-
ment of the signal at 8 8.02 after irradiation of the formyl proton at 8 10.56.
(Reproduced from Whitnall et al. (1997) /. Organomet. Chem., 529, 35, copy-
right (1997), with permission from Elsevier Science.)

cyclopentadienyl protons of one molecule are pre-irradiated. Some of

the phenyl proton resonances show a positive NOE. This shows that
a phenyl group of this molecule is pointing towards the cyclopentadi-
enyl ring. Contrast this with Fig. 8.9(e) where the cyclopentadienyl
protons of the other molecule are pre-irradiated and no NOE is
observed from the phenyl protons. At the same time an NOE is
observed into H8/ and CH3lr enabling these signals to be assigned to
the molecule with H8/ pointing towards and the phenyl group pointing
away from the cyclopentadienyl ring. The result in Fig. 8.9(e) does not
permit a clear distinction between CH^' and CH37/, but this comes
from the spectrum in Fig. 8.9(d) where H8/ is pre-irradiated and an
NOE is observed from the phenyl ortho-protons on the same carbon
atom as H8/, C5H512/, and the adjacent CH37', showing that the methyl
signal giving an NOE in Fig. 8.9(e) is by default CH3P. It is also notable
that the ortho-protons at 8 6.95 which show an NOE when H8/ is pre-
irradiated are different from those at 8 7.1 which show an NOE when
C5H512 is pre-irradiated. The relative assignment of CF^1 and CH37
comes from Fig. 8.9(c), where H8 is pre-irradiated and both the phenyl
groups on the same carbon atom and CH37 show an NOE.
NOE difference spectroscopy is not restricted to proton-proton
NOE measurements but can be used between different nuclei.

www.pdfgrip.com
244 One-dimensional NMR spectroscopy

(e)-

(d)-

(c)-

(b)-

(a)-
8 7 6 5 4 3
1
H/ppm
Figure 8.9 The 400 MHz 1H NMR spectra of [(Ti5-C5H5)Mo(CHPhNMeCPh=NMe)(CO)2] in CDC13. (a)
The normal spectrum and (b) to (e), the NOE difference spectra with pre-irradiation at (b) 8 4.9, H12, rf-
C5H5, (c) 8 5.6, H8, (d) 8 5.5, H8/, and (e) 8 5.4, H12', Ti5-C5H5, with the shift scales progressively displaced
to the right to give a clearer picture. (Reproduced from Brunner et al (1983) /. Organomet. them., 243,
179, copyright (1983), with permission from Elsevier Science.)

8.2.3 Some applications of the X[1H] nuclear Overhauser effect

The heteronuclear NOE can provide a very useful increase in signal
strength. For the observation of 13C, for instance, when protons in the
molecule are double-irradiated, the ratio is 1.99 and 1 + r^max is effec-
tively 3. This is a very useful gain in intensity for such low receptivity
nuclei, and is one good reason why 13C spectra are routinely obtained
with proton broad-band irradiation. Some values of T]max are shown in
Table 8.1 for various combinations of nuclei. We should note that some
nuclei have magnetogyric ratios with negative sign and that nqmax is
then negative also. In such cases, provided Timax is more negative than
-1 (15N and 29Si in the table), then the signal is inverted relative to
the normal one and the total enhancement is less than T|max. If the
relaxation mechanism contains other contributions than the direct
dipolar one, then Timax will be reduced, and it is not uncommon to find

www.pdfgrip.com
 The nuclear Overhauser effect 245

Table 8.1 Maximum nuclear Overhauser effects for several pairs of nuclei
Irradiate >1i 19p
13 19 p 29 31 ! 13 19p
Observe *H C 15N Si P H C
0.5 1.99 -4.93 0.53 -2.52 1.24 0.47 1.87 0.5
i+^max 1.5 2.99 -3.93 1.53 -1.52 2.24 1.47 2.87 1.5

that 1 + T]max ~ 0 and all signal is lost. In such cases, it is necessary to

suppress the NOE in some way, and this can be achieved either instru-
mentally or by adding a paramagnetic salt (a relaxation agent), which
completely dominates the relaxation process. Alternatively, INEPT or
DEPT can be used, see section 8.3.

8.2.4 Application to 13C NMR spectroscopy

Figure 8.10 illustrates the changes brought about in the 13C spectrum
of carvone by broad-band proton double irradiation. In the absence
of irradiation, all the 13C-1H spin coupling interactions are observed.
Thus the carbon atoms of the methyl groups, seen to low frequency,
are split into quartets by the protons bonded directly to the carbon
atom, and these lines are further split by longer-range couplings to the
other ring protons. Only the quaternary carbon atoms give an apparent
singlet, though there will be some long-range splitting. A triplet is seen
in the centre of the trace at 8 77 due to the solvent CDC13 whose 13C
resonance is coupled to the 2H nucleus. The CH and CH2 resonances
are doublets and triplets, respectively, due to the directly bonded
hydrogen nuclei, and with longer-range two-bond coupling to the
vicinal hydrogens, which produces fine structure in the resonances. Not
all the resonances of these multiplets can be seen in the spectrum illus-
trated as the conditions were chosen to produce a noisy baseline. The
difference in intensity between the phenyl quaternary carbon signal
and the remaining phenyl carbon signals is due to the difference in
relaxation time. That of the quaternary carbon atom is long and, at
the pulse repetition rate used to obtain the spectrum, has insufficient
time to recover its z magnetization after each pulse and so its signal
is reduced in amplitude. The spectrum is much simplified by double
irradiation, and each type of carbon nucleus appears as a singlet. In
Fig. 8.10(b), the spectrum was recorded without NOE and Fig. 8.10(c)
with NOE. The enormous improvement in the signal-to-noise ratio is
immediately evident.
Double irradiation thus permits us to obtain 13C spectra at natural
abundance routinely with good signal-to-noise ratio but such spectra
contain reliably only the chemical shift information. Often, this is suffi-
cient, but there are instances where it may be useful to observe the
coupling patterns, and if this could be done while retaining the NOE
then much accumulation time could be saved. Alternatively, it might
be useful to know the correct intensities, and for this we need to

www.pdfgrip.com
 246 One-dimensional NMR spectroscopy

(c)

(b)

(a)
240 220 200 180 160 140 120 100 80 60 40 20 0
13
C/ppm
Figure 8.10 The 100.62 MHz 13C spectra of carvone, dissolved in CDC13. (a) Obtained without proton irra-
diation, (b) With irradiation at 400.13 MHz with NOE suppression, (c) With irradiation at 400.13 MHz
and full NOE, otherwise the spectrometer conditions were identical. A large improvement in signal-to-
noise ratio is obtained with double irradiation, especially when the NOE is permitted.

decouple but suppress the NOE. Such techniques may be necessary if

we are to observe quaternary carbon atoms with very long relaxation
times. If we are happy to contaminate the sample, then the relaxation
agents can be very useful since these effectively suppress the NOE
and also shorten the 13C relaxation times, so permitting a shorter delay
between pulses. If contamination cannot be tolerated, or if the natural
parameters of the molecule are required, then special pulse sequences
are used, which depend upon the different timescales needed to set
up decoupling and NOE. Thus we can have the following:
1. Full decoupling without NOE. This is achieved by decoupling only
for the short time needed to collect the FID and then waiting suffi-
cient time to allow any small NOE population changes to return
to normal. The short pulse of B2 does not allow appreciable build-
up of the NOE (Fig. 8.11).
2. No decoupling but full NOE. Here the decoupling power is left on
for sufficient time for the NOE to build up to its full value and is
then switched off while the FID is collected (Fig. 8.12). This

www.pdfgrip.com
 The nuclear Overhauser effect 247

decouple
'H

13, FID

d1 P1
Figure 8.11 The NOE suppression pulse sequence, dl is a relaxation delay of
at least 57\. pi is a 90° pulse. Decoupler timing to allow a fully decoupled
spectrum to be obtained without any distortion of intensity due to the NOE.
There will be a small build-up of NOE during the B2 pulse (typically 0.5 s)
and this must be allowed to die away completely before the next pulse. The
long delay time means also that the nuclear magnetization has decayed fully
before the next 90° pulse and there are no intensity distortions due to relax-
ation effects.

decouple

13, FID

d1 P1
Figure 8.12 The pulse sequence used to record a 1H coupled 13C NMR spec-
trum with full NOE. dl should be 5T{ or at least considerably longer than
the acquisition time. Decoupler timing is arranged to allow a coupled spec-
trum to be obtained but with the benefit of the full NOE increase in inten-
sities. There will be very little fall-off in NOE during the short time needed
to collect the FID data. This technique allows up to nine times reduction in
time over the basic method used to obtain a non-decoupled spectrum.

technique allows normal spectra to be obtained more quickly. If

bad overlap of the multiplets occurs, a different approach may be
needed, but normally it is possible to distinguish quartets, triplets,
doublets and singlets by a comparison with the decoupled spec-
trum and so to assign resonances to CH3, CH2, CH and quaternary
carbons. It is well worth taking the trouble to carry out this exper-
iment because the NOE enhancement permits a time reduction in
accumulation of as much as nine times.
NOE difference measurements can be very useful in determining the
spatial relationship between !H and 13C. This is particularly valuable
where the 13C does not have a directly attached proton so that its
dipole-dipole relaxation arises from more remote protons. The assign-
ment of connectivity between 13C and directly attached protons is far
more easily carried out using two-dimensional NMR spectroscopy, see

www.pdfgrip.com
 248 One-dimensional NMR spectroscopy

Chapter 9. The technique is illustrated by considering a problem that

was solved using 13C-1H NOE measurements. When 2-methylnaphtho-
1,4-quinone is treated with Me2CN2 a product was obtained which is
an adduct incorporating a Me2C and a Me2CN2 group.
There were four possible products. It was very easy to unambigu-
ously identify the single hydrogen on the sp3 carbon from it being a
singlet of intensity 1 at 8 2.37. The 13C NMR spectrum is shown in
Fig. 8.13(a), while the NOE difference spectrum with pre-irradiation
(8.1) of this proton is shown in Fig. 8.13(b). Examination of the spectrum
shows a clear NOE into the 13C at 8 195.0, which must be due to the
C=O carbon, showing that the structure must be (8.1) or (8.4). There
are also NOEs to the carbon atoms at 8 90.0 and 90.5. Now, the CH
group in (8.1) is attached to the C=O and one other carbon atom,
whereas in (8.4) it is attached to C=O and two other carbon atoms.
Hence the correct structure is (8.4).
A variety of other experiments is also possible. Off-resonance decou-
pling will give multiplet structure with apparently reduced coupling
constants, which will reduce overlap in crowded spectra and permit
assignment of the carbon spectrum and its correlation with the proton
(8.2)

(8.3)

(a)
200 150 100 50 0
(8.4) 13
C/ppm
13
Figure 8.13 The 100.6 MHz !
C NMR spectra of (8.4) in CDC13. (a) The 13C
NMR spectrum with no H irradiation, (b) The difference spectrum obtained
by pre-irradiating the proton at 8 2.37 at low power to develop an NOE in
the 13C NMR spectrum and subtracting spectrum (a), (c) An expansion of the
signals at 8 90.0 and 8 90.5 in (b). (Reproduced with permission from Aldersley
et al. (1983) /. Chem. Soc., Chem. Commun., 107.)

www.pdfgrip.com
 The nuclear Overhauser effect 249

spectrum. Such spectra obtained at several spot frequencies will indi-

cate at which JH frequency a given 13C multiplet becomes a singlet,
and so permit an even more precise correlation between the two. Such
experiments are, however, time-consuming and are now better carried
out by two-dimensional techniques.

8.2.5 Detailed relaxation mechanisms by 13C NOE and Tl

measurements
The relaxation time of 13C nuclei can be obtained by the inversion-
recovery method outlined in section 6.1.2.1, though this is not the only
method available. The NOE factors nq are obtained by comparing inte-
grals with and without double irradiation, and if the relaxation mech-
anism is dipole-dipole then TJ should have a value near to 2. If T) is
less than this, then other mechanisms are present. An alternative
method of measurement of 7\ and NOE in the same experiment is
also possible and is done by a method that is a combination of the
two previous ones. The NOE is allowed to build up for a time t prior
to producing the FID and this is collected while irradiation continues.
B2 is then switched off, the system allowed to equilibrate and the
process repeated. A series of experiments are run with different values
of t (Fig. 8.14). Such experiments are known as dynamic NOE
measurements and give both the Overhauser enhancement for each
nucleus from the intensity of the signals when t = 0 and t = °°, and
the value of Tl from the plot of the change in signal intensity as a
function of t. The spectra obtained for such an experiment with
biphenyl are shown in Fig. 8.15, and the results for biphenyl and some
for toluene are illustrated in Table 8.2. The CH ring carbon atoms of
biphenyl have an NOE only slightly less than 3 and so must be almost
totally relaxed by the directly bonded hydrogen atoms. The carbon

Figure 8.14 The dynamic NOE experiment. This is essentially a combination of the two experiments
described by Figs 8.11 and 8.12. The decoupler is first gated ON to allow the NOE to build up. The
amount of NOE increases if t is increased, reaching a maximum when t>9Tv At the end of time t, the
FID is produced by the 90° pulse and collected with continuing irradiation. This is removed when the
data collection is finished and the NOE allowed to decay to zero. Sufficient FIDs are collected to give
the required signal-to-noise ratio, and the experiment is repeated with different values of t, but always
the same number of FIDs are collected, so that spectral intensities can be compared directly.

www.pdfgrip.com
 250 One-dimensional NMR spectroscopy

0 100 200 300 400/Hz

Figure 8.15 Fourier transform 13C spectra of biphenyl in CDC13 solution excited by 90° read pulses sepa-
rated by intervals of 300 s in order to ensure the re-attainment of equilibrium after each pulse. The spectra
are stacked as a function of the time t of proton irradiation prior to the application of the pulse. (From
Freeman et al. (1972) /. Magn. Reson., 7, 327, with permission.)

nuclei 2 and 3 have similar relaxation times but the T{ of carbon 4 is

appreciably shorter, and this is because the CH bond is not reoriented
relative to BG by rotations around the axis of the molecule, whereas
this motion reduces the correlation time of the carbon atoms 2 and 3,
if only slightly. Much more marked is the long 7\ of the quaternary
carbon nucleus, which relies on the long-distance effect of the protons.
Its NOE, however, is reduced and indicates that only half the rate of
relaxation is due to this dipole-dipole interaction and that another
influence is operating, which we can suggest is that of chemical shift
anisotropy. The 7\ values for the ring carbon atoms of toluene show
similar behaviour, though the actual relaxation times are longer, as
would be expected for a smaller, more rapidly rotating molecule. The
relaxation time of the methyl carbon nucleus has, of course, to be
multiplied by 3, the number of directly bonded protons, NH, if it is to
be meaningfully compared with the 7\ values of the CH carbon nuclei,
and this gives a value of 48 s. This is really very long when compared
with the ring CH carbons, and it is not surprising to observe that the
NOE is also small, with only 30% of the relaxation rate arising from
the dipolar mechanism. In this case, the dipolar effect must be reduced
because of a very short correlation time, much shorter than for the
rest of the molecule. This will be the free, almost unhindered, rotation
of the methyl group, which in addition causes spin rotation relaxation

www.pdfgrip.com
 The nuclear Overhauser effect 251

Table 8.2 Experimental spin-lattice relaxation times 7\, and nuclear

Overhauser effect, T|
Carbon atom Tj (s) /y/o = 1 + 7] rj
(a) Biphenyl
C4 3.4 ± 0.6 2.72 ± 0.16 1.72
C3 5.4 ±0.7 2.80 ± 0.21 1.80
C2 5.2 ± 0.8 2.76 ±0.19 1.76
Cl 54 ± 0.4 2.00 ± 0.16 1.00
(b) Toluene
CH3 16 (7VVH = 48) 1.61 0.61
Cl 89
C2 24
C3 24
C4 17

and so accounts for the small NOE. One can calculate that the part
of the relaxation which is purely dipolar, 3riDD, has the value 160s,
which implies a reduction of TC for the methyl group of around eight
times compared with the rest of the molecule.
A more complex and more informative example concerns the relax-
ation behaviour of the carbon nuclei in [Cr(CO)5(NC5H5)]. This is a
a-bonded complex in which one CO ligand in [Cr(CO)6] has been
replaced by an N-bonded pyridine molecule. The chemical interest in
such molecules arises because it is uncertain what sort of energy
barriers exist to the rotation of the Cr(CO)5 moiety relative to the
aromatic ligand. One way of probing the intramolecular motion is to
measure the relaxation rates of the 13C nuclei. Provided the mecha-
nism of relaxation is known, then correlation times of individual atoms
can be calculated. Two types of mechanism are expected to coexist in
such molecules: dipole-dipole for the protonated carbon atoms and
chemical shift anisotropy relaxation for the CO carbon atoms. The
first mechanism can be confirmed if the NOE is high, and low NOE
establishes its absence for the CO atoms. Another experiment is,
however, needed if we are unequivocally to establish the presence of
CSA relaxation, and this is achieved by measuring 7^ and T] at different
magnetic fields, though the rates of motion must be such that the
extreme narrowing condition is met at the higher field and spectro-
meter frequency
The Tv values were in this case obtained by a saturation-recovery
technique in which the 13C spin populations are equalized by the appli-
cation of a series of closely spaced 90° pulses and the recovery of the
signal intensity monitored as a function of time. The NOE was
determined by comparing signal intensities obtained with continuous
decoupling with those using gated decoupling of the type depicted in
Fig. 8.11. CDC13 was used as solvent and solutions were degassed
to ensure that there was no contribution to relaxation from solvent
nuclei or dissolved paramagnetic oxygen. The two spectrometer
frequencies used were 125.7MHz (11.7 T, 1H resonates at 500 MHz)

www.pdfgrip.com
 252 One-dimensional NMR spectroscopy

and 50.29 MHz (4.7 T, 1H at 200 MHz). Certain assumptions had to

be made in order to calculate the chemical shift anisotropy of the
carbonyl carbon atoms. Since the trans CO is on the molecular axis,
its relaxation is not affected by rotation around this axis since the CSA
will be axially symmetric. The same comment applies to the relaxation
of the ^-carbon atom of the pyridine since axial rotation does not
reorient the C-H bond. One can therefore assume that these two
carbon atoms have the same correlation time of motion, that of the
end-over-end tumbling of the complex. The correlation time for the y
carbon is then calculated from its TIDD using equation (4.3) and this
value is used to calculate a,, - cr± for CO using equation (4.6). The full
results are shown in Table 8.3. First we see from the chemical shifts
of the carbonyl carbons that the cis and trans carbonyl groups are very
similar, so that we can assume similar chemical shift anisotropy in
calculating the correlation times of the cis carbonyl carbons. The values
of Tl9 jRj (= l/7\) and T| are given for each carbon at both magnetic
fields. For the nuclei of the ring carbon atoms, the TJ values are quite
substantial, and there is only a relatively small decrease in relaxation
time with increase in magnetic field. The mechanism is predominantly
dipolar, and we can extract /?IDD w^h reasonable precision, and so the
correlation times. In the case of the carbonyl carbons, r\ is essentially
zero, and there is little long-range dipolar interaction, 7\ is very field-
dependent and 7"1CSA can be calculated from the values obtained at
each field and, thus, the correlation time of the cis carbonyls. It is
evident that the rate of movement of the cis carbonyl groups is about
half that of the reorientation of the plane of the pyridine ring, so that

Table 8.3 Relaxation and NOE data for the 13C nuclei in two chromium carbonyl complexes

50.29 MHz 125.7MHz

Atom Tj(s) iRj (s-1) 77 Tj(s) R, (s-1) n *1DD RICSA TC (PS)
(ppm)
(a) [Cr(CO)5(NC5H5)]
a 155.3 7.52 0.133 1.7 5.86 0.171 1.4 0.166 0.017a 5.5 ± 0.9
P 124.8 7.01 0.143 1.6 4.47 0.224 1.2 0.123 0.014a 6.0 ±1.0
137.1 2.44 0.410 1.6 2.16 0.463 1.3 0.319 0.102a 16.3 ± 2.8
cis-CO 220.7 10.3 0.097 0.0 2.63 0.38 0.0 0.000 0.338 Assumed
same as
above
trans-CO 214.3 25.6 0.039 0.1 3.96 0.253 0.0 0.000 0.143 12.8 ± 2.8
6
(b) [(7! -C6H6)Cr(CO)3]
CH 92.7 11.7 0.085 1.8 10.8 0.093 1.4 0.070 0.021a 3.6 ± 0.6
CO 232.8 31.3 0.032 0.1 6.61 0.151 0.1 0.005 0.142 6.3 ± 1.4
a
These values probably embrace contributions from several mechanisms. /?1DD is calculated from the results at both
frequencies and the average is taken. The calculated ^1DD + ^ICSA should equal R{ found by experiment at the higher
frequency, but the averaging process destroys the equality. The values calculated for #1CSA apply to the higher frequency
only. The accuracy of the experimental T} values is approximately ± 7% and the error in TQ is ± 0.2.
Source: Data adapted from Gryff-Keller et al (1990) Magn. Reson. Chem., 28, 25, copyright (1990) John Wiley and Sons
Ltd, reprinted with permission.

www.pdfgrip.com
 One-dimensional multipulse sequences 253

the two halves of the complex are rotating independently around the
N-Cr bond, with the rates of motion determined primarily by inter-
actions of the groups with the solvent. The data for the (T]6-arene)
complex, [(7i6-C6H6)Cr(CO)3], are also given in Table 8.3. They
resemble closely the previous set of data, though the correlation times
calculated are much shorter and the speed of rotation of the benzene
is high indicating its very small interaction with the solvent.

8.3 ONE-DIMENSIONAL MULTIPULSE SEQUENCES

Several pulses can be used during the preparation period to extract

information from NMR spectra. This opens up a wide range of exper-
iments. Here we are only going to examine a selection of the more
useful pulse sequences.

8.3.1 J Modulation, JMOD, and the Attached Proton Test, APT

This pulse sequence (Fig. 8.16) is used almost exclusively to edit 13C
NMR spectra. The resulting NMR spectrum has the 13C signals from
the C and CH2 groups pointing up and CH and CH3 13C signals
pointing down (Fig. 8.17). This sorting of 13C signals is a valuable aid
to assignment. There is no convention as to whether the C and CH2
signals are phase adjusted to be positive or negative, and it is usual
to have to decide the phase used from the solvent signal. In Fig. 8.17,
the 13CDC13 signal at 8 77 is positive, showing that the C and CH2
signals have been phased positive.
How does this pulse sequence work? The spin dynamics are shown
in Fig. 8.18. Figure 8.18 uses the rotating frame, and starts where a
90° pulse has been applied to bring the 13C magnetization into the x'y'
rotating frame. The diagrams have been constructed on the basis that
the 13C NMR signal is on resonance, so that the magnetization vectors
do not rotate with respect to the rotating frame. The simplest case is

13C

d1 p1 d2 p2 d2

Figure 8.16 The pulse sequence for the / modulation pulse sequence, dl is
for relaxation, pi is a 90° pulse, d2 is 1/J(13C,1H), and p2 is a 180° refocusing
pulse.

www.pdfgrip.com
254 One-dimensional NMR spectroscopy

-i—|—i—|—i—|—i—|—i—|—i—|—i—|—i |—i | i | i | i |—i—r

240 220 200 180 160 140 120 100 80 60 40 20 0
13
C/ppm
Figure 8.17 The application of the /-modulation pulse sequence to the 100.62 MHz 13C NMR spectrum
of carvone in CDC13. Note that the 13C and 13CH213C NMR signals are displayed as positive signals, while
the 13CH and 13CH3 signals are displayed as negative signals.

a 13C signal with no 1H coupling (Fig. 8.18(a)). As the 13C NMR signal
is on resonance, it does not matter how long we wait after the initial
90° pulse as the signal will remain with its original alignment. However,
in a normal 13C NMR spectrum, many of the signals will be away from
resonance, and will precess relative to the rotating frame at different
frequencies. Thus, by the time we collect the FID, they will be very
much out of phase and enormous phase corrections will be needed to
rephase the spectrum. This is avoided by placing a 180° refocusing
pulse in the middle of the sequence so that nuclei are all in phase
again when the collection of the FID is initiated, see section 6.3. This
is a general feature of these pulse sequences.
Let us now consider a 13CH group, again on resonance (Fig. 8.18(b)).
The rotating frame is at the frequency of the centre of the 13CH
doublet. One line of the doublet will be +J/2 Hz away and the other
will be -JI2 Hz away. The result is that the 13C nuclei attached to 1H
nuclei with a or p spins will be precessing at rates different from that
of the rotating frame. After 1/27 s, one group will have precessed +90°

www.pdfgrip.com
 One-dimensional multipulse sequences 255

Figure 8.18 The behaviour of the nuclear spins in the x'y' rotating frame
during the /-modulation pulse sequence. In order to simplify the picture, it
is assumed that the 13C NMR signal is on resonance, (a) A 13C not bearing
1
H. (b) A 13CH group, (c) A 13CH2 group, (d) A 13CH3 group. The 180° pulse
and second 1/J wait are omitted as they follow from the refocusing pulse
description in section 6.3.

and the other -90°. After l//s, they will have processed ±180° and
refocused. The decoupler is now switched on and the doublet collapses
to a singlet, which precesses at the same frequency as the rotating
frame. The signal from the 13CH group is 180° out of phase compared
with the non-proton bearing 13C.

www.pdfgrip.com
256 One-dimensional NMR spectroscopy

A similar argument applies to the 13CH2 and 13CH3 groups. After

I// s, the central line of the 1: 2 :1 triplet of the 13CH2 group has not
moved, and the outer lines, separated by / Hz from the central line,
have rotated 360° and are realigned with the central line. Decoupling
then produces a singlet with the same phase as that of a non-proton
bearing 13C. After IIJ s, the inner pair of lines of the 1: 3 : 3 :1 quartet
of the 13CH3 group have rotated 180°, like the 13CH doublet, and the
outer lines, separated by 3//2 Hz from the central line, have rotated
540° and are realigned with the inner two lines. Decoupling then
produces a singlet, 180° out of phase compared with the non-proton
bearing 13C.
The choice of a value for / is normally easy. For most carbon atoms,
1 13
J( CH) is in the range 125-170 Hz. A compromise of 145 Hz or
142.86 Hz, where IIJ = 0.007 s, is normally used. The technique does
run into problems with sp 13CH groups of alkynes, where 1J(13CH)
~ 250 Hz, and if d2 in Fig. 8.18 is set as 0.007 s, the arms of the 13CH
doublet rotate ±315° and give a signal with the same phase as a 13C
or 13CH2 group!
Note that the decoupler is left on during the relaxation period as
well as during acquisition. This is to allow the NOE to build up and
to obtain the strongest possible signals with 13CpH} NOE.
/ modulation has also been used to obtain signals selectively from
only the non-proton bearing 13C nuclei, by using a d2 of 1/2/s. The
way that this works is also apparent from Fig. 8.18. As mentioned
earlier for non-proton bearing 13C nuclei in Fig. 8.18(a), the length of
the waiting time does not affect the alignment of the signal. In contrast
for 13CH, 13CH2, and 13CH3 groups, after 1/2/s, one half of the multi-
plet is pointing in the opposite direction to the other half. The result
is after decoupling, they cancel each other, and there is no signal. In
practice, this technique is difficult to use. Non-proton bearing 13C
nuclei give weak signals compared with proton bearing 13C nuclei due
to their longer Tl values and reduced NOEs. There is normally a range
of 1/(13C1H) values, making the choice of d2 a compromise, and
residual signals due to proton bearing carbon atoms are normally
observed. Frequently it is preferred to obtain an INEPT or DEPT
spectrum which does not contain signals due to non-proton bearing
13
C nuclei and identify them in the complete spectrum by comparison.
/ modulation has been described here as it gives a relatively simple
description of the use of multiple pulses and delays to edit NMR
spectra. It is a technique which was introduced in the early days of
multiple pulse NMR spectroscopy as it could be implemented on most
spectrometers. To a large extent it has now been replaced by DEPT
and PENDANT, vide infra.
A common variation on the sequence is the Attached Proton Test
or APT (Fig. 8.19). The advantage of APT over / modulation is
that a shorter relaxation delay time, dl can be used. This can easily
be understood. In the absence of phase cycling, the /-modulation
pulse program applies a total of a 270° pulse, leaving all the 13C

www.pdfgrip.com
 One-dimensional multipulse sequences 257

'H decouple decouple

13C FID

d1 p1 d2 p2 d2 d3 p2d3
Figure 8.19 The APT pulse sequence, dl is a relaxation delay, pi is a 45° pulse, d2 is I//, p2 are 180°
pulses, and d3 is the pre-acquisition delay.

magnetization in the .ry-plane which needs 57\ to regain the z-direc-

tion. In contrast, the APT pulse program applies a total of a 405°
pulse, equivalent to a 45° pulse, leaving M z /v2 magnetization in the
z-direction, ready for the next pulse sequence to be applied.

8.3.2 INEPT
There is a family of pulse sequences which gain an enhancement in
signal intensity from a higher frequency coupled nucleus, usually 1H.
Equation (1.3) gives the population difference between the upper and
lower energy levels for an / = 1/2 nucleus. The population excess in
the lower energy level is proportional to jjuj or ylt Hence for :H, with
yH = 26.7510 x 107 rad T'1 s-1, the population excess is approximately
four times greater than for 13C, with yc = 6.7263 x 107 rad T'1 s'1. If it
were possible to transfer this population excess from !H to 13C, then
the 13C NMR signal strength would increase by a factor of approxi-
mately four times. This can be done, but the situation is not all win
as the NOE is lost. Despite this, population transfer is generally
preferred as the gains, y$lyi, are normally greater than those from NOE,
1 + 7s/27j, see Tables 8.1 and 8.2, while the uncertainty associated with
NOE, that dipole-dipole relaxation is required to be dominant, is
removed. The gain is spectacular for very low frequency nuclei such as
103
Rh, especially as in this case NOEs are not normally observed.
The principle behind population transfer can be illustrated for
an AX system consisting of a 13C1VL pair of spins (Fig. 8.20). The
relative populations of the energy levels are given. Before we apply a

Table 8.4 The intensity gains on polarization transfer from *H to several

nuclei
13 29 31p 103
Observe C "N 19p Si Rh
Is 3.98 24.7 1.06 5.03 2.47 315

www.pdfgrip.com
258 One-dimensional NMR spectroscopy

selective 180° pulse, the population differences for the two 1H transi-
tions are both 48N, while for the two 13C transitions they are both 8N,
in Fig. 8.20(a). In Fig. 8.20(b), after applying the selective 180° pulse
to one !H transition, the population differences, and hence intensities
of the 13C transitions have changed to 58N: (-38N). Hence one tran-
sition has increased in intensity by 48N units and the other decreased
by -48N units. If we were to now repeat the measurement, applying
the selective 180° !H pulse to the other !H transition, we would get a
13
C doublet with intensities (-38N): 58N. If these signals are now
subtracted, the result is a 88N: (-88N) doublet compared with a
28N : 28N doublet if a simple 13C NMR spectrum had been taken using
two acquisitions. The gain has been a factor of four, or more accu-
rately VYc-
This experiment is impractical to carry out routinely as it would be
necessary to carry it out for each separate carbon environment in
the molecule. However, exactly the same result can be achieved
by the pulse sequence in Fig. 8.21 for all the 1H coupled to X in a
molecule, where X is an / = 1/2 nucleus. This is the Insensitive Nuclei
Enhanced by Polarization Transfer, or INEPT, pulse sequence.

(a)

(b)

Figure 8.20 The effect of applying a 180° selective pulse to one line of the
!H AX doublet of a 13CH group, (a) Before, (b) After.

www.pdfgrip.com
 One-dimensional multipulse sequences 259

'H
d1 p1 d2 p2 d2 p3

X.
P4 P5

Figure 8.21 The INEPT pulse sequence, dl is a relaxation delay, pi is a (90°)^

1
H pulse, d2 is l/4/XH, p2 is a 180° !H pulse, p3 is a (90°) !H pulse, p4 is a
180° X pulse, and p5 is a 90° X pulse.

Let us examine the effect of these pulses on the !H nuclei (Fig. 8.22).
We need to keep track of both the !H and X nuclei. In order to do
this, 1VL nuclei attached to X nuclei with a-spin are labelled a, and
those attached to X nuclei with p-spin are labelled b. The first (90°)r
pulse brings the XH magnetization into the xy plane pointing in the y'
direction. If the !H nuclei are on resonance, the two lines of the
doublet rotate at ±/XH/2 Hz- After 1/4^xn s > theY wili have rotated ±45°.
A 1H(1SQ°)X refocusing pulse is applied. If nothing else were to be
done, the two lines of the doublet would refocus in the -y' direction
after a further l/4/XH s. Instead a X(180°) pulse is also applied. This
has the effect of exchanging the X spin states. The protons which were
attached to X with a-spin are now attached to X with (3-spin and vice
versa. The result is that the !H nuclei precession direction in the
rotating frame is reversed, and after the further l/4/XH s they end up
aligned 180° with respect to each other. The final lH(90°)y pulse rotates

Figure 8.22 The effect of the INEPT pulse sequence on the !H magnetiza-
tion in the xy plane. The diagrams start following the 1H(9Q°)X pulse.

www.pdfgrip.com
260 One-dimensional NMR spectroscopy

their magnetization back into the z-direction. One line of the doublet
is pointing in the +z direction, while the other is pointing in the -z
direction. This has achieved the same effect as the selective 180° pulse
described above, but has produced the effect for all the !HX spin
systems in the sample. The X(90°) pulse then produces an X FID with
polarization transfer. Phase cycling is used to alternately invert the
lines of the !HX doublet. For example, if instead of using a lH(9Q°)y
pulse as the final 1H pulse, a lH(9Q°)_y pulse was used, the phase of
the 1H doublet is reversed. Phase cycling of the receiver permits
addition or subtraction of the spectra. The resulting 1H coupled X
spectrum does not have the usual 'Pascal triangle' intensities, but
rather those given in Table 8.5. This is illustrated in Fig. 8.23 for
some rhodium hydrides. Note that to distinguish between a !:(-!)
doublet and a 1:0: (-1) triplet, the separation of the lines has to be
measured, and is / in the former case and 2J in the latter, with / being
pre-determined from the 1H NMR spectrum.
In order to be able to *H decouple, the pulse sequence has to be
extended. This is because in a !H coupled INEPT spectrum there
is as much positive intensity as negative intensity. The result of
decoupling is to give zero intensity. Fortunately the answer is very
straightforward. A delay is placed between the basic INEPT pulse
sequence of Fig. 8.21 and acquisition (Fig. 8.24). Of course, it is neces-

-1360 -1380 -1920 -1940 -1600 -1620 -1640

103
Rh/ppm
Figure 8.23 The 12.62 MHz 103Rh NMR spectra obtained using the INEPT pulse sequence with T = 0.0074 s,
corresponding to y^RtfH) of 34 Hz. The spectra are referenced to S = 3.16 MHz. (a) [Oi5-C5Me5)
RhH(SiEt3)(Ti2-C2H4)] in CD3C6D5, showing a -1:1 doublet, (b) [(ti5-C5Me5)RhH2(SiEt3)2] in C6D6,
showing a -1:0:1 triplet, (c) [(T]5-C5Me5)RhH3(SiEt3)] in CD3C6D5, showing a -1:-1:1:1 quartet.
(Reproduced by permission of Academic Press from Mann (1988) Adv. Organomet. Chem., 28, 397.)

www.pdfgrip.com
 One-dimensional multipulse sequences 261

Table 8.5 The INEPT Pascal triangle. It is constructed in the same way as
the normal Pascal triangle adding together the pair of numbers on the line
above, except that it is extended by putting 1 and -1 at the ends of the
next line.

0
-1 1
0 1 -1
1 1 - 1 - 1
1 2 0 - 2 - 1
1 3 2 - 2 - 3 - 1
1 4 5 0 - 5 - 4 - 1
1 5 9 5 - 5 - 9 - 5 - 1
1 6 14 14 0 -14 -14 -6 -1
1 7 20 28 14 -14 -28 -20 -7 -1
etc.

sary to include a refocusing 180° pulse for both !H and X, p4 and p7,
halfway through the delay.
Figure 8.25 shows the response of the X-magnetization in the xy
plane for an AX group. The first diagram shows the state of the magne-
tization following the X(90°);c pulse, p6. The two lines of the X doublet
are 180° out-of-phase. It is necessary again to follow the spin states
of the other nuclei. The X-nuclei attached to XH nuclei with a-spin
are labelled a and those attached to !H nuclei with (3-spin are labelled
b. After 1/4/ s, the magnetizations have rotated ±45° in the rotating
frame, assuming again that the X signal is on resonance. As usual it
is necessary to use an 180° X refocusing pulse. It is then necessary to
use a 180° !H pulse to reverse the 1H spin states and hence the labels
on the magnetization vectors. The result is that they continue to
precess in the same direction and refocus after a further 1/4/s. The
1
H decoupler is then turned on and the 1H decoupled FID collected,
enhanced by polarization transfer.

Figure 8.24 The refocused INEPT pulse sequence with !H decoupling, dl is

a relaxation delay, pi is a (90°),, !H pulse, d2 is l/4/XH, p2 and p4 are 180°
X
H pulses, p3 is a (90°)y 1H pulse, d3 is a variable delay, see text, p5 and p7
are (180°), X pulses, and p6 is a (90°),, X pulse.

www.pdfgrip.com
262 One-dimensional NMR spectroscopy

Figure 8.25 The effect of the INEPT pulse sequence on the X magnetization
in the xy plane.

The introduction of the delay d3 into the refocused INEPT pulse

sequence provides another method to distinguish between CH, CH2
and CH3 groups. The variation of their intensity with d3 is plotted in
Fig. 8.26. It will be noted that when d3 is chosen to be 0.125// all the
signals are positive, when d3 = 0.25/J only CH 13C signals are detected,
while when d3 = 0.375/7, CH and CH3 groups are positive, while
CH2 groups are negative. In each case, only 1H bearing 13C atoms are
detected. It is therefore possible to distinguish between each type of
group.
In practice, INEPT is not used to sort CH, CH2 and CH3 13C NMR
signals, but DEPT is normally used. The problem with INEPT is that
it is far more sensitive to the value of /XH than is DEPT. A typical
13
C NMR sample has a range of 1J(13C1H) and the resulting refocused
INEPT spectra show varying intensity and phase. Where INEPT wins
over DEPT is when the coupling constant is small. During the delay(s)
between the pulses, relaxation occurs if the delays are long. It is then
advisable to minimize the delays, and INEPT has the shorter delays.
INEPT is valuable for recording :H coupled 13C NMR spectra and this
is illustrated in Fig. 8.27.

8.3.3 DEPT
DEPT is the pulse sequence of choice to edit 13C NMR spectra. The
pulse sequence is given in Fig. 8.28. It is not possible to use a simple
vector model to describe this pulse sequence. The 13C editing is carried
out by choosing a suitable p3. Values of 45°, 90°, and 135° are used.
With a 45° pulse, the CH, CH2, and CH3 signals are all positive, a 90°
pulse only gives CH signals, and the 135° pulse gives CH and CH3
positive, while the CH2 signal is negative. Suitable addition and

www.pdfgrip.com
 One-dimensional multipulse sequences 263

1.2

0.8

0.4
&
g 0
o>
J:
-0.4

-0.8

-1.2
0.125 0.25 0.375 0.5
d3 in units of -j- s
j
Figure 8.26 The response of CH, CH2, and CH3 groups as a function of d3
when the refocused INEPT pulse sequence is used.

22 21 20 19
13
C/ppm
Figure 8.27 A partial 100.62 MHz 13C NMR spectrum of one of the methyl
groups of carvone in CDC13. The spectrum was recorded using the INEPT
pulse sequence and shows !H coupling. The -1 : -1:1 :1 pattern arises
from the INEPT pulse sequence which was set up for a J(13C1H) - 145 Hz
corresponding to an average ^(^CH). The doublet of doublet of doublets
multiplet structure of each of the four lines arises from the much smaller
3 13
/( C?H).

www.pdfgrip.com
264 One-dimensional NMR spectroscopy

'H
d1 p1 d2 p2 d2 p3

13C
P4 p5 d2

Figure 8.28 The DEPT pulse sequence with 1H decoupling, dl is a relaxation

delay, pi is a 90° H pulse, d2 is l/2/XH, p2 is a 180° !H pulse, p3 is a !H
1

pulse of variable length, see text, p4 is a 90° X pulse and p5 is a 180° X pulse.

i£H3

CH2

CH

150 100 50 0
5(13C)

Figure 8.29 Edited 13C DEPT spectra of ristocetin at 100.6 MHz. The lower spectrum is the normal broad-
band decoupled spectrum showing all the carbon resonances. Note that even the resonance of the CD2H
groups of the solvent dimethylsulfoxide-d6 appears in its appropriate trace. (From Bruker, with permission.)

subtraction of DEPT-45, DEPT-90 and DEPT-135 spectra yield

spectra only showing CH, CH2 or CH3 signals. An example is given
in Fig. 8.29.

8.3.4 PENDANT
Although DEPT is an excellent pulse sequence to use for the
observation of CH, CH2, and CH3 groups, it does not detect non-
protonated 13C atoms. / modulation or APT used to be the initial

www.pdfgrip.com
 One-dimensional multipulse sequences 265

method of choice to measure a 13C NMR spectrum, followed by

DEPT as required to differentiate between C and CH2 groups and
between CH and CH3 groups. Recently, / modulation and APT have
been replaced by the pulse sequence, PENDANT, Polarization
ENhancement During Attached Nucleus Testing. PENDANT has
nearly the same sensitivity as DEPT for CH, CH2, and CH3 groups,
and also gives signals for quaternary carbon atoms. The pulse sequence
is given in Fig. 8.30.
An example of a PENDANT 13C NMR spectrum is given in Fig.
8.31. Comparison of this spectrum with the corresponding JMOD spec-
trum in Fig. 8.17 shows that the non-proton bearing carbon atoms give
stronger signals. As these carbons generally give weaker signals than
proton bearing carbon atoms due to their longer Tl values, and
possibly due to smaller NOE's, it is these carbons which set the number
of acquisitions and hence time required for a given spectrum.

8.3.5 The use of spectral editing pulse sequences for 13C

assignments
For most compounds, it is now pointless to acquire a simple ^CpH}
NMR spectrum, when, in the same length of time a PENDANT spec-
trum can be acquired. The sorting of the 13C NMR signals into C and
CH2 signals with one phase and CH and CH3 signals with the oppo-
site phase is often sufficient to assign the signals, when combined with
the chemical shift information. Where there is ambiguity, the appro-
priate DEPT spectra can be measured.
Signals can be missing from spectra recorded using pulse sequences
such as / modulation, APT, INEPT, DEPT, or PENDANT. In all these
pulse sequences, there are delays between pulses and relaxation can
occur during these delays. This is a severe problem where there are
broad signals such as occur where there is substantial broadening due
to exchange or scalar relaxation by a quadrupolar nucleus.

'H
p1 p2 p3 p4

X
dl p5 d2 p6 d2 P7 d3 p8 d3 '

Figure 8.30 The PENDANT1 pulse sequence with 1H decoupling, dl is a 0relax-

ation delay, pi is a (90°)^ H pulse, d2 is l/4/XH, p2 and p4 are (ISO ), *H
pulses, p3 is a (90°) !H pulse, d3 is 5/8/XH, p5 is a (90°)^ X pulse, p6 and p8
are (180°)^ X pulses, and p7 is a (90°)y X pulse.

www.pdfgrip.com
266 One-dimensional NMR spectroscopy

C/CH2
CDCI3

CH/CH3

- i | i | i | i | i | i | i | i | i | i | i | i | i |

240 220 200 180 160 140 120 100 80 60 40 20 0

13
C/ppm
Figure 8.31 The 100.62 MHz 13C NMR spectrum of carvone in CDC13 recorded using the PENDANT
pulse sequence.

8.3.6 One-dimensional INADEQUATE

The observation of coupling between low abundance nuclei is diffi-
cult. For example, 13C is only 1.08% abundant, and the 13C NMR spec-
trum of an organic molecule consists of sharp singlets flanked by weak
13
C satellites. These satellites are often obscured by the strong central
resonance. The one-dimensional INADEQUATE, Incredible Natural
Abundance Double Quantum Transfer Experiment, pulse sequence is
given in Fig. 8.32. This pulse sequence gives the 13C coupling as an
out-of-phase doublet. A longer pulse sequence is required to bring the
doublet into phase (Fig. 8.33). It is this pulse sequence that was used
to produce the spectra in Fig. 8.34. d2 can be chosen to select 1J(13C13C)
or a longer range coupling (Fig. 8.34). This is the experiment of choice
to measure /(13C13C), while two-dimensional INADEQUATE is the
experiment to use where 13C-13C connectivity is required.

decouple

13C

d1 p1 d2 p2 d2 P3 d3 p4

Figure 8.32 The one-dimensional INADEQUATE pulse sequence with !H

relaxation delay, pi, p3 and p4 are 90° 13C pulses, d2 is
decoupling, dl is a 13
l/4/cc, p2 is a 180° C pulse, d3 is 3 JJLS.

www.pdfgrip.com
 Exercises in spectral interpretation 267

'H decouple

13C
d1 p1 d2 p2 d2 P3 d3 p4 d2 p2 d2

Figure 8.33 The one-dimensional INADEQUATE pulse sequence with !H

decoupling and rephasing. dl is a relaxation delay, pi, p3 and p4 are 90° 13C
pulses, d2 is l/4/cc, p2 is a 180° 13C pulse, d3 is 3 JULS.

(b)-

(a)
n i r
5910 5860 3600 3550 3040 2990
Hz
Figure 8.34 The application of the one-dimensional INADEQUATE pulse sequence to determine
/(13C13C) in [(T]5-C5H5)Ni(l,3,4-Ti3-2,2-dimethylbutenyl)] at 100.62MHz in d8-THF. (a) The signals due to
C3, C2, and C6. (b) INADEQUATE NMR spectra with d2 - 0.0062 s corresponding to /(13C13C) - 40 Hz.
(c) INADEQUATE NMR spectra with d2 - 0.08 s corresponding to /(13C13C) - 3 Hz. (Reproduced from
Benn and Rufinska (1982) /. Organomet. Chem., 238, C27, copyright (1982), with permission from Elsevier
Science.)

8.4 EXERCISES IN SPECTRAL INTERPRETATION

The determination of the structures of organic molecules using proton

and carbon spectra together will now be considered in some exercises.

www.pdfgrip.com
268 One-dimensional NMR spectroscopy

We have shown in the previous chapters that the carbon spectra are
capable of yielding much information through a series of quite time-
consuming experiments. These facilities are normally reserved for the
more difficult samples, and often it will be sufficient to have simply
the broad-band proton-decoupled spectrum. This gives the chemical
shift information, details of coupling to nuclei other than hydrogen,
and a carbon count, provided all the likely errors in line intensity are
taken into account. It is also certain that the proton spectrum will be
available, since this is so easy to obtain, and interpretation will
be based on the two sets of data taken together. The carbon spectra,
of course, give information about atoms not bonded to hydrogen,
which is not available in the proton spectra.
The carbon chemical shifts are much more widely dispersed than
are the proton shifts. The ranges within which different types of carbon
atom resonate are shown on the chart in Fig. 8.35. The chemical shift
of a given carbon atom in a family of compounds is sensitive to the
influence of all four substituents, and for alkanes, for instance, can be
predicted using the Grant and Paul rules
8; = -2.6 + 9.1na + 9.4/10 - 2.5ny + 0.3na

Tl\US
ni-uv/i ^^H 1
vi ry*, ^^^^M

cyclopropanesl
CH3R 1•
C BpR2™
Ch«R3"¥
CHsHal 1 •
r.Hj\i^«
'*•' • vy " "
•
CHp •
QP.M^n ^^H
1 IVXI l^\.X
••

R2CHO •
p_/^n
~ ^
ig^v^ -f-

RCH2S 1 •I
Alkynes •l
Alkenes ^" •1
1'CH ••
Aromatics < CN
•
.CO" •
i—»/-\x-\ 1ijmm•
i-iu U2' ^^H
RCHOI
R^CO •1
•
M(CO)
••

200 160 120 80 40 0

13
C/ppm
Figure 8.35 Chart of approximate chemical shift ranges of carbon atom nuclei
in organic and organometallic compounds. M represents a metal and Hal a
halogen.

www.pdfgrip.com
 Exercises in spectral interpretation 269

7.3 7.2 7.1 2.7 2.6 2.5

1.3 1.2

9 8 7 6 5 4 3 2 1 0
1
(a) H/ppm

Exercise 4 (a) The 400.13 MHz 'H NMR spectrum of a solution of ethyl benzene in C6D6.

130 128 126

200 180 160 140 120 100 80 60 40 20 0

13
(b) C/ppm
Exercise 4 (b) The 100.62 MHz 13C JMOD NMR spectrum phased so that C and CH2 groups are nega-
tive and CH and CH3 signals are positive. In each case, expansions are given as insets.

www.pdfgrip.com
270 One-dimensional NMR spectroscopy

8.3 8.2 8.1 8.0 7.9 7.8 7.7

1
H/ppm

200 180 160 140 120 100 80 60 40 20

13
C/ppm

N=N N N-

Exercise 5 The 400 MHz 1H NMR spectrum (upper) and 100.62 MHz 13C{1H}
JMOD NMR spectrum (lower) of the aromatic compound C12H8N2. The 13C
NMR spectrum shows signals at 8 143.3,130.4, and 129.6 due to the compound
and 77.0 due to CDC13. The proton spectrum is diagnostic of an [AB]2 four-
spin system in which there are two pairs of protons with the same chemical
shift 8A, 5B, but different coupling constants /(AB), /(AA'), /(AB') and
7(BB'). Only three lines are observed in the carbon spectrum. This informa-
tion is sufficient to identify the compound, some suggestions for which appear
below the spectra. Example supplied by A. Romer.

www.pdfgrip.com
 Exercises in spectral interpretation 271

The chemical shift of carbon i can be calculated from the number

of directly bonded carbons (na) and the number of carbon atoms two
(n^9 three (ny) and four (ns) bonds removed; -2.6 ppm is the chem-
ical shift of methane. Similar rules exist for ethylenic hydrocarbons
and substituent effects have been documented.
The use of the two sets of information together is illustrated in
Exercise 4 for the simple case of ethyl benzene. The 400 MHz lTrl NMR
spectrum contains the unmistakeable quartet-triplet pattern due to an
ethyl group and five protons resonate in the aromatic region. The
aromatic protons split into two sets of signals with an intensity ratio
of 2:3. The signal of intensity 2 at 8 7.26 is a triplet with additional
structure. This must be due to the raeta-hydrogen atoms, which are
coupled to the ortho and para ones. The ortho- and para-hydrogen
signals overlap at 8 7.16. This is indeed sufficient to give the structure
in this case, especially if the molecular weight were known. We never-
theless turn our attention to the carbon JMOD spectrum. We find the
two carbons of the ethyl group and four aromatic resonances with
appropriate polarities. The aromatic signal to high frequency is rather
small, as would be expected for quaternary carbon and its polarity
indicates a lone C. The remaining three are CH and two are sig-
nificantly larger than the third so may arise from two carbon atoms
each. The pattern has the form to be expected for a monosubstituted
six-membered ring, and allows us to over determine the structure of
this molecule. All its features are apparent, though we should note

180 170 160 150 140 130 120 110

13
C/ppm
Exercise 6 The 100.62 MHz 13C JMOD NMR spectrum of C10H2O6 in (CD3)2CO. The spectrum is plotted
so that signals due to CH and CH3 groups are positive and those due to C and CH2 groups are negative.
Example supplied by A. Romer.

www.pdfgrip.com
272 One-dimensional NMR spectroscopy

carefully the ambiguities of the intensity of the resonances. Which line

should we take as representing one carbon? Some more complex
examples are given in Exercises 5 and 6.
Exercise 6 shows the 13C JMOD NMR spectrum of the compound
C10H2O6 with proton double irradiation. In a 1H coupled spectrum,
not shown, / = 179 Hz is observed to the resonance at 130.1 ppm and
/ = 6 Hz to the resonance at 135.6 ppm. The proton spectrum is a
singlet in the aromatic region. What is the structure of the molecule,
given that it is an anhydride?

www.pdfgrip.com
 Two-dimensional NMR
spectroscopy 3G
Two-dimensional NMR spectroscopy has become an important tool
for chemists. Several experiments are possible and new variants are
being continually invented, such is the power of the method. The two
dimensions are dimensions of time as previously discussed at the begin-
ning of Chapter 8. One of these is already familiar, and is the time
domain within which we collect the FID output from the spectrom-
eter and which contains frequency and intensity information. The
second dimension refers to the time elapsing between the application
of some perturbation to the system and the onset of the collection of
data in the first time domain. This second time period is varied in a
regular way and a series of FID responses are collected corresponding
to each period chosen.
In order to illustrate how two-dimensional NMR spectroscopy
works, we will show how the one-dimensional / modulation described
in section 8.3.1 can be converted into a two-dimensional experiment.
The delay d2, which is related to /, is replaced by a variable delay
fj/2, where ^ is made to assume many values. The pulse sequence is
shown in Fig. 9.1 which is essentially that of Fig. 8.16 though we should
note that in practice it is more common for the decoupler to be off
during the first delay and on during the second, though this makes no
difference to the outcome. Examination of the behaviour of the spin
magnetization in Fig. 8.18(b) shows that the intensity of the 13C signal
from a CH entity is a function of the delay time and varies as
cos(2ffrtlIJ). It will also decay by the T2 process though the 180° pulse
refocuses the magnetization and this is the true T2\ see section 6.3.
Similarly, the 13C signal of a CH2 group will be seen from Fig. 8.18(c)
to remain always positive and to follow a [cos(2(Trr1//) + 1] law. In the
two-dimensional form of the experiment, a series of FIDs are collected
corresponding to different values of ^. Each FID is then Fourier trans-
formed individually to give a series of spectra, one for each value of
*! chosen. So we have spectra varying as a function of/ 2 (corresponding
to the t2 dimension) and covering a range of ^ values as in Fig. 9.2.
If we look at the intensity at 8 42.45, which arises from a CH, we
will observe that it changes across the spectra as a decaying cosine
wave. This is made clearer by the section through the spectra
shown in Fig. 9.3. The frequency of this wave gives the C-H coupling
constant. The signal is not however, a simple decaying cosine wave

www.pdfgrip.com
274 Two-dimensional NMR spectroscopy

but is modulated by smaller, long distance couplings to give lower

frequency waves. This is not very apparent for the signal at 8 42.45 as
this unresolved coupling broadens the lines and gives the rapid decay.
This modulation is much clearer for the signal at 8 43.10 as two and
three bond couplings to single protons introduce negative components
into the pattern which without these smaller couplings would remain
always positive.

decouple decouple

13C FID

d1 p1 A. p2 A.
2 2
Figure 9.1 The J modulation pulse sequence, dl is for relaxation, pi is a 90°
pulse, f/2 and t2 are the two time dimensions for the two-dimensional exper-
iment, and p2 is a 180° pulse.

0.040 s

0.035 s

0.030 s

0.025 s

0.020 s

0.015s

0.010s

0.005 s
0.001 s
543.10 6 42.45

Figure 9.2 The individual spectra of a partial 7-resolved 100.62 MHz 13C NMR spectrum of carvone in
CDC13 obtained using the pulse sequence in Fig. 9.1. The individual FIDs have been transformed in the
r2//2 direction and plotted for ^ values from 0.001 to 0.04 s, incremented by 0.001 s, given on the right.
Note the decaying cosine oscillations of each signal in the fj//i direction.

www.pdfgrip.com
 Two-dimensional NMR spectroscopy 275

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8s

Figure 9.3 Cross-section through the signal at 8 43.10 after transformation in the/ 2 direction of a /-resolved
100.62 MHz 13C NMR spectrum of carvone obtained using the pulse sequence in Fig. 9.1. A partial stack
plot is shown in Fig. 9.2. The *-axis corresponds to the delay, t±. The high frequency components are due
to 1J(13C1H), while the low frequency components are due to 2J(13C1H) and 3J(13C1H).

It should be clear that chemical shifts are contained in the /2 dimen-

sion and the coupling information is in the ^ dimension. We can
measure the coupling frequency from the plots of Fig. 9.3 but it is
much more convenient to get the computer to do it for us by
performing a Fourier transform across the spectra in the ^ direction,
especially as this will give us both the evident frequency and the
smaller components. This is now the /i dimension. The result is shown
in Fig. 9.4. The set of spectra shown in Fig. 9.4 is very difficult to use,
especially when there are many signals. It is essentially the same
problem that was solved many years ago by cartographers. In exactly
the same way as mountains are represented on maps by contours, the
intensity of the NMR signals of a two-dimensional NMR spectrum are
represented by contours on a two-dimensional plot of/ 2 , normally hori-
zontal, against fl9 normally vertical (Fig. 9.5). The result is that over-
lapping multiplets are separated (Fig. 9.6).

• ' • i • • • i • < • i • > • i • - • i ' < • i • i ' i • ' ' i • ' • i ' ' ' i • i • i —
200 180 160 140 120 100 80 60 40 20 0
13
C/ppm
Figure 9.4 The individual spectra of a /-resolved 100.62 MHz 13C NMR spectrum of carvone obtained
using the pulse sequence in Fig. 9.1. The individual FIDs have been transformed in the t2/f2 and ^//i direc-
tions and plotted for a selection of ^ values.

www.pdfgrip.com
276 Two-dimensional NMR spectroscopy

-100

- -50

OHz

- 50

- 100

200 150 100 50 0

13
C/ppm

Figure 9.5 The contour plot of a /-resolved 100.62 MHz 13C NMR spectrum
of carvone in CDC13 obtained using the pulse sequence in Fig. 9.1.

There are some experimental considerations. Two-dimensional

NMR spectra can produce very large data sets and can take very
long times. The data set for the FIDs giving rise to the spectrum in
Fig. 9.5 is 32 768 data points in the t2 direction and 128 data points
in the ^ direction. The sweep width in both dimensions was kept to
a minimum to maximize the digitization. As each word is 32 bits,
this gives rise to a dataset of 16 megabytes. The major consideration
is time. In order to complete the phase cycling to remove artefacts, a
multiple of 4 spectra have to be acquired for each tl value. The acqui-
sition plus dl time needs to be long enough for recovery, 3.3 s in this
case. The result is that the minimum time required for this experiment
is 4 x 3.3 x 128 s = 1690s = 28 m 10 s though on top of this there is the
time associated with the pulse sequence. In practice, considerably
longer times can be required. There is not normally a time penalty in
increasing the /2 resolution by increasing the acquisition time, as a
corresponding decrease can be made in dl. However, to increase the
/! resolution by a factor of two requires a doubling of time. The time
for the experiment can be reduced by decreasing dl, but then arte-
facts become a problem. Real gains are made by using magnetic field
gradient pulses. Then dl can be shortened, and phase cycling is not
required or is considerably reduced.

www.pdfgrip.com
 Two-dimensional NMR spectroscopy 277

We will describe some of the experiments possible in more detail

below. If the molecule to be studied is complex, it is necessary to carry
out several different two-dimensional experiments in order to discover
all the correlations necessary to provide an unequivocal structure. We
will use the simple molecule carvone to introduce how several of the
sequences work and use D-amygdalin, whose formula is in Fig. 9.7, to
demonstrate the advantages of the two-dimensional technique.

50 0 -50 100 50 0 -50 -100

(a) Hz (b) Hz

Figure 9.6 Cross-sections of the signals at (a) 8 42.45 and (b) 43.10 of a /-resolved 100.62 MHz 13C NMR
spectrum of carvone in CDC13 obtained using the pulse sequence in Fig. 9.1, showing the multiplicity due
to !H coupling. Note that due to the pulse sequence used the coupling constants have been halved.

OH
(a) (b) 10

Figure 9.7 The structures of (a) D-amygdalin and (b) carvone.

www.pdfgrip.com
278 Two-dimensional NMR spectroscopy

9.1 /-RESOLVED TWO-DIMENSIONAL NMR

SPECTROSCOPY

9.1.1 Homonuclear J-resolved two-dimensional NMR spectroscopy

This experiment separates chemical shift information from coupling
constant information. The result is a !H NMR spectrum with all the
first-order 1H coupling removed and cross-sections giving the coupling
patterns.
This technique is based on the Carr-Purcell pulse sequence
described in section 6.1.2.2 for the determination of T2, which only
works if the resonances studied are singlets, since the refocusing of
doublets caused by spin-spin coupling modulates the output intensity
as a function of the waiting time. This is easily transformed to advan-
tage into a two-dimensional experiment. The pulse sequence is shown
in Fig. 9.8.
This two-dimensional NMR experiment was applied to carvone.
After transformation in both the t2 and ^ dimensions, the two-
dimensional NMR spectrum shown in Fig. 9.9 is obtained. The
chemical shift information is in the /2 dimension and the coupling
information is spread out in the /t dimension. It is then normal to 'tilt'
the spectrum so that all the lines of each multiplet line up in the /j
dimension. This is a very simple operation which is done automati-
cally by the computer. Each row of the two-dimensional matrix is
shifted by an amount proportional to its distance from the centre of
the spectrum. Thus the row corresponding to a 15 Hz offset in the /j
dimension is shifted to a 15 Hz increase in frequency, and this produces
a shift to the left, higher frequency. The result is shown in Fig. 9.10.
If all the individual spectra are now added together, the result is a !H
decoupled !H NMR spectrum, which is shown above the two-dimen-
sional spectrum. It is possible to clean up the spectrum by
symmetrizing it about the /j = 0 Hz row. This consists of comparing
the intensities in the spectra equally spaced above and below the /t =
OHz row and taking the smallest value. This is shown in Fig. 9.11
and the improved !H decoupled 1H NMR spectrum is shown above
the two-dimensional spectrum. A section taken along the length of a

FID

d1 p1 tl P2 ll
2 2
Figure 9.8 The /-resolved pulse sequence, dl is for relaxation, pi is a 90°
pulse, tJ2 and t2 are the two time dimensions for the two-dimensional exper-
iment, and p2 is a 180° pulse.

www.pdfgrip.com
 Two-dimensional NMR spectroscopy 279

-15

-10

-5

10

15
Hz
2.6 2.5 2.4 2.3
1
H/ppm

Figure 9.9 A contour plot of the /-resolved 400 MHz 1H NMR partial spec-
trum of carvone in CDC13 before tilting. A projection of the spectrum is shown
above.

multiplet, parallel with the /i axis (a cross-section) gives the coupled

spectrum for each proton of that chemical shift, and an example of
one such section is given in Fig. 9.12. This technique is very valuable
for the separation of overlapping multiplets and producing !H broad-
band 'decoupled' 1H NMR spectra.
A model compound such as carvone used to illustrate the experi-
mental aspects of the technique does not fully show its power. In a
much more congested spectrum, such as that for D-amygdalin, the
signals at 8 3.9 and between 8 3.3 and 3.6 are much more difficult to
disentangle (Fig. 9.13), but the /-resolved experiment separates the
multiplets. A typical plot is shown in Fig. 9.14 for the low frequency
sugar resonances of D-amygdalin, where most spectral congestion lies.
The chemical shifts are plotted on the horizontal scale, whose length
is equivalent to 800 Hz, and coupling constants are in the vertical axis.
The plot below the chart is the normal one-dimensional spectrum. That
at the top is the projection of the contour plots onto the chemical shift
axis and is, in fact, a spectrum in which all the protons have been fully
decoupled from one another, i.e. the experimentally unachievable
homonuclear broad-band decoupling experiment. Clearly, this plot
helps to distinguish resonances that overlap in the one-dimensional

www.pdfgrip.com
280 Two-dimensional NMR spectroscopy

-15

-10

-5

10

15
Hz
2.6 2.5 2.4 2.3
1
H/ppm

Figure 9.10 A contour plot of the /-resolved 400 MHz !H NMR partial spec-
trum of carvone in CDC13 after tilting. A projection of the spectrum is shown
at the top.

trace. There are, for instance, two 1:1:1:1 quartets at 8 4.6 and 3.73,
and we can see immediately that the former consists of two equal over-
lapping doublets whereas the latter is a doublet of doublets involving
only one chemical shift value. The multiplet structure appears in the
coupling dimension. 128 values of T were used and 128 FIDs were
obtained, which limits the digital resolution to 0.39 Hz. It is a simple
matter to obtain all the coupling constants to within this limit. What
can we deduce about the spectrum of D-amygdalin from this chart?
The lines are already assigned on the figure but we will attempt to
proceed as if this had not already been done. Four lines are marked
as originating from proton-6. They are all doublets of doublets, but
one of the couplings is rather large at about 11 Hz and is found
nowhere else in the spectrum, and we can think immediately in terms
of a possible geminal pair. The two large couplings are slightly
different, sufficiently so that it is possible to separate the four reso-
nances into two pairs linked by identical coupling constants. These are
then the CH2 protons linked to the two rings, although we do not actu-
ally know yet which is which. Each proton of these geminal pairs is
further coupled to proton 5, though the responses from both of these

www.pdfgrip.com
 Two-dimensional NMR spectroscopy 281

•15

•10

-5

10

15
Hz
2.6 2.5 2.4 2.3
1
H/ppm

Figure 9.11 A contour plot of the /-resolved 400 MHz 1H NMR partial spec-
trum of carvone in CDC13 after tilting and symmetrizing about a horizontal
running through the middle of the spectrum at 0 Hz. A projection of the spec-
trum is shown at the top.

20 15 10 5 0 -5 -10 -15 -20

Hz
Figure 9.12 The cross-section of the 400 MHz 'H NMR spectrum of the signal
at 5 2.46 of carvone in CDC13.

is rather weak and we cannot reliably proceed to further assignment.

The two doublets near 8 4.6 have a chemical shift that is character-
istic of the anomeric proton HI of (3-pyranosides of glucose. Further,
these protons are the only ones that will be split into simple doublets,
and so the assignment is confirmed. Finally, we can see that all the

www.pdfgrip.com
282 Two-dimensional NMR spectroscopy

i ' ' ' ' r

8 7 6 .
1
5 4 3
H/ppm

—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i i—i i—i—i—i—

7.65 7.60 7.55 7.50

Figure 9.13 The normal, i.e. one-dimensional, 500 MHz !H NMR spectrum
of D-amygdalin. There is a pattern in 2 : 3 intensity ratio at 7.6 ppm due to
phenyl, a CHCN singlet at 5.88 ppm, HOD at about 4.75 ppm and a complex
pattern to low frequency due to the remaining 14 sugar protons. (From Ribiero
(1990) Magn. Reson. Chem., 28, 765-7, copyright (1990) John Wiley and Sons
Ltd, reprinted with permission.)

coupling constants other than those between proton-6 are about

7-8 Hz, so that we have axial-axial stereo-chemistry rather than equa-
torial-equatorial or equatorial-axial.

9.2 HOMONUCLEAR COSY NMR SPECTROSCOPY

COSY stands for Correlation SpectroscopY, and was the first two-
dimensional technique to be proposed by Jeener in 1971. As we shall
see, it has given rise to many variants, which depend upon the exis-
tence of spin-spin coupling between nuclei to provide supplementary
responses that relate the chemical shift positions of the coupled nuclei.
It is equivalent to carrying out simultaneously a series of double-reso-
nance experiments at each multiplet in the spectrum and looking for
the part of the spectrum where a perturbation has occurred. It is thus
rapid and avoids the need to know what irradiation frequency to use

www.pdfgrip.com
 Homonuclear COSY NMR spectroscopy 283

10

10
Hz
4.8 4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2
I 11 I /~ ~.

4.8 4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2

1
H/ppm

Figure 9.14 Partial homonuclear /-resolved two-dimensional spectrum of D-

amygdalin. The responses are shown as contour plots on the //8 chart. The
normal one-dimensional spectrum is below and the fully decoupled version
of this is above the chart. All coupling constants can be read off on the right-
hand scale. (From Ribiero (1990) Magn. Reson. Chem., 28, 765-73, copyright
(1990) John Wiley and Sons Ltd, reprinted with permission.)

for each experiment, information that in any case is often difficult to

obtain for a complex or crowded spectrum. Figure 9.15 shows the basic
pulse sequence used. There are many variations on this pulse sequence
and some of the more important ones will be illustrated in the
following sections.

9.2.1 COSY-90 NMR spectroscopy

The simplest and most sensitive version of the COSY experiment is
the COSY-90 experiment where pi is a 90° pulse. PI produces a
normal FID and p2 distorts this in a way which depends upon the
delay t^ Treatment of the set of distorted spectra which result, gives
a two-dimensional diagram with the proton chemical shift along both

www.pdfgrip.com
284 Two-dimensional NMR spectroscopy

FID

d1 p1 * P2

Figure 9.15 The basic pulse sequence for the COSY experiment, dl is a relax-
ation delay, pi is a variable pulse, fixed for a given experiment, p2 is a 90°
pulse, and ^ is the variable delay for the second dimension.

axes and the spectrum along the diagonal, which projects onto either
axis as the normal spectrum. The distortion introduced by the pulse
sequence introduces extra responses, off the diagonal, for those
protons which are spin coupled and none for those which are not.
These extra resonances appear on the two-dimensional map at the
points which describe the two chemical shifts of the coupled protons
and so enable the shifts of coupled resonances to be discovered, or
correlated. These peaks are often called cross-peaks.
A partial COSY-90 spectrum of D-amygdalin is shown in Fig. 9.16,
concentrating on the low frequency part of the spectrum where all the
interest lies. The diagonal runs from top left to bottom right of
the map and when projected onto a chemical shift axis is seen to
be the one-dimensional spectrum, in this case shown at the top of the
figure. Several off-diagonal responses are evident, disposed symmetri-
cally about the diagonal, and it is these we must examine for infor-
mation about correlation. The anomeric signals at about 4.6 ppm can
be used as a point of entry for analysis of the spectrum, and the connec-
tivities to the H2 signals can easily be traced. Unfortunately, the diag-
onal is very cluttered in the 3.4 to 3.6 ppm region, and we cannot move
on with any degree of certainty to the connectivities to protons further
around the ring. This is a disadvantage of COSY-90, which we will
see how to circumvent shortly. The doublet of doublets at 4.22 ppm
has a cross-peak to the complex multiplet at 3.92 ppm and to the multi-
plet at 3.62 ppm, the latter also being correlated with the 3.92 ppm
resonance. This pattern is that of an ABX spin system and so has to
be due to H6, H6' and H5 of one ring. The multiplet at 3.62 ppm is an
octet and so has to be H5, in agreement with the /-resolved data, which
has already allowed us to locate the protons 6. It follows from the
cross-peak near the diagonal that H4 of this ring is at about 3.5 ppm.
A second, similar pattern, connects signals at 3.9, 3.75 and 3.49 ppm
and presumably arises from the protons 5 and 6 of the second ring.
H4 cannot be assigned in this case. It remains to assign this and the
H3 protons.
Figure 9.16 shows the spectrum before symmetrization. The
symmetrized spectrum is in Fig. 9.17.

www.pdfgrip.com
 Homonuclear COSY NMR spectroscopy 285

HOD

5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2
1
H/ppm

Figure 9.16 Partial 400 MHz !H COSY-90 NMR spectrum of D-amygdalin

before symmetrization showing the sugar resonances.

There is a problem with symmetrization and care is necessary in its

use. Examination of the peaks on the diagonal in Fig. 9.16 shows
that the peaks are not round or oval, but have horizontal and vertical
'tails'. When peaks are close in chemical shift, the 'tails' can reinforce
each other, giving an apparent cross-peak. This is obvious before
symmetrization but after, it is easy to be fooled.
COSY is not restricted to !H NMR spectroscopy or indeed to nuclei
where coupling is resolved. As long as the coupling contributes to the
linewidth of the signals, then COSY can be used. For example, COSY
is very valuable in elucidating the structure of boron hydrides and
carboranes. This is illustrated here for 9-SMe2-7,8-rad0-C2B9H11, (9.1)
(Fig. 9.18). Examination of the COSY spectrum shows strong corre-
lations, as shown by the large number of contour lines for each cross
peak, acting through 1/(11B11B) and weak correlations through
2 n n
/( B B) and/or 3/(nBnB). This spectrum then allows the resonances
to be assigned to the nearest neighbour boron atoms and to confirm
this via the longer range correlations.

9.2.2 COSY-45 NMR spectroscopy

The COSY-45 NMR spectrum only differs from the COSY-90 NMR
spectrum by using a 45° pulse for pi in the pulse sequence in Fig. 9.15.
There are three reasons for measuring a COSY-45 NMR spectrum in

www.pdfgrip.com
286 Two-dimensional NMR spectroscopy

5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2
1
H/ppm

Figure 9.17 Partial 400 MHz 1H COSY-90 spectrum of D-amygdalin after

symmetrization, showing the sugar resonances.

2
B5B9 B B11B3B4B6B10 B1

-35
-30
-25
-20 Q.
Q.
-15,23
-10
-5
(9.1) 0
0 -5-10-15 -20-25 -30-35
11
B/ppm
Figure 9.18 The 128.4 MHz nB COSY-90 NMR spectrum of 9-SMe2-7,8-mdo-
C2B9Hn, (9.1), in CDC13. Numbers are used to identify the boron atoms and
• to indicate the carbon atoms. (Reproduced from Rosair et al (1997) /.
Organomet. Chem., 536, 299, copyright (1997), with permission from Elsevier
Science.)

www.pdfgrip.com
 Homonuclear COSY NMR spectroscopy 287

preference to the more sensitive COSY-90 NMR spectrum. First, the

peaks on the diagonal are weaker, so that it is easier to identify corre-
lations close to the diagonal. Second, it is easier to decide which
coupling constant is operating for a particular correlation, and third,
the relative signs of coupling constants can be determined. As noted
earlier in sections 3.1 and 3.7.3, coupling constants have sign. If we
only measure the magnitude of a coupling constant, then we are
throwing away valuable information. The most common case where
sign is important is in -CH2CH< fragments in molecules. 2/(1H1H)
normally lies between -12 and -14 Hz, while 3/(1H1H) can lie between
0 and +16 Hz depending on the torsion angle and substituents. It is
quite common to have situations where in order to distinguish between
assignments, we need to know which coupling is positive and which is
negative.
A COSY-45 *H NMR spectrum of D-amygdalin is shown in
Fig. 9.19. Examination of the spectrum shows that the diagonal is
cleaner making it easier to observe cross peaks close to the diagonal
and cross-peaks are no longer oblong, like the COSY-90 1H NMR
spectrum, but lean. In order to compare a COSY-90 and a COSY-45
NMR spectrum two partial spectra are given in Fig. 9.20. The peak
on the diagonal at 8 3.5 in the COSY-90 spectrum also consists
of off-diagonal signals which map out a square for both protons B6
and A6/. Comparing this with the same peak in the COSY-45 NMR

5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2
1
H/ppm

Figure 9.19 Partial 400 MHz !H COSY-45 spectrum of D-amygdalin in D2O

showing the sugar resonances.

www.pdfgrip.com
288 Two-dimensional NMR spectroscopy

spectrum shows that the off-diagonal signals are not present. The result
is that the spectrum close to the diagonal is also much cleaner, and it
is easier to identify cross peaks in crowded regions of the spectrum,
for example between 8 3.3 and 3.6, compare Figs 9.16 and 9.19.
Examination of the cross peak at 8 3.5/3.9 shows that in the COSY-
90 NMR spectrum the responses map out a rectangle (Fig. 9.20(a)),
while in the COSY-45 NMR spectrum, half the responses are missing
(Fig. 9.20(b)). The result is that the cross peak appears to 'lean'. This
provides us with useful information. If we compare the two B6 cross
peaks at 8 3.5 and 3.74, they lean in opposite directions. This is because
the cross peak at 8 3.74 is to protons B6' and B5.2/(B6,B6') is a geminal
coupling constant and is negative, while 3/(B6,B5) is a vicinal coupling

3.2

3.4

3.6 _
5
•o
3
3.8

4.0

4.2

4.4

1 1
(a) H/ppm (b) H/ppm

Figure 9.20 Partial 400 MHz 'H (a) COSY-45 and (b) COSY-90 NMR spectrum of D-amygdalin in D2O.
The active coupling constant is indicated by dotted lines and arrows. The diagonal resonances are near
8 3.95.

www.pdfgrip.com
 Homonuclear COSY NMR spectroscopy 289

constant and is positive. The direction of 'lean' gives the relative sign
of the coupling constant that is operating. The missing peaks also tell
us which coupling constant is operating between a pair of protons. It
is the one where both components are present.

9.2.3 Phase-sensitive COSY NMR spectroscopy

The COSY-90 and COSY-45 NMR spectra described above are trans-
formed in the magnitude mode, i.e. the intensity of each data point is
calculated as V(w 2 + v2), where u and v are the intensities of the real
and imaginary parts of the transform spectra. This removes the prob-
lems of phasing and makes all the signals positive. If high resolution
is used in both the /j and /2 directions, then it is possible to obtain a
phase-sensitive COSY spectrum. The spectra have to be phased in
both the /j and /2 directions, and both positive and negative peaks are
obtained. The peaks on the diagonal are out-of phase and give long
tails in the /i and /2 directions leading to very messy looking spectra.
However, if cross-sections are taken, then the resulting spectrum
makes it easy to determine in the cross peak the coupling which
connects the two resonances (Figs 9.21, 9.22). There is a 180° phase
change between lines associated with the active coupling constant
which can then be identified.

5.0 4.8 4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2
1
H/ppm

Figure 9.21 Partial phase-sensitive 400 MHz 1H COSY spectrum of D-amyg-

dalin showing the sugar resonances.

www.pdfgrip.com
290 Two-dimensional NMR spectroscopy

4.0 3.5
1
H/ppm

Figure 9.22 A cross-section through a signal at 8 3.74 fron the partial phase-
sensitive 400 MHz *H COSY spectrum of D-amygdalin shown in Fig. 9.21.
The cross-section is through B 6 , which gives the out-of-phase signal at 8 3.74,
and shows the coupling to B6 at 8 3.94 and B5 at 8 3.49.

9.3 HETERONUCLEAR COSY NMR SPECTROSCOPY

The next stage in understanding the spectroscopic properties of our

model molecule is to assign the 13C resonances. This is done by inves-
tigating the connectivities imparted by JH-13C coupling paths. The task
is facilitated by the fact that the one-bond CH coupling constants are
much larger than the two- or three-bond (CCH or CCCH) coupling
constants, so that the pulse sequence used can differentiate between
different types of coupling path. Decoupling of protons from 13C is
achieved in the usual way by irradiating the protons in t2, while accu-
mulating the 13C FID. The sequences used also illustrate how broad-
band decoupling of the protons may be achieved simultaneously by
using a specially adapted pulse sequence. The pulse sequences needed
are illustrated in Fig. 9.23, and are shown in order of increasing
complexity so that the experimental technique can be built up in stages.
The first sequence shown produces 1H-13C connectivities via the one-
bond coupling. The first 90° pulse at the !H frequency produces trans-
verse y magnetization. This then evolves with ^ and the position of
the magnetization of a given proton depends upon its chemical shift.
Those protons attached to carbon-13 will also have two magnetization
components, which precess at different frequencies, and these are refo-
cused by inverting the carbon magnetization half-way through the ^
period. In principle, then, the proton magnetization at the end of ^ is
a function of tl9 and this can be transferred into the carbon magneti-
zation by a second proton 90° pulse, which places the y magnetization
into the z direction and permits polarization transfer. Unfortunately,
the inversion of the 13C nuclei also causes the proton lines of the
!
H-13C doublet to be of opposed phases with zero resultant. It is

www.pdfgrip.com
 Heteronuclear COSY NMR spectroscopy 291

cpd
d1 p1 jj_ f| d2 p2 d3
2 "2

FID
(a) P3 P4

'H cpd
d1 p1 p5 d4 p6 d5 p7 d2 p2 d3

13C FID
P3 P4

(b) BIRD pulse

Figure 9.23 Pulse sequences used to produce ^-^C hetero COSY spectra, pi, p2, and p5 are (90°),
pulses, p7 is a (90)% pulse, p4 is a 90° pulse, p6 is a (180°), pulse and p3 is a 180° pulse, dl is the relax-
ation delay, ^ is the time for the second dimension, (a) Upper sequence is the one used for detecting
correlations through 1J(13C1H) as used for Fig. 9.24. d2 and d3 are 1/2J for ^("CH) and2 0.0036 s was
used, (b) The lower sequence is used to produce Fig. 9.25 where the correlation is through J(13C1H) and
3 13 1
J( C H). d2 and d3 are 1/2J for 2*J(13C1H) and 0.14 s was used. d4 and d5 are 1/2J for 1J(13C1H)9 0.0036 s.
cpd represents composite pulse decoupling.

www.pdfgrip.com
292 Two-dimensional NMR spectroscopy

therefore necessary to wait a period of dl for the two components to

come into phase before the 90° pulse is applied. The delay dl, which
is fixed, is made to be l/2/(CH). A 90° pulse is applied at the carbon
frequency at the same time as the second proton pulse, which creates
transverse carbon magnetization and produces an output. It is conve-
nient to remove the multiplicity due to coupling to the protons from
the carbon signals and this is done using broad-band decoupling.
However, again because of the different phases of the multiplet signals,
it is necessary to introduce a refocusing delay d2, which often is of
the same length as dl, before decoupling and signal acquisition are
initiated. The experiment is repeated for many values of ^ and a stack
of FIDs obtained, which contain both carbon and proton chemical shift
information for all the C-H pairs in the molecule. The pulse sequence
is tuned to ^(CH) because of the values chosen for dl and d2 and
which also suppresses any responses due to the much smaller two- and
three-bond couplings. It follows that, in this type of hetero COSY
spectrum, quaternary carbon atoms will not give a response. Numerical
treatment of the data produces a two-dimensional chart in which there
is a response for each CH pair at coordinates (8C),(8H).
A spectrum obtained in this way for D-amygdalin is shown in
Fig. 9.24, with the 13C spectrum along the horizontal axis and proton
on the vertical. The response was tuned to a 13C-1H correlation through
1 13 1
J( C H) by choosing values of d2 and d3 of 0.0036 s corresponding to
an average coupling constant of 139 Hz. The lack of significant response

-3.5
-4.0
-4.5
HOD
-5.0 3f
"D
-5.5 I
CH- -6.0
-6.5
-7.0
m,p -7.5
130 120 100 90 80 70 60 50
13
C/ppm

Figure 9.24 100.62MHz 13C-1H NMR correlation plot for hydrogen and
carbon atoms that are directly bonded in the molecule D-amygdalin in D2O.

www.pdfgrip.com
 Heteronuclear COSY NMR spectroscopy 293

for the phenyl quaternary carbon and CN carbon atoms is evident. It

is also possible to read off the carbon chemical shifts corresponding to
the now known proton assignments. Also note that the CH2 groups due
to A6 and B6 give two responses in the 1H dimension corresponding to
the two different hydrogen environments. This can be a valuable tool
in identifying chemically inequivalent CH2 protons.
A correlation through 1J(13C1H) is very valuable for assigning indi-
vidual 13C NMR signals. However, no correlation is observed for 13C
nuclei which do not bear protons. Such nuclei can be detected through
2 13 1
J( C ¥L) and 3J(13C1H). In principle, a correlation can be achieved
using the pulse sequence given in Fig. 9.23(a), using values of d2 and d3
of about 0.14 s, appropriate for the much smaller values of 2J(13Clt£)
and 3J(13C1H). This can also result in correlations through V^CH).
This problem can be reduced by lengthening the pulse sequence. This
sequence is shown in the lower part of the Fig. 9.23(b) and gives an idea
of the way it is possible to manipulate spin systems to obtain particular
objectives. Three pulses are added to the proton sequence and one to
the carbon sequence, with a particular relationship between them, which

140 120 100 80 60

13
C/ppm
Figure
3
9.25 Heteronuclear proton-carbon chemical shift correlation of D-amygdalin optimized for 2/(CH)
and /(CH) long-range couplings (6.25 Hz). Direct responses due to ^(CH) were suppressed with the use
of a BIRD pulse. Major inter-residue, long-range responses have been labelled. Additional intra-residue,
long-range correlations include:
Ring A: AH2 -» AC 1; AH2 -> ACS; AH3 -> AC2; AH3 -> AC4; AH4 -> AC3; AH4 -> ACS;
AH4 -> AC6; AH5 -> AC3.
Ring B: BH2 -» BC3; BH3 -> BC2; BH3 -> BC4; BH4 -^ BC5 and BH6 -» BC4.
Aryl ring: oriho H —> para C; meta H —> quaternary; meta H —> ortho C; para H —> ortho C; para H —> meta
C. (From Ribiero (1990) Magn. Reson. Chem., 28, 765-73, copyright (1990) John Wiley and Sons Ltd,
reprinted with permission.)

www.pdfgrip.com
294 Two-dimensional NMR spectroscopy

is called a BIRD (Bilinear Rotation Decoupling) pulse, see section 6.3.2.

d4 and d5 in the BIRD pulse are chosen as 0.0036 s to optimize the
rejection of correlations through 1J(13C1H). The resulting spectrum is
shown in Fig. 9.25 and provides a great deal of extra data, which gives
unequivocal assignments. For instance, the BjH to A6C and A6H to BjC
responses show the connectivity between the two rings.
We have thus used a series of two-dimensional experiments to make
a full assignment of the proton and carbon spectra of D-amygdalin and
obtain unequivocal evidence about its structure, despite the very
congested and unpromising appearance of the one-dimensional spec-
trum as a source of useful information. We have described five two-
dimensional techniques. A recent review (see Bibliography) lists some
40 pulse sequences, and there are said to be perhaps 500 variations
of two-dimensional spectroscopy proposed currently. There exists,
then, a whole battery of techniques that permit the successful exami-
nation of molecules much more complex than D-amygdalin, and
two-dimensional NMR is being extensively used by biochemists to
understand the properties (and, indeed, structures in solution) of many
molecules found to regulate the behaviour of living systems.

9.4 HOHAHA OR TOCSY

In many molecules there are groups of independent coupled protons,

for example in a peptide where each amino acid residue gives a
coupled spin system, but there is negligible coupling between each
residue. Amygdalin is another example, where there is coupling within
each ring, but not between them. HOHAHA (HOmonuclear HArtman
HAhn), or TOCSY (TOtal Correlation SpectroscopY) provide a
method to separate the various spin systems and hence identify the
single residues. The technique uses a spin-lock using the MLEV-17
pulse sequence. Spin-locking has been described in section 6.1.2.3. The
MLEV-17 pulse sequence is similar to the WALTZ sequence and is
based on a sequence of phase-cycled composite 180° pulses,
(90°)JC(1800)y(90°)Jc. The pulse sequence is given in Fig. 9.26 and its
application to D-amygdalin in Fig. 9.27.

d1 p1 ?1 p2 p3 p2
Figure 9.26 The basic pulse sequence for the HOHAHA or TOCSY experi-
ment, dl is a relaxation delay, pi is a 90° pulse, ^ is the variable delay for
the second dimension, p2 is a pulse, typically 2.5 ms, attenuated to the spin
lock level, and p3 is the spin lock pulse sequence, where the transmitter power
has been attenuated so that the 90° pulse is of the order of 40 JJLS.

www.pdfgrip.com
 HOHAHA or TOCSY 295

3.5

4.0 f
3

4.5

4.5 4.0 3.5

1
H/ppm

Figure 9.27 A partial 400 MHz JH HOHAHA or TOCSY experiment applied

to D-amygdalin in D2O. Note the two separate sets of signals for the protons
of the two sugar residues.

4.6 4.4 4.2 4.0 3.8 3.6 3.4 3.2

1
H/ppm

Figure 9.28 Two cross-sections from the 400 MHz 1H HOHAHA or TOCSY experiment shown in Fig.
9.27. (a) A cross-section though 8 4.58. (b) A cross-section through 8 4.63. Note the two separate sets of
signals for the protons of the two sugar residues.

www.pdfgrip.com
296 Two-dimensional NMR spectroscopy

9.5 TWO-DIMENSIONAL INADEQUATE

INADEQUATE (Incredible Natural Abundance DoublE QUAntum

Transfer Experiment) is the ultimate experiment to determine connec-
tivity in the carbon skeleton of a molecule. It uses 1/(13C13C) to map
out which carbon atoms are attached to each other. The problem is
that 13C is only 1.1% abundant. In general, this means that only 0.012%
of the molecules have two connected 13C atoms. The result is that very
concentrated solutions are required. The use of field gradients
improves the sensitivity of the experiment by at least a factor of ten.
COSY will produce the same results in theory, but the molecules
containing isolated 13C nuclei give very strong signals which obscure
the correlations. INADEQUATE is a double quantum transfer exper-
iment and selects AX spin systems, and hence selects coupled nuclei.
The pulse sequence is given in Fig. 9.29. An INADEQUATE 13C
NMR spectrum is shown in Fig. 9.30.
INADEQUATE can be applied to detect homocoupling between
any pair of low abundance nuclei. An example is given in Fig. 9.31
where the technique is applied to 183W in [SiVWuO40]5-. This
anion has a structure closely related to [AlC^Al^OH^H^O)^7*
(Fig. 7.30), where the central tetrahedral aluminium has been replaced
by silicon, eleven octahedral aluminium atoms have been replaced by
tungsten, and one octahedral aluminium atom has been replaced
by vanadium so that the majority of the tungsten atoms are in distin-
guishable sites.

9.6 OVERHAUSER AND MAGNETIZATION TRANSFER

BASED TWO-DIMENSIONAL NMR SPECTROSCOPY

It remains to be ascertained which of rings A and B of D-amygdalin

is attached directly to the CHCN group. This is done simply in this
case by irradiating the CHCN proton at 5.88 ppm and observing which
of the doublets around 4.6 ppm (the HI protons) show an NOE
enhancement. It proved to be the one slightly to high frequency at

d1 p1 d2 p2 d2 p3 p4

Figure 9.29 The INADEQUATE pulse sequence, dl is for relaxation, pi, p3,
and p4 are 90° pulses, d2 is 1/47', where J is the coupling constant between
the nuclei it is wished to detect, usually, 1J(13C13C), tl9 is the second dimen-
sion for the two-dimensional experiment, and p2 is a 180° pulse.

www.pdfgrip.com
 Overhauser and magnetization transfer based spectroscopy 297

(9.2)

200 175 150 125 100 75 50 25 0

13
C/ppm

Figure 9.30 100.62 MHz 13C INADEQUATE experiment applied to carvone,

(9.2), in CDC13. There is a lone signal at 843 which arises from the close
chemical shifts of C5 and C6, giving an AB spin system.

4.61 ppm, and this is then in the A ring, as, conveniently, we have
labelled the peaks throughout. The inverse experiment is also possible,
in which the doublets at 4.61 and 4.58 ppm are irradiated in turn and
the effect observed on the CHCN proton. This leads to the same
conclusion, though with less clarity, since it is not possible to irradiate
such close resonances with sufficient selectivity so that one gives an
NOE and the other none. A weaker effect is detected in the latter
case due to spill-over of the irradiation power into the 4.61 ppm
doublet region.
It is, of course, possible to carry out a two-dimensional version of
this experiment on D-amygdalin. Such experiments are very important
in the case of large, flexible biomolecules such as peptides. In solu-
tion, it is possible, once the proton resonances of identifiable residues
have been assigned, to determine which are in close proximity in space.
Thus the way the chains of such molecules are folded can be ascer-
tained, and the data currently being obtained in solution studies of
large molecules are comparable in accuracy with crystallographic data
on the same molecules. This experiment is called Nuclear Overhauser

www.pdfgrip.com
298 Two-dimensional NMR spectroscopy

--1000

- -500
f. *

- 500

1000
500 -500 Hz

-80 -100 -120 -140 -160

183
W/ppm

Figure 9.31 The 15 MHz two-dimensional INADEQUATE 183W NMR spec-

trum of 51V decoupled l^SiVWuOJ, 0.6M, 30°C, 1536 transients, 128 x 2k
files, 115 h, 20 mm sideways tube. Note that the signals due to W(2) and W(3)
are weak. (Reproduced by permission of the American Chemical Society from
Domaille (1984) /. Am. Chem. Soc., 106, 7677.)

Effect Spectroscopy (NOESY), and while it will not be illustrated in

detail here, we will shortly discuss the investigation of chemical
exchange by two-dimensional spectroscopy, which uses essentially the
same technique.

9.6.1 NOESY
The NOESY (Nuclear Overhauser Enhancement SpectroscopY)
experiment is the two-dimensional version of NOE difference NMR
spectroscopy, see section 8.2. The NOESY pulse sequence is given in
Fig. 9.32.
The NOE builds up during the mixing time d2. This presents
an experimental problem. The individual contributions to 7\ from an

www.pdfgrip.com
 Overhauser and magnetization transfer based spectroscopy 299

FID

d1 P1 t\ p2 d2 p3

Figure 9.32 The NOESY pulse sequence, dl is the relaxation delay, pi, p2,
and p3 are 90° pulses, ^ is the time for second dimension, and d2 is the mixing
delay for the NOE to build up.

individual H-H dipole-dipole interaction are always longer than the

overall 7\ and T2. The result is that the signal is decaying while the
NOE is building up. The result is that the NOESY experiment is less
sensitive that the NOE difference experiment. It can be advantageous
to use the experiment if many correlations are required due to the
time that would be required to carry out all the NOE difference exper-
iments, but if only a few correlations are required it is better to use
NOE difference NMR spectroscopy. The application of NOESY to
D-amygdalin is given in Fig. 9.33. As with the one-dimensional exper-
iment, a correlation is found between the CH and A1 protons providing
the relative assignments of A1H and B1H.

7 6 5 4 3
1
H/ppm

Figure 9.33 A symmetrized 400 MHz 1H NOESY NMR experiment applied

to D-amygdalin in D2O. Note particularly the NOE observed between the CH
and A1 protons. This identifies the sugar ring next to the CH group.

www.pdfgrip.com
300 Two-dimensional NMR spectroscopy

9.6.2 Two-dimensional chemical exchange NMR spectroscopy,

EXSY
Chemical exchange has perforce to be a homonuclear process and is
studied in two dimensions by the pulse sequence shown above for
NOESY. If we have two uncoupled spins with different chemical shifts
that undergo slow exchange, then the pulse sequence affects them as
follows. The first 90° pulse places the magnetization in the xy plane
and the two components then precess at their individual frequencies
for a time tr At the end of tl9 they will each have a particular orien-
tation, and in general both will have x and y components of magne-
tization. The second pulse then turns this magnetization into the xz
plane, where the y components continue to precess about the z axis.
A short mixing time (r m ), typically 0.05 s, is then given during which
time there will be an exchange of magnetization. However, the z
magnetization of the two components depends upon ^; indeed, for
some ranges of ^ one of the two components will be inverted and the
exchange will cause quite marked changes in magnetization. Thus at
the end of Tm we have the two component frequencies modulated
by the exchange process as a function of £15 and this gives cross-peaks
in the transformed two-dimensional trace. A FID is produced by
moving the z magnetization back into the xy plane at the end of im.
We should also note that, in non-exchanging molecules, if there exists
through-space relaxation, then there can also be an exchange of mag-
netization through the NOE effect, since the second pulse produces

Figure 9.34 128.37 MHz nB EXSY NMR spectrum of a 1.05 :1.0 M mixture
of BC13 and BBr3 at 400K using a mixing time of 0.05 s. (Reproduced with
permission from Derose et al (1991) /. Magn. Reson., 93, 347.)

www.pdfgrip.com
 Overhauser and magnetization transfer based spectroscopy 301

non-equilibrium spin distributions. The same pulse sequence is thus

also used for NOESY experiments as mentioned above. The acronym
EXSY is often used for the exchange experiment.
A relatively simple example of the use of the EXSY pulse sequence
to map out exchange is for a mixture of BC13 and BBr3 where the
halogens scramble (Fig. 9.34). The cross peaks clearly demonstrate that
exchange involves the transfer of only one halogen. Hence BBr2Cl
exchanges with BBrCl2 and BBr3, but not with BC13. This result was
obtained by using a mixing time of 0.05 s. At 400K this time is only
long enough for a significant number of single exchanges to occur. If
a longer mixing time had been used, then cross peaks corresponding
to the transfer of two or three halogens would have been detected due
to there being time for two or three exchanges to occur.
A more complicated EXSY spectrum is given in Fig. 9.35 for
exchange between isomers of [Cr(CO)2(CS){P(OMe)3}3]. In solution,
the compound exists as three isomers, (9.3), (9.4), and (9.5). All three
isomers give AB2 31P NMR spectra. It was proposed that the mer-
isomers undergo interconversion via the trigonal twist mechanism (Fig.
9.36).

178-
179-
180-
181 -
182-
c 183-
£184-
o: 185-
£ 186-
187-
188-
189-
190-
191 -
' -/
192-
• n—i—i—i—i—i—i—i—i—i—i—i—i—i—r~^
192 190 188 186 184 182 180 178
OH .
V/ppm

Figure 9.35 121.5MHz 31P EXSY NMR spectrum of [Cr(CO)2(CS)

{P(OMe)3}3] in CD3C6D5 at 61°C. The three isomers are identified by symbols,
X, (9.3), •, (9.4), and V, (9.5). Note that the doublet at 8 189 in the one-
dimensional NMR spectrum has been truncated. (Reproduced with permis-
sion from Ismail et al (1985) Organometallics, 4, 1914, copyright (1985)
American Chemical Society.)

www.pdfgrip.com
302 Two-dimensional NMR spectroscopy

Figure 9.36 The trigonal twist mechanism applied to [Cr(CO)2(CS)

{P(OMe)3}3]. Two of the P(OMe)3 ligands rotate about the pseudo C3 axis via
a trigonal bipyramidal indermediate.

9.6.3 CAMELSPIN or ROESY

NOE changes its sign as TC increases, see section 8.2. The result is that
when the molecular mass is in excess of 1000, NOESY measurements
are generally unsuccessful. The problem is removed by working under
spin-lock conditions where the effective co in equations (8.1) - (8.3) is
reduced from the NMR frequency range to the kHz range. The result
is that TC is very short compared with the effective o> and the experi-
ment operates in the positive NOE region. The experiment is called
ROESY (Rotating frame Overhauser Enhancement SpectroscopY)
and uses the pulse sequence given in Fig. 9.37.
A typical experiment is shown in Fig. 9.38. In this example,
composite 180° spin-locking pulses were used.
ROESY generally gives stronger correlations than NOESY as the
irradiation is continued for typically 300 ms. However, it does suffer

d1 P1 p2 p3

Figure 9.37 The ROESY pulse sequence, dl is the relaxation delay, pi is a

(90°X- pulse, ^ is the time for second dimension, p2 is a (180°) pulse and p3
is a (180°)y pulse and the transmitter is attenuated so that 180° pulses are
approximately 180 JJLS. The 180° pulses make up a train of n pulses taking
typically 300 ms.

www.pdfgrip.com
 Overhauser and magnetization transfer based spectroscopy 303

8 7 6 5 4 3
1
H/ppm

Figure 9.38 The 400 MHz two-dimensional 1H ROESY spectrum of D-amyg-

dalin in D2O.

from the problem that correlations can occur through coupling, arising
from the same process as is exploited in TOCSY. Consequently more
caution is necessary in interpreting the results. The key observation
in Fig. 9.38 of the correlation between the CH group and the A:H
proving the relative correlation between these two protons, is safe as
no coupling pathway exists between them.

9.6.4 HOESY
Two-dimensional NOE can be used to link different species of nuclei,
when the technique is called HOESY (Heteronuclear Overhauser
Effect SpectroscopY). The technique works well whenever there is a
significant dipole-dipole relaxation process operating between the
nuclei. This normally means pairs of nuclei such as !H and 13C or 31P,
but in favourable cases quadrupolar nuclei can be involved. The
example chosen here is of 6Li-lH HOESY. Due to the low quadrupole
moment of 6Li, dipole-dipole relaxation is an important relaxation
pathway, and hence the Overhauser effect is observed. The compounds

www.pdfgrip.com
304 Two-dimensional NMR spectroscopy

(9.6) (9.7)

(9.6) and (9.7) (where L is monodentate Me2NCH2CH2NMe2) are in

equilibrium in solution and so their proton spectra are always seen
together. In this case we use the Overhauser effect between protons
and 6Li to show which protons are close to which lithium. The pulse
sequence is given in Fig. 9.39 and it will be seen that the inversion of
the 6Li magnetization will perturb the magnetization of proximate
protons. The two-dimensional plot is shown in Fig. 9.40 where it will
be seen that the monomer shows a strong response to the bidentate
ligand and also to protons 9, 14 and 18, whereas in the more tightly
arranged dimer, the pendant ligands show no interaction and interac-
tion is seen with almost all the protons of the unsaturated ligands.

9.7 INVERSE DETECTION

Inverse detection is a very powerful way of detecting insensitive nuclei

attached to sensitive nuclei. The method uses the sensitivity of the

P3 P4

Figure 9.39 The HOESY pulse sequence, dl is for relaxation, d2 is the mixing
time for the NOE to develop, pi, p2, and p4 are 90° pulses, f/2 is the second
dimension for the two-dimensional experiment, and p3 is a 180° pulse.

www.pdfgrip.com
 Inverse detection 305

JL

1.6 1.41.2 1.0 0.80.60.40.2

6
Li/ppm

Figure 9.40 A contour plot of a phase-sensitive 58.8 MHz 6Li/H HOESY

NMR spectrum of an equilibrium mixture of (9.6) and (9.7) in THF-d8 at
-64°C. (Redrawn by permission of John Wiley and Sons Ltd, from Bauer and
Schleyer (1988) Magn. Reson. Chem., 26, 827.)

sensitive nucleus, frequently !H, to observe other spin-coupled nuclei.

Inverse detection through two-dimensional correlation spectroscopy
can, in theory, give a gain in sensitivity of (yH/yx)5/2 over straight
detection. For 13C this gives an increase in signal strength of 31.6
compared with no enhancement or 7.9 compared with an INEPT or
DEPT spectrum. For lower y nuclei the gain is even more dramatic
(Table 9.1). These gains are actually overstated as the XH NMR spectra
are 1H-1H coupled multiplets and consequently the full gain is not
achieved. However, the gains are useful and inverse detection is gener-
ally preferred.
There are problems associated with inverse detection. Firstly, low
frequency nuclei such as 103Rh have relatively long 90° pulses, with
the consequence that, as 1/4PW is small, they can only be detected
over a fairly narrow frequency range. However, the correct frequency
can be estimated by using a BIRD type pulse sequence. Secondly,

www.pdfgrip.com
306 Two-dimensional NMR spectroscopy

Table 9.1 The intensity gains on using inverse detection of several nuclei,
X, by 1H. Two gains are given. ("W/x)572 *s tne §am relative to simple
detection and (^/p/Yx)372 *s tne £am relative to an INEPT or DEPT spec-
trum.
Observe nucleus 31p
13 15 29 m 15
Inverse detect C N Si Rh N 103
Rh
(V7x) 5/2 31.6 306 56.8 5634 31.9 587

(V7x)3/2 1.9 31 11.3 178 8.0 45.8

nuclei such as 13C have shifts spread over a wide frequency range. In
order to obtain reasonable resolution in the fj/jfj dimension, a large
number of spectra have to be acquired.
We will describe two inverse detection experiments, HMQC or
Heteronuclear Multiple Quantum Coherence and HMBC or Hetero-
nuclear Multiple Bond Correlation. HMQC is the simplest 1H,X corre-
lation experiment and is normally combined with BIRD selection (see
section 6.3.2) of the magnitude of the coupling to be observed and
GARP decoupling of the X nuclei (see section 6.1.5) and it is this
combination which is given here. The pulse sequence is given in
Fig. 9.41. Delay d2 is l/n/(XH) and d3 is set so the 1H spin vectors
are out of phase. HMQC is an alternative to heteronuclear COSY (see
section 9.3) but is generally more sensitive.
HMBC is used to study the two or three bond coupling correlations.
This can be done using HMQC or :H,X COSY (see section 9.3) but
HMBC suppresses more effectively the one bond coupling correlation.
The pulse sequence is given in Fig. 9.42. Both sequences start with the
BIRD sequence which, for HMQC is tuned to the coupling interaction
desired and for HMBC is tuned to ^(XH), the smaller couplings being
selected by d4.

Figure 9.41 The pulse sequence used for the HMQC experiment with BIRD selection and GARP decou-
pling, dl is a relaxation delay, pi, p3, p4, p8, and p9 are 90° pulses, p2, p5, and p6 are 180° pulses, d2 is
1/27, d3 is the BIRD delay, optimized for minimum 1VL NMR signal, and ^ is the variable delay for the
second dimension.

www.pdfgrip.com
 Inverse detection 307

Figure 9.42 The pulse sequence used for the HMBC experiment with BIRD selection, dl is a relaxation
delay, pi, p4, p7, p8, and p9 are (90°)^ pulses, p3 is a (90°).,, pulse, p2, p5, and p6 are 180° pulses, d2 is
1/21/, d3 is the BIRD delay, optimized for minimum 1H NMR signal, d4 is l/2n/ where n = 2 or 3, d5 =
d2 - d4, and ^ is the variable delay for the second dimension.

9.7.1 The study of

[RuCii^CsMesXTfi^Z-C^S-r^-C^S-Z-C^S)] (9.8)
The HMQC plot of this compound, tuned for ^(CH), is shown in
Fig. 9.43. The CH correlations are easily seen and it is also evident
which are the quaternary carbon atoms with no response. The proton

(9.8)

1
H/ppm
Figure 9.43 A 500 MHz 1H-{3C HMQC NMR spectrum of [Ru(-q5-C5Me5)
(Ti5-2-C4H3S-2/,6/-C4H2S-2-C4H3S)], (9.8), in (CD3)2CO. The experiment
permits the assignment of the hydrogen bearing carbon atoms. (Reproduced
with permission from Graf et al (1995) Inorg. Chem., 34, 1562, copyright
(1995) American Chemical Society.)

www.pdfgrip.com
308 Two-dimensional NMR spectroscopy

80

90
100 _,
CO

110§
•O

•120
•130

140
7.5 7.0 6.5
1
H/ppm

Figure 9.44 A 500 MHz !H-13C HMBC NMR spectrum of [Ru(Yi5-C5Me5)(T]5-

2-C4H3S-2',6'-C4H2S-2-C4H3S)], (9.8), in (CD3)2CO. The experiment permits
the assignment of the non-hydrogen bearing carbon atoms. (Reproduced with
permission of the American Chemical Society, from Graf et al. (1995) Inorg.
Chem., 34, 1562, copyright (1988).)

-5850 £
Ol
-o
c*
T3

-5800 3

-5750
6 5 4 3 2 1
1
H/ppm

Figure 9.45 400 MHz lH,l95Pt two-dimensional inverse correlation NMR spec-
trum for the two isomers of [Pt(SnCl3)(Ti3-C4H7)(Ti2-PhCH=CH2)]. The normal
!
H NMR spectrum is superimposed on the spectrum. The 195Pt chemical shifts
are referenced to H2PtCl6. (Reproduced by permission of the American
Chemical Society, from Pregosin et al (1988) Organometallics, 7, 2130, copy-
right (1988).)

www.pdfgrip.com
 Inverse detection 309

5(1H) = -16.28

•100

100
Hz
-16.2 -16.4
1
(a) H/ppm

8(31P) = 36.1 ;

1291
1292
CJl
1293 -^
(D^

1294 •§
1295
1296

5950 5900 5850 5800 5750

31
(b) P/Hz

Figure 9.46 Inverse !two-dimensional

57
NMR 31spectra
57
of [(iq5-C5H5)Fe(H)
(PMe3)2] at 300K. (a) H, Fe spectrum and (b) P, Fe spectrum with single
frequency decoupling of the methyl protons. (Reproduced from Benn, in
Transition Metal Nuclear Magnetic Resonance, ed. P.S. Pregosin, Elsevier,
Amsterdam, 1991, p. 112, copyright (1991), with permission from Elsevier
Science.)

www.pdfgrip.com
310 Two-dimensional NMR spectroscopy

X
p7 p8 p9 p10

Figure 9.47 The pulse sequence used for the HSQC experiment with BIRD selection, dl is a relaxation
delay, pi, p5, p8, and p9 are (90°)x pulses, p3 and p5 are (90°) pulses, p2, p4, p6, p7, and plO are 180°
pulses, d2 is 1/41/', and ^ is the variable delay for the second dimension.

spectrum is partially assignable since protons 3' and 4' give an

AB spectrum and the other two rings give two doublets and a triplet
with the ring bonded to ruthenium giving a low frequency chemical
shift.
The HMBC plot is shown in Fig. 9.44 with impurity peaks marked
with an asterisk. This permits the assignment of the non-proton bearing
carbon atoms by the observation of correlations through the longer
distance coupling paths. It also helps to sort out ambiguities in the
proton spectrum. The quaternary carbon correlated with protons 3, 4,
and 5 is also correlated with a resonance of the AB spectrum, which
is thus proton 3', and this leads us on to assign C2/ via its weak corre-
lation with H4'. Similarly, H3 is correlated with C2, and we can assign
the whole spectrum in this way.

9.7.2 Other applications of the HMQC method

A typical application of HMQC is shown in Fig. 9.45. The complex
[Pt(SnCl3)(T]3-C4H7)(Ti2-PhCH=CH2)] exists as two isomers in solution
and the 1H spectrum is complex as the two spectra are inter-
leaved. The HMQC spectrum in which X = 195Pt, [(yH/yPt)512 = 44.95]
shows two platinum resonances at 8 -5770 and 8 -5875 ppm in the /j
dimension and the associated proton resonances in the /2 dimension.
The normal !H spectrum is superimposed on the two-dimensional
NMR spectrum and consists of a series of resonances flanked by their
195
Pt satellites which are the only ones to appear in the two-dimen-
sional NMR spectrum, where it is clear which proton resonances are
associated with which isomer resonance.
Inverse detection has been used with considerable success to observe
very insensitive nuclei such as 57Fe or 187Os either through 1H or 31P
[(V^Fe)5/2 = 5263; (V^Fe) c
5/2
= 55°; (V70s)5/2 = 1257264; (yp/y0s)512 =
1282; the receptivities R of the two elements are Fe = 4.25 x KH

www.pdfgrip.com
 Three-dimensional NMR spectroscopy 311

3-

4-
H^~

1
Q.
,0.
I

" 5

6•* i i

10 9 . 8 7
(a) ^/ppm/^

4-
s-T

Q. .
^Q.
I

5-

6-
10 9 H1 8 7
(b) H/ppm//3

Figure 9.48 Two- and three-dimensional 600 MHz :H NMR spectra of inter-
leukin-lp. In both cases, NOESY is used to identify the connection 15
between
the NH and the aCH protons of individual amino acid residues. N editing
was used so that only 15NH groups are observed, (a) The two-dimensional
NOESY spectrum showing the large number of responses from the amino
acid residues, (b) A three-dimensional NOESY spectrum with the 15N chem-
ical shift being used as the /2 direction. One slice is plotted corresponding to
a 8(15N) = 123.7. The introduction of the third dimension has considerably
simplified the problem. (Reproduced with permission from Bax et al (1989)
Biochemistry, 28, 6150, copyright (1989) American Chemical Society.)

www.pdfgrip.com
312 Two-dimensional NMR spectroscopy

and 187Os = 1.15 x 10~3]. An example is shown in Fig. 9.46 where the
57
Fe signal of [Fe(Ti5-C5H5)H(PMe3)2] is observed through (a) 1H or
(b) 31P. The 1H spectrum of the 57Fe containing species is a triplet of
doublets though some intensity shows through from the non 57Fe
complex, which is almost 50 times more abundant, to give a triplet of
triplets. The 57Fe resonance is also a triplet of doublets. In (b) only
the couplings to P and H show.

9.7.3 HSQC
The HSQC (Heteronuclear Single Quantum Coherence) experiment
produces similar correlations to those obtained using the double
quantum experiment, HMQC, but has the advantage of removing WH
coupling leading to sharper signals. This offers considerable advan-
tages for larger molecules with crowded spectra. The pulse sequence
is given in Fig. 9.47.

9.8 THREE-DIMENSIONAL NMR SPECTROSCOPY

It will come as no surprise that the techniques of two-dimensional

NMR can be combined to produce experiments into which further
dimensions are introduced. Possibilities include NOESY-COSY
spectra or correlations between 15N in 15N-enriched molecules with
the proton-proton COSY spectrum. While such techniques are in
their infancy, they should assist in improving the resolution of the
spectra of complex molecules, and are likely to take up an increasing
amount of the time of NMR spectroscopists. The example chosen here
is where a ^^H NOESY spectrum is greatly simplified. The spec-
trum is of interleukin-lp (Fig. 9.48). The two-dimensional NMR
spectrum was first simplified so that only NH groups are observed by
using 15N filtering via a BIRD like sequence. NOEs are observed to
the a-CH protons. Due to the very large number of amino acid
residues in the molecule, the spectrum is still very complex. A third
dimension, that of the 15N chemical shift, is then added and the
signals are sorted in this third dimension giving sets of much simpler
spectra.

9.9 QUESTIONS

The following selection of problems is chosen to test you on the knowl-

edge that you have acquired from the first nine chapters of this book.
They do not refer exclusively to the information in Chapter 9.
9.1. Figure 9.49 shows the 119Sn NMR spectrum of a mixture of
[But(F)Si(OSnBut20)2Si(F)But] and [But(F)Si(OSnBut2)2O(fju-F)2
SnBul2] in CDC13. The two compounds in the solution are:

www.pdfgrip.com
 Questions 313

Bu* Bu*2

Assign the 119Sn signals and give the chemical shifts of each
different type of tin. Account for the multiplicity of the signals.
Derive the /(119Sn19F) values, where observed.

-150 -200 -250 -300

119
6( Sn)

Figure 9.49 149.2MHz 119Sn{1H} NMR spectrum of a mixture of

[But(F)Si(OSnBut2O)2Si(F)But] and [But(F)Si(OSnBut2O)2(fji-F)2
SnBul2] in CDC13. x marks an impurity. (Reproduced with permission
from Jurkschat et al (1998) Organometallics, 17, 5697, copyright (1998)
American Chemical Society.)

9.2. Figure 9.50 shows the 31P and 195Pt NMR spectra of a mixture of
two products formed by the addition of H~ to

Given that [(V-C7H7)Mo(CO)3]+ reacts with H- to give [(r|6-

C7H8)Mo(CO)3], suggest two products which would account for
the NMR spectra and derive the 1/(195Pt31P) values.
Given that [PtCl4]2- is at 8 2887 relative to S(195Pt) = 21.4 MHz,
re-reference the signals in Fig. 9.50(a) to E(195Pt) = 21.4 MHz.

www.pdfgrip.com
314 Two-dimensional NMR spectroscopy

(a)
• i i p i rrp-m-| i 1 1 1 , 1 1 . 1 p i i i | M i i p M i | i i i i p i i i p i i i p i i i p i i i p i i i | i i i i | M i i | i i i i p i i i p i i i p i i i | i i i i p M i pn . I , . . . . , ,

-4400 -4450 -4500 -4550

5(195Pt)

(b) JJL
\ l^ r
40 35 30 25 20 15 10 5
8(31P)
Figure 9.50 The NMR spectra of the two products of the reaction of [{Ti2-(Ph3P)2Pt}{7i6-Mo(CO)3}C7H5]+
with H-. (a) 64 MHz 195Pt NMR spectrum, referenced to K2PtCl4 in D2O. (b) 121 MHz 31P NMR spec-
trum. (Reproduced with permission from Jones et al (1996) Organometallics, 15, 596, copyright (1996)
American Chemical Society.)

9.3. Figure 9.51 shows the cyclopentadienyl region of the !H and 13C
NMR spectra of [{7i5-C5H4P(OC6H4NMe)2}Fe(CO)2Me]. Account
for the number of signals and derive the chemical shifts of the
signals and in the case of 13C the /(31P13C) values.

www.pdfgrip.com
 Questions 315

1 1

(a)
1 ' ' ' 1 ' ' 1
1 i I > i i I i i
5.4 5.2 5.0 4.8 4.6
8(1H)

(b)
^lAri^*Vw/vVV^VW^Vv''/s'v^^^
I . . . . I >
97.5 95.0 92.5 90.0 87.5 85.0
13
5( C)

Figure 9.51 The cyclopentadienyl region in the (a) 300 MHz JH NMR spectrum and (b) the 75 MHz
} NMR spectrum of [{ri5-C5H4P(OC6H4NMe)2}Fe(CO)2Me] in CDC13. (Reproduced with permission
from Nakazawa et al. (1998) /. Am. Chem. Soc., 120, 6715, copyright (1998) American Chemical Society.)

9.4. Figure 9.52 shows the 31P{1H] NMR spectrum of 3-Cl2P-5-Ph2P-

1,2-P2C3H. Explain how the coupling pattern arises.

p2_p3

www.pdfgrip.com
316 Two-dimensional NMR spectroscopy

31000 30000 18300 17900 2000 Hz

Figure 9.52 The 109.38MHz 31PpH} NMR spectrum of 3-Cl2P-5-Ph2P-l,2-P2C3H in pyridine. x marks an
impurity signal. (Reproduced with permission from Schmidpeter et al. (1998) Eur. J. Inorg.Chem., 1907,
copyright Wiley-VCH.)

9.5. Figure 9.53 shows the !H NMR spectrum of the SMe2 signal of
[Pt 2 (C 6 H 4 CH=N-l-C 6 H 10 -2-N=CHC 6 H 4 )Me 4 Br 2 (^-SMe 2 )],
complete with 195Pt satellites. Account for the coupling pattern.

1.70 1.69 1.68 1.67 1.66 1.65 1.64 1.63

5( 1 H)
Figure 9.53 The 300 MHz ^H NMR spectrum of the SMe2 group of
[Pt2(C6H4CH-N-l-C6H10-2-N=CHC6H4)Me4Br2(fjL-SMe2)] in CDC13. x
marks an impurity. (Reproduced with permission from Puddephatt
et al. (1998) Organometallics, 17, 32, copyright (1998) American
Chemical Society.)

www.pdfgrip.com
 Questions 317

9.6. Figure 9.54 shows the ^P^H} NMR spectrum of [Pd(cw-Ph2

PCMe=CHPPh2)2]2+ Assign the signals and account for the
coupling pattern. {Remember that 2/(31P31P) trans across platinum
is several hundreds of Hz, while cis is typically 10-20 Hz.}

80 70 _. 60 50
8(31P)

Figure 9.54 The 101 MHz 31P NMR spectrum of [Pd(cw-Ph2PCMe=

CHPPh2)2]2+ in CD3CN. (Reproduced with permission from Higgins
et al (1998) /. Chem. Soc., Dalton Trans., 1787.)

(d)_

(c)_

(b)

(a)

-50 0 50 Hz
Figure 9.55 The magnetic field dependence of the *H NMR signal of
H4 of [HB(5-Bul-pyrazolyl)3Tl] in CDC13 at room temperature, (a)
200 MHz. (b) 300 MHz. (c) 400 MHz. (d) 500 MHz. (Reproduced with
permission from Parkin et al (1998) /. Am. Chem. Soc., 120, 10416,
copyright (1998) American Chemical Society.)

www.pdfgrip.com
318 Two-dimensional NMR spectroscopy

9.7. Figure 9.55 shows the effect of magnetic field on H4 of [HB(5-

But-pyrazolyl)3Tl]. Explain why there is a doublet observed at
200 MHz, but a broad singlet is observed at 500 MHz.

w_

(a)

6.9 6.8 6.7 6.6 6.5 6.4 6.3 6.2

1
8( H)

Figure 9.56 The magnetic field dependence of the *H NMR signal of

H4 of [HB{3-(2-thienyl)-3-CF3-pyrazolyl}3Tl] in CDC13 at room temper-
ature, (a) 200 MHz. (b) 500 MHz. (Reproduced with permission from
Parkin et al (1998) /. Am. Chem. Soc., 120, 10416, copyright (1998)
American Chemical Society.)

Figure 9.56 shows the effect of magnetic field on the :H NMR

spectra of [HB{3-(2-thienyl)-3-CF3-pyrazolyl}3Tl]. Explain the
differences in the appearance of the spectra between 200 and
500 MHz.
31
9.8. Figure 9.57 shows the P{!H} NMR spectrum of [{(dppe)(2,6-
Me2C6H3NC)Pt}2Hg].

Analyse the spectrum and derive as many coupling constants

as you can.

www.pdfgrip.com
 Questions 319

x
olL li il JlL kJL
110 100 90 80 70 60 50 40
31
6( P)
Figure 9.57 The 100.6 MHz ^P^H) NMR spectrum of [{(dppe)(2,6-Me2
C6H3NC)Pt)2Hg] in CD2C12. x marks impurities. (Reproduced with
permission from Puddephatt et al. (1996) Organometallics, 15, 1502,
copyright (1996) American Chemical Society.)

9.9. Partial !H NMR spectra of [Ru{l,5-(C3H3N2CH2)2-2,4-Me2-C6H4}

(terpy)]2+ are shown in Fig. 9.58. Use the NOE difference spectra
to assign the 1H NMR signals due to the CH2 protons, H2, H5,
H5', H6", and H6'". Explain why the CH2 protons are inequiva-
lent. Explain the marked difference in chemical shifts of H6", and
H6'". What extra experiment(s) would you do to assign the
remaining protons?

www.pdfgrip.com
320 Two-dimensional NMR spectroscopy

(e)

(d)'

(c)<

(b)~

(aM • ••I••••i• •••i

10 9 8 7 6
1
5( H)

Figure 9.58 Partial 300 MHz *H NMR spectra of [Ru{l,5-(C3H3N2CH2)2

-2,4-Me2-C6H4}(terpy)]2+ in (CD3)2CO. The only signal omitted is that
of the methyl groups, (a) The simple spectrum, (b)-(e). NOE difference
spectra with presaturation at (b) CH3 (c) 86.0; (d) 85.6; (e) 87.4.
(Reproduced with permission from Steel et al. (1998), Organometallics,
17, 3487, copyright (1998) American Chemical Society.)

9.10. The 15N NMR spectrum of [Ru(NH3)4(2-bzpy)]2+ in Fig. 9.59

shows the three different types of 15NH3 groups. Use stick
diagrams to show the three separate 15NH3 INEPT signals.

2+

www.pdfgrip.com
 Questions 321

-400 -405 -410 -415 -420 -425 -430 -435 -440

8(15N)

Figure 9.59
2+
The partial 40.56 MHz 15N NMR spectrum of [Ru(NH3)4
(2-bzpy)] in d6-DMSO at room temperature recorded using the INEPT
pulse sequence. (Reproduced with permission from de Paula et al. (1999)
Polyhedron, 18, 2017, copyright (1999), with permission from Elsevier
Science.)

I T T
300 200 100 0 -100 -200 Hz

Figure 9.60 The 7.91 MHz 103Rh INEPT NMR spectrum of

[Rh{C(C6H4-2-OH)-N-2-6-Me-C5H4N}H{P(OMe)3}(PPh3)2]+ in CDC13
recorded using polarization transfer from 1H. x marks an impurity.
(Reproduced with permission from Pregosin et al. (1987) Magn. Res.
Chem., 25, 158, copyright John Wiley and Sons Ltd.)

www.pdfgrip.com
322 Two-dimensional NMR spectroscopy

9.11. Figure 9.60 shows the 103Rh INEPT NMR spectrum of

[Rh{C(C 6 H 4 -2-OH)-N-2-6-Me-C 5 H 4 N}H{P(OMe) 3 }(PPh 3 ) 2 ] + .
Analyse the coupling pattern and derive ^(^Rh1!!),
1 103
/{ Rh(31PPh3)2} and lJ{mRh31P(OMe)?}. Suggest a structure
which is consistent with this spectrum, given that the hydride is
trans to nitrogen.
9.12. Figure 9.61 shows the variable temperature XH NMR spectrum
of the methyl region of [AsH(SiBulMe2)2]. Account for the obser-
vation of two methyl signals at 10°C. On warming, they exchange.
Estimate AG^. Identify sources of error in AG*. Suggest a
mechanism for the exchange.

I I I I
0.5 0.4 0.3 0.2 0.1 0.0
8(1H)

Figure 9.61 The methyl region in the 400 MHz !H NMR spectrum of
[AsH(SiButMe2)2] in CDC13. The spectra are at (a) 10°C, (b) 20°C, (c)
30°C, (d) 40°C. (Reproduced with permission from Westerhausen et al
(1998) Inorg. Chem., 37, 619, copyright (1998) American Chemical
Society.)

www.pdfgrip.com
 Questions 323

9.13. Figure 9.62 shows the variable temperature XH NMR spectrum

of the CH2 protons of [Mo(CO)3{MeN(CH2-2-C5H3N-6-CH2)2
NMe}]. Explain why there are four signals. Explain why they
broaden equally at 280K. Explain why there is only one doublet
observed at 310K. Estimate AG*310. Identify sources of error in
AG*. Suggest a dynamic process which accounts for the changes
in the 1H NMR spectra.

(j)

(f)

(e)

(0_ (d).

(c)

(b)

(h) (a) UUL

6 5 4 6 5 4
8(1H)

Figure 9.62 The 360 MHz 1H NMR spectrum of [Mo(CO)3{MeN

(CH2-2-C5H3N-6-CH2)2NMe}] in (CD3)2NDO at (a) 250K, (b) 270K, (c)
280K, (d) 293K, (e) 300K, (f) 310K, (g) 320K, (h) 330K, (i) 350K, (j)
380K. (Reproduced with permission from Kelm, and Krtiger (1998) Eur.
J. Inorg. Chem., 1381, copyright Wiley-VCH.)

www.pdfgrip.com
324 Two-dimensional NMR spectroscopy

9.14. Figure 9.63 shows the central signals of the variable temperature
29
SipH} NMR spectra of ds-[Pt(PEt3)2(SiFMe2)2]. The 195Pt satel-
lites are outside the spectral range. Explain the multiplicity of
the spectra at -80°C and 40°C. Account for the changes in the
spectra with temperature. Suggest a mechanism of fluxionality
which is consistent with the spectra.

I I
64 62 60 58
5(29SJ)
Figure 9.63 The 79.3 MHz 29Si NMR spectrum of ds-[Pt(PEt3)2(SiFMe2)2]
in d8-toluene at (a) -80°C, (b) -50°C, (c) -30°C, (d) 20°C, (e) 40°C
(Reproduced with permission from Tsuji et al. (1998) Organometallics, 17,
507, copyright (1998) American Chemical Society.)

9.15. At 25°C, the 1H NMR spectrum of the phenyl group of [l-Ph-3-

(-ri5-C5Me5)-7-SMe2-3,l,2-c/050-RuC2B9H9] shows two sharp
triplets at 8 7.12 and 6.95 with further fine structure and a broad
signal at 8 6.80, with relative intensities 2 : 1 : 2 . On cooling to

www.pdfgrip.com
 Questions 325

-60°C, the signal at 8 7.12 appears to be a quartet, the signal at

8 6.95 does not change significantly, and two doublets appear at
8 6.85 and 6.54. The relative intensities of the four signals are
2 : 1 : 1 : 1. See Fig. 9.64.

(a)

(b)
7.2 7.1 7.0 6.9 6.8 6.7 6.6 6.5
8(1H)

Figure 9.64 A partial 400 MHz ]H NMR spectrum of [l-Ph-3-Cn5-

C5Me5)-7-SMe2-3,l,2-c/osi0-RuC2B9H9] in CDC13 showing the phenyl
protons, (a) At 25°C. (b) At -60°C. (Reproduced with permission from
Welch et al (1998) Organometallics, 17, 3227, copyright (1998) American
Chemical Society.)

www.pdfgrip.com
326 Two-dimensional NMR spectroscopy

20 10 0 -10 -20 -30

Figure 9.65 The 80.2 MHz nB NMR spectrum of [(9-BBN)B 10H12]- in

d7-DMF. (a) The *H coupled spectrum showing lJ(nBlH). (b) The
"B^H) NMR spectrum. (Reproduced with permission from Shore et al
(1998) Inorg. Chem., 37, 3276, copyright (1998) American Chemical
Society.)

^--30

|--20

|--10

I- o
I- 10

i- 20

20 10 0 -10 -20 -30

Figure 9.66 The 80.2 MHz 11E{1H] COSY NMR spectrum of [(9-BBN)
B10H12]~ in ^7-DMF. (Reproduced with permission from Shore et al.
(1998) Inorg. Chem., 37, 3276, copyright (1998) American Chemical
Society.)

www.pdfgrip.com
 Questions 327

(i) Assign the signals as far as possible,

(ii) Explain the observation of four signals at -60°C,
(iii) Explain why one signal is broad and the other two signals
are sharp at 25°C.
9.16. Assign the n B NMR signals of [(9-BBN)B10H12]- using the
spectra in Figs 9.65 and 9.66. Note that 1/(11B1H) coupling is only
observed between U B and terminal hydrogens. Only 1/(11B11B)
is observed in the COSY spectrum.
9.17. The XH NMR spectra in Figs 9.67-9.69 show the 1H NMR spectra
of [Y(Ph2CCPh2)]. Assign all the !H NMR signals.

7.5 7.0 6.5 6.0 5.5 5.0 4.5

1
5( H)

Figure 9.67 The 500 MHz !H NMR spectrum of [Y(Ph2CCPh2)] in d8-THF.

The numbers above each signal give their relative intensities. (Reproduced
with permission from Roitershtein et al (1998) /. Am. Chem. Soc., 120, 11342,
copyright (1998) American Chemical Society.)

www.pdfgrip.com
328 Two-dimensional NMR spectroscopy

A/1

(b)

(a)

4.68 4.64 4.60 4.56

6(1H)
Figure 9.68 The 500 MHz 'H NMR partial spectrum of [Y(Ph2CCPh2)] in dg-
THF. (a) Normal spectrum, (b) 'Hf^Y} NMR spectrum. (Reproduced with
permission from Roitershtein et al. (1998) /. Am. Chem. Soc., 120, 11342,
copyright (1998) American Chemical Society.)

7.5 7.0 6.5 6.0 5.5 5.0

5(1H)

Figure 9.69 The 'H enhanced gradient COSY spectrum of [Y(Ph2CCPh2)] in

dg-THF. (Reproduced with permission from Roitershtein et al. (1998) J. Am.
Chem. Soc., 120, 11342, copyright (1998) American Chemical Society.)

www.pdfgrip.com
 Questions 329

Fig. 9.70 shows the variable temperature 1H NMR spectra of

[Y(Ph2CCPh2)]. Explain the changes in the spectra as the temper-
ature is increased.

(en

[!_

(b)^_

(*!>

7.5 7.0 6.5 6.0 5.5 5.0 4.5

1
5( H)

Figure 9.70 The 400 MHz 'H variable temperature spectrum of

[Y(Ph2CCPh2)] in ^-dioxan. (a) 295K, (b) 323K, (c) 348K, (d) 373K. Note
that the sample is impure. (Reproduced with permission from Roitershtein
et al. (1998) /. Am. Chem. Soc., 120, 11342, copyright (1998) American
Chemical Society.)

www.pdfgrip.com
330 Two-dimensional NMR spectroscopy

9.18. The !H NMR spectra of the hydrides of [Ir2H4(N2C3H3)2

(NCMe)(PPri3)2] are shown in Fig. 9.71. Selectively decoupling
one 31P affects the signals at 8 -21.6, -23.3, and -24.15. Selectively
decoupling the other 31P affects the signals at 8 -20.3 and -24.15.
Assign the hydride signals.

Prj3P l/PPr's
NCMe

(a)
I
-20 -21 -22 -23 -24
5(1H)

24

22 5(1H)

20
-20 -22 -24 -20 -22 -24
5(1H) 6(1H)

Figure 9.71 (a) The 300 MHz 'H NMR spectrum of the hydride signals of
[Ir2H4(N2C3H3)2(NCMe)(PPri3)2]. (b) The COSY spectrum, (c) The NOESY
spectrum. (Reproduced with permission from Oro et al. (1998)
Organometallics, 17, 683, copyright (1998) American Chemical Society.)

www.pdfgrip.com
 Questions 331

9.19. Partial !H NMR spectra of [OsH3(NH-CPhC6H4))(PPri?)2] are

shown in Fig. 9.72. Use the NOESY spectrum to assign the
hydride signals. Explain the variable temperature 1H NMR spec-
trum. Derive two values of AG* from the variable temperature
!
H NMR spectrum and identify sources of error in AG*. Predict
the appearance of the hydride !H NMR spectrum at 263K if the
spectrum was measured at (i) 100 MHz, (ii) 600 MHz.

353 K

343 K

333 K
323 K
iaig
293 K
283 K
273 K
CH6^
263 K
253 K

243 K

233 K
NH -
223 K

(a) 213 K

203 K

;193K

Figure 9.72 (a) Partial 300MHz !H NOESY spectrum of [OsH3

(NH=CPhC6H4) (PPri3)2] in CD2C12 at 193K. (b) Variable temperature partial
300 MHz 1H NMR spectrum of [OsH3(NH=CPhC6H4) (PPri3)2] in 6?8-toluene.
(Reproduced with permission from Esteruelas et al. (1998) Organometallics,
17, 4065, copyright (1998) American Chemical Society.)

www.pdfgrip.com
332 Two-dimensional NMR spectroscopy

7.0 6.5 -8.5 -9.0 -9.5 -10.0

6(1H)

X 1
I | I I I I ' ' ' 'I
0 -5 -10
(d) 8(1H)

Figure 9.73 300 MHz *H NMR spectra of [Ru(CO)H2(PHPh2)(PPri3)2] in

C6D6. (a) A partial spectrum with broad band 31P decoupling, (b) A partial
spectrum with continuous wave 31P decoupling centred on the PHPh2 reso-
nance, (c) {A partial spectrum !with continuous wave 31P decoupling centred
on the PPr 3 resonance, (d) A H COSY NMR spectrum. Note the arrowed
cross peaks. (Reproduced with permission from Esteruelas et al. (1998)
Organometallics, 17, 3346, copyright (1998) American Chemical Society.)

www.pdfgrip.com
 Questions 333

9.20.1H NMR spectra of [Ru(CO)H2(PHPh2)(PPri3)2] are shown in

Fig. 9.73. The 31P[1H} NMR spectrum is AX2. Deduce the stereo-
chemistry of the compound.
There are two hydride signals at 8 -8.5 and -9.6 in the 1H{31P}
NMR spectrum (a) Explain why the signal at 8 -8.5 is a triplet
or a double doublet and the signal at 8 -9.6 is a doublet.
Spectra (a) to (c) show the !H{31P}, ^{^PHPhJ, and
iRf^PPr^} spectra. Deduce the appearance of 1VL NMR signals
due to PH, and the two hydrides in the 31P coupled NMR spec-
trum. Estimate the values ofJ(3lPlH) and /(1H1H) for these three
hydrogens. Account for the broadness of the apparent quartet at
8 6.9 in spectrum (b).

www.pdfgrip.com
www.pdfgrip.com
 Magnetic resonance
imaging and
biomedical NMR
10
Magnetic resonance imaging has made immense strides over the last
decade. Initially, it was applied to medicine where it has proved to
give an extremely useful extra diagnostic method to radiographers and
has now become in addition, a medical research tool of increasing
usefulness. If these were its only uses, imaging might not be an appro-
priate subject for a book aimed at chemists but applications in biology
and technology are now well established and a basic understanding at
least of the subject is necessary in today's world for the aspiring NMR
spectroscopist. Note that the word 'nuclear' has been quietly dropped
from the description of the technique. This is because the public has
difficulty in understanding the difference between stable and unstable
isotopes. It also avoids any confusion with Nuclear Medicine, which
discipline does use unstable isotopes. We will describe briefly how
images are obtained, remembering that there now exist a multitude of
RF pulse/magnetic field gradient sequences designed to obtain various
ends, and then look at a number of more chemically oriented appli-
cations. Biomedical NMR looks at the chemistry taking place in living
matter and, apart from straightforward spectroscopy, may use imaging
with spatially resolved spectra or simply place a coil on the surface of
a sample near an organ of interest, and watch what happens when
various constraints are imposed on the system.

10.1 PRODUCING AN IMAGE

The spectra that we have been considering so far contain resolved

structure, which arises from chemical differentiation of the nuclei
observed and which are obtained in an accurately defined, homoge-
neous, fixed magnetic field. If, on the other hand, we were to use a
magnetic field that had a gradient in one direction and a sample
with only one type of nucleus, say, 1H in H2O, then the frequency of
the water resonance would vary across the sample, and the output
would thus contain information about the gradient and the disposition
of the sample within it; in other words, the gradient encodes spatial

www.pdfgrip.com
336 Magnetic resonance imaging and biomedical NMR

information for us. Since the human body contains a large proportion
of water, it became evident that, if one could map its spatial distrib-
ution, then one might be able to investigate the nature of the soft
body tissue. Water is present as about 55% of body weight and the
proportion varies widely in different parts of the body, as does
the water mobility, and so its relaxation rates. In general, in body fluids
T2<T^ We will describe first a basic experiment to indicate the general
principles used in obtaining an image following Fig. 10.1.
The basic idea behind producing an image can be understood as
follows. An object is three-dimensional but the display of its image
has to be two-dimensional. It is necessary therefore to first define a
plane whose two-dimensional image is to be reproduced. This is done
by subjecting the object to be imaged to a field gradient (and so
frequency gradient) in one direction only, and we will take the z direc-
tion, which may be vertical or horizontal in the case of imaging. A 90°
pulse is applied, but with a fairly long duration or with a gaussian
shape, to give a small frequency spread, so that the nuclei are in
resonance and swing into the xy plane over only a short distance
along the magnetic field gradient. For instance, if the field gradient

Magnetic
field
gradient in |z| Directions

Slice
selective_
pulse

Output f(x) f(y, x)

Figure 10.1 Pulse and field gradient timing for obtaining a proton density
image of a slice of a subject. The finishing point of the signal produced for
each volume element, or voxel, under the x gradient depends on the x coor-
dinate of the resonating fluid and this is encoded into the signal produced
under the y gradient as a phase change. The output thus contains informa-
tion of both x and y coordinates of every part of the subject in the slice
defined by the z gradient and RF pulse

www.pdfgrip.com
 Whole body imaging 337

dB0/dz = 5 juiT mm'1 along the main field axis, then the frequency
change for 1H that occurs for a 2 cm displacement along the z axis is
4150 Hz, and a rectangular pulse of the order of 240 JJLS long will define
a slice of the object 2 cm thick. The z field gradient is switched off at
the end of the pulse and, say, gradient x is then switched on. The
water in the 2 cm slice thus produces a signal, the frequency of which
depends upon the x coordinate only so that the rows of spin density
perpendicular to the gradient direction each have different frequency.
The magnetization evolves with time and, after some time tx9 that of
the row at each x coordinate will have a particular phase, which is a
function of tx and x coordinate. This is often called the phase encoding
gradient for this reason. The x gradient is then switched off and
replaced by a y gradient. The output frequencies now depend upon
the y coordinate but with phase determined by the x coordinate. The
output is collected in the time domain t. Outputs are collected for
several values of tx and the data are then subjected to a two-dimen-
sional Fourier transform, to give a map of proton density in the slice
originally defined by the combination of RF pulse and z gradient. The
timing of the experiment is summarized in Fig. 10.1. In practice, the
technique shown has a number of drawbacks, principally because
the relaxation times T2* of body fluids are rather short due to the non-
homogeneous nature of the subject so that the signal has very little
intensity by the end of the collection period and the intensity is also
a function of tx. The example given is however, clear and follows
directly from the techniques for two-dimensional spectra already
discussed.

10.2 WHOLE BODY IMAGING

Many different sequences have now been developed for whole body
imaging which avoid the disadvantages mentioned above and also
permit contrast weighting of the image by utilizing the differences in
7\ or T2 of body tissue. In addition the effects of blood flow and the
unavoidable motion of body tissue have to be compensated for and
this necessitates adding further complexity to the sequences used. We
will describe one basic sequence here which is shown in Fig. 10.2. The
first point to note is that the slice selective z gradient (Gz) is followed
by a period of z gradient of opposite sign. During the short slice selec-
tive period, the spins lose some phase coherence in the xy plane
through the depth of the slice and this reversal of gradient compen-
sates for this, refocusing the spins. Secondly, tx is kept constant and
the changes in phase encoding are brought about by varying the inten-
sity of the phase encoding x gradient (GJ, which is shown as a pulse
cut by horizontal lines. The y gradient (Gy) is next switched on to give
the frequency encoding, though by the end of this time the signal will
be of almost zero intensity. A 180° RF pulse is then applied, in conjunc-
tion with a z gradient so that it is also slice selective and this produces

www.pdfgrip.com
338 Magnetic resonance imaging and biomedical NMR

90° 180°

Echo
RF

Slice selection

Gz

Phase encoding
Gx

Frequency
encoding
Gy

time

Figure 10.2 A typical sequence used for whole body imaging. The vertical
arrow implies that the various phase encoding gradients are applied sequen-
tially. Gz, Gx, Gy, are the magnetic field gradients applied along the three
axes of the system.

an echo which is frequency encoded by a further period of y gradient.

This procedure is repeated for each value of Gx This is called the spin
echo sequence. An alternative approach is to omit the 180° RF pulse
and form the echo by applying a negative y gradient then a positive
y gradient. This is known as the gradient recalled echo sequence.
The output of such sequences is weighted by the spin density in each
voxel of the slice and also by the real T2 of the voxels. The shorter
the sequence, the less the signal intensity lost due to T2 relaxation.
Evidently, by varying the length of the sequence, we can introduce
T2 weighting. The sequences are of quite short duration, insufficient
for full 7\ relaxation to take place so that 7^ also affects the intensi-
ties, and in this case the weighting can be changed by altering the
repetition rate with which spectra are produced. It is equally possible
to start the sequence with a 180° RF pulse to invert all the spins
and take the image after some recovery period, though this adds to
the time needed to take an image, which can be unacceptable to the

www.pdfgrip.com
 Whole body imaging 339

patient. In summary, spin density, 7\ and T2 are all manifested in the

image but one can be chosen to predominate by choosing appropriate
imaging conditions. Typical images of cross sections of a human head
are shown in Fig. 10.3. The advantages of the method are that it is
completely uninvasive and the contrast between, for instance, the
different types of brain tissue is some six times greater than for X-ray
radiography. We shall also see below that it is capable of very subtle
developments.
One such is based on detecting changes in blood flow in localized
parts of an image. If the organism is at rest, then the blood flow will

Figure 10.3 Two sections of a human head showing the location of a brain
tumour (Images supplied by Bruker Medizintechnik.)

www.pdfgrip.com
340 Magnetic resonance imaging and biomedical NMR

be steady and the ratio between the concentrations of oxyhaemoglobin

and deoxyhaemoglobin will have a certain value. Now, oxyhaemo-
globin is diamagnetic and the deoxy form is paramagnetic, so that
the blood will have a particular value of magnetic susceptibility in the
resting state. This will distort the applied magnetic field and so
contribute to relaxation via the T2* effect. If the organism is
stimulated, the blood flow will increase and this will increase the
proportion of oxyhaemoglobin present and so change the magnetic
susceptibility of the blood. T2 will change and if the imaging sequence
is chosen so as to be sensitive to T2* then the contrast of the image
will change in the affected region. The difference between two images
taken before and after stimulation, will show clearly where there were
changes in the blood flow. This, of course, means that the refocusing
pulse cannot be used in the imaging sequence, otherwise inhomo-
geneity effects would be refocused. An application of this method is
shown in Fig. 10.4 where two difference images of a human brain are
shown. In each case, images were taken while the subject had periods
of repose when nothing was presented to his vision followed by images
taken in periods when a visual stimulus was given and the difference
was obtained to highlight the changes. If he was asked to look at words
then the major part of the activity took place in the left hemisphere

Observe words Observe faces

Figure 10.4 Difference images of thick sections of a human brain showing the different parts of the brain
brought into action by the visual stimuli of words (left) and faces (right). (Example supplied by M.
Raybaudi, INSERM, Grenoble.)

www.pdfgrip.com
 Diffusion and flow 341

of the brain, whereas the opposite was the case when faces were in
his field of view. The centres of activity are quite dispersed and some
of the centres are involved in more than one activity.

10.3 DIFFUSION AND FLOW

Diffusion of molecular species has been studied since the mid-1950s

using a method based on the Carr-Purcell 90° - T - 180° - T - echo
pulse sequence. Two short periods of magnetic field gradient are added
to the sequence as shown in Fig. 10.5 and modify the behaviour of the
output as follows. The first 90° pulse turns the magnetization into
the xy plane and this then loses phase by the T2 process and due to
inhomogeneity effects. The first field gradient pulse causes further
dephasing which depends upon the position of each spin in the sample.
The 180° RF pulse inverts the spins and refocuses the inhomogeneity
effects but not the effect of the first gradient pulse. This is cancelled
by the second field gradient pulse, which after inversion of the spins
acts as if it were of opposite sign to the first field gradient pulse. In
the absence of translational motion of the spins, we get the usual echo
as if the field gradient pulses had not existed. If the spins move
however, then a spin which is dephased by an angle $ by the first field
gradient pulse will be refocused by another angle, say 6, by the second
field gradient pulse. Since the two angles are not equal, the magneti-
zation will remain dephased and the intensity of the echo will be
reduced, so permitting the estimation of the average spatial displace-
ment of spins between the two magnetic field gradient pulses. In this
way, self-diffusion can be estimated (a random process) or flow
can be measured in the direction of the magnetic field gradient of
the pulses. In the latter case, the result will depend also upon whether
the flow is plug flow, turbulent flow or non-turbulent flow, with flow

90° 180°

g g

RF Echo

Figure 10.5 The basic pulse sequence used to measure the spatial displace-
ment of spins in short intervals of time. The two RF pulses of 90° and 180°
are the usual Carr-Purcell sequence and two short pulses of magnetic field
gradient (g) are added equally displaced from the 180° RF pulse. This
sequence is followed by an echo whose intensity is reduced if there is displace-
ment of spins between the two pulses g.

www.pdfgrip.com
342 Magnetic resonance imaging and biomedical NMR

faster in the centre of the conduit than at its walls. It will be evident
by referring to Fig. 10.2 that this pulse sequence can easily be combined
with an imaging sequence and so measure flow or diffusion through
an image. Other methods are also possible and are often called time
of flight measurements. Thus if the repetition time between pulse
sequences is short then stationary spins are saturated and flowing
spins are not and so appear with different intensity. Alternatively, if
all the spins in a slice are inverted by a 180° pulse and then a waiting
period given such that all the stationary spins are at null intensity, new
spins flow into the slice wherever there is flow and these appear in
the image.
Diffusion weighted images of the brain have been found to be invalu-
able in the early diagnosis of stroke damage since the diffusion of
water is slower in ischemic tissue and, indeed, imaging is the only tech-
nique capable of doing this. Other medical applications include the
measurement of blood flow in phase contrast magnetic resonance
angiography. If flow occurs in a linear gradient there is a phase change
induced which is identical for every position along the gradient and
allows the flowing fluid to be differentiated from stationary fluid. The
magnitude of the phase change depends upon the timing of the pulse
sequence, the value of the gradient, and the velocity, which can thus
be calculated from the data.

10.4 CHEMICAL SHIFT IMAGING

10.4.1 Water suppression

The techniques discussed so far depend upon the presence of water
to provide the signal to generate the image. The water has a strong
signal and, with the exception of the signal from fat, obscures the much
weaker resonances from metabolites present. In order to carry out
spectroscopy of biological samples then it is necessary to eliminate in
some way the water resonance, and this holds for one-dimensional
spectroscopy, two-dimensional spectroscopy and spectroscopy associ-
ated with imaging. One technique is to use a weak continuous wave
at the water frequency (long and therefore very selective) during the
relaxation delay so as to saturate the water resonance, then produce
a spectrum in the normal way. This is useful but not fully effective
and resonances near to the water may well be obscured by the remnant
water signal. A variety of schemes have been developed to improve
water suppression and all are based on the use of magnetic field gradi-
ents. One such is WATERGATE (WATER suppression by GrAdient
Tailored Excitation). The sequence is shown in Fig. 10.6. The spec-
trometer frequency is first centred on the water resonance. Transverse
magnetization is created by a non-selective 90°^ pulse then a gradient
is applied which defocuses all the spins. This is followed by a specially
designed pulse which consists of two long, selective 90% pulses

www.pdfgrip.com
 Chemical shift imaging 343

90° (X) 90° (-X) 180°(X) 90°(-X)

RF FID

Figure 10.6 The WATERGATE pulse sequence. The two tall RF pulses are
non-selective and the two broad, short ones are selective and at the water
frequency. The two magnetic field gradient pulses G refocus only the non-
water resonances.

separated by a non-selective 180°,, pulse. The water magnetization is

left unchanged by this combination but the rest of the spins are flipped
180°. A second field gradient pulse refocuses the spins that have been
subjected to the 180° flip, but not the water protons. This gradient
pair technique is effective because it compensates for imperfections in
the Bl pulses.
If large biomolecules are being examined, then we can take advantage
of the differences in diffusion rate between them and water, and diffu-
sion weight the spectra using the principles outlined in the previous
section. If there is a 20-fold difference in diffusion constants then it is
possible to design an experiment such that the signal required is atten-
uated by only 70% whereas the water intensity is reduced 1000-fold.
In the following section, we will shortly encounter a third water
suppression method called CHESS which is used with chemical shift
imaging (Fig. 10.8).

10.4.2 Imaging by spectra

By working at high enough magnetic field, it is possible to produce a
!
H spectrum from each voxel of the sample. The lines are broad but
the significant features of the spectrum can be observed provided
the water resonance is attenuated. The basic pulse sequences used
to produce such spatially localized spectra are called STEAM
(STimulated Echo Acquisition Mode) or PRESS ( Point RESolved
Spectroscopy; a spin echo sequence). In STEAM, a series of three 90°
pulses are applied, each in the presence of a slice selective gradient
along one of the three axes so that a voxel is defined at the intersec-
tion of the three slices and a spectrum is obtained from this voxel.

www.pdfgrip.com
344 Magnetic resonance imaging and biomedical NMR

One can do even better than this, however, and produce the spectra
of all voxels in a slice using an imaging scan. The slice selective 90° -
T - 180° sequence is used similar to that described above (Fig. 10.2)
but gradient pulses are applied along the other two axes, say x and y,
just before the 180° RF pulse, with many values of Gx and many values
of Gy for each value of Gx. Fourier transformation of this data stack
gives a spectrum for each voxel of the slice. The sequence has to be
preceded by some sort of water suppression sequence such as that
included in Fig. 10.8. Figure 10.7 shows such an image. It comprises
a matrix of 16x11 spectra at relatively low resolution but within
which it is possible to discern major changes over the space occupied
by the sample. The sample is the brain of an anaesthetized rat in
which a tumor had been implanted. Each spectrum corresponds to a
voxel dimension of 1 x 1 mm. The resonances which are visible are,
starting from low frequency, lactate in the tumour, 7V-acetyl aspartate,

Figure 10.7 1H spectra of a slice of a rat brain into which a tumour had
been implanted. Each spectrum represents a pixel of 1 x 1 mm. The tumour,
which is rich in choline and lactate, is located at the centre left of the figure.
Note that the oval shaped brain produces no responses in the lower right
and left corners of the matrix. (Figure provided by A. Ziegler, INSERM,
Grenoble. To be published in Magn. Res. in Medicine)

www.pdfgrip.com
 Chemical shift imaging 345

Figure 10.8 X H NMR COSY imaging pulse sequence used to obtain the two-
dimensional spectra of Fig. 10.9. First the water is subjected to three selec-
tive pulses followed by pulses of gradient which dephase its transverse
magnetization, so that the water signal is much reduced; this is called the
CHESS sequence. The RF sequence that follows is a simple COSY sequence
but the magnetic field gradient pulses which are added produce an image.
The first 90° selective RF pulse is thus converted into a slice selective pulse,
and the second is accompanied by gradients in the three axes, several values
of Gx being used for each value of Gz. Note that the gradient pulses can
overlap.

glutamate, glutamine, etc., creatine, and choline and related species in

the tumour. The normal tissue contains relatively little choline or
lactate. The tumour is at the centre left of the figure, and is quite clear
in the spectra presented though, naturally, it shows up better in a
colour representation based on the intensity of each resonance in
each spectrum.
Resolution in such spectra is limited but can be improved by using
COSY imaging in which the COSY spectrum is obtained for each voxel
and the various compounds present are characterized by the cross
peaks on the COSY spectrum of each voxel. A typical pulse sequence
is shown in Fig. 10.8 and some representative spectra are shown in
Fig. 10.9. The initial CHESS sequence (CHEmical Shift Selection
imaging) is used to minimize the water resonance and consists of three
selective 90° pulses, each followed by a gradient pulse (in the x and
z directions in this case, and of varying amplitudes to avoid echo
formation). Each RF pulse produces transverse water proton magne-
tization which is then destroyed by the gradient pulses. Two slice selec-
tive 90° pulses are applied next, which produce the COSY spectra as
described in Chapter 9 and which are localized in space by the later
gradient pulses. Note that the 90° RF pulse/gradient pulse slice selec-
tive pair is bracketed by two gradient pulses which compensate for
imperfections in the RF pulse. The result of this acquisition is that
each voxel produces a COSY spectrum in which there is the diagonal,
a still non-negligible water resonance (the vertical response) and a
number of cross peaks which arise through the ^^H correlations in

www.pdfgrip.com
 346 Magnetic resonance imaging and biomedical NMR

Figure 10.9 Correlation peak imaging of rat brain in which a tumour had been implanted, (a) Normal
brain tissue, (b) Tumour tissue. The cross peaks correspond to (1) Af-acetyl aspartate, (2) glutamine/gluta-
mate, (3) glucose, (4) aspartate, (5) taurine, (6) inositol, perhaps choline, (7) lactate, (8) alanine and (9)
hypotaurine. The nominal voxel volume for each spectrum is 23 fjil and slice thickness 5 mm. (Figure
provided by A. Ziegler, INSERM, Grenoble.)

each molecule and which are far better resolved than in the simple
spectrum. It is thus possible to detect some fourteen compounds in
the rat brain, spatially localized (Fig. 10.10). The change in the metabo-
lites present in tumour and normal tissue is very marked. Note that
the strong TV-methyl resonance of choline is not seen in these COSY
plots since there is no coupling. The choline resonance remains in the
diagonal.

10.5 BIOLOGICAL USES OF IMAGING - IMAGING

MICROSCOPY

Imaging of small biological samples can be carried out in smaller appa-

ratus and the reduction in scale also permits an increase in resolution
or decrease in voxel size, typically to 25 x 25 juim with a slice thickness
of 300 to 2000 jjim. Otherwise, the techniques are similar to those
already described. The technique has been named NMR microscopy.
The resolution is less than available with optical microscopy, but
different things are imaged by the two techniques and, more impor-
tantly, NMR is non-invasive, the sample needs little preparation and
studies can be made in vivo. In Fig. 10.11, we show the image of a

www.pdfgrip.com
 Biological uses of imaging 347

a)NAA b) Glu / Gin c)Glc

Figure 10.10 Metabolic images for a whole rat brain corresponding to the cross peaks of Fig. 10.9. The
contours of brain and tumour are superimposed on the images, (a) shows the distribution of Af-acetyl
aspartate, (b) that of glutamine/glutamate, (c) that of glucose, (d) that of lactate, (e) that of hypotaurine,
(f) that of alanine, (g) that of choline/inositol, (h) that of aspartate and (i) that of phosphoethanolamine.
(Figure provided by A. Ziegler, INSERM, Grenoble. To be published in Magn. Res. in Medicine.)

cross-section of the hypocotyl stem of a castor bean seedling, essen-

tially growing in the probe, in which the structures are well visible.
COSY images permitted the localization of metabolites in various parts
of the cross-section of the stem and using other techniques, water flow
has also been detected moving both up and down the stem. In the
latter case, the method used had to be able to differentiate between
flow in both directions in a single voxel!

www.pdfgrip.com
348 Magnetic resonance imaging and biomedical NMR

Figure 10.11 Image of a cross-section of the hypocotyl of a castor bean seedling. The eight vascular bundles
(b) are connected by the meristem ring. The cellular structures of pith parenchyma (a) and cortex
parenchyma (c) are visible. (From Ziegler et al (1996) /. Magn. Reson., 112B, 141, with permission.)

10.6 INDUSTRIAL USES OF IMAGING TECHNIQUES

The measurement of the distribution of fluid throughout a sample is

of interest in a number of other fields. Imaging has, for instance, been
used to detect the position of material on a chromatography column,
which otherwise was difficult to detect. It has a number of techno-
logical uses, such as the study of water in food materials or the cooking
of foods, where the non-destructive nature of the technique is useful.
The method is also coming to be used in chemical engineering where

www.pdfgrip.com
 Industrial uses of imaging techniques 349

such things as the distribution of fluid flow in packed bed columns can
be visualized or the mixing of fluids of different NMR relaxation times
which have contrast in the images.
Some specific examples are the study of packed bed reactors. The
fluids present may be gas and liquid when it is easy to determine how
the two are distributed because of the different spin densities, and the
way the distribution changes with flow rate is easy to follow. Such
reactors are most efficient in the trickle flow regime when the liquid
falls in rivulets from one catalyst particle to the next. The rates of
reaction also depend upon the shape and size of the catalyst particles
and the way they pack in the reactor, and these factors can now be
studied directly. Packed reactors through which two-phase liquid
systems are being passed can also be examined and the way the phases
flow and co-dissolve can be observed. The distribution of flow rates
in the interstices of the reactor can be measured using techniques
already discussed. An example is given in Fig. 10.12.
The deactivation of catalyst beads can also be followed by imaging.
A catalyst bead, whose pores may have become partially blocked by

8.3 mm s~1

0.0 mm s~1

Figure 10.12 Flow rates of a water/hydrocarbon mixture through the beads (black) in a packed bed reactor.
The rates are calibrated by the shade scale at the left. (From Gladden (1998) Chem. in Britain, 34, No.
3, 35, with permission.)

www.pdfgrip.com
350 Magnetic resonance imaging and biomedical NMR

carbon deposits is soaked in a solvent and then the cross-section exam-

ined by NMR microscopy and the points where there is a solvent signal
show the areas where the pores are still open. The distribution of the
carbon in a partially deactivated pellet is not as expected from current
theory and this observation should lead to new developments in the
theory of deactivation.

10.7 BIOMEDICAL NMR

In addition to the mobile protons in a living organism, there are mole-

cules containing 13C and 31P nuclei that can also be used for study of
how the chemistry of life operates. Early, pre-imaging experiments
with whole, anaesthetized, small animals in large-bore magnets were
promising, but signals were obtained from all parts of the animal. It
quickly became clear that it would be of most use if signals could be
localized to a specific known organ. One approach is to create a
specially contoured magnetic field that has the correct value for reso-
nance over only a small, controllable volume. Alternatively, a coil may
be placed on the body surface. This coil produces a radiofrequency
field that penetrates below the surface by approximately its own radius
and so can stimulate responses from nuclei inside the body. This is
called 'topical NMR'. It is possible, for instance, to follow the 31P
signals of nucleotide phosphates and inorganic phosphate in muscle
and monitor how these respond to different conditions imposed on
the muscle or to various disease conditions. Figure 10.13 demonstrates
the effect on the 31P spectrum of a subject's arm of applying a tourni-
quet. The normal spectrum shows peaks for the three phosphorus
atoms in adenosine triphosphate ATP (I, II and III), for phosphocre-
atine (IV) and for inorganic phosphate (V). Application of the tourni-
quet leads to oxygen starvation and breakdown of organic phosphate
to inorganic phosphate. The effects of exercise on muscle chemistry,
the spectra of skin tumours or the brain can all be obtained in this
way. It is, for instance, possible to follow changes during treatment of
a baby suffering from birth asphyxia using such techniques.
13
C spectroscopy offers wider dispersion of resonances in work with
non-homogeneous biological samples and 13C spectra can be obtained
routinely with surface coils using modern equipment. Proton irradia-
tion is used and is applied using a second surface coil. Power dissipa-
tion has to be limited (8Wkg -1 is the recommended maximum
dissipation) and the WALTZ technique is ideal for this reason. Tracer
studies using 13C enriched molecules are very informative. For instance,
it is possible to follow the liver metabolism in anaesthetized rats. The
rat is perfused with either [1-13C] glucose or [1,2-13C ] glucose and
the evolution of the spectra observed as a function of time. In fact,
difference spectra are used in which the spectrum obtained before the
start of perfusion is subtracted from those obtained under perfusion.
A steady increase is seen with [1-13C] glucose perfusion over about

www.pdfgrip.com
 Biomedical NMR 351

(a)

(b)

ppm 10 0 -10 -20

Figure 10.13 The 31P topical NMR spectra obtained at 32.5 MHz using a
single-turn coil placed on the surface of a subject's arm, in a contoured
magnetic field, (a) The normal spectrum (64 2-s scans). Peak assignments are
given in the text, (b) The same subject 50 min after the application of a tourni-
quet to the upper arm. (Reproduced with permission from Gordon (1981)
Eur. Spectrosc. News, 38, 21.)

130 minutes of the concentrations of the metabolites glycogen, a and

P forms of glucose, glycerol and after about 87 minutes, resonances
appear due to glutamate/glutamine. If [1,2-13C] glucose is used for infu-
sion, we have the situation that metabolites may be formed with the
13
C-13C bond intact and so showing 13C-13C coupling. On the other
hand, if this bond is ruptured by the metabolic pathway then this
coupling will disappear. Two spectra are shown in Fig. 10.14 obtained
with the two differently labelled glucose molecules. The a and (3 forms
of glucose are evident (resonances 3,9 and 2,8 respectively) as is a
doublet resonance of glycerol which confirms that the pair of 13C atoms
from the glucose have been incorporated into the glycerol via the
glycolytic intermediate [2,3-13C2] dihydroxyacetone phosphate.
Our final example returns to imaging with the measurement of the
diffusion of metabolites in vivo. This is a combination of the voxel
selection sequence PRESS with the addition of diffusion weighting
gradient pairs as shown in Fig. 10.15. PRESS is designed to select one
voxel in the sample and produce a one-dimensional spectrum. The
RF pulses are all selective, the first being 90° and the second two 180°.

www.pdfgrip.com
352 Magnetic resonance imaging and biomedical NMR

[1,2-13C2]glycerolC1+C3

(B)

[1-13C]glycerolC1+C3

(A)
Vw^^^^X^' V*vyR\*vX^rv^*w~»*

ppm 100 80 60 40

Figure 10.14 In vivo 13difference 13C spectra of13rat liver after 125 minutes of
infusion of either [1- C] glucose (A) or [1,2- C] glucose (B). The doublets
indicate C-C bonds which have been metabolized without bond fission.
Resonances 1, 4 and 6 arise from glycogen, the others are named
in the text. (From Kiistermann et al (1996) Bruker Report 143, 33, with
permission.)

A double echo is thus formed and the second is collected. Each RF

pulse is accompanied by a gradient pulse in one of the three orthog-
onal axes so that we select three orthogonal slices and the voxel where
they intersect is the only part of the sample to give a signal. In addi-
tion, each 180° pulse is bracketed by a gradient pair which encodes
the diffusion. In order to suppress the water resonance, the sequence
shown can be preceded by a CHESS sequence. A set of results
are shown in Fig. 10.16 for a 45 juil volume localized in the brain of
an anaethetized rat. On the left are the spectra with the CHESS pulses
omitted which give essentially only the water resonance. On the
right, the water is suppressed and the metabolites are now visible.
The spectra are taken with the diffusion gradients G increasing up
the series. It will be evident that water diffuses the fastest and that

www.pdfgrip.com
 Biomedical NMR 353

90° 180° 180°

Echo
RF-

Gradient slice 1

Gradient slice 2

Gradient slice 3

Figure 10.15 The PRESS sequence for obtaining the spectrum of a single
voxel, modified to diffusion weight the spectra so obtained. The PRESS pulses
are open, the diffusion weighting pulses are marked with a diagonal. D is the
interval between pairs of these pulses and d is the pulse duration.

the metabolites diffuse appreciably more slowly. Diffusion in the pres-

ence of living cells is a complex process and may be anisotropic due
to the structures present. Evidently, the type of experiment described,
which is in its infancy, can give access to much detailed information
on the behaviour of living systems. The sensitivity is incredible when
it is recalled that the molecules investigated move only micrometres
during the experiment.

www.pdfgrip.com
354 Magnetic resonance imaging and biomedical NMR

6.5 6 5.5 5 4.5 4 3.5 3 5.5 5 4.5 4 3.5 3 2.5 2

Chemical shift (ppm) Chemical shift (ppm)

Figure 10.16 In vivo diffusion weighted !H spectra of a voxel of a rat brain

showing water (left) and (right) the water suppressed by a CHESS sequence
to show the metabolites. The spectral conditions were: TE = 136 ms; repe-
tition rate = 3 s; d = 8 ms; D = 26.2 ms. The metabolites are: tCre, creatine
and phosphocreatine; Glx, glutamine/glutamate; Ins, myoinositol; Cho, the
W-methyl protons of choline; NAA, the methyl protons of Af-acetyl aspartate.
(From K. Nicolay et al (1995) NMR Biomed., 8, 365, with permission, copy-
right (1995) John Wiley & Sons Ltd.)

www.pdfgrip.com
 High-resolution
solid-state NMR 11
We have already mentioned in Chapter 4 that, in the solid state,
the relaxation time 7\ is long due to the lack of modulation of the
dipole-dipole interaction and T2 is short due to mutual spin flips occur-
ring between pairs of spins. In a static solid, each nucleus produces a
rotating magnetic field as it precesses in the applied magnetic field,
and this can cause direct exchange of energy between nuclei. The life-
times of the spin states are thus reduced and so T2. In addition, each
spin has a static field component that influences the Larmor frequen-
cies of its neighbours. An individual nucleus will experience the fields
of several neighbours, but their spin directions will vary randomly,
so that there will be a range of frequencies that will add to the line
broadening due to the rapid rate of relaxation. Finally, particularly for
the heavier nuclei, including 13C, there will exist a chemical shift
anisotropy, which will also contribute to the broadening, assuming that
the sample is a powder or a glass and not a single crystal, because the
chemical shift varies with orientation relative to the BQ direction. Thus
solid materials, particularly if they contain nuclei with high magnetic
moments such as *H or 19F, will have broad, structureless resonances,
which will not permit the type of investigation that we have shown
can be carried out in the liquid phase. This state of affairs has proved
a challenge to the NMR community, who have over the last two
and a half decades found means to render ineffective the apparent
physical restraints to the spectroscopic examination of solids at high
resolution.
Before discussing this work in detail, however, it is necessary to
mention two useful aspects of the broad lines. In the first place, because
the broadening is determined by the dipole-dipole interactions, it is
sensitive to the distance separating interacting spins. The spectrum of
a solid can thus be used to obtain internuclear distances, which in the
case of the proton are difficult to obtain by other means. The mole-
cules studied must be static in the solid state and must be sufficiently
simple, preferably an isolated spin pair, that the resonance width can
be interpreted in terms of a single, principal distance. If the molecules
reorient in the solid in some way, then this modulates the internuclear
interaction and its magnitude is reduced, and this is the second prop-
erty that proves useful. If the linewidth of a solid material is measured
as a function of temperature from very low temperatures, it is often

www.pdfgrip.com
 356 High-resolution solid-state NMR

found that there are quite rapid changes in linewidth at certain tran-
sition temperatures. These mark, if the temperature is being gradually
increased, the onset of motion within the solid lattice. Figure 11.1
shows an example for the solid complex adduct of boron trifluoride
with trimethylamine, Me3N—>BF3. The !H resonance linewidth below
80K is 85 kHz (the old unit of 1 gauss is equivalent to 4250 Hz in this
case). Heating from 68 to 103K reduces the linewidth to about 21 kHz,
and this corresponds to the onset of rotation of the methyl groups
around the C-N bonds. Further narrowing to 13 kHz occurs on raising
the temperature to 150K owing to the onset of rotation of the whole
NMe3 moiety around the B^N bond. Finally, just below 400K, the
line narrows to a few hundred hertz as the whole molecule starts to
rotate and diffuse isotropically within the still solid crystal. The 19F
resonance can also be examined and is found to be broad only below
77K and the BF3 rotates around the B«-N axis at all higher temper-
atures. Of course, when the sample melts, the linewidth falls to a frac-
tion of a Hertz.
Such studies using broad lines are, however, of relatively limited
application, and solid-state NMR formed only a small part of the total

80 200 400
Temperature (K)

Figure 11.1 The proton resonance linewidth of the methylamine protons in the solid adduct (CH3)3N —> BF3
as a function of temperature. Note that 1 gauss = 10"4 tesla = 4250 Hz. (After Dunnell (1969) Trans.
Faraday Soc., 65, 1153, with permission.)

www.pdfgrip.com
 Magic angle spinning 357

NMR work that was undertaken. The situation has now changed
dramatically with the application of the modern techniques to be
described below. Both I = 1/2 and quadrupolar nuclei are studied,
though the treatment of these two classes of nuclei is rather different,
and we will have to discuss them separately. One technique is common
to both, one which has revolutionized solid-state NMR more than any
other, and we will describe this first.

11.1 MAGIC ANGLE SPINNING

The magnetic field produced by a nucleus with magnetic moment jx

at a second nucleus a distance r away will, in general, have a compo-
nent in the z or BQ direction, which influences the frequency of the
second nucleus and also couples the two spins. The z component Bz
is given by

Bz = - L (3cos2e - 1)

where K is a constant and 9 is the angle between the direction of J?0

and the line joining the two nuclei. At one particular angle, shown in
Fig. 11.2, Bz is zero. This is the angle for which 3cos26 - 1 is zero or
0 = 54°44/. It is the angle at which all dipolar interactions disappear.
In a real sample, of course, which typically will be a powder, the inter-
nuclear vectors take all possible angles and the trick is to make them
behave as if all had 0 = 54°44/. This is done by mounting the sample
in the rotor of a small air-driven turbine whose axis is inclined at an
angle of 54°44/ to the magnetic field direction. Means are provided to
adjust the angle so as to obtain the optimum results. The turbine is
rotated at high speed, and this gives all the internuclear vectors an
average orientation at the rotor angle, which produces dramatic
changes in linewidth. This is illustrated in Fig. 11.3. The angle has to
be adjusted quite finely to obtain the best results and is accordingly
called the 'magic angle', and the technique is called 'magic angle spin-
ning', usually abbreviated to MAS. The chemical shift anisotropy also
follows a 3cos20 - 1 law, and broadening due to this cause is removed
as well by MAS. Quadrupole broadening for nuclei with / > 1/2 is also
reduced by MAS, but the situation is more complex for the
quadrupolar nuclei, as we shall see. The spinning speed is important,
since if the static linewidth of the resonance to be studied is F Hz,
then the spinning speed must be greater than this if all the broadening
interactions are to be nullified. Since the basic physical strengths of
materials are limited, there is a limit to the centrifugal forces that they
will withstand, and so a limit to the speed of spinning. Currently, the
normal limiting speed is around 35 kHz for small ceramic rotors. It is
also useful to know the speed at which a rotor is spinning, and these
are provided with marks on the rotor cap that can be detected by a
suitable frequency counter.

www.pdfgrip.com
358 High-resolution solid-state NMR

IX

Figure 11.2 A line of the magnetic field originating from a magnetic dipole has zero z component at a
point situated on a line originating at the centre of the dipole and at an angle of 54°44' to the direction
of the dipole. (From Bruker CXP Application Notes, with permission.)

Figure 11.3 Showing how a solid sample is mounted for magic angle spinning and how this gives the inter-
nuclear vectors an average orientation at the spinning angle. (From Bruker CXP Application Notes, with
permission.)

www.pdfgrip.com
 Spin-1/2 nuclei with low magnetogyric ratios 359

11.2 SPIN-1/2 NUCLEI WITH LOW MAGNETOGYRIC RATIOS

Common examples of compounds within this class are 13C in organic

compounds or 31P in inorganic or organic phosphates. If protons are
present, they, of course, cause strong dipolar broadening and their
influence has to be removed. This can be achieved by high-power
double irradiation, high power being required because the proton
resonance will be of very large width. The 13C spectra of adamantane
(Fig. 11.4) show the improvements in resolution that can be obtained.
The normal solid-state spectrum has a linewidth of some 5000 Hz, and
the two types of carbon present in the molecule do not have resolved
resonances. MAS alone reduces the linewidth to 200 Hz and enables
the two types of 13C to be distinguished (Fig. 11.4(a)). High-power
decoupling of the protons alone also reduces the linewidth to 450 Hz
in the case illustrated (Fig. 11.4(b)). Combination of the two tech-
niques, however, reduces the linewidth to 2 Hz, which is little worse
than in liquid-state samples (Fig. 11.4(c)). Obtaining solid-state spectra

(a)

(b) (c)

Figure 11.4 The 13C spectra of solid adamantane (formula inset), (a) With MAS. (b) With high-power
proton double irradiation, (c) With both MAS and double irradiation. (From Bruker CXP Application
Notes, with permission.)

www.pdfgrip.com
360 High-resolution solid-state NMR

in this way has certain drawbacks. The rather long 13C Tl values mean
that pulse rates have to be sufficiently slow so as not to saturate the
resonances, and the natural insensitivity of the nucleus means that long
accumulation times are needed. These disadvantages can be circum-
vented by a modification of the double-resonance technique, which
permits the exchange of polarization between the !H and 13C spins,
and which is called cross-polarization (CP); when combined with MAS,
the whole is abbreviated to CP-MAS.
In order that energy exchange shall be possible between the two
nuclear species, we must introduce components of motion with the
same frequency for each. This is done as follows. Referring to
Fig. 11.5, we first prepare the protons for the cross-polarization by
applying a short 90° pulse, which swings the proton magnetization into
the xy plane. We will call this field Bm. We then change the phase of
B1H by 90° (and perhaps reduce its intensity also), so that the BIH
vector becomes parallel with the magnetization in the xy plane, which
therefore remains in the xy plane and stays there locked to Bm. This
second pulse is known as a spin locking pulse. The magnetization
precesses in the xy plane at the Larmor frequency, and can be thought
of as also precessing around BIH at a frequency of yHBm/2TT Hz,
behaving as if it was very strongly polarized in the weak BIH. Reference
to equation (1.3) will show that such a polarization, correct under
normal circumstances for Z?0, can only be attained if the temperature
is very low for a weak field BIH, and we can consider that the spin
locking has cooled the spins, which can now act as an energy sink.
Energy transfer is obtained by applying a long pulse at the 13C
frequency, U1C, which has an amplitude such that the 13C nuclei precess
around it at a frequency of ycBlc!2^ Hz which is equal to yHBlH/2^
Hz. This is known as the Hartmann Hahn matching condition. The
identical frequency components allow energy transfer, which follows
an exponential curve, and when this has reached a maximum the BIC
is cut off and is followed by a 13C FID of enhanced intensity. The i?1H
field remains on during this time and provides the decoupling of the
protons from the carbon nuclei. The experiment can be repeated when
the proton spins have reached equilibrium again and so the effective
relaxation time is the shorter one of the proton system. The cross-
polarization increases the 13C population difference by the factor
'Yi/Yc and so produces a useful improvement in signal strength. An
example of a 13C CP-MAS spectrum is shown in Fig. 11.6 for a more
complex molecule, the steroid deoxycholic acid, together with a partial
assignment of the resonances. Deoxycholic acid forms inclusion
compounds, and, when the guest molecule is ferrocene, [Fe(T]5-C5H5)2],
the methyl singlets due to CIS, C19 and C21 become doublets due to
differentiation of the deoxycholic acid molecules in the solid lattice.
The cross-polarization technique is much used with low--y nuclei where
the compound studied contains hydrogen; 13C, 15N, 29Si, 31P and 113Cd
are some examples of / = 1/2 nuclei studied. Quadrupolar nuclei can
also benefit from the CP technique.

www.pdfgrip.com
 Spin-1/2 nuclei with low magnetogyric ratios 361

If the spinning speed is appreciably lower than the width of the static
resonance of the compound studied, then sidebands are produced sepa-
rated by the spinning speed. Figure 11.7 shows the solid-state 31P
spectra of aminomethanephosphonic acid, H2NCH2PO3tl2, which has
a zwitterionic structure. All the three spectra shown benefited from

Irradiation 90° pulse

at 1 H
Frequency

Spin
Decoupling
lock

Irradiation
at 13 C
Frequency

1H

Figure 11.5 The timing of the various events in a cross-polarization experi-

ment. A 90° pulse at the proton frequency is followed by a long spin locking
pulse changed in phase from the initial pulse by 90°, which prevents the normal
proton
13
relaxation processes. Energy is caused to flow from the assembly of
C nuclei to the cooled proton nuclei by applying a long pulse at the 13C
frequency, which introduces a precession frequency equal to that already
established for the protons. When equilibrium is reached, this field is switched
off and is followed by a 13C FID of enhanced intensity. Note that the contact
time (about 2 ms) is much shorter than the observation time (about 50 ms).
(From Oldfield et al. (1984) /. Magn. Reson. Chem., 60, 467.)

www.pdfgrip.com
362 High-resolution solid-state NMR

C10
C13

Molecular structure of deoxycholic acid, showing the

atom numbering system used
C24

180 140 100 60 20 8(ppm)

Figure 11.6 The 13C CP-MAS spectrum of deoxycholic acid. The 13C frequency was 50.32 MHz and that
for *H was 200 MHz. Some 350 mg of sample was packed into the MAS rotor and some 800 transients
were acquired. The contact time was 1 ms and recycle delay 3.5 s. (From Heys and Dobson (1990) Magn.
Reson. Chem., 28, S37-46, copyright (1990) John Wiley and Sons Ltd, reprinted with permission.)

CP, but the uppermost one (a) was from a static sample and shows
the shielding anisotropy of the phosphorus nucleus, which does not
have axial symmetry and so has three principal components crn, cr22
and a33 (see Chapter 2). MAS at 813 Hz (b) produces a group of
narrow lines separated by the spinning frequency. The number of lines
is reduced on increasing the spinning speed to 2950 Hz (c), and it is
evident that only one does not move, so that this (arrowed) is the
true resonance with the isotropic chemical shift and the remainder
are spinning sidebands. It is particularly important to note that at
low spinning speeds, the envelope enclosing the sidebands approxi-
mates the static lineshape and so retains the form of the shielding or
chemical shift anisotropy. The 31P spectrum of Af,Af-dimethylamino-
methanediphosphonic acid, Me2NHC(PO3H2)2H, is also shown for a
spinning speed of 1740 Hz (d). In this case, there are two resonances
that are not sidebands, since the two phosphorus atoms in a single
molecule are not related by symmetry due to the crystal structure. In
solution, of course, only a singlet 31P resonance is produced by this
compound.

www.pdfgrip.com
 Spin-1/2 nuclei with low magnetogyric ratios 363

(a)

(b)-

(c)

(d) 100 50 0 -50

6 (ppm)

Figure 11.7 The 31P spectra of aminomethanephosphonic acid, (a) Static

sample showing the shielding powder pattern. The !H-31P dipole interaction
is eliminated by double resonance, (b), (c) With MAS at 813 Hz and 2950 Hz.
The arrow is at 18.3 ppm and the bar represents 2 kHz. (d) MAS spectrum
of Af,Af-dimethylaminomethanediphosphonic acid showing the existence of two
crystallographically differentiated sites. The chemical shift scale applies only
to this spectrum. The arrows show the centre band lines with the isotropic
chemical shifts. The other lines are spinning sidebands. (From Harris et al.
(1989) Magn. Reson. Chem., 27, 470, with permission, copyright (1989) John
Wiley and Sons Ltd, reprinted with permission.)

Such high-resolution spectra allow access to parameters such as

chemical shift anisotropy and permit comparison of molecular
structure in solution and crystal. In addition, they permit insoluble
substances to be studied at reasonably high resolution, and investiga-
tions are being carried out into the structures of materials such as plas-
tics or coals. The latter have been notoriously difficult to study without
degrading the coal structure, and this is now possible using NMR. An

www.pdfgrip.com
364 High-resolution solid-state NMR

example is shown in Fig. 11.8 for both a plastic and a coal and, while
the resolution for the coal is not exceptional, it has to be remembered
that it is a most complex mixture of structures and that it is very useful
to be able to distinguish aromatic and aliphatic resonances and perhaps
to be able to do this quantitatively.
It will be no surprise to find that two-dimensional techniques have
also been found useful for the study of solid state systems. We give
an example of an EXSY experiment used to study motion in the
solid polymer [(Me3Sn)4Ru(CN)6]00; in which the tin coordination is

(a)

(b)

\ '
250 200 150 100 50 -50
8 (ppm)
Figure 11.8 The 13C CP-MAS spectra at 15 MHz of (a) a sub-bituminous coal
and (b) polycarbonate solids. (From Wind (1991) Modern NMR Techniques
and Their Application in Chemistry, eds Popov and Hallenga, Marcel Dekker
Inc., New York, p. 186, with permission.)

www.pdfgrip.com
 / = 1/2 nuclei with high magnetogyric ratios 365

F2 / PPM
-4-
-3-
-2 -
-1 -
0-
1-
2•
3-
4-
5-
6-
7-
8-
9-
10-
1 0 9 8 7 6 5 4 3 2 1 0 - 1 - 2 -3 -4
F1 / PPM

Figure 11.9 Decoupled 13C proton two-dimensional exchange spectrum at 75.4

MHz and with CP-MAS of [(Me3Sn)4Ru(CN)6]00. The off-diagonal signals
link methyl carbon atoms belonging to the same Me3Sn group, there being
two sets of three as shown by the dashed lines. (Harris (1995) Frontiers
of Analytical Spectroscopy, eds Andrews and Davies, Royal Society of
Chemistry, p. 77.)

trigonal bipyramidal with the methyl groups in equatorial positions.

The crystal symmetry is such that there are two distinguishable SnMe3
groups and so six methyl 13C signals. However, the SnMe3 groups can
rotate about the axial bonds and line shape calculations could in prin-
ciple give kinetic data, except that it is not known which three signals
belong to a given SnMe3 group. This can be determined by the two-
dimensional exchange experiment, the results of which are shown in
Fig. 11.9 and where it will be seen that the cross peaks neatly connect
two sets of three carbon resonances on the diagonal.

11.3 / = 1/2 NUCLEI WITH HIGH MAGNETOGYRIC RATIOS

Here we are thinking specifically of the nuclei !H or 19F, where the

homonuclear interactions are very strong and so difficult to remove
by MAS. 1H spectra are the most difficult to deal with, since not only

www.pdfgrip.com
366 High-resolution solid-state NMR

are the static linewidths very large, but the chemical shifts are small,
so making big demands on the resolution ability of the system. MAS
at the very highest spinning speeds and using the highest magnetic
fields to maximize the chemical shift dispersion is capable of reducing
a static linewidth of the order of 10 kHz to around 1500 Hz, and
this can give resolvable resonances though the actual improvement
obtained is very sample dependent. Alternatively, the spins can be
swung around by a succession of pulses so that they appear to adopt
the magic angle in the rotating frame. A typical sequence, called
MREV-8, is shown in Fig. 11.10 and has the effect that the magneti-
zation is shifted quickly between the three orthogonal axes, which, as
it will be remembered from section 4.3 and Fig. 4.7, are placed at the
magic angle from their three-fold symmetry axis. The spins thus hop
around the magic angle axis and their dipole-dipole interaction is
much reduced, though the chemical shift anisotropy and heteronuclear

(I)

One MREV-8 cycle

(II)

20 10 0 8(ppm)

www.pdfgrip.com
 / = 1/2 nuclei with high magnetogyric ratios 367

OH)
(f) Expansion

10

(e) CRAMPS

(d) 11.0 kHz

(c) 8.0 kHz

(b) 6.0 kHz

(a) Static

40 20 0 -20 -40
kHz

Figure 11.10 (I). The eight-pulse MREV-8 cycle. Each pulse is a 90° pulse,
which rotates the magnetization around the x or y axes in one direction or
its opposite. The large spacings are double the length of the small spacings.
The nuclear signal is sampled between pulses. (II) The *H CRAMPS spec-
trum at 200 MHz of aspartic acid HOOCCHNH 2CH2COOH. (From Bruker
Report 1/1988, with permission.) (Ill) ]H NMR spectra of solid adipic acid
(a) static, showing the broadening due to dipolar interactions, (b, c, d) with
MAS at various speeds, none exceeding the linewidth of the static spectrum
(40 kHz) and (e) with CRAMPS, (f) is an expansion of (e) and it will be seen
that the three types of proton are well resolved. (From Maciel, in Gerstein
(1996) Encyclopedia of Nuclear Magnetic Resonance, 3, 1501, eds Grant and
Harris, copyright John Wiley and Sons Ltd, with permission.)

www.pdfgrip.com
368 High-resolution solid-state NMR

interactions are not affected. As might be expected, then, since these

will be reduced by MAS, a combination of the two methods proves
very successful. This technique is called combined rotation and
multiple-pulse spectroscopy (CRAMPS). Resolution of the order of
180 Hz is possible in 1H spectroscopy. The CRAMPS spectrum of
aspartic acid, HOOCCHNH2CH2COOH, is also shown in Fig. 11.10,
where resonances due to the four different types of proton can easily
be distinguished.
A second example of a CRAMPS spectrum is shown in Fig. 11.10
for adipic acid (HOOCCH2CH2CH2CH2COOH with three different
types of proton), which also shows the way the spectrum changes from
its static shape with the introduction of MAS at various speeds and
finally the improvement in resolution gained by introducing the multi-
pulse sequence as well as MAS. The pulse sequences are arranged to
return the spin system to its original state at the end of each cycle of
pulses and the spectrometer output is then sampled, digitized and
stored to give the usual FID in memory.
In favourable cases the resolution required can be obtained using
only MAS with double irradiation if appropriate. An example is given
of the study of the ring inversion of fluorocyclohexane in its solid
thiourea inclusion compound. The host thiourea contains tunnels in
which the guest molecule resides and many studies have been made
of a variety of monosubstituted cyclohexanes in this environment to
try to understand the constraints put upon the conformation of the
cyclohexane in these tunnels. It has been found that if the substituent
is Me, NH2 or OH then the substituent takes predominately the equa-
torial position whereas with halide substituents the axial conformer
predominates. This work was done using 13C NMR but for the fluoro
isomer it is possible to use the much more receptive 19F nucleus.
Double irradiation of the protons is required but this presents tech-
nical problems due to the proximity of the XH and 19F frequencies and
the high power used for double irradiation and a specially designed
probe is required. The 19F spectra obtained in this way are shown
in Fig. 11.11, obtained as a function of temperature to give the
typical two site exchange bandshapes. Rate constants were obtained
by computer fitting of the bandshapes. The equatorial and axial
conformers have chemical shifts of -163.9 and -187.0 ppm respectively.
It will be evident in this case that the two conformers have apparently
equal populations though careful assessment of the data shows that at
lower temperatures, there is a slight preference for the equatorial
conformer.

11.4 MAS OF QUADRUPOLAR NUCLEI

Quadrupolar nuclei in the solid state usually have weak dipolar inter-
actions with their surroundings, and this will not concern us here.
Where they do exist then a cross-polarization experiment may be

www.pdfgrip.com
 MAS of quadrupole nuclei 369

(a) (b) kS'1

142.8

32.0

14.2

• 5.4
• 2.1

0.94

0.18

0.0015

-150 -160 -170 -180 -190 -200 -150 -160 -170 -180 -190 -200
8F/ppm

Figure 11.11 19F spectra of fluorocyclohexane in its solid inclusion compound

with thiourea. The temperatures are, from bottom to top: 177, 217, 237, 247,
258, 268, 278 and 300K. (a) is the calculated spectrum and (b) is the actual
spectrum. The ring inverts readily at 300K. (From Harris et al (1999) Magn.
Reson. Chem., 37, 15, copyright John Wiley and Sons Ltd, reproduced with
permission.)

possible. Quadrupole nuclei are, of course, polarized by the magnetic

field but are also subject to any electric field gradient present at their
position in the molecular and crystallographic environment, and which
can arise from the bonding electrons and, in contrast with the liquid,
from more distant electronic distribution. The nature of the interac-
tion depends upon whether the nucleus examined has an integral or
half-integral spin. As there are only effectively three important nuclei
with integral spin, namely 2H, 6Li and 14N, we will discuss initially only
the half-integral spin nuclei. If a single crystal of a solid that contains
such a nucleus is placed in a strong magnetic field at some particular
orientation of a crystal axis to that field, then we have also determined
the way the electric field gradient interacts with the several possible
spin states. If the electric field gradient (EFG) is zero, then the energy
gap between spin states is the same for any pair: that is, the energy
of the transitions possible for a I = 3/2 nucleus, 3/2<->l/2, 1/2^-1/2,
-l/2<-^-3/2 are all equal and a single resonance results. If there is an
EFG, then the energy of each spin state is altered. The 1/2 and -1/2
states move in parallel, so that the transition energy is unaltered, but
the 3/2 and -3/2 states change in opposite directions and so reduce

www.pdfgrip.com
370 High-resolution solid-state NMR

the energy of one transition and increase the energy of the other. The
degeneracy of the three possible transitions is now lifted and we detect
three resonances. This is shown schematically in Fig. 11.12. If we
change the orientation of the crystal in the magnetic field, we change
the orientation of the EFG tensor and so the interaction of the nucleus
with the EFG. This alters the energy levels: though the l/2<-»-l/2 tran-
sition frequency is unchanged, the other two will both be affected, one
being increased and one decreased. At certain orientations (say 0°)
they will coincide with the 1/2^-1/2 transition and, as the crystal is
rotated, they will move away from this l/2<-»-l/2 central line, reach a
maximum displacement, return to the central line, cross over, reach
maximum again and, at 180° rotation for crystals with axial symmetry,
again coincide with the central line. This is known as the first-order

Figure 11.12 Energy level diagram for a nucleus / = 3/2. The transition ener-
gies are equal if only the magnetic field BQ is taken into account, but the
quadrupole interaction causes one to become larger, one to become smaller
and one, the central transition, to remain unchanged.

www.pdfgrip.com
 MAS of quadrupole nuclei 371

quadrupole effect. An example of how the outer bands move as a

function of rotation angle is given in Fig. 11.13, which shows the results
for a single crystal of sodium nitrate placed so that its three-fold axis
of symmetry could be rotated to describe a plane parallel with U0. The
maximum displacement of the satellites is proportional to the magni-
tude of the quadrupole coupling constant and to 3cos 2 0-l, where 6
is the angle between the EFG tensor axis (Vzz) and the direction of
the magnetic field BQ. In a powder sample, where the crystallites are
arranged at random, the satellite lines have all possible positions and
so smear out into the baseline, leaving only the central line invariant
and detectable.
Because of the tensor nature of the EFG, this description is over-
simplified and we have to take into account a second-order perturba-
tion of the energy levels and so of the frequencies of all the lines. If
we define a quantity VQ as a measure of the magnitude of the quadru-
pole interaction, where
3^6
VQ
2//z(2/-l)
and provided the quantity vQ2/vQ is appreciable in magnitude, where
vQ is the Larmor frequency of the nuclear species examined, then the
frequency of the central l/2<-> -1/2 transition also depends upon the
angle 6 but in a more complex way. Thus this second-order shift v2 is

200-

-200
60 120 180
Degrees rotation

Figure 11.13 The positions of the two satellite lines in the 23Na spectrum of
a single crystal of NaNO3 as a function of angle between the B0 magnetic
field direction and the three-fold symmetry axis of the crystal. The origin of
the frequency axis is the frequency of the central line. (Derived from Andrew
(1958) Nuclear Magnetic Resonance, Cambridge University Press, Cambridge.)

www.pdfgrip.com
372 High-resolution solid-state NMR

+
v2 = - (^[W !) - IJC1 - cos26)(9cos2e - 1)

for an axially symmetric field gradient with iq = 0. These terms have

already been defined in Chapter 4. In a powder sample, this second-
order angular dependence of the line frequency imparts width, which,
it has to be emphasized, is a shift effect, not a relaxation effect as in
the liquid phase. The line has breadth v1/2 and shift from the isotropic
resonance frequency 82 of approximately
25vQ2 vQ2
v,1/2/9 = — and -o2? = -----
18v0 20v0
The line shape of a powder sample is thus seen to be a valuable source
of information about the nuclear environment, though the centre of
the line is not at the chemical shift position and the true chemical shift
has to be calculated if a precise value is required. The form of the
spectrum is also a function of the spectrometer magnetic field strength,
and the second-order effects become less evident as this is increased.
Thus high magnetic fields are essential for the high resolution study
of quadrupolar nuclei in the solid state, though if the value of the
quadrupolar coupling constant is required then this is better obtained
at lower magnetic fields.
The question we must next ask is, of course, will magic angle spin-
ning reduce this linewidth? Evidently, if there is any contribution from
chemical shift anisotropy or dipolar coupling then this will be reduced
or eliminated. Otherwise MAS is not so magic in the case of the
quadrupolar nuclei. The second-order term, which does not vary as
3cos26 -1, is not eliminated by MAS, though it is reduced by a factor
of about four times. Thus if nq = 0, the width and shift of the reso-
nance under MAS are approximately
v VQ2 anc , a vQ2
i/2 - 3v
o—0 * ~ ^22 = ~r~
4v0
One has to use the term 'approximately' in describing these quan-
tities because the line shape is complex and quite unlike the Lorentzian
lines obtained with liquid samples. A theoretical line shape is sketched
in Fig. 11.14 for a nucleus of / = 5/2 in a powder sample undergoing
MAS with T] = 0. Perhaps the most important point to note from this
sketch is the fact that the true isotropic chemical shift of the reso-
nance is not at the centre of the resonance but is to high frequency
always. In practice, the line shape is rounded off from the rather
angular shape shown, and an estimate of the true shift can be made
as being the point where the high frequency side of the resonance has
lost almost all its intensity. Clearly, this is approximate, and any accu-
rate estimate requires the line shape to be calculated for a particular
spectrum. It is, however, a better approximation than that often
encountered currently of giving the shift as the centre of the resonance.

www.pdfgrip.com
 MAS of quadrupole nuclei 373

0 ± JL ±
2 1 6 3

Figure 11.14 A sketch of the line shape of a / = 5/2 nucleus in a powder

sample undergoing MAS. The frequency scale is calibrated in units of vQ2/2v0
and the origin of the shift scale is the Larmor frequency in the absence of
any second-order quadrupole interaction. (From Akitt (1981) Prog. NMR
Spectrosc., 21, 1, copyright (1981) Elsevier Science with permission.)

This adequately describes the position of the peak on the spectrum

but is not its chemical shift.
The line shape of a quadrupolar nucleus in a spinning sample is
strongly dependent on the angle of the turbine set relative to jB0. In
fact, the best angle for reducing the second-order linewidth is not the
magic angle, and it is better to choose some other angle. The actual
value to use is a compromise between reducing the second-order
effects and those effects, if present, that depend upon 3cos20 - 1 terms.
It is thus necessary to be able to vary the spinning angle, and this has
been given the name VAS - variable angle spinning.
Some of the aspects of solid-state NMR that we have just discussed
are summarized in Fig. 11.15, which shows the 27A1 spectra of the tride-
cameric aluminium cation already discussed in Chapter 4. This has
the formula [A1O4A112(OH)24(H2O)12]7+, with one Al in a site of high
tetrahedral symmetry and so low quadrupole coupling constant and
the remaining 12 in sites of distorted octahedral symmetry and so with
high quadrupole coupling constants. The solution-state spectrum
contains one narrow line (A1O4) and a very broad, usually undetectable
line 58ppm to high frequency (A1O6). The solid-state spectrum
obtained at low field is similar, though it must be appreciated that the
A1O6 resonance is broadened by quite different mechanisms in
the two phases. As the field and v0 are increased, the second-order
broadening is reduced and the octahedral resonance becomes visible
((c), (b) and (a)). If the spinning angle is changed to 75°, then the
A1O6 linewidth is further reduced (d) and two spinning sidebands are

www.pdfgrip.com
374 High-resolution solid-state NMR

(a) (d)

(b) (e)

(c) (f)

200 0 -200 200 0 -200

ppmfrom Al (H2O)g3+

Figure 11.15 The 27A1 solid-state spectra of the tridecameric aluminium cation
at different magnetic fields, with and without sample spinning. Spectrometer
frequencies were (c), (f) 39 MHz, (b), (e) 93.7 MHz and (a), (d) 129.7 MHz.
Samples (e) and (f) were static, samples (a), (b) and (c) were spinning at the
magic angle, and sample (d) was spinning at an angle of 75° to the magnetic
field direction. (After Oldfield et al (1984) J. Magn. Reson., 60, 467, with
permission.)

also visible. It is remarkable to note that, in this spectrum, we have

better resolution of the octahedral aluminium resonance than we do
in solution. The spectra of static samples are also shown and make
very evident the improvement in resolution obtained with spinning.
The static spectra can nevertheless be used to calculate the quadru-
pole coupling constants, which are to be used to interpret the spectra
of the spinning samples.

www.pdfgrip.com
 MAS of quadrupole nuclei 375

A further possibility for improving the resolution of resonances

exists in the solid state which is absent for liquid samples. Since the
second-order quadrupole effect produces frequency shifts, then sites
with similar chemical shifts but different quadrupole coupling constants
may well have resonances with centroids that are well separated.
A striking example of this is shown in Fig. 11.16, which shows the
27
Al MAS spectrum of CaO3Al2O3-3H2O obtained at 130MHz.
The crystal contains two crystallographic types of four-coordinate
aluminium but which have a chemical shift of only 1 ppm, or 130 Hz
in the spectrometer used. The quadrupole coupling constants at the
two sites are, however, very different, so the line shapes are different
and their centroids are separated by some 10 ppm, sufficient to observe
the resonances separately and to compute their line shapes. Note also
how the isotropic chemical shifts, shown by two marks on the chemical
shifts axis near 80 ppm, coincide with essentially zero signal intensity.

80 70 60 50
8 (ppm)

Figure 11.16 The 27A1 MAS spectra at 130 MHz of CaO3Al2O3-3H2O showing
the two superimposed signals and their computed line shapes. Note particu-
larly the two close values of the true isotropic chemical shifts, which are
marked on the shift axis and correspond with hardly any signal intensity. Note
also how two types of chemically very similar aluminium are well differenti-
ated by different quadrupole couple constants. (Reprinted with permission
from Muller et al (1986) /. Chem. Soc., Dalton Trans., 1277.)

www.pdfgrip.com
376 High-resolution solid-state NMR

11.5 SOME APPLICATIONS

Solid-state NMR can usefully be applied to the determination of the

state of the cations in the alkalide salts. These substances typically
have formulae MM'Ln, where M and M' are an alkali metal or metals
and L is a strong complexing ligand, either a cryptand or a crown
ether capable of enclosing or partially enclosing the alkali-metal cation.
The second alkali metal is present as the anion, M~. The 23Na MAS
spectrum of the alkalide Na[C222.Na] is shown in Fig. 11.17. C222 is
the cryptand N(C2H4OC2H4OC2H4)3N. Two resonances are observed,
one with an isotropic shift near that of the standard at 0.0 ppm,
Na+(aq), and one with a shift significantly to low frequency, as would
be expected for the greater electron density on Na~. If the alkalide
contains two different alkali-metal atoms, then the solid-state chem-
ical shifts of each will indicate which is the anion and which the cation.

20 0 -20 -40 -60 -80 -100

5 (ppm)

Figure 11.17 The 23Na MAS NMR spectra of the homonuclear alkalide
Na[C222.Na]. (From Dye et al (1991) Modem NMR Techniques and
Their Application in Chemistry, eds Popov and Hallenga, Marcel Dekker Inc.,
New York, p. 291, with permission.)

www.pdfgrip.com
 Some applications 377

The examples shown up to present have all given data comparable

with crystallographic work, and the work could in many cases have
been achieved equally well by this older technique, though NMR is
perhaps less time-consuming and can be used to follow the effect of
small changes in sample conditions with relative ease. There is,
however, a whole class of compounds where crystallography is of little
help, such as disordered solids, glasses and amorphous substances,
heterogeneous substances and solids with minor but important compo-
nents that are not picked up by diffraction experiments. NMR is ideal
for the study of such materials and we give several examples below.

11.5.1 The setting of cement

Monocalcium aluminate, CaOAl2O3, is the main constituent of high-
alumina cement and the ability to investigate its hydration, i.e. how it
sets, is of obvious general interest and technological importance. The
dry starting material contains four-coordinate aluminium, which
becomes six-coordinated upon hydration. There is a chemical shift of
some 70 ppm between the two types of aluminium, and so the hydra-
tion can be followed through all its stages. A typical 27A1 MAS spec-
trum is shown in Fig. 11.18, together with plots showing how the
proportion of octahedral aluminium changes with time and also the
amount of heat of reaction evolved with time. Setting is more rapid
at high temperatures and the conversion to the octahedral form is
more complete. The rate of reaction also varies quite widely at the
two higher temperatures and a series of steps are observed. The initial
reaction is believed to produce a mixture of phases, which covers the
unreacted material and causes the reaction to slow down. This covering
then suffers transformation to new phases, which expand and loosen
the coating and so permit reaction to proceed again.

11.5.2 Zeolites
These substances are formed of networks of aluminosilicates that
contain pores of certain fixed sizes and are active as catalysts. They are
produced by crystallization from a gel formed upon mixing, say, an
aluminate salt with a soluble silicate, and the structure of the zeolite
formed depends upon the nature of the components used to form the
gel. The ratio of silicon to aluminium present in the zeolite can be
varied within certain limits and the silicon is always in excess. The solids
thus contain Si-O-Si and Si-O-Al linkages but not A1-O-A1 linkages,
which appear to be forbidden. These substances contain two magneti-
cally active nuclei, 29Si with / = 1/2 and the quadrupolar 27A1 with
I = 5/2, and both are used extensively in their study. The 29Si spectra
may contain up to five resonances, which correspond to tetrahedral
SiO4 units with zero, one, two, three or four attached aluminium atoms.
A spectrum of zeolite Na-Y with Si/Al = 2.61 is shown in Fig. 11.19,
where the resolution of the different environments is seen to be

www.pdfgrip.com
378 High-resolution solid-state NMR

10 15 20 25 3 5
Hydration time (min)

Figure 11.18 Hydration of calcium aluminate. A typical 27A1 MAS spectrum is shown at the top left of
the figure. The three other plots show the progress of hydration with time at three different temperatures.
The full curves show the heat evolution and the broken curves show the percentage of six-coordinate
aluminium formed. SB on the spectrum signifies spinning sidebands. (Reprinted with permission from
Rettel et al (1985) Br. Ceram. Trans. J. 84, 25.)

excellent. With care, the intensities are quantitative and the pattern
allows the Si/Al ratio to be calculated. In natural zeolites, this ratio is
always less than about 5, but materials with much lower aluminium
contents can be synthesized. The series of highly siliceous zeolites called
ZSM-5 with Si/Al typically 31 are a well-known range of catalysts with
extra stability conferred by the high silicon content. The 29Si spectra of
such substances are simple with essentially a single line due to Si(OSi)4
units. A second material isostructural with ZSM-5 but with only a trace
of aluminium and called silicalite is also known. If the aluminium
content is particularly low, then the single-line 29Si spectrum is found
to be resolved into a group of lines, which arise from the various crys-
tallographic sites in the as-yet uncertain structure of this material. This
remarkably well-resolved spectrum is also shown in Fig. 11.19. Because
the aluminium content of silicalite is so low, it was argued in order
to be able to patent its use, that it was not a zeolite and that any
aluminium was present as alumina impurity. 27A1 is a good, receptive
nucleus and can be detected at quite low levels in solids, which are
concentrated states of matter. The 27A1 MAS spectrum of the same

www.pdfgrip.com
 Some applications 379

sample of silicalite is shown in Fig. 11.19, and it is evident that the 27A1
is detectable, even though an accumulation time of over two days was
required to collect the 176214 FIDs needed. The chemical shift is
diagnostic for tetrahedrally coordinated aluminium, so that the alu-
minium is to be found within the silicalite framework and, further, there
are at least two different aluminium environments. Note, again, that it
is the peak centroids that are indicated on the figure. The same type
of structure has since been observed in ZSM-5 that has been thoroughly
de-aluminated to reach Si/Al = 800.
An alternative approach to these catalysts is to take, for instance,
zeolite Y and subject it to what is known as decationation and ultra-
stabilization. The ammonium form of the zeolite is subjected to heat
treatment under vacuum conditions, when it loses ammonia and water.
The resulting crystalline material has much greater stability than the
starting material and is a good catalyst used for hydrocracking in
the petroleum industry. It has a much reduced ion-exchange capacity
and this indicates that aluminium has been lost from the framework,
the vacancies created being reoccupied by silicon. The aluminium
remains but can subsequently be leached out of the solid catalyst.

(a

-80 -100 -120

8 (ppm) from IMS

(b) (C)

-108 -110 -112 -114 -116 -118 75 50 25

+
8 (ppm) from IMS 8 (ppm) from AI(H2O)i

Figure 11.19 (a) The 29Si MAS spectrum of zeolite Na-Y with Si/Al = 2.61 obtained at 79.8 MHz. Five
Si environments are indicated, (b) The 29Si MAS spectrum of silicalite with Si/Al > 1000 obtained at
99.32 MHz. The resonances are all from SiO4 with no directly linked Al. (c) The 27A1 MAS spectrum of
the same sample taken at 104.2 MHz and the result of accumulating 176 214 FIDs. (From Klinowski and
Thomas (1985) Adv. Catal, 33, 199, and Fyfe et al (1982) /. Phys. Chem., 86, 1247, and Klinowski et al
(1984) Prog. NMR Spectrosc., 16, 237, copyright (1984) Elsevier Science, reprinted with permission.)

www.pdfgrip.com
380 High-resolution solid-state NMR

This ultrastabilization process has been studied by both 29Si and 27A1
spectroscopy, as shown in Fig. 11.20. The starting material (a) had
Si/Al = 2.61, four lines in the 29Si spectrum and all tetrahedral Al.
After calcining in air at 400°C for one hour (b), there are evidently
fewer aluminium atoms linked to the silicon and some octahedral
aluminium has appeared. Si/Al was calculated to be 3.37, a calcula-
tion that does not include the octahedrally coordinated metal, since it
is based on the 29Si intensities. More drastic treatment, heating in steam
at 700°C, produces even greater spectral changes and an Si/Al ratio
of 6.89 (c). Repetition of this procedure followed by leaching with
nitric acid (d) removes most of the aluminium to give Si/Al = 50 and
a single 29Si line, which indicates good crystallinity as is required if the
Al vacancies are filled. The octahedral aluminium resonance becomes
very narrow on leaching and represents remaining Al in the form of
[A1(H2O)6]3+ free to rotate in lattice cavities. These changes can also
be achieved by treatment with SiCl4 vapour or the aluminium can be
put back into the structure using A1C13 vapour, both processes having
been monitored by 27A1 MAS.

Si(2AI) Tetrahedral
\Si(1AI) (a)

Tetrahedral
(b)

Tetrahedral Octahedral
Si(OAI)
(c)

Octahedral
Si(OAI)l (d)

Tetrahedral
yy
_8Q -90 -100 -110 -120 200 100 0 -100
8 (ppm) from TMS 6(ppm)from[AI(H2O)6]3+

Figure 11.20 High-resolution MAS spectra monitoring the ultrastabilization of

zeolite-Y as described in the text. The left-hand spectra are 29Si at 79.8 MHz
and those on the right are 27A1 at 104.2 MHz. (From Thomas and Klinowski
(1985) Adv. CataL, 33,199 and from Thomas and Klinowski (1982) Nature, 296,
533-6, copyright (1982) Macmillan Magazines Ltd, reprinted with permission.)

www.pdfgrip.com
 Deuterium, an integral-spin nucleus 381

11.6 DEUTERIUM, AN INTEGRAL-SPIN NUCLEUS

The patterns obtained in the solid state with the nucleus 2H, for which
7 = 1 , are somewhat different from those described above. A nucleus
with 7 = 1 has three energy levels and two degenerate transitions in
the absence of any quadrupole coupling. The interaction of the nucleus
with the magnetic field and the electric field gradient causes the three
energy levels to be modified, so that there are two transition frequen-
cies disposed symmetrically about the isotropic chemical shift value
and with a frequency difference that is proportional to the quadru-
pole coupling constant and to the orientation of the bond to deuterium
(and so the electric field gradient) relative to the magnetic field. In a
powder sample all orientations exist and the resulting 2H spectrum,
shown in Fig. 11.21, has a particular shape and is known as a Pake
powder pattern. The separation of the sharp edges of this spectrum is
three-quarters of the value of the quadrupole coupling constant and
usually lies in the range 120-150 kHz. If the moiety in which the
deuterium lies is capable of rotation in the solid, then the width of
the Pake pattern will be reduced and the extent of the reduction and
the shape of the pattern will depend upon the details of the motion.
The motion has to be fast relative to the value of the quadrupole
coupling constant. These comments will be illustrated by reference to
the 2H spectra of deuteriobenzene absorbed on graphite to form a
multilayer some ten molecules thick. Spectra were obtained at several
temperatures and are shown in Fig. 11.22. At 298K a singlet narrow
signal is observed, and shows that the absorbed benzene is reorienting
as if it were in the liquid phase. Indeed, this is found to be the case
even if the amount of benzene absorbed is reduced until it forms a
monolayer. At 170K the spectrum is of mixed form, with a Pake
pattern and a minor singlet. This latter is due to the absorbed mono-
layer, which is still undergoing fast two-dimensional motion, while the
Pake spectrum arises from benzene crystallites that form further out
from the surface. The splitting between the sharp edges of the Pake
pattern is 70 kHz, and this is typical of the value found for benzene
undergoing fast rotation around its hexad axis. The spectra obtained
at 90K were run using two different sets of conditions. In one, the

Figure 11.21 The shape of the 2H resonance in a solid powder sample, or

Pake powder pattern.

www.pdfgrip.com
382 High-resolution solid-state NMR

pulse repetition rate was slow, so that all the deuterons were detected.
In the other, the pulse repetition rate was much faster, so that the
component with the longer relaxation time T^ was saturated and effec-
tively was removed from the spectrum. In the first case, two compo-
nents are observed, both Pake patterns, and with splitting of 70 and
140 kHz. In the second case, the spectrum of the less mobile phase
has disappeared and the 70 kHz pattern remains. Note, though, that
this is not the same as that seen at 170K, since there is a distinct asym-
metry in the base. This is due to the now solid absorbed monolayer
rotating only around its hexad axis, and it follows that the broader
pattern arises from the now static benzene crystallites.
This type of spectroscopy is also much used in the study of liquid-
crystal phases, where the Pake-type patterns obtained from these
partially ordered materials can give much information about the
degree of order and the rates of motion, and how these change with
the experimental conditions.

50kHz
(a) (b)

50kHz
(c)
(d)

-v — «-v—

Figure 11.22 The 2H NMR spectra of deuteriobenzene (benzene-^) absorbed

on graphite to a thickness of 10 molecular layers: (a) temperature 298K; (b)
at 170K; (c) at 90K and time between read pulses of 10 s; (d) at 90K but with
the time between pulses 0.2 s so that the broad component is saturated. (From
Boddenberg and Grosse (1987) Z. Naturforsch, 42a, 272-4.)

www.pdfgrip.com
 Questions 383

11.7 QUESTIONS

11.1. Figure 11.23 shows the 13C NMR spectrum at 9.4 T of a solid
copper cyanide sample enriched in both 13C and 15N. The sample
is stationary. The shape of the spectrum shows principally the
chemical shift anisotropy though there are also contributions from
coupling (direct and indirect) between the nuclei present.
Calculate the approximate principal chemical shifts and the
isotropic chemical shift of the 13C given that TMS is at 0 ppm.
What information does this spectrum give about the structure of
(CuCN),7?

400 300 200 100 0 -100 -200

ppm

Figure 11.23 The 13C NMR spectrum at 9.4 T of solid copper cyanide enriched
in 13C and 15N. The sample was non-spinning. (After R.E. Waylishen et al.
(1999) /. Am. Chem. Soc., 121, 1528, with permission, copyright (1999)
American Chemical Society.)

11.2. Figure 11.24 shows the 31P CP NMR spectrum of a stationary

powder sample of (Ph2P)2CH2. As in the previous example, calcu-
late the principal chemical shifts and the isotropic chemical shift
of 31P.

50 0 -50 -100
ppm

Figure 11.24 The 31P CP NMR spectrum of a non-spinning powder sample

of (Ph2P)2CH2. (After R.E. Waylishen et al. (1999) Inorg. Chem., 38, 639, with
permission, copyright (1999) American Chemical Society.)

www.pdfgrip.com
www.pdfgrip.com
 Bibliography

NMR is a subject of sufficient importance to have a considerable liter-

ature devoted entirely to various aspects of both its technology and
its use. The following list represents only a fraction of the total avail-
able but is sufficient to give an entry into the field.

COMPREHENSIVE WORKS FOR GENERAL REFERENCE

Abragam, A. (1961) The Principles of Nuclear Magnetism, Oxford

University Press, Oxford, reprinted 1983, ISBN 019852014X.
Grant, D.M., Harris, R.K. (ed.) (1996) Encyclopedia of Nuclear
Magnetic Resonance, Wiley, ISBN 0471938718.
Jackman, L.M. and Sternhell S. (1969) Applications of Nuclear
Magnetic Resonance Spectroscopy in Organic Chemistry, Pergamon
Press, Oxford, ISBN 0080125425.
Pople, J.A., Schneider, W.G. and Bernstein, H.J. (1959) High
Resolution Nuclear Magnetic Resonance, McGraw-Hill, New York,
ISBN 70505160.

INTRODUCTORY TEXTS, SOME WITH A VARIETY OF

PROBLEMS

Fisher, J., Loftus, P. and Abraham, R. J. (1988) Introduction to NMR

Spectroscopy, Wiley, Chichester, ISBN 0471918938.
Glinther, H. (1995) NMR Spectroscopy : Basic Principles, Concepts,
and Applications in Chemistry, 2nd edn, Wiley, New York, ISBN
047195201X.
Harris, R.K. (1986) Nuclear Magnetic Resonance Spectroscopy,
Addison Wesley Longman Higher Education, ISBN 0582446538.
Sanders, J.K.M. and Hunter, B.K. (1993) Modern NMR Spectroscopy,
2nd edn, Oxford University Press, Oxford, ISBN 0198555679.
Sanders, J.K.M., Constable, E.G. and Hunter, B.K. (1995) Modern
NMR Spectroscopy : A Workbook of Chemical Problems, 2nd edn,
Oxford University Press, Oxford, ISBN 0198558120.

www.pdfgrip.com
386 Bibliography

THE TECHNIQUES OF NMR

Braun, S., Kalinowski, H.-O., and Berger, S. (1996) 150 and More Basic
NMR Experiments: A Practical Course, 2nd edn, Wiley, New York,
ISBN 3527295127.
Canet, D. (1996) Nuclear Magnetic Resonance: Concepts and Methods,
Wiley, New York, ISBN 0471961450.
Derome, A.E. (1987) Modern NMR Techniques for Chemistry
Research, Pergamon, Oxford, ISBN 0080325149.
Fukushima, E. and Roeder, S.B.W. (1998) Experimental Pulse NMR,
10th edn, Addison Wesley, Reading, MA, ISBN 0201627264.
Martin, M.L., Delpeuch J.J. and Martin, GJ. (1980) Practical NMR
Spectroscopy, Heyden, London, ISBN 0855014628.
Mullen, K. and Pregosin, P.S. (1976) FT NMR Techniques-A Practical
Approach, Academic Press, New York, ISBN 0125104502.
Shaw, D. (1984) Fourier Transform NMR Spectroscopy, 2nd edn,
Elsevier, Amsterdam.
Neuhaus, D. and Williamson, M. (1989) The Nuclear Overhauser
Effect, VCH, Weinheim, ISBN 3527266399.

SHIMMING

Conover, W.W. (1983) Top. Carbon-13 NMR Spectrosc., 4, 37-51.

SAM 1.0, Shimming Simulation Software Package for IBM-PC
Compatible Computers, ACORN NMR, 46560 Fremont Blvd.,
Fremont, CA 94538-6482.

DYNAMIC NMR

Jackman, L.M. and Cotton, F.A. (eds) (1975) Dynamic NMR

Spectroscopy, Academic Press, New York.
Kaplan, J.I. and Fraenkel, G. (1980) NMR of Chemically Exchanging
Systems, Academic Press, New York.
Sandstrom, J. (1982) Dynamic NMR Spectroscopy, Academic Press,
London, ISBN 0126186200.
Oki, M. (1985) Applications of Dynamic NMR Spectroscopy to Organic
Chemistry, VCH, Deerfield Beach, FL, ISBN 0895731207.

TEMPERATURE CALIBRATION

Van Geet, A.L. (1968) Anal. Chem., 40, 2227; ibid, 1970, 42, 679.

www.pdfgrip.com
 Bibliography 387

PULSED FIELD GRADIENTS

Berger, S. (1997) Prog. NMR Spectrosc., 30, 137-56.

Norwood, TJ. (1994) Chem. Soc. Rev., 23, 59.

ORGANIC SPECTRAL INTERPRETATION

Atta-ur-Rahman, Choudhary, M.I., Ernst, R.R. and Jackman, L.M.

(1995) Solving Problems With Nmr Spectroscopy, Academic Press,
San Diego, ISBN 0120663201.
Breitmaier, E. (1993) Structure Elucidation by NMR in Organic
Chemistry: A Practical Guide, Wiley, Chichester, ISBN 0471937452.
Duddek, H. (1998) Structure Elucidation by Modern NMR; A
Workbook, Springer-Verlag, Darmstadt, ISBN 3798509301.
Lambert, J.A., Shurvell, H.F., Lightner, D.A., and Cooks, R.G. (1998)
Organic Structural Spectroscopy, Prentice Hall, ISBN 0132586908.
Pretsch, E. and Clerc, J.-T. (1997) Spectra Interpretation of Organic
Compounds, Wiley, ISBN 3527288260.
Williams, D.H. and Fleming, I. (1995) Spectroscopic Methods
in Organic Chemistry, 5th edn, McGraw-Hill, London, ISBN
0077091477.

RELAXATION

Wehrli, F.W. (1976) Topics in 13C Nuclear Magnetic Resonance

Spectroscopy, (ed. G.C. Levy), Vol. 2, Chap. 6, Wiley-Interscience,
New York.

TWO-DIMENSIONAL NMR SPECTROSCOPY

Brey, W.S. (ed.) (1988) Pulse Methods in ID and 2D Liquid-Phase

NMR, Academic Press, San Diego, ISBN 0121331555.
Croasmun, W.R. and Carlson, R.M.K. (eds) (1994) Two-Dimensional
NMR Spectroscopy, VCH, Weinheim, ISBN 1560816643.
Ernst, R.R., Bodenhausen, G. and Wokaun, A. (1987) Principles of
Nuclear Magnetic Resonance in One and Two Dimensions,
Clarendon Press, Oxford, ISBN 0198556470.
Freeman, R. (1987) A Handbook of NMR, Longman, Harlow.
Freeman, R. (1996) Spin Choreography: Basic Steps in High Resolution
NMR, Spektrum Academic Publishers, ISBN 1901217043.
Friebolin H. (1998) Basic One- and Two-Dimensional NMR
Spectroscopy, 3rd edn, Wiley, New York, ISBN 3527295135.
Popov, A.I. and Hallenga, K. (eds) (1991) Modern NMR
Techniques and Their Application in Chemistry, Marcel Dekker,
New York.

www.pdfgrip.com
388 Bibliography

MULTINUCLEAR NMR SPECTROSCOPY

Berger, S., Kalinowski, H.-O. and Braun, S. (1996) NMR Spectroscopy

of the Non-Metallic Elements, Wiley, New York, ISBN 0471967637.
Harris, R.K. and Mann, B.E., NMR and The Periodic Table (1978)
Academic Press, London, ISBN 0123276500.
Mason, J. (ed.) (1987) Multinuclear NMR, Plenum Press, New York,
ISBN 0306421534.
Pregosin, P.S. and Kunz, R.W. (1979) 31P and 13C NMR of Transi-
tion Metal Phosphine Complexes, Springer-Verlag, Berlin, ISBN
3540091637.
Pregosin, P.S. (ed.) (1991) Transition Metal Nuclear Magnetic
Resonance (Studies in Inorganic Chemistry, No. 13), Elsevier,
Amsterdam, ISBN 044488176X.

COLLECTIONS OF NMR DATA

Mann, B.E. and Taylor, B.F. (1981) 13C NMR Data for Organometallic
Compounds, Academic Press, London, ISBN 0124691501.
Pouchert, C.J. and Behnke, J. (1993) The Aldrich Library of 13C and
2
H FT-NMR Spectra, Aldrich Chemical, Milwaukee, WI, ISBN
0941633349.
Tebby, G. (1991) CRC Handbook of Phosphorus-31 Nuclear Magnetic
Resonance Data, CRC Press, ISBN 0849335310.

SOLID STATE NMR SPECTROSCOPY

Stejskal, E. O. and Memory, J.D. (1994) High Resolution NMR in the

Solid State: Fundamentals of CP/MAS, Oxford University Press,
USA, ISBN 0195073800.
Klinowski, J. and Kolodziejski, M. (1999) Solid-State NMR Techniques,
Imperial College Press, ISBN 1860940897.

IMAGING

Bushong, S.C. (1995) Magnetic Resonance Imaging: Physical and

Biological Principles, C.V. Mosby, St Louis, ISBN 0815113420.
Callaghan, P.T. (1993) Principles of Nuclear Magnetic Resonance
Microscopy, Clarendon Press, Oxford, ISBN 0198539975.
Kimmich, R. (1997) NMR: Tomography, Diffusometry, Relaxometry,
Springer-Verlag, Berlin, ISBN 3540618228.
Mansfield, P. and Morris, P.G. (1982) NMR Imaging in Biomedicine,
Academic Press, New York, ISBN 0120255626.
Mansfield, P. and Hahn E.L. (1991) NMR Imaging, Cambridge
University Press, ISBN 0521404606.

www.pdfgrip.com
 Bibliography 389

Rajan, S.S. (1997) Magnetic Resonance Imaging: A Conceptual

Overview, Springer-Verlag, Berlin, ISBN 0387949119.

The student may also care to read the following few original short
papers which summarize the early and unexpected results that
heralded the development of NMR as a subject useful to chemists:

Arnold, J.T., Dharmatti, S.S., and M.E. Packard, (1951) First obser-
vation of chemical shifts in a single chemical compound, /. Chem.
Phys., 19, 507.
Dickinson, W.C. (1950) Observed chemical shifts in fluorine
compounds and noted the effect of exchange, Phys. Rev., 77, 736.
Gutowsky, H.S. and D.W. McCall, (1951) An early observation of
spin-spin coupling, Phys. Rev., 82, 748.
Gutowsky, H.S., D.W. McCall and Slichter, C.P. (1951) A theory of
spin-spin coupling, Phys. Rev., 84, 589.
Proctor, W.G. and Yu, F.C. (1950) 14N chemical shift between [NH4]+
and [NO3],Phys. Rev., 77, 717.

It should be remembered in reading the two Gutowsky papers that

the hertz separation of the 31P doublet and the 19F doublet are the
same. The gauss separation can be calculated from AU0 = (//v0)B0 and
is greater for 31P since v0 is smaller for the fixed field used.

www.pdfgrip.com
www.pdfgrip.com
 Answers to Questions

Outline answers are given here. More detailed answers can be found
at http://www.thorneseducation.com

CHAPTER 1

1.1 250, 1000, -1000, -250 Hz.

1.2 5000 Hz, 20ms.
1.3 Null point is 25 000 Hz from the carrier. Shorten the pulse
length. Move the carrier frequency closer to the signal.
1.4 Mz = Mxv = 0.707M. Mz = 0, Mxy = M.MZ = -M, M = 0.
1.5 32S, 40Ar, MZn.
1.6 4000.
1.7 54.74°.
1.8 48.2.
1.9 2130 Hz, 3.2 x 10-10.
1.10 2 Hz.

CHAPTER 2

2.1 CHC12.
2.2 8 7.30, 2190 Hz, 8 7.30.
2.3 Worse.
2.4 -15950 ppm.
2.5 Ring current.
3537
2.6 C1 isotopomers.
2.7 The sign of the susceptibility correction is dependent on the
alignment of the magnetic field.
2.8 126.241 609MHz, 8981.
2.9 8 4519.7, 8 -204.0, 8 0, 8 -4701.5.
S-L _ Sll
2.10 ^^ ^ = Aref
y f - Asample*
y

www.pdfgrip.com
392 Answers to questions

CHAPTER 3

3.1 8 3.43, 8 1.67, 8 Hz.

3.2 2.97 :1. 1 : 4 : 6 : 4 :1 quintet.
3.3 14.3 Hz.
3.4 1: 6 :15 : 20 :15 : 6 :1 septet.
3.5 8.6 Hz, 86.479, 86.394, 1.86:0.14, 3244.2 Hz, 3235.6 Hz,
3200.9 Hz, and 3192.3 Hz, 8 6.437, 1.20 : 0.80.
3.6 Unresolved coupling to the ortho-pheny\ protons.
3.7 A2X3 or A2B3, A2X2, [AX]2 or [AB]2, ABC or ABX or AMX,
AX2 or AB2, ABC or ABX or AMX, [AX9]2, [AX]3.
3.8 Doublet of quartets of triplets.
3.9 [IrH5(PEt2Ph)2], mer-frans-[IrH3(PEt2Ph)2(AsMe2Ph)].
3.10 2, 3, 4.
3.11 -3.3 ppm.
3.12 125 Hz, 19 Hz, -0.17 ppm.
3.13 cw-mer-[RuH2(PPh3)3(CO)].
3.14 Diastereotopic protons. Triplet.
3.15 Monomeric, [Li(Me2NCH2CH2NMe2)Bu«].
3.16 Each signal is two doublets of quartets. There are two doublets
as the 107Ag and 109Ag have different coupling constants.
3.17 CHMe2 and H2.
Ex. 1 CH=CHCH2SO2.
Ex. 2 CF3CH2OH.
Ex. 3 CF2HCF2CH2OH.

CHAPTER 4

4.1 22 s.
4.2 1 :1.15. Note !H and 13C relax 1H, whereas only 1H relaxes 13C.
4.3 Since the z coordinate is zero, then no terms can cancel and the
EFG is proportional to -3qr3. For zero EFG, each term of the
sum has to be zero, and if z is the coordinate of the displaced
charge, then 3z2 - r2 = 0 and z = ± r/V3.
4.4 46.9 s, 11.7 s.
4.5 1:0.103.
4.6 3.5 x 10-13 s.
14
4.7 N coupling. Cooling increases the rate of 14N relaxation and
decouples it.
4.8 Preferential rotation about the long axis. The sp carbon relax
by CSA, which depends on B02.
4.9 Scalar relaxation by 79Br.
103
4.10 Rh relaxes by CSA.
4.11 R2 = 282.74 s-1, T2 = 0.0035 s, R2SR = 280.51 s~\ T2SR = 0.0036 s.

www.pdfgrip.com
 Answers to questions 393

CHAPTER 5

5.1 Dwell time = 56 jxs, 32K, 0.910 s, not fully relaxed, maximum
pulse width = 56 JJLS.
5.2 Mz, 0.368MZ.
5.3 Small error in frequency. Poor shimming.
5.4 4V2.
5.5 68.4°.

CHAPTER 6

6.1 In homonuclear decoupling, time sharing prevents interference

of the decoupler frequency with the receiver.
6.2 z direction.
6.3 0.000 117 T.

CHAPTER 7

7.1 Intra-molecular exchange.

7.2 10 Hz, 1:100.
7.3 A//* = 51.7 kJ moH, AS* = -5.2 J Kr1 mol'1.
7.4 1,2-shift, 31.42 s, 45.5 kJ moH.
7.5 Ca at 8 21.01, Cb at 8 128.37, Cc at 8 128.54, Cd at 8 31.01.
7.6 2.6 s-1, 53.0 kJ moH.
7.7 Coalescence temperature depends on the frequency separation
of the signals which is field dependent. 1.5 Hz, 1.5 Hz, 1.14 Hz,
4.55 Hz.

CHAPTER 8

www.pdfgrip.com
394 Answers to questions

CHAPTER 9

9.1 OSnBu'20(jA-F), 8-232, 7(119Sn19F) = 1800 Hz; (u.-F)2SnBu'2,

8-282, V(119Sn19F) = 3300 Hz. 8-163 and -165: cis- and trans-
isomers of [But(F)Si(OSnBut2O)2Si(F)But].

(Ph3P)2Pt

Mo(CO)3

(Ph3P)2Pt-p

Mo(CO)3

9.2 8(195Pt) = -4481; 8(3IP) = 22.1, 23.8; '/(195Pt,31P) = 3325, 3415 Hz;
8(195Pt) = -4449; 8(31P) = 22.9; 7(195Pt,31P) = 3325 Hz;
8(195Pt) (21.4) = -1607, -1575.
9.3 Five coordinate P group is chiral, hence C5H4R CH groups are
diastereotopic.
8(!H) = 4.73(2). 5.17(1), 5.23(1).
S(13C) = 92.3 (/ = 255 Hz); 85.8, 86.4, 94.6, 95.9 (/ = 15 for all).
9.4 P2 and P3 form an AB pattern centred on 30 700 Hz with 7AB =
490 Hz, /AM = 55 Hz, 7AX = 20 Hz, 7BM = 150 Hz.
3 195 1
9.5. /( Pt H) = 7.5,12 Hz. Viewed from platinum, the methyls are
inequivalent.
9.6.
2+

2+

[AX]2 /AX large. Two N doublets with fine structure.

9.7 203T1> 2osT1 csa relaxation.
9.8. 8 92.6: y(195Pt31P) = 2157 Hz; 37(195Pt31P) = 286 Hz; 2/(199Hg31P)
= 1632 Hz; 4/(31P31P) = 100 Hz.
8 62.5: V(195Pt31P) = 3090 Hz; 2/(199Hg31P) = 218 Hz.

www.pdfgrip.com
 Answers to questions 395

9.9. H2 8 9.39; H5 8 8.38; H5' 8 8.30; H6" 8 10.1; H6'"8 6.5. Use COSY.
9.10. Three -1 : -1 :1 :1 quartets centred at 8 -414 (1), -420 (2), -422
(1).
9.11. yO^Rh'H) = 11 Hz produces 1 : -1 doublets. 7(lo3Rh31P) (PPh3)
= 109 Hz (triplet). '/(^Rh^P) {P(OMe)3} = 120 Hz (doublet).

CH,

(MeO)3P-Rh
L
'H

9.12. Me are diastereotopic. AG* = 63 kJ mol"1.

9.13. NMe(CH2)2 exchange. AG* =66.5 kJ mol-1.
9.14. The -80°C AMM'XX' pattern collapses to AM2XX' at 40°C.
The N doublet does not change and remains sharp.
9.15. Restricted rotation of the phenyl. At 25°C, ortho H form a broad
hump, meta and para are sharp. At -60°C there are two ortho
and two para signals.
9.16. BBN 8 11, B1 8 5, B2-4 8 -29, B3 8 3, B5-10 8 -2, B6-9 8 17, B 78 8 -9.
9.17. Ha 8 7.55, Hb 8 6.7, Hc 8 6.1, Hd 8 6.8, He 8 4.6, H° 8 7.15, Hm
8 6.9, Hp 8 6.05. Restricted agostic phenyl rotation.
9.18. H1 8 -23.3, H2 8 -21.6, H3 8 -24.15, H4 8 -20.3.
9.19. H1 8 -9.9, H2 8 -10.4, H3 8 -7.3. AG* = 44 and 56 kJ mol-1.
9.20. PH. Doublet of triplets of doublets. 7(31P'H) = 310 Hz;
V(31P1H) = 5 Hz; VOHRuP'H) = 4 Hz.
Hydride at 8 -8.5. Doublet of triplets of doublets of doublets.
VOHRu'H) = 3 Hz; 2/(31P'H) (PPr'3) = 10 Hz; 2/(31P1H) (PHPh2)
=16 Hz.
Hydride at 8 -9.6. Doublet of triplets of doublets. ^('HRu'H)
= 3 Hz; 2/(31P1H) (PPr'3) = 10 Hz; 2J(31P1H) (PHPh2) = 43 Hz.

CHAPTER 11

11.1. 8|| = -84, 8± = +267, 8iso = + 151. Axially symmetric, so linear.

11.2 8n = 26, 822 - -28, 833 = -70, 8iso = -24.4.

www.pdfgrip.com
www.pdfgrip.com
Index

Absolute frequency 37f magnetization transfer 198f, Coupling 2, 45

Absorption mode 150 211 Coupling constant, / 45
Absorption signal 150, 163 multisite exchange 198, 208f interproton spin coupling
Activation parameters 204, 205 saturation transfer, see constants 46
Air turbine 132 magnetization transfer reduced coupling constant, K
D-Amygdalin 277 unequally populated two site 46
Analogue filters 140f exchange 194f relative sign of the coupling
Analogue to digital converter Chemical shift 4, 32, 48 constant 287
141, 146f, 170 anisotropy 120f through-space 45
Angular momentum 1, 7 contact shifts 27 Coupling pattern 47f
Anisotropic screening 42 diamagnetic screening 18f, 22 AB 65, 69f, 86
Arrow diagram 49f diamagnetic term ad 18 AX 71
Assisted Solvent-Induced Shifts, halogen dependence 22 A2X 65
ASIS 41 heavy-atom screening effect ABX 65, 77f
Asymmetry factor, nq 110 22 AMX 65
Attenuation 138 high field 34 AA'X 81
AB energy levels 72 high frequency 33f [AX]2 63, 66, 81, 84
AX energy levels 240 isotope shifts 24f AB2 76
low field 33f AX2 76
Biomedical NMR 335f, 350 low frequency 34 A3B2 65
Bloch-Siegert shift 167, 234, 236 paramagnetic term, ap 18f A3X2 65
Block diagram of a NMR paramagnetic screening 18, 22 A3M2X 65
spectrometer 13, 130 pseudo-contact shift 27 [AX9"]2 82
Blood flow 339f temperature effects 42, 134 analysis of first order
Bohr relation 5 Chemical shift imaging 342f multiplets 57f
Boltzmann distribution 7, 174, Chemical shift ranges satellites 61
13
198, 233, 236f C 268 second-order effects 69
Bond angle 46 'H93 sub-spectrum 61
Bond length 46 Grant and Paul rules 268 Credit cards 131
Brain tissue 339 Chemical shift references 37 Crystallization of amorphous
Broadening by unpaired external standardization 39, polyethylene 230
electrons 27, 108 42 Curly brackets 233
Brownian motion 45, 104, 105, Chemical shift scales 33f
109 delta scale, 8 34 Data size 155
Bulk susceptibility contribution tau scale, T 34 Debye theory of electric
42 chi scale, E 37f dispersion 105
Composite pulses 180f Decibel 138
Carvone 277 Contours plot 275, 276, 279 Decoupling 167f
CD rom 166 Cooley-Tukey algorithm 162, broad-band heteronuclear
Chemical Exchange 189f 163 decoupling 172, 174
Berry pseudo-rotation Correlation time difference spectroscopy 234f
mechanism 198 rotational correlation time, TC GARP 175
coalescence temperature 193f 104f heteronuclear decoupling 170f
equally populated two site spin-rotation correlation time, homonuclear decoupling 168f
exchange 191f, 368f TCP 117f saturation transfer 198

www.pdfgrip.com
398 Index

WALTZ 172, 174f Hubbard's equation 118 Magnetic resonance imaging,

Detection system 139 Human head 339 MRI 335f
Diastereotopism 66f Hybridization 46 Magnetic susceptibilities 39
Difference spectrum 200, 234f Hydrogen bonding 42 Magnetization 8f, 48, 101f
Diffusion 341f Hydrolysis of aluminium salts Magnetogyric ratio, y 1, 4f,
Digital filtering 145 227 107
Dihedral angle 46, 47 Mixing period 233
Dispersion mode 136, 150 Imaging Microscopy 346f Molecular mass 104f
Dispersion signal 150, 163 INEPT Pascal triangle 261 Moment of inertia, / 118
Double quantum transition Inhomogeneities in the magnetic Multiplicity 47
probablity 238 field 129
Dwell period 141 Integration 165 Neighbour anisotropy 22, 27,
Dynamic range 146 Intermediate frequency, IF 139 234
Inverse detection 304f Nephelauxetic effect 22
Electric Field Effect 23 Inversion-recovery method 175, Nuclear magnetic vectors 5
Electric field gradient, EFG 369f 249 Nuclear moment 7
Electromagnetic radiation 5 Nuclear Overhauser effect,
Electron cloud 17f Jeener 282 NOE 170, 172, 200, 237f
Electronegativity 22, 46 ^Irnax 238
Energy separation of the spin Karplus relationship 46f dynamic NOE 249f
state 5f no decoupling but full NOE
Energy states 5 Lanthanide and actinide 246
Ernst equation 157 compounds 29 NOE difference spectroscopy
Evolution period 233 Larmor precession 8, 10f, 14, 242f
Excess population 7 103, 106, 179, 360 heteronuclear nuclear
Exponential multiplication 158f Line broadening 27 Overhauser effect 244f
Exponential relaxation 147f Line intensities 48 Nuclear paramagnetism 7, 8
Extreme narrowing 106f Line shape 147 Nuclear precession frequency 9f
Eyring equation 204 Gaussian 160 Nuclear spin quantum number, /
Lorentzian 149, 153, 160, 1, 3f
Field gradient 133, 335f 165 Nyquist frequency 141, 162
Field-frequency lock 135 powder sample 381f
Floppy discs 131, 166 super-Gaussian 151, 153 Off-resonance decoupling 173
Flow 341f, 349f super-Gaussian decay 151, One-dimensional NMR
Folding 142 153 spectroscopy 233f
Fourier relationship 149 Line shape transformation 158
Fourier transform 12, 14, 150f, Lorentzian-Gaussian Pake powder pattern 381f
162f transformation 160f Paramagnetic molecules 27
Fourier transform (FT) sine-bell function 161f Paramagnetic relaxation 108f
spectrometer 13f Linewidth 104, 155 Paramagnetic transition metal
Free induction decay, FID 12, Liquid helium 130 ions 27
149f, 155 Longitudinal magnetization 102, Pascal's triangle 53f
Frequency domain 150, 155 116, 129 Periodic Table 4
Frequency synthesizer 137 Periodicity 19f
Magnetic field 5, 6 Permeability of a vacuum 107
Gyromagnetic ratio 1 drift of field 131 Phase 138, 145f
electromagnets 39 correction 163f
Hartmann Hahn matching field-frequency lock 136f CYCLOPS phase cycle 145f
condition 360 gradients 187f, 335f phase cycling 145
Heart pacemakers 131 hazard associated with the phase-sensitive detector 137,
Heisenberg uncertainty principle stray magnetic field 131 139f
191 permanent magnet 39 phase shift 138
Helmholtz double coils 9, 135 superconducting solenoid 12, Planck's constant 1
High-pressure NMR 39, 130f Powder pattern 381f
spectroscopy 219f Magnetic equivalence 49, 66 Precession 8, lOf, 167
Homogeneity of the magnetic Magnetic moment If, 5, 7 Probe 134f
field 39, 130f Magnetic records 166 Projection 278f

www.pdfgrip.com
 Index 399

Pulse angle 157 asymmetry factor, 110 T1SR117f

Pulse lengths 157f coupling to quadrupolar T2 102
Pulse sequences nuclei 53f, 116 T~2DD 106f
attached proton test, APT oblate 109 T2Q 110f
253f quadrupole coupling constant r2SC 121f
BIRD 186f, 294, 305, 312 115 Resonance frequency 3f
CAMELSPIN 302f quadrupolar electric field 2, Resonance spectroscopy 10
Carr-Purcell 179f, 278, 341 109f Ring current anisotropy 22f
CHESS 345, 352f quadrupole moment 2f, 111 Rotating frame 10, 167
Correlation SpectroscopY, quadrupolar relaxation 109f
COSY 282f, 345 prolate 109 Sample preparation 36, 134
COSY-45 265f Quantum number m 1 Satellites 61
COSY-90 283f Saturation 169
coupling with NOE 247 Radio-frequency field 9 Scalar relaxation 121f
DANTE 184f Reaction field 42 Screening 22, 34
DEPT 262f Receptivity 2f, 15 screening anisotropy 20, 22,
EXSY 296, 300f, 365 Relaxation 101 120f
HMBC 306f chemical shift anisotropy screening constant, a 18f, 27,
HMQC 306f relaxation 120f 32
HOESY 393f dipole-dipole relaxation 104f, screening tensor 20
HOHAHA 294f 238 Secondary isotope effects 24f
Heteronuclear COSY NMR longitudinal relaxation 102 Selective population transfer
spectroscopy 290f relaxation mechanisms 101f 236f
HSQC 312 quadrupolar relaxation 111f Shift reagents 29f
INADEQUATE 266f separation of relaxation chiral shift reagents 30, 32
INEPT 257f mechanisms 249f Shim coils 131
JMOD 253f spin-lattice relaxation 102 Shimming 132f
J-resolved 13C NMR spectrum spin-rotation relaxation 117 computer shimming 133
274f spin-spin relaxation 103 manual shimming 133
J-resolved 1H NMR spectrum transverse relaxation 103, Signal intensity 7
278f 129 Solid state NMR 355f
MLEV-17 294 Relaxation field 106 CP-MAS 360f
MREV-8 366 Relaxation rates, R{ and R7 CRAMPS 367f
NOE measurement 241 101f cross-polarization 360
NOE suppression 247 R{ 102 magic angle spinning, MAS
NOESY 296f RICSA 120f, 252 357, 372
NOESY-COSY spectra 312 RIDD 106f, 252 quadrupolar nuclei 386f
PRESS 343f, 351f RIQ lllf variable angle spinning, VAS
PENDANT 264f RISR 117f 373
Phase-sensitive COSY 289f #2 102 Solvent exchange 214, 222f
Refocussing pulse 182 R2DD 106f Solvent effects 40, 42
ROESY 302f R2Q lllf Spectral density 106
shaped 185 Relaxation times, T{ and T2 101f Spectral editing 265f
13
spin-lock pulse sequence C relaxation 124f Spectral interpretation 92
179 Measurement of T{ and T2 Spectrometer 129f
STEAM 343f 175f Spin 1, 2, 8
TOCSY 294f Measurement of Tl by Spin coupling pattern 45f
two-dimensional population inversion 102, simulated spectrum 77
INADEQUATE 296 175f multiple combinations of
WALTZ 294, 296f, 350 Measurement of Tlr and Spin- I > ^60
WATERGATE 342f Locking 179 deceptively simple triplet 80,
Pulse shape 185, 336 Measurement of T2 103, 178 82
Pulse width, PW 11, 13 T{ 102 effect of exchange 219
T1CSA 120f, 252 splitting due to groups of
Quadrature detection 137, 143f T1DD 106f, 252
Quadrature image 145 r1Q ii9f sub spectrum 77
Quadrupolar nuclei -isc 121f Spin locking 179

www.pdfgrip.com
400 Index

Spin states 1, 2, 5, 7 Through-space 27 Unpaired electron 27, 108

Spinning 12, 40, 132, 357f Time domain FID 141, 150,
side bands 133 151 Van der Waals interactions 42
speed 132 Time sharing 136, 170 Viscosity of the liquid 104
spinner 131 Topical NMR 350
Superheterodyne principle 139 Torsion angle 47 Water suppression 342f
Surface coil 335 Transition probability, W 238 Whole body imaging 337f
Transmitter 137f
Temperature measurement 203 Transverse magnetization 101 Viscosity of solvent 106
thermocouple 134 Two-dimensional NMR 273f
Temperature stability 36, 42, 134 cross-section 277, 281 Zero-filling 155
Three-dimensional NMR symmetrization 278f Zero quantum transition
spectroscopy 312 tilt 278f probability 238

www.pdfgrip.com