Gibb

STUDIES IN CHEMICAL PHYSICS
General Editor
A.D. Buckingham F.R.S., Professor of Chemistry,
University of Cambridge
Series Foreword
The field of science known as 'Chemical Physics' has
greatly expanded in recent years. It is an essential part
of both physics and chemistry and now impinges on
biology, crystallography, the science of materials and
even on astronomy. The aim of this series is to present
short, authoritative and readable books on different
topics in chemical physics at a level that is appreciated
by the non-specialist and yet is of prime interest to the
expert-in fact, the type of book that we all welcome
and enjoy.
I was grateful to be given the opportunity to help
plan this series, and warmly thank the authors and
publishers whose efforts have brought it into being.
A.D. Buckingham,
University Chemical Laboratory,
Cambridge, U.K.
Other titles in the series

Advanced Molecular Quantum Mechanics, R.E. Moss
Chemical Applications of Molecular Beam Scattering,
M.A.D. Fluendy and K.P. Lawley
Electronic Transitions and the High Pressure Chemistry
and Physics of Solids, H.G. Drickamer and C.W. Frank
Aqueous Dielectrics, J.B. Hasted
Statistical Thermodynamics, B.J.M. McClelland
Principles of
Mossbauer Spectroscopy
T.e. GIBB
Department of Inorganic and Structural Chemistry,
University of Leeds
SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

© T. C. Gibb 1976
Originally published by chapman and hall in 1976
Soficover reprint of the hardcover 1st edition 1976
Typeset by Mid-County Press

and printed in Great Britain by
Fletcher & Son Ltd, Norwich
ISBN 978-0-412-13960-4 ISBN 978-1-4899-3023-1 (eBook)

DOI 10.1007/978-1-4899-3023-1
All rights reserved. No part of

this book may be reprinted, or reproduced
or utilized in any form or by any electronic,
mechanical or other means, now known or hereafter
invented, including photocopying and recording,
or in any information storage or retrieval
system, without permission in writing
from the publisher
Preface
The emergence of M6ssbauer spectroscopy as an important experi-

mental technique for the study of solids has resulted in a wide range
of applications in chemistry, physics, metallurgy and biophysics.
This book is intended to summarize the elementary principles of
the technique at a level appropriate to the advanced student or
experienced chemist requiring a moderately comprehensive but
basically non-mathematical introduction. Thus the major part of
the book is concerned with the practical applications of Mossbauer
spectroscopy, using carefully selected examples to illustrate the
concepts. The references cited and the bibliography are intended
to provide a bridge to the main literature for those who subseouent-
ly require a deeper knowledge.
The text is complementary to the longer research monograph,
'Mossbauer Spectroscopy', which was written a few years ago in
co-authorship with Professor N.N. Greenwood, and to whom I am
deeply indebted for reading the preliminary draft of the present
volume. I also wish to thank my many colleagues over the past
ten years, and in particular Dr. R. Greatrex, for the many stimu-
lating discussions which we have had together. However my
greatest debt is to my wife, who not only had to tolerate my eccen-
tricities during the gestation period, but being a chemist herself
was also able to provide much useful criticism of the penultirna te
draft.
Leeds, May 1975 T.e. Gibb

Contents
Preface
1 The Mossbauer Effect 1

1.1 Resonant absorption and fluorescence 2
1.2 The Mossbauer effect 5
1.3 The Mossbauer spectrum 9
1.4 The Mossbauer spectrometer 13
1.5 Mossbauer isotopes 16
1.6 Computation of data 19
References 21
2 Hyperfine Interactions 22
2.1 The chemical isomer shift 23
2.2 Magnetic hyperfine interactions 28
2.3 Electric quadrupole interactions 30
2.4 Combined magnetic and quadrupole interactions 36
2.5 Relative line intensities 38
References 43
3 Molecular Structure 45
3.1 Iron carbonyls and derivatives 46
3.2 Geometrical isomerism in Fe and Sn compounds 54
3.3 Linkage isomerism in cyano-complexes of Fe 56
3.4 Conformations in organometallic compounds of Fe 60
Contents
3.5 Stereochemistry in tin compounds 63
3.6 Molecular iodine compounds 67
Appendix Quadrupole splitting in cis- and trans-isomers 69
References 71
4 Electronic Structure and Bonding:

Diamagnetic Compounds 73
4.1 Formal oxidation state 74
4.2 Iodine 77
4.3 Tellurium and antimony 84
4.4 Tin 86
4.5 Covalent iron compounds 93
References 100
5 Electronic Structure and Bonding:

Paramagnetic Compounds 102
5.1 Quadrupole interactions 102
5.2 Magnetic hyperfine interactions 109
5.3 Spin cross-over 118
5.4 Pressure effects 121
5.5 Second and third row transition elements 124
5.6 Lanthanides and actinides 129
References 133
6 Dynamic Effects 135

6.1 Second-order Doppler shift and recoilless fraction 135
6.2 The Goldanskii-Karyagin effect 143
6.3 Electron hopping and atomic diffusion 145
6.4 Paramagnetic relaxation 148
6.5 Superparamagnetism 156
References 157
7 Oxides and Related Systems 159

7.1 Stoichiometric spinels 160
7.2 Non-stoichiometric spinels 165
7.3 Exchange interactions in spinels 168
7.4 Rare-earth iron garnets 174
7.5 Transferred hyperfine interactions 179
References 181
Principles of Mossbauer Spectroscopy
8 Alloys and Intermeiallic Compounds 182
8.1 Disordered alloys 183
8.2 Intermetallic compounds 188
References 196
9 Analytical Applications 197

9.1 Chemical analysis 197
9.2 Silicate minerals 201
9.3 Surface chemistry 206
References 212
10 Impurity and Decay After-effect Studies 213

10.1 Impurity doping 214
10.2 Decay after-effects 221
References 232
11 Biological Systems 234

11.1 Haemoproteins 236
11.2 Ferredoxins 241
References 244
Bibliography 245
Observed Mossbauer Resonances 247
Index 248
CHAPTER ONE
The Mossbauer Effect
The study of recoilless nuclear resonant absorption or fluorescence

is more commonly known as Mossbauer spectroscopy. From its
first origins in 1957, it has grown rapidly to become one of the
most important research methods in solid-state physics and
chemistry.
Resonant nuclear processes had been looked for without success
for nearly thirty years before R.L. Mossbauer made his first acci-
dental observation of recoilless resonant absorption in 191Ir in
1957 [1]. He not only produced a theoretical explanation of the
effect which now bears his name, but also devised an elegant ex-
periment which today remains almost unmodified as the primary
technique of Mossbauer spectroscopy.
The Mossbauer effect is of fundamental importance in that it
provides a means of measuring some of the comparatively weak
interactions between the nucleus and the surrounding electrons.
Although the effect is only observed in the solid state, it is precise-
ly in this area that some of the most exciting advances in chemi-
stry and physics are being made. Because it is specific to a parti-
cular atomic nucleus, such problems as the electronic structure of
impurity atoms in alloys, the after-effects of nuclear decay, and
the nature of the active-centres in iron-bearing proteins are just a
few of the diverse and many applications.
2 Principles of Mdssbauer Spectroscopy
1.1 Resonant absorption and fluorescence
Before delving into the details of the subject, it is worthwhile con-
sidering the historical perspective of what has come to be consider-
ed as a discovery of prime importance.
Atomic resonant fluorescence was predicted and discovered just
after the turn of the century, and within a few years the under-
lying theory had been developed. From a simplified viewpoint, an
atom in an excited electronic state can decay to its ground state
by the emission of a photon to carry off the excess energy. This
photon can then be absorbed by a second atom of the same kind
by electronic excitation. Subsequent de-excitation re-emits the
photon, but not necessarily in the initial direction so that scat-
tering or resonant fluorescence occurs. Thus if the monochromatic
yellow light from a sodium lamp is collimated and passed through
a glass vessel containing sodium vapour, one would expect to see a
yellow glow as the incident beam is scattered by resonant fluore-
scence.
A close parallel can be drawn between atomic and nuclear reso-
nant absorption. The primary decay of the majority of radioactive
nuclides produces a daughter nucleus which is in a highly excited
state. The latter then de-excites by emitting a series of r-ray pho-
tons until by one or more routes, depending on the complexity
of the 'Y-cascade, it reaches a stable ground state. This is clearly
analogous to electronic de-excitation, the main difference being in
the much higher energies involved in nuclear transitions. It was
recognized in the 1920's that it should be possible to use the 'Y-ray
emitted during a transition to a nuclear ground-state to excite a
second stable nucleus of the same isotope, thus giving rise to nuc-
lear resonant absorption and fluorescence.
The first experiments to detect these resonant processes by
Kuhn in 1929 [2] were a failure, although it was already recog-
nized that the nuclear recoil and Doppler broadening effects (to
be discussed shortly) were probably responsible. Continuing
attempts to observe nuclear resonant absorption [3] were inspired
by the realization that the emitted 'Y-rays should be an unusually
good source of monochromatic radiation. This can easily be
shown from the Heisenberg uncertainty principle. The ground
state of the nucleus has an infinite lifetime and therefore there is
no uncertainty in its energy. The uncertainty in the lifetime of
The Mossbauer Effect 3
the excited state is given by its mean life T, and the uncertainty
in its energy is given by the width of the statistical energy distri-
bution at half-height r. They are related by
0.1 )
where h (=2rrtt) is Planck's constant. T is related to the more fami-

liar half-life of the state by T = In 2 x t 1/2. If r is given in electron-
volts 0 eV = 1.60219 x 10- 19 J and is equivalent to 96.49 kJ
mol-I) and t1/2 in seconds, then
r = 4.562 x 10-I6/t1/2 0.2)
F or a typical nuclear excited-state half-life of t 1/2 = 10- 7 s, r =

4.562 x 10-9 eV. If the energy of the excited state is 45.62 keY,
the emitted 'Y-ray will have an intrinsic resolution of 1 part in 1013.
It should be borne in mind that the maximum resolution obtained
in atomic line spectra is only about I in 10 8 .
The nucleus is normally considered to be independent of the
chemical state of the atom because of the great disparity between
nuclear and chemical energies. However, the 'Y-ray energy is so
sharply defined that r is smaller in magnitude than some of the
interactions of the nucleus with its chemical environment. The
typical magnitudes of some of the energies involved are indicated in
Table 1.1.
Unfortunately there are several mechanisms which can degrade
the energy of the emitted 'Y-ray, particularly the effects of the
nuclear recoil and thermal energy [3]. Consider an isolated
nucleus of mass M with an excited state level at an energy E and
moving with a velocity V along the direction in which the 'Y-ray
is to be emitted (the components of motion perpendicular to this
direction remain unaffected by the decay and may be ignored).
The energy above the ground state at rest is (E +-tMV2). When a
'Y-ray of energy E'Y is emitted, the nucleus recoils and has a new
v-e1ocity V + v (which is a vector sum in that Vand v may be
opposed) and a total energy of ~ M( V + v)2. By conservation of
energy
E +lMV2
2
= E 'Y +lM(V
2
+ v)2 (1.3)
4 Principles of M6ssbauer Spectroscopy
Table 1.1 Typical energies of nuclear and chemical interactions
Mossbauer 'Y-ray energies (E'Y) 1OL I0 7 kJ mol- 1

Chemical bonds and lattice energies 10 2-10 3 kJ mol- 1
Electronic transitions SO-SOO kJ mol- 1
Molecular vibrations S-SO kJ mol- 1
Lattice vibrations O.S-S kJ mol- 1
Nuclear recoil and Doppler energies (ER, ED) 10- 2-1 kJ mol- 1
Nuclear quadrupole coupling constants <10- 3 kJ mol- 1
Nuclear Zeeman splittings <10- 3 kJ mol- 1
Heisenberg linewidths (J') 10- 7-10- 4 kJ mol- 1
1 eV = 1.60219 x 10- 19 J and is equivalent to 96.49 kJ mol- 1
so that the actual energy of the photon emitted is given by
E'Y =E -1..Mv
2
2- Mv V
0.4)
The 'Y-ray is thus deficient in energy by a recoil kinetic energy

(ER =tMv2) which is independent of the initial velocity V, and by
a thermal or Doppler energy (ED =Mv V) which depends on Vand
can therefore be positive or negative.
Momentum must also be conserved in the emission process. The
momentum of the photon is E'Y/c where c is the velocity of light,
so that
MV=M(V+v) +E'Y/c 0.5)
and the recoil momentum is Mv = -E'Y/c. Hence the recoil energy

is given by
(1.6)
and depends on the mass of the nucleus and the energy of the 'Y-
ray. However, the Doppler energy ED is dependent on the thermal
motion of the nucleus, and will therefore have a distribution of
values which is temperature dependent. A mean value, ED, can be
defined which is related to the mean kinetic energy per translation-
al degree of freedom, Ek =tkT, by
- _/~
ED = 2v (EkE R) = E'Y
j2Ek
Mc 2 0.7)
where k is Boltzmann's constant and T is the absolute temperature.

As a result, the statistical distribution in energy of the emitted
'Y-rays is displaced from the true excited-state energy by -ER and
broadened by ED into a Gaussian distribution of width 2E D. The
distribution for absorption has the same shape but is displaced by
+E R . This is illustrated schematically in Fig.l.l, and the order of
magnitude of ER and ED is indicated in Table 1.1. Nuclear reso-
nant absorption (which in practice can only be observed following
an emission) will only have a significant probability if the emission
and absorption energy distributions overlap strongly. The recoil
and thermal broadening clearly prevent this. It is possible to com-
pensate partially for ER by moving the emitter towards the observ-
er at a very large (supersonic) velocity, or to increase the Doppler
broadening by raising the temperature. Both result in only a
marginal increase in the resonant overlap. Furthermore, the intrin-
sic resolution in the energy of the photon has been degraded to
about 1 in 10 7•
It should be noted that these equations are not peculiar to nuc-
lear processes and apply equally well to atomic absorption. How-
ever, in the latter case the low energy of the transition results in
the recoil energy being significantly less than the thermal broaden-
ing. Consequently the overlap for absorption is large and the
effects of recoil may be ignored.
1.2 The Mossbauer effect

The Mossbauer effect is unique in that it provides a means of eli-
minating the destructive effects of the recoil and thermal energies.
The key to the problem lies in the behaviour of the recoiling nuc-
leus when it is no longer isolated (as is implicit in the preceeding
discussion), but instead is fixed in a crystal lattice. As can be seen
from Table 1.1, the recoil energy is much less than the chemical
binding energy, but is similar in magnitude to the lattice-vibration
phonon energies. If the recoil energy is transferred directly to
6 Principles of Mossbauer Spectroscopy
(a)
-Ey
-Ey
Fig. 1.1 The statistical distribution in the -y-ray energy for (a) emission,
(b) absorption, and (c) the resonant overlap for successive emission and ab-
sorption.
vibrational energy, then the -y-ray energy is still degraded. How-

ever, the phonon energies are quantized, and the recoil energy can
only be transferred to the lattice if it corresponds closely to an
allowed quantum jump.
The simplest mathematical treatment is one in which the vibra-
tional characteristics correspond to an Einstein solid with one vi-
brational frequency, w. Transfer of energy to the lattice can only
take place in integral multiples of 1lw (0, ±lIw, ±211w, ... ). If the
recoil energy ER is less than nw, then either zero or one unit (nw)
of vibrational energy may be transferred to the lattice. It has been
The Mossbauer Ellect 7
shown by Lipkin [4] that when many emission processes are con-
sidered, the average energy transferred per event must exactly
equal E R . If a fraction, I, of emission events result in no transfer
of energy to the lattice (zero-phonon transitions), and a fraction
(1 - f) transfer one phonon of energy tLw, then
ER = (1 - f)tLw
or
(1.8)
In a zero-phonon emission, the whole crystal rather than a

single nucleus recoils. Equations (1.6) and (1.7) for ER and ED
contain the reciprocal mass, 11M. If the mass M is increased to
that of a crystallite containing perhaps 10 15 atoms, then both the
recoil energy and the Doppler broadening become very small and
much less than r. Hence zero-phonon transitions are referred to
as recoilless. In any real solid there will be a wide range of lattice
frequencies, but fortunately it is very difficult to excite the low
frequencies and there is still a significant fraction of nuclear events
in which the ,),-ray energy is not degraded.
To summarize, the Mossbauer effect is the resonant emission or
absorption of ,),-rays in a solid matrix without degradation by
recoil or thermal broadening, and it gives an energy distribution
dictated by the Heisenberg uncertainty principle.
The parameter I is known as the recoilless or recoil-free frac-
tion, and to increase the relative strength of the recoilless resonant
process it is important that I be as large as possible. On a more
quantitative theory [5, 6] it is possible to relate I to the vibration-
al properties of the crystal lattice by
1= exp (_ _E-,-;_(x-:::2,--») (1.9)

. (tLc)2
where <x 2) is the mean-square vibrational amplitude of the nucleus

in the direction of the ')'-ray. From the form of the exponent, I
will only be large for a tightly bound atom with a small mean-square
displacement, and for a small value of the ,),-ray energy, E'Y. For
the latter reason the highest energy for which a Mossbauer reson-
ance is known is 187 keVin 1900s.
The precise form of <x 2 ) depends on the vibrational properties
of the lattice. In any real solid these are usually complex and ade-
quate models are not available. However, it is instructive to con-
sider the behaviour predicted [6] by the Debye model, which em-
bodies a continuum of vibrational frequencies in the harmonic
oscillator approximation following the distribution N(w) = con-
stant x w 2, up to a maximum value of wo. A characteristic tem-
perature 80 is defined by 1!wo = k80 and is known as the Debye
temperature. The final expression obtained is
f= exp {_ 6E~ [~+ (~) 2

k80 4 80
J 001 T xdx ]}
0 eX - 1
0.10)
From the form of the equation it can be seen that f is large when
80 is large (a strong lattice) and when the temperature T is small.
This decrease in f with rise in temperature follows in general from
equation (1.9), in that thermal energy will increase the amplitude
of vibration of the nucleus, but the precise form of the tempera-
ture dependence can only be predicted using a model for the lat-
tice vibrations. This problem is discussed in more detail in
Chapter 6.
It ·should be noted that although the recoilless fraction can be
increased by lowering the temperature there is a limiting value at
absolute zero which is still less than unity. For instance, from
equation (1.1 0) when T = 0
(1.11)
which still depends on ER and 00.

In summary, recoilless emission or absorption is optimized for
a low-energy r-ray with the nucleus strongly bound in a crystal
lattice at low temperature.
1.3 The Mossbauer spectrum
Having detennined the conditions under which recoilless resonant
absorption can occur, it is possible to devise a simple experiment
to demonstrate the effect. A solid matrix containing the excited
nuclei of a suitable isotope is used as the source of -y-rays. It is
placed alongside a second matrix of identical material containing
the same isotope in its ground state, which becomes the absorber.
The intensity of the -y-radiation transmitted through the absorber
is measured as a function of temperature. If there is no resonant
absorption, the counting rate is independent of temperature. If
resonant absorption occurs, then there should be a decrease in the
transmission as the temperature is lowered, and the recoilless frac-
tion increases. The resonantly absorbed -y-rays are effectively lost
from the transmitted beam of radiation. Some are re-emitted in
a random direction (Le. resonant fluorescence), while others are
lost by an internal conversion process during decay of the excited
level (see p.ll).
Although this was the experiment first used by R. L. Mossbauer,
it is not particularly useful. The method which he also pioneered
and has now been adopted exclusively is far more subtle. The
definition of the energy of the -y-ray emitted by the source in a
recoilless Mossbauer event is about 1 in 10 12 _10 13 , and corresponds
conveniently to the Doppler energy shifts produced by small move-
ments. If the source and absorber are in relative motion with a
velocity v. then the effective value of E'Y 'seen' by the absorber
differs from the true energy by a small Doppler shift energy of
€ = (v/c)E'Y. In the event that v is zero, the emission and absorp-
tion profiles completely overlap and the absorption is at a maxi-
mum. Any increase or decrease in v can only decrease the overlap.
If v is very large (of either sign) there will be no overlap and no
absorption. It therefore follows that a record of transmission as a
function of the velocity v will show an 'absorption spectrum '.
This is illustrated schematically in Fig.1.2. By convention a posi-
tive velocity is taken to be a closing velocity as this represents an
increase in the apparent energy of the photon arriving at the
absorber.
The lineshape of the absorption is derived simply from the
Heisenberg uncertainty in the energy. The distribution of the
recoilless source radia tion is given by
Transmission
as
recorded Radiation
incident
on
absorber
_ _ NonIsonan~
Radiation { - - - --
lost in Non -resonant
absorber by
non-resonant
- - - - - ---t---
Resonant
absorption
o +
Relative Ooppler shift E" fEy
Fig. 1.2 A schematic representation of a Mossbauer transmission spectrum

produced by Doppler scanning.
N(E) dE =-
Ir dE
(1.12)
21T (E -Ky + (r(2)2
where N(E) is the probability of the energy being between E and
E + de, f is the recoilless fraction of the source and r is the Heisen-
berg width. This function which is usually referred to as the
Lorentzian distribution has a maximum value at E = E'"(" The cross-
section for resonant absorption, aCe), is similarly expressed [6, 7]
as
(r/2)2
aCe) = ao (1.13)
(E - Ky + (r/2)2
where ao is a nuclear constant called the absorption cross-section
given by
(1.14)
where Ie and Ig are the nuclear spin quantum numbers of the excit-
ed and ground states and ex is the internal conversion coefficient
of the -y-ray [Note: -y-emission does not always lead to an observ-
able -y-ray. In a proportion of events an atomic s-electron is eject-
ed instead in a process known as internal conversion. ex is defined
as the ratio of the number of conversion electrons to the number
of -y-ray photons emitted. Values for ao are often quoted in units
of barns (1 bam = 10-24 cm 2 = 10- 28 m 2)]. For ao to be large,
both E'Y and ex should be small.
In any real absorber of finite thickness the resonant absorption
is in direct competition with other non-resonant processes such as
Compton scattering. A general evaluation of the integrated absorp-
tion intensity or 'transmission integral' is extremely difficult, but
to a first approximation the basic behaviour is comparatively
simple [8]. The recorded absorption spectrum for infinitely thin
sources and absorbers has a Lorentzian distribution with a width
at half-height of rr = 2r. For an absorber of finite thickness con-
taining t a resonant atoms per unit area of cross-section and with a
recoilless fraction of faJ an 'effective thickness' is usually defined
by the dimensionless parameter Ta = faaot a. The absorption shape
approximates very closely to Lorentzian but with a broadened
wid th r r given by
( 1.15)
provided that Ta < 5; i.e. the line broadens with an initially linear
dependence on thickness. Self-resonance in the source matrix
results in an additional line broadening which is approximately
independent of absorber thickness.
The probability that any given photon will be resonantly absorb-
ed can be enhanced by increasing the thickness Ta. However, the
intensity of the incident -y-ray beam is also strongly attenuated by
conventional non-resonant absorption processes, and this places
an upper limit on the value of Ta which can be used effectively.
The absorption may be defined in terms of the transmitted inten-
sity at the resonant maximum (10) and the transmitted intensity
at a large Doppler velocity where the absorption is zero (I ..) by
(1.16)
It has been evaluated [8] to be
(1.17)
where f is the recoilless fraction of the source and J o(x) is a zero-

order Bessel function. The value of Alf as a function of Ta is plot-
ted in Fig. 1.3, and illustrates how the absorption shows a satura-
tion behaviour with increasing thickness. The effects of non-
resonant attenuation combined with more practical problems con-
cerning background radiation levels make it advisable to use the
smallest value of Ta which gives an adequate absorption. In prac-
tice there is usually an optimum range of absorber thickness for
each isotope which is quickly derived from experience.
One of the unfortunate problems which arises is that it is dif-
ficult to determine absolute values for the recoilless fractions f
and fa directly from the absorption intensity without tedious and
difficult experimentation. However, in the majority of applica-
tions accurate values for these parameters are not required.
1·0.--------------------------,
0·8
N
i::
~0·6
...,
.
N
i:::
I
I
";;" 0·4
....
.......
~
4 6 8 10
Thickness (Tl
Fig. 1.3 The function A If = 1 - e- TI2 J oUTI2) showing how the absorption,
A. shows a saturation behaviour with increasing thickness, T.
1.4 The Mossbauer spectrometer
The Mossbauer spectrum is a record of the transmission of reso-
nant r-rays through an absorber as a function of the Doppler velo-
city with respect to the source. It is therefore quite simply a
record of transmission as a function of the energy of the incident
radiation. The major difference from other forms of transmission
spectroscopy is that the hazard to health in using high-energy radia-
tion, combined with the high cost of producing the radioisotopes,
necessitates the use of relatively low photon flux-densities. In con-
sequence, much longer counting times are required to achieve a
given resolution (hours rather than minutes). The statistics of the
random counting predicts that the standard deviation in N register-
ed events is VN: thus the standard deviation in 10 000 counts is
1% and in 1 000000 counts is 0.1 % of the total count. A prospec-
tive gain in resolution must be balanced against the longer experi-
mental time involved, which brings with it the problem of long-
term reproducibility in the measuring equipment. To double the
resolution requires four times the counting time, and makes it
imperative to obtain as large a percentage absorption as possible.
The measurement of a Mossbauer spectrum is almost exclusively
carried out by repetitively scanning the whole velocity range re-
quired, thereby accumulating the whole spectrum simultaneously,
and allowing continuous monitoring of the resolution. This can
be achieved electromechanically, and a typical modem Mossbauer
spectrometer is illustrated schematically in Fig.lA. The following
description is a brief outline of the major principles involved. For
a more detailed account of instrumentation and experimental
methods the reader is referred to a recent review by Kalvius and
Kankeleit [9] .
The major component is a device known as a multichannel
analyser which can store an accumulated total of r-counts, using
binary memory storage like a computer, in one of several hundred
individual registers known as channels. Each channel is held open
in tum for a short time interval of fixed length, which is derived
from a very stable constant-frequency clock device. Any r-counts
registered by the detection system during that time interval are
added to the accumulated total already stored in the channel. The
sequential accessing of the channels is completed in about 1/20
of a second, and is repeated ad infinitum.
r-------,
'---r---,;---:-r---' :s,,~ ,
I AbsorberI
I I
I I
I Cryostat I
' - _ _ _ _ _ .J Pulse
Waveform
generator amplifier
Constant Multichannel
frequency Discriminator
clock Address analyser
control
Data output to
1
C.R.T. display
Typewriter
Tape punch
Magnetic tape
x-v plotter
Fig. 1.4 A schematic arrangement for a Mossbauer spectrometer.
The timing pulses from the clock may also be used to synchro-
nize a voltage waveform, which is used as a command signal to the
servo-amplifier controlling an electromechanical vibrator. The
latter moves the source relative to the absorber. A waveform, with
a voltage increasing linearly with time, imparts a motion with a
constant acceleration in which the drive shaft and the source
spend equal time intervals at each velocity increment. The multi-
channel analyser and the drive are synchronized so that the velo-
city changes linearly from -v to +v with increasing channel num-
ber. In this way the source is always moving at the same velocity
when a given channel is open.
The 'Y-ray detection system is conventional. A scintillation
counter, gas proportional counter, or lithium-drifted germanium
detector may be used according to circumstances [10]. The pulses
from the detector are amplified, passed through a discriminator
which rejects most of the non-resonant background radiation, and
finally are fed to the open channel address.
The geometric arrangement of the source, absorber and detec-
tor is important. A 'Y-ray emitted along the axis of motion of the
source has a Doppler shift relative to the absorber of E"{v/ c, but
any 'Y-ray which travels to the detector along a path at an angle 0
to this axis has an effective Doppler shift of only E"{ v cosO/c. This
'cosine-effect' of solid-angle will therefore cause a spread in the
apparent Doppler energy of the 'Y-rays and hence line broadening
[11]. The effect can be reduced by maintaining an adequate
separation between the source and detector, or by collimation.
The velocity range required to completely encompass the absorp-
tion is usually less than ± 1a mm s-l, so that the overall displace-
ment of the source during a scan is barely discernible. Absolute
calibration of the drive is not easy, but can be achieved by mount-
ing a diffraction grating or mirror on the shaft as part of an inter-
ferometer. For most purposes it is more convenient to rely on an
internal standard. As will be seen in the next chapter, it is possible
for an absorber to show more than one resolved absorption line.
If such a material is chemically reproducible and stable, and the
lines are measured on an instrument with absolute calibration, it
can then be used in other laboratories to establish the velocity
amplitude and linearity of un calibrated instruments. For example,
it is common practice to use a magnetic a-iron foil which has six
lines as a calibrant for 57Fe work, and to quote all measurements
of velocity relative to the centroid of this reference spectrum.
Very few Mossbauer resonances are easy to record at room tem-
perature, the main exceptions being 57Fe and 119Sn, and in many
instances it is necessary to cool at least the absorber, and some-
times the source as well, to increase the recoilless fractions.
Because the source is moving, this is not without its difficulties.
There is also the possibility that some of the hyperfine interactions
(to be described in Chapter 2) are temperature dependent. For
this reason alone, low-temperature measurements have become
popular [9]. Numerous commercial cryostats are now available,
the commonly used refrigerants being liquid nitrogen for tempera-
tures down to 78 K and liquid helium down to 4.2 K. Cryostats
with a fully-variable temperature control are also used, particularly
in the study of phase transitions. In some applications it is desir-
able to apply a large external magnetic field to the absorber, and
superconducting magnet installations are available which can pro-
duce magnetic flux densities of up to lOT.
Although this discussion has been weighted in favour of using a
16 Principles o! Mossbauer Spectroscopy
transmission experiment to observe resonant absorption, it is also
possible to detect resonant fluorescence by a scattering experiment.
The resonant absorption event may be recorded by detecting the
scattered 'Y-ray on de-excitation (alternatively, if the internal con-
version coefficient IX is large, the conversion-electron or X-ray
produced may be registered). This has the advantage that non-
resonant radiation from the source is not recorded, and the major
remaining contributions to the radiation background are scattered
higher-energy 'Y-rays and non-resonant Rayleigh scattering. The
main disadvantages are that the intensity of the scattered radiation
is very weak, and the large solid geometry requirement for a high
count rate results in broadening of the Mossbauer line. A further
problem, which may arise if scattered 'Y-rays are counted, is the
possibility of anomalous effects dependent on the scattering angle.
These are due to interference between resonant Mossbauer scatter-
ing and the non-resonant Rayleigh scattering [12] and to diffrac-
tion [13].
The principal application of scattering experiments is in the stu-
dy of surface phenomena, and examples of this are given in Chap-
ter 9.
1.S Mossbauer isotopes

Several requirements can be formulated which must be fulfilled
if there is to be an easily observable Mossbauer resonance:
(1) The energy of the 'Y-ray must be between 10 and 150 ke V,
preferably less than 50 keY, because the recoilless fraction! and
resonant cross-section ao both decrease as E"{ increases. This is
the main reason why no resonances are known for isotopes lighter
than 40K. The 'Y-transitions in light nuclei are usually very energetic
(2) The half-life of the first excited state which determines r
should be between about I and 100 ns. If t1/2 is very long, then r
is so narrow that mechanical vibrations destroy the resonance con-
dition, and if t1/2 is very short then r will probably be so broad as
to obscure any useful hyperfine effects.
(3) The internal conversion coefficient IX should be small « 10)
so that there is a good probability of detecting the 'Y-ray.
(4) A long-lived precursor should exist which can populate the
required excited level. This usually means an isotope which decays
by ~-decay, electron capture, or isomeric transition. Although
some experiments have been done with energetic nuclear reactions
in situ such as Coulombic excitation or (d,p) reactions, these
methods are not applicable to routine measurements.
(S) The ground-state isotope should be stable and have a high
natural abundance so that isotopic enrichment of absorbers is
unnecessary.
Despite these restrictions, a Mossbauer resonance has been
recorded in 100 transitions of 83 different isotopes in 44 elements,
the majority of the heavy elements in fact. These are listed at the
end of the book. However, many of these resonances can only
be recorded with difficulty at the present time. Some of the more
useful resonances, many of which are illustrated with examples in
later chapters, are listed in Table 1.2. Some of their nuclear pro-
perties are also given, including the nuclear spin states Ig and Ie
and their parity, and the internal conversion coefficient ex.
The most popular Mossbauer resonance is undoubtedly the
l4.4l-keV 'Y-transition in 57Fe, and the decay scheme for the 57Co
parent is shown in Fig.1.S, together with the decay schemes for the
119Sn, 129 1 and 1931r resonances. The conveniently long-lived 57Co
nucleus decays by electron-capture with high efficiency to the
136.32-keV level of 57Fe, from whence 8S% of the decays pro-
ceed to the 14.4I-keV level. The lifetime of this state is 99.3 ns,
giving a Heisenberg width of rr = 0.192 mm s-1 in the Mossbauer
resonance. The large value of ex for this level (8.17) unfortunately
reduces the source efficiency to less than 10%, but the 'Y-ray is
well resolved from the two higher energy 'Y-rays and the 7-keV
X-rays produced by internal conversion. Although the natural
abundance of 57Fe is only 2.19%, the absorption cross-section of
ao = 2.S7 x 10- 18 cm 2 is unusually large, and results in a satisfac-
tory resonance at room temperature. Metals and refractory
materials can give an acceptable absorption even above 1000 K,
although organometallic compounds with a low effective Debye
temperature are more effective when cooled.
One of the most important experimental aspects is in the choice
ofa source matrix. It is very desirable to have a high recoilless
fraction and a single emission line unbroadened by hyperfine
interactions. The best choice of host matrix is usually a high-
melting metal or a refractory oxide of cubic structure; for example
the most popular host for 119mSn is the oxide BaSn03, and 57Co is
usually diffused into a metal, such as palladium, platinum or rho-
Table 1.2 Nuclear parameters for selected Mossbauer transitions
Natural
rrl abund-
Isotope E.y/keV (mm s-1) Ig Ie a ance (%) Nuclear decay*
.---~~-
57Fe 14.41 0.192 1/2- 3/2- 8.17 2.17 57Co(EC

270 d)
61Ni 67.40 0.78 3/2- 5/2- 0.12 1.25 61Co(~-99 m)
99Ru 90 0.147 5/2+ 3/2+ 12.63 99Rh(EC 16 d)
119Sn 23.87 0.626 1/2+ 3/2+ 5.12 8.58 119mS n (IT
250 d)
121Sb 37.15 2.l 5/2+ 7/2+ -10 57.25 121mSn(~-
76 y)
125Te 35.48 5.02 1/2+ 3/2+ 12.7 6.99 1251(EC 60 d)
1271 57.60 2.54 5/2+ 7/2+ 3.70 100 127mTe
(~-109 d)
1291 27.72 0.59 7/2+ 5/2+ 5.3 nil 129mTe
W-33 d)
129Xe 39.58 6.85 1/2+ 3/2+ 11.8 26.44 129I(~-
1.7 x 107 y)
149S m 22.5 1.60 7/2- 5/2- -12 13.9 149Eu(EC
106 d)
151Eu 21.6 1.44 5/2+ 7/2+ 29 47.8 151Gd(EC
120 d)
161Dy 25.65 0.37 5/2+ 5/2- -2.5 18.88 161Tb(~-
6.9 d)
169Tm 8.40 9.3 1/2+ 3/2+ 220 100 169Er(~-
9.4 d)
182W 100.10 2.00 0+ 2+ 3.2 26.4 182Ta(~-
115 d)
1890s 69.59 2.41 3/2- 5/2- 8.2 16.1 189Ir(EC
13.3 d)
193Ir 73.0 0.60 3/2+ 1/2+ -6 61.5 1930s(~-
31 h)
197 Au 77.34 1.87 3/2+ 1/2+ 4.0 100 197PtW-18 h)
237Np 59.54 0.067 5/2+ 5/2- 1.06 nil 241 Am
(a458 y)
*EC =electron capture, p- =beta-decay, IT = isomeric transition, a = alpha-decay
dium. However, these were chosen as the result of experience,

and the development of a good source is of prime importance in
studying a new resonance.
In some instances where the absorption is too weak, a substan-
tial improvement can be obtained by using isotopic enrichment.
The Mdssbauer Effect 19
~iCo
~- ---r--270d
EC (99.84 % )
119mSn
50
250d 89.54
IT
2H7keV
t-'-t---L-...--- 14'41 keY
t- ----''---1..__ 0
, 0
t + -----"'---
119s
50 n
129mTe 193 05
33 d _ _52~_~. 76
70m-:-=-:!--.-
" - - . - - - 138·9
- - i - - - 80·2 IT 12d
'---rrrH-If- 487
'--t--..-- 73-0 keY
----..,,.-LH-+t+- 278
1'-r''+--':1:±Ll1- 2H keY
'-"-L......l~r- 0 '--''--....IL..--O
~-
1-7 x 107y
Fig. 1.5 The nuclear decay schemes for the Mossbauer resonances in
57Fe, 119S n , 1291 and 1931r.
Thus 57Fe has an abundance of only 2.17% in natural iron, but

material enriched to 90% can be used in compounds with a very
low total iron content (eg. dilute iron alloys or haeme-proteins) to
obtain an adequate cross-section for absorption.
1.6 Computation of data

The characteristic digital data comprising a Mossbauer spectrum
are in an ideal form for statistical calculations by digital computer.
For this reason the multichannel analyser is usually equipped to
output data in a form suitable for input to one of the standard
computer peripherals.
Each spectrum comprises N digital values, Yj, of the number of
'Y-counts registered at velocities Xi. The statistics of spontaneous
radioactive decay processes follow a Poisson distribution [14].
If the mean value of the number of counts recorded in a given time
is Y, then the standard deviation in Y is"; Y. In its simplest form
the Mossbauer spectrum comprises a single absorption line which
has a Lorentzian shape, and can therefore be specified completely
by four parameters; the linewidth, the line position, the intensity
of the absorption, and the baseline count for zero absorption.
These parameters may always be found by visual inspection, but
more precise values, together with their standard deviations, are
obtained by computing a least-squares fit to the data. Appropriate
starting values of the n parameters are 'guessed' and used to cal-
culate the 'theoretical' value of data point i to be Ai(n). The dis-
crepancy between the observed data and the assumed model can
be represented by the weighted sum of the squares of the dif-
ferences
(1.18)
This expression can be used as a 'goodness-of-fit' function and

minimized by standard mathematical routines to give the best
values of the n variables. The function F has N-n degrees of free-
dom and the properties of a chi-squared distribution at the mini-
mum [14] ; it may be related to the pro ba bility that the theoreti-
cal model is a valid description of the data, and in conjunction
with statistical tables [15] its value indicates the degree of confi-
dence to be placed on the final answers. For example, a computed
fit to data with 400 degrees of freedom is within the 25-75% con-
fidence limits of the X 2 distribution if 380 < F < 419, and within
the 5-95% limits if 355 < F < 448. A value outside these limits
is generally indicative of either faulty data or an inappropriate
theoretical model.
The most commonly used theoretical function is a summation
of Lorentzian lines, but the same basic principles are valid for any
function which correctly describes the data. However, one word
of caution must be given. A good fit to data is not unambiguous
proof that the theoretical function is the correct one. In a com-
plicated spectrum it is quite feasible to fit a function which has no
physical significance. The final data analysis must be compatible
with other scientific evidence, and inevitably there will be instances
where it is not possible to distinguish between alternative hypo-
theses.
References
[1] Mossbauer,R.L.(1958)Z.Physik, 151, 124.
[2] Kuhn, W. (1929) Phil. Mag., 8,625.
[3] Metzger, F. R. (1959) Progr. Nuclear Phys., 7,54.
[4] Lipkin, H. J. (1960) Ann. Phys., 9,332.
[5] Petzold, J. (1961) Z. Physik, 163, 71; Wisscher, W. M. (1960) Ann.
Phys., 9, 194.
[6] Frauenfelder, H. (I962) The Mossbauer Effect, W. A. Benjamin Inc.,
N.Y.
[7] Jackson, J. D. (1955) Can. J. Phys., 33, 575.
[8] Margulies, S. and Ehrman, J. R. (1961) Nuclear Instr. Methods, 12, 131.
[9] Kalvius, G. M. and Kankeleit, E. (1972) Recent improvements in in-
strumentation and methods of Mossbauer spectroscopy. In Mossbauer
Spectroscopy and its Applications, International Atomic Energy
Agency, Vienna, p.9.
[10] (1965) Alpha-, beta-, and gamma-ray spectroscopy VoLl, ed. K. Sieg-
bahn, North-Holland, Amsterdam.
[11] Riesenman, R., Steger, J. and Kostiner, E. (1969) Nuclear Instr.
Methods, 72, 109.
[12] Black, P. J., Longworth, G. and O'Connor, D. A. (1964) Proc. Phys.
Soc., 83,925.
[13] Black, P. J. and Duerdoth, I. P. (1965) Proc. Phys. Soc., 84,169.
[14] Mulholland, H. and Jones, C. R. (1968) Fundamentals of Statistics,
Butterworths, London.
[15] Beyer, W. H. (editor), (1968) Handbook of Tables for Probability and
Statistics, 2nd edition. The Chemical Rl.!1.Jber Co., Cleveland.
CHAPTER TWO
Hyperjine Interactions
In Chapter I it was shown that the Mossbauer spectrum is a record

of the intensity of a transmitted r-ray beam as a function of the
Doppler velocity between the source and absorber. The resonant
absorption line has a Lorentzian shape of widthr r and is centred
at zero relative velocity between source and absorber. The practi-
cal application of this measurement would be limited were it not
for the fact that the energies of the nuclear states are weakly influ-
enced by the chemical environment. It is possible to detect these
extremely small effects because of the high definition (better than
I in 10 12) of the r-ray energy.
There are three principal interactions to consider. These are:
(1) A change in the electric monopole or Coulombic interaction
between the electronic and nuclear charges, which is caused by
a difference in the size of the nucleus in its ground and excited
states. This is seen as a shift of the absorption line away from
zero velocity and is variously known as the chemical isomer
shift, isomer shift, or centre shift, and designated by the
symbol o.
(2) A magnetic dipole interaction between the magnetic moment
of the nucleus and a magnetic field. The origin of the latter may
be intrinsic or extrinsic to the a tom. The result is a multiplet
line structure in the spectrum known as magnetic hyperfine
splitting.
(3) An electric quadrupole interaction between the nuclear
22
Hyperfine Interactions 23
quadrupole moment and the local electric field gradient tensor
at the nucleus. This also results in a multiple-line spectrum.
Magnetic hyperfine interactions were first observed by Pound
and Rebka in 1959 [1], and the chemical isomer shift and quadru-
pole interactions by Kistner and Sunyar in 1960 [2]. At first sight
the rapid development of Mossbauer spectroscopy from then on-
wards seems extraordinary, but it should be realized that most of
the underlying theory was already available. The chemical iso-
mer shift is similar in origin to the isotope shift in atomic spectra.
Electric quadrupole interactions are the basis of nuclear quadrupole
resonance, while the magnetic dipole interactions together with the
theory of time-dependent processes were familiar in nuclear mag-
netic resonance and electron spin resonance spectroscopy.
All three effects may occur together, but only the magnetic and
quadrupole interactions are directional and thus have a complicated
interrelationship. The chemical isomer shift behaves independently
of the other two interactions, and is conveniently considered first.
Although this chapter emphasizes the ways in which the hyper-
fine interactions of the nucleus are used to investigate the chemical
environment, it should be noted that many of the early measure-
ments on each Mossbauer resonance were made to obtain values
for the nuclear parameters. For example, although the spin, mag-
netic moment, and quadrupole moment are usually known for
the ground state of the nucleus, this is not necessarily the case
for an excited state. It has often been necessary to determine one
or more of these parameters from the hyperfine interactions in
appropriately selected compounds.
2.1 The chemical isomer shift

In most instances it is adequate to consider the Coulombic inter-
action between the electrons and the nucleus as if the latter were a
point charge. This assumption is made for example in solving the
Schrodinger wave equation for the electronic structure of the atom.
Such a simplification predicts that there will be no change in the
Coulombic interaction energy when the nuclear excited state
decays to the ground state. However, the nucleus does have a finite
size which may change fractionally during the transition. It must
also be realized that an s-electron wavefunction has a finite value
inside the nuclear radius, and is directly responsible for the change
in electrostatic energy observed.
The integrated Coulombic energy for an electron of charge -e
moving in the field of a point nucleus of charge +Ze is given by
Ze 2
Eo = - - -
foo'1'2-
dr
(2.1)
41TEO 0 r
where EO is the permittivity of a vacuum, r is the radial distance,

and -e'P2 is the charge density of the electron in volume element
dr. If the nucleus is spherical with a radius R, then equation (2.1)
is valid for r > R, but overestimates Eo for r < R. It is possible to
calculate an approximate correction, W, by assuming a model for
the proton charge density within the nucleus [3]. If this is taken
to be unifonn, and certain simplifying approximations are made
in the mathematics, the comparatively simple expression obtained
is
(2.2)
where l'Ps(O) 12 is the non-relativistic Schr6dinger wave function at

r = O.
If the nucleus changes its radius by a small increment 0 R during
its transition from the excited state to the ground state, there will
be a simultaneous change in electrostatic energy given by
(2.3)
The value of oR/R is characteristic of each transition, but is typi-

cally of the order of 10-4 and may be of either sign. A positive
sign means that the nucleus shrinks on de-excitation. The M6ss-
bauer experiment compares the difference in energy between the
nuclear transitions in the source and absorber, so that the chemical
isomer shift as 0 bserved is given by
I 2 2 oR 2 2
o= 5€Q Ze R If (W's(O)abm>er I - IlJI"s(O)source I ) (2.4)
This can then be related quite simply to the measured shift in

Doppler velocity units, v, by v = (c/E,()o. (The customary symbol
for the chemical isomer shift, 8, is not to be confused with the
change in nuclear radius, oR).
Equation (2.4) can be seen to be the product of a chemical term
(the electron density at r = 0) and a nuclear term (the change in
nuclear radius). The latter is a constant for a particular transition,
and thus it is possible to study relative changes in electron density
directly. Equations similar to (2.4) are sometimes used which
express the nuclear radius as a mean-square value to indicate that
it is not necessarily spherical, but these may be converted to the
expression above by using the relationships o<R'2)/<R2) = 2oR/R
and <R~> -<Ri> = ~R2(oR/R). For chemical applications it is often
sufficient to compare chemical isomer shift values using the sim-
plified expression
(2.5)
where A and B are two different chemical environments of which

B is either the source matrix or a reference absorber.
It is important to realize that IlJIsCO)12 is the s-electron density
at the nucleus, and not the s-electron occupation in the formal
chemical sense. If oR/R is positive, a positive value of the chemi-
cal isomer shift, 8, implies that the s-electron density at the nuc-
leus in A is greater than in B. IlJIs(O) 12 includes contributions from
all the occupied s-electron orbitals in the atom, but is naturally
more sensitive to changes which take place in the outer valence
shells. Although the values of IIJI(O)12 for p-, d- andf-electrons
are zero, these orbitals nevertheless do have a significant indirect
interaction with the nucleus via interpenetration shielding of the
s-electrons. For example a 3d54s 1 configuration will have a larger
value of IlJIsCO)12 than 3d64s 1 because in the latter case the extra
d-electron shields the 4s-electron from the nucleus.
In a number of instances a Mossbauer resonance can be observed
for two or more transitions to the same ground state, or for several
isotopes of the same element. The chemical isomer shifts of two
different resonances in the same series of compounds should be
interrelated purely by the difference in oR/R, and this appears to
be the case. An interesting illustration is provided by the 1271 and
129 1 spectra of Na3H2106 shown in Fig.2.l. The source matrix was
zinc telluride in each case but con taining 127m Te and 129m Te respec-
tively [4]. These isotopes (3-decay to populate the 57.6-keV level
of 127 1 and the 27.7-keV level of 1291. The absorptions occur at
positive and negative velocities because oR/R is negative in 1271
and positive in 129 1, while the s-electron density at the nucleus is
greater in the source than the absorber. The relative magnitudes
of the shift also differ because [oR/R(1271)] /[oR/R(1291)] = -0.65.
The resonance lines have different widths because rr = 2.54 mm
s-1 for 1271 and 0.59 mm s-1 for 1291.
An accurate calibration of the chemical isomer shift scale for a
particular isotope is not without difficulties. The change in nuc-
lear radius, oR/R, cannot be determined independently of the
chemical environment, which means that the electron density
PPiO) 12, has to be estimated by molecular orbital methods in at
least two compounds before a value for oR/R can be obtained.
Some of the problems which this creates are discussed in Chapter 4.
The observed line-shift is not entirely caused by the chemical
isomer shift as described above. There is another generally smaller
contribution termed the second-order Doppler shift, which was
first observed by Pound and Rebka in 1960 [5]. The emitting or
absorbing nucleus is not stationary but is vibrating on its lattice
site. The period of vibration is much shorter than the Mossbauer
lifetime so that the average displacement and velocity are effective-
ly zero, but the mean-squared values of the velocity, (v 2 ), are
finite.
The relativistic equation for the Doppler effect on the apparent
frequency v of the emitted photon (as recorded at the absorbing
nucleus) is
v = vo (1 _ ~ cos a) (1 _ ;~) - 1/2 (2.6)
where Vo is the frequency for a stationary nucleus, and v is the ap-

parent relative velocity of the emitting nucleus along a direction
making an angle a with the 'Y-ray direction (Note: v is not the velo-
city along the direction of travel of the -y-ray). For an oscillatory
1·00
0·95
50·90
;;;
'"E
'"g 1·00
L.
~
Source: Zn1Z7m Te
0·98
Absorber : No 3HZ 1Z7106
-10 -5 0·0 +5 +10

Velocity I (mm 5- 1)
Fig. 2.1 The 1271 and 1291 spectra of Na3H2I06 showing the smaller line-
width and larger chemical isomer shift of opposite sign for the latter ([ 41 ,
Fig. 1)
motion the mean value (v) is zero so that only the second-order term
in (v2) can influence the Mossbauer resonance. Hence for v ~ c
(V2»)
II =vo (1 +-- (2.7)
2 2c
This gives rise to a shift in the Mossbauer line of
DE vo - v <v 2 )
(2.8)
It is obvious that (v 2 ) increases with rise in temperature. Accord-

ingly, the Mossbauer resonance in an absorber moves to a more
negative velocity as the temperature is raised. The second-order
Doppler shift contribution to the observed chemical shift is smallest
at absolute zero, but is still finite because of a zero-point motion
of the nucleus.
The second-order Doppler shift is often ignored when compar-
ing values of the total observed shift because it always acts in the
same sense, but such a comparison should only be made for simi-
lar compounds at the same temperature, and small differences
regarded as not significant. A more detailed discussion of the
second-order Doppler shift will be found in Chapter 6.
2.2 Magnetic hyperfme interactions

The nucleus has a magnetic moment, Ji, when the spin quantum
number, I, is greater than zero. Its energy is then affected by the
presence of a magnetic field, and the interaction of Ji with a mag-
netic flux density of B is formally expressed by the Hamiltonian
:J['= -p.' B = -gJiNI . B (2.9)
where JiN is the nuclear magneton (eh/4mnp = 5.04929 x 10- 27 A

m 2 or J T-l) and g is the nuclear g-factor [g = Ji/(IJiN)]' Solving
this Hamiltonian [6] gives the energy levels of the nucleus in the
field to be
(2.10)
where mzis the magnetic quantum number and can take the values
I, 1 - 1, ... , - I. In effect, the magnetic field splits the energy
level into 21 + 1 non-degenerate equi-spaced sublevels with a separa-
tion of JiB/I.
In a Mossbauer experiment there may be a transition from a
ground state with a spin quantum number Ig and magnetic moment
Jig to an excited state with spin Ie and magnetic moment Jie- In a
magnetic field, both states will be split according to equations
(2.9) and (2.10). Transitions can take place between sub-levels
provided that the selection rule llm z = 0, ± 1 is obeyed [this is cal-
led a magnetic dipole (M I) transition; for other selection rules see
Section 2.5 on line intensities]. The resultant Mossbauer spectrum
contains a number of resonance lines, but is nevertheless symmetri-
cal about the centroid.
HyperJine Interactions 29
A typical example of magnetic hyperfine splitting is illustrated
schematically in Fig.2.2, which is drawn to a scale appropriate to
+,
119Sn. For this isotope Ig = Ie =T, Ilg = -1.041IlN and
Ile = +0.67IlN· The change in sign of the magnetic moment results
in a relative inversion of the multiplets. The six lines are the allow-
ed t:..m z = 0, ± I transitions, and the resultant spectrum is indicated
in the stick diagram. The lines are not of equal intensity, but the
3:2: I: I: 2:3 ratio shown here is often found for example in the
57Fe and 119S n resonances in randomly oriented polycrystalline
samples. A more detailed account of relative line intensities is
given later in the chapter.
_1
I
I
2
I 1
I.e=l2 I /
(,./ -2
\'\ ' +12
\
\ +12
+1
123 2
/
/
I
I
Ig=~ /
\
\
\
\
\
\ 1
456 -2
Velocity
-I' 0, +1'
3
I
4
I
2
I
5
I I6
Fig. 2.2 The energy-level scheme and resultant spectrum for magnetic hyper-
t
fine &p!itting of an. I g =} ~ Ie = transition. The relative splittings are
scaled In accord WIth the magnetIc moments of 119Sn ; Ilg = -1.041 IlN and
Ile = +0.67IlN- The line intensity ratios of 3:2: I: 1:2:3 are appropriate to a
poly crystalline absorber_
The magnetic field may be applied by an external magnet, or it
may be intrinsic to the compound in which case it is usually refer-
red to as an 'internal' field. An unpaired electron in the atomic
environment can induce an imbalance in electron spin-density at
the nucleus, and thereby generate a local magnetic flux density
which can be as large as 100 T. Examples of this may be found
in Chapter 5.
The sign of an internal magnetic flux density, B, can sometimes
be obtained by applying an additional externally generated mag-
netic flux density, B o, with a large magnet. If the resultant flux
density in the Mossbauer spectrum has increased to B + Bo by
parallel alignment of B with Bo (i.e. the internal field is parallel
to the magnetization) then the sign is positive, whereas a field of
B - Bo signifies antiparallel alignment and a negative sign. This
method fails if the magnetic anisotropy of the matrix is large
enough to prevent rotation of the internal field into the direction
of the applied field.
Time-dependent effects such as relaxation are discussed separ-
ately in Chapter 6.
2.3 Electric quadrupole interactions

The electric quadrupole interaction in Mossbauer spectroscopy is
very similar to that in nuclear quadrupole resonance spectroscopy
[7]. The main difference is that the latter is concerned with radio-
frequency transitions within a hyperfine multiplet of a ground state
nucleus, whereas the former is a 'Y-ray transition between the
hyperfine multiplets of the nucleus in its ground and excited
states. Only those nuclear states with I > ~ have a nuclear quadru-
pole moment and hence show a quadrupole hyper fine splitting. In
consequence it is often possible to observe a quadrupole interaction
in the Mossbauer spectrum which derives from the excited state
of the nucleus, even though the ground state has Ig = 0 or Ig =1-
and therefore does not give an N.Q.R. signal.
The nuclear quadrupole moment, Q, is a measure of the devia-
tion from spherical symmetry of the nuclear charge. It is expressed
by
eQ = Jpr2(3 cos 2 (J - l)dr (2.11 )

where +e is the charge on the proton, and p is the charge density
in the volume element dr at a distance r from the centre of the
nucleus and at an angle e to the axis of the nuclear spin. The mag-
nitude of Q is often referred to in units of barns (1 bam = 10- 28
m 2). The sign of Q can be positive or negative according to whe-
ther the nucleus is respectively elongated (prolate) or flattened
(oblate) along the spin-axis.
The electrostatic potential at the nucleus due to a point charge
q at a distance r is given by V = q/(4rreor) where eo is the permitti-
vity of a vacuum. The interaction of the nuclear quadrupole mo-
ment with the electronic environment is expressed by the
Hamiltonian
1
:Jf' = - - eQ • 'VE (2.12)
6
where'VE represents the electric field gradient at the nucleus

(which is the derivative of the electric field, E, and hence the
negative of the second derivative of the potential, V). 'VE is a ten-
sor quantity which can be written as
a2 v
'ViEi = - = -Vii (2.13)
ax/-'Oxi
(Xi, Xi = X, y, z)
There are thus nine values of Vii to express 'VE in a Cartesian axis
system, X, y, z. A 'principal' axis system may always be defined
*"
such that all the Vii terms with i j are zero (the matrix of the
tensor is diagonal), leaving the three finite 'principal' values Vxx ,
Vyy , and Vzz . Furthermore, ~E is a traceless tensor, so that
(2.14)
As a result it is only necessary to specify two parameters to com-

pletely describe an electric field gradient tensor in its principal
axis sytem. Vzz = eq is taken to be the largest value of IVii I, and
an asymmetry parameter, T/, is defined by
(2.15)
such that I Vzz I > I Vyy I ~ I Vxx I and 0 .;;;; 7'/ .;;;; 1. The z axis is then
referred to as the 'major' axis, and x is the 'minor' axis. The cor-
rect assignment of these axes is often obvious from the molecular
geometry if the overall symmetry is high; e.g. in a molecule with
axial symmetry, the z axis is the symmetry axis. (Note: in an arbi-
trary axis sytem it is necessary to specify five parameters which
maybe Vxx, VX)', Vzz , Vyy,and Vyz,oraltemative1ytheprincipal
values Vzz and 7'/ with three angles to specify their orientation with
respect to the arbitrary x, y, z axes.)
The Hamiltonian in equation (2.12) is more often written as
2
= e qQ [31; _ 1(1 + 1) + 1/(/; -1;)] (2.16)
41(21 - 1)
where Ix, ly and lz are quantum-mechanical spin operators. A

completely general solution of this Hamiltonian is not possible,
but exact expressions may be given under certain conditions.
If the electric field gradient tensor has axial symmetry (7J = 0)
the energy levels are given by
EQ = e2qQ [312 - 1(1 + 1)] (2.17)

41(21 - 1) z
where the quantum number 1z can take the 21 + 1 values of

1, 1 - I, ... ,-1. If 1 = h hi-etc. the energy level is split into
a series of Kramers' doublets with ±1z states degenerate. An
important example is 1= h which has two energy levels at
+e 2qQ/4 for 1z = ±%- and -e 2qQ/4 for I z = it. If the symmetry is
lower than axial (i.e. when 7J > 0), an exact expression can only be
given for I = h and is
EQ = e2qQ [31; - 1(1 + 1)] (1 + 7] 2/3)1/2 (2.18)

41(21 - 1)
with energy levels at ±(e 2qQ/4)(1 + 7]2/3)1/2. Values for higher spin
states must be calculated numerically.
A Mossbauer transition between states can take place according
to the selection rule f1mz = 0, ± 1 for Ml dipole radiation. The
most common example is the Ig = } -+ Ie = f- transition in which
the ground state with Iz = ±} is unsplit, and the excited state has
two levels with I z = ±f- and ±t separated by (e 2qQ/2) (1 + T/ 2/3 )1/2.
The resultant Mossbauer spectrum is a doublet with a separation
called the quadrupole splitting of f1 = (e 2qQ/2)(1 +7]2/3)1/2. When
7] = 0, f1 = e 2qQ/2 or half the quadrupole coupling constant defin-
ed in nuclear quadrupole resonance spectroscopy. Typical energy
level schemes for} -+ %and T-+ i (scaled to 1291) transitions and
the resultant spectra with relative line intensities for randomly
oriented polycrystalline samples are illustrated in Fig.2.3.
The quadrupole spectrum is symmetrical only in the case of a
} -+ %transition. In this particular instance there is no direct means
of establishing the magnitude of T/ or the sign of Vzz . However,
as detailed later in the chapter, it is possible to determine these
parameters, either from the angular dependence of the line inten-
sities in single crystals, or from the spectrum observed when a
large external magnetic field is applied.
Once e 2qQ and T/ have been measured, they may be related to
the chemical environment of the resonant nucleus. The coupling
constant e 2qQ is the product of a nuclear constant (eQ) and the
maximum value of Vii (eq). Considerable confusion is often caused
by referring to the sign of Vzz = eq without specifying whether e
is taken to be the charge of the electron or the proton. For this
reason it is preferable to refer to the sign of e 2qQ or of q.
°
If Ig = or} so that Qg = 0, then the magnitude of Qe has to be
established empirically by comparing the observed quadrupole split-
ting values with estima ted values of Vzz. Fortunately, use of the
quadrupole splitting is not entirely dependent on these intrinsi-
cally difficult calculations. The relative order of magnitude of
e2qQ can be related comparatively easily to the valence orbitals
of the atom. The numerical value of Vzz due to an occupied elec-
tron orbital expressed by the wave function lJf is given by the
integral
-
Vzz = q = - (1 -R) JlJf * (3COS 28-1) lJf dr (2.19)
e 41T€Q r3
where (1 - R) is an empirical quantity less than unity, called the

+~ 2 678 +2
-2 \ I -2
\ I
\ I
\ I
\ I
\
\ le=~ L=2
e 2 I
I
I \" 345 +1
I
I
\ -2
I \
I \
I \ 12 +1
+1 I -2
-2
+1
-2
I
I
/
19 =12 Ig =1 I
1.---
I
+~
-2
<1:,
",-..
,, +~
-2
+1
-2
li
Velocity Velocity
-V 0 +V
4
! ! 8
:5
Z
6
Fig. 2.3 The energy-level schemes and resultant spectra for quadrupole
t t
splitting of Ig = -+ Ie = ~ and Ig = -+ Ie = (scaled to 1291) transitions. f
Sternheimer shielding factor, which allows for an opposing polari-

zation of the inner core electrons [8]. Similarly
I
11=-
q
f IJI* (3 sin e cos 2r!J) IJIdr
2
r3
(2.20)
Hyper/ine Interactions 35
Table 2.1 The magnitude of q and 1/ for the p- and d-orbitals
Orbital 41TEoq/(i - R) 1/
_~(r-3) 0
Pz 5
Px +~(r-3) -3
5
Py +~(r-3) +3
5
d X 2_y2 +~(r-3) 0
7
dz 2 -!(r- 3) 0
7
dxy +~(r-3) 0
7
dxz -~(r-3) +3
7
dyz -~(r-3) -3
7
The magnitudes of q and 1/, expressed in units of the expectation

value of 1/r 3 , are given in Table 2.1 for the conventional p- and
d-orbitals.
The total value of q for a completely filled or half-filled shell of
electrons is zero. An excess of electrons along the z axis (occupy-
ing pz. d;. dxz or d yz ) results in a negative value of q. while an
excess in the xy plane (Px. Py. d xy or d x2_ 2) gives a positive value
of q. The value of (r- 3) for a 4p-electron '~ill be significantly smal-
ler than that of a 3p-electron because of the greater radial extent
of the former, and similarly a 3d-electron will give a smaller value
than a 3p-electron, but probably greater than a 4p-electron. The
spherically symmetric s-electron gives no electric field gradient.
The sign and magnitude of e2qQ can therefore be used empirically
to measure the asymmetric occupation of the atomic valence orbi-
tals, and many examples of this will be found in later chapters.
In some compounds the Mossbauer atom has an intrinsically higl
symmetry (e.g. the Fe3+ d 5 ion has a half-filled shell and is a sphe-
rical S-state ion) but may still show a quadrupole splitting. The
latter originates from charges external to the atom, such as other
ions, which polarize the spherical inner core and can induce a very
large electric field gradient at the nucleus. This 'lattice' contribu-
tion summed over individual charges Zie may be written as
Vzz (l - Yoc) '" 3 cos 2 e - 1

- = qlatt = - - - - ~Zi ---
r3
(2.21)
e 41TEQ
i
where Yoo is the Sternheimer antishielding factor [8]. This para-
meter expresses the core polarization and is usually large and nega-
tive. The lattice contribution also exists when the valence orbitals
are asymmetrically occupied, but is usually acknowledged to be a
minority contribution unless the co-ordinate geometry is very
distorted. Formal attempts to subdivide q into valence and lattice
terms are generally arbitrary and not altogether convincing.
2.4 Combined magnetic and quadrupole interactions

Both the magnetic and quadrupole hyperfine interactions express
a directional interaction of the nucleus with its environment. How-
ever, when the two are present together, their respective principal
axes are not necessarily co-linear, and it is not surprising therefore
to find that the resultant behaviour can be much more complex.
The formal Hamiltonian which is the sum of equations (2.9) and
(2.12) has no general solution, but the predicted spectrum may
always be obtained by numerical computation.
One of the few useful restricted solutions for a ~ -+ f transition
is the case where the quadrupole interaction is very much weaker
than the magnetic term, and can be treated as a small perturbation
upon the latter. The resultant energy levels are given by
I'm 1+ 1/2 e 2qQ (3 cos 2 e - 1)

EQM = -gllNBmz + ( -1) z -4- 2 (2.22)
where e is the angle between the magnetic axis and the major axis
of the electric field gradient tensor. All the magnetic hyperfine
lines are shifted by a quantity
e2qQ (3cos 2 e-l)

lel=-- (2.23)
4 2
but the angle e and the value of e2 qQ cannot be determined

separately from the line positions. A schematic illustration of this
is given in Fig.2.4. The presence of a small quadrupole perturba-
tion is easily visible because the spectrum is no longer symmetrical
about the centroid. If by chance cos e = 1;'''';3 then the second term
in equation (2.22) vanishes and the spectrum appears to be that
6 _l
, - - /..../ 2
I
I
,
I
I
I .l
/
I ........ /
/--,
, 3 5
_e_2_ _ ~""
,,
,, ..... _- " '
........
\
' ....
,, Z 4
,, - - /.... /
1
-- -
......-
,,_____ --L....L.L.++-+-_ +1
" 2
1=1 //
/ "
---<.
9 2 "
"" ,
" ,
,-------....&-.. . . . - _12
Magnetic Magnetic
+
quadrupole
I
Velocity
----'1
-V 0 +V
I~ ---,----'-,-3
' -----'---51
6
Fig. 2.4 The effect of a small quadrupole perturbation on a } ~ magnetic t

hyperfine splitting. Lines 1 and 6 are shifted to more positive velocity by
+E, while lines 2, 3, 4 and 5 are shifted by -E. The difference in the separa-
tions 1-2 and 5-6 is therefore 4E.
of an unperturbed magnetic hyperfine splitting. Examples of com-

bined magnetic/quadrupole interactions may be found on pp 115-116.
In the preceding section it was found that the sign of e 2qQ and
the magnitude of 1/ could not be determined from the quadrupole
splitting of a 4- ~ ~ transition in a polycrystalline absorber because
the spectrum is symmetrical. One means of obtaining this informa-
tion is to re-measure the spectrum in a large externally applied mag-
netic field with a magnetic flux density of 30-50 T. It is custo-
mary to apply the field parallel or perpendicular to the 'Y-ray
beam. The resultant spectrum is-of complex shape, reflecting the
random orientation of the electric field gradient tensor with
respect to the applied field, but is not symmetrical, so that the
sign of e 2qQ is indicated. The shape of the spectrum may be cal-
culated approximately by summing numerically the individual cal-
culated spectra for a large number of orientations of the electric
field gradient tensor.
One of the best examples of this method is given by the Fe57
resonance. The applied field splits the it
component of the quad-
rupole spectrum into an apparent doublet, and the i~ component
into an apparent triplet. Typical calculated spectrum envelopes
are shown in Fig.2.S. The spectrum tends towards a symmetrical
shape as f/ approaches unity (Le. when Vzz = - VY.Y the sign of the
major axis is indeterminate), and can be used to determine f/
to an accuracy of about ±O.OS.
2.5 Relative line intensities

It was intima ted in the preceding sections that the individual lines
in a spectrum showing a hyperfine interaction have characteristic
intensities which can be used to identify particular transitions. The
magnitudes of the chemical isomer shift, the magnetic hyperfine
field and the quadrupole splitting, can all be determined from the
line positions alone, and the intensities merely used to confirm the
assignment. If it is also desired to obtain the orientation of the
magnetic axis or the electric field gradient tensor in an anisotropic
sample such as a single crystal, then a more detailed knowledge of
line intensities is essential.
The intensity of a particular hyperfine transition between quan-
tized sub-levels is determined by the coupling of the two nuclear
angular momentum states [9]. It can be expressed as the product
of two terms which are angUlar-dependent and angUlar-independent
respectively. Since the former averages to unity when all orienta-
tions are equally probable, e.g. in a randomly oriented polycrystal-
line powder sample, it is convenient to consider the angular-
independent term first.
1/=0
1/-0,4
c:
o
';;;
'e... ~--
...
oil
c:
I-
-2 o 2 4
Velocity /( mm 5- 1)
Fig, 2,5 Calculated spectra for a polycrystalline absorber giving an 57Fe

quadrupole splitting of ~ = 3 mm s-1 and in an applied field with a flux
density of 5 T parallel to the direction of observation, The spectrum becomes
symmetrical as 1/ tends towards unity, The sign of e 2qQ is positive in all
cases,
The intensity in this instance is given by the square of the appro-

priate Oebsch-Gordan coefficient [10]
(2.24)
where the two nuclear spin states It and 12 have I z values of m 1

and m2, and their coupling obeys the vector sums J = 11 + 12 and
m = ml - m2. J is referred to as the multipolarity of the transi-
tion, and the intensity is greater if 1 is small: if 1 = I it is referred
to as a dipole transition, while with 1 = 2 it is a quadrupole transi-
tion. Most of the Mbssbauer transitions take place without a
change in parity, so that the radiation is classified as a magnetic
dipole (M I) or electric quadrupole (E2) transition. One of the few
exceptions is the 25.65-keV ~.+-+ %- transition in 161Dy which is
an electric dipole (El) transition. In some instances such as
99Ru(90-keV) and 197 Au(77-keV) the radiation is a mixture of Ml
and E2, and both terms must be added in the appropriate pro-
portion. The selection rule for an M1 or E 1 transition is !::.m z =
0, ±1, and for an E2 transition is !::.m z = 0, ±I , ±2.
The most frequently used coefficients are those for the ~ -+ ~
MI transition, and these are given in Table 2.2 (coefficients for
other spin states have been tabulated in [ II ]). h may be either
the ground or excited state spin. Although there are nominally
eight transitions, the +~ -+ -~ and -~ -+ +-i transitions have a zero
probability (forbidden). The six finite coefficients, (;2, which
express the angular-independent intensity have a total probability
of unit intensity and give directly the 3:2: I: 1:2:3 intensity ratios
for a magnetic hyperfine splitting, shown in Fig.2.2. The corres-
ponding terms for a quadrupole spectrum are obtained by summa-
tion and give a I : I ratio.
The angular dependent terms, eel, m), are expressed as the
radiation probability in a direction at an angle 6 to the quantiza-
tion axis (Le. the magnetic field axis or Vzz : note that the values
in the table in the latter case are only correct if 1/ = 0). The inten-
sity for a polycrystalline sample is obtained by integration over all
6 to obtain eel, m); e.g.
~ sin2 6 = -4nI J21f J 1f

(~ sin 2 6) sin 6d6drp = I (2.25)
2 2
o 0
and the total of emitted radiation is independent of 6 and nor-

malized to unity, Le.
ml m 2
L {<hl - ml m II2m 2)2e(l, m) - I (2.26)
Coefficients such as those in Table 2.2 are necessary to interpret

the angular dependence of the spectrum from a single crystal or
oriented absorber. For example, a magnetically ordered metal alloy
or oxide absorber may often be 'polarized' by magnetizing in a
small external magnetic field to give a unique direction of the inter-
nal field. The expected line intensities can then be predicted from
Table 2.2 to be in the ratios 3:x: 1: 1 :x:3 where x = 4 sin 2 8/0 +
cos 2 8); in particular the m = 0 transitions have a zero intensity
when observed along the direction of the field (8 = 0°) and a maxi-
mum intensity perpendicular to the field (8 = 90°). This is illustrat-
ed schematically in Fig.2.6.
The equivalent behaviour in the quadrupole spectrum is a 1:3
ratio for the 'Y-ray axis parallel to the direction of Vzz and a 5: 3
ratio perpendicular to V zz. In this instance the angular dependence
Table 2.2 The relative pro babilities for a ~,~ transition
Magnetic spectra (M 1)
C C2 e (J, m)
m2 -m1 m (1) (2) (2)
+!. I
+~
2 2
+1 1 4 ~O + cos 2 8)
+!.2 +!.
2
0 vi I
6" ~2 sin 2 8
I
-"2 +!. -1 V~
I
io + cos 2 8)
- .,.
2 12
-z3 +!.
2
I
0 0
+~ -"2 +2 0 0
2
+!.2 2
I
+1 V} I
12 ~(l + cos 2 8)
I
-2"
I
-2 0 vi- I
6" ~
2
sin 2 8
-2
3 I
-2" -1 1 I
"4 io + cos 2 8)
Quadrupole spectra (Ml) when 11 = 0

C2 8(J, m)
Transition (2) (2)
+!. +!. 1
-2' -2 "2 !.
2
+ ~4 sin 2 8
+~ +!. I
~(1 + cos 2 8)
-2' -2 "2
(1) The Oebsch-Gordan coefficient (!..t- mlmltm2>

(2) C2 and e(J, m) are the angular-inaependent and angular-dependent terms normalized
to a total radiation probability of
L
mlm2
C 2 e(J, m) = 1
n
Random
r
I I
m2 +~
2
+12 -21 +12 -21 -23
+12 +12 +12 1 1
m1 -2 -~ -2
Fig. 2.6 The effect of orientation upon the relative line intensities of a
t
magnetic hyperfine splitting and a quadrupole splitting of a } -+ transition
in an oriented absorber with a unique principal axis sytem.
can be used to determine which of the two lines is the ±i -+ ±%

transition, and hence to obtain the sign of e2qQ.
If there are combined magnetic and quadrupole interactions, or
the asymmetry parameter 71 is non-zero, the previous discussion is
no longer strictly valid. In these circumstances the energy states
are no longer 'pure' states with a defined quantum number, and
must be represented as linear combinations of terms. Likewise the
intensity is obtained by an amplitude summation to allow for
interference effects. These calculations are somewhat complex
[ 11] and will not be described here.
The total cross-section for resonant absorption is a nuclear con-
stant, so that the individual hyperfine lines are proportionately
weaker than in an unsplit resonance line. This unfortunately neces-
sitates a longer counting time to resolve the extra detail. In this
discussion the absorption intensity has been considered for a single
nucleus, but it is also important to consider the effects of finite
a bsorber thickness (see p.11-12). The effective thickness of the
absorber Ta = nauofa is now multiplied by the fractional line inten-
sity for each component C2e(l, m), with the result that saturation
of the absorption intensity occurs more quickly for the more in-
tense resonance lines. This is seen as a dependence of the relative
line intensities on increasing thickness which tends to decrease
any asymmetry. In addition, the more intense lines have a larger
line width due to the thickness broadening effect. Because of this,
a quadrupole doublet spectrum of for example a single crystal
with the direction of observation parallel to Vzz will show a line
intensity ratio much less than 1:3.
An additional complication which also affects the intensities
appreciably is polarization of the incident r-ray beam as it pro:-
gresses through the crystal [12]. For these reasons, unless it is
proposed to attempt a somewhat difficult thickness correction, it
is necessary to obtain relative intensity data from very thin ab-
sorbers. However, the possibility of underestimation of any asym-
metry should always be borne in mind.
There are two other important influences on relative line inten-
sities which are of particular importance in powders. If the com-
pacted polycrystalline powder has a tendency towards partial
orientation (texture), then the values ofe(l, m) do not average to
unity, and there is some residual asymmetry in the spectrum which
is angular dependent. This effect can be quite large, and exceed-
ingly difficult to eliminate in fibrous or platelike materials. A
second effect which is similar in many ways is the Goldanskii-
Karyagin effect, in which anisotropy of the recoilless fraction
results in an asymmetry which is not angular dependent but is
usually small. This is discussed in more detail in Chapter 6. In
many instances deviations from the 1: 1 line intensity ratio in a
i ~ ~ quadrupole doublet have been attributed to the Goldanskii-
Karyagin effect without convincing proof that there was no
residual texture in the sample.
References
[1] Pound, R. V. and Rebka, G. A. (1959) Phys. Rev. Letters, 3, 554.
[2] Kistner, O. C. and Sunyar, A. W. (1960) Phys. Rev. Letters, 4, 412.
[3] Breit, G. (1958) Rev. Mod. Phys., 30,507.
[4] Reddy, K. R.,Barros, F. de S. and DeBenedetti, S. (1966)Phys.
Letters, 20, 297.
[5] Pound, R. V. and Rebka, G. A. (1960) Phys. Rev. Letters, 4, 274.
[6] Abragam, A. (1961) The Principles of Nuclear Magnetism. Clarendon
Press, Oxford.
[7] Kubo, M. and Nakamura, D. (1966) Adv. Inarg. Chern. and Radiachem.,
8,257.
[8] Ingallis, R. (1962) Phys. Rev., 128, 1155.
[9] Rose, M. E. (1955) Multipale Fields. Chapman and Hall, London.
[10] Condon, E. U. and Shortley, G. H. (1935) The Theory of Atomic
Spectra. Cambridge Dniv. Press.
[11] Greenwood, N. N. and Gibb, T. C. (1971) Mossbauer Spectroscopy,
Chapman and Hall, London.
[12] Housley, R. M., Grant, R. W. and Gonser, U. (1969)Phys. Rev., 178,
514.
CHAPTER THREE
Molecular Structure
In the preceding chapter it was shown how the hyperfine inter-

actions in a Mossbauer spectrum contain information about the
immedia te electronic environment of the resonant nucleus. In
particular, they can be related to the geometrical symmetry and
to the nature of the chemical bonding involved. For example, a
change in the formal oxidation state of the resonant nucleus (re-
moval or addition of a valence electron) may be expected to result
in a change in the s-electron density at the nucleus and thence in
the chemical isomer shift. The remaining chapters of this book are
therefore concerned with various aspects of the chemical applica-
tion of Mossbauer spectroscopy. In this chapter, some examples
of its use to determine molecular structure, i.e. the geometrical
disposition of atoms in a compound, are discussed. The study of
electronic structure and the way in which these atoms are chemi-
cally bonded is deferred until the following chapter.
The successful determination of structural data from the Moss-
bauer spectrum of a molecular complex is very much governed by
individual circumstance. In the majority of compounds studied
there has been only one suitable resonant isotope. The advantage
of measuring parameters specific to particular atoms is offset by
the difficulty of determining the number and disposition of the
atoms or groups to which they are bonded. This is particularly
true in diamagnetic compounds where the primary parameters of
the chemical isomer shift and the quadrupole splitting are a func-
tion of all the valence electrons, which are of course usually par-
45
ticipating in covalent bonding to more than one atom. The unpair-
ed electrons in the valence shells of paramagnetic compounds pro-
duce characteristic effects which are discussed in Chapter 5.
The resonant atom in a diamagnetic compound gives a simple
spectrum, usually with a quadrupole splitting. In the event that
there are several non-equivalent resonant atoms in the structure,
then each should give a different contribution to the spectrum;
indeed the observation of two quadrupole spectra from a com-
pound with two resonant atoms is proof that these are not in the
same environment. The converse, that one quadrupole spectrum
indicates identity of environment, is not necessarily true, since
any differences may be less than the resolution of the measurement.
3.1 Iron carbonyls and derivatives

The 14.4-keV 57 Fe resonance has been widely used in structural
studies of the iron carbonyls and their derivatives where multiple
iron environments are quite common. A good example is provided
[I] by the spectra of poly crystalline [Fe(pY)6] 2+ [Fe4(CO)13] 2-
and [NEt4L2[Fe4(CO)13j2- illustrated in Fig.3.l. Quadrupole
splitting of the Ig = -} -+ Ie = i- transition results in a symmetrical
doublet for each distinct iron environment. The extra pair of lines
in the former case are thus a quadrupole doublet from the high-
spin Fe 2+ complex cation. This assignment is borne out both by
comparing the value of the chemical isomer shift with data from
other Fe 2+ compounds, and by the approximate absorption area
which is one quarter of that due to the anion. However, the anion
itself is known to be a triangular-based pyramid of Fe(COh groups
with the thirteenth carbonyl in a triply-bridging configuration
below the basal plane. Although there are three equivalent iron
atoms and a unique one, it is not possible to visually resolve the
apparently small differences in the Mossbauer spectrum.
As a further example, the hydride anion [Fe2(CO)8H]- shows
only a simple quadrupole splitting in the 57Fe spectrum, suggesting
that there is only a single iron environment [2]. The three most
likely structures are (1 )-(3). Structures (1) and (2) are derived
from that of Fe2(COh by substitution of a bridging and a terminal
carbonyl group respectively. Only (1 ) and (3) have identical iron
environments, but (2) cannot be completely eliminated as simply
even though data for other carbonyl hydrides leads oile to predict
Molecular Structure 47
.,
-1 o 2 3 4
Velocity /(mm 5- 1)
Fig.3.1 57Fe spectra at 80 K for the [Fe(PY)61 2+ and [NEt41 + salts of

[Fe4(CO)1312-. (PY = pyridine) Note the extra lines showing quadrupole
splitting from the cation in the former, and the failure to distinguish the
unique iron atom in the carbonyl anion. ([ 1] , Fig. 2.10).
g
\,8
OC - -Fe-H -
I
Fe - -CO
1
o
/\
3
a detectable difference in parameters. In a situation like this it
is often advantageous to consider the Mossbauer data in combina-
tion with other spectroscopic measurements. In the event, the pre-
sence of bridging carbonyl bands in the infra-red spectrum elimi-
nates structure (3), and the total number of bands corresponds to
prediction from the symmetry of structure (I). As may be seen
from Table 3.1, the Mossbauer parameters for [Fe2(CO)gH]- are
similar to those for Fe2(CO)9. The replacement of a bridging car-
bonyl in Fe2(CO)9 by hydrogen causes a reduction in chemical
isomer shift, [) , and an increase in the quadrupole splitting, D.,
exactly analagous to the observed changes in Fe3(COh2 and
[Fe3(CO)11H]-. As a result, the Mossbauer data in conjunction
with the infra-red evidence constitute satisfactory proof that struc-
ture (I) is correct.
The chemical isomer shift is not a direct function of stereo-
chemistry because it involves the electron wavefunctions with
spherical symmetry. Nevertheless, a significant difference in shift
between two related compounds is usually evidence of either a
change in co-ordination number, or in the co-ordinated ligands, or
both. Such a change may have a large effect on the p- and d-
electron shielding or on the bond hybridization. On the other hand,
the quadrupole splitting derives from asymmetric occupation of p-,
d- and f-electron orbitals and is therefore related directly to the
geometry of the compound. A study of these parameters for a
large number of related compounds reveals systematic relationships
Table 3.1 Mossbauer parameters from the 14.4-keV 57Fe resonance at 80 K

in some derivatives of Fe2(CO)9 and Fe3(COh2
Outer doublet Inner doublet

Compound D.* [)* D.* 8*
Fe2(CO)9 0.42 +0.16

[Fe 2(CO)SH)- 0.50 +0.06
Fe3(COh2 1.13 +0.11 -0 +0.05
[Fe 3(CO)l1H)- 1.41 +0.04 0.16 +0.02
[(OChFePMe2Ph)3 1.15 +0.09 0.57 +0.02
(ffars) Fe3(COho 1.52 +0.16 -0 +0.03
HFe3(COhoCNMe2 0.94 -0.04 0.16 +0.04
* Values in rom s-1. Ii is given relative to iron metal at room temperature. ffars = 1,2-
bis( dimethy larsino )tetrafluorocyclo butene.
with such features as co-ordination number, site symmetry, and
the nature of the bonding groups. Diagnostic interpretation of the
spectra from new uncharacterized compounds can then become
possible.
As an example, the Mossbauer parameters for iron carbonyls are
not very sensitive to substitution, but the quadrupole splitting is
often indicative of geometry [3]. An iron atom in 6-coordination
tends to give only a small quadrupole splitting as seen in for
example Fe2(CO)9, [Fe2(CO)sH]- and (OChFe(PMe2hFe(COh-
The latter (structure 4) has an Fe-Fe bond to complete the 6-co-
ordination (with a quadrupole splitting of .l = 0.68 mm s-l). In
2
ac,,1
"Fe-CO
if- I
C
o
4 5
o
C
I 2
Oc I 2
~"le-p~e2 I,,-CO
I
Oc ........ 1 ___ PMe,_I __ CO
OC .:--.. ~Fe __ Fe
Me,P-Fe ..... OC- -PMer , .......CO
8 I Co
(, I
C
o
6 7
contrast, Fe(CO)s (.l = 2.57 mm s-l) and (OC)4Fe.PMe2.PMe2.Fe(CO)4

(.l = 2.58 mm s-l) are 5-coordinated, but (OChIFe(PMe2hFeI(COh
(.l = 0.99 mm s-l) is again 6-coordinated with a small splitting
(structures 5-7). However, the range of structural types is so
large that reliable generalizations by empiricism can become
difficult.
In appropriate instances it is helpful to determine the sign of
e2qQ and the magnitude of the asymmetry parameter, 1/, by apply-
ing a large external magnetic field (see p.38). An example is given
by the results in Table 3.2 for some triphenylphosphine and
Table 3.2 S7Fe quadrupole coupling parameters for phosphine derivatives
of Fe(CO)S
Compound }e2qQ/(mm s-1) 'T/
Fe(CO)S +2.60 0.0

Ph3PFe(CO)4 +2.54 0.0
(Ph3PhFe(CO}J +2.76 0.0
(diphos)[Fe(CO)412 +2.46 0.0
(diphos)Fe(COh -2.12 0.8
1,2-bisdiphenylphosphinoethane (diphos) substituted derivatives

of Fe(CO)s [4]. The magnitude of the quadrupole splitting is
insensitive to the substitution, but when the axial symmetry of
the trigonal bipyramid is broken by a chelated substitution as in
(diphos)Fe(COh the sign of e 2qQ changes and 'T/ increases dramati-
cally. However, it is still not possible to distinguish whether the
bi-dentate substitution is axial-equatorial or equatorial-equatorial.
Nevertheless, it is clear that the triphenylphosphine groups in
(Ph 3PhFe(COh are both in the axial positions.
The usefulness of M6ssbauer spectroscopy in following structu-
ral changes in a family of related compounds can be well illustrated
by some studies made on derivatives of the iron carbonyl, Fe3(CO)J.2.
The parent carbonyl shows a spectrum with three lines of equal
intensity [5]. This is illustrated in Fig.3.2. Comparison with other
carbonyl data allows confident assignment of the outer lines to a
quadrupole splitting of two apparently identical iron atoms, and
the centre line to the third atom in a more symmetrical environ-
ment such that any quadrupole splitting is not resolvable. These
observations were considerably at variance with the then accepted
crystal structure (which was incomplete due to complications from
crystal disorder effects) [7]. This structure proposed an equi-
lateral triangle of iron atoms, each with two terminal carbonyls
and two carbonyls bridging to each of the other iron atoms, so that
all three iron atoms were equivalent.
The spectrum of the related [Fe3(CO)l1H] - anion is similar to
that of Fe3(CO)J.2 but with a larger splitting in the outer doublet
[8J. The parameters are given in Table 3.1 for comparison. It
can be deduced that the hydrogen has replaced a carbonyl group
bridging to the two equivalent iron atoms, and the structure pro-
3090
~.......
-.
A..~
...\ :
'. .'' ...

.." '
,..,;-. ,-.:-'
~
.
.....
C
: ......".
::I HFe3(CO)10CNMez
<32'30 \
:'
:"'\
..
,',
......
NO
-1 ,0 -0·5 o +0·5 HO
Veloci ty I( mm s-') relat ive to iron
Fig. 3.2 57Fe Mossbauer spectra at 80 K for Fe3(CO)12 and

HFe3(COhoCNMe2. The two compounds are clearly related structurally.
The central line in Fe3(CO)12 does not show a resolved quadrupole splitting.
([6), Fig. O.
posed for the hydride (structure 8) has been verified by X-ray

techniques. A recent re-investigation of the parent Fe3(COh2 has
confirmed [9] the now suspected structure (9), showing the iso-
sceles triangular arrangement and the close relationship to
[Fe 3(COh 1H]-.
The substituted derivative Fe3(COh(PMe2Phh can have one of
many possible structures depending on where the substitution takes
place. The spectrum is again closely similar to that of Fe3(CO)12,
but shows a small decrease in the chemical isomer shift at both
types of iron atom, as well as small changes in quadrupole splitting
(see Table 3.1). Here again, the tentative structure (10) involving
substitution at each iron atom has been fully confirmed at a later
date by X-ray analysis [10].
Another derivative, (ffars)Fe3(COho is also clearly related to
Fe3(CO)12 [11]. The two equivalent iron atoms remain so, but
with a considerable increase in quadrupole splitting, leading to
the proposed structure (11 ).
A final example is given by a compound initially reported as
oc co
oc co
0
(j
/Me
OC
I"
Me Ph
/Ph
°0
jP--Me
Me
10
oc
11
oc co
12
HFe3(CO)uNMe2. The suggested structure featured a terminal

-NMe2 group. The Mossbauer spectrum indicates a strong simi-
larity to Fe3(CO)12 with any structural change affecting the bridg-
ed iron atoms equally (see Fig.3.2), and is incompatible with the
initial proposition. Further investigation by mass-spectrometry
has established that the compound is in fact HFe3(COhoCNMe2
[5] ,and combination of the Mossbauer data with the fragmenta-
tion pattern from the mass spectrometer led to the novel struc-
ture (12). The mass spectrum shows stepwise loss of the carbonyl
groups to give a skeletal fragment HFe3CNMel, and other posi-
tively identified fragments were HFe3C+ and HFe2(CO)CNMel,
good evidence for the adopted formulation.
3.2 Geometrical isomerism in Fe and Sn compounds
In many series of co-ordination compounds it is possible to find
examples of a structural ambiguity from isomerism. One of the
common forms is cis-/trans-geometrical isomerism in octahedral
compounds of the type MA2B4 where the A groups are either
adjacent (cis) or diametrically opposed (trans). It is easy to show
that the quadrupole splitting of atom M can be expected to be
twice as great in the trans-isomer and of opposite sign to the cis-
isomer. An elementary proof of this by a point charge calculation
is given in the appendix at the end of the chapter, where the prin-
cipal values of the electric field gradient tensors in MAB s, cis-
MA2B4 and trans-MA2B4 are found to be in the ratio 1: -1 :2
respectively. The asymmetry parameter is zero in all three cases.
Experimental verification of this was first given by Berrett and
Fitzsimmons [12] for some isocyanide complexes of iron:
[Fe(CN)(CNEt)s] (Cl 04) Ll = 0.17 mm s- 1

cis- F e( CNh(CNEt)4 Ll = 0.29 mm s-1
trans-Fe(CNh(CNEt)4 Ll = 0.56 mm s-1
and in one instance the difference in sign between the cis- and
trans-isomers has been verified by the applied field method [13] :
[FeQ(p-MeO.C614.NC)s] (CI04) Ll = 0.70 mm s-1

cis-FeQ2(p-MeO.C614·NC)4 Ll = -0.83 mm s-1
trans-FeCI2(P-MeO.C614·NC)4 Ll = +1.59 mm S-1
This general rule for distinguishing isomers will remain valid pro-
vided that the geometry is close to octahedral and provided that
the electron distribution in one M-A bond is not significantly influ-
enced by its geometrical relationship to the other, i.e. there is no
large 'trans-effect' (note that a gross distortion of the geometry
may lead to a large value of the asymmetry parameter and/or a
change in the principal axis system of the electric field gradient
tensor).
Similar results have been found in 6-coordinate organotin com-
pounds using the 23.87-keV Ig = ~ -+ Ie = %resonance of 119Sn. An
empirical study of observed quadrupole splittings reveals a charac-
teristic distinction between on the one hand groups bonding
through carbon such as alkyl and aryl groups, and on the other
hand halides and nitrogen and oxygen bonded donor ligands. In
consequence, a compound such as Ph2SnCl2.phen behaves as if it
were of the type MA2B4 where both the chlorine atoms and the
1,IO-phenanthroline are classed as B ligands. This rather peculiar
effect will be mentioned again later in the chapter. Such com-
plexes also show the 1:2 ratio for cis- and trans-structures as may
be seen from Table 3.3, and it is therefore possible to derive un-
known stereochemistries [14]. The earlier assignment of trans-
phenyl groups to Ph 2Sn(acach is evidently incorrect. The cis-
isomers in chelated tin compounds are usually distorted from an
octahedral geometry. As described on p.67 this may result in a
change in the sign of the quadrupole splitting so that it corresponds
to that of a trans-isomer; however the magnitude of the splitting
still remains approximately the same as expected for a more regu-
lar geometry, and the method is still diagnostic of configuration if
this problem is borne in mind.
As a cautionary note, mention must be made of the syn- and
anti-isomers of the iron cyclopentadienyl derivative Cp(OC)Fe-
(PMe2hFe(CO)Cp (structure 13) which have the cyc10pentadienyl
ring on the same and opposite sides of the Fe2P2 plane respectively.
In this case the 57Fe spectrum is the same in both compounds.
The immediate Fe environment is identical, and is not influenced
significantly by geometrical changes at a much greater distance [3].
Table 3.3 The relationship between the quadrupole splitting and cis-/trans-
isomerism in organotin compounds
Stereochemistry Stereochemistry
Compound by Mossbauer by other methods
Ph2Sn(acac)z 2.14 cis trans

Me2Sn(acac)z 3.93 trans trans
Ph2SnCl2·phen 3.70 trans
Ph2SnCl2·bipy 3.90 trans
Me2SnCl2·phen 4.03 trans
Pr~Sn.oxin2 2.20 cis
Ph2Sn.oxin2 1.78 cis cis
Bu~Sn.oxin2 2.21 cis
Me2Sn.oxin2 1.98 cis cis
acac =acetylacetone, phen = 1,1 O-phenanthroline, bipy = 2,2'-bipyridyl, oxin = 8-hydro-

xyquinoiine.
13
3.3 Linkage isomerism in cyano-complexes of Fe

Some simple inorganic ligands such as CN- and NCS- are capable
of co-ordinating to a metal cation from either end of the molecule,
and this leads to a possible structural ambiguity. The hexacyano-
ferrate(II) complexes with the transition metals usually form a po-
lymeric lattice structure with FeH(CN)t- anions co-ordinated
through the nitrogens to other metal cations. Excess cations
required for charge neutrality can be accommodated in the large
interstitial sites. The idealized cubic structure is shown in Fig.3.3.
There exists the possibility of a kind of isomerism known as link-
age isomerism involving the interchange of the cations between
the sites co-ordinated to carbon and to nitrogen. This is equiva-
lent to reversing the C-N molecule, thereby changing the metal co-
ordination from [M(CN)61 n- to [M(NC)61 n-.
The four possible electronic configurations of iron in these com-
plex cyanides are low-spin iron(lI) or iron(III) co-ordinated to
carbon, and high-spin Fe 2+ and Fe 3+ co-ordinated to nitrogen or in
interstitial sites. High-spin Fe 2+ and Fe 3+ have characteristically
different chemical isomer shifts and are easily distinguished, but
empirical observations on simple hexacyanoferrate(II) and hexa-
cyanoferrate(III) compounds show that these low-spin configura-
tions will have similar chemical isomer shifts. However, a hexa-
cyanoferrate(II) usually gives a single-line resonance, whereas the
hexacyanoferrate(IIl)may show a substantial quadrupole splitting.
It is therefore possible to determine which sites are occupied by
iron, and also the cation oxidation states, in a compound with the
structure shown in Fig.3.3.
For example, the reaction product precipitated from solutions
of TiCl3 and H3Fe(CN)6 gives the single sharp absorption correspon-
ding to a hexacyanoferrate(II) [151, and is clearly Ti4+[FeII(CN)61 4 -
OM
Fig. 3.3 The crystal structure of the transition-metal hexacyanoferrates.
A mutual redox reaction has taken place. The parameters are

given in Table 3.4. TiCl3 or TiCl4 with HtFe(CN)6 give the identi-
cal product, showing that oxidation occurs with the former. How-
ever, COCl2 with K4Fe(CN)6 and K3Fe(CN)6 gives a hexacyano-
ferrate(II) and a hexacyanoferrate(III) respectively, Co~+[Fen(CN)6]4-
and Co~+[Feill(CN)6] ~-, and no redox reaction occurs.
An interesting technique which has been successfully applied to
these systems is that of selective isotopic enrichment. The reaction
of Fe2(S04h with K3Fe(CN)6 gives a hexacyanoferrate(III) com-
plex, Fe 3+[FeI1\CN)6j3-. The spectrum (Fig.3.4) has a complex
shape because components from the two types of iron overlap.
The two resolved peaks are due to a quadrupole splitting of the
Fe 3+ion. The low-spin component is not resolved, and it is not
easy to distinguish the oxidation state. However, if the prepara-
tion is repeated using K 3Fe(CN)6 made with 57 Fe enriched iron,
the absorption cross-section for the [Fe ill (CN)6j3- can be increas-
ed by a factor of about 40. The observed resonance spectrum is
then completely dominated by the low-spin component to the
effective exclusion of the Fe 3+ lines, and 8S shown in Fig.3.4 the
quadrupole splitting is then clearly seen, thereby confirming the
presence of [Fe ill (CN)6] 3-.
Table 3.4 Mossbauer parameters* for complex iron cyanides
Fell FeW Fe 3+
Compound A -------- (j A--------- (j A --------- (j

Ti[ Fe( CN)61 0 -0.01
C02[Fe(CN)61 "'0 -0.01
C03[Fe(CN)6h 0.85 -0.08
Fe[Fe(CN)61 0.43 -0.06 0.52 +0.50
Prussian blue 0 -0.08 0.57 +0.49
Turnbull's blue 0 -0.07 0.51 +0.49
* All data at 77 K in mm s-1 {j is relative to iron metal at room temperature.
The classic compounds 'Prussian blue' made from Fe 3+ and

[Fe(CN)6]4- and 'Turnbull's blue' made from Fe 2+ and [Fe(CN)6P-
pose the question as to whether they are essentially the same
material. There is the possibility of different formulations such as
Fe~+[FeII(CN)6b and Fe3+Fe~+[FeIl1(CN)6b, as well as the pos-
sibility of fast electron transfer to give an averaged electronic con-
figuration. Their M6ssbauer spectra are identical, and show quite
elegantly that both are in fact Fe~+[FeII(CN)6h [16]. Selective
enrichment has also been used successfully in this case. Prussian
blue prepared with enriched 57Fe2(S04h and Turnbull's blue pre-
pared with enriched 57FeCl 2 both give a spectrum dominated by
the Fe 3+ components. The formation of Turnbull's blue does not,
therefore, involve any breaking of cyanide linkages, but is merely
an electron transfer process. There is no substantial exchange
between the cations and anions, which would have resulted in the
57Fe apportioning between both.
Cyanide linkage isomerism has also been found to occur in the
solid state in an iron-chromium complex [ 17]. The mixing of
solutions of K3Cr(CN)6 and FeS04 generates a light brown preci-
pitate (A) which approximates in composition to Fe1.6 [Cr(CN)6](OH)o.2
with an additional amount of water which cannot be removed by
heating under vacuum. The M6ssbauer spectrum shows all the iron
to be present as Fe 2+ ions (the chemical isomer shift of -1.25 mm
s-1 is characteristic) and the two partially resolved doublets with
quadrupole splittings of A = 2.92 and 2.11 mm s-1 have intensities
in the ratio 0.6/1.0. This agrees with prediction from the formula
assuming 37% of the Fe 2+ to be in interstitial sites and the remain-
-....-
•..,....... 0#\ .............,.., •• ,.#.,••• -
~ "~'-', :'
".
'.
-
(0)
. ".:
II I I ,'.:
-4 -3 -2 -1 0 1 2 3
Velocity relative to Fe/(mm 5- 1 )
o ,.......... . ..
;." .~...~..... :.:....:....,.,...
'\,.
~ •••••0:...., ...;',.........;':.-::•••
". -,
-
,
".
.~
'" .......".
~c
.' -
o .' -
~5- (b) .'
-
..,~
...
-
. .
..... -
.....: I .
.J 1 I
-2 -1 0
Velocity relative to Fe/(mm s-1)
Fig. 3.4 57Fe Mossbauer spectra at 77 K for Fe[Fe(CN)6J. Spectrum (a)

is for the normal material. Spectrum (b) is for a preparation using 57Fe en-
riched K3Fe(CN)wand shows t?e quadrupole splitting (on an expanded.
scale) of the [Fe Ill(CN)61 3- amon. The Fe 3+ lines are now too weak to In-
fluence the spectrum. ([ 15 1, Figs. 9, 10).
der in the sites with co-ordination to nitrogen.

Mild heating generates a new darker brown complex (B). The
lines from interstitial Fe 2+ disappear (although the Fe 2+ in nitrogen
sites remains), and are replaced by a single absorption characteri-
stic of a hexacyanoferrate(II). The interstitial Fe 2+ atoms have
therefore interchanged with the Cr 3+ in the sites co-ordinated to
carbon.
Heating above 80°C gives a dark green complex (C) which
requires both oxygen and water for its formation. Its structure is
still enigmatic, although the indications are of low-spin Fell in car-
bon sites with oxidized high-spin Fe 3+ in nitrogen holes or inter-
stitial sites.
Reduction of (C) with hydrazine hydrate causes a reversion to
the original brown colour, but this product (D) is unstable in air.
The Mossbauer spectrum shows that all of the iron not in the carbon
sites has been reduced to Fe 2+, with a quadrupole splitting of -2.8
mm s-1 similar to the value already shown for interstitial sites.
The overall reaction scheme can be most conveniently summarized
as
(Carbon (Nitrogen (Interstitial

sites) sites) sites)
A Crill Fe 2+ 0.6 Fe 2+
B 0.4 Crill, 0.6 Fell Fe 2+ 0.6 Cr 3+
C [Fell (0.6 Fe 3+, Cr3+) ?]
D Fell Cr3+ 0.6 Fe 2+
The reduced complex is apparently the true isomerization product

with Cr 3+ in nitrogen holes, Fe 2+ in the carbon holes and the origi-
nal amount of Fe 2+ in interstitial sites. However, it is convincingly
shown that the linkage isomerism does not take place by a simple
cyanide inversion, but by a complicated series of cation migrations.
3.4 Conformations in organometallic compounds of Fe

Many large molecules have considerable freedom of movement
of the atoms within the structure, and several configurations may
exist which are energetically very similar. However, the energy
barrier between these may be very small so that interchange
between the various 'conformations' can take place very easily.
The cycIooctatetraene compound (COT)Fe(COh provides an
interesting example of the way in which structural conformation
can be established. The proton N.M.R. spectrum of the compound
in various solvents at room temperature shows all eight protons to
be equivalent; the obvious but incorrect inference is that the CgHg
ring is planar and bonding symmetrically to the Fe(COh moiety.
The X-ray crystal structure of the pure solid reveals asymmetric
bonding of the iron to a C414 section of the ring, which is in fact
14
biplanar (structure 14) [18]. It can be shown that the room tem-
perature N.M.R. spectrum results from a fast averaging effect by
cooling the solutions to about 120 K, where the spectrum becomes
more complicated as the molecular motion slows down. Clearly
the molecule can adopt more than one conformation but with
only a very small energy barrier between them which cannot pre-
vent their interconversion at room temperature. However, three
different interpretations of the nature of the low-temperature
'static' form of the complex have been given: a biplanar COT ring
acting as a I ,3-diene as in the solid state [19] ; a tub-shaped COT
ring acting as a I ,S-diene [20]; and a fast isomerization in which
the Fe(COh moiety moves between two equivalent structures in
the I ,3-diene and S,7-diene positions of a tub-shaped COT ring [21].
Although the Mbssbauer effect cannot be observed in a solute
molecule in the liquid phase, it is comparatively easy to freeze a
solution into a glassy matrix where the individual molecules are
not subject to the same nearest neighbour interactions as in the
pure solid. It is imperative of course that the solute should not
precipitate during freezing, and that the solvent should be both
chemically inert and form a glass rather than a well-defined crystal-
line matrix.
Freezing (COT)Fe(COh in several solvents is seen to have a
negligible effect on the Mbssbauer parameters (Table 3.S, [22]).
By way of contrast, the Fe(COh complex with cycloocta-l ,S-
diene and the 1,4-diene with norbornadiene show substantially
different parameters (Table 3 .S), and it is clear that the conjugated
1,3-diene structure found in the solid is retained in the low-
temperature solution.
There are also examples where conformational changes are seen
in solution. The compound (tr-CsHs)Fe(COhSnCI3 shows the
Table 3.5 S7Fe Mossbauer parameters in complexes with cyclooctatetraene
Compound Ll* 8*
(COT)Fe(CO>3 1.24 +0.05

in EPA solution t 1.16 +0.08
in nitrobenzene 1.21 +0.10
in cyclooctane 1.23 +0.10
in methyltetrahydrofuran 1.20 +0.09
cycloocta-l,5-diene Fe(CO}J 1.83 -0.03
norbornadiene Fe(CO}J 2.15 +0.03
* in mm s-1. 0 is with respect to iron metal

t EPA is 16/42/42 % v/v ethanol/l-propanol/diethyl ether COT =cyclooctatetraene.
same S7Pe and 1l9Sn spectra in both the solid state and dispersed
in a glassy polymethylmethacrylate matrix. The Mossbauer para-
meters are not influenced by the nature of the matrix. The S7Pe
spectra of [(1T-CsHs)Fe(COhhSnC12 are also uninfluenced by the
matrix, but the 119Sn spectrum shows four lines in the polymethyl-
methacrylate rna trix as against only two in the solid state [231.
This is illustrated in Pig.3.5. Assignment of the 4-line spectrum
into two quadrupole doublets can be made in more than one way,
but the ambiguity does not affect the prime result, that the 119 Sn
can have two different environments. Conformational effects have
also been reported in solutions of [(1T-C sHs)Fe(COhl SiC1 2CH 3 and
[(1T-CsHs )Fe(COhl SnC1 2 CH3 , and it would appear that in the pre-
sent instance two rotational conformers exist related by a 1200
rotation about an Sn-Pe bond (structures 15 -16). Structure 16
is that found in the solid state.
CI
15 16
100
99
98
97
100
98
3·0 4·0
Fig. 3.5 The 119Sn Mossbauer spectra of [(1T-CsHs)Fe(COh]2SnCI2 in the

solid state (a) and in a polymethylmethacrylate matrix (b). The extra lines
in the latter result from the presence of two rotational conformers in solu-
tion. ([23], Fig.2) .
3.5 Stereochemistry in tin cQmpounds

The study of organo-tin compounds has become an important area
of research in inorganic chemistry. The 119Sn Mossbauer spectrum
has been recorded for a large number of organo-derivatives, but
considerable difficulties have been experienced in interpreting
the observed values of the chemical isomer shift and the quadru-
pole splitting. For example, it is disturbing to find no detectable
quadrupole splitting in many compounds such as Me3SnPh and
Me3SnH which are clearly unsymmetrical about the tin atom,
whereas Me3SnCI and Me3SnOH for example show large splittings.
There appears to be a clear distinction between bonds to carbon,
and bonds to halides, oxygen, nitrogen or sulphur; only when the
co-ordinated ligands are a mix of the two types is a large splitting
observed. There is no obvious correlation with electronegativity,
and current evidence favours a difference in a-bonding characteri-
stics which is not yet fully understood. In consequence, the lack
of a quadrupole splitting is not necessarily indicative of a high
symmetry in the tin environment.
A major complication has been the realization that many com-
pounds which nominally have four co-ordinated ligands such as
Me2SnCI2, Me3SnF and Me3SnCN do in fact feature a higher co-
ordination (S or 6) in the solid state as a result of one or more
ligands being in a bridging position between two tin atoms. In
the few cases where X-ray structures are known, unexpected devia-
tions from regular geometry are sometimes found. Rationaliza-
tion of the Mossbauer parameters in terms of stereochemistry is
therefore not easy. A correspondence between geometry and the
magnitude of the quadrupole splitting has been formulated on the
basis of point-charge calculations (see p.66), but unfortunately
does not allow an unambiguous assignment. It has also been pro-
posed that polymeric bridged structures should have a significant-
ly higher Debye temperature (and recoilless fraction) than purely
monomeric structures, but again the differences are not sufficient-
ly distinct for reliable diagnostic application.
The two parameters of the chemical isomer shift and quadrupole
splitting are insufficient to make clear distinctions as to the nature
of the geometrical environment in many instances. Extra informa-
tion can be obtained by applying an external magnetic field with a
flux density of the order of 3 -S T to determine the sign of e2qQ.
Since Q is negative for 119Sn, e2 qQ is positive for Pz, d Z 2, d xz and
dyz electrons, and negative for Px, Py, d x 2 _ yl and d xy . The applied
field spectrum has a complex shape, but the sign can be determined
by visual comparison with the spectrum predicted by theoretical
calculation [24]. A typical example is shown in Fig.3.6. Although
it would be useful to know the value of T/, the method as applied
to 119Sn is not very sensitive to values of the asymmetry parameter
of less than O.S. However, a knowledge of the sign of e 2qQ alone
tells us something about the electrons producing q and can fre-
quently allow prediction of the stereochemistry.
Me2SnF2 and Me2SnCh which have a bridged-octahedral stereo-
chemistry in the solid state with trans-axial methyl groups both
give e2qQ as positive [2S]. The sign of e2qQ for orbitals bonding
(0)
.."
c:
",;;
"e
on
c:
"
L.
I-
(b)
-6 -4 -2 0 2 4 6
Velocity I (mm 5- 1)
Fig. 3.6 Computer simulated spectra for a polycrystalline sample giving a

quadrupole split 119Sn spectrum with e 2qQ/2 = 2.215 mm s-l, 11 = 0.6 and
a linewidth of 0.8 mm s-l: (a) in zero field; (b) with a magnetic field with a
flux density of 5 T (50 kG) applied perpendicular to the direction of observa-
tion. If e 2qQ were to be negative in sign, the spectrum would be the mirror
image of that shown.
in the z direction is also positive, consistent with the greater elec-

tron density being in the less-ionic Sn-C bonds, and since this situa-
tion may be anticipated to be general, we expect any similar trans-
octahedral structures to also show a positive sign. Me3SnNCS and
Me3SnOH which are known to have a trigonal bipyramidal geo-
metry in the solid state with axial bridging NCS and OH groups
respectively have a negative value of e 2qQ (consistent with asym-
metric a-bonding, rather than 1T-donation from the bridging groups,
or lattice effects). Ph3SnF and Ph 3SnCl also have a negative sign
for e2qQ and may be presumed to have similar 5-coordinate
structures.
Some of the available data are summarized in Table 3.6. As an
approximate guide to the magnitude of Ll in different geometries,
simple point charge calculations [26] give
RSnX 5 : trans-R2SnX4 : cis-R2SnX4 : R3SnX2: R3SnX

+2 +4 -2·' : -(3 to 4) : -2
These values are approximately those found in practice, although

differences in the R and X groups also cause large variations.
The method has been used for diagnostic determination of the
Table 3.6 119Sn Mossbauer parameters for organotin compounds
Sign of Ll Stereo-
Compound e 2qQ /(mm s-l) chemistry Ref.
(Me4Nh[ EtSnCI51 + 1.94 } { 26

octahedral
PrnSnC13.2 piperidine + 1.99 27
r
Me2SnF2 +
Me2SnCI2
CS2[Me2SnCl41
+
+
3.4
465
4.28
~ octahedral
28
26
K2[Me2SnF41 + 4.12 trans-alkyl 26
Me2SnCl2(pyOh + 4.1 29
Pr2SnCI2.2J}picoline + 3.99 27
Me3SnNCS 3.77 ,25
Me3SnOH 2.91 25
Ph3SnF 3.62 25
Ph3SnCI 2.51 trigonal 25
(Me4N)[Me3SnCI21 3.31 bipyramidal 26
(Me4N) [Ph3SnCI21 3.02 26
Et3SnCN 3.17 26
Ph3SnCl.piperidine 2.95 27
Me3SnC6H5 1.39 } { 26
tetrahedral
r
Ph3SnC6F5 0.97 26
Me2Sn(oxinh +
2.06
1.67 }
Ph2Sn(oxinh + octahedral 26
Ph2Sn(NCS)2.1, I O-phenanthro- cis-alkyl
line + 2.36 26
Pr~SnCI2.2 morpholine + 2.41 27
stereochemistry of new compounds [27]. Ph3SnCl.piperidine
for example has e2qQ negative in sign as predicted for a trigonal
bipyramidal structure with equatorial phenyl groups, and contrary
to prediction for equatorial Cl and piperidine ligands. Pr~SnCI2.2{3-
picoline evidently has trans-propyl groups. However, although
Pr~Sna2.2 morpholine has a quadrupole splitting approximately
half that of the {3-picoline compound, leading one to suspect a cis-
geometry, the sign of e2 qQ is positive in both cases and contrary
to the arguments given earlier in the chapter regarding cis-/trans-
isomerism. Moreover, other cis-complexes such as Me2Sn(oxin)z
for which an X-ray structure is available also show a positive value
of e 2qQ. The problem here is that chelating ligands cause substan-
tial distortion of the octahedral bond angles. The result is that
the principal direction of the electric field gradient tensor has to
be redefined and the value of 11 may be quite large. In this example
it is therefore partly fortuitous, although convenient, that the 2: 1
trans/cis ratio described earlier still holds.
3.6 Molecular iodine compounds

Several good examples are available of the determination of geo-
metry in an iodine compound. The covalently bonded molecule
12Cl6 has a planar bridged structure with two identical4-coordinated
iodine atoms (structure 17). The 129 1 resonance gives a complex
quadrupole pattern from the ~ ~ t decay. An example is shown on
p.Sl.. The derivative compound 12a4Br2 can have several hypo-
thetical structures (l8a-lSf). The 129 1 spectrum of this compound
[30] shows two superimposed quadrupole patterns of equal inten-
sity, indicative of two dissimilar iodine atoms. This eliminates
symmetrical structures b, c and d. Closer inspection reveals that
one iodine atom is apparently identical to iodine in 12C16, which
eliminates structures e and f and establishes structure a, in which
both bromine atoms are terminally substituting on the same iodine
atom, to be the correct one.
The quadrupole splitting of 129 1 in complexes of la and IBr
with pyridine bases, e.g. ICl.pyridine, has a negligible asymmetry
parameter [31 ]. This is good evidence for collinearity of the
CI-I-N bonds in these compounds.
The 129 1 spectrum of [IF 6] +[AsF6]- is a single sharp line (Fig.
3.7). This confirms the regular octahedral symmetry of the 1Ft
17
CI
CI
CI
"'1/ "'1/
/'"CI / "Br
Br Br
CI
'"
/1", /
/ /1",
CI",
CI
Br
CI
(0) (b)
CI",
/
CI
/1", /1",
"/ Br CI",
"/
/1", /1"
/
Br CI
Br CI CI CI Br CI
(C) (d)
CI Br CI CI"", Br
',/Br
'" '"
/
""'/ "'/ /1" / \
CI / CI / Br CI CI CI
(e) (I)
18
6·07.--...,---,.--;---.--,--.----,-,---,-.,.---,--.--...,---r-,--r-,-"'"1
5·84 , ............
,
f~ 5·61
0° "' ......
'"
~5-38 (0)
<3
5·15
4-92 '---:-8:--'----:6r---'----'4;---'----!2r--...L.----:-
O--'----::-2-J......~4-J......-6~--'----,8!:--'
3077 ...-, ,
.... ... . ~
.;;-
I
~
."
~H4
0
<..>
3069
-16 -12 -8 -4 0 4 8 12 16
Velocity f(mm 5- 1)
Fig. 3.7 The 1291 spectra of (a) the 1Ft cation and (b) the IF6" anion at
90 K. The single line in the former case is consistent with regular octahedral
symmetry, while the quadrupole splitting with a large asymmetry parameter
for IF6" indicates a stereo chemically active lone-pair of electrons. ([ 32],
Fig. O.
F
c~{>,
F
19
cation [32], and contrasts strongly with the large quadrupole

splitting found for the IF6" anion in CsIF6 . The spectrum in this
case is nearly symmetrical due to a large asymmetry parameter, 1/,
and indicates that the additional lone-pair of electrons is stereo-
chemically active. A suggested structure is derived from the penta-
gonal bipyramid of IF 7 by replacing an equatorial fluorine with the
lone-pair (structure 19).
Appendix
Quadrupole splitting in cis- and trans-isomers

The electric field gradient at the nucleus is derived from the second
derivative of the electric potential due to all the extranuclear
charges, and in particular to valence electrons and neighbouring ions.
At a naive level, a complex MABs is considered as comprising a
charge q A at distance R, and five charges qB at distance r from M.
MABS c;s-MA ZB4

The cis- and trans-isomers of MA2B4 can be treated similarly, and
the principal axis of the electric field gradient tensor is indicated
in each case in the figure as the z axis.
Considering MAB s , the potential at point s due to q A is given by
VA=~
S R - z
thence
_d_V_~ = __q_A-c:-
dz (R - z)2
and
2
d VA
_ _ s = ___
2qA ::-
dz 2 (R - z)3
The value as z -+ 0 becomes
which is the contribution of charge q A to Vzz . Including all terms

in the potential,
d2Vs 2qA 2qB 4qB(r 2 - 2z2)

-- - --~ + ----:=-
dz 2 (R - z)3 (r + z)3 (r2 + z2)S12
and as z -+ 0
A101ecular Structure 71
Vs = 2q A + 2qB + ~ + qB
(R2 + z2)1/2 (r2 + z2)1/2 (r + z) (r - z)
hence
. 2qA 2qB
V (ClS)=- - + - = - V (mono)
zz R3 r3 zz
Likewise for trans-MA2B4
Thus we have the result
2 Vzz(mono) = -2Vzz (cis) = Vzz(trans)

The result is valid as long as the bonds are basically similar in all
three compounds, and is independent of the magnitudes of q A and
qB. However, if for example there were a stronger interaction
between the A groups when trans- to each other, then deviations
may be expected.
References
[1] Gibb, T. C. (1970) In Spectroscopic Methods in Organometallic
Chemistry, Chapter 2, p.47, ed. W.O. George, Butterworths, London.
[2] Fannery, K., Kilner, M., Greatrex, R. and Greenwood, N. N. (1968)
Chen!. Comm., 593.
[3] Gibb, T. C., Greatrex, R., Greenwood, N. N. and Thompson, D. T.
(1967) J. Chem. Soc. (A), 1663.
[4] Clark, M. G., Cullen, W. R., Garrod, R. E. B., Maddock, A. G. and
Sams, J. R.(1973) Inorg. Chem., 12,1045.
[5] Herber, R. H., Kingston, W. R. and Wertheim, G. K. (1963) Inorg.
Chem., 2, 153.
[6] Greatrex, R., Greenwood, N. N., Rhee, I., Ryang, M. and Tsutsumi, S.
(1970) Chem. Comm., 1193.
[7] Dahl, L. F. and Rundle, R. E. (1957)J. Chem. Phys., 26,1751.
72 Principles of lIfossbauer Spectroscopy
[8] Erickson, N. E. and Fairhall,A. W.(1965)Ino~. Chern., 4, 1320.
[9] Wei, C. H. and Dahl, L. F. (1969) J. Arner. Chern. Soc., 91, l351.
[10] McDonald, W. S., Moss, J. R., Raper, G., Shaw, B. L., Greatrex, R. and
Greenwood, N. N. (1969) Chern. Cornrn., 1295.
[11] Cullen, W. R., Harbourne, D. A., Liengme, B. V. and Sams, J. R.
(1969) Inorg. Chern., 8,95.
[12] Berrett, R. R. and Fitzsimmons, B. W. (1966) Chern. Cornrn., 91;
(1967)1. Chern. Soc. (A), 525.
[13] Bancroft, G. M., Garrod, R. E. B., Maddock, A. G., Mays, M. J. and
Prater, B. E. (1970) Chern. Cornrn., 200.
[14] Fitzsimmons, B. W., Seeley, N. J. and Smith, A. W. (1969)J. Chern.
Soc. (A), 143.
[15] Maer, K., Beasley, M. L., Collins, R. L. and Milligan, W. O. (1968) J.
Arner. Chern. Soc., 90,3201.
[16] Kerler, W., Neuwirth, W., Fluck, E., Kuhn, P. and Zimmermann, B.
(1963) Z. Physik, 173, 321.
[17] Brown, D. B., Shriver, D. F. and Schwartz, L. H. (1968) Inorg. Chern.,
7,77.
[18] Dickens, B. and Lipscomb, W. N. (1962)J. Chern. Phys., 37, 2084.
[19] Kreiter, C. G., Maasbol, A., Anet, F. A. L., Kaesz, H. D. and Winstein,
S. (1966) J. Arner. Chern. Soc., 88,3444; Anet, F. L., Kaesz, H. D.,
Maasbol, A. and Winstein, S. (1967) 1. Arner. Chern. Soc., 89,2489.
[20] Cotton, F. A., Davidson, A. and Faller, J. W. (1966) J. Arner. Chern.
Soc., 88,4507.
[21] Keller, C. E., Shoulders, B. A. and Pettit, R. (1966) 1. Arner. Chern.
Soc., 88,4706.
[22] Grubbs, R., Breslow, R., Herber, R. H. and Lippard, S. J. (1967)
J. Arner. Chern. Soc., 89, 6864.
[23] Herber, R. H. and 9oscinney, Y. (1968)Inorg. Chern., 7, 1293.
[24] Gibb, T. C. (1970) J. Chern. Soc. (A), 2503.
[25] Goodman, B. A. and Greenwood, N. N. (1971)J. Chern. Soc. (A),
1862.
[26] Parish, R. V. and Johnson, C. E. (1971) J. Chern. Soc. (A), 1906.
[27] Goodman, B. A., Greenwood, N. N., Jaura, K. L. and Sharma, K. K.
(1971) J. Chern. Soc. (A), 1865.
[28] Goodman, B. A. and Greenwood, N. N. (1969) Chern. Cornrn., 1105.
[29] Fitzsimmons, B. W. (1970) 1. Chern. Soc. (A), 3235.
[30] Pasternak, M. and Sonnino, T. (1968)J. Chern. Phys., 48,1997.
[31] Wynter, C. I., Hill, J., Bledsoe, W., Shenoy, G. K. and Ruby, S. L.
(1969) 1. Chern. Phys., 50,3872.
[32] Bukshpan, S., Soriano, J. and Sharnir, J. (1969) Chern. Phys. Letters,
4,241.
CHAPTER FOUR
Electronic Structure and Bonding:

Diamagnetic Compounds
Once the molecular structure of a compound has been established,

it becomes possible to study the nature of the chemical bonding in
more detail. Mossbauer data now assume a more important role,
because once the number and disposition of the groups bonding to
a resonant nucleus are known, the hyperfine interactions can be
used to investigate the electron distribution in individual bonds.
The important hyperfine effects were introduced in Chapter 2.
The chemical isomer shift is a direct function of the s-electron den-
sity at the nucleus, but shows secondary induced effects due to
changes in shielding of the s-electrons by p-, d- or [electrons. It is
therefore sensitive to most changes of orbital occupation in the val-
ence shell of the atom. The quadrupole splitting is derived from
asymmetric occupation of the non-spherical p-, d- or I-electron
orbitals, and its magnitude and sign can be used to determine the
directional orientation of the asymmetry with respect to the chemi-
cal bonds. Diamagnetic compounds do not show a magnetic hy-
perfine splitting unless a large external magnetic field is applied.
The generation of a large internal magnetic field at the nucleus by
an unpaired electron will be described in the next chapter on para-
magnetic compounds.
The presence of the unpaired electrons in paramagnetic compounds
results in a wide variety of effects, many of which are strongly tem-
perature dependent. In contrast, the electronic properties of dia-
magnetic compounds are largely unaffected by temperature varia-
tion, resulting in a significant reduction in the amount of informa-
73
tion obtainable from the hyperfine interactions. The primary ob-
servable parameters become the chemical isomer shift and the quad-
rupole splitting, and these are both intimately linked to the covalent
bonding of the molecule.
Because of this distinction, it is convenient to discuss covalency
in diamagnetic substances in some depth before turning to some of
the more exotic phenomena associated with unpaired electrons. As
an introduction, it is instructive to consider, simply, the effects of
a change in the formal oxidation state of the resonant atom.
4.1 Formal oxidation state

Some intimation of the effect on the Mossbauer spectrum of a
change in the formal oxidation state of the resonant atom was
given by several of the examples in Chapter 3. A change in oxida-
tion state involves the addition or removal of a valence electron,
and manifests itself as a change in the chemical isomer shift. This
is true of both diamagnetic and paramagnetic electron configura-
tions. For example, the oxidation of tin(II) to tin(lV) formally
involves the loss of two 5s-electrons which will result in a large
change in the 119Sn chemical isomer shift. Similarly, the oxidation
of Fe 2+ to Fe 3 + involves loss of a 3d-electron and causes a reduc-
tion in the shielding of the outer s-electrons.
Accurate estimation of the changes in electron-density is diffi-
cult. The problems involved can be illustrated by considering the
different oxidation states of iron. Figures are available for several
free-ion configurations of iron, based on restricted Hartree-Fock
calculations by Watson [1, 2] ,and show the effect of a decreasing
number of 3d-electrons on the value of Il/Is(0)12. As may be seen
f
from Table 4.1, the largest effect is felt in the contribution from
11/I3sCO) 12. A progressive reduction in the number of 3d-electrons
causes a non-linear increase in the total s-e1ectron density.
Unfortunately, the value of the fractional change in the nuclear
radius in equation (2.4), oR/R, cannot be determined independently
of the chemical environment. It is therefore necessary to construct
an empirical calibration of the chemical isomer shift scale using the
observed shifts of compounds whose electron configurations are
known. In this way the sign of oR/R and its approximate magni-
tude are obtained. However, no known compound of iron can be
assigned the free-ion configuration 3d5 or 3d 6 because all Fe 3+ and
Electronic Structure and Bonding: Diamagnetic Compounds 75
Table 4.1 Electron densities at r = 0 for different configurations of an Fe
atom [1,2]
Electrons per
cubic Bohr 3dB 3d7 3d6 3d5 3d 6 4s 2
radius Fe+ Fe 2+ Fe 3+ free atom
ilhS<O) 12 5378.005 5377.973 5377.840 5377.625 5377.873

11/1 2s<. 0) 12 493.953 493.873 493.796 493.793 493.968
11/13.10) 12 67.524 67.764 68.274 69.433 68.028
11/14.10) 12 3.042
2 L
n
Il/Ins(O) 12 11878.9 11879.2 11879.8 11881.7 11885.8
Fe 2+ compounds show some measure of covalent character. The

classic work on the calibration of the 57Fe chemical isomer shift by
Walker, Wertheim and Jaccarino [3], although valid in principle, is
incorrect in detail because it neglects the considerable 4s-character
in the covalent bonding of the Fe 3+ compounds.
A more realistic approach is to carry out a full molecular-orbital
calculation on a model compound such as the FeF~- cluster, and
to equate the derived electron-densities with the observed shift [4].
A newer and alternative method is to isolate atoms and ions of the
element in an inert matrix such as solid xenon. Experiments which
have produced neutral iron atoms in the 3d64s 2 configuration and
Fe+ ions in the 3d6 4s 1 and 3d 7 configurations are described in
Chapter 10. However, in all instances the results are dependent
on the accuracy of the theoretical wave-functions used.
Despite these difficulties it is comparatively easy to determine
the relative change in s-electron density between oxidation states
and thence the sign of oR/R. Thus the Fe 2+ oxidation state gives a
chemical isomer shift at higher velocity than Fe 3+; Table 4.1 shows
that the s-electron density at the nucleus is smaller in Fe 2+, so that
oR/R for the l4.4-keV transition in 57Fe is negative in sign.
Although each oxidation state has a different chemical isomer
shift, within any given oxidation state the shift can vary from com-
pound to compound because of the effects of covalent character
already alluded to. The approximate ranges for iron, compiled
empirically from experimental data, are shown in Fig.4.l. This
figure also illustrates an important secondary consideration; name-
Fe(J) 5=~ 0
o FeW 5=~
W~ Fe (II) 5=2
D Fe(Il) 5=1
Fe(Jj) 5=0
Fe (III) 5= ~
Fe (III) 5 = ~
Fe(II1) 5=~
Fe (IV) 5=2
D Fe(IV) 5=1
D Fe (VI) 5=1
-1,0 -0'5 o + 0'5 + 1'0 +1'5 +2,0

Chemical isomer shift 6/(mm 5-')
(relative to iron metal)
Fig. 4.1 Approximate representation of the ranges of chemical isomer

shifts found in iron complexes. The values are related to iron metal at room
temperature, but do allow for temperature variations in the absorber. The
width of the individual distributions may be attributed to covalency, and as
a general rule the more ionic compounds have the higher shift. The most
common configurations are cross-hatched, and the diamagnetic covalent
classification includes carbonyls and ferrocenes etc. In most series the shift
decreases as spin-pairing takes place. Some of the less common configura-
tions may in fact extend beyond the limits shown. Note that the Fe(IV)
S = 2 subgroup refers to oxides with collective-electron bonding which may
account for their anomalous position as regards the Fe(IV) S = I compounds
which have a true low-spin configuration.
ly the effect of differing electron correlation. For example the

3d6 configuration of iron(II) can exist in high-spin (S = 2), inter-
mediate-spin (S = I), and low-spin (S = 0) forms. These differ sub-
stantially in their shielding effect on the s-electrons, and therefore
show distinct chemical isomer shifts. An increase in covalent charac-
ter may be expected to cause an effective decrease in 3d-shielding
due to the nephelauxetic ('cloud-expanding') effect; thus the s-
electron density increases and the chemical isomer shift decreases.
For the more ionic compounds, the degree of covalent character can
usually be estimated approximately from the chemistry of the atoms
or groups bonding to the resonant nucleus, so that the chemical iso-
mer shift can be predicted more closely than Fig.4.l would seem to
indicate.
The diagnostic use of the determination of the oxidation state
has already been indicated in Fig.3.l, and in the discussion of link-
age isomerism. Further examples will become apparent elsewhere.
It is now also clear that differences in electronic configuration and
covalent bonding produce characteristic effects. Similar arguments
hold for resonant nuclei other than 57Fe, although the distinctions
become less clear for elements with strong covalent bonding such as
iodine. For example, tin(II) compounds give a 119Sn resonant
absorption between +2.9 and +4.5 mm s-l (relative to BaSn03)
while the range for tin(IV) is -0.5 to +2.0 mm s-l. Europium(II)
compounds give a 21.6-ke V 151 Eu absorption in the region -13.9
to -10.9 mm s-1 (relative to EuF 3), compared to -0.2 to +0.9
mm s-l for europiumOII). The large difference between Eu 2+ and
Eu 3+ is due entirely to the reduced shielding of the s-electron core
upon removal of an [-electron, and indicates oR/R to be positive
in sign.
The foregoing discussion has been based solely on deductions
from the chemical isomer shift. In some instances where unpaired
electrons are present, the quadrupole splitting can also be used to
indicate the oxidation state. Examples of this are given in Chapter 5.
4.2 Iodine
The fortunate occurrence of suitable Mossbauer isotopes for the
contiguous series tin, antimony, tellurium, iodine and xenon has
greatly facilitated the study of chemical bonding in these elements.
Their compounds are diamagnetic, and the orbitals participating
most strongly in the bonding are the 5s- and Sp-orbitals, with some
contribution from 5d- in appropriate instances. The ability of
iodine to form compounds involving only one covalent bond to
another atom makes it possible to develop a quantitative interpreta-
tion of the Mossbauer parameters which is not always possible for
other elements. In this discussion of covalency, it is instructive to
consider iodine first, before tackling the more difficult problems
which occur elsewhere.
Data can be obtained from both the 27.7-keV 1291 and 57.6-keV
1271 resonances, the former giving better spectra (see Fig.2.1) but
with the inconvenience of being an artificial radioactive isotope.
Both levels are populated by the ~-decay of a tellurium parent, so
that chemical isomer shift data are usually quoted with respect to
the customary source matrix, zinc telluride. The 1291 excited state
spin of Ie = ~ and ground state spin of Ig = ~ are interchanged in
127 1, but otherwise the analysis of the respective quadrupole spectra
are equivalent. The energy-level scheme appropriate to 129 1 is
illustrated in Fig.2.3.
The chemical isomer shift for 1271 or 1291 can be expressed in
terms of the deviation from the closed-shell electronic configura-
tion 5s 25p6 of the 1- anion [5,6]. The nephelauxetic effect of
covalent bonding causes an electron to become less effective in the
hyperfine interactions. In particular, a 5p-electron in a covalent
bond has a very much smaller value of (r- 3 ) so that it appears to
be a 'hole' in the c10sed-shell configuration. This concept has al-
ready been well proven in the interpretation of nuclear quadrupole
resonance (N.Q.R.) data. If we define the number of 5s- and 5p-
electron 'holes' in the 1- configuration to be hs and hp respectively,
the chemical isomer shift with respect to I - can be written as
where K and yare constants. The parameter K in fact contains

oR/R and is therefore numerically different for 1271 and 1291. The
term -Khs represents the decrease in s-electron density at the
nucleus due to the loss of 5s-electrons, the proportionality being
assumed linear. The (2 - h s ) remaining 5s-electrons are then de-
shielded by a total decrease in the number of electrons of hp + hs .
This assumes that the 5s- and 5p-electrons produce an equivalent
shielding effect. In the event that hs is negligible, we have
The value of y has been determined by calculation to be 0.07 [7],

but the derivation of a numerical value for K requires a more cir-
cumspect approach.
The relative occupation of the 5s- and 5p-orbitals is not given
directly by the chemical isomer shift, but has to be derived in con-
junction with the quadrupole splitting to which hs does not contri-
bute. The relevant theory is a direct adaptation from nuclear quad-
rupole resonance spectroscopy [8]. The Townes and Dailey theory
relates the principal value of the molecular field gradient (eqrnol)
to the atomic electric field gradient from a 5pz-electron (eqaJ by
where Up is the effective 5p-electron imbalance. Up is defined in

terms of the 5p-electron populations in the x, y, and z directions,
Ux , Uy , and Uz by
This follows directly because if the major axis of the electric field
gradient tensor is defined as the z axis, then eq(pz) = -} eq(px, py).
The asymmetry parameter is given by
so that
For axial symmetry, Ux = Uy and T/ = O. The value of (e 2qat 127 Qg/h)

is known accurately from N.Q.R. work to be +2293 MHz, and it is
customary to quote both 1271 and 129 1 quadrupole coupling constants
in the same units [the appropriate conversions from a quadrupole
splitting in units of velocity, v, are obtained from (v/c)(E'Y/h) and
are: 1271 I mm s-1 == 46.46 MHz, 129 1 I mm s-1 == 32.58 MHz]. Up
can now be obtained directly by proportionality, and having cal-
culated Uz , Ux and Uy from e 2q 127 Qg and T/, hp can be derived as
Calibration of the chemical isomer shift is made by using data

so Principles of Mossbauer Spectroscopy
for the iodine molecule isolated in frozen solutions (see below) in
which hp = 0.99, and for the iodide anion in which hp = O. This
gives the equations
0(1291) = -S.2hs + 1.5h p - 0.54 mm s-1
0(1271) = +3.44h s - 0.56hp + 0.16 mm s-1
where in both cases the reference zero is the source matrix ZnTe.
Note the inversion of sign caused by the opposing signs of the
oR/R values.
It can be seen from these equations that, in addition to permit-
ting a determina tion of 5p-im balance in the equivalent manner to
N.Q.R. spectroscopy, the M6ssbauer spectrum also allows a deter-
mination of the 5s-orbital occupation. The two figures in combina-
tion give a much more detailed description of the covalent bonding.
The 129 1 spectrum of solid iodine at 100 K is shown in Fig.4.2
and illustrates the good resolution which can be obtained with this
isotope [9]. The large number of lines in the quadrupole split
spectrum allows an unambiguous determination of both the value
and sign of e2qQ, and also of the asymmetry parameter 17. The
iodine atom in solid iodine has e2q 127Qg/h = -20S5 MHz, but sur-
prisingly has a non-zero asymmetry parameter of 17 = 0.16. This
is unexpected for a linear molecule and is contrary to the naive
assumption that the bonding is purely of p-character. The results
are consistent with admixture of, for example, an s2p 4d configura-
tion into the s2p 5 state as a result of intermolecular interactions.
However, frozen solutions of iodine in hexane, CC4, and solid
argon show a similar spectrum with e 2q 127Qg/h ~ -2250 MHz but
here the asymmetry parameter is zero [ 10]. Clearly in the latter
case a closer approach to a true isolated 12 molecule is found, and
the calculated hp value of ~0.99 is in good agreement with the
predicted value for a free molecule of 1.00.
Having established the calibration of the isomer shift scale
using 1- and the 12 molecule, it becomes feasible to study the
chemical bonding in other compounds. Sample experimental data
are collected in Table 4.2. Some of the numerical values for Up
and hp differ from their original source because all data have been
recalculated using the latest calibration figures.
The value of e2q127 Qg/h in IBr is -2892 MHz, and since this
~
~
(';
.....
d
;::
Velocity/(mm ,-" ;::;.
- 2'13 -1·17 -1'30 -0·88 -0'47 - 0·05 0 + 0'37 +1·20
1 . 80 ir----------r----------r----------r----------,----------r-r--------~--------_,----------._--,
Vl
~
(';
'"="'. ~~. .... <.~.,
2"
4 ~
~1 ' 76 I:::l
;::
2
" ~
1/2
§'" 1·74 5/2 ! 3/2 ttl
8 ex . s -r--- o;::
'0 \
.2: 1-72 512 ~
!j 26·8 k.v 2 3 ~ 5 678 ~'
z
HO 1 1/2
7/2 3/2 e:,
g.s, scum obs \ 5/2
7/2 IS'
1·68 ~
o so 400
Channel number ~
~
(SOchann.l. 4 ·175 mm
= ,-'I .....
;::;.
Fig.4.2 The 1291 resonance in solid 12912 at 100 K. The line numbered 1
9
on the decay scheme is off the scale of the spectrum. It is usually ignored ~o
because of its low intensity. ([ 91, Fig. 3). >::
;::
~
00
Table 4.2 Mossbauer data for the 1291 resonance
Compound T 1291i(ZnTe) e 2q 127Qg h- 1 1/ Up hp hs Reference

K /(mm s-l) /MHz
12 (solid) 100 +0.82 -2085 0.16 +0.91 9

12 (inert matrix) 88 +0.94 -2250 0 +0.99 1.00 - 10
KI 80 -0.51 0 0 0.02 - 5
IC1 80 +1.73 -3131 0.06 +1.38 1.48 - II
IBr 80 +1.23 -2892 0.06 +1.26 1.18- II
ICN 80 +1.19 -2640 0.00 +1.15 1.15 - 12
CH31 85 +0.20 -1739 0 +0.76 0.76 0.04 13
CH2Ii 4.2+0.42 -1920
CHI3 85 +0.53 -2029 0 +0.88 0.88 0.03 13
CI4 85 +0.65 -2102 0 +0.91 0.91 0.02 13
SiI4 85 +0.26 -1335 0 +0.58 0.53 - 15
Gel4 85 +0.48 -1500 0 +0.65 0.68 - IS
Snl4 85 +0.43 -1364 0.00 +0.59 0.64 - 14
• values estimated from 1271 data.
is a linear molecule gives directly a value for Up of 1.26 [11]. If

the bonding involves only pz electrons, then hp = 1.26. Alternative-
ly, using the chemical isomer shift calibration with hs = 0 gives
hp = 1.18. The discrepancy is probably within the accuracy of
the calibration, but it is noteworthy that it corresponds to a value
for hs of only 0.01. We can therefore be confident that the s-
character in the bonding is very small, and that ~0.25 e- is trans-
ferred from the iodine to bromine. Similarly in ICN 0.15 e- is
transferred to the cyanide [12], and in ICI 0.48 e- is transferred
to the chlorine [10]. In both IBr and 10 there is a small asymmetry
parameter, and the observed discrepancies may well be the result
of intermolecular interactions in the solid.
The iodomethanes CH31, CH212, CHI 3 and CI4 are not expected
to involve 7T-bonding, and Up then gives hp directly [ 13]. The p-
electron transfer to iodine decreases in the order 0.24 e-, 0.16 e-,
0.12 e- and 0.09 e- per iodine as the number of substituent io-
dines increases, and at the same time the s-character in the bond-
ing decreases. The sensitivity of the data to the s-character of the
bond is shown in FigA.3 where the chemical isomer shift and qua-
drupole splitting values are plotted around the line calculated for
pure 5p-bonding. The deviations shown by the iodomethanes are
2r----------------------------------------------,
IC1~
•
12
I", 1
E
--
E
~
c:
N
2
...
-
.~
a
~ 0
CO
-1~--~----------~----------~----------~----~
o 1000 2000 3000
e 2 q127Qg / MHz
Fig. 4.3 Values for the chemical isomer shift and quadrupole coupling con-
stant in iodine compounds plotted around the line calculated for pure 5p-
bonding. Note the substantial deviations in the iodomethanes due to an
s-character of less than 5%.
substantial, despite the fact that CH31 has only about 4% s-

character.
Compounds in the series CI4, SiI 4, Gel4 and Snl4 [14, 15] all
feature essentially pure Pu bonding and show no direct evidence of
appreciable s-character or p1r-character in the bonding; the amount
of electron density transferred to iodine is 0.09, 0.47, 0.32 and
0.36 e- respectively.
Compounds which show chemical isomer shifts outside the range
spanned by 1- and 12 are usually (if perhaps unwisely) interpreted
by means of the same calibration. However, the large differences
between for example 1Fs (+3.00 mm s-l), 1F7(-4.56), 1F6 (-4.68)
and IF;; (+2.45) make the estimated orbital occupations very sen-
sitive to the numerical values of the constants in the calibration.
It is also noteworthy that there is no obvious correspondence bet-
ween the chemical isomer shift and the formal oxidation state.
The large amount of s-character in the bonding of IF 7 and IF6
causes apparently anomalous shifts. The difficulties arise mainly
because of the large number of orbitals directly involved in bonding
when the co-ordination is high. This problem becomes more acute
in the elements immediately preceeding iodine.
4.3 Tellurium and antimony

The 35.5-keV resonance in 125Te has an observed range of chemi-
cal isomer shifts which is less than the naturallinewidth of 5.02
mm s-l, and the quadrupole splittings from the Ig = 4- -+ Ie = i
transition are also poorly resolved. Comparatively few data for
inorganic compounds are available, but several important generaliza-
tions have emerged [16]. The octahedral tellurium(IV) complexes
such as TeCl~-, TeBr~- and TeI~- show the most positive chemical
isomer shifts. They are unique in having a stereochemically inac-
tive pair of non-bonding electrons which may be presumed to have
considerable 5s-character, and to be highly contracted towards the
Te nucleus. Other tellurium (IV) compounds such as square pyra-
s,
midal TeF pyramidal TeXt in TeCl4, TeBr4 and TeI4, TeO~
anions in tellurites, and bridged square-pyramidal TeF 4 all have a
stereochemically active pair of non-bonding electrons with pre-
sumably greater 5p-character. These all show a more negative
chemical isomer shift, and the predicted reduction in the lone-pair
contribution to the s-electron density at the nucleus is compatible
with 8R/R being positive. The tellurium(VI) compounds such as
Te(OH)6 and TeO~- show the most negative shifts because the 5s-
orbital is now involved in covalent bonding, the nephelauxetic
effect of which causes a further reduction in the s-electron density
at the nucleus. Unfortunately, it is not possible to make reliable
ab-initio molecular-orbital calculations for compounds in this area
of the periodic table, and the relative occupations and hybridiza-
tion of the various orbitals involved in bonding can only be
estimated.
Neighbouring 121 Sb shows large chemical isomer shifts in the 37.1-
keY resonance with a range of about 20 mm s-1, and oR/R is
negative. The two common oxidation states SbOIl) and Sb(V)
are comparatively distinct with the latter absorbing at more positive
velocities. Some typical values of the M6ssbauer parameters are
collected in Table 4.3. The most negative chemical isomer shifts
found in antimony are for the octahedral [SbmCI6P- anions. Here
Table 4.3 121Sb Mossbauer parameters for antimony compounds
Compound S/(mm 8- 1)
relative to
121Sb in Sn02
Antimony halides (at 80 K) [17]

K3[ SbC16] -17.8
CS3[Sb06] -17.7
[NH4]3[SbC16] -16.8
[Co(NH3)6] [SbC16] -19.3
-19.2
Rb2[SbC16]
-2.4
Rb[ SbC16] -2.3
Organo-antimony(III) (at 4.2 K) [18]
Ph3Sb -9.4 +17
Organo-antimony(V) (at 4.2 K) [18]
Ph3SbF2 -4.69 -22.0
Ph3SbC12 -6.02 -20.6
Ph3SbBr2 -6.32 -19.8
Ph3SbI2 -6.72 -18.1
Ph4SbF -4.56 -7.2
Ph4SbC1 -·5.26 -6.0
Ph4SbBr -5.52 -6.8
P~SbI (at 80 K) -5.6 ?
Ph4SbN03 -5.49 -6.4
Ph4SbC104 (at 80 K) -5.9 ?
again the lone-pair has a high Ss-character and is sterically non-

active [17] , and with the negative sign for SR/R results in the large
negative shifts. The values are sensitive to the nature of the cation,
there being a large difference between for example the [NH4]+ and
[Co(NH3)6] 3+ salts. The caesium and potassium salts are both
known to be distorted from octahedral symmetry in the solid
state, and the evidence supports a small involvement of the 5s-
orbital in the bonding scheme to cause a change in the s-electron
density at the nucleus. No such cation effects were found in the
isoelectronic tellurium compounds [16]. The compound Rb2SbCl6
shows two distinct absorptions [17] at -19.2 mm s-1 from Sb m
and at -2.4 mm s-1 from Sb Y and should be reformulated as
Rb4[SbillSb VCI 12 ]. This is a further example of the determination
of formal oxidation states.
In compounds where there is a quadrupole splitting, the complex
pattern from the Ig = t -+ Ie = t transition allows, in principle, the
determination of both e 2qQ and 1/, but in many cases the resolu-
tion is such that only a single asymmetric envelope is seen and only
an approximate value of e 2qQ can be obtained.
Data for some simple organo-antimony derivatives can be inter-
preted readily in terms of the covalent bonding [18]. Triphenyl-
antimony, Ph3Sb, is believed to be pyramidal with a stereochemi-
cally active lone-pair of electrons. It shows a substantial quadru-
pole splitting with e 2qQg positive. Since Qg is known to be nega-
tive, the implication is that there is an excess of electron density
in the z-direction (along the pyramidal axis), i.e. the lone-pair
contains considerable pz character.
Triorgano-antimony(V) halides all have large negative quadru-
pole splittings (Table 4.3), indicative of excess p-electron density
in the equatorial plane. The halogens are apical along the axis of
the trigonal bipyramid in all positively characterized examples.
As p-electron density is withdrawn along the z-axis, the quadru-
pole splitting increases in the order Ph3SbI2 < Ph3SbBr2 <
Ph 3Sb0 2 < Ph 3SbF 2. At the same time s-electron density is also
withdrawn to the axial ligands, and the chemical isomer shift
becomes less negative. Thus the Sb-X bonds have both p- and s-
character.
Compounds of the type Ph4SbX also show significant variation
in parameters with change in X, and clearly favour covalent bond-
ing of the halogen rather than a Ph4Sb+ cationic structure. A pos-
sible exception to this is the perchlorate which has the most nega-
tive shift. If they are all isostructural with P~SbOH which has an
apical OH group, then the same interpretation already given to the
Ph3SbX2 series can be applied.
4.4 Tin
Some of the difficulties encountered in using the 119 Sn resonance
to study molecular compounds were referred to in Chapter 3. In
particular, the failure of many unsymmetrical tin(lV) molecules to
show a detectable quadrupole splitting of the Ie = ~ excited state
is both unexpected and still unexplained quantitatively. The con-
fusion caused by uncertainties in the stereochemistry of many
organotin compounds, and our poor knowledge regarding the ori-
gins of the electric field gradient at the tin nucleus have resulted
in the extensive use of empiricism. However, the problems are no
greater than with any of the other spectroscopic techniques, and
it is gratifying to observe the way in which Mossbauer spectroscopy
has drawn attention to many fundamental issues in the understand-
ing of the chemistry of tin, which in turn it is helping to solve.
Compounds in the tin (II) oxidation state are known to feature,
with few exceptions, a pyramidal bonding arrangement of three
ligands about the tin. Early theories regarding the origin of the
electric field gradient predicted that the sign of e 2qQ in these
compounds could be either positive or negative. However, recent
measurements on SnF 2, SnO, SnS, Sn3(P04h, SnC204, NaSnF3,
NaSn2FS, SnS04, Sn(HC02h, Sn(CH3C02h and K2Sn(C204h.H20
in applied magnetic fields [19, 20] have shown that e2 qQ is posi-
tive in every case. As Q is known to be negative, the sign of e2qQ
is consistent with an excess of electron density in the pz orbital
(for which q is negative, see p.35). Thus the lone-pair has a sig-
nificant pz character which becomes the dominant contribution
to the electric field gradient because of tighter binding to the nuc-
leus. However, neither simple covalent nor point-charge arguments
have been able to predict successfully the magnitude of the gradient.
A salutory warning as to the dangers inherent in the interpreta-
tion of small differences between related compounds is provided
by data for the trihalogeno complexes, [SnXYZ] - where X,Y,Z =
Cl, Br, I. The chemical isomer shifts for the [BulN] + salts of these
compounds are correlated in Fig.4.4 with the average Mulliken
electronegativity of the ligands, XM, and a linear relationship is
found [21]. A similar correlation exists between the chemical iso-
mer shift and the quadrupole splitting. The electric field gradient
is influenced by the polarity of the tin-ligand bonds and increases
with increasing polarization, presumably because the pz character
of the lone-pair becomes in greater imbalance with respect to
the covalent bonding orbitals. The trend in quadrupole splitting
values is MSnI 3 ) = 0.79 mm s-l < A(SnBr3) = 1.02 mm s-l
< A(SnCIJ) = 1.13 mm s-1. This does not necessarily imply that
the pz character has increased.
An unexpected and startling observation is the large difference
in both Mossbauer parameters between the [Bu~N]+ salts and cor-
3·0
Z·9
-;
..""
~
>
Z'8
Cl BrI
" Br~
Z'6
Z.5 L..-_-"-_----'-_ _-'---_-L...._-...L _ _-'---_-'-_-...L_----'

~ H H ~ H
BI{mm 5- 1)
Fig. 4.4 A correlation of the chemical isomer shift in tin(II) compounds

of the type [SnXYZ]-, where X, Y, Z = CI, Br, I, with the average Mulliken
electro negativity of the ligands, XM. Note the dependence on the nature
of the cation.
responding [Ph4As] + and [Et4N]+ salts. In fact the chemical iso-

mer shift difference between [Bu~N] Sn03 and [Et4N] SnCl3 is
greater than between [Bu~N] SnCl3 and [Bu~N] SnI3. While the
vibrational spectra of the [Bu!lN]+ salts in solution are essentially
the same as in the solid, this is not the case with the [Et4N] + salts.
The [Ph4As]+ salts appear to be intermediate. One can only con-
clude that there is a substantial solid-state interaction within the
[Et4N]+ salts, although whether this is between the anion and
cation, or between two anions has not been ascertained. It seems
possible that many difficulties experienced in rationalization of
data for tin(II) compounds are a result of nominally weak inter-
molecular interactions having a disproportionately large influence
on the lone-pair of electrons.
A chemical isomer shift-electronegativity correlation has also
been shown for tin(lV) compounds. The analogous [SnX4Y2]2-
anions (X, Y = CI, Br, I) also give a linear correlation [22, 23] .
Although any quadrupole splitting in the unsymmetrical [SnX4Y2]2-
species is unresolved, a small degree of line broadening is seen which
is proportional to the electro negativity difference in X and Y. This
implies that a small electric field gradient is in fact present. The
corresponding fluorides appear to be anomalous. However, it is
interesting to note that the (CI02hSnF6 salt has a large splitting
of 1.01 mm s-1 as well as an unusually positive chemical isomer
shift for an SnF~- salt [24]. A more general spectroscopic study
of several fluoride salts has indicated that anion-cation interactions
are the rule rather than the exception.
The tetrahalides Sn~ (X = Cl, Br, I) in which the tin is tetra-
hedrally co-ordinated also show a linear correlation of the chemical
isomer shift with electronegativity, as do the oxyhalides SnOX2
(X = CI, Br, I). The slope of the line is different for each series,
and in the case of the oxyhalides is of opposite sign. The behaviour
observed is strongly dependent on stereochemistry, and results
from a subtle balance between the 5p- and 5s-character in the vari-
ous bonds and the shielding effect of 5p-electrons on the 5s-elec-
trons. The importance of 5d-character in this context cannot be
assessed at the present time, other than that the effect on the quad-
rupole splitting would seem to be less important. For any series
of compounds where the co-ordination and geometry are the same
for all members, and one of the ligands is systematically varied, a
gradation of properties can be expected which allows the bonding
characteristics of unfamiliar ligands to be assessed in relation to
the series.
The factors governing the Mossbauer parameters of organotin
compounds can be particularly obscure. For example in any series
RxSn~_x ex =0-4, R =alkyl or aryl, X = halogen) the chemical
isomer shift and quadrupole splitting are both considerably greater
when x = 2 or 3. There is no obvious correlation between the shift
and the nominal degree of substitution as one would expect if the
major factor were the electron withdrawing power of the groups
bonded to tin. Thus in the series Me4Sn, Me3SnBr, Me2SnBr2,
MeSnBr3 and SnBr4 the values for the chemical isomer shift and
quadrupole splitting are 5 = 1.21, 1.38, 1.59, 1.41, 1.15 mm s-1
and A = 0,3.20,3.41, 1.91,0 mm s-1 respectively.
It was shown in Chapter 3 that the magnitude of the quadrupole
splitting in tin(lV) compounds does show a certain degree of cor-
relation with the stereochemistry. Nevertheless, a direct determi-
nation of the sign of e 2qQ is of great assistance in determining the
origin of the p-electron imbalance, and enables an assignment of
the co-ordination number and stereochemistry in many cases (see
Chapter 3 for a fuller discussion). Although the use of an external-
ly applied magnetic field to determine the sign is a comparatively
recent development, such data as are available show convincingly
that the electric field gradient originates primarily from asymmetric
occupation of the a-bonding orbitals, rather than from 'IT-bonding
to bridging groups or indirectly from lattice effects. The relative
importance of p- and d-electrons in a-bonding remains unknown,
although it may be presumed that the contribution of d-electrons
to the electric field gradient will be comparatively small.
The effects of distortion from regular geometry are probably
greater than has been generally assumed. A good example is given
by Ph2SnCl2 which has a distorted tetrahedral geometry in the so-
lid state [25]. The mean bond angles are Cl-Sn-CII 00°, CI-Sn-C
107° and C-Sn-C 125.5° with a mean Sn-Cl bond distance of 2.346
A (234.6 pm). The third-nearest chlorine neighbour is at 3.77 A,
so that the structure cannot be described as polymeric. The 119 Sn
quadrupole splitting in the pure solid is 2.80 mm s-1, but a differ-
ent value is found in crystalline and glassy matrices made by freez-
ing solutions of Ph2SnCl2 [26]. In a non-basic solvent the quadru-
pole splitting decreases by about 0.2 mm s-1 to -2.61 mm s-1
and it is thought that this value is more representative of an iso-
lated unrestrained molecule. Basic solvents give an increased split-
ting of -3.5 mm s-1. Presumably in this case two solvent mole-
cules co-ordinate to the Ph2SnCl2 to give a distorted trans-octa-
hedral complex with phenyl groups in the axial positions.
The anomalous sign of e 2qQ in cis-Me2Sn(oxin)z referred to in
Chapter 3 is undoubtedly a result of the large distortion from
pseudo-regular geometry (the C-Sn-C bond angle is 111°). In the
case of Me2SnCl2 the stereochemistry is effectively a badly distort-
ed tetrahedron, but the crystal packing is such that it approxi-
mates to a polymeric chain with trans-methyl groups and bridging
chlorine atoms [27]. It therefore seems likely that many of the
apparent anomalies in the magnitude of the quadrupole splittings
for series of related compounds can be directly attributed to distor-
tions from pseudo-regular tetrahedral, trigonal bipyramidal or
octahedral stereochemistry.
One of the few tin compounds where a more quantitative descrip-
tion of the bonding has been possible is [(1T-(:::sHs)Fe(COhhSnCI2.
The X-ray structure is known, and the geometry of the Fe2SnCl2
unit has C2 symmetry. The 119Sn spectrum in an applied magnetic
field is shown in Fig.4.5, together with computer simulations for
various values of the asymmetry parameter 1'/ [28] ,and it can be
seen that 1'/ is approximately 0.65 with e2qQ positive. It is pos-
sible to represent the contribution of each ligand to the electric
field gradient using a point charge approximation in which the
effective value of eq for the Sn-Fe bond is represented as [Fe] etc.
The molecular geometry is shown in FigA.6. The Fe-Sn-Fe plane
is perpendicular to the Cl-Sn-Cl plane, and therefore the three
-6 -4 -2 0 2 4
Velocity I(mm s-\ )
Fig. 4.5 The Mossbauer spectrum of [(1r-CSHS)Fe(CO)212SnCl2 at 4.2 K

with a magnetic field of flux density 5 T applied perpendicular to the direc-
tion of observation. The solid curves are computed predictions for different
values of 1/. ([28), Fig. 1).
Fe
Cl
Fig.4.6 The molecular geometry of [(1T-CSHS)Fe(COhl 2SnCl2.
principal axes of the electric field gradient tensor lie in these

planes. A priori it is not known which axis is the principal z axis,
so the three axes are labelled i, j and k. Then the relationships
can be derived
y..
II
= -2[Fe] + 2 (3 sin 2 ~
2
- 1) [Cll
Vkk=2(3sin 2 ~-l) [Fe] -2[Cl]
The angles are O! = 128.6 0 and {3 = 94.1 o. The direction and sign of
Vzz is a complex function of the ratio R = [Fe] /[Cl]. The quadru-
pole splitting in a single crystal of the compound shows unequal
line intensities (see p.4D) which establish that the principal axis, z,
is along k. In combination with the observed value for the asym-
metry parameter this leads to a unique value for R of 1 .2. It may
therefore be concluded that the more electronegative halogen
atoms withdraw a greater proportion of Sp-electron density from
the tin atom in a-bonding, and therefore make a smaller contribu-
Electronic Stmcture and Bonding: Diamagnetic Compounds 93
tion than the iron to the electric field gradient.
This compound is also of interest in being one of several for
which the crystal structure shows unusually short Sn-Fe and long
Sn-Cl bond distances. It was originally thought that this was the
result of intermolecular interactions or crystal-packing effects, but
the similarity of the Mossbauer spectra of the pure solid with those
of frozen solutions in inert rna trices confirms that the structure
reflects the internal bonding of the molecule itself.
4.5 Covalent iron compounds

The transition metals differ from the main-group elements in that
the valence d-orbitals are only partially involved in chemical bond-
ing, and therefore retain a degree of identity which is particularly
useful. Many compounds have one or more unpaired electrons,
and being of essentially d-character, these cause characteristic
effects which are considered separately in the next chapter. Where
the covalency is strong, this proves to have a substantial influence
on the Mossbauer spectrum.
In considering covalent effects in more detail, it is convenient to
consider compounds of iron(II) in the low-spin (S = 0) configura-
tion where paramagnetic effects are absent. Nevertheless, the dis-
cussion is directly applicable to for example the chemical isomer
shift in iron(lII) low-spin (S =~) complexes. A typical compound
is potassium hexacyanoferrate, ~[Fe(CN)6] .3H20, which has
octahedral co-ordination in the [Fe(CN)6]4- anion. The chemical
isomer shift of -0.04 mm s-1 (with respect to iron metal) at room
temperature contrasts strongly with the typical values of +1.0 to
+1.3 mm s-1 for octahedral high-spin iron(II) compounds (S = 2).
An octahedral ligand-field splits the five-fold degeneracy of the
3d-electrons into an upper doublet (e g ) and a lower triplet (t]g)
level. In the S = 2 compounds, the electron configuration is
t1 e 2 , but for S = 0 it becomes t~g, i.e. a fully occupied triplet
le~ef. In discussing the cyanide complexes, it is particularly impor-
tant to consider the effect of 1T-bonding between the t2g levels and
the 1T-bonding and 1T*-antibonding orbitals of the CN- ligands.
With this in mind, the chemical isomer shift observed will be com-
pounded from several factors:
(1) a direct contribution from 4s-electrons, and
(2) indirect contributions from 3d-electrons by shielding the s-
electrons, which may be considered in terms of:
(a) the purely ionic or non-bonding effect of a 3d n configuration.
(b) covalency with the ligands in which filled ligand orbitals
donate an effective total of n2 electrons to the 3d-orbitals.
(c) covalency with the ligands in which the metal 3d-orbitals
donate an effective total of n3 3d-electrons to the 1T* ligand
orbitals.
Thus the effective number of 3d-electrons is
In molecular orbital terms, one can say that the metal t'4: (3d)
levels and the 1T and 1T* ligand orbitals all have t2g symmetry and
combine to form molecular orbitals with the same symmetry. The
eg (3d) orbitals have the same symmetry as ligand a-orbitals, so
that a large part of n2 will be a-donation from the ligands.
Calculation of the relative importance of these effects is diffi-
cult, but one interesting example concerns a comparison of
[Fe(CN)6]4- and [Fe(CN)6j3- made by Shulman and Sugano [29].
They estimated many of the molecular orbital parameters from
electron spin resonance (E.S.R.) data, and found that the degree of
1T back-donation as typified by n3 is about one electron greater in
[Fe(CN)6]4- than in [Fe(CN)6j3-, while the term n2 remains
about the same. Therefore the value of neff is similar in both com-
pounds, and explains why they have almost the same chemical
isomer shift. Although the hexacyanoferrate(II) nominally has an
extra 3d-electron, its effect is largely nullified by an additional
delocalization of the t2gorbitals to the ligands. However, it is im-
portant to realize that the two configurations are distinct. The
full molecular orbital schemes still feature an unpaired-electron in
the hexacyanoferrate(III), whereas the hexacyanoferrate(II) is
diamagnetic.
The Mossbauer spectrum of a hexacyanoferrate(II) is only a
single line because of the regular octahedral geometry, but the same
is not true of substituted cyanides where a quadrupole splitting
will be seen. A simple example is the so-called nitroprusside ion,
[Fe(CN)s(NO)]2-, which shows a splitting of 1.705 mm s-l [30] .
The ground state configuration for the axial C4v symmetry is
(dxz • dyz)4(dxy )2. The (dxz. dyz) orbitals are strongly delocalized
by 1T-donation to the 2p-orbitals of the nitrosyl group, and there is
also some 1T-donation from the dxy orbital to the equatorial
cyanides.
The electric field gradient may be presumed to originate mainly
from the unequal contributions of these 3d-orbitals. If, as with
the hexacyanoferrate(II), we define effective occupation numbers
nxy etc., it becomes possible to write the value of q in the quadru-
pole splitting e2 qQ in terms of the value for a single d-electron of
(4/7)(1 - R)(,-3)/(41TEO) as
q = (1
-41TEO
- - (r- 3) + -7 nxy
- R) [4 2
- - (n
7 xz
+n
yz
) ]
The value of nxy (incorporating the delocalization to the cyanides)
can be estimated to be approximately 1.74 via the E.S.R. orbital
reduction factor in [Fe(CN)6]3-. Using an estimated value for
the quadrupole splitting from a single d-electron of -4.0 mm s-l,
and the experimental value of 1.705 mm s-l, it is possible to der-
ive (nxz + nyz) = 2.63. If the delocalization to the nitrosyl is
indeed greater than to the cyanide, then this figure for the occu-
pation of dxz and dyz can be used to derive an approximate molecu-
lar orbital
t/J xz = 0.81 dxz + 0.58 1T* (NO)
The resultant occupation figure for the 1T* (NO) orbital of 34%
is an upper limit, and is in good agreement with independent esti-
mates of about 25%.
The nitroprusside ion has an unusually low chemical isomer shift
for an iron(II) cyanide complex (-0.258 mm s-l relative to iron
metal at 298 K). The effective 3d-popUlation, neff, is only -4.4,
and there is an unusually high 4s-population of -0.5. These two
factors combine to produce the exceptionally low shift observed.
The related complex ion [Fe(CN)s(NO)]3- has an extra electron
in an orbital of essentially d;- character, and in frozen solutions
its parameters are a quadrupole splitting of Ll = + 1.90 mm s-l and
a chemical isomer shift of 0 = +0.1 mm s-l [31 ]. In this case the
value of q is given by
(1 - R) -3 4 2 4
q= - - - ( r >[ +"7nxy -"7(n xz + n yz ) -"7nz2]
47T€Q
One would expect the value of nxy to be little changed from that in
the nitroprusside, and consideration of E.S.R. data leads to an esti-
mate for n z 2 of 0.6. The experimental value for the quadrupole
splitting can then be used to derive the values (n xz + nyz ) '='='- 1.6,
nxy ~ 1.9, nz 2 ~ 0.6, giving neff '='='- 4.1. Although these values are
approximate, the large decrease in (n;xz + nyz) compared to the nitro-
prusside reflects the large increase in 7T-donation to the NO+ ligand
as a result of a formal reduction from Fe(II) to Fe(l). This offsets
the increase in electron density about the central atom associated
with the addition of a non-bonding electron.
It will now be apparent that it is difficult to estimate the relative
effects of a- and 7T-bonding from the comparatively small amount
of information obtainable from a single compound. A more fruit-
ful approach is to compare data for large numbers of related com-
plexes. In Chapter 3 it was shown that cis- and trans-isomers of
low-spin iron(II) compounds may be distinguished by a 1 :2 ratio
of the quadrupole splittings. From data for large numbers of low-
spin iron(II) compounds, the chemical isomer shift has been found
to be an approximately additive property of the number and type
of the ligands [32, 33]. Any contributions to the shift from either
the second-order Doppler shift or zero-point motion are ignored
on the grounds that for comparative purposes they are self-
cancelling in similar compounds. Similarly, the quadrupole split-
ting, ~, can be expressed in terms of partial quadrupole splittings
such that for the three stereochemistries FeAB s , cis-FeA2B4 and
trans-FeA 2B4
~AB5= 2(PQS)A = 2(PQS)B
~cis = -2(PQS)A + 2(PQS)B
~trans = 4(PQS)A - 4(PQS)B
The justification for these equations can be seen by considering the

Appendix at the end of Chapter 3. The partial chemical isomer
shift (PCS) and partial quadrupole splitting (PQS) values for a num-
ber of simple monodentate ligands are given in Table 4.4.
Table 4.4 Partial chemical isomer shift (PCS) and partial quadrupole split-
ting (PQS) data for low-spin 6-coordinate iron(II) compounds
Ligand PCS/(mm s-1)(a) PQS/(mm s-1 ) (b)
NO+ -0.20 +0.02

H- -0.08 -1.04
RNC 0.00 -0.69
CN- 0.01 -0.84
SnCl] 0.04 -0.43
NCS- 0.05 -0.49
NCO- 0.06 -0.50
NH3 0.07 -0.51
Nj"" 0.08 -0.38
H2O 0.10 -0.44
CI- 0.10 -0.30
Br- 0.13 -0.28
I- 0.13 -0.29
(a) based on stainless steel at 295 K as reference. Calculated values for /j can be con-
verted to iron metal as standard by subtracting 0.09 mm s-l.
(b) based on -0.30 mm s -1 for a-as an arbitrary value.
The compound cis-Fe(SnCI3h(RNC)4 would be expected to show

A = -2 X (-0.43) + 2 X (-0.69) = -0.52 mm s-1 (observed 0.50)
and {) = 2 X (0.04) + 2 X (0.00) = +0.08 mm s-1 (observed +0.11)
relative to the reference point of 57Co in stainless steel. The partial
shift and splitting values can be related to the bonding characteri-
stics of the ligands. This can be illustrated by considering cyanides
of the type [Fe(CN)s L]n- where L is a variable ligand. For inst-
ance the chemical isomer shift increases in the order NO+ < CO
z
< CN- < Ph3P< SO~- < NO ~ NH3 ~ Ph 3As ~ Ph3Sb, cor-
responding to a decrease in the 1r back-donation to the ligand anti-
bonding orbitals.
The partial chemical isomer shift value will decrease with increas-
ing a-bonding from the ligand to hybrid d 2sp3 orbitals on the
metal. The direct increase in 4s-density is greater than the increas-
ed shielding due to 3d-augmentation, and oR/R is negative. An
increase in 1r back-bonding to the ligand from the t2g (3d) orbitals
decreases the shielding of the inner s-orbitals and also causes a
decrease in chemical isomer shift. The partial quadrupole split-
ting will become more negative with increased a-bonding and more
positive with increased 1r back-bonding.
PCS = -k(a + rr)
PQS = +K(1T - a)
where k and K are constants of proportionality. Thus the best a-

donor (H-) and the best rr-acceptor (NO+) give the most negative
PCS values, but are at opposite extremes of the PQS scale.
This type of treatment is very useful in assessing the relative im-
portance of a- and 1T-bonding for a series of ligands, but cannot be
applied to series of compounds which differ in co-ordination or
geometry, or show non-additive effects in the bonding such as a
strong 'trans-effect'. For this reason the covalent iron carbonyl
derivatives do not readily lend themselves to this kind of interpre-
tation, although where series of related compounds with similar
stereochemistry can be found, it is still possible to make empirical
correlations with the a- and rr-bonding characteristics of particular
ligands.
One particularly interesting organometallic derivative is
(rr-CsHshFe, bis(pentahapto-cyclopentadienyl)iron(II) commonly
known as ferrocene, which gives a large temperature-independent
quadrupole splitting of 2.37 mm s-l. The corresponding cation,
(rr-CsHshFe+, contrasts in that splitting is almost absent. This was
suspected and later confirmed by Collins [34] to be the result of
a chance cancellation of the various contributions to the electric
field gradient tensor. Several types of theoretical bonding scheme
are available for the ferrocene molecule. A crystal field calcula-
tion by Matsen predicted a negative sign for e 2qQ, whereas mole-
cular orbital schemes by Ballhausen (using Watson's wavefunctions
for iron) and by Shustorovich and Lyatkina (using Slater wave-
functions) both predict e2qQ to be positive. Collins determined
the sign experimentally by the first application of the magnetic
perturbation method to be reported in the literature, and found
that it was positive.
The molecular orbital scheme, adopted from Dahl and Ballhausen,
is shown in Table 4.5. The molecular orbital wavefunctions (1/1)
are taken as linear combinations of metal (f1) and ligand (p) orbi-
tals. The effective occupation of the metal atomic orbitals is then
given by the squares of the metal coefficients. The 4s-orbitals do
not contribute to the electric field gradient tensor, and there is
good evidence to show that any contribution from the 4p-orbitals
Table 4.5 Molecular-orbital calculations for ferrocene
Wave
function
No. of (metal
elec- coeff.
Y net Ll*
14PO>
net Ll*
13dO>
trons
>. 1/I(e2g) = 0.898J.L(3d~ 4 0.8064 -3.222

....
Ol)
Cl)
~
+ 0.440p(e!g)
Cl)
Ol)
1/I(etg) = 0.454J.L( 3d 1) 4 0.2061 +0.411
.S + 0.891p(etg)
'"c<:S 1/I(a 19) = J.L( 3dO) 2 1.000 +2.000
-
....
Cl)
u
~
1/I(e1u) = 0.591J.L(4P1) 4 0.3493 -0.693
+ 0.807 p(eiu)
1/I(a2u) = 0.4 71J.L( 4PO) 2 0.2218 +0.444
+ 0.882p(a2u)
1/I(a1g) = 0.633J.L(4s) + 2 0.4007
+ 0.774p(a1g)
-0.249 -0.811
(1T - CsHshFe e2qQ/2 = -0.24914pO> - 0.81113dO>

(1T - CsHshFe+ e2qQ/2 = -0.24914PO> - 0.00513dO>
* The contributions to the quadrupole splitting in units of that from a 14po> or 13do>
electron. Both quantities are defined as negative.
will be small in comparison to that from the 3d, because of the

greater radial expansion of the former. Each individual orbital
generates a contribution to the field gradient and to e2 qQ along
the symmetry axis of the molecule, and in the last two columns of
the Table these are normalized in terms of that from the 14po>
function and the 13do> function (which will be of the order of
-3.6 mm s-1 per electron). We can now see that in ferrocene one
predicts a quadrupole splitting of -0.81 I 13d o> = -+2.9 mm s-1.
Removal of an electron from the 1/1 (e2g) orbitals (to generate the
cation) will effectively remove 0.8064 of a 13do> electron and
reduce the splitting to -0.00SI3d o> = ~+0.2 mm s-1. This is in
excellent agreement with observation.
The same principles are of course applicable to the paramagnetic
compounds, but the metal-ligand interaction in for example a high-
spin Fe2+ compound is much weaker than in a low-spin Fe(II)
compound. The observed s7Fe chemical isomer shifts at 4.2 K (in
mm s--1 with respect to Fe metal) in for example the anhydrous
iron(II) halides are FeF2 (1.48), FeCl2 (1.16), FeBr2 (1.12),
Fel2 (1.04). These values correspond to an increase in s-electron
density due to reduced 3d-shielding and an increase in covalency
along the series. This is confinned by the crystal structures which
change from the rutile type for FeF2 to the Cda2-type for FeCl 2
and the Cdlrtype for FeBr2 and FeI 2. All feature a distorted 6-
coordination to the halogen. In a tetrahedral ion such as [FeCI4] 2-
however, the shift of 0.90 mm s-l is lower than in an octahedral
co-ordination because of a substantial increase in covalency. This
effect is also found in iron oxides and sulphides.
The influence of covalency in paramagnetic compounds has not
been studied systematically. Perhaps this is because of the greater
interest which attaches to phenomena associated with the unpaired
electron. These are considered in detail in the following chapter.
References
[1] Watson,R. E.(1960) Phys. Rev., 118,1036.
[2] Watson, R. E. (1960) Phys. Rev., 119,1934.
[3] Walker, L. R., Wertheim, G. K. and Jaccarino, V. (1961)Phys. Rev.
Letters, 6,98.
[4] Duff, K. J. (1974) Phys. Rev. B, 9, 66.
[5] Hafemeister, D. W., Pasquali, G. de and Waard, H. de (1964) Phys Rev.,
135, BI089.
[6] Perlow, G. J. and Perlow, M. R. (1966) J. Chern. Phys., 45, 2193.
[7] Ehrlich, B. S. and Kaplan, M. (1969) J. Chern. Phys., 50,204l.
[8] Das, T. P. and Hahn, E. L. (1958) Nuclear Quadrupole Resonance
Spectroscopy, Supplement I of Solid State Physics, Academic Press
Inc., New York.
[9] Pasternak, M., Simopoulos, A. and Hazony, Y. (1965)Phys. Rev., 140,
A1892.
[10] Bukshpan, S., Goldstein, C. and Sonnino, R. (1968) J. Chern. Phys.,
49,5477.
[11] Pasternak, M. and Sonnino, T. (1968) J. Chern. Phys., 48, 1997.
[12] Pasternak, M. and Sonnino, T. (1968)J. Chern. Phys., 48, 2009.
[13] Bukshpan, S. and Sonnino, T. (1968) J. Chern. Phys., 48,4442.
[14] Bukshpan, S. and Herber, R. H. (1967) J. Chern. Phys., 46,3375.
[15] Bukshpan, S. (1968) J. Chern. Phys., 48,4242.
[16] Gibb, T. C., Greatrex, R., Greenwood, N. N. and Sarma, A. C. (1970)
J. Chern. Soc. (Aj, 212.
[17] Birchall, T., Della Valle, B., Martineau, E. and Milne, J. B. (1971)
J. Chern. Soc. (Aj, 1855.
[18] Long, G. G., Stevens, J. G., Tullbone, R. J. and Bowen, L. H. (1970)
J. Arner. Chern. Soc., 92,4230.
(19] Gibb, T. C., Goodman, B. A. and Greenwood, N. N. (1970) Chern.
Cornrn., p.774.
[20] Donaldson, J.D., Fihnore, E. J. and Tricker, M. J. (1971) J. Chern. Soc.
(A), 1109.
[21) Goldstein, M. and Tok, G. C. (1971) J. Chern. Soc. (A), 2303.
[22] Herber, R. H. and Cheng, Hwan Sheng (1969) Inorg. Chern., 8,2145.
[23] Clausen, C. A. and Good, M. L. (1970) Inorg. Chern., 9,817.
[24] Carter, H. A., Qureshi, A. M., Sams, J. R. and Aubke, F. (1970)
Canad. J. Chern., 48,2853.
[25] Greene, P. T. and Bryan, R. F. (1971)J. Chern. Soc. (A), p.2549.
[26] Herber, R. H. (1973) J.Inorg. Nucl. Chern., 35,67.
[27] Davies, A. G., Milledge, H. J., Pux1ey, D. C. and Smith, P. J. (1970)
J. Chern. Soc. (A), 2862.
[28] Gibb, T. C., Greatrex, R. and Greenwood, N. N. (1972) J. Chern. Soc.
(A), p.238.
[29] Shulman, R. G. and Sugano, S. (1965)J. Chern. Phys., 42,39.
[30] Danon, J. and Iannarella, L. (1967) J. Chern. Phys., 47, 382.
[31] Raynor,J. B. (1971)J.Inorg. Nucl. Chern., 33,735.
[32] Bancroft, G. M., Mays, M. J. and Prater, B. E. (1970)J. Chern. Soc.
(A), p.956.
[33] Bancroft, G. M., Garrod, R. E. B. and Maddock, A. G. (1971)J. Chern.
Soc. (A), 3165.
[34] Collins, R. L. (1965)J. Chern. Phys., 42, 1072.
CHAPTER FIVE
Electronic Structure and Bonding:

Paramagnetic Compounds
In Chapter 4 it was shown how the chemical isomer shift and

quadrupole splitting in diamagnetic compounds are dependent on
the oxidation state and covalent bonding of the resonant atom.
The chemical isomer shift in paramagnetic compounds is governed
by the same principles, and it was therefore convenient to consider
the influence of covalency in high-spin Fe 2+ and low-spin Fe(III)
compounds in Section 4.5. However, the presence of an unpaired
electron can result in magnetic and quadrupole interactions which
are intimately related to the orbital state of the atom. It is there-
fore appropriate to consider these separately. The major part of
this chapter is concerned with iron chemistry, but the ideas expres-
sed are also applicable to other elements, and some of these are
discussed separately at the end of the chapter.
5.1 Quadrupole interactions

The quadrupole interaction in a diamagnetic compound is deter-
mined mainly by asymmetry in the covalent bonding of the reso-
nant atom. As a rule, the first excited electronic level is not acces-
sible by thennal excitation (being at a much higher energy than kT
at room temperature where k is Boltzmann's constant and T is the
temperature). The electronic configuration and the orbital occu-
pation are therefore not sensitive to change in temperature, and
in consequence the quadrupole splitting is almost independent of
temperature. This behaviour is found in for example tin and low-
102
Electronic Structure and Bonding: Paramagnetic Compounds 103
spin iron(II) compounds. However, in many paramagnetic ions
there is a substantial degree of thermal excitation to electronic
levels which are only of the order of kT above the ground state.
The resulting temperature dependence of the orbital occupation
produces characteristic effects on the quadrupole splitting.
The most familiar example of this is the high-spin Fe 2+ ion. The
free-ion configuration 5 D (tigei) signifies a single 3d-electron out-
side a spherical half-filled shell. In a ligand field of cubic sym-
metry such as regular octahedral, the t2g levels remain degenerate
(as do the eg) and there is no finite electric field gradient. How-
ever, if the symmetry is lowered to trigonal or tetragonal, further
degeneracy is removed. The sixth electron is in the appropriate
lowest-lying unfilled state, and now generates an electric field
gradient at the iron nucleus. A ligand field of axial symmetry (i.e.
a tetragonal distortion of the octahedron) splits the t2g state into
a lower d xy singlet and an upper dxz,yz state. The sixth electron
will be in the d xy level and (from Table 2.1) will generate a quad-
rupole splitting in proportion to Vzz/e = q = +~(r-3}(l - R)/(4rre.o).
If the doublet lies lowest, then the electron is in dxz,yz and gives
q = _~(,-3) (1 - R)/( 4rre.o). If the distortion is trigonal, i.e. an
elonga tion along the (111) axis, the singlet ground state corresponds
to dz2 and gives q = -~(r-3}(l - R)/(4rre.o).
A singlet d xy ground state is found for example in
[Fe(H20)6](NH4S04h for which e2 qQ is positive, and a singlet
d z 2 state is found in the trigonally distorted [Fe(H20)61 SiF 6 for
which e 2qQ is negative. One of the few known examples of a
doublet ground state is in trigonally distorted FeC03, and the low-
temperature quadrupole splitting value of +2.06 mm s-1 is about
half that of -3.61 mm s-1 for the fluorosilicate. All three com-
pounds have a distorted 6-coordination to oxygen. The quadru-
pole splitting originates not from asymmetric covalency with the
ligands, but from the sensitivity of the orbital state of an essen-
tially non-bonding electron to the geometrical environment. This
effect provides a powerful means of studying small distortions
from regular co-ordination.
At an infinitely high temperature, all the levels of the t 2g multi-
plet would be equally populated by thermal excitation, and the
electric field gradient would once again be zero. At intermediate
temperatures, a partial degree of thermal excitation will cause a
partial cancellation of the electric field gradient, so that the quad-
rupole splitting should decrease with rise in temperature. A full
mathematical treatment of the temperature dependence, (an out-
line of which is given below) was first given by Ingalls [I] .
The effects of the ligand field on the free-ion wavefunctions
are calculated using a perturbation Hamiltonian
where the successively smaller terms are Vo the axial field, VT the
tetragonal (or trigonal) distortion, VR the rhombic distortion, and
'!I.l.§ the spin-orbit coupling. This last term has a considerable
effect on the quadrupole splitting at low temperatures because
it causes admixture of excited states into the ground state. The
free-ion 5D state of Fe 2+ has S = 2 and L = 2 so that there are
(2S + 1)(2L + 1) = 25 energy states in the ligand field. Some typi-
cal energy level schemes from such calculations are illustrated in
Fig.5.1 for the lowest 15 levels only.
If the energy separation of a given excited level is Ej , its relative
population is given by Boltzmann's theory as e-Ej/kT for a ground
state population of unity. The thermal averaging takes place
within a time-scale less than the Mossbauer event (ca. 100 ns for
57Fe), so that the total resultant quadrupole splitting will be a
statistically averaged quantity. Each excited level has an electric
field gradient represented by V'z= eqj and f/j' and the quadrupole
splitting at any temperature T is given by
[(t qje-Ej/kT) 2 +~ (~ f/jqje-Ej/kT) 2 ] 1(2
t:.T = ~e2Q
This summation is made over all 25 levels. In the event that VT is

much greater than vR and '!I.L.S, the latter terms may be ignored.
The problem then reduces to one of thermal excitation from a
pure dxy ground state to the dxz,yz excited state at an energy Eo,
and results in a very simple expression for the temperature depend-
ence of
Electronic Structure and Bonding: Paramagnetic Compounds lOS
,,
,,...---2
,',,-----
dxr ,/,,-----
1''--'';';';;'''---''\ /"
,II , , - - - - -
2
1
I \\ /I~---
~'11/~
dlz, yz I
,
\ {,
,,,
,....--""-{
I,I,
\ 1
\ 1 \
,,
\ I
\ \ ___--'-_~I
dyz I
I,
( \\,'----
\
\
\
1------ 2
\
\
\
\
\
\
\
\
\
\
" ----2
\'
\\
"',1----
Cubic Axial Rhombic
field + distortion + distortion 1\
1\
2
'------1
Axial distortion
+ spin - orbit coupling
Fig. 5.1 The effect of axial (tetragonal), rhombic, and spin-orbit coupling
perturbations on the t2g energy levels of the Fe 2+ ion. The level scheme on
the right is a combined axial distortion + spin-orbit coupling, and the num-
bers indicate the degeneracy of the states.
(1 - e-Eo / kT )
I::. T = 1::.0 (1 + 2e-Eo / kT)
where 1::.0 is the quadrupole splitting due to a single 3d-electron in

the d xy state (-4 mm s-I).
This function is approximately valid except at low temperatures
where the spin-orbit coupling causes a reduction from the value
1::.0. The effects of the axial field and spin-orbit coupling when the
full calculation is made [2] are illustrated in Fig.S.2. The curves
represent the observed quadrupole splitting I::.T/I::. O = F for a
t2g- egseparation of 10 Dq = 10000 cm- 1 (=1.9878 x 10- 19 J
== 119.7 kJ mol-I) and a spin-orbit coupling of A = -80 em-I.
The numerical figures refer to the total splitting of the t2g levels,
1.0E===~~==::::::::::::===------ co
1500
Ground
state
d xy
900
600
300
oL__________~==========::==::====~;;~~1~50~ 60 0-
150
Ground
state
600
dxr,yr
-0·5L-------1.-------L------L-----------l co
o 200 400 bOO 800
T/K
Fig. 5.2 Calculated values for the quadrupole splitting flT/ flO = F in units
of the value for a 3d-electron. The curves are drawn for different values of
the splitting of the t2g levels (in cm- 1), and for a spin-orbit coupling para-
meter of A = -80 cm~1.
Eo in cm -1, by the axial field. A positive value of F corresponds

to a singlet ground state and a positive value for e 2qQ, while a
negative value corresponds to a doublet ground state and a nega-
tive value for e 2qQ of approximately half the magnitude. Note in
particular the large reduction in flT at zero temperature which is
a direct result of the spin-orbit coupling.
The magnitude of the quadrupole splitting is very sensitive to
the axial (or trigonal) distortion and the spin-orbit coupling, but
not to the rhombic distortion which has a greater effect on the less-
populated excited states. The presence of a rhombic distortion
does however cause the asymmetry parameter to be non-zero and
temperature dependent, and an example of this can be found in
[Fe(H20)6] (N14S04h where the large rhombic field causes state-
mixing with the nominal dxy ground state so that 11 = 0.3 at 4.2 K
and 0.7 at 300 K [3].
Although the above theory is precise, there are a number of addi-
tional factors which must also be considered. Firstly, the value
of ~o is dependent on the value of <r- 3 ) for a 3d-electron, and this
effectively decreases in the presence of covalent bonding. The
upper limit for ~o is about 4.0 mm s-l in ionic compounds, but is
not known accurately. A second factor which is not so easy to
estima te is the contribution to the electric field gradient tensor
from charges external to the central atom, the so-called 'lattice'
term. Ingalls considers the lattice correction to be of opposite
sign to that of the 3d-orbital contribution [I], but direct lattice-
sum calculations from known crystal structures have suggested
that there need be no direct relationship between the two [4]. The
lattice-term is neglected in many instances on the grounds that it
appears to be usually less than 10% of the total field gradient,
but there have been recent reports of some Fe 2+ compounds with
unusually large quadrupole splittings {e.g. 4.65 mm s-l at 300 K
in Fe(NH2CSNHNH 2hS04 which is also unusual in having two geo-
metrically distinct iron sites, both with six co-ordination, and
with identical M6ssbauer parameters [5]} and this has raised
fresh conjecture on the problem.
A third major aspect is the possibility of temperature depend-
ence of the ligand field splittings and the lattice-term as a direct
result of thermal expansion. A major attempt has been made to
include all these corrections in the analysis of data from F eF 2
between 78 K and 1023 K [6]. The experimental data and the
final computed fit are shown in Fig.53. However it is generally
acknowledged that such sophisticated theoretical models are
unlikely to provide a unique interpretation of the experimental
data.
Of the other common paramagnetic oxidation states of iron,
tetrahedral high-spin Fe 2+ and octahedrallow-spin Fern can be
treated in an analogous manner. The E ground state of Fe 2+ in
a cubic tetrahedral field is also split by an axial distortion, but
the spin-orbit coupling is only effective in second-order terms
and can be neglected. The resulting thermal population of the
d z2 and d X 2_y2 levels results in a limiting value of the quadrupole
splitting at zero-temperature which is not reduced by the spin-
orbit coupling and is independent of the axial field. Thus the
observed variation [7] in the values at 4.2 K for FeCli- (3.27
mm S-l), FeBrl- (3.23), Fe(NCS)i- (2.83), and Fe(NCSe)i-
(2.73) can be accounted for solely in terms of a difference in
3.0.-----~--l
'", 2·0
-
E
E
~
c
Ia
~
1·0
o 50
Temperature I K
1000
Fig. 5.3 The temperature dependence of the quadrupole splitting in FeF2

with a theoretical curve from a model incorporating thermal lattice expan-
sion. ([ 6] , Fig. 2).
lattice tenn or a change in (r- 3 ) due to covalency. The tempera-

ture dependence relationship is particularly simple in this instance,
being
=L\o tanh (Eo/2kT)

where Eo is the separation of d z 2 from d x 2_ y 2.
Octahedral Fern has a 2T ground state (t1.) and corresponds to
a single electron-hole in an otherwise cubic triplet level. It is mathe-
rna ticalIy similar to octahedral Fe 2+ except that because S = ~ there
are only six sub-levels to consider in the t2g multiplet, and the spin-
orbit coupling parameter is much larger. The available data con-
finn the expected temperature dependence, but the effect of the
greatly increased covalency on the value of (r- 3 ) has not been
established satisfactorily. Similar arguments can be used for other
elements. However in the second and third-row transition metals
the spin-orbit coupling constant is much larger, and may result in
an almost temperature-independent and much reduced quadrupole
splitting below 300 K.
Not all paramagnetic configurations of iron show a temperature
dependent quadrupole splitting. The high-spin Fe 3+ ion is in a 6S
state (t1-ei) which has spherical symmetry and therefore no
inherent electric field gradient. Any observed splitting arises
solely from charges external to the cation, either by direct con-
tribution or by indirect polarization. Point charge summations
of the lattice contribution correlate approximately with the
observed splitting, but tend to be unrealistic because the major
contributions arise very close to the central atom where the point
charge model is particularly inadequate.
The principles outlined above can also be applied to the less
common configurations of iron, but with varying results. For
example the quadrupole splitting in phthalocyanine iron(II) is
2.70 mm s-1 at 4.2 K and only decreases to 2.62 mm s-1 at 293 K.
The iron has two unpaired electrons (S = I), being in an inter-
mediate spin-state appropriate to the square-planar geometry.
The lack of orbital degeneracy and low-lying excited states in this
configuration prevents significant temperature dependence. In
consequence, very little information is obtained from the quadru-
pole splitting in isolation. However, in such cases it is often pos-
sible to make more positive deductions from a combination of
quadrupole and magnetic hyperfine data.
5.2 Magnetic hyperfine interactions

It was shown in Chapter 2 that the presence of a magnetic hyper-
fine field causes line splitting, and it was intimated that this field
could be intrinsic to the resonant atom. If the atom has one or
more unpaired d- or [-electrons, it is intuitively obvious that their
interaction with the s-electrons of parallel spin will be different to
that with the s-electrons of opposed spin. This interaction results
in a slight imbalance of spin density at the nucleus [8], thereby
generating a magnetic field with a flux density which can be ex-
pressed as
where J1.0 is the permeability of a vacuum (41T x 10-7 kg m s-2 A-2)

and J1.B is the Bohr magneton (9.273 x 10- 24 A m 2). lI/It(O) 12 re-·
presents the electron density at the nucleus with spin parallel to
the magnetic moment, and 11/1.\.(0) 12 the antiparallel spin-density.
B therefore has the units of tesla (T) or kg s-2 A-1 (1 T = 1 Wb
m- 2= 104 G where Wb denotes Weber and G denotes Gauss). Bs
is usually referred to as the Fermi contact term, and is directly pro-
portional to the total spin-imbalance of the s-electrons with a
magnitude which can be as high as 200 T. It is not, however, the
only possible intrinsic contribution to the field.
If there is a non-zero orbital magnetic moment on the resonant
atom, then there is a further contribution to the flux density
where (,-3) is the expectation value in m- 3 of l/r 3 for the electrons

producing the moment (L z ). The orbital moment (L z ) is normally
quenched in for example Fe 2+ ions, but spin-orbit coupling can in-
duce a flux density of
where (Sz) is the expectation value of the spin moment, and gz is

the electronic Lande splitting factor. In high-spin Fe2+ compounds
where S = 2 the orbital contribution, BL, is opposite in sign to Bs
and varies from about +20 to +60 T.
A third major contribution to the magnetic flux density arises
from a dipolar interaction of the spin moment of the atom with
the nucleus, which for axial symmetry is
Electronic Structure and Bonding: Paramagnetic Compounds III
BD =- (~~) J1B (,-3> <3 cos 2 8 - 1> (Sz>
This expression is closely related to that for the quadrupole split-

ting e 2 qQ. For example, a high-spin Fe 2+ ion has a total spin of
S = 2; it has an electric field gradient Vzz = eq given by writing
equation 2.19 in alternative form as
so that
Estimated values for BD are usually in the range 0-8 T, and BD is

of course zero in cubic symmetry.
The total magnetic flux density can therefore be written as
B=Bo+Bs+BL+BD
where Bo is the flux density arising from an externally applied

field (not necessarily collinear with Bs etc.). From the foregoing
discussion one might expect that all compounds with unpaired
valence electrons would show a magnetic splitting, but this is not
so. The Hamiltonian in equation 2.9 contains I.B as a vector pro-
duct, and the Mossbauer event in for example 57Fe lasts for about
100 ns. In the absence of an externally applied field, the electro-
nic spin which generates B will normally be undergoing electronic
spin relaxation, and if the relaxation rate is far faster than 100 ns,
the time-average <B) is zero. This situation applies for example to
paramagnetic Fe 2+ (S = 2) ions, so that the Mossbauer spectrum
shows only a quadrupole interaction. The relaxation rate is much
slower in Fe 3+ (S = ~) ions and in some of the rare-earth ions, and
in these cases one can see magnetic splitting. However, as these
also frequently embrace the intermediate case where both the
relaxation and Mossbauer processes operate on the same time-
scale, discussion of these is deferred until Chapter 6 on dynamic
effects.
A further possibility is that the spins on neighbouring ions are
strongly interacting so as to have a preferred direction, and that
this leads to ferromagnetic or antiferromagnetic coupling. The
nucleus sees a time average (Sz> of the electronic spin S which is
non-zero. The relaxation rates for (Sz> in such cooperative pheno-
mena are generally very much slower, and magnetic hyper fine
splitting is seen without dynamic effects. The internal magnetic
field at any given site in a magnetically ordered material is general-
ly proportional to the magnetization at that site. The temperature
dependence of the magnetization usually follows a Brillouin func-
tion, which may be written as
MT =Bs (~. Tc MT)

Mo S+ 1 T Mo
where
Bs(x) 2S +
=- 1 (2S + 1)x
- coth - - 1coth (1)
-- x
2S 2S 2S 2S
MT is the magnetization at a temperature T, Mo is the saturation

magnetization, and Tc is the Curie temperature. The function is
illustrated in Fig.5A for a value of S = 1 in terms of the reduced
variables MT/Mo and T/Tc· The magnetization and hence the in-
ternal field decrease with temperature rise and become zero at the
Curie or Neel temperature. Whereas the bulk magnetization is an
average effect of all magnetic sites in the crystal, the Mossbauer
spectrum is capable of distinguishing the magnetization at more
than one type of crystal site. This is particularly useful in the study
of antiferromagnetic compounds where the bulk magnetization is
unduly sensitive to ferromagnetic impurities.
Although several Mossbauer isotopes have been used to study
magnetic ordering in intermetallic alloy systems (see Chapter 8)
the only nuclide which commonly shows magnetic ordering in co-
ordination compounds is 57Fe. The simplest case to consider is
the Fe 3 + (S =~) ion, in for example FeF 3 which is antiferromagne-
tic below 363 K. The S-state ion is in a nearly octahedral environ-
ment, and above 363 K shows a single resonance because (Sz> time-
averages to zero [9]. Below this temperature, the co-operative
ordering makes (Sz> non-zero so that the flux density due to the
1 · 0 r - - - - - -_ __
c::
:3 0.6
:a'"...
c::
""
'"
E
t:::>
0·4
~
""
0·2
o 0·2 0·4 0'6 0'8 1·0

Reduced temperature TITe
Fig. 5.4 The S = I Brillouin function in terms of the reduced variables

MT/MO and T/TC.
contact field, Bs, becomes non-zero, and a six-line magnetic

hyperfine splitting appears (Fig.5.5). The magnetic flux density
increases very rapidly at first, approximately following the S = ~
Brillouin function, with a saturation value at zero temperature of
62 T (620 kG). The saturation value is found to vary consider-
ably from compound to compound, being 55 Tin Fe2(S04h.
47 Tin FeC13 and FeC14 and 42 T in FeBr4' A decrease in flux
density corresponds to an increase in covalency, and represents an
effective decrease in polarization of the Fe inner core as the spin-
density is delocalized towards the ligands. As BL and BD are
always small for the S-state Fe 3+ ion, they can be neglected to
first order. A further noteworthy feature of the Fe 3+ ion is that
it has a very low magnetic anisotropy. Application of an external
magnetic field causes the intrinsic magnetization to rotate into
the axis of the external field, and the sum B ::: Bo + Bs becomes
algebraic in the manner described in Chapter 2. This feature will
94
L = 1·0009
TN
88
. ..
82
100
...,<»
0
L-
96
...,""
c::
~ =0·9989
c::
:>
0
U
92
100
96
~ =0·9955
92
88L---~----~----~----~--~----~----~--~
-3 -2 -1 0 2 3
Fig. S.S(a) The Mossbauer spectrum of FeF3 in the vicinity of the Neel
tern perature. ([ 9 ], Fig. 2)
88
84~~----~----~----~----~----~----~--~
-12 -8 -4 0 4 8 12
Doppler velocity / (m m !i-I)
Fig. S.S(b) The Mossbauer spectrum of FeF3 at 4.2 K. ([9], Fig. 1)
be mentioned again more specifically in connection with iron

oxides in Chapter 7.
Excluding the variations due to covalency, the value of Bs for
any configuration of iron is given approximately (in tesla) by
-22 (Sz>. Measurements in an applied external field show the sign
to be negative in all cases. The flux density in Fe 3+ (S = %) com-
pounds averages about 55 T, while Bs for an Fe 2+ (S = 2) com-
pound is only 44 T.
The total saturation flux density in ordered Fe 2+ compounds is
very variable, being -32.9 T in FeF 2, -25 Tin FeCI 2.2H 20 and
+3 Tin FeBr2. This is because the values of BL and BD may be
quite large for this electron configuration. Although Bs and BL
are usually much greater than B D , they are usually in opposition
to each other so that the resultant flux density can be quite small
and of either sign.
An example of magnetic hyperfine splitting from a high-spin
Fe 2+ion is given by data for FeF 2 [10]. Typical spectra are illu-
strated in Figs. 5.6 and 5.7. At 78.21 K which is 0.09 degrees
103rr-r----r-..,----y--,--,---;---.--r---n
100r:..~~"W-=-
97
94
91
7B·21 K
BB
85
94
94 70'23 K
91 88
Y
88 78·11 K
.
85 ~
~
~ 94
98
.
2-
~
0
67·6 K
~ 8S
96 .'=
~ 94
~
~ 92
Eo
~ 90
g' 88
""g1001-"'.~~~,:..,
u
98
96
94
92
90
n96K
88
100 ,....-_____ 4·2 K
98
BB
96 B
94
Fig.S.? Hyperfine structure in FeF2 when the
92
magnetic coupling is stronger than the quad-
90 rupole coupling. ([ 111, Fig.2).
8B
86
84_U4--L-_2~~-~0--L-~2-~~4-~-6~
Doppler velocity/(mm 5")
Fig.S.6 Magnetic hyperfine structure of 51Fe in FeF2 within I K of the

Neel temperature. The quadrupole coupling is stronger than the magnetic
coupling. ([ 101, Fig. 1).
above the Neel temperature a sharp doublet is seen. At 78.11 K,
only 0.01 degrees below the critical point, considerable line broad-
ening is found, and the subsequent splitting res em bles the triplet-
doublet pattern found for magnetic perturbation of a diamagnetic
compound by an applied field. This is because the principal axis
of the electric field gradient in FeF 2 is perpendicular to the spin
axis and thus approximates to the most probable relationship
between an external field and the electric field gradient in a pow-
dered sample. At low temperatures all six lines of the spectrum
are clearly resolved, and there are indications of weak absorption
in the two formally forbidden lines (Am z = ±2) because in this
example of combined magnetic/quadrupole interactions the energy
levels are no longer pure eigenstates of m z . The line positions and
rela tive intensities contain far more information than is available
from the quadrupole spectrum alone. Mossbauer spectra from
single crystals show that the magnetic spin axis is the crystal 'c'
axis. The data illustrated at 4.2 K lead to a value for B of 32.9 T,
a value for A = Ie 2qQ of +2.85 mm s-1, and an asymmetry para-
meter of 1/ = 0.40. The line positions and intensities require that
the principal axis of the electric field gradient tensor is perpendi-
cular to the spin axis which is itself along the minor axis. Thus
analysis of a combined magnetic/quadrupole interaction gives
detailed information about the crystal symmetry and the magnetic
structure.
It is still possible to use an applied magnetic field perturbation
to measure the sign of e 2qQ in a paramagnetic Fe 2+ compound, but
under more restrictive conditions. At moderate temperatures the
spin relaxation remains fast and the method works. At very low
temperatures the magnetic susceptibility becomes high, the elec-
trons tend to align with the applied field, and the observed magne-
tic splitting may not only be far greater than that due to the applied
field alone, but becomes more difficult to interpret [11]. The
anisotropy of the Fe 2+ ion environment must be considered in
detail.
The other electronic configurations of iron all have a lower
value of S, and although for example K3 [Fe(CN)6] with S =I
shows antiferromagnetic ordering, the Neel temperature is only
0.13 K. Co-operative phenomena as such are therefore not easy to
study. In paramagnetic configurations where the geometry is
highly distorted such as phthalocyanine iron(II) (square-planar,
S = 1) and bis-(N,N-diethyldithio-carbamato)iron nitrosyl
(approximately square-based pyramidal, S = ~), the iron environ-
ment is highly anisotropic. The quadrupole splitting observed is
usually temperature independent because the ligand field splitting
of the non-bonding 3d-orbitals is much larger than in the more
regular complexes. It therefore gives little information when taken
in isolation. Application of a magnetic field at very low tempera-
tures can induce a large magnetic splitting in a similar manner to
Fe2+, and makes it possible to determine the electronic ground state.
A good example of this is given by the S = ~ nitrosyl complex
just mentioned [12]. The molecular unit (Et2NCS2hFe(NO) is
formally an iron(l) complex if the nitrosyl is considered to be NO+.
In attempting to deduce the electronic configuration the chemical
isomer shift is uninformative because there are no comparative
data foriron(l) systems. The quadrupole splitting decreases from
1.03 mm s-1 at 1.3 K to 0.89 mm 8- 1 at 300 K. The sign of e2qQ
is positive, but it requires a detailed analysis of measurements in
applied magnetic fields to establish unequivocally that it is an
S = ~ complex with a d z2 ground state. As such it would be expect-
ed to have a negative sign for e2qQ were it not for the large con-
tributions to the electric field gradient from other delocalized
bonding electrons and from lattice contributions, which in the
event cause a change of sign.
Diamagnetic ions do not show intrinsic magnetic splitting, but
in some circumstances the presence of a nearby magnetically-
aligned ion can induce a small spin polarization resulting in an
effect known as transferred hyperfine splitting. This is known in
for example tin, antimony, tellurium and iodine, but will be con-
sidered more explicitly in Chapter 7.
5.3 Spin cross-over

In the majority of paramagnetic transition metal compounds it is
sufficient to distinguish between a small ligand-field splitting of
the d-orbitals and thence a 'high-spin' configuration, and a large
ligand-field splitting which induces spin-pairing to give a 'low-spin'
configuration. However, in a small minority of compounds one
finds that both configurations can exist, sometimes even in appa-
rently stable coexistence, and these are known as spin-crossover
Electronic Strncture and Bonding: Paramagnetic Compounds 119
complexes. In octahedral geometry the phenomenon is limited
to d 4 , d S , d 6 and d 7 configurations, and is far from understood.
For this reason Mossbauer spectroscopy has been extensively used
to study examples of spin-crossover in iron compounds.
In iron(II) complexes the change is from a ST2 (S = 2) state to
the lAl (S = 0) state. These configurations are particularly distinct
in the Mossbauer spectrum and their relative proportions can be
determined easily. Most iron(II) complexes with 1,1 O-phenanthro-
line are in either the ST2 or lA 1 configuration, but in two instances
a cross-over has been found, Fe(phenh(NCSh and Fe(phenh(NCSeh-
At high temperatures the magnetic moments are about 5.2 JlB, typi-
cal of the high-spin state, but at 174 K and 232 K respectively
there is a comparatively sudden change to the low-spin state,
although the low-temperature form usually has a small residual
moment which varies from sample to sample [13]. If the cross-
over were to involve a simple thermal population of close-lying
ST2 and lAl states, there would be no sudden change. In other
compounds such as Fe(bipyh(NCSh and complexes with 2-
aminomethylpyridine, 2-(2' -pyridyl)imidazole and hydro-tris-
(l-pyrazolyl)borate ligands, the transition is equally abrupt.
In the region of change-over both spin-states appear to co-exist,
and this is particularly marked in the borate complex as illustrated
in Fig.5.8, where both states coexist over a range of 50 degrees but
show most change over only a few degrees [14]. In this instance at
least, the effect seems to show signs of hysteresis as well as a depen-
dence on particle size, and slow thermal cycling of a crystal causes
pulverization.
In several cases there is good evidence for substantial structural
change involving alterations in molecular dimensions. The equili-
brium bond distances in the ST2 and lA 1 forms are unlikely to be
identical because of different electron correlation, and this could
induce crystallographic changes. Furthermore, some of the large
ligands involved could, in principle, form different geometric
isomers. Such changes would be consistent with the apparent
slowness of the interconversion (> 10-7 s) as deduced from the
Mossbauer spectrum, and with the retention of a small proportion
of one form well below the nominal transition temperature.
The iron (III) cross-over from 6A 1 (S = %) to 2T2 (S = ~) is less
common and is best known in the N,N-dialkyldithiocarbamates
such as (Me2NCS2hFe. These compounds do not usually show
~:::r~~ t~~ r-~I..

0·96
0-94 269 K :
:. :- .:
0-92 '. :
l'OO~lI.".~"
... ~ .:.... h-. 1""';'-'
-I-"~:
0'98
.
' , . : .-
~~ '::
0·96 : ~ ; i
245 K ': :,
0·94 '; :' ~
l'OOI~':"
0.98 ~ r ~ ..~~
; ... i ::
0'96 ' ; "
•• , l'
:'
.
0·94 236K:: :.;
.2
n~I~'" ~ ,'i ~
~ 0·98
.~
... ~~
".
~
f~·, a~
~:
~ ,'.t#. .•
~
:; 0·96 \
.:.~.
II" ,:
~
., 0·94
'ii
234K':
" p'
;
erc 0'92 ;
1'COI'~.;t:t*-
0-98
. '1 ..
.}I;,
',.'
,i!'"
.il.?
t ~ ~ .'
0·96 . ::
....
~.
.
o!) "."
0-94 230 K <
"
:
0·92 ~ .
l'OOl ~'!\"
0,98 \; . '
,-..~
.'
:;
" :: f
0·96 :;
194K '.
0·94 ::
1'001~
0·98 : .'--
:
0·96 : :
0'94 .:
147K '.
0·92 ~
! ,I ,
-4 . -2 0 4
V.loci\y/(mm s")
Fig. 5,8 TYf,ical spectra of a hydro-tris-(l-pyrazolyl)borate iron(II) complex

showing the T2-1A 1 crossover. ([ 141, Fig. 5).
any sudden discontinuity in the magnetic susceptibility curve,
and in the case of the methyl compound the susceptibility rises
steadily from about 1.8 ilB at 4.2 K to about 5 J1.B at 400 K. At
temperatures above 78 K the Mossbauer spectra comprise a simple,
poorly resolved doublet [15]. There are no obvious signs of two
components, and although the lines broaden on cooling, this can
if desired be ascribed to spin relaxation (see Chapter 6) which in
any case becomes clearly significant below 78 K. The evidence
seemed at first to favour a fast rate of exchange between the 6Al
and 2T2 states. However, recent data for some tris(monothio-{3-
diketonato)iron(III) complexes [16] have shown clear evidence
for the coexistence of both high-spin and low-spin species during
the Mossbauer lifetime. Whether by chance the two species are
unresolved in the former case has not been ascertained. Another
problem raised by the new data concerns the relative proportions
of the two species. If the high-spin form is achieved by thermal
excitation from a 2T2 to a 6Al state, then even at very high tem-
peratures the proportion of the 6A 1 form can never exceed 50%.
However, in the diketonato complexes there is evidence to suggest
that a more complete transfer to the 6A 1 state takes place.
Mossbauer spectroscopy is particularly useful in these studies of
spin-crossover, because, unlike magnetic susceptibility measure-
ments, it is not an averaged measurement but records uniquely
the different species present.
5.4 Pressure effects

The convenience of carrying out experiments at atmospheric pres-
sure has led to a tendency to measure only the temperature depend-
ence of physical properties. However, it is equally feasible in
principle, if technically difficult in practice, to observe phenomena
which are pressure-dependent. This applies to both diamagnetic
and paramagnetic materials, but as the most interesting effects
concern the latter, they are conveniently discussed here.
Extreme pressure (up to 200000 atmospheres; 1 atm = 1.01325
x 105 Pa = 1.01325 bar) will compress a material, which in the
context of 57Fe can be thought of as an increase in the overlap
between the inner s-shells of iron and the valence orbitals of the
ligands. This generates an overlap-induced increase in s-electron
density at the nucleus, and a decrease in the chemical isomer shift.
This decrease in shift with increasing pressure has been observed
in a wide range of compounds by Drickamer and co-workers [17],
but has only been fully confirmed with wave-function calculations
in the case of KFeF 3 [18]. The effect is of course fully reversible
on release of the pressure.
If the iron environment is non-cubic so that a quadrupole split-
ting is present, the latter also changes with pressure, but the be-
haviour is less easy to rationalize because of the possibility of
pressure-induced distortions.
The most interesting aspect of pressure studies is the tendency
for iron in many compounds to show a change in oxidation state
or electronic configuration at high pressures. This usually affects
only a fraction of the total iron, and is fully reversible on pressure
release. The most common instance of this is found in high-spin
iron(lII) compounds such as Fe2(S04h and FeP0 4 which show a
partial conversion to high-spin iron (II). Low-spin iron(III) can
change to low-spin iron(II) as in K3Fe(CNk Such a change in
oxidation state results from an electron transfer from ligand to
metal. Compression alters the spacing of the electronic energy
levels of the system in such a way that thermal excitation can cause
electron transfer to a level which was inaccessible under ambient
conditions. K3Fe(CN)6 proves particularly interesting because
there is also a first-order phase transition at about 50 kbar which
presumably relieves the internal compression of the hexacyanofer-
rate(III) ion and causes a large decrease in the observed fraction
of the reduced species.
A particularly clear example of high-spin to low-spin conversion
is given in Fig.5.9 which shows the effect of pressure on the
Mossbauer spectrum of 57Fe impurity atoms in MnS2 (the impurity
concentration was 2 at.%.).MnS2 is isostructural with FeS2 in
which the iron is in the low-spin configuration. The impurity
atoms in MnS2 are in the high-spin configuration at atmospheric
pressure because the lattice constant of the pyrite lattice has
increased to 6.1 02 A in MnS2 compared to 5.504 A in FeS2. At
comparatively modest pressures the impurity atoms revert to
the low-spin configuration found in the iron analogue [19].
The occurrence of such changes is often intimately linked to
the 1T and 1T* orbitals on the ligands and the changing ability of
the metal to back-donate to the 1T* system as the overlap and
energy separations alter. That the balance is a subtle one can be
0·000
......
. ... .....
c
o
~ 0'020
5l
.0
00·040
":5
'il 0'060 4 Kbor
at
0·080
0'000 .. -..,
c 0'020
o
'z
e-
5l0'040 Low
.g spin
'0
:5 0'060
'z
"~
"- 0'080
65 Kbor
0'100
., .
0000
: ..... .... -.
:5 0'020
:;:;
Q,
g (}040
.0
o
'0
:5 0060
z
"o
.t 0{)80
0'100 138 Kbor
-4'0 -3'0 -2'0 -1'0 0 4'0

Doppler velocity (mm 5-')
Fig. 5.9 The 57Fe Mossbauer spectra of MnS2 containing 2 at.% 57Fe
impurity atoms under a pressure of 4,65 and 138 kbar. Note the change
from high-spin Fe 2+ to low-spin Fe 2+ with increase in pressure. ([ 19], Fig,l).
seen from the fact that high-spin iron(II) in bis-phenanthroline

complexes converts to 1ow-spin iron(II) with increase in pressure
[20], whereas low-spin iron(II) in tris-phenanthroline complexes
converts to high-spin iron(II).
5.5 Second and third row transition elements
The compounds of the second and third row transition elements
differ significantly from those of the first row. The ligand-field
effects on the d-electrons are usually much stronger, and result in
more frequent adoption of low-spin configurations. The magnetic
properties are also more complicated. In the second row, signifi-
cant Mossbauer data are only available for ruthenium, which is
of particular interest because of its relationship to iron in the
Periodic Table. Many of the third row elements have at least one
known Mossbauer resonance, and although none have been studied
extensively as yet, the main features of several of them are now well
established.
The 90-keV resonance of 99Ru is comparatively difficult to
observe, partly because the high 'Y-ray energy makes the resonance
weak except at liquid helium temperature, and partly because of
the short 16-day half-life of the 99Rh parent. However, with the
recent improvements in experimental technique, several laboratories
have obtained good data for a wide range of compounds. Although
the transition is from the Ie = ~ ground state to the Ie = ~ excited
state, the quadrupole moment of the ground state is very small.
As a result, any quadrupole effect is usually seen as an apparent
doublet splitting of the excited state.
The quadrupole splitting and chemical isomer shift data at 4.2 K
for a small selection of compounds are given in Table 5.1 [21, 22].
As with iron, the ruthenium oxidation states are characterized by
the d-electron configuration. The chemical isomer shift decreases
as the number of 4d-electrons increases, which by analogy with
iron indicates oR/R to be positive. Within the range of shifts
shown by Ru(II) compounds, the higher values for the cyanide
and nitrosyl derivatives results from back-donation to the ligand
7T* orbitals. This behaviour is comparable with that found in the
iron cyanides. In general the chemical isomer shifts correlate
with the spectrochemical series of the ligands.
Diamagnetic compounds with a regular geometry such as RU04,
[Ru(NH 3)6]2- and [Ru(CN)6]4- show no quadrupole splitting.
However, Ru04" and RuO~- with one and two unpaired electrons
respectively have a distorted geometry in analogy with [FeX4 ]2-
compounds, and in consequence have a large quadrupole splitting.
[Ru(SCN)6P- with one unpaired electron is distorted in the same
Table 5.1 Mossbauer parameters for the 99Ru resonance [21,22]
Oxidation Configuration Compound O(Ru i e2 qQe

state metal) /(mm s-l)
/(mm
s-1 )
Ru(VIII) 4dOS=0 Ru04 +1.06

Ru(VII) 4d 1 S = 1...
2
KRu04 +0.82 0.37
Ru(V!) 4d 2 S = 1 BaRu04·H20 +0.38 0.44
Ru(V) 4d 3 S = ~
2 RuFS +0.15 0.56
Ru(IV) 4ct4 S = 1 Ru02 -0.26 0.50
K2[ RuC161 -0.31 0.41
Ru(III) 4dS S = 1...
2 K3[ RuF 61 -0.84
K2[RuC1SH201 -0.71 0.32
[Bu~N13[Ru(SCN)61 -0.49 0.53
[Ru(NH3)61 Cl3 -0.49
[Ru(NH3)SCll Cl2 -0.53 0.38
[Ru(bipyridylhl (CI04)3 -0.54
Ru(II) 4d6 S=0 [Ru( NH3)61 0 2 -0.92
[Ru(NH3)SN2] Br2 -0.80 0.26
[Ru(NH3)SNO] C13·H20 -0.19 0.39
K2[RuC1SNO] -0.36 0.18
K4[ Ru( CN)61 -0.22
K2[Ru(CN)SNO]. 2H20 -0.08 0.39
Ru(CO)zCI2 -0.23
RU(1T-CSHS)z -0.75 0.43
way as [Fe(CN)6]3-, and [RuC1 6]2- with two unpaired electrons

is similar. The substituted Ru(II) derivatives [Ru(NH3hN0]3+
and [Ru(CNhNO]2- both show the splitting expected from their
non-cubic symmetry. Quantitative analysis of the 99Ru quadru-
pole splittings is difficult as yet, particularly in paramagnetic com-
pounds where the large spin-orbit coupling parameter (A. = -1000
cm- 1) causes substantial state-mixing and consequent deviations
from the value of the quadrupole splitting expected for a particu-
lar 4d-configuration.
It is noteworthy that magnetic ordering has not been detected
in any ruthenium co-ordination compound even at 4.2 K, although
such ordering has been seen in ruthenium oxide systems at much
higher temperatures.
i
The 73-keV transition in 193Ir is from the Ie = excited state to
the Ig =~ ground state. The quadrupole spectra are therefore
simple doublets, but a large E2/Ml mixing ratio results in all eight
magnetic transitions having a finite intensity. Some typical para-
meters for iridium compounds are given in Table 5.2 [23], and
show how the chemical isomer shift decreases with an increase in
the number of 5d non-bonding electrons in typical 'ionic' halogen
compounds. This is analogous to ruthenium and iron. The anoma-
lous shifts of Na3[1r(N02)402] and K 3[1r(CN)6] parallel experi-
ence with K 4[Ru(CN)6] and result from the strong back-donation
to 1T* orbitals.
Although the Ir(lII) compounds are diamagnetic because of the
t~g configuration, several formally octahedral [IrX6]3- compounds
show a substantial quadrupole splitting, implying considerable
distortion from regular geometry in the solid state. Iridium(lV)
compounds have one unpaired electron, and K2IrCl6 and
(NH4hIrCI6 for example order antiferromagnetically below 3.08 K
and 2.16 K respectively. The Fermi contact term, Bs, is large at
about 50 T per unpaired electron.
The 77.34-keV transition in 197Au is from the Ie= ~ excited
state to the Ig = ~ ground state and here again the quadrupole
spectrum is a doublet. The two common oxidation states of gold
are Au(I) and Au(III). The Au+ ion has the c1osed-shell 5dlO con-
figuration. Its structural chemistry is dominated by a tendency
Table 5.2 Mossbauer data for the 73-keV transition in 1931r
Oxida- Configu- Compound T ot ~f B

tion ration /K /(mm s-l) /(mm s-l) /T
state
h(VI) 5d3 S =} IrF6 4.2 +1.45 -0.47* 187.6

h(V) 5a4 s =1 IrF5 4.2 +0.06 0.74
h(IV) 5d5 s =!.
2 K2[ IrC161 1.52 -0.95 -0* 39.5
(NH4h[IrCI61 1.42 -0.99 -0* 40.8
K2[ IrBr61 4.2 -1.10 -0.20* 39.8
Ir(III) 5d6 S = 0 K3[ IrC161 4.2 -2.26 0.52
K3[ IrBr61 4.2 -2.23 0.40
Na2[IrClSH201 4.2 -1.97 0.54
Na3 [Ir( N0 2)4CI21 4.2 -1.27 2.00
K3[Ir(CN)61 4.2 +0.26 0.30
1le 7
relative to Ir meta
2qQ (1 + l1l2)12
* from p~rturb"ation of the magnetic spectrum.
to form linear compounds of the type AuX 2 or LAuX. If the
bonding is mainly by a-donation to the gold, the orbitals involved
will be 6s and 6pz, although there is also the possibility of some
Sdz2 admixture into the 6s-orbital.
The Au3+ ion has a nominal Sd B configuration and forms square-
planar complexes of the type AuX4", which for a-bonding involves
the 6s, 6px, 6py and Sdx 2_y2 orbitals. Molecular orbital calcula-
tions for AuCI4" have indicated an effective 0.9 e- hole in the
Sdx 2_ y2 orbital. In comparing the anions AuX2 and AuX4" where
X is electron withdrawing, the reduced Sd-occupation in AuX4
would be expected to give a corresponding increase in the s-electron
density at the gold nucleus as a result of reduced Sd-shielding.
oR/R is believed to be positive for 197 Au, so that AuX4" should
have a more positive chemical isomer shift than AuX 2, and as may
be seen in Fig.S.IO, this is in fact observed. In this plot of some
Mbssbauer parameters for typical compounds [24] , KAu( CN)4 and
AuCl3 have larger shifts than in KAu(CNh and Aua respectively.
However, these differences are much less than the overall range of
shift values for a given oxidation state.
Although neither the shift nor the quadrupole splitting in isola-
tion give a reliable indication of oxidation state, both parameters
in conjunction can be used diagnostically. For example CS2AU2Cl6
shows lines from the AuCl 2 and AuCI4" ions and does not contain
gold(II).
The isotope shifts observed in the atomic spectrum of mercury
have shown that the shielding effects of a Sd-electron or 6p-
electron on the 6s-electron density are only 2S% and IS% respec-
tively, and the large variation in chemical isomer shift in gold can
therefore be attributed mainly to the 6s-orbital occupation. If
the linear X-Au-X unit is 6s6p hybridized, then the 6s-electron den-
sity should increase linearly with the 6pz occupation. There should
therefore be a linear relationship in gold(1) compounds between
the chemical isomer shift (from the 6s-density) and the quadrupole
splitting (from the asymmetric 6p-occupation). This is found to
be the case for those compounds with symmetrical co-ordination
shown in Fig.S.1O (AuCI, AuBr, Aul, Na3Au(S203}z.2H20, KAu(CNh),
and also for AuCN with one bond to carbon and one to nitrogen.
Other available data (not shown) for compounds not characterized
structurally but believed to possess linear bonding lie close to this
line, and there is a good correlation between the observed parame-
12
KAu(CN)2
10
Iva 6
e KAu( CNIZi 2
e
:::::: KAu(CN)Z 8r2
'"
~4
KAu(CN)zCl z
Ci.
'"
~2
0.
..,::>
L
'"
::>
etO
-2
-4
-3 -2 -1 o 1 2 3 4 5 6
Chemical isomer shift / (mm 5- 1 )
Fig. 5.1 0 The chemical isomer shifts and quadrupole splittings of some
simple gold compounds of the type AuX, AuX 2, AuX3, AuX4" and AuX2Y2".
Note the linear correlations shown by gold(I) and gold(III) compounds.
ters and the positions of the ligands in the spectrochemical series.

It is noteworthy that the cyanides correlate with the other com-
pounds, thereby confirming the high a-donation of the cyanide
group with only weak 1r back-donation in these compounds. The
observed linear relationship argues against a varying amount of
5dz 2 admixture. The value of q in e 2 qQ is negative for the 6pz
orbital, and since Q is positive, e 2qQ should be negative for all
the gold(l) compounds.
In the square-planar gold(lll) complexes, the large reduction in
the 5d-density in the xy plane when the gold-ligand bonding is
ionic will result in an excess of 5d-density in the z axis and a nega-
tive value of e 2qQ. Increasing covalency in the dsp2 bonding will
increase the electron density in the xy plane. The effect of the
excess of 5d-density in the z axis will be reduced both directly
Electronic Structure and Bonding: Paramagne tic Compounds 129
and by the positive contribution from the Px and Py orbitals. The
electric field gradient from a Sd-electron might be anticipated to
be greater than from a 6p-electron because it is in a lower quantum
shell with a different (,-3) value. However, the large values of
e2qQ found for gold(l) compounds suggests that this is not the
case, and that it is the 6p-electrons which dominate. Increasing
covalency might therefore be expected to cause a reversal in the
sign of e 2qQ, and as seen in Fig.S.9 the correlated data for the
gold(lII) compounds seem to indicate that this has indeed taken
place. The more ionic fluorides may be assumed to have a nega-
tive value of e 2qQ, whereas the other compounds have the oppo-
site sign. As with the gold(l) cyanides, the gold(II!) cyanides show
only weak 1T-bonding. The isomer shift for KAu(CN)zBr2 is close
to the arithmetic mean for KAuBr4 and KAu(CN)4, implying that
the Au-ligand bonds are strongly directional and largely independ-
ent of each other.
5.6 Lanthanides and actinides

At least one Mossbauer resonance has been observed in each lan-
thanide element with the exception of cerium, there being no less
than 44 resonances in the fourteen elements including lanthanum.
Many of the isotopes have a deformed nucleus with a large quadru-
pole moment, and there has been considerable interest in deter-
mining the nuclear parameters of the excited states. From a more
chemical point of view, the lanthanide resonances unfortunately
provide less useful information. This is partly because the chemi-
stry of each element is dominated by the ionic M3+ cation, and the
4[-electrons are not significantly involved in the bonding.
A typical example is the 21.6-keV resonance of 151Eu for which
a substantial quantity of data is now available. The observed
chemical isomer shifts (quoted relative to EuF3) may be divided
into three distinct groups:-
ionic Eu 2+ compounds (4f') -13.9 to -10.9 mm s-1

metallic band systems -ll.4to -7.6mms- 1
ionic Eu 3+ compounds (4[6) -0.2 to +0.9 mm s-1
The change in electron configuration from 4[1 to 4[6 leads to an

increase in the s-electron density at the nucleus and, because oR/R
is positive, a net increase in the chemical isomer shift of -13 mm
s-l. In any series of related compounds the chemical isomer shift
increases as the bonds become less ionic and the degree of 6s-
character in the bonding increases [25]. For example in the series
EuF 3, EuC13, EuBr3 and EuI3 the shift increases by 0.53 mm s-1
and in EuOF, EuOCl, EuOBr and EuOI, it increases by 0.74 mm
s-l. However, in the series EuO, EuS, EuSe and EuTe the shift
decreases by 1.00 mm s-1 because in these compounds the 5p-
electrons participate in the bonding and alter the screening of the
s-electrons. The alloys correspond approximately to an Eu 2+ con-
figuration, but with an increased shift due to 6s-participation in
the band structure.
The Eu 2+ ion has a 4f7 half-filled shell with an electronic ground
state of 8S7(2' while the Eu 3+ ion has a 4f6 configuration with a
7FO ground state which has zero total angular momentum. In
neither case is their a valence-electron contribution to the electric
field gradient from the 4f-electrons. Consequently any quadru-
pole splitting can only arise from external lattice effects, and being
generally less than the linewidth is seen only as line broadening.
Thus it is not possible to obtain information easily about the co-
ordination or site symmetry.
The 7Fo ground state of Eu 3+ is non-magnetic, but exchange and
crystal-field interactions cause mixing with the TFJ excited states,
and magnetic hyperfine splitting of the Ig =i ~ Ie =t transition is
not uncommon in Eu 3+ oxides (see Chapter 7). Magnetic splitting
has also been observed a t very low temperatures in anum ber of
Eu 2+ compounds such as EUS04, EuC12 and EUC03 [26] .
The Mossbauer spectra of the other lanthanides may be inter-
preted in a similar manner. One major difference is that in some
of them (e.g. Sm, Dy, Er and Vb) the electronic ground state is
magnetic and the paramagnetic relaxation time may be long enough
to allow a magnetic hyperfine splitting at low temperatures. This
is considered in more detail in Chapter 6. The values of the magne-
tic flux density and quadrupole interaction are determined largely
by the electronic ground state, and although small variations are
observed due to crystal field interactions with the ligands, insuffi-
cient data are available to interpret the latter in detail.
Although Mossbauer work with the actinides involves some dif-
ficult experimental problems, outstanding results have been obtain-
ed for the 59.5-keV Ig = i ~ Ie = %- resonance of 237Np. The com-
Electronic Structure and Bonding: Paramagnetic Compounds l31
monly used parent is 241Am which a-decays to 237Np. The Heisen-
berg linewidth is only 0.067 mm s--I, but the narrowest experi-
mentallinewidths are at least 15 times broader. This is probably
caused by radiation damage in the source because the a-decay dis-
places the Np nucleus from its original lattice site.
Neptunium forms compounds with oxidation states from Np3+
to Np7+. The chemical isomer shifts which span a huge range of
about 100 mm s-1 correlate well with the expected Sf configura-
tions of 5f4 to 5jO. This is illustrated in Fig.5.11 [27]. The majo-
+40 ~pF3' ~pCI3' NpBr 3

NPZ S3
+30
+20
+10 NpBr4
NpCl"
Am(Thl NpOZ
0 t..::...::':::"::":::..:F=- NpF"
~pAlz ~...........- - KINpOz'C0 3
-10
(NpOZ'CZ04.2HzO
I", (NpOZ' OH. x HZO
-20
E
E
--
'0
-30
~~~- Rb(NpOZ)(N0 3' 3
NpFS
Na( NpOZ)(CZH30Z13
-40 (NH"IZNPz07' HZO
K3( NpO Z' Fs
-50
NpF6
-60 Ba3(NpOslz. x HZO
CQ3 (NpOS'Z. xHZO
[Coen3J NpOS. x HZO
-70 [Co(NH3'6J NpOS. x HZO
Lis Np 06
-80
-90
Fig. 5.11 A correlation of the 237N p chemical isomer shifts with oxidation
state. The shaded areas represent the predicted ranges. The more positive
shifts in those compounds with CO-<lrdination to oxygen rather than fluorine
are indicative of considerable covalent character in the former. (after [27)).
100,0
99·0
96'0 NpF,
(NpHI
97,0
100{)
99 ·0
li 96{)
~
~ 97'0
~.
i!
~ 100-0
co:
99·9
996
99-7
1000
99,5
99-0
- 160 -120 -60 -40 0 40 80 120 160

Velocity I (mm s· l)
Fig. 5.12 The S9.S-keV 237Np spectra of four neptunium compounds with
different oxidation states showing the large chemical isomer shifts, quadru-
pole splitting (N pF 3), and combined quadrupole/magnetic splitting
(KNp02C03 and K3Np02FS)' ([28], Fig.4).
rity of spectra have been recorded at 4.2 K, and at this temperature

many compounds show a paramagnetic hyperfine splitting due to
slow relaxation analogous to the behaviour found in the lantha-
nides. Quadrupole splittings are also large. Typical spectra [28]
are shown in Fig.S.12.
Compared with the lanthanides, the greater variation in chemi-
cal isomer shifts within a given oxidation state is a reflection of the
greater covalency in these compounds, although the way in which
the Sf, 6d and 7s orbitals are utilized is not yet fully understood.
The more positive shifts associated with the NpOi and NpO~+
cations (in comparison to the fluorides) reflect this increase in
covalency. Of particular interest is the compound LisNp06 [27].
It was originally thought to be analogous to LisRe06 in that the
NpO~- anion has Oh symmetry. However, the Mossbauer spec-
trum shows a large quadrupole splitting with 1/ = 0.33, and the
neptunium environment is therefore non-cubic. On this basis it is
proposed that the octahedron of oxygens is compressed along the
z axis and rhombically distorted in the xy plane to emphasize the
O-Np-O linear grouping which is so stable in the NpOi and NpO~+
cations.
References
[1] Ingalls, R. (1964)Phys. Rev., 133, A787.
[2] Gibb, T. C. (1968) J. Chern. Soc. (A), 1439.
[3] Ingalls, R., Ono, K. and Chandler, L. (1968) Phys. Rev., 172,295.
[4] Nozik, A. J. and Kaplan, M. (1967) Phys. Rev., 159,273.
[5] Campbell, M. J. M. (1972) Chern. Phys. Letters, 15,53.
[6] Dale, B. W. (1971) J. Phys. C, 4,2705.
[7] Edwalds, P. R., Johnson, C. E. and Williams, R. J. P. (1967) J. Chern.
Phys., 47, 2074.
[8] Watson, R. E. and Freeman, A. J. (1961) Phys. Rev., 123,2027.
[9] Wertheim, G. K., Guggenheim, H. J. and Buchanan, D. N. E. (1968)
Phys. Rev., 169,465.
[10] Wertheim, G. K. and Buchanan, D. N. E. (1967) Phys. Rev., 161,478.
[11] Johnson,C.E.(1967)Proc.Phys. Soc., 92,748.
[12] Johnson, C. E., Rickards, R. and Hill, H. A. O. (1969) J. Chern. Phys.,
50,2594.
[13] Konig, E. and Madeja, K. (1967) Inorg. Chern., 6,48.
[14] Jesson, J. P., Weiher, J. F. and Trofimenko, S. (1968)J. Chern. Phys.,
48,2058.
[15] Rickards, R., Johnson, C. E. and Hill, H. A. O. (1968) J. Chern. Phys.,
48,5231.
[16] Cox, M., Darken, J., Fitzsimmons, B. W., Smith, A. W., Larkworth,
L. F. and Rogers, K. A. (1972) J.es. Dalton, 1192.
[17] Drickamer, H. G. and Frank, C. W. (1973) Electronic Transitions and
the High Pressure Chernistry and Physics of Solids, Chapman and Hall,
London.
[18] Simanek, E. and Wong, A. Y. C. (1968) Phys. Rev., 160,348.
[19] Bargeron, C. B., Avinor, M. and Drickamer, H. G. (1971) Inorg. Chern.,
10,1338.
[20] Fisher, D. C. and Drickamer, H. G. (1971) J. Chern. Phys., 54,4825.
[21] Kaindl, G., Potzel, W., Wagner, F., Zahn, U. and Mossbauer, R. L.
(1969) Z. Physik, 226, 103.
[22] Potzel, W., Wagner, F. E., Zahn, U. and Mossbauer, R. L. (1970) Z.
Physik, 240, 306.
[23] Wagner, F. and Zahn, U. (1970)Z. Physik, 233, 1.
[24] Faltens, M. O. and Shirley, D. A. (1970) l. Chern. Phys., 53,4249.
[25] Gerth, G., Kienle, P. and Luchner, K. (1968) Phys. Letters, 27A, 577.
[26] Kalvius, G. M., Shenoy, G. K., Ehnholrn, G. J., Katila, T. E., Lounasmaa,
O. V. and Reivari, P. (1969) Phys. Rev., 187,1503.
[27] Frohlich, K., Gutlich, P. and Keller, C. (1972) l.C.S. Dalton, 971.
[28] Dunlap, B. D., Kalvius, G. M., Ruby, S. L., Brodsky, M. B. and Cohen, D.
(1968) Phys. Rev., 171,316.
CHAPTER SIX
Dynamic Effects
Although it is appropriate in many instances to consider the reso-

nant nucleus as being in a static environment, there are several
dynamic processes which have important effects on the Mossbauer
spectrum. Firstly, the nucleus is vibrating on its lattice site, and
the mode of these vibrations depends on the nature of the chemi-
cal bonding. Secondly, the nucleus may jump to a different lattice
site by some kind of diffusion process, and thirdly there may be
large fluctuations in the electronic environment of the nucleus
due to relaxation. These will be considered in turn.
6.1 Second-order Doppler shift and recoilless fraction

In describing the dynamic motion of the nucleus at its lattice site,
it should be noted that the atomic vibrations take place on a time-
scale faster than the lifetime of the Mossbauer excited state.
Accordingly, the mean velocity (v) and the mean displacement (x)
are both effectively zero. However, the corresponding mean-square
parameters are both finite. It was shown in equation (2.8) that
the mean-square velocity (v 2 ) is related to the second-order Doppler
shift by
(6.1 )
Similarly from equation 1.9, the mean-square displacement (Xl)
135
in the direction of observation j is related to the recoilless fraction
by
(6.2)
Note that <xJ >is the mean value along direction j, so that it is
dependent on direction, whereas <v 2 > is averaged in all directions
and therefore can have only a single value.
Unfortunately the values of <v 2 >and <xJ> at any given tempera-
ture are a complex function of the dynamic motion of the atom.
However, the temperature dependence of either quantity can be
compared with the predictions of a suitable theoretical analysis.
It is customary to consider the dynamics using the harmonic-
oscillator approximation. One such treatment by Housley and
Hess [I] calculates the mean-square velocity and mean-square dis-
placement of the particle in direction j to be given by
(6.3)
<x?> = !2.. ~ [.!.. + _~_] blt (6.4)

I m ~
I
2 ehw;/kT - I Wi
where m is the mass of the particle, T is the temperature, k is

Boltzmann's constant, Wi is the frequency of the i'th normal vibra-
tion mode, 21Th = h is Planck's constant, and the bji are constants
related to the force constants connecting the particle with its
neighbours.
Both these functions have interesting general properties which
are illustrated schema tically in Fig.6.1. Note that in each case the
slope and curvature are zero at T = 0, and that the curvature is
always greater or equal to zero. The slope approaches a limiting
value at high temperatures. If a weighted mean frequency w/n)
is defined as
Wj(n) =[~ bAw?] lin (6.5)

Dynamic Effects 137
~3 Slope = m
3kT
..'"'"
c:
>.
~.'" 2
Intercepts - tm Wj (1)
100 200 300 400

Temperature (arbitrary units)
4
kT 1
Slopes = In w2(-2)
J
Intercepts = 2m
" 1
Wj (-1)
100 200 300 400

Temperature (arbitrary units)
Fi~. 6.1 Schematic representations of the temperature dependence of

(v.) and (xf). In the former case the intercept is frequency dependent and
tHe limitin~ slope is frequency independent. In the latter case both para-
meters are frequency dependent. Two feasible curves for each are labelled
I and 2. ([1], Fig. 1).
and an expansion is made which is valid for high temperatures, one

obtains the limiting conditions
2 kT
(Vj )T-+oo = -- (6.6)
m
kT
(6.7)
<xJ>r-> wJ( -2)
0 0 = _ _ __
m
00
The form of equation 6.6 is particularly interesting as it is indepen-

dent of frequency. Summing over all three orthogonal directions j
to obtain (v 2 ) (that is (v 2 ) = (v~) + (v;) + (vi») gives
(6.8)
Substitution in equation (6.1) shows that the second-order Doppler

shift reaches a limiting value at high temperature of
(6.9)
In the low-temperature limit
(6.10)
2 1'1 1
(Xj )T->O = 2m -W-j(---l-) (6.11)
Note that these values correspond to a zero-point motion and,

being frequency-dependent, will differ from compound to com-
pound.
The evaluation of the general expressions in equation 6.3 and 6.4
requires a detailed knowledge of the frequency distribution which
is not usually available, and has therefore to be approximated. The
simplest approach is to use the Einstein model with one vibration
frequency WE such that the Einstein temperature is BE = l'IwE/k.
This then leads to
(6.12)
Dynamic Effects 139
(6.13)
Such an expression for (v 2 ) has been used for example to fit the
second-order Doppler shift of FeC12, the resulting value being
OE= 169 K [2].
Better agreement with experiment can usually be obtained using
a Debye model in which there is a distribution of frequencies. The
number of oscillations with frequency w isN(w), given by N(w) =
3w 2/wg where Wo is the maximum value of w, and 'hwo/k = OD
defines the 'Debye temperature' of the lattice. Thence
(v 2 ) =-911- JWo [!. +

w3
1] w 3dw
om 0
2 eft.-J/kT - 1
m
[i
= 9kO D + (~)4J8DIT x 3dx
OD 0 eX - 1
] (6.14)
2 -
(Xj ) -
311
wgm
IWo [1 + 1]
0 2" ehwlkT _ 1 wdw
=~ [!. + (~) J8DfI' xdx ]

2 (6.15)
kmOD 4 0 D eX - 1 0
This latter expression is most commonly used to evaluate the tem-

perature dependence of the recoilless fraction. The lattice tem-
perature (0 E or OD depending on which model is used) can be
determined directly from the temperature dependence of f. The
limiting values for the zero-point motion are then
(~~ t---!-:-:-2 (6.16)

or
9kBD 9kB D
<V 2 >D= - - (6.l7)
8m 16mc 2
The importance of the zero-point motion as a contribution to

the second-order Doppler shift was first recognized by Hazony [3]
who showed tha t it was a far from negligible contribution to the
total observed chemical isomer shift. His calculations for K4Fe(CN)6
and K4Fe(CNk3H20 revealed that even closely related compounds
can show significantly different zero-point motion contributions,
and that comparisons of small differences in the chemical isomer
shift are meaningless unless suitable correction is made.
Some of the problems in using the simplified theory are illustra-
ted by recent data for the absorbers iron metal, Na2Fe(CNhNO.2H20,
Na4Fe(CNkl OH 20 and K4Fe(CNk3H20. The temperature de-
pendence of the absorption intensity and the chemical isomer shift
between 78 K and 293 K were used [4] to derive independent
pairs of values for both BE and BD. In general, BE and 8D as deriv-
ed from the recoilless fraction data are always lower than from the
shift data. This is because the value of <x}> in the general expres-
sion is weighted by wi-l, while <v}> is weighted by Wi. Thus <x}>
is weighted towards lower frequencies than <v}>. This effect is
significantly greater in the salts than in the metal because of the
greater importance of high-frequency 'optical' modes in the salts.
Whereas for iron metal the values of BD = 358 ± 18 K from <x} >
and 421 ± 30 K from <v 2 > agree reasonably well, in sodium nitro-
prusside for example, the corresponding figures are 203 K and 788
K. Clearly neither the Einstein nor the Debye model is adequate
for compounds.
From equation (6.7), the high-temperature limiting value for
<x}> is linearly proportional to temperature, so that an extrapola-
tion of high-temperature data should pass through the origin as
shown in Fig.6.1. As may be seen in Fig.6.2, this is certainly tflle
in hydrated sodium hexacyanoferrate(lI). In hydrated potassium
hexacyanoferrate(II) however, the extrapolation does not have a
zero intercept, and is indicative of considerable anharmonicity in
the lattice vibrations. There is a particularly large mean-square
displacement at low temperatures. Potassium hexacyanoferrate(II)
Dynamic Effects 141
3r-----------------------------~
N
E
u
co
I
o
Temperature! K
Fig.6.2 The temperature dependence of <x2> in hydrated potassium hexa-

cyanoferrate(II), (PF), and hydrated sodium hexacyanoferrate(II), (SF).
The large intercept obtained by extrapolation from high temperatures in the
former case is indicative of the inadequacy of the harmonic approximation
in this compound. ([41, Fig. 3).
becomes ferroelectric below 251 K, and this is known to be often

associated with a change in frequency of some of the optical modes
of lattice vibration. Low-temperature anharmonicity has also
been clearly demonstrated in FeCl 2 [2].
Thus far in this discussion of the recoilless fraction, <xl> has been
considered to be an isotropic function. This concept is adequate
when the absorber is a random microcrystalline sample with a
single resonance line and an average value <x2> can be defined.
However, the vibrations of an atom a tala ttice site of non-cubic
symmetry will in general be anisotropic, and this should result in
an anisotropy of the recoilless fraction which should be 0 bserve-
able in single-crystal samples. The recoilless fraction in direction
j is given by f = exp (-<X!>/'fi.2) where A= flC/ Ey. If we define three
principal values in the co-ordinate system x, y, z such thatj is de-
fined by the usual polar angles () and ¢, then
<xl> = [<x~> cos 2 ¢ + <x;> sin 2 ¢] sin 2 () + <xi> cos 2 () (6.18)

In the case of axial symmetry, <x; > = <x];> and <xJ > = <x;> sin 2 e
+ <x;> cos 2 8. It is then customary to write <x;> as <xi> and <x;>
as <xI> so that
<xl> = [<xi> - <XI>] cos 2 e + <XI> (6.19)
The angular dependence of f can then be expressed as
(6.20)
An example of anisotropic recoilless emission may be found in

experiments using a source of 57Co doped into a single crystal of
zinc metal. The crystal was cut at 45° to the c axis so that the
recoilless emission can be studied both parallel (II) and perpendicu-
lar (1) to the c axis [5]. The results at room temperature of
f1 = 0.64 and fll = 0.41 reflect the large anisotropy in the vibra-
tions of the cobalt impurity atom.
It is more difficult to study anisotropy in single crystal absorbers
rather than sources because the observed spectrum has to be cor-
rected for the saturation effects of the finite thickness of the ab-
sorber. It was only realized comparatively recently that this
requires a rather complicated calculation because of the way in
which the absorber polarizes the transmitted 'Y-ray beam [6, 7].
One compound in which the vibrational anisotropy at the 57Fe
nucleus has been correctly determined is sodium nitroprusside,
Na2[Fe(CNhNO] .2H 20 [8]. Both the electric field gradient and
mean-square displacement tensors can be represented by three
principal values. In the former, only two such values are needed
to specify the tensor completely because the sum of all three must
be zero, but all three mean-square displacements must be given
to specify the vibration completely. Both tensors must satisfy the
point symmetry of the lattice site, but in general the two sets of
principal axes need not be collinear if the symmetry is low. In
sodium nitroprusside, one of the principal values of the mean-square
displacement tensor at the 57Fe site must be parallel to the crystal
c axis. The other two are therefore in the ab plane and can be
determined from any three measurements within this plane. Full
analysis of the experimental data shows that the principal values
are directed very close to the crystal axes with fe = 0.3 77 «x;>
Dynamic Effects 143
= 1.83 x 10- 18 cm 2),fa =0.367 «x;> = 1.88 x 10- 18 cm 2 ) and
fb = 0.332 «xl> = 2.07 x 10- 18 cm 2). However, although Vxx is
directed along the c axis, Vzz is at 36° to the a axis in the ab plane
and is approximately along the N-C-Fe-N-O axis. This is in fact
quite reasonable as the electric field gradient will reflect the C4v
symmetry of the immediate nearest neighbours, whereas the mean-
square displacement will be influenced by long-wavelength vibra-
tions in the crystal reflecting the lattice symmetry, rather than
the approximate local symmetry at the resonant atom.
6.2 The Goldanskii-Karyagin effect

It was found in Chapter 2 that the quadrupole spectrum of a
~ -+ ~ transition in a polycrystalline sample comprises two lines
of equal intensity. From Table 2.2, the angular dependence of the
±~ -+ ±~ line in a single crystal is as (1 + cos 2 e), and of the ±~ -+ ±~
line is as (} + sin 2 e) where e is the angle between the direction of
observation and the principal z axis of the electric field gradient.
The averaged values for a random polycrystalline sample are then
given by integrating over all orientations, and the relative line in-
tensities become
1_3_/2_ =
f +
'If
(1 cos 2 e)sin ede

_0_ _ _ _ _ _ _ _ _ =1 (6.21)
h/2
f (j- +
'If
o
sin 2 e) sin ede
These expressions are strictly valid only for axial symmetry, and
become more complex for a non-axial field gradient because of
state mixing. However, the final result in the more general analy-
sis is still a value of unity.
The above argument is only rigorous if the probability for recoil-
less absorption (or emission) is the same in all directions. If it is
not isotropic, then the intensity of the absorption will be weighted
in favour of a particular orientation of the crystallites. Writing the
recoilless fraction f as a function of e, we have for a thin absorber
f1T(l + cos 2 e)f(e) sin ede

1_3_/2 = 0_ _ _ _ _ _ _ _ _ i= 1 (6.22)
/t/2
f (j
1T + sin2 e )f(e) sin ede
°
That is to say, an anisotropic recoilless fraction should result in an
asymmetry in the intensities of the quadrupole doublet which is
independent of the orientation of the sample. This concept is
referred to as the Goldanskii-Karyagin effect after its original
proponents [9, 10] .
That anisotropy of the recoilless fraction should cause unequal
line intensities is quite simple to predict, but far harder to verify
in practice. This is because there are other phenomena which can
cause the same effect. In particular it is extremely difficult to pre-
vent some microcrystallites from orienting on compacting, and of
course partial orientation or texture induces asymmetry. The first
onset of paramagnetic relaxation (see later) is also seen as an appa-
rent intensity asymmetry which could be mistaken for a Goldanskii-
Karyagin effect. Many claimed observations of the effect could
well prove to be the result of other causes.
One example which will stand close scrutiny is the behaviour
found in Me2SnF2. The experimental ratio for the line intensities
has been measured as h/2lh/2 = 0.78 at 294 K [11]. The X-ray
structure of this compound shows a pseudo-octahedral co-
ordination about the tin with axial methyl groups and four bridg-
ing fluorine atoms. The root-mean-square amplitUdes of vibration
of the tin as deduced from the X-ray data have been given as 13.7
pm along the Sn-F bonds and 21.0 pm along the Sn-C bonds. The
mean-square vibrational amplitude is related to the recoilless frac-
tion by equation (6.20). This expression in terms of the known
values for <xi> and <x~> given above can be used to evaluate a pre-
dicted intensity ratio of 0.72, which is in good agreement with
the experimental value.
The principles of the Goldanskii-Karyagin effect can also be
applied to magnetic hyperfine spectra and to isotopes with higher
spin states; in all cases the effect will be seen as a deviation from
predicted line intensities.
Dynamic Effects 145
6.3 Electron hopping and atomic diffusion
In all the dynamic effects discussed so far, the atom or ion has been
restricted to a particular site in the lattice. However, there are
two situations where this is not strictly true. In the fIrst an elec-
tron has a high probability of 'hopping' from the resonant atom
(within the Mossbauer lifetime) to another atom on a neighbour-
ing site, and in the second, the resonant nucleus itself changes its
lattice site by jump diffusion.
The europium oxide EU304 has the orthorhombic CaFe204 struc-
ture with Eu2+ ions at the Ca sites and Eu 3+ at the Fe sites. These
two oxidation states are distinct in the 21.6-ke V 151 Eu Mossbauer
spectrum with chemical isomer shifts of -12.5 mm s-l (Eu 2+)
and +0.6 mm s-l (Eu 3+) relative to EU203 [12]. On the other
hand, the sulphide EU3S4 has a structure in which all the europium
cations are in equivalent sites. At low temperatures the Eu 2+ and
Eu 3+ resonances are distinct with an intensity ratio of 1 :2 as ex-
pected [13]. Above 210 K the two lines broaden and coalesce
until at about 273 K the spectrum comprises a single line at the
centre of gravity of the low-temperature spectrum. This behaviour
is illustrated in Fig. 6.3. The data can be completely explained
by introducing the idea that the extra electron on an Eu 2+ ion can
hop to a neighbouring Eu 3+ ion by overcoming the energy barrier
E. All europium sites are involved. This process is characterized
by a relaxation time T which will decrease with temperature rise
because of an enhanced jump probability. The temperature depen-
dence of T is given by T = TO exp (-ElkT). Theoretical calculations
of the spectrum predicted for various T can be compared with
experiment, and were used to derive the relaxation times shown
in the figure. As expected, T varies exponentially with temperature
and leads to an estimate for the activation energy of the hopping
process of E = 22 kJ mol-I.
Hopping processes are also known in iron oxides. Fe304 is an
inverse spinel ferrite, Fe 3+[Fe 2+Fe 3+] 04, which shows a transition
in many physical properties between 110 and 1 20 K known as the
Verwey transition. Below 110 K, the Fe 2+ and Fe 3+ oxidation
states are distinct and give magnetic hyperfIne splittings which can
be separately identifIed (see Chapter 7 for further details of spinel
oxides). Above the transition there is a fast hopping on the B sites
which causes substantial changes in the magnetic spectrum as the
~:~~.~\ (.~ '.
"\.' .
V ." '" \ f ..'.
~:~':'~':"':..
t=~'5<1O:S ... •
.",\
.!
1'\ 1\ 1foo \.1'J
K
.
•••••,~ .' . ~ \\ \'7/
f
.'
:.. :-~.. 'tJ .;i'~
t=3·5<10··s ...'::-..•.., ~:.:...... \
'" /.: \.. I
"'-'. .~...
.\ 'ti 213K
:t.:...
:'~:e:'. " .......
.' \. '/' ."""'
...-:' /'
§ t=17x10··s
:::
~•
. ~~'..
":..
,'. '\
\U.
'.,. /" ••
...... '-'.-.. .
. j ~. .~ ,"
...
228 K
c. •• • • \
'. . .......~""
....
~. ~ ~.
t=85<10·'Os \ .1'
~ "\ :/'250K
. .... .-.. ...., .~..". . I.. 7 ... -..-

. •
~~....
-.~.
----.~
~., .. .'~---:
/ . . . ..
t =8·5 X 10·" 5 • ~ ~/./.. . 273 K
\
... :
"1--·
. -........e._,TL,..-~~.~.
..... :
\
r-::'"
. .
. . ...~ \\""
. ~.
t=3'5x 10-"5 Vr.. ' ....

. ~ '~".'.
• '.\ 325 K
Source: lS1Sm,O, .~ 1
\ t
,
Absorber: Eu,S,
-
• Experimental spectra
Calculated spectra
~ I
f
~
Statistical error ~/
-20 -10 o 10
Velocity / I mm 5-1)
Fig. 6.3 The 151Eu spectra of EU3S4 between 83 K and 325 K. The weaker
line at 85 K is due to Eu 2+, and the stronger line to Eu 3+. The fast electron
hopping which occurs at higher temperatures results in an averaged spectrum,
and the solid curves are computed with a relaxation model whose time con-
stant is T. ([ I 31, Fig. l).
Dynamic Effects 147
Fe 2+ and 50% of the Fe 3+ cations combine to give an averaged
contribution [14, 15].
The phenomenon of atomic diffusion is more difficult to
study. If the nuclei jump instantaneously between lattice sites,
then the effect should be manifested as a broadening of the reso-
nance line without a change in recoilless fraction. If 70 is the mean
resting time between jumps, then the fractional increase in the
width of the energy distribution at half-height is given by
Ar = 212/70 (6.23)
This can be related to the self-diffusion coefficientD = [2/670

where [2 is the mean-square jump distance. Basically similar
broadening effects can be expected when the diffusion is a con-
tinuous Brownian motion such as is found for instance in very
viscous liquids, but this aspect will not be considered further.
Diffusion broadening has been clearly shown for 57Co impurity
atoms in both copper and gold at high temperatures. However,
the broadening observed [16] is less than predicted, and it would
seem that any satisfactory theory must include consideration of
the probability that a vacancy occurs next to the diffusing ion
and is available to accept it. A model incorporating vacancy dif-
fusion has shown that the broadening may be expected to be less
than predicted for an un correlated jump [17] .
Diffusion processes have also been studied in frozen aqueous
solutions, using the Mossbauer nucleus as an impurity probe. For
example, an aqueous solution containing Fe 2+ ions forms an un-
stable phase when quenched to liquid nitrogen temperature. This
was formerly thought to be a cubic ice lattice, but more recent
evidence favours the formation of a glassy or amorphous variety
of ice of near eutectic composition (at least in the region surround-
ing the impurity ions), so that the freezing causes segregation into
two components, a eutectic and a pure ice phase [18]. At about
163 K the glass softens and at about 183 K the supercooled
liquid recrystallizes irreversibly to a new phase based on the stable
hexagonal ice lattice which is then retained until the eutectic
melting point. While the transition from the low to the high tem-
perature form is taking place, the 57Fe Mossbauer resonance com-
pletely disappears, presumably because of the effect of diffusional
motion on the recoilless fraction. The 57Fe quadrupole splitting
has different values above and below the phase transition
temperature.
Quenched solutions containing EuCl3 show an entirely analogous
behaviour, but for some reason a quenched solution of Fe 2+ and
Eu 3+ is completely different [19]. Both the 57Fe and 151Eu reso-
nances decrease rapidly above 183 K and vanish at about 208 K
presumably due to diffusion, but do not reappear above this tem-
perature. Re-cooling causes a reappearance of the original spectra,
showing that in this case the effect is fully reversible and that the
ice transformation is strongly inhibited in the mixed solution.
There is no unique interpretation of these observations at present,
but clearly the Mossbauer resonance has a valuable role to play
in the study of diffusion in quenched solutions.
6.4 Paramagnetic relaxation

During the discussion of paramagnetic complexes in Chapter 5, it
was shown that although there is an interaction between the
nuclear spin and the electronic spin, the relaxation of the latter
is generally faster than the lifetime of the Mossbauer excited state
so that the average value (Sz) defined in the nuclear axis system is
zero. This contrasts with magnetically ordered materials where
the long-range coupling induces a non-zero value of (Sz) which is
seen directly as the Fermi contact interaction.
However, if the electronic relaxation rate in a paramagnetic
material is very slow, there will again be a static hyperfine inter-
action between the nucleus and the spin S (this paramagnetic
interaction differs from the magnetically ordered case in which
<Sz) is still a dynamic average). It is necessary to consider this
static case corresponding to a very slow relaxation in some detail,
before turning to the intermediate dynamic situation. The static
spin S may be able to couple in several ways with the nucleus.
For example the energy levels of the Fe 3+ ion (S = ~) in an axial
ligand field are three Kramers' doublets which are normally repre-
sented as ISz = ±i>, 1±1->' I±~>. The Fem(S= i) ion has a strong
spin-orbit interaction, and the 6-fold degenerate 2T2 term is also
split into three Kramers' doublets. If we consider a I±i> Kramers'
doublet state, both the Sz = +i and Sz = -i- orientations couple
with the nuclear spin to give an observed magnetic flux density
of magnitude B, but with opposite sign. There is no preferred
Dynamic Effects 149
direction for Sz (unlike the situation in ordered systems) and be-
cause the Mbssbauer spectrum does not indicate the sign of B, we
expect to see only a single magnetic hyper fine splitting correspond-
ing to the flux density 1B I. The three Kramers' doublets of the
Fe3+ ion are generally separated by only a very small energy. At
temperatures above 1 K all three are partly populated and one can
in principle expect to see three superimposed hyperfine splittings.
A full mathematical treatment o( the interactions is often com-
plex. A comprehensive discussion of the principles involved has
been given by Wickmann and Wertheim [20]. A useful representa-
tion is to consider each Kramers' doublet as having an effective
spin of S' = } and accumulating all other terms into a hyperfine
tensor A so that the general form of the Hamiltonian operator
becomes
(6.24)
Considering the Fe 3+ ion, the ISz = ±i> doublet corresponds to

the uniaxial symmetry A z > Ax = Ay !:::!. 0 and reduces the Hamil-
tonian to !lrS/2 = A z S~Iz. The corresponding spectrum is a simple
six-line spectrum similar in appearance to the magnetically ordered
pattern. This form of Hamiltonian is also commonly applied in
the rare-earth elements. The case Ax = Ay = 3A z corresponds to
the Fe3+ ISz = ±} > doublet which is more complicated with a total
of ten resolvable hyperfine components in the spectrum.
In general there may be an additional quadrupole interaction in
the Hamiltonian, plus a term representing an externally applied
field and written as an isotropic interaction ~E =A I.S. The
observed behaviour can vary drastically in external fields having a
flux density of only a few tesla. The static situation as described
here can rarely be achieved for all the Kramers' doublets involved.
Before discussing the spectra to be expected, it is necessary to con-
sider the effects of relaxation.
The electronic spin can change its quantized orientation by three
types of relaxation process. In spin-lattice relaxation it can trans-
fer or receive energy from a lattice phonon with a mean-time
between transitions denoted by T 1 • The interaction takes place
via the spin-orbit interaction AL.S so that if the ion has considerable
orbital momentum, i.e. (L) is large, then the relaxation will be
rapid. Fe2+ and low-spin Femcompounds do not normally show
paramagnetic hyperfine effects because the spin-lattice relaxation
is fast. On the other hand, Fe 3+ which is an S-state ion has a cor-
respondingly long spin-lattice relaxation time. Because the process
involves lattice phonons, this type of relaxation will slow down
with decreasing temperature.
Spin-spin relaxation involves a mutual spin-flip between neigh-
bouring ions of the type SI+S2- ~ SI-S2+ and is energy conserving
if the spins are identical. Because the spins are on neighbouring
ions, the relaxation is strongly concentration-dependent and will
be much slower as the distance between ions increases. However,
unlike spin-lattice relaxation, spin-spin relaxation is not temperature-
dependent. The third type of relaxation is less important and
involves a cross-relaxation between two different spin types. In
the current context this could mean a nuclear spin and an electronic
spin.
The result of these various relaxation processes on a particular
energy level is usually expressed by a single effective parameter T,
which represents the lifetime of the electronic level. This must be
compared with the nuclear Larmor precession frequency WL, and
ifr> wr: 1 the static spectrum is seen. When T ~ wi 1, the relaxa-
tion randomly changes the energy of any particular Mossbauer
transition. For example the transition ISz = +}> ~ ISz = ~ >
reverses the sign of the flux density B. The detailed theory is
again too complex to give here [20], but the behaviour to be
expected can be illustrated by examples.
The rare earth elements are comparatively simple to treat in that
a static paramagnetic hyperfine splitting is usually produced by a
highly anisotropic hyperfine tensor with A z > Ax. A y . Thus the
Hamiltonian reduces to 2= AzS~/z as already mentioned, and
the nuclear level with spin 1 splits into 21 + 1 equi-spaced levels as
with the normal Zeeman interaction for long-range order. If the
anisotropy were reduced, the spin-spin relaxation time would
decrease and the hyperfine splitting disappear.
An example is given by the Dy3+ ion which has a 6H15{l electro-
nic ground state. The large spin-orbit interaction in the rare-
earths causes J to be a good quantum number, and the excited
±1;
states are not thermally accessible from the IJz = > ground
state doublet level at low temperatures. The 161Dy resonance in
DyCr03 at 4.2 K comprises a static paramagnetic hyperfine splitting
from the Dy3+ cation, and is illustrated in Fig. 6.4. There are a
Dynamic Effects 151
...... .
.- ....
100 .' ... ~.II',.':
.... ..:\:..,.. .."r'~.
..-. :.··;:.:......
,
99 300 K
98
. '.~.,
r."ot' I I - . : ....
... ....
.,...~•• ; - .·I~"·
•.." ...".....,..,...... ...;:.
100 ... ..... , ..." ,
.,.... .. ___........ "" ••-~
~~
,... -.:.
."..
78K
99
...-.......:•~.
f.ro::,1'01;;...•·t,..... .;.: .;:........•••• ';0'-.:","':" .
..
100 ~ •...,~.... '. "~'" :r\ -•
AI._ .. .." - . . " ••-
99 54 K
, ,
...'~V"A.:·.. ._.,;" -:.".' ..,"~
...
....
'" ..."":
.. ..-
.to .. .,
...
.. .. , .. -
.: .... "
.' .. ~
99 : '. :
20 K
.'."'-.." -: .:--..",.... .,-:...,. .:' ..

.:.-...
. .. ...
.~:--
.: ....
~
~-
..
.-~
"" _'_..'.' e.
. .' .
•• ..1 ••
~
'.
99
4·2 K
98 L-------~
20-0-------1~0~
0 ------0~----~10~O~--~2~O~O----~
Voloc lty/ Imm ,-1)
Fig.6.4 The 25.65-keV 1610y resonance in OyCr03. At 4.2 K the spec-

trum is a resolved paramagnetic hyperfine splitting from the IJz = ±!?.>
ground state. Increasing temperature causes a reduction in the spin-fattice
relaxation time and a complete collapse of the splitting. ([21], Fig. 1).
152 Principles of Moss bauer Spectroscopy
total of 16 resonance lines from the Ie = ~ -+ 19 =%E1 decay [21]. As
the temperature is raised the effective relaxation time decreases, being
7 ns at 20 K and 0.05 ns at 78 K, and results in complete collapse
of the magnetic structure. The inward 'motional narrowing' of
the spectrum is typical of relaxation processes. The decreasing
relaxation time is a result of a faster spin-lattice relaxation as
thermal excitation takes place to the excited Jz levels. The static
spectrum at 4.2 K is slightly asymmetrical because of an axially
symmetric quadrupole splitting. Very similar spectra are known
in a large number of dysprosium compounds at 4.2 K including
DyF 3.5H 20, Dy(N0 3h6H 20 and DyP04.5H20 [22] .
Fully resolved paramagnetic hyperfine splitting has also been
recorded for 149Sm (Jz = ±~) in SmC13.6H20 and Sm(N03h.6H20
at 4.2 K [23], and in several of the erbium and ytterbium reson-
ances. Only one case is known of paramagnetic hyperfine struc-
ture in a rare-earth compound which is not highly anisotropic.
Erbium ethyl sulphate, Er(EtS04h9H20, has short relaxation
times, but magnetic dilution with 97.6% of the non-magnetic
ytterbium compound slows the spin-spin relaxation rate and allows
the appearance of a static field in the 80.6-ke V 166Er resonance [24].
In this case the abbreviated Hamiltonian is not valid and results in
incorrectly calculated line positions.
There are many examples of paramagnetic relaxation in the
Mossbauer spectra of iron compounds, the majority of these con-
taining Fe3+ ions. This S-state ion does not relax by a spin-lattice
mechanism, and the spin-spin process dominates. In iron(III)
compounds with a small Fe-Fe distance such as FeCl3.6H20
(-6 A = 600 pm) the relaxation is fast and the lines are compara-
tively sharp, whereas compounds with large chelating ligands such
that the Fe-Fe distance increases to about 9.5 A tend to show defi-
nite magnetic structure [25]. The spectrum of iron(III) acetylace-
tonate (7.6 A) is very broad at 1.8 K with a non-Lorentzian line-
shape of about ten times the naturallinewidth and only narrows to
about 7 linewidths as the temperature is raised to 300 K.
The relaxation behaviour 0 bserved may be particularly complex
when the resonance is also quadrupole-split. A common type of
behaviour in 57Fe spectra is that one of the two lines becomes
increasingly broadened with rise in temperature. This is the in-
verse of the normal situation, and is found for example in iron(III)
haemin chlorides (see Chapter 11), and in the acetylacetonate com-
Dynamic Effects 153
p1ex, Fe(acachC1 [26]. These compounds all feature a square-
pyramidal geometry with an apical chlorine atom and are in the
high-spin S = 1- configuration rather than the intermediate-spin
S = ~ configuration found in the dithiocarbamato derivatives. How-
ever, the phenomenon also occurs in octahedrally co-ordinated
compounds such as FeC13.6H20.
Typical spectra for Fe(acachQ are shown in Fig. 6.5. The sym-
metrical doublet seen at 4.2 K becomes grossly broadened by
300 K, implying that the relaxation slows down as the temperature
increases which is the opposite of spin-lattice relaxation. An ex-
planation has been provided by Blume [27]. The ground state
of the Fe 3+ ion in the crystal field is the I±}> state with the I±%>
and I±1-> excited states at energies ~ and 3~ above the ground
state (Fig. 6.6). The separation ~ expressed in terms of kT is
usually of the order of 10-20 K (80-160 J mol-I) so that at 4.2
K only the I±}> level is occupied. The spin-spin relaxation pro-
cess involves a SI+S2- -* SI-S2+ transition and the rate is tempera-
ture independent provided that the spin populations remain con-
stant. In fact, the I±}> level popUlation decreases substantially
as an increase in temperature causes progressive thermal popula-
tion of the higher levels. The result is an increase in the relaxation
time of the I±}> doublet. The excited states themselves have much
longer relaxation times, and the net increase in relaxation time
results in the line broadening. It is usually the Ig = ±} -* Ie = ±%
component of the quadrupole splitting which broadens most, so
r.':"~~:~;.~..
' ..
'
,:,..
c: ., "'
o
'.~ ;",,"'..;. / ~ :
: '. !,
'E
II>
c:
C
.....L. (0) (b) ',' (c)
-2 -1 0 1 2 -2 -1
.
..
0 1 2 -2 -1 0 1 2
Velocity/( mm S-1) Velocity/( mm s- '} Velocity/( mm S-I}
Fig. 6.5 Mossbauer spectra of Fe(acachCI at (a) room temperature,

(b) 80 K and (c) 4.2 K, showing the progressive decrease in asymmetry as
the relaxation rate increases. ([ 26], Fig. I ).
1
If
2.1l
.Il
~ Itt>
Fig.6.6 The energy-level scheme of the Fe3+ ion.
that e 2qQ is seen to be negative in FeC13.6H20 and positive in the

square-pyramidal compounds.
The compounds just discussed are magnetically dilute in that
the iron atoms do not strongly interact with each other. It is
interesting to note that in compounds where strong intramolecu-
lar antiferromagnetic coupling takes place between pairs of Fe 3+
ions such as [Fe(salen)hO, [Fe(salen)Clh where salen-l ,2-di(sali-
cylideneamino)ethane, and [(phenhFe.O.Fe(phenh] where phen
= I ,IO-phenanthroline, the quadrupole lines are sharp at low tem-
perature because the ground state is non-magnetic with a resultant
spin of S' = 0 [26]. However, the excited states with S' = I, 2,3,
4 and 5 are magnetic and result in similar relaxation broadening
to the previous case except that it is not usually as pronounced
[28]. A clear distinction between the two cases can be made by
applying a large external field at low temperature [29]. The I±{>
state in Fe(acac)~l shows a large augmentation of the external
magnetic flux density (e.g. B = 13.5 T for an applied field of flux
density Bo =3 T) whereas for a non-magnetic S' =0 state the ex-
ternal flux density is essentially unmodified [28] .
In general, the application of a small external field to a broaden-
ed Fe 3+ spectrum often causes substantial narrowing of the lines
due to a field dependence of the relaxation, but although such
narrowing has been studied in several compounds, it is not fully
understood.
It is noteworthy that Fe3+ relaxation is not seen in oxides con-
taining more than 5% of Fe3+ cations because the short cation-
cation distances make them magnetically concentrated, but resolv-
Dynamic Effects 155
ed splitting can be expected below the 2% Fe 3+ level. The spectra
of Fe 3+ ions (0.14%) in Al 2 0 3 are typical of the behaviour
found, and the 57Fe spectrum at 77 K is shown in Fig. 6.7
[30]. Although the ground state is the I±t> level, the relaxa-
tion rate of this level is faster than those of the excited states,
and consequently it contributes only a broad background to the
spectrum. The best resolved lines are from the I±i> level with a
flux density of about 54.8 T (close to that expected for an S = %
ion), and it appears that relaxation is also slow for the I±%> state.
Increasing the iron content decreases the spin-spin relaxation
I
±S/2
I
I
, ,
... '
,,
±3/2 I
o ....•.
.' I I I' I'
· . >~S~. ±1/2
\
....
~ ~.
c:
....0-
.Q
I-
oof)
.J:J
«
-10 -5 0 5 10
Velocity/(mm 5-')
Fig.6.7 The 57Fe spectrum at 77 K of 0.14 at.% Fe 3+in A1203. The pre-
dicted lines for the three Kramers' doublets are shown as bar diagrams, but
the I±~> contributions are not resolved. ([301, Fig. 2).
times and causes collapse of the magnetic splittings, and this pro-
cess is largely complete at 1.4% Fe.
Of the other common iron electronic configurations, low-spin
iron(lII) shows magnetic effects only if magnetically dilute «5%)
to prevent spin-spin relaxation and at low temperature to slow the
spin-lattice relaxation. Such conditions have been achieved by
doping K 3Fe(CN)6 into K 3Co(CN)6 [31] .
6.S Superparamagnetism
Having considered the case of paramagnetic ions which can show
magnetic hyperfine interactions, it is appropriate finally to con-
sider the inverse situation of a magnetically ordered material which
appears to be paramagnetic. Using a naive description, one can say
that a magnetically ordered ion with Sz :::: ~ will still fluctuate bet-
ween the Sz :::: +~ and Sz :::: -~ states as for a paramagnetic ion,
but that the spin will statistically spend more time in one state
than the other. In magnetically ordered solids the spin-exchange
interactions are strong. The spin-wave description of the magnetism
results in an averaged value (Sz) which is responsible for the magne-
tic hyperfine splitting. This value of (Sz) is defined along the
direction of magnetization and has a long relaxation time because
the co-operative interaction is long-range. Thus a magnetically
ordered solid usually shows a static hyperfine splitting at all tem-
peratures below the ordering temperature.
The value of (Sz) is not, however, a static quantity, and within
each magnetic domain fluctuates with a relaxation time 7
:::: 70 exp (KV/kT) where KV is effectively the energy barrier to be
overcome in changing the direction of the magnetization, and kT
is the thermal energy. Since V is the volume of the domain, K
is the energy per unit volume. Thus the relaxation time can be
decreased by decreasing the volume of the domain or by raising
the temperature. Provided that the relaxation rate is slower than
the Mossbauer lifetime then the full hyperfine splitting is seen,
but if the two are comparable then a relaxation narrowing can be
expected.
This effect has been clearly demonstrated in several iron oxides
such as a-Fe203 [32] and a-FeOOH [33] by decreasing the par-
ticle size and thus the magnetic domain size to the order of <200
A (20 nm). Although the magnetic hyperfine splitting is still
Dynamic Effects 157
present at low temperatures, it is seen to collapse with increase in
temperature to give the spectrum characteristic of the paramagne-
tic phase well before the Neel temperature is reached. The finer
details of the relaxation are often obscured because most methods
of generating small particles tend to give a distribution in particle
size with a consequent distribution in the relaxation time.
Probably the most common type of magnetic relaxation occurs
in non-stoichiometric compounds where the ordered cations on a
particular lattice site have been diluted with a non-magnetic sub-
stituent. This weakens the exchange interactions to the point
where they become localized to small clusters of cations. It is
thus analogous to the small particle-size effect, and a contributory
cause of the collapse of the magnetic hyperfine splitting. Several
examples will be found in Chapter 7.
References
[1] Housley, R. M. and Hess, F. (1966) Phys. Rev., 146,517.
[2] Johnson, D. P. and Dash, J. G. (1968)Phys. Rev., 172,983.
[3] Hazony, Y.(1966) 1. Chern. Phys., 45,2664.
[4] Lafleur, L. D. and Goodman, C. (1971) Phys. Rev. B, 4,2915.
[5] Housley, R. M. and Nussbaum, R. H. (1965),Phys. Rev., 138, A 753.
[6] Housley, R. M., Gonser, U. and Grant, R. W. (1968) Phys. Rev.
Letters, 20,1279.
[7] Housley, R. M., Grant, R. W. and Gonser, U. (1969) Phys. Rev., 178,
514.
[8] Grant, R. W., Housley, R. M. and Gonser, U. (1969) Phys. Rev., 178,
523.
[9] Goldanskii, V. 1., Gorodinskii, G. M., Karyagin, S. V., Korytko, L. A.,
Krizhanskii, L. M., Makarov, E. F., Suzdalev, 1. P. and Khrapov, V. V.
(1967) Doklady Akad. Nauk S.S.S.R., 147, 127.
[10] Karyagin, S. V. (1963) Doklady Akad. Nauk S.S.S.R., 148, 1102.
[11] Herber, R. H. and Chandra, S. (1970)J. Chern. Phys., 52,6045.
[12] Wickman, H. H. and Catalano, E. (1968) J. Appl. Phys., 39,1248.
[13] Berkooz, 0., Malamud, M. and Shtrikman, S. (1968) Solid State Cornrn.,
6,185.
[14] Banetjee, S. K., O'Reilly, W. and Johnson, C. E. (1967) 1. Appl. Phys.,
(1967) 38,1289.
[15] Hargrove, R. S. and Kundig, W. (1970) Solid State Cornrn., 8, 803.
[16] Knauer, R. C. and Mullen, J. G. (1968) Appl. Phys. Lett., 13,150.
[17] Knauer, R. C. (1971) Phys. Rev. B, 3,567.
[18] Cameron, J. A., Keszthelyi, L., Nagy, G. and Kacs6h, L. (1971) Chern.
Phys. Letters, 8,628.
[19] Dilorenzo, J. V. and Kaplan, M. (1968) Chern. Phys. Letters, 2, 509.
[20] Wickman, H. H. and Wertheim, G. K. (1968) In Chemical Applications
of Mossbauer Spectroscopy, Chapter 11, ed. V. I. Goldanskii and
R. H. Herber, Academic Press, New York.
[21] Eibschutz, M. and van Uitert, L. G. (1969) Phys. Rev., 177,502.
[22] Wickman, H. H. and Nowik, I. (1967)J. Phys. Chern. Solids, 28,2099.
[23] Ofer, S. and Nowik, I. (1967) Nuclear Phys., A93, 689.
[24] Seidel, E. R., Kaindl, G., Clauser, M. J. and Massbauer, R. L. (1967)
Phys. Letters, 25A, 328.
[25] Wignall,J. W. G.(1966)J. Chem. Phys., 44,2462.
[26] Cox, M., Fitzsimmons, B. W., Smith, A. W., Larkworthy, L. F. and
Rogers, K. A. (1969) Chem. Comm., 183
[27] Blume, M. (1967) Phys. Rev. Letters, 18,305.
[28] Buckley, A. N., Herbert, I. R., Rumbold, B. D., Wilson, G. V. H. and
Murray, K. S. (1970) J. Phys. Chem. Solids, 31,1423.
[29] Fitzsimmons, B. W. and Johnson, C. E. (1970) Chem. Phys. Letters, 6,
267.
[30] Johnson, C. E., Cranshaw, T. E. and Ridout, M. S. (1964) Proc. Int.
Cant on Magnetism, Nottingham, (Inst. Phys. Physical Soc., London,
1965), p. 459.
[31] Oosterhuis, W. T. and Lang, G. (1969)Phys. Rev., 178,439.
[32] Kundig, W., Bammel, H., Constabaris, G. and Lindquist, R. H. (1966)
Phys. Rev., 142,327.
[33] van der Kraan, A. M. and van Loef, J. J. (1966) Phys. Letters, 20,614.
CHAPTER SEVEN
Oxides and Related Systems
There are a large number of ionic compounds in which the crystal

structure is determined largely by a close-packed lattice of the
anions. The most familiar examples are found among the oxides,
sulphides, and anhydrous halides. The anionic lattice has the power
to accommodate a wide range of different cations in the largest
interstices, not only to give stoichiometric compounds, but also
to form the important category of non-stoichiometric compounds.
Any given cation in the lattice is co-ordinated to several anions
which frequently number four or six and are arranged in a near
regular geometry. The electronic interaction of the cation with
this environment can be described in terms of the ideas outlined
in the previous chapters on electronic structure. However, although
all cations of a given type may have the same co-ordination to the
anion, the successive co-ordination spheres containing other cations
may well be very different. It is this distinction between a stoichio-
metric ordered lattice, such as in FeF 2, and a non-stoichiometric
disordered lattice as in COl_xZnxFe204 which results in important
new phenomena.
The examples which follow will be restricted mainly to oxides
with the spinel or iron garnet lattices in the knowledge that the
principles outlined for these compounds are directly applicable to
other structures. It is appropriate in the first instance to develop
the concepts introduced in the earlier chapters with particular
reference to those oxides and sulphides which are close to stoichio-
metry, as these serve as model compounds for their more com-
159
plicated derivatives. Some topics involving dynamic effects such
as electron hopping, paramagnetic relaxation and superparamagne-
tism have been described separately in Chapter 6.
7.1 Stoichiometric spinels

One of the most widely adopted oxide structures is the spinel
lattice, and because it has both tetrahedral and octahedral co-
ordination at the cation sites, it serves as a convenient illustrative
example. The idealized lattice is shown in Fig. 7.1, and consists of
tetrahedral (A) sites and octahedral [B] sites in a face-centred cubic
oxide sub1attice, with stoichiometry AB 20 4 . Where necessary the
convention of round parentheses for the A-site and square brackets
for the B-site will be used.
The tetrahedral site has cubic (Td) symmetry, and hence there
is no 'lattice' contribution to the electric field gradient, but the
octahedral site has trigonal symmetry and one can therefore anti-
cipate a large electric field gradient along the [ 111] trigonal axis
of the octahedron. Although many oxides with a 'normal'
(A)[B2] 0 4 structure exist, others are of the type (B)[AB] 04 and
are said to be 'inverse'. Partial inversion is also common. The
Mossbauer spectrum can be used to determine the site occupancy
in the spinel (and in other oxides with multiple site symmetries).
Many of the spinels can order magnetically. The major magnetic
interactions are between A-site and B-site cations, the A-A and B-B
interactions being much weaker. In the context of the 57Fe reso-
nance, this may involve both Fe-Fe and Fe-M interactions where
M is another paramagnetic cation.
Several general observations can be made about the 57Fe spectra
of spinels which derive from the discussion in Chapters 4 and 5.
The chemical isomer shifts increase in the sequence o(tetrahedral
Fe 3+) < o(octahedra1 Fe 3+) ~ o(tetrahedra1 Fe 2+) < o(octahedra1
Fe 2+) and are therefore diagnostic of the oxidation state and co-
ordination. This generalization is also valid for sulphide spinels
except that the shifts are uniformly smaller because of increased
covalency. Where quadrupole splitting occurs, that for the Fe 3+
cation is usually quite small.
The internal magnetic field is the most important of the three
hyperfine interactions in these systems. The magnetically ordered
Fe 3+ cation has a saturation magnetic flux density of between 49
Oxides and Related Systems 161
I
I
I
I I I
I /- I
;'
I'-"---_ r " <
~ ----~-/~~---- -~
--+~ -+-jq-;;--~;' :
I B I
I I
I I
I I
Fig. 7.1 The normal spinel structure, A [B21 04, contains eight octants of
alternating A04 and B404 units as shown on the left; the oxygens build up
into a face-centred cubic lattice of 32 ions which co-ordinate A tetrahedrally
and B octahedrally. The unit cell, Ag[B161 032, is completed by an encom-
passing face-centred cube of A ions, as shown on the right in relation to two
B404 cubes.
and 54 T (490-540 kG), essentially originating in the Fermi con-

tact term B s , and is usually much greater than any field from Fe 2+
which is subject to the orbital and dipolar contributions, BL and
B D • An Fe 3+ cation at a tetrahedral site will have a smaller magnetic
flux density than at an octahedral site because of the greater cova-
lency, but the difference is only of the order of 5% and may not
be fully resolved.
Fe 3+ cations at the trigonally distorted B sites in a spinel show a
quadrupole splitting in the paramagnetic state. For an S-state ion,
this can only arise from the lattice contributions to the electric
field gradient. A number of calculations of this lattice term have
been made, but are often of dubious value . As an example, the
normal spinels Zn[Fe21 04 and Cd[Fe21 04 show quadrupole split-
tings of 0.333 and 0.284 mm s-l respectively at 298 K [I J. The
sign of e 2qQ has been determined by applying an external magne-
tic field and in both cases is negative. However, a simple monopole
calculation of e2qQ, using point-charge summations based on the
known positions of the ions in the crystal structure, gives com-
pletely the wrong sign . It transpires that in this instance the dipole
polarizability of the oxygen anion gives the major contribution
to the electric field gradient, which is of opposite sign to the for-
mal ionic charge summation. The latter is therefore of no value in
isolation. The Fe-O bond distance is shorter in the zinc spinel.
This, together with the decrease of 0.018 mm s-l in the chemical
isomer shift from the value in the cadmium spinel, indicates a
higher degree of covalent bonding in Zn[Fe2] 04.
Fe 2+ cations at B sites show a large quadrupole splitting consi-
stent with the trigonal symmetry (see Chapter 5). The 5A 19 singlet
level (dz2) lies lowest, and is the ground-state for the sixth d-elec-
tron. Thermal population of the 5Eg excited level causes a tem-
perature dependence of the quadrupole splitting. Data for
Ge[Fe2] 04 between 77 and 1020 K are illustrated in Fig. 7.2,
together with the theoretical fit [2] for a 5Alg_5Eg separation of
0= 1145 cm- 1 (=13.7 kJ mol-I; o/k = 1650 K). Distortions from
regular octahedral symmetry in these oxides are usually much
3·0
2·5
C
2·0
, D 'E,
~ 1·5
E
II Sr'g SEg
6 ~
Free Ion Octahedral Trigonal 19
1·0 field field
0·5
°0~--~----~----~--~-----10~00--~
Fig. 7.2 The temperature dependence of the 57Fe quadrupole splitting in

Ge[Fe2104. The solid line corresponds to a 5Alg-5Eg separation of 0 = 1145
cm- 1 (=13.7 kJ mol-I, o/k = 1650 K). ([ 21, Fig.l).
greater than in co-ordination complexes. Thus a particularly large
splitting is also seen in the 99Ru Mossbauer spectrum [3] of
C02+[ ComRum] 0 4. The low-spin d 5 Ru 3+ ion in this oxide has a
similar level scheme to Ge [Fe2] 0 4, but with a filled 2A 19 level
and an eJectron 'hole' in the 2Eg level.
Fe 2+ cations at A sites behave somewhat differently. Fe[Cr2]04
is a cubic spinel phase at high temperatures, but transforms to a
tetragonal symmetry below 135 K. The room temperature spec-
trum is a single line compatible with Fe 2+ ions at the regular tetra-
hedral A site [4]. As the temperature is lowered, a small quadru-
pole splitting appears at ~ 170 K, indicating a departure from cubic
symmetry, and increases steadily until the Curie-temperature is
reached at 90 K. This is illustrated in Fig. 7.3. A tetragonal distor-
tion wi11lower the symmetry of the A site, but it is noteworthy
that the process begins well above the nominal crystallographic
transition temperature, in the vicinity of which the line widths are
particularly broad. Similar behaviour is found in Fe [V 2] 0 4. The
Fe 2+ion is a lahn-Teller ion, and should be unstable in a regular
tetrahedral co-ordination. It is assumed that each Fe 2+ distorts
its surrounding;; into a tetragonal symmetry, but that at high tem-
peratures the direction of its tetragonal axis reorients dynamically
among the three equivalent directions, thus maintaining the effec-
tive cubic structure. The reorientation frequency will be tempera-
ture-dependent. At temperatures well below the crystallographic
phase transition the reorientation is slower than the observation
time of the Mossbauer event so that the electric field gradient is
observed. The broad lines at the transition region are directly due
to this dynamic process.
Comparable behaviouris not found in FeA120 4 because this
spinel is partly inverse and does not therefore have strict cubic
symmetry at the A site. It appears that two different situations
can be defined. In a spinel where the B-site cations are all identi-
calor are very similar, then the A-site symmetry remains cubic at
high temperatures by dynamic motion and gives a static distortion
on cooling. If the B-site cations are dissimilar, and this frequently
means a difference in ionic charge, then there is a static distortion
at the A site with a large electric field gradient. This applies to
FeA1 20 4 and also to the inverse spinel Fe 2+[Fe 2+Ti 4+]04. In both
cases the static A-site distortion persists to beyond 800 K [5].
FeV 204 and FeCr204 both become magnetic at temperatures
100"--~"'-- 294K
98
..
96
94 170K
92
90
-1,0 0 1·0 2·0 3·0 4·0

Velocity/lmm 5- 1)
Fig. 7.3 Mossbauer spectra of Fe[Cr2] 04 showing the quadrupole splitting

associated with the tetragonal distortion. ([4], Fig.3).
well below the cubic-tetragonal transition, but a more curious

behaviour is found in for example Fe[Cr2] S4 which is cubic with
an unsplit resonance line above its ordering temperature of 180 K.
In the ferrimagnetic phase below 180 K there is a quadrupole
splitting which appears to be magnetically induced [6]. This beha-
viour is not unexpected, but proves difficult to interpret theoreti-
cally.
A detailed discussion of magnetically ordered spinels will be
deferred for the moment until Section 7.3 on exchange interactions.
7.2 Non-stoichiometric spinels
The introduction of non-stoichiometry causes many additional
problems in the study of oxides. The cation site symmetry may
no longer be unique, and the M6ssbauer spectrum recorded is then
a weighted summation which can no longer be analysed explicitly.
In the spinel structure for example it is possible to tolerate a large
number of vacant cation sites which we can represent as D. The
classic example of this is in the spinel 'Y-Fe203 which, as there are
32 oxide ions in the unit cell, may be written as Fe21!.D21.032. The
A and B sublattices in this material are magnetically coupled anti-
parallel to each other in the ferrimagnetic phase. The A and B-site
magnetic hyperfine flux densities can be distinguished by the appli-
cation of an external field, Bo. The two internal fields align with
Bo (see Chapter 2) so that the observed flux densities are B'tf =B A
+ Bo and B~ff =BB - Bo. The relative intensities of the A- and B-
site patterns can be used to estimate the site occupancy ratio [7] .
For a thin absorber the resonance intensity is approximately pro-
portional to the concentration, and the recoilless fractions of the
two sites are similar. Thus the site occupancy ratio can be deduced
to be AlB = 0.62 ± 0.05. This result indicates the structure to be
(Fe)[Fes/3Dl/3]04 for which AlB = 0.6; that is, all the vacancies
occur at the B sites. This procedure has been widely used to de-
duce site occupancies in other oxides, and further examples will
be found later in the chapter. The flux densities of B A = 48.8 T
and BB = 49.9 T at room temperature, together with the chemical
isomer shifts relative to iron metal, [i A = 0.27 mm s-l and [iB =
0.41 mm s-l, illustrate the generalizations made earlier on p. 160.
A second possible way of incorporating non-stoichiometry is
by oxygen deficiency, although this is not a usual feature in
spinels. However, a particularly clear example of this is found in
the perovskite phases derived from the iron(lV) oxide SrFe03.
The exact composition is very dependent on the experimental
details of the preparation; for example, in SrFe02.86 the loss of
oxygen is compensated by a partial reduction to Fe 3+. This results
in a tetragonal distortion of the cubic perovskite lattice and the
presence in the M6ssbauer spectrum of contributions from both
oxidation states [8]. Charge compensation in SrFe02.86 requires
the formation of two Fe 3+ cations for each oxygen vacancy, and
there is some evidence that these are closely related as a 'defect
cluster' in the lattice. This may explain why the Fe 3+ ion shows a
remarkably low magnetic flux density of only 42 T in the magneti-
cally ordered phase at 4.2 K.
The conditions of preparation, and the subsequent thermal
history of an oxide sample are very important when there are
multiple site symmetries and/or a defect structure. For example
in the spinel CoFe204, a slowly cooled sample has been prepared
with only 7% inversion, whereas a quenched sample had 24% in-
version [9]. Apart from the obvious effect on the relative inten-
sities of the A and B sublattice spectra, there are also more subtle
changes which are extremely important with regard to the magne-
tic properties. For this reason the effects of partial inversion and
isomorphous replacement in general will be discussed more fully
in connection with exchange interactions in Section 7.3.
It should be noted that the Mossbauer spectrum is not particu-
larly sensitive to small concentrations of lattice defects. Even in
the case of r-Fe203 the presence of a large proportion of B-site
vacancies is seen only as a difference in site occupancy, although
in this example the Fe 3+ is expected to be particularly insensitive
because it is an S-state ion. The primary effect of random fluctua-
tions in the resonant site environment on the spectrum of the
paramagnetic phase is usually a small degree of line broadening.
For this reason the doping of oxides with impurity atoms of for
example 57Fe is not as successful a technique as one might have
hoped.
One of the few systems where a systematic study of the effects
of a defect structure has been made is in the binary oxide, FeO
[10]. This oxide has the rock-salt face-centred cubic lattice, but
is metastable at room temperature. Its preparation involves
quenching from high temperatures and results in cation deficiency,
so that the formula can be written as Fet3xFe~:DxO. Samples
between FeO.9470 and FeO.8600 all show an asymmetrical doublet
spectrum at room temperature with a splitting of about 0.93 mm
s-l. The Fe 3+ ions appear to give a distinct contribution to the
spectrum which is however not resolved. The Fe 2+ environment
is distorted from the ideal cubic symmetry by the presence of
neighbouring defects (hence the quadrupole splitting), but the
random nature of the structure appears to result in a line broaden-
ing which cannot be simply interpreted. For this reason the mag-
netically ordered samples at low temperatures also show complex
spectra.
Thermal annealing of Fel_xO for short periods at temperatures
above 400 K leads to partial decomposition into Fe304, but with
the formation of a residual FeO phase which is very close to
stoichiometry. The reaction can be represented as
This disproportionation can be seen quite clearly in the Mossbauer

spectrum (Fig. 7.4) because the quadrupole splitting and line
broadening of the Fel_xO phase decrease until only a single reso-
nance line compatible with a cubic phase is seen superimposed
.......~...~'.;..~J.:~'
.l 423 K
.'
..
..
'
I I
,.'
" , , I
-1 0 1 2
Velocity/(mm s-')
3
"
Fig. 7.4 The Mossbauer spectrum of FeO.9400 recorded during the precipi-
tation of Fe304 by disproportionation. The sample was held at successive
temperatures for about 2 -3 hours. Note the gradual appearance of an intense
line due to Fe304 on the right, and the narrowing of the Fel_xO line as the
phase tends towards stoichiometry. The line components due to Fe304 are
indicated. ([ 101, Fig. 2) .
168 Principles ofM6ssbauer Spectroscopy
upon the growing Fe304 resonance. Variations in the intensity of
the various spectra with time enable the kinetics of the reactions
to be followed. The spectra of Fel_xO at 1074 K where the com-
pound is thermodynamically stable comprise an unsplit broad line
because of rapid diffusion of the vacancies (see Chapter 6).
7.3 Exchange interactions in spinels

The study of magnetic phases by Mossbauer spectroscopy has
attracted considerable attention because it has permitted a much
more detailed study of the mechanisms of spin-ordering and of the
contributions of different sites to the sublattice magnetization.
More recently, the effects of randomization of cations have been
analysed in detail, and some of the properties of typical magnetic
phases will now be described.
There are two aspects to consider, firstly the nature of the
magnetic ordering and the configuration of the interacting spins,
and secondly the magnitude of the interactions. Both may be
determined from the Mossbauer spectrum. In the simple spin
arrangements for ferromagnetic, ferrimagnetic, or antiferromagnetic
ordering, the one or more sublattices show a collinear spin coupl-
ing. Thus a simple collinear antiferromagnet has two sublattices
with antiparallel spin alignment (Neel ordering). In such an
instance it is sufficient to determine the direction of spin alignment
at anyone site. For example if the site is distorted and there is a
large electric field gradient at the nucleus which may be related to
the geometry of the site, then the combined magnetic-quadrupole
interaction may contain sufficient information to define the direc-
tion of the spin axis. An example of this in FeF2 was given in
Chapter 5. If on the other hand a single crystal is available, the
rela tive intensities of the magnetic lines are angular-dependent and
give the result more directly.
However, many oxide systems and particularly those which are
nonstoichiometric do not have a collinear spin system, and a more
detailed knowledge of spin orientation is a pre-requisite for the
understanding of the magnetic properties in general. One of the
more complicated ordering arrangements in spinels is the Yaffet-
Kittel triangular ordering arrangement, in which there are effec-
tively six sublattices in the antiferromagnetic structure. A further
possibility is a canting of the spin away from the magnetic axis,
which in a ferrimagnet may reduce the total magnetization (but
not the moment at each site) by effectively introducing an anti-
ferromagnetically coupled component in the perpendicular plane.
Alternatively, spin canting in an antiferromagnet can introduce a
small ferromagnetic contribution which becomes the major term
in the susceptibility. Helical spin arrangements in which the spin
is canted from the magnetic axis and successive spins along it pre-
cess in a spiral arrangement are common in antiferromagnets, an
example of this being in the perovskite SrFe03 mentioned earlier.
The exchange interactions which lead to ordering usually take
place via an intermediate oxygen ion and are therefore highly
directional. The coupling of two magnetic ions depends more upon
the M-O-M bond angle than upon the M-M distance. Thus in the
spinel oxides the A-B site coupling far exceeds the A-A or B-B
exchange. This results in some magnetically concentrated oxides
such as Zn[Fe2]04 and Ge[Fe2]04 being paramagnetic except at
extremely low temperatures «20 K), while MgFe204 which is
partly inverse and therefore has some A-B interactions is ferri-
magnetic at room temperature. The value of the Neel or Curie
temperature is in proportion to the strength of the exchange inter-
actions, but more detailed information on the directional nature
of the exchange is usually obtained by magnetic dilution. Thus
the phase (Mn~+Zntx)[Fe~+]04 will show a progressive increase in
the number of A-B interactions as x increases because Mn 2+ is a
magnetic cation, whereas Zn 2+ is diamagnetic.
Some of the recent Mossbauer data for substituted spinels have
shown conclusively that the magnetic field of an Fe 3+ cation is
dependent on the nearest-neighbour cation environment, parti-
cularly when it occupies the B site. For example NiFe204, which
is completely inverse and therefore has Fe 3+ at all six A sites about
each B site, shows very sharp resonance lines, whereas CoFe204,
MnFe204 and MgFe204 which are all partly inverse show a broad-
ening of the hyperfine lines from the B sites due to the variations
in the cation distribution at the nearest A sites [9]. A slowly
cooled sample of CoFe204 with a cation distribution of (COo.07FeO.93)
[CoO.93Fel.07]04 shows much less broadening than a quenched
sample with a distribution of (Coo.24Feo.76)[CoO.76Fe1.24] 04· A
detailed analysis of the lineshapes from this sample has been made.
The probability of finding n C02+ ions on the six nearest-neighbour
A sites when the site occupancy is truly random is given by the
binomial distribution
6'
Pen) = ---' -- (1 - c)6-n c n
n!(6-n)!
where c is the fractional occupation of the site in the bulk sample.

For a value of c = 0.24, P(O) = 0.19, P(1) = 0.37, P(2) = 0.29 and
P(3) = 0.12. The measured magnetic flux densities at room tem-
perature were B(O) = 51.5, B(1) = 49.9, B(2) = 47.5 and B(3)
= 44.5 T with a flux density at the A site of 49.9 T. The analysis
of one of the component lines, that at extreme negative velocity,
is shown in Fig. 7.5. A similar analysis has been made for MnFe204.
These differences in flux density have been attributed to a super-
transferred hyperfine interaction, that is a transfer of spin-imbalance
from one metal ion to another via the oxygen anion [11]. The
details are too involved to give here, but this phenomenon is rein-
troduced later in the chapter in a different context (see p. 179).
Very complex spectra are found for magnetic oxides in which
I ,
( l
o ~
I
c:
o
';;;
'"
'e .'.
.,c:'" ....
~
Co FtZ04 quenched
-9 -8 -7 -6
Velocity I (mm 5- 1 )
Fig. 7.5 One of the lines in the Mossbauer spectrum of CoFe204 (quenched
sample) in an applied flux density of 1.7 T. The A site is relatively unaffected
by the cation distribution, but contributions from iron atoms at B sites with
0, I, 2 and 3 Co2+ cations on the six nearest neighbour A sites are identified.
([9], Fig. 4).
the magnetic cations are diluted by a non-magnetic substituent.
At low temperatures the spectrum of such an oxide containing Fe 3+
ions is usually a six-line spectrum, if somewhat broadened, but as
the temperature increases there tends to be a progressive inward
collapse, often with the appearance of an apparently paramagnetic
central component before the ordering temperature is reached.
This is illustrated in Fig. 7.6 for some data for (Zno.sFeo.s)
[Co o.s Fel.5] 0 4 where the most probable cation distribution is of
three Fe 3+ and three Zn 2+ cations about each B site [12]. How-
ever, if the cation distribution is truly random, no less than 11 %
455K :,,_ ,:" '(!' 340 K

f-:~l
:. ,
:,:.,
.~ ::.l~~~~;\~ .~ ~ .//~~'¥;' :~:
! " ;:\:,
.~ ""!<:< : .,; .;!~t{'
l~'f~q~ ;.!tt.~.:~:~
/'1:/387 K
:.",~i'f,,'!')·,,//\r\,:~ 222 K
,..,.
'il? 367 K
:~~4i:~~-) :. a.:.;JJ.:,,-::.;
J '
'."" " I ~,i

'\.f~ ~;,,/ .
359 K
Velocity
Fig. 7.6 The temperature dependence of the S7Fe M6ssbauer spectrum of

(ZnO.SFeO.S)[CoO.SFeLSI 04 showing the inward collapse of the hyperfine
splitting with temperature increase and the appearance of a 'paramagnetic'
line below the Neel temperature of 448 K. ([ 121, Figs. I, 2).
of the B sites have only one or no iron neighbours, and if occupied
by iron can be coupled only weakly to the lattice.
The spectra can be simulated by a stochastic model for spin
fluctuations which gives excellent agreement with experiment over
the whole temperature range. It therefore seems likely that those
Fe 3+ cations which are only weakly coupled to the lattice are ex-
periencing a fast relaxation well below the Curie temperature.
However, it is possible to explain at least some of the broadening
without invoking relaxation. From molecular field theory it is
possible to show that those Fe 3+ cations which have fewer neigh-
bours will have a different temperature dependence of the hyper-
fine field [13 l. In particular, the initial rate of decrease in the
hyperfine field is greater and results in the observed inward broad-
ening. This effect has a similar appearance to motional narrowing,
but even for high magnetic dilution, as in the present example, it
seems that the initial change in shape of the spectrum that accom-
panies the rise in temperature need not invoke a dynamic spin
fluctuation. However, the molecular field theory approach appears
to be inadequate in the region just below the ordering temperature.
The spinel phase COl_xZnxFe204 also serves as an example of the
determination of a complex spin configuration [14 l. The Zn2+
preferentially enters A sites as in Zn[ Fe2l 04, but the degree of
inversion in ferrimagnetic CoFe204 is dependent on thermal treat-
ment of the sample. Fe 3+ is an S-state ion, and this means that
whatever the zero-field direction of the magnetic sublattice, it
can be easily rotated by the application of an external field with
which it aligns parallel or antiparallel. However, antiparallel Neel-
coupled sublattices for example remain antiparallel (except in
extremely large applied fields) so that the respective flux densities
are increased or decreased by the value of the applied field. Even
where the flux densities of different sublattices are very similar,
they can often be distinguished by appropriate choice of the appli-
ed field strength (this method has already been described in con-
nection with r-Fe203). If spin-canting occurs, it cannot be easily
destroyed, and although the resultant magnetic axis may align
with the field, the individual spin axis will remain canted. This
provides a means of distinguishing different cases. If an external
field is applied to an Fe 3+ oxide to align the magnetic axis parallel
to the direction of observation, the Llm = 0 components of the
Mossbauer magnetic splitting should have zero intensity. This
only remains true if each individual spin moment lies along this
axis; any canting from the direction of observation will introduce
a finite intensity into the Ilm = 0 lines.
In the present example, the application of an external field of
flux density 8 T to CoFe204 separates the A and B sublattice
hyperfine lines, providing a means of determining the A and B
site occupancy and of detecting spin canting. The spectra for
x = 0 and x = 0.6 at 4.2 K are shown in Fig. 7.7. The variation
in the intensities of the ouer lines can be used to determine the
site occupancies which are (FeO.81COO.l9)[CoO.81Fe1.l91 04 and
90 -..... -:..... . ...,., ............ ...... ::...-.. " ."

..-:.::'.....
"'- ..'.:....: ..: ........ ~
85
':.:
(Q)
80
51
-',: ;,'.... .....~. . :::.:...........0.-·
.:~ : ....
.
-:' ::.:.. .:::01'::" .::,.•...: ..:.::. ~.:.........:':. .... . .,.:.......~:- '::.:
'"
I 49 "
S2
....
" :
~ 47
C
:::I
o
(.)
(b)
.......... ...... :..._.
192 - ••.:..,........... ,
... ',' .....
...........'_.
0" ,"
..
.....
..··0, ....... .:..: ...... .,....... ..... . '.,~': :"'.
168 :.:
t i
184 A B BA
180
(e)
-10 -8 -6 -4 -2 0 2 4 6 8 10
Velocityf(mm s-')
Fig. 7.7 57Fe Mossbauer spectra for CoFe204 at 4.2 K in (a) zero magnetic
field (b) a flux density of 8 T applied along the direction of observation and
(c) C00.4ZnO.6Fe204 at 4.2 K in an externally applied flux density of 8 T.
The variation in the intensities of the outer lines allows a determination of
the site occupancy, and the weak intensity of the Ilm = 0 lines in (b) and
particularly in (c) indicates the presence of spin canting. ([ 14] , Figs. 1, 3).
(ZnO.6Fe0.4) [CoO.4Fe 1.6] 04· As can be seen the cobalt is quickly
displaced from the A sites by zinc. The f:..m = 0 lines are weakest
in CoFe204, but increase rapidly in intensity above x = 0.4, and
are very intense for x = 0.8 (not shown). Although CoFe204
approximates to a Neel ordered ferrimagnet, spin canting is pre-
sent at both sites and increases with zinc substitution. Detailed
study of the lineshapes for x = 0.8 provides evidence that individual
B-site spins become increasingly canted as the number of Zn 2+ ions
in the nearest neighbour A sites increases. Therefore in any given
sample, the spins are canted by different angles with a sta tistical
distribution.
7.4 Rare-earth iron garnets

Some good examples of the complex exchange interactions which
can be studied are provided by the rare-earth iron garnets,
R3Fes012, and their substituted derivatives. The large unit cell
contains 24 dodecahedral (c) sites occupied by R3+ cations, 16
octahedral (a) sites occupied by Fe 3+, and 24 tetrahedral (d) sites
also occupied by Fe 3+ cations; it is conventionally written as
{R3}[Fe2] (Fe3)012. The structure is represented in Fig. 7.8 [15].
Mossbauer spectroscopy is particularly useful in complicated
systems such as these because it is possible to study independently
the occupancy and magnetizations of all three sites, and to deter-
mine the relative magnitudes of the different exchange interactions.
Both iron sites are distorted, the a-site symmetry being trigonal
and the d-site tetragonal. As a result there are four differently
oriented but otherwise identical a-sites and three d-site orienta-
tions in the unit cell. These are not always equivalent orientations
in the magnetically ordered phase because there is then a unique
direction of the magnetization in the unit cell. In polycrystalline
samples the small Fe 3+ quadrupole interactions tend to average out
as an apparent line broadening of the magnetic lines.
In some garnets, such as yttrium iron garnet (YIG), Y3FeS012,
the tervalent rare-earth ion is diamagnetic so that exchange forces
concern solely the a and d sites, but in others such as the samarium,
europium and dysprosium iron garnets, the rare-earth also has a
sublattice magnetization, and the potential number of exchange
interactions is much greater. The relative importance of these can
be determined by selective magnetic dilution with a non-magnetic
(±-Z1 0 )
(fto)~----------------------~'
~----·---------------4-----~(t11)
I
I
I
I
I
I
I
I
I
-~
"-
"-
"-
"-
Fig. 7.8 A schematic representation of part of the unit cell of the

{R3} [Fe2l (Fe3)012 lattice showing the inter-relation of the three cation
site symmetries. (after [ lS 1).
cation. Thus in the phase {Y 3} [Fe2l (Fe3_xAlx)012 the number

of a-a interactions remains constant, whilst the numbers of a-d and
d-d interactions decrease as the aluminium content increases.
As a general rule the major exchange coupling is between a and d
sites, and gives a ferrimagnetic anti-parallel coupling of the two
sublattices. The rare earth is then weakly coupled anti-parallel to
the resultant moment of the iron sublattices, but the dominance of
the a-d coupling results in the Curie temperature being close to
560 K regardless of whether or not the rare earth is magnetic.
However, some of the finer details of the magnetic ordering in a
stoichiometric iron garnet are dependent on the rare-earth. For
example, in Y3FeS012 the magnetization lies along the [Ill]
direction so that there are two non-equivalent octahedral a-site
orientations with all tetrahedral d sites equivalent. The correspond-
ing s7Fe spectrum at 300 K shows magnetic flux densities of 49.0
and 48.4 T at the a sites and 39.3 T at the d site with the relative
intensities of 3: 1 :6 as predicted [16]. As with the spinels, the
flux density at the tetrahedral site is much smaller than those of
the octahedral sites. By contrast, in Sm3FeS012 the magnetization
is along the [110] direction and there are now two non-equivalent
d-site orientations. The flux densities at 295 K in this case are
49.5 T at the a sites and 40.8 and 40.6 T at the d sites with rela-
tive intensities of 4:2:4. It is possible to distinguish the two d sites
because they show different quadrupole splittings of e 2qQ/2
= -0.20 and +0.10 mm s-l respectively which perturb and separ-
ate the magnetic lines.
The importance of the a-d interaction can quite easily be seen
from a comparison of Curie temperatures. For example Y3FeS012
(Tc = 550 K) and Gd3Fes012 (560 K) are typical of garnets with a
strong a-d exchange, whereas compounds with mainly a-a inter-
actions such as {Ca3} [Fe2] (Si3)012 and {Ca3} [Fe2] (Gd3)012, or
d-d interactions as in {YCa2}[Sn2](Fe3)012 and {NaCa2}[Sb2] (Fe3)012,
or c-c interactions as in {Fe~+} [A1 2] (Si 3)012 are all paramagnetic
above 80 K [1 7] .
Each a site has 6 exchange linkages via oxygen to d sites, and
each d site has 4 linkages to a sites. Any given magnetic ion must
have two or more linkages with magnetic cations in the other sub-
lattice to maintain continuity in the ferrimagnetic ordering. Thus
in the phase {Y3}[Fe2](Fe3_xAlx)012, when x = 1.7 18% of the
a-site Fe 3+ ions have either none or one exchange linkage.
As we have already seen, weakly coupled atoms show a large
decrease in the magnetization and a consequent decrease in the
internal magnetic flux density. This effect will be particularly
important at temperatures close to the Curie temperature. For
example in the phase {Y 3-xCax} [Fe2] (Fe3-xSix)012 there is a
substantial inward collapse of the spectrum above x = 1.5 at room
temperature (see Fig. 7.9) as the magnetic interactions weaken
[ 18]. Note the reversal in the intensities of the outer pairs of
lines between x = 0.15 and x = 1.2. The extreme outer lines are
due to a sites and become more prominent as the contribution
from the d sites is reduced. This provides a means of determining
which site is substituted.
Although the ground state of Eu 3+ is nominally the non-
magnetic 7Fo level, crystal-field and exchange interactions mix in
the TFJ excited states so that EU3Fes012 shows ordering at both
T=78K
.:.•.:;:":.::-":. :;.. ' .
., .._'::'0....."" ',":;.. :::..:.
...
. ',
, ... .. :.I.:~
".~ "
;. : ..........
'.'
',:. ::'
.:": x =0,15
.. ':'
'.'
.. ..
....: :
(.'r.:...: . . . :.:..":. ~':""':. .. . . . .
....r
~: . ':...
. :~:. . " ....'.... x =0-75
• ':.0.:
",'::.:'
. ........... ...... -:.~:':. ',::',. .
'0.:-, ............
.. . '::.' .
.:.:::: ',:::
c:::
"
'~ " .--. ....
x = 1·2
·e "", '',..
".
','
'"c:::
"'-
I-
. ....
. : / . ........ ,,',',,:. x=1·5
.. :.' ' . :'... • . ... 0'
" ",
,",' ......:...:.::. ....:-:...: .....

.....~ ......
. ".: :'" .... :.............. .
......: ... :-'
.:: .....
Velocity
Fig. 7.9 Spectra of the garnets Y 3-xCaxFeS-xSix012 showing how the

magnetic hyperfine splitting is affected by the magnetic dilution for values
of x greater than 1.5. ([ 18] , Fig. I).
the europium and iron sites. All the c-sites are not in fact equiva-
lent because the iron sub1attice magnetization is in the [111] direc-
tion, and there are two 151 Eu hyperfine flux densities of 56.1 and
62.3 Tat 4.2 K [19]. Substitution of gallium for iron takes place
preferentially at the d sites to give {EU3} [Fe2] (GaxFe3-x)012. The
exact site occupancy can again be determined from the s7Fe spectra,
and in fact 85% of the gallium is in d sites for x = 0.66 and 73% for
x = 3.03 [20]. It is interesting to note that the a- and d-site mag-
netizations are opposed to each other, and the net magnetization
vanishes at X""" 1.4 because of 'magnetic compensation'.
The lSlEu resonance at 4.2 K is altered dramatically with increase
in x, and as can be seen in Fig. 7.1 0, the magnetic splitting collaps-
es completely. The europium field is generated by an exchange
interaction with the two nearest-neighbour Fe atoms at d sites,
which will therefore be weakened by the gallium substitution.
The observed spectra can be simulated successfully using a simple
model. Each europium ion can have one of three environments:
(a) two Fe 3+ neighbours at d sites (statistical weight y2 where y is
the concentration of iron on d sites determined from the 51Fe
spectra); (b) one Fe 3+ and one Ga 3+ (weight 2y[ 1 - y]); (c) two
Ga 3+ neighbours (weight [1 - y]2). The exchange mechanisms
produce a flux density at the Eu 3+ of Be, O.52Beand 0 respectively.
The solid curves in the figure were calculated using essentially
1'14
1'12
0,42
0'41
0·62
X=1'26
0'60
§0'59
~
.~ X=I'60
c:
E
... 0'57
0'55
X=2,37
0·53
1,40
X=3·03
1·37
-60 -40 -20 0 20 40 60

Velocity I (mm 5-1 )
Fig. 7.10 151Eu spectra of gallium-substituted europium iron garnets of the

type Eu~Fe5_xGax012 showing the reduction in the exchange interaction at
the Eu 3 ion with increasing gallium substitution. ([20], Fig. 3).
this simple model, but including an additional contribution to the
exchange field of 12% from the four third-nearest neighbours on
d sites.
Scandium substitution takes place exclusively at the a sites,
as in {EU3} [ScxFe2_x](Fe3)012, and one would expect that the
Eu-Fe exchange via d sites would be unaffected. However, partial
collapse of the 151Eu magnetic spectrum is also seen here [21] .
It appears that in this case the iron d-site spins are canted from
the [111] direction by an angle determined by the number of a-d
site exchange linkages, and as a result the c-d interactions show a
statistical variation. Complex spin-canting can also be identified
in {EuxSml_x} [Fe2] (Fe3)012 where the magnetization is along the
[111] direction for x = 1 and along the [110] direction for
x = 0 [22] .
It will now be clear that the effects of substitution on the magne-
tic exchange interactions can be very subtle and complex. Where
multiple sublattices are involved, M6ssbauer spectroscopy is unique-
ly successful in elucidating the nature of the operative mechanisms.
7.5 Transferred hyperfine interactions

Although a diamagnetic cation or anion has no resultant spin and
therefore cannot order, it may nevertheless be influenced by the
close proximity of another cation with a magnetically ordered
spin. The electrons on the diamagnetic atom then become polarized
by exchange interactions so that there is a small resultant imbalance
in the spin density at the nucleus and consequently a magnetic
hyperfine field. These transferred hyperfine interactions can be
quite large, and in 119Sn for example, magnetic flux densities of
over 20 T have been found in some oxide systems.
An interesting example is given by the garnets {Y3-xCax}
[Fe2-xSnx] (Fe3)012. The Sn4+ diamagnetic cations at the a site
are all surrounded by six F e3+ cations at d sites. It was made clear
in the preceding section that the a-d exchange between iron atoms
is not appreciably affected by the initial magnetic dilution, and in
this case there is no effect until x> 0.9. The 119Sn spectrum shows
a large magnetic flux density (Fig. 7.11) which, being produced
by six Fe 3+ neighbours is insensitive to the degree of tin substitution
below x = 0.9 [23, 24]. However, the breakdown of long-range
'"
~
!
c:
"
o
u
180 L-__ ~ ______________- L______________ ~ __ ~
- 15 o 15
Velocity/(mm s-' )
Fig. 7.11 The 119Sn spectrum of CaO.25Y 2.75SnO.25Fe4.75012 at 77 K show-

ing a magnetic hyperfine splitting of over 20 T from a transferred hyperfine
interaction. ([23], Fig. 1).
order between x = 0.9 and 1.2 results in a rapid collapse of the flux
density at both the 57 Fe and 119Sn nuclei.
The cation Sb 5+ is isoelectronic with Sn 4+, and transferred hyper-
fine interactions are also known for antimony in the spinels
Nil+2xFe2-3xSbx04 [25]. The antimony substitutes at the octa-
hedral B sites, but here the average number of Ni 2+ neighbours at
A sites alters with increasing substitution. The result is a more
complex spectrum in which each possible combination of A-site
neighbours produces a different hyperfine flux density at the
antimony.
Transferred hyperfine interactions can also occur at the anion,
and an example of this can be found in the ferromagnetic spinel
CuCr2Te4 where the 125Te resonance reveals a flux density of 14.8
T at 80 K [26] .
References
[1] Evans, B. J., Hafner, S. S. and Weber, H. P. (1971)J. Chern. Phys., 55,
5282.
[2] Eibschutz, M., Ganiel, U. and Shtrikman, S. (1966) Phys. Rev., 151,
245.
[3] Gibb, T. C., Greatrex, R., Greenwood, N. N., Puxley, D. C. and
Snowdon, K. (1973) Chern. Phys. Letters, 20, 130.
[4] Tanaka, M., Tokoro, T. and Aiyama, Y. (1966)J. Phys. Soc. Japan,
21,262.
[5] Ono, K., Chandler, L. and Ito, A. (1968) J. Phys. Soc. Japan, 25, 175.
[6] Eibschutz, M., Hermon, E. and Shtrikman, S. (1967)J. Phys. Chern.
Solids, 28, 1633.
[7] Armstrong, R. J., Morrish, A. H. and Sawatzky, G. A. (1966) Phys.
Letters, 23,414.
[8] Gallagher, P. K., MacChesney, J. B. and Buchanan, D. N. E. (1964)
J. Chern. Phys., 41,2429.
[9] Sawatzky, G. A., van der Woude, F. and Morrish, A. H. (1969)Phys.
Rev., 187, 747.
[10] Greenwood, N. N. and Howe, A. T. (1972) 1. Chern. Soc. (A), pp. 110,
116 and 122.
[11] van der Woude, F. and Sawatzky, G. A. (1971)Phys. Rev., 4, 3159.
[12] Iyengar, P. K. and Bhargava, S. C. (1971)Phys. Stat. Sol. B, 46, 117.
[13] Coey, J. M. D. and Sawatzky, G. A. (1971) Phys. Stat. Sol. B, 44,673.
[14] Pettit, G. A. and FOf€!ster, o. W. (1971) Phys. Rev. B, 4,3912.
[I5] Gilleo, M. A. and Geller, S. (1958) Phys. Rev., 110, 73.
[16] van Loef, J. J. (1968) 1. Appl. Phys., 39, 1258.
[17] Lyubutin, I. S. and Lyubutina, L. G. (1971) Soviet Physics-Crystallo-
graphy, 15,708.
[I8] Lyubutin, I. S., Belyaev, L. M., Vishnyakov, Y. S., Dmitrieva, T. V.,
Dodokin, A. P., Dubossarskaya, V. P. and Shylakhina, L. P. (1970)
Soviet Physics-JETP, 31,647.
[19] Stachel, M., Hufner, S., Crecelius, G. and Quitmann, D. (1969) Phys.
Rev., 186,355.
[20] Nowik, I. and Ofer, S. (1967) Phys. Rev., 153,409.
[21] Baurninger, E. R., Nowik, I. and Ofer, S. (1967) Phys. Letters, 29A,
328.
[22] Atzmony, U., Bauminger, E. R., Mustachi, A., Nowik, I., Ofer, S. and
Tassa, M. (1969) Phys. Rev., 179,514.
[23] Goldanskii, V. I., Trukhtanov, V. A., Devisheva, M. N. and Belov, V. F.
(1965) ZETF Letters, 1, 19.
[24] Belov, K. P. and Lyubutin, I. S. (1966) Soviet Physics-JETP, 22,518.
[25] Ruby, S. L., Evans, B. J. and Hafner, S. S. (1968) Solid State Cornrn.,
6,277.
[26] Ullrich, J. F. and Vincent, D. H. (1967) Phys. Letters, 25A, 731.
CHAPTER EIGHT
Alloys and Intermetallic Compounds
One of the major applications of Mossbauer spectroscopy is in the

field of metal physics. Many of the physical properties of a metal
such as the electrical conductivity and magnetic susceptibility are
macroscopic; i.e. they derive from the bulk solid and in particular
from its collective-electron band structure. The Mossbauer tech-
nique differs in that it records individual atoms in the metal. This
enables a detailed study of the near-neighbour interactions with
the resonant nucleus, and of the effects of changing composition
on the electronic and magnetic interactions of particular atoms
within the alloy. Many of the principles enunciated in earlier
chapters are still applicable, but some new phenomena are found
and the more important of these will now be discussed. Some of
the analytical applications in applied metallurgy are indicated in
Chapter 9.
Most interest is centred on magnetically ordered alloys, and in
this context it is important to consider the origin of the internal
magnetic field at the resonant nucleus. If the Mossbauer isotope
is a transition metal such as iron, then the band-structure of the
alloy may be such that there is still an effective localized d-electron
magnetic moment on the atom. It is no longer possible to speak
in terms of an integral number of d-electrons, as in Chapter 5, but
this unpaired d-band spin-density can induce an imbalance in the
s-e1ectron spin-density at the nucleus, and hence a magnetic flux
density equivalent to the Fermi interaction described on p. 110.
This effect is known as 'core-polarization'. In addition, there may
182
Alloys and Intermetallic Compounds 183
be a contribution from unpaired conduction-electron spin-density
which usually has considerable s-character and therefore also con-
tributes significantly to the observed magnetic flux density. This
is referred to as the 'conduction-electron polarization'. The rela-
tive importance of these two effects and the way in which they
are influenced by change in composition of the alloy can often
be determined by Mossbauer spectroscopy. The temperature
dependence of the magnetic flux density below the ordering tem-
perature is essentially similar to the behaviour found in ionic com-
pounds (see p. 112) and will not be discussed further.
It is convenient to consider any metallic phase as belonging to
one of two classifications. A disordered alloy is one in which two
or more elements occupy the same crystallographic sites with a
random probability. Such a phase frequently has a wide range of
composition. An intermetallic compound differs in that each ele-
ment shows strong preference for a particular lattice site. The
result is a regular structure with a composition close to a simple
stoichiometry and a very restricted range of composition. How-
ever, there are intermediate examples where an alloy shows a vari-
able amount of atomic order/disorder according to its previous
thermal and mechanical history.
8.1 Disordered alloys

The chemical isomer shift and the magnetic flux density at the
nucleus in an alloy are intimately related to the electron band-
structure. A systematic study of the variation of these parameters
with change in composition, can therefore be expected to yield
significant information about the structure of the alloy.
A good example of the use of the chemical isomer shift in non-
magnetic alloys is provided by the solid-solution of palladium with
0-16.5 at.% of tin. This alloy has the face-centred cubic struc-
ture [1]. The paramagnetic susceptibility of the alloy decreases
rapidly with increasing tin content from the value in pure palla-
dium, and approaches zero at the limit of solubility. Surprisingly,
the 119Sn chemical isomer shift is invariant over the whole range
of composition (-1.0 mm s-l with respect to (Hin metal). This
shift is also characteristically different from that in any of the tin-
rich intermetallic compounds Pd3Sn, Pd2Sn, PdSn, PdSn2 and
PdSn4, thereby indicating that the electronic configuration of the
tin in the solid-solution phase is indeed independent of composi-
tion. The large negative value of the shift denotes a significant
decrease in the 5s-electron density at the tin compared to the pure
metal. The paramagnetic susceptibility of palladium can be attri-
buted to a spin-moment from partial filling of the 4d-band. Both
the magnetic and Mossbauer data for the alloy can be explained
by a simple rigid band model in which the tin valence electrons
are donated to the unfilled 4d-band of the palladium. The latter
apparently becomes filled at the limit of solid solubility. The
effectively complete loss of electrons from the 5s-band of the tin
results in the large negative chemical isomer shift.
The itinerant nature and the high s-character of the binding
conduction-electrons tend to minimize the effects of variations
in the near-neighbour environment on the Mossbauer spectrum.
Consequently, any influences of such variations on the quadru-
pole splitting or chemical isomer shift are usually obscured by
the width of the Mossbauer line, and can be tacitly ignored. How-
ever, this is not always the case with magnetic interactions.
In a pure metal phase such as ferromagnetic a-iron (body-
centred cubic lattice) there is only one type of iron site. The
57Pe Mossbauer spectrum is therefore a single 6-line magnetic
hyperfine pattern, the magnetic flux density of 33.0 Tat 294 K
originating largely from core-polarization. Consider in the first
instance a dilute substitution of an impurity metal into the a-iron
structure. Each impurity atom may be expected to have a signi-
ficant effect on the charge distribution and spin-density at the
neighbouring iron sites. The effect on the Mossbauer spectrum
will be a change in the observed magnetic flux density at all the
iron sites affected which is dependent on the nature of the im-
purity atom and on its distance from the resonant atom. As a
result, the observed 57Fe spectrum is a statistical summation of
all the different contributions, and the single hyperfine pa ttern is
replaced by a broadened envelope with only a poorly resolved fine
structure.
Four examples of the spectra of alloys of a-iron containing a
small proportion of a second metal are shown in Fig. 8.1 [2]. The
complicated line-shapes result directly from the multiple environ-
ments. The major experimental problem lies in the correct analy-
sis of these line-shapes. A systematic survey [2] of the transition
metals alloyed with iron has shown that Ti, V, Cr, Mn, Mo, Ru, W,
Alloys and In term eta llic Compounds 185
1 n nr,!0 (, ,',.-
:i V ~. V \' \. ,
~
'{
. ~r j ::"'V' '\ r\ .'--

~
i
i V J/
"'" ", r \ ,-... r~: r'\
'. r;it....-.. r
\/ \/
~: ".
V V \:
Pt ; j
\!
:;
~ +. "-; ~
i
-6 -4 -2 0
V.locity I (.,," ,-I)
Fig. 8.1 The 57Fe spectra at room temperature of alloys derived from a-iron
by substitution of (top to bottom) 2.2 at.% Cr, 1.0 % Mo, 5.0 % V, 3.0 % Pt.
([2], Fig. I) .
Re and Os impurities (i.e. those elements to the left of or below

iron in the periodic table) give a substantial reduction in the effec-
tive hyperfine flux density with observable satellite lines in the
spectrum. On the other hand, Co, Ni, Rh, Pd, Ir and Pt (i.e. ele-
ments to the right of iron) give only slight broadening but with an
increased magnetic flux density. This is probably indicative of
the relative effects on the iron 3d-band.
More detailed analysis is best achieved in conjunction with com-
plimentary measurements by spin-echo and continuous-wave
nuclear magnetic resonance experiments, which can also be used
to study variations in magnetic flux-density at the iron nucleus.
It is usually only possible to determine the flux density for iron
atoms contributing to the two most prominent satellite resonances
(generally assumed to be due to an impurity in the first or second
near-neighbour co-ordination shells; a total of 8 + 6 sites in the
body-centred cubic lattice). For example in the Mo alloy, the
field is reduced by 3.87 ± 0.05 T for an iron with one molybdenum
atom in the first co-ordination shell, and by 3.16 ± 0.07 T for the
second co-ordination shell.
A recently described experiment circumvents some of the prob-
lems in an ingenious way [3]. A single crystal foil of Fe containing
4.8 at.% of chromium was magnetized successively along the [111]
and [001] axes of the crystal. The resulting differences between
the two corresponding Mossbauer spectra lead to a more positive
identification of the satellite components. It is suggested that it
is the 1st and 5th (not the second) co-ordination shells which show
large reductions in the magnetic flux density of -3.30 T and
-2.37 T respectively, the 2nd, 3rd and 4th co-ordination spheres
producing little or no change. These conclusions lead to an un-
expected picture of the spin-density disturbance about the chro-
mium. It has frequently been assumed that the effect will either
decrease with increasing distance or show a damped oscillatory
behaviour. In the body-centred cubic lattice, the first and fifth
neighbours occupy adjacent sites along the diagonal axes of the
cubic cell, e.g. (111) and (222). It would therefore appear that
the spin-density disturbance is large along the diagonal axes (1 st
and 5th neighbours), but close to zero along the cube axes (2nd
neighbour) and intermediate directions (3rd and 4th neighbours).
It remains to be seen whether this directional model is applicable
to any of the other dilute impurities in a-iron.
Alloys with a higher concentration of the second element fea-
ture a significant probability that any iron atom has two or more
impurity atoms in its first co-ordination shell, resulting in a large
increase in the number of site environments. As just described,
the gross effect of one nearest neighbour is to produce a shoulder
(albeit of complex shape) on the main resonance peaks, correspond-
ing to an average change in magnetic flux denc;ity of about 2 T.
Those atoms with two, three, four etc. substituted neighbouring
sites generate additional shoulders, the average change in flux den-
sity being approximately proportional to the number of neighbours.
This behaviour is found for example in the disordered Fe-Si alloys
« I 0 at.% Si) where components corresponding to 0, 1 and 2 near-
neighbours may be discerned [41.
Near-neighbour effects are not confined to the 57Fe resonance,
although much less data are available for other isotopes. An
interesting example is given by the nickel-palladium alloy system
[5]. The magnetic flux density at the 61Ni nucleus in ferromagnetic
nickel metal at 4.2 K is -7.6 T. The effective electronic moment
per nickel atom as determined by magnetic measurements is 0.70
J1 B. The Ni-Pd alloys are ferromagnetic over the whole of the con-
centration range 0-98 at.% Pd, the effective moment per nickel
atom, rising to about 3J1B in the Pd-rich alloys because of an induc-
ed magnetic effect on the palladium host.
The 61Ni Mossbauer spectra show considerable broadening of
the hyperfine pattern in the intermediate concentrations, due to
variations in the flux density caused by near-neighbour effects.
Unfortunately the resolution of the fine-structure is inadequate for
a detailed analysis. However, it is significant to note that the
'average' flux density changes in a regular manner, and reverses
sign at about 53 at.% Pd. The data are illustrated in Fig. 8.2. The
normal contributions to the magnetic flux density from core-
polarization by the nickel 3d-moment and from conduction-
electron polarization are both expected to be negative in sign in
accord with measurements on other Ni-based alloys. However, the
observed behaviour in the Ni-Pd alloys can only be explained if a
third positive term is invoked which arises from a polarization
of the nearest neighbour Pd atoms by the nickel magnetic moment
via the conduction electrons. This contribution to the 'average'
flux density observed is directly proportional to the 'average' num-
ber of Pd atoms in the first co-ordination shell of the nickel. The
solid curve in Fig. 8.2 represents a theoretical model on this basis,
and the good agreement with experiment confirms the idea that
the bulk magnetic properties of the Pd-rich alloys are determined
not by the d-band of the palladium host but by clusters of polariz-
ed Pd atoms about each Ni atom.
-
I-..
z
Palladium (at. Dfo)
Fig.8.2 The variation of the magnetic flux density at 61Ni at 4.2 K as a

function of concentration in the Ni-Pd alloy system. The triangles represent
the experimental average value of IB I. The circles are values of B where the
sign was observed to be negative. The squares represent estimated values
from a distribution of fields. The solid line is a proposed theoretical model.
([5], Fig. 6).
8.2 Intermetallic compounds

A binary intermetallic compound usually has a restricted range of
composition close to a simple stoichiometry such as AB 4 , AB3,
AB 2, A2B3, AB etc., and has a correspondingly simple crystal struc-
ture in which the two elements occupy distinct crystallographic
sites. In a sense these compounds are intermediate between the
ionic lattice with localized electrons and the true random alloy.
The uniformity of the atomic environments ensures that the
Mossbauer spectrum no longer shows the line broadening found
in a solid-solution alloy. Consequently, any small quadrupole
interaction usually remains visible and can be related directly to
the local site symmetry.
There has been considerable interest in the study of magneti-
cally ordered intermetallic compounds, and it is not uncommon
to find subtle complexities in the magnetic behaviour. A good
example is given by the cubic Laves phases, MFe2, where M is a
rare-earth atom. The iron atoms are all crystallographically equi-
valent and occupy corner-sharing tetrahedral networks with a
three-fold symmetry about the [111] directions. As a result, the
electric field gradient at the iron is axially symmetric with Vzz
along this axis. The direction of the magnetic axis is determined
by the rare-earth element. For example DyFe2 and HoFe2 give a
single 57Fe magnetic hyperfine pattern (HoFe2 illustrated in Fig.
8.3) showing that the iron sites are all magnetically equivalent
[6, 7]. This can only be true if the axis of magnetization is along
[001] so that Vzz has the same relative orientation at each site
with respect to the magnetic flux density, B. In other phases such
as ZrFe2, YFe2, TbFe2, ErFe2 and TmFe2, the Mossbauer spectrum
shows two very distinct magnetic patterns in the area ratios of
3: 1 (ErFe2 illustrated in Fig. 8.3). In these compounds the mag-
netic axis is along [111] so that Vzz is parallel to B for {- of the iron
atoms and at 1080 toBfor~ of the atoms.
A compound containing a magnetic element for which no
Mossbauer resonance is available may often be studied indirectly
via the transferred hyperfine interactions at a non-magnetic
nucleus. For example some of the compounds of gold with the
3d-metals show ordering of the 3d-magnetic moments, which
results in transferred magnetic hyperfine interactions in the 197Au
resonance. A particularly interesting series of examples are pro-
vided by some of the Au-Mn phases.
The 197Au Mossbauer resonance in a solid-solution alloy of gold
with 5 at.% manganese shows two superimposed hyperfine pat-
terns corresponding to none or one manganese nearest neighbour
[8]. The magnetic flux densities are 12.7 and 34.4 T respectively.
The gold has no intrinsic moment, but the core and 6s-conduction
electrons are spin-polarized by the neighbouring manganese mo-
ments. The magnetic flux density at the isolated gold atom is
dependent on manganese concentration, and rises from 12.7 T
at 5 at.% Mn to 16.5 Tat 10 at.% Mn. The most likely origin of
this field would therefore seem to be a long-range spin-polarization
HS
c:
HO
o
'Vi
'"
'£1
VI
c:
d
~
3-48
3-44 ErFez
I' I
I
-4 -2 0 2 4
Velocity I (mm S-I)
Fig. 8.3 The 57Fe Mossbauer spectra at 77 K of HoFe2 and ErFe2 showing
that only one iron site exists in the former, but two in the latter because of a
different direction of the magnetic axis. ([6], Figs. 1,3).
of the gold 6s-electrons in the alloy. This is additional to the

short-range polarization caused by an immediately neighbouring
manganese 3d-moment which gives the extra flux density of about
20T.
On the other hand, in the intermetallic compounds AU4Mn,
AU3Mn, AUsMn2 and AU2Mn, the observed magnetic flux density
is in direct proportion to the number and magnitude of the neigh-
bouring Mn 3d-moments, and clearly in this case the long-range
polarization of the conduction electrons is less significant and is
dominated by the short-range polarization.
AU2Mn has a layered body-centred tetragonal lattice and orders
antiferromagnetically below 363 K. Each gold site has axial sym-
metry along the c axis and lies between a plane of gold and a
plane of manganese atoms (see Fig. 8.4). The 197Au resonance at
4.2 K shows- a magnetic hyperfine splitting of the ~ ~ ~ transition
with a large magnetic flux density of 157 T. The spectrum in
Fig. 8.5 shows eight lines because of a substantial quadrupole (E2)
admixture in the radiation which allows transitions with flmz = ±2
to take place [9]. The spectrum is asymmetric because of a
quadrupole interaction. The solid lines represent the calculated
spectrum assuming Vzz (which must be along c from the crystal
symmetry) to be perpendicular to the magnetic axis. The trans-
ferred hyperfine field at the gold will be parallel or antiparallel
to the 3d-moment on the manganese which therefore must also
be normal to the c axjs.
AuMn2 has the same lattice but with the atoms reversed. Each
gold layer lies between two manganese layers, and curiously the
structure is not magnetically ordered at 4.2 K. Both AU2Mn and
AuMn2 show large quadrupole effects, the values of e2qQg being
2.9 and 4.64 mm s-1 respectively [8]. From Fig. 8.4 it can be
seen that the Au environment will be considerably distorted in
both instances. The near-neighbours comprise a flattened cube
with 4 Mn and 4 Au on opposite faces in AU2Mn, and 8 Mn atoms
in AuMn2. The larger value of e 2qQ in the second case shows that
it is the manganese and not the gold neighbours which are largely
responsible for the presence of the electric field gradient.
In the whole series of Au/Mn alloys and compounds there is
an almost linear increase in the 197Au chemical isomer shift as the
gold content decreases. The total change of 6.6 mm s-1 is much
more than can be explained by the compression of the gold elec-
trons by the decrease in cell volume. The most likely explanation
[8] is a transfer of electrons from the transition metal to the 6s-
band of gold amounting to about 0.7 e- per atom.
The effects of compositional variation are particularly marked
in the alloys based on the ordered body-centred cubic AuMn struc-
A~Mn ~H~
c/o ·2·55 c/o - 2·56
t1 phose of AuMn t2 phose of AuMn

c/o == 0·97 c/o = 1·03
Fig.8.4 The structures of AU2Mn, AuMn2, and the t1 and t2 phases of AuMn.
The manganese spins are aligned parallel within any shaded plane, and anti-
parallel in adjacent planes (ignoring any spiral effects) to give an antiferro-
magnetic structure. AuMn2 is non-magnetic.
ture (40-53 at.% gold). Typical 197Au spectra at 4.2 K are shown
in Fig. 8.6. Upon cooling, the cubic high-temperature phase trans-
forms initially to a tetragonally distorted t1 phase (cIa < 1), but
for less than 50 at.% Au there is a further phase transition to a
second tetragonal phase, t2 (cIa> 1), without any change in cell
volume. Both tetragonal phases are antiferromagnetically ordered
with alternating ferromagnetic sheets of Mn atoms perpendicular
to a short axis (see Fig. 8.4). In the ordered 50 at.% alloy the
1·01, ~
......
~
..,
-
I::l
;::
I::l..
~
~
0·991- tV \ \ I IA-' ,- Il, 1\ I V II' -i ~

~
~
~, :::::
;;; ("')
-.
c:
.~
'" 0·98 g
'"
-}; ~
'1;:j
0
0; a
:::::
'" ;::
0·97 I , ~
AU2Mn
0·96
I
0·95' , , " '"
- 12 -8 -4 0 ~. - .-
Velocity/(mm 5· ')
Fig.8.5 The 77.34-keV 197 Au resonance at 4.2 Kin AU2Mn. The eight component lines are
from an Ig = -} ~ Ie = transition with E2/M 1 mixing. The magnetic hyperfine splitting appears
1.
asymmetnc because ot a small quadrupole interaction. ([9), Fig.2).
.-
\0
W
• ~ ~..... ' . . . . 1. " , ' , ' , ............. .
•. ,.- ." .- ~ - -.\. , ~'''''. ~t·

'.'
_ . __ .....;.-
'"
i
- -
-" ,-' ..• ...• '.___ , _, ~5'/' Au
~'
_ " . ' , '-,"' , , ' / _ 1.•• ,.: . . .... _, ,-,'.- • •..-1 ..... .. I -~"'I."'- 1"',- ". ... ...,.. SO·/oAu
., ". o9.d
.. J......:..
. .
,..
. ...... f .....;--' ...·.-'.. •~ . -•.•. ' ... 52 % AU
:
\ ... .. ' .-.;, . ..... . . -',- , . '.-
.r .'
" _:,' '... _',- 53'/. Au
, :
'r '.
"
- 10 -5 0 10 15
V.loc;ty/(mIlU-')
Fig,8.6 The 77.34-keV 197Au spectra at 4.2 K of Mn/Au alloys close to

the composition AuMn, Note the line broadening upon departure from
stoichiometry in either direction. ([8] , Fig. 3).
gold environment comprises eight Mn neighbours of which four

have their spins opposed to the other four (being in different
magnetic planes). The spin-polarization effects at the Au atom
are therefore self-cancelling, and no field is expected or observed
in the 197Au resonance below the ordering temperature. For a gold
content of <50 at.%, the environment of ea,ch gold atom remains
at 8 Mn, but the placement of Mn atoms in gold sites will disturb
the magnetic arrangement, and this is reflected in the magnetic
broadening of the Mossbauer spectrum. For a gold content of
> 50 at.%, some of the gold atoms have only seven nearest neigh-
bour Mn atoms, and the resultant spin-imbalance will also result
in a magnetic flux density at the gold nucleus. Thus a departure
from stoichiometry in either direction results in some magnetic
polarization of the gold ru-electrons and a broadening of the 197Au
resonance.
A further example of order-disorder effects is given by the Fe-AI
system which maintains a body-centred cubic lattice from 0 to 54
at.% Al but tends to produce ordered structures near compositions
corresponding to Fe3AI and FeAl. The degree of order-disorder is
variable according to the thermal and mechanical treatment of the
alloy [10]. The 57Fe Mossbauer resonances of the ordered and
disordered forms differ because of the changes in the near-
neighbour environment of the iron. For example an alloy corres-
ponding to FeAI and ordered by thermal annealing shows a narrow
single-line absorption corresponding to the unique non-magnetic
iron site in the ordered CsCI structure (illustrated in Fig. 8.7).
The crushed alloy shows significan t line broadening and in parti-
cular a gross broadening in the wings of the resonance characteri-
100
.
99 (a) ' \ r'F~:1
98 annealed
97
96
9S I
(;1
100
98
(b)
96 cr~5hed
94
co 92
'iii 90 , .. .
.~ 100
c
.-. .,... .
"
~ 98 (c)
Fe,., Al 0.Q
96 annealed
9-4
92
90
88
86
84 -8 -6 -4 -2 0 2 4 6 8
Velocity f(mm 5-')
Fig. 8.7 The 57Fe spectra at 4.2 K of (a) ordered FeA1 (b) crushed FeA1
(c) Fel.1A10.9. ([10], Fig. 2).
stic of an inhomogeneous magnetic hyperfine interaction. Plastic
deformation of the body-centred cubic lattice will generate slip
planes, giving antiphase boundaries or stacking faults containing
an excess of like-atom pairs. This increase in the number of con-
tiguous iron atoms is responsible for the magnetic behaviour.
An indirect verification of this interpretation can be obtained
by examining a non-stoichiometric ordered alloy of the type
Fel+xAll-x. Each iron atom on an Al site has eight Fe nearest
neighbours, while each Fe site has dominantly between 6 and 8 Al
nearest neighbours. The spectrum for an alloy with x = 0.1 is
shown in Fig. 8.7. The magnetic area in the wings of the spectrum
(about 9% of the total area) corresponds closely to the fraction
of iron on AI sites which are thus centred in an Fe9 cluster. The
eight neighbouring iron atoms on Fe sites are less strongly coupled,
but are nevertheless affected as seen by the broadening of the cen-
tral component. The ferromagnetic behaviour of nearly stoichio-
metric FeAl is therefore determined by the presence of magnetic
clusters, the number of which may be varied by appropriate treat-
ment of the alloy.
References
[1] Cordey Hayes, M. and Harris, 1. R. (1967) Phys. Letters, 24A, 80.
[2] Vincze, I. and Campbell, I. A. (1973) J. Phys. F, 3, 647.
[3] Cranshaw, T. E. (1972)J. Phys. F, 2, 615.
[4] Stearns,M. B.(1963) Phys. Rev., 129,1136.
[5] Tansil, J. E., Obenshain, F. E. and Czjzek, G. (1972)Phys. Rev. B, 6,
2796.
[6] Bowden, G. J., Bunbury, D. st. P., Guimaraes, A. P. and Snyder, R. E.
(1968)J. Phys. C, 1,1376.
[7] Bowden, G. J. (1973) J. Phys. F, 3,2206.
[8] Longworth, G. and Window, B. (1971) J. Phys. F, 1,217.
[9] Thompson,1. 0., Huray, P. G., Patterson, D. O. and Roberts, L. D.
(1968) In Hyperfine Structure and Nuclear Radiations, ed. E. Matthias
and D. A. Shirley, p. 557, North Holland, Amsterdam.
[10] Wertheim, G. K. and Wernick, J. H. (1967) Acta Met., 15,297.
CHAPTER NINE
Analytical Applications
The preceding chapters have been chiefly concerned with the

Mossbauer spectra of single-phase materials containing a small num-
ber of distinct atomic sites for the resonant atom. The more com-
plicated behaviour seen in non-stoichiometric phases has been
found to relate to the statistical variations in the near-neighbour
environment. However, there are other solid-state systems which
are either heterogeneous multi-phase materials, or have a com-
plicated structure in which the atomic environment is variable
and may even be unknown. Typical examples in this category are
supported catalysts and ion-exchange resins. Nevertheless,
Mossbauer spectroscopy can be invaluable as a means of investi-
gating such systems. The sensitivity of the resonance to changes
in oxidation state and site symmetry, and the essentially non-
destructive nature of the technique can be used to good effect.
This chapter describes a miscellaneous selection of analytical appli-
cations grouped under three general headings, but in no way
attempts to be comprehensive.
9.1 Chemical analysis

It would be misleading to imply that Mossbauer spectroscopy is
ideally suited to chemical analysis in general. Although in prin-
ciple the intensity of an absorption can be related exactly to the
amount of the resonant isotope in the absorber, the problems
involved in correcting for the effects of the radiation background,
197
and the need to know the recoilless fraction accurately, tend to
make the experimental result subject to errors larger than those
obtained by more conventional analysis. However, there are
many examples where the elemental analysis cannot fully charac-
terize a compound because two or more elements have alternative
oxidation states.
An interesting example is provided by a blue iron-titanium
anhydrous double sulphate which approximates in composition
to the formula FeTi(S04h with iron in the Fe 2+ oxidation state
[1]. Repeated attempts to prepare a stoichiometric compound
have failed; in every case the product appears to be a single phase
material, but the chemical analyses always show an excess of iron
and a significant but variable proportion of Fe 3+. The Mossbauer
spectra at 295 K are comparatively simple, with a quadrupole
doublet from Fe 2+ (A = 0.98, D = 1.26 mm s-l with respect to Fe
metal) and a broad singlet from Fe3+ (D = 0.64 mm s-l). However,
in every preparation the proportion of iron in the Fe 3+ oxidation
state, as estimated approximately from the relative absorption
areas, is found to be appreciably higher than indicated by chemical
analysis (the figures for the sample closest to stoichiometry being
28% by Mossbauer spectroscopy as opposed to 5.5% by analysis).
The formula of the hypothetical ideal compound is
(Fe 2+)(Ti4+)(SOl-)3' and indeed the ratio of Fe + Ti to sulphate is
always found to be very close to 2 :3. In order to incorporate the
excess of iron,a formulation of (FetxFe~1(Fe~+Titx)(SOl-)3 must
be adopted, which is then consistent with the analyses, but not with
the Mossbauer data. The solution to the problem is to propose the
existence of a proportion of Ti3+ in the solid phase. The new formu-
lation is (Fetx_yFe~!y)(Fe~irTitx_y)(SOl-)3' which in the limit
of apparent stoichiometry becomes (FetyFer)(Ti~~ity)(SOl-h.
The unstable Ti 3+ ion will reduce Fe 3+ to Fe 2+ on dissolution, so
that the solution analysis gives a low value for the Fe 3+ content of
the solid. Unfortunately, the crystal structure of this interesting
compound has yet to be determined.
A second example is provided by the solid-solution SrFexRul_x~_y
derived from SrFe03 and SrRu03, both of which have the perov-
skite lattice with M4+ cations but are antiferromagnetic and ferro-
magnetic respectively [2]. The perovskite lattice is often oxygen
deficient, and in characterizing the solid-solution phase it is im-
portant to know the oxygen content. Unfortunately, for this
Analytical Applications 199
combination of elements it is difficult to obtain analyses of suf-
ficient accuracy to determine the value of y. However, the oxygen
deficiency can be determined indirectly from the 99Ru and 57Fe
Mossbauer spectra which give the oxidation states of the cations.
Thus substitution of iron in SrRu03 takes place entirely as Fe 3+
and oxygen deficiency increases, so that compositions such as
3+ RuO.9
Sr F eO.1 4+ O2.95d
anSr F eO.2
3+ RuO.802.90
4+ .
are obtamed. From
about x = 0.3 it appears that the increase in oxygen deficiency is
halted by the introduction of some Ru 5+, and at x = 0.4 the com-
position is approximately SrFet4Rut39Ru5~2102.91. At x = 0.5
the oxygen deficiency appears to be minimal with the unexpected
composition of SrFet5Rug~503. Such detailed information could
not be obtained by conventional analysis.
One rarely reported but very useful application of the Mossbauer
spectrum is as a 'fingerprint' to check the efficacy of a chemical
preparation. This is particularly true for co-ordination and organo-
metallic compounds of iron and tin, where unwan ted products are
often easily detected by the presence of additional components
in the Mossbauer spectrum. A preliminary examination in this way
can be carried out quickly, and is completely non-destructive of
the material.
This non-destructive aspect of the technique can be particularly
useful. Furthermore, almost any size or shape of object can be
examined if the experiment is carried out in a scattering mode.
One unusual application is in the classification of works of art.
Many of the yellow and brown pigments used in oil paintings such
as ochres, siennas and umbers contain a substantial proportion of
iron, mainly in the form of the magnetic oxides a-Fe203 and a-
FeOOH. However, the naturally occuring deposits from which
they were originally derived are variable in quality and average
particle size. A systematic survey [3] of iron pigments has shown
that the room-temperature Mossbauer spectrum is often characteri-
stic, with a degree of superparamagnetic collapse of the hyperfine
pattern which relates directly to the particle size. Pigments have
been successfully identified in 18th-century oil paintings, and a
systematic study of the work of a particular artist should prove
valuable in the corroboration of doubtfully attributed paintings.
The scattering method is also applicable to analysis of terracotta
statuary whose colour derives mainly from the presence of iron
[3]. Typical spectra from two terracotta works are shown in
Fig. 9.1. The appearance of magnetic components in the sample
TC-19 has been established to be the result of firing the clay to a
higher temperature than was the case for TC-3. Individual crafts-
men undoubtedly varied in their methods of furnace handling,
and this distinction is also useful in the attribution of works to a
particular artist. Fake statues made from different material can
be readily distinguished from the genuine articles without damage
to either.
The application of scattering techniques in the study of surface
effects is considered in a separate section later in the chapter.
0·25
0·20
0·1 5
TC-3
0·10
0·05
~' "t '",,~. ~l~ ~,'" .'.e~ ~ . &
.~
'"
>
~
o
"iii
a:
-0·05'--~-----l---...L----L _ _ _l....-...J
-10 -5 0 5 10
Velocity / (mm 5- 1)
Fig. 9.1 The 57Fe Mossbauer spectrum in a scattering geometry from two
samples of terracotta statuary. The magnetic components in the spectrum
from TC-19 establish that it was fired to a higher temperature than TC-3 .
([3], Figs. 20,21).
9.2 Silica te minerals
The many naturally occurring silicate minerals are all structurally
derived from the tetrahedral bonding of silicon to oxygen. The
discrete orthosilicate anion, SiO:-, is found in only a small pro-
portion of them. In the vast majority the Si04 tetrahedra are
joined by oxygen sharing into infinite chains or sheets. Examples
are the pyroxenes with single-strand chains of composition (SiO~-)n'
the amphiboles with double-strand cross-linked chains of composi-
tion (Si 40Yl)n and the micas with infinite sheets of composition
(Si20g-)no The silicon-oxygen anionic groupings are held together
by metal cations between the chains or sheets in sites with usually
four or six co-ordination to oxygen. The strength and rigidity of
the silicon-oxygen lattice allow an unusually large variation in the
nature of the cations which serve primarily to produce e1ectro-
neutrality. Partial replacement of silicon by aluminium results
in the closely related feldspars, zeolites and ultramarines. The
silicates therefore show a greater degree of compositional variation
than is found for example in simple oxides.
Many of the silicates contain a substantial proportion of iron as
Fe 2+ and Fe 3+ cations, and therefore give a good 57Fe Mbssbauer
absorption. The number of distinct cation sites is usually limited
to between one and four. The distribution of the various cations
within these sites is not completely random. A substantial degree
of ordering is often found, determined largely by the differing sizes
of the lattice sites and of the cations which occupy them. To
determine quantitative site populations is not an easy matter,
and Mbssbauer spectroscopy represents one of the most useful
methods for this purpose.
The effects on the Mbssbauer spectrum of the immediate site
co-ordination in an iron oxide are described in Chapter 7, and
these principles still hold in the silicates. The main difference lies
in that the somewhat random occupation of the near-neighbour
cation sites by widely different cations tends to produce a notice-
able broadening of the resonance lines. Magnetic effects are almost
unknown, and any deductions must be made from the chemical
isomer shift, quadrupole splitting, and intensity of absorption.
The large difference in chemical isomer shift between Fe 2+ and
Fe 3+ enables the approximate proportions of the two oxidation
states to be determined quite easily from the relative areas of the
absorption lines. The recoilless fractions at different sites are
usually similar, and the error introduced by assuming them to be
equal is small. The quadrupole splitting of an Fe 3+ component is
generally small, and it is difficult to distinguish multiple site sym-
metries for this cation. However, the comparatively large splitting
found for an Fe 2+ site and its sensitivity to environment usually
results in a resolved fine structure, leading directly to an estimate
of the relative site occupancy by the Fe 2+ ions. Both the chemical
isomer shift and quadrupole splitting of an Fe 2+ ion decrease as
the co-ordination decreases from eight to six to four. As a result
it is possible to make deductions concerning site symmetry and
occupation in silicates of unknown structure. Many mineral
samples are strongly textured (particularly the amphiboles and
micas) and cannot be made into an absorber without some residual
orientation. As a result the two components of a quadrupole
doublet may have an unequal intensity.
These general remarks can be illustrated by the spectra in Fig.
9.2. The mineral howieite is believed to have a chain structure and
(0] (b]
2 2
~3 ~3
c:: c:
0 0
A A'
'e. 4
0
~4
L..
0
VI VI
J:J J:J
'"" 5 '""5
6
6
7
7
8
-2 -1 0 2 3 -1 0 2 3
Velocity / (mm 5- 1]
Fig. 9.2 The room temperature 57Fe Mossbauer spectra of (a) howieite and
(b) deerite. ([4], Fig. 1).
has the approximate composition NaFe~+Mn3Fe~+Si12031(OH)13.
The room-temperature spectrum [4] shows two quadrupole doub-
lets, that with the larger splitting (AA') having the high chemical
isomer shift and large quadrupole splitting (see Table 9.1) typical
of high-spin Fe 2+ ions in octahedral co-ordination. The inner
doublet (BB') is typical of high-spin Fe 3+ ions, also in octahedral
co-ordination. The mineral deerite has the approximate composi-
tion FerjFefSi13044(OH)l1. The spectrum shows three quadrupole
doublets. AA' arises from Fe 2+ ions in distorted six co-ordination,
BB' from Fe 3+ in distorted six co-ordination, while the CC' lines
with a significantly lower shift than AA' but still within the range
for high-spin Fe 2+ are clearly from a four co-ordinated tetrahedral
site. Note that the two components of each doublet are not neces-
sarily equal in intensity because of texture in the absorber.
A better known class of silicates is the pyroxenes which contain
single-stranded chains of Si04 groups of overall composition
(SiO~-)n" The general formula can be written as ABSi206 where
A and B represent distinct cation sites known as M1 and M2
respectively. The M1 sites have a near-regular six co-ordination to
oxygen and are usually occupied by small cations such as Mg2+,
Al3+, Mn2+, Fe 3+ or Na+. The M2 site is grossly distorted, and being
larger tends to be occupied by Na+, Mg2+, Ca 2+, Mn 2+ and Fe 2+. All
pyroxenes do not have a common space group; for example
pigeonite has the space group P2dc, CaMgSi206 and CaFeSi206
belong to C2/c, and Mg2Si206 to Pbca, but the small differences
involved have only a minor influence on the Mossbauer spectra.
In all the pyroxenes the Fe 2+ cations at the more distorted M2
sites show a smaller quadrupole splitting than at the M1 sites. This
Table 9.l Mossbauer parameters for howieite and deerite at room tem-
perature
Compound Site o/(mm s-l)* A!(mm s-l)
Howieite A l.l8 2.81

B 0.40 0.59
Deerite A l.l3 2.57
B 0.45 0.59
C 1.02 1.40
* relative to Fe metal.
is probably because of a large 'lattice' contribution to the electric
field gradient tensor which is opposite in sign to the 'valence' term
from the 3d6 configuration (see p. 103). The 57Fe spectra comprise
two superimposed doublets which are partially resolved (ignoring
for the present a possible third contribution if Fe 3+ ions are also
present; some examples of spectra from orthopyroxenes are shown
in Fig. 9.4).
It is convenient to limit further discussion to the orthopyroxenes,
which have a composition close to the system MgSi03-FeSi03.
The Fe 2+ cations prefer to occupy the M2 sites, but considerable
disorder can be induced by heating to 1270 K or above. Anneal-
ing at lower temperatures causes a preferential occupation of the
M2 sites by Fe 2+ which increases with decreasing temperature
until at about 750 K the ionic diffusion is apparently frozen [5,6].
If the relative occupations of the MI and M2 sites by Fe 2+ are Xl
and X2 respectively, then a parameter k can be defined by
For a system at equilibrium. k corresponds to a true equilibrium

constant for the reaction
and can be related to the free-energy difference AGE by
AGE = -RTln k
If A GE is independent of temperature, k should decrease with

decreasing temperature (i.e. Xl decreases, X2 decreases).
The values of Xl and X2 for a wide range of natural orthopyro-
xenes are plotted in Fig. 9.3. The open circles correspond to
natural samples re-heated at 1270 K until equilibrium had been
reached and then rapidly quenched to preserve the disorder. The
solid curve represents the theoretical prediction for k = 0.235
(AGE = 15.30 kJ mol-I), and shows that the relationship holds
for Xl < 0.6. The filled circles represent metamorphic and pluto-
nic rocks which have cooled slowly and have therefore reached a
1·0~-'--r---r-r---r--'---'---.---,---"
o·g
0·8 • •
.".
•
+
lit
~
•
+
0.7·. '9.
0·6
004
0·3 •
0'2
0·1
o 0·1 0·2 0·3 0'4 0'5 0·6 0·7 0·8 O·g 1·0
x1
Fig. 9.3 A plot of the measured site occupancies x I and x2 of the M 1 and
M2 sites in a selection of natural orthopyroxenes. The open circles represent
natural samples reheated to 1270 K and quenched. The filled circles repre-
sent natural slowly cooled metamorphic and plutonic rocks. The crosses re-
present quickly cooled volcanic rocks. ([ 6] , Fig. 1).
final equilibrium at about 750 K with k = 0.028. Note the pre-

ferential occupation of M2 sites by Fe 2+. The crosses represent
volcanic rocks which have cooled more quickly and have therefore
retained a proportion of non-equilibrium disorder.
The pyroxenes are now known to comprise a major proportion
of the rocks and soil on the lunar surface (together with other sili-
cates such as olivine and plagioclase, opaque minerals such as
ilmenite, troilite, metallic iron/nickel alloy and spinel oxides, and
vitrified material). Being non-destructive, Mossbauer spectroscopy
has proved to be a useful method of studying the material returned
to earth by the American 'Apollo' and Russian 'Luna' spacecraft.
In some instances it has proved possible to obtain orthopyroxene
separates from lunar rocks. Such a sample separated from an igne-
ous rock returned by the 'Apollo 14' mission (sample 14310,
l16-Pl) has a composition of Mg1.40Fe0.4sCaO.12Si2 06 [7]. The
Ml and M2 site occupancies have been determined to be Xl = 0.076
and X2 = 0.398, leading to a value for k of 0.097. As may be seen
by reference to Fig. 9.3, the degree of cation disorder is higher
than in slowly cooled terrestrial orthopyroxenes. The cooling rate
of this rock after solidification from the melt was certainly faster
than expected for a deep-lying rock, but probably slower than one
would find for the surface of a lava flow. The rock is believed to
have crystallized from a molten pocket in the ejecta blanket at
Fra Mauro following the catastrophic event which created Mare
Imbrium, but was then excavated in a subsequent event without
substantial reheating after its initial cooling to below 750 K.
In this connection it is important to note that an ordered pyro-
xene can be disordered by an intense shock [8]. A natural ortho-
pyroxene of composition Mg1.72Feo.28Si206 (highly ordered with
k = 0.041) was subjected to a shock of between 900 and 1000
kbar. The Mossbauer spectrum (Fig. 9.4) was unchanged apart
from an increase in the relative intensity of the outer doublet to
give a value of k = 0.27. One of the effects of the shock is to
cause a short-term heating to as high as 1650 K. The result is a
rapid diffusion of cations which are then quenched into a dis-
ordered state. It therefore follows that intensively shocked meteo-
rite or lunar basalt will have lost any thermal record held prior
to the catastrophic event. It is also clear that lunar rock 14310,116
cannot have been intensively shocked during excavation to the
surface by a meteorite impact.
9.3 Surface chemistry

There are several important aspects of chemistry which are inti-
mately concerned with the properties of surfaces, including such
topics as corrosion, adsorption, and catalysis at solid-liquid and
solid-gas interfaces. In propitious cases it is still possible to make
Mossbauer measurements in conventional transmission geometry,
but for surface studies in general it is often more appropriate to
use a scattering experiment. The detected radiation can then be
either the resonantly scattered r-ray, or alternatively the X-rays
or electrons produced by internal conversion. The final choice is
dictated partly by the properties of the r-decay, and partly by the
nature of the surface.
For example, the detection of resonant scattering from 57Fe in
steel plates is best monitored by using the 6.3-keV X-rays rather
than the primary r-ray, and provides a means of studying corro-
o~~~~~~~~~~
0·02
(a)
0·04
0·06
0·08
0·10
0·12
Fez+(M z)
0·14 Fe2+(M 111
....- - - - - - - '
c:
o
'';:;
..
a.
L.
Sl
.0
« 0·02 (bl
0·04
0·06
0·08
0·10
0·12
0·14
Velocity / (mm 5- 1I
Fig. 9.4 The 57Fe Mossbauer spectrum at 77 K of an orthopyroxene of

composition Mgl.72FeO.2SSi206 (a) in natural form and (b) after intensive
shock. ([ 8 J , Fig. 1).
sion [9]. The spectrum for scattering from a steel plate exposed
to fumes of HCl in air is shown in Fig. 9.Sa. The dominant doub-
let in the spectrum can be identified from the chemical isomer
shift and quadrupole splitting as being that of j3-FeOOH and is
distinct from the spectra of the other oxides and oxyhydroxides
of iron. The intensity of the scattering can be related to the depth
of the surface layer by evaluation of the appropriate scattering
integral, and in this case gives an approximate thickness of a bout
2 x 10- 3 cm. The additional weak components on the baseline are
the magnetic hyperfine pattern of the underlying steel. Fig. 9.Sb
(a)
(e)
+12 +8 +4 0 -4 -8 -12
Veloeity/(mm S-I)
Fig. 9.S The 57Fe Mossbauer spectrum at room temperature obtained by

scattering from a corroded steel plate: (a) the corroded plate showing the
strong central absorption from J3-FeOOH; (b) as in (a) but with the rust
scraped off; (c) as in (b) but with the surface milled clean. ([ 9] ,Fig. 2).
shows the spectrum recorded after attempting to scrape off the

rust, and reveals that there is still a substantial rust residue. Spec-
trum (c) was obtained after milling 1 mm off the steel to give a
fully cleaned surface. Those resonant nuclei close to the surface
give a larger contribution to the observed spectrum than deeper-
lying nuclei because the non-resonant attenuation of the X-rays
increases with depth. The effective 'surface' is of the order of
10- 3 cm.
In some ways a more valua ble technique for surface studies is
to record the back-scattering of the conversion electrons. The
attenuation of electrons by a solid is much greater than for the
'Y-rays, so that it is only possible to record events which take place
at a depth of less than a bout 400 A (4 x 10- 6 cm) from the imme-
diate surface. A convincing demonstration of this method has
been given in a study of the thermal oxidation of iron [10]. To
improve the quality of the spectra the natural iron foils were elec-
troplated with 57Fe and annealed in hydrogen to form an isotopi-
cally enriched surface layer. Oxidation at 225° C for varying times
showed the gradual formation of a surface layer of non-stoichio-
metric Fe304, and this is illustrated in Fig. 9.6.
The rate of growth of the surface layer (as measured by the
change in the thickness W as a function of time t) will follow a
different rate law according to the mechanism of the corrosion
[11]. For example a parabolic growth law of the form W2 = At
+ B is usually found when the movement of ions takes place through
the lattice of the surface layer by diffusional migration (i.e. the
cations move outwards or the anions inwards) and the resultant
surface film is fairly uniform in thickness. Penetration of oxygen
into the interior in a 'leakage' manner rather than by uniform dif-
fusion results in a logarithmic law of the form W = A In (Bt + C)
and a less uniform layer. The intensity of the Mossbauer resonance
lines from the growing surface layer in this experiment follow a
logarithmic time dependence, consistent with the 'leakage' mech-
anism being dominant at 225°C. Spectrum (d) in Fig. 9.6 was
obtained after oxidation for 120 minutes, and corresponds to a
thickness of about 100 A (10-6 cm).
Oxidation at 350°C for five minutes gave a two-component sur-
face layer in which additional components due to o:-Fe203 and Fe304
could be distinguished. The minimum thickness of an oxide coat which
can be detected by this method is estimated to be about 40 A.
A more sophisticated development of this technique is to in-
corporate a focussing ~-spectrometer and to measure the exact
energies of the electrons emitted from the surface. The energy of
the electron is decreased with increasing depth of the emitting
atom below the surface as the result of scattering. It is therefore
possible to distinguish electrons originating in different sub-surface
layers. This has been convincingly demonstrated [12] using a
metallic tin plate exposed to bromine vapour. The Mossbauer
~--------------------- -,
,-,-__...' ''----'-''-'--""'---"__~" rf,o,
'---'---'__'--"'----', Fr
.. '. ' ,. (fl
:. . ~ : ~:~ ;\ j ~ ,.
- ~V\;~\/ '....J, .
(dl
.. .,
(el
,.,
:: :;;; . \ ;:
"
.! :~
; v\JV\. h./ .
.'
.. ..
" (bl
(QI
.'
:JJvJJ ~
-10'0 -5-0 0-0 5·0 10·0
VfIOCily/( mm ,-'I
Fig.9.6 The 57Fe Mossbauer spectra at room temperature (obtained by

counting the conversion electrons emitted after resonant absorption in a
scattering geometry) of an iron foil oxidized at 225°C for varying periods of
time: (a) before oxidation (b) 5 minutes (c) 15 minutes (d) 120 minutes
(e) 1000 minutes. ([101, Fig.4).
Analy tical Applications 211
spectrum corresponding to a considerable depth below the surface
is that of pure tHin. Layers nearer to the surface show an increas-
ing absorption due initially to SnBr2 and ultimately to SnBr4.
The study of adsorption at surfaces presents considerable prob-
lems, but is of immense importance with regard to catalysis and
ion-exchange. One type of catalyst which has been extensively
studied is made from a-Fe203 microcrystallites supported on high-
area oxides. For example silica gel can be impregnated with ferric
nitrate and then calcined in air to give an 'active' deposit on the
Si02 matrix. This preparation can easily be carried out with en-
riched 57Fe, and the Mossbauer spectrum recorded by transmis-
sion methods [13] .
There are two aspects to these investigations. Firstly, the pro-
perties of the catalyst are very dependent on the method of pre-
paration and its resultant influence on the surface area and the
particle size of the Fe203. The latter can be conveniently investi-
gated using the 57pe Mossbauer spectrum because the magnetic
hyperfine splitting of bulk antiferromagnetic a-Fe203 is partially
or completely collapsed by superparamagnetic relaxation in the
microcrystalline state (see p. 156). The temperature-dependence
of the spectrum can therefore be easily related to the mean particle
size.
Secondly, the adsorption of ions or radicals onto the surface
of the iron oxide often has a marked effect on the Mossbauer
spectrum, although it is not easy to determine why the changes
take place. Adsorption of ammonia [14] , water or H2S [ 15]
completely changes the Mossbauer absorption; for example addi-
tion of water reduces the Fe 3+ quadrupole splitting from 2.24 mm
s-1 to 1.56 mm s-1 at room temperature. All the iron atoms appear
to be affected, and the exact nature and mode of action of the
activated surface are not yet fully understood.
The investigation of zeolites by Mossbauer methods differs in
that the iron is not incorporated into the support but is adsorbed
onto the interior surface. The Mossbauer parameters of Fe 2+ or
Fe 3+ cations in the zeolite can in principle be related to the sym-
metry of the adsorption sites in the aluminosilicate framework
[ 16]. Once these sites have been characterized, it then becomes
possible to undertake selective replacement adsorption with other
species, and in this way to indirectly identify the adsorption sites
for atoms and molecules containing only non-Mossbauer elements.
Where the adsorption is slow, the examination of a series of samples
exposed to reagents for varying periods of time can be used for
kinetic studies.
References
[1] Gibb, T. C., Greenwood, N. N., Tetlow, A. and Twist, W. (1968) J.
Chern. Soc. (A), 2955.
[2] Gibb, T. C., Greatrex, R., Greenwood, N. N. and Snowdon, K. G.
(1974) J. Solid State Chern., 14,00.
[3] Keisch, B. (1973) Archaeornetry, 15, 79.
[4] Bancroft, G. M., Burns, R. G. and Stone, A. J. (1968) Geochirn.
Cosrnochirn. Acta, 32,547.
[5] Virgo, D. and Hafner, S. S. (1970) Arner. Mineral, 55,201.
[6] Hafner, S. S. and Virgo, D. (1969) Science, 165,285.
[7] Schiinnann, K. and Hafner, S. S. (1972) Proc. Third Lunar Sci. Conf.,
Geochirn. Cosrnochirn. Acta, Suppl. 3, vol. 1,493.
[8] Dundon, R. W. and Hafner, S. S. (1971) Science, 174,581.
[9] Terrell, J. H. and Spijkerman, J. J. (1968) Appl. Phys. Lett., 13, 11.
[10] Simmons, G. W., Kellennan, E. and Leidheiser, H. (1973) Corrosion,
29,227.
[11] Davies, D. E., Evans, U. R. and Agar, J. N. (1954) Proc. Roy. Soc.,
225A,443.
[12] Bonchev, Zw., Jordanov, A. and Minkova, A. (1969) Nuclear Instr.
Methods, 70,36.
[13] Hobson, M. C. and Gager, H. M. (1970) J. Catalysis, 16,254.
[14] Hobson, M. C. and Gager, H. M. (1970) J. Colloid & Interface Sci.,
34,357.
[15] Gager, H. M., Lefelhocz, J. F. and Hobson, M. C. (1973) Chern Phys.
Letters, 23,387.
[16] Delgass, W. N., Garten, R. L. and Boudart, M. (1969)J. Phys. Chern.,
73,2970.
CHAPTER TEN
Impurity and Decay After-effect Studies
One of the first applications of Mossbauer spectroscopy was the

study of the behaviour of an impurity atom in a metallic host
matrix. The vibrational characteristics of such an atom are of
fundamental importance to an understanding of lattice dynamics,
and may be measured using the temperature dependence of the
chemical isomer shift and recoilless fraction (see Chapter 6).
There are however a number of other ways in which impurity
effects can be used. In some instances the interest centres upon
the impurity atom itself, as for example in semiconductors where
the physical and chemical properties may be drastically altered
by the nature of the impurity. In other cases the impurity atom is
incorporated into the host with the intention of monitoring the
properties of the latter. The incorporation of isotopically enriched
impurities into metals, oxides and other ionic salts for use as
Mossbauer absorbers has been extensively used for both purposes.
There is however a lower limit to the concentration of an im-
purity which can be used to obtain a good resonant absorption,
and this is of the order of 0.1 at.%. If very dilute concentrations
of impurity atoms are required, it may be better to use the alterna-
tive procedure of incorporating the radioactive precursor into the
host matrix, and then comparing this source with a single-line refer-
ence absorber. [Note:- A resonant absorption occurring at a posi-
tive velocity with respect to a standard source represents an energy
greater than the standard. If the roles are now reversed, a reso-
nant emission occurring at negative velocity with respect to a
213
standard absorber also represents a higher energy. For this reason
a hyperfine emission spectrum is the mirror image of a correspond-
ing absorption spectrum. The sign of the chemical isomer shift
must therefore be reversed for comparison with conventional
absorption data.]
However, the radioactive decay of the parent nucleus may cause
drastic chemical modification of the environment of the daughter
nucleus before the M6ssbauer -y-emission takes place. If the host
matrix is metallic, then the high mobility of the electrons ensures
that any after-effects of the decay are obliterated within a time
much less than the M6ssbauer excited-state lifetime. Therefore
the doping of metals with for example 57Co or 119m Sn can be
regarded simply as a means of producing very low concentrations
of iron or tin impurity atoms. If on the other hand the host matrix
is an insulator, then after-effects may be observed. These will
be discussed separately. The after-effects of nuclear decay may be
studied advantageously by M6ssbauer spectroscopy because the
technique is specific to the site of the disturbance and gives infor-
mation concerning the atomic environment within 10- 7 s of the
primary event.
10.1 Impurity doping

Impurity doping can be used most effectively in simple lattice
compounds such as halides and oxides where the co-ordinate geo-
metry at the impurity site can be easily established. An excellent
example is provided by cadmium fluoride doped with 57Fe [1].
CdF 2 is isostructural with fluorite, CaF 2. Each Cd 2+ ion is at the
centre of a cube of F- anions, and each F- is tetrahedrally sur-
rounded by Cd2+ ions. The ionic radius of Cd2+ is 1.07 A (107
pm). However, the radius of high-spin Fe 2+ is only about 0.81 A
so that substitution of Fe 2+ on a Cd 2+ site will leave a significant
amount of 'rattle-room'. The doping of iron into CdF 2 can pro-
duce impurities in both the Fe 2+ and Fe3+ oxidation states, but
with appropriate thermal treatment it is possible to obtain samples
containing only Fe 2+ ions. The M6ssbauer spectra at 296 K and
5 K of CdF 2 doped with enriched 57Fe (CdO.99Feo.01F2) are shown
in Fig. 10.1. The single-line resonance at 296 K has a chemical
isomer shift of +1.444 mm s-1 (with respect to Fe metal at the
same temperature) which is typical of the Fe 2+ ion in a site with
Impurity and Decay After-effect Studies 215
. ..
0 · 91~ -
0·86 r- -
e
''E:
0
..
.e
0
• (a)
....
"-
0·81 t I I I
0 2
1·01 r---:------..,.-----..,.---,----- , . . . - - - - - - - - - ,
"
l
\ , 8
0·95 '.;. 0:
". (b)
0 .92L---_~1--~0----Ll-----!2~--~--4L---..L5-....J
Veloc i ty I (mm S-I)
Fig. 10.1 The Mossbauer spectra of CdO.99FeO.01F2 at (a) 296 K and (b) 5 K
The weak doublet ~B) seen at low temperatures corresponds to a near-
neighbour Fe 2+-Fe + cation pair, while the central resonance (A) corresponds
to isolated Fe 2+ ions. ([ 1] , Fig. 1).
little or no covalency. The lack of a quadrupole splitting shows

that the impurity ion is occupying the cubic Cd 2+ site. The line-
width of 0.26 mm s-l broadens slightly at lower temperatures (e.g.
0.34 mm s-l at 195 K) and becomes non-Lorentzian until at about
90 K a quadrupole-split doublet (B) comprising about 13% of
the total area emerges from the wings of the central peak. This
doublet has a chemical isomer shift which is smaller than that of
the central line by about 0.04 mm s-l, and a quadrupole splitting
which increases rapidly with decreasing temperature from Ll = 1.735
mm s-l at 89.9 K to 3.599 mm s-l at 2.08 K. The central com-
ponent also broadens slightly and shows a small splitting (A).
The quadrupole-split doublet (B) shows that a proportion of the
F e2+ ions experience a local symmetry lower than cu bie. If the
Fe2+ ions occupy Cd2+ sites by a completely random replacement,
then the probability for any Fe 2+ ion that the twelve nearest-
neighbour cation sites are all occupied by Cd 2+ (i.e. the environ-
ment is cubic) is 88.6%. The probability that there is one Fe 2+
nearest-neighbour is 10.0%, and the remaining 1.4% are distributed
in higher-order clusters. The Fe 2+ ions showing a large quadrupole
splitting can therefore be associated with the occurrence of Fe 2+-
Fe 2+ near-neighbour cation pairs in the lattice. The cubic sym-
metry of an isolated Fe 2+ site is thereby lowered to tetragonal C4v
symmetry.
The ground-state of the 3d 6 ion in this distorted site is a level of
eg symmetry with a t2gexcited state. The eg level will split when
the symmetry is lowered. The quadrupole-splitting behaviour is
therefore essentially equivalent to that for tetrahedral site sym-
metry described on p. 107. This is borne out by the fact that the
temperature dependence of the splitting follows a function of the
type Ll = LlO tan h(Eo /2kT), giving an eg level-splitting of Eo
= 8.5 X 10- 3 eV (Eo = 0.82 kl mol-I; Eo/k = 98 K).
The 3d6 Fe 2+ ion is a lahn-Teller ion, but by analogy with the
behaviour found in for example FeV204 and FeCr204 (see p. 163)
it is clear that in this instance there is still no static lahn-Teller
distortion of the cubic Fe 2+ site at 5 K. The very small splitting
of the central component can be explained satisfactorily as being
due to random strains from distant defects distributed throughout
the lattice.
The use of impurity doping to study the host matrix can be
illustrated by data from single crystals of MnF 2, grown from a
melt of MnF2 containing isotopically enriched FeF 2 [2]. MnF2
has the rutile (Ti02) structure and is antiferromagnetic below 67.3
K. Substitution of Fe 2+ into the lattice at Mn 2+ sites causes a
slight increase in the Neel temperature (by about 1 K), but other-
wise has no catastrophic effects on the magnetic properties. At
4.2 K the 57Fe spectrum shows magnetic hyperfine splitting with
a flux density of 22.8 T. Because the absorber is a single crystal,
the line intensities are angular dependent. As a result it is easy
to show that the 57Fe atomic moments which produce the flux
density B, and hence the Mn moments of the host with which they
are aligned, are parallel to the crystal c axis. The magnetic axis
in MnF2 is therefore in the same direction as in isostructural FeF 2,
The observed Mossbauer spectrum is complicated by a combina-
tion of magnetic and quadrupole interactions. Computation from
the line positions reveals that the electric field gradient tensor has
a principle value Vzz which is perpendicular to B and to the c axis.
The coupling constant e 2qQ/2 = 2.80 mm s-1 and the asymmetry
parameter 7'/ = 0.5. The tensor at the Fe 2+ site in MnF2 is therefore
very similar to that in FeF 2,
Apart from the previously mentioned application to the study of
lattice dynamics, impurity doping in metals can also serve to
examine electronic and magnetic structure. An interesting example
is given by the systematic variations in the chemical isomer shift
shown by 99Rh, 1930s and 197Pt impurities in the 3d-, 4d- and 5d-
transition metals [3]. These isotopes are the respective source
isotopes for the 90-keV 99Ru, 73-keV 193Ir, and 77-keV 197Au
Mossbauer transitions. The observed shifts relative to metallic
Ru, Ir or Au as appropriate are shown in Fig. 10.2. The following
general trends may be observed:
(1) The value of 11/Is'(0) 12 decreases nearly monotonically with an
increase in the d n configuration of the host metal. [Note:- fjR/R
is positive for all three isotopes]
(2) For hosts in any given column of the periodic table, 11/IlO)12
decreases from the 3d- to the 5d-host.
Some of the deviations are probably accounted for by structural
differences; e.g. Fe metal has a body-centred cubic structure
while Ru and Os have a hexagonal close-packed lattice. Surprising-
ly, these influences are largely subordinated to the more general
periodic relationship.
A more specific investigation concerns the magnetic properties
of the alloy Ni3Al which shows a change from being strongly para-
magnetic to weakly ferromagnetic as the nickel content increases.
Furthermore, the presence of a minute proportion of Fe atoms
can induce a giant magnetic moment by polarization of the host
matrix. This can be as high as 39 fJB per iron atom. Doping with
57Co is a useful technique in that the impurity can be kept low
V Cr Mn
-0·5
2
"j'
'"
E
E
~
:.c
'" 0
L..
E
0
'"
.!!!
a 8
.~
E
~
(J
'"
Fe Co Au
~
RU Rh N, Cu
4 w Os
Ir Pd
Ag
PI
o Au
3 5 7 9 11
Number of outer electrons
Fig. 10.2 The chemical isomer shifts of 99Ru, 193Ir and 197 Au impurity
atoms in the 3d-, 4d- and Sd-metals. The values were obtained from the emis-
sion spectra of 99Rh, 1930s and 197Pt, and are quoted with respect to Ru, Ir
and Au metal respectively but in the sense of a conventional absorption ex-
periment. 8R/R is positive in all cases, so that a more positive shift cor-
responds to a larger s-electron density at the nucleus. ([3 J, Fig. I).
enough to obtain the maximum magnetic effect [4]. The 57Co

emission spectra of phases with different compositions are shown
in Fig. 10.3. They were recorded at room temperature where they
are all paramagnetic.
The structure of Ni3A1 is of the AU3CU type with a face-
c
o
·iii
·e
VI
VI~_ _ _
c
.='"
..
\i I 5%
absorption
1 -1
Velocity/(mm S-l)
Fig. 10.3 The 57Co emission spectra at room temperature of doped alloys
near the composition Ni3Al. ([4], Fig. 1).
centred cubic lattice in which the AI atoms occupy the cube-

comers and Ni atoms the face-centred positions. An Fe atom at
the AI site should not show an electric field gradient because the
site symmetry is cubic, but this is not true of the Ni site which
has only axial symmetry. The dominant single-line resonance for
a nickel content in excess of 75 at.%, and thus a deficiency of
aluminium, may therefore be taken to indicate substitution at the
AI sites. In the nickel deficient samples the converse takes place
and iron is substituted at the Ni sites. The spectrum now shows a
quadrupole splitting, but the shape is complex and appears to
indicate the presence of a second weak doublet whose origin is
not conclusively established. The magnetic spectra at low tem-
peratures are particularly complex and will not be discussed here,
but many of the apparent inconsistencies in earlier data for these
alloys appear to stem from a tendency towards inhomogeneity in
the alloy composition.
A comparatively new technique in M6ssbauer spectroscopy is
to isolate small molecules in an inert matrix and thereby obviate
the effects of the strong intermolecular interactions which are
normally present in the solid phase. This is achieved by condens-
ing a vapour containing the atoms or molecules of interest (diluted
at least 50-fold with an inert gas such as argon, xenon or nitrogen)
under controlled conditions onto a surface at 4.2 K until a layer
with an adequate absorption cross-section can be obtained. Iso-
topic enrichment is generally desirable.
Some of the early experiments of this kind were directed to-
wards the isolation of single atoms of 57Fe in matrices of argon,
krypton and xenon [5]. Such an iron atom should have the free-
atom configuration of 3d64s2(5D4) and provide a valuable refer-
ence point in the absolute calibration of the chemical isomer shift
scale for 57Fe. The spectra obtained at 4 K for 57Fe in krypton
with concentrations of Kr/Fe = 136 and 44 are shown in Fig. 10.4.
The unsplit resonance line at -0.75 mm s-1 (with respect to Fe
metal at 300 K) corresponds to an isolated Fe atom on a Kr site
in the close-packed cubic lattice. The weaker quadrupole doublet
also recorded increases in intensity as the impurity concentration
increases and can be attributed to an Fe2 dimer. The splitting of
4.05 mm s-1 is close to that expected for a single unpaired 3d-
electron. The 5D atomic configuration of the free-atom will split
in the axial crystal field produced by the adjacent Fe atom. The
exact electronic ground state is not known, but the sixth electron
(apart from the spherical d 5 half-filled shell) which produces the
electric field gradient will occupy either dxy or d x 2_y2 or dz'l. This
could be resolved by determining the sign of e2qQ.
The chemical isomer shift for the dimer is the same in both
argon and krypton (-0.14 mm s-I), and suggests that the dimer is
a bound state with a constant internuclear separation. The value
indicates an electronic configuration of about 3d64s 1. 47 ; i.e. the
large overlap of the 4s-orbitals of the two atoms causes a large
reduction in effective 4s-density and a large increase in the shift
observed.
Similar monomer/dimer behaviour has been recorded in 119Sn
[6]. Another tin species which has been observed is the isolated
SnO molecule in matrices of argon and nitrogen [7]. The quadru-
......
.... . .. ...
. ,.... ".,
. .,"'. ... ....... .
. -.
, ..
1001-
.
.,-,'~...'..
,. :. I • ••
.' . l'
'" . # Kr/Fe -136
'"
.:s;;; 99,5-
..
'E
." .....
..•
<
.,
c:
...........
"
I
,=
... .-
~
lOOt-- •
..... /', .<1.' _~
Ie· .--.-,
t~ • •."., '. :
.....
•
.~.
.
..
'10 ,
\
. ,,.,.'
).... I.
:
. , Kr/Fe=44
99·5 -
..
,I
-,.
-2 -1 0 2
Fig. 10.4 The 57 Fe Mossbauer absorption spectrum of iron atoms isolated

in a matrix of krypton at 4 K with relative concentrations of Kr/Fe = 136
and 44. ([5],Fig.2).
pole splittings of 4.10 and 4.40 mm S-1 respectively are unusually

large for tin(II), and reflect the large imbalance in the occupation
of the p-orbitals when only one covalent bond is formed. A more
quantitative discussion of this splitting has not yet been devised.
10.2 Decay after-effects

In the majority of M6ssbauer experiments the effects of the nuclear
decay in the source can be ignored. This is because the source
matrix has been carefully chosen to minimize any chemical after-
effects. However, in insulating materials where the mobility of
electrons is low it is not unusual to find that the decay of the
M6ssbauer precursor has a substantial effect on the chemical
environment of the nucleus. The electron-capture, isomeric trans i-
tion, and {3-decays are of comparatively low energy and do not dis-
place the atom from its original lattice site. They may be distin-
guished from the more energetic reactions such as a-decay, Coulom-
bic excitation and (n, 'Y) in situ reactions, where the large recoil
energy will cause displacement. As a result the emission of Moss-
bauer 'Y-rays in the latter may take place from a number of differ-
ent sites in the lattice.
The electron-capture process (EC) removes an electron from an
inner s-shell of the atom and is then followed by an Auger cascade.
The net effect is to produce a highly ionized daughter state which
for an isolated uncharged parent atom can in principle have a
charge as high as +6. In the solid state this charge is quickly reduc-
ed to a more modest value by an influx of electrons from neigh-
bouring atoms, and it is generally accepted that the final equili-
brium charge-state is achieved within a time-scale much shorter
than the lifetime of a Mossbauer excited level (i.e. < 10- 8 s). For
this reason, in metals, where the mobility of electrons is high, no
effects of the decay are recorded. Nevertheless, in insulating
solids, there remains the possibility that the equilibrium charge
state is not that predicted from the chemical state of the parent
atom, because the energetic Auger cascade may have had import-
ant effects on the immediate chemical environment. It is also
possible, though more difficult to prove, that the atom is still in
a metastable excited electronic state at the time of the Mossbauer
measurement.
A recent experiment has clearly demonstrated the formation of
multiple charge states by the EC decay of 57Co. 57Fe and 57Co
atoms were isolated in solid xenon [8]. The 57Co emission and
57Fe absorption spectra of such a matrix containing -1.1 at.%
iron and -10 J,LCi 57Co are shown in Fig. 10.5. Only a minor pro-
portion of the 57Co decays result in an isolated iron atom with
the same 3d64s 2 configuration as the 57Fe. These give a resonance
line at -0.76 mm s-l as described earlier (positive velocity defined
in the sense of an absorption experiment). The chemical isomer
shift of the major component in the emission spectrum at +1. 77
mm s-l can be attributed to Fe+ atoms with a 3d7 configuration.
The large value of the shift is greater than found for 3d 6Fe 2+ in for
example FeF 2. This 3d7 configuration must be metastable as the
free-ion ground-state is normally considered to be 3d 64s 1• The
latter configuration would give a very different chemical isomer
Fig. 10.5 The 57Co emission (top) and 57Fe absorption (bottom) spectra of
a matrix of solid xenon containing "'1.1 at.% of iron and "'10 fJ.Ci of 57Co.
On1~ a port.ion of the electron-capture decays give .t~e 3d64s 2 configuration
of Fe WhICh absorbs at -0.76 mm s-l. The additIOnal component at
+1. 77 mm s-1 corresponds to Fe+ ions in a metastable 3d7 state. The velo-
cities in the source experiment have been reversed in sign to facilitate com-
parison. ([8], Fig. 1).
shift, and has been recently observed following ultraviolet irradia-
tion of matrix-isolated 57Fe atoms, and has a shift of +0.26
mm s-1 [9].
The situation can be more complex when the decaying atom is
in a chemically bound state. The EC decay of 57Co in a simple
ionic lattice compound involves insufficient energy to eject the
atom from its lattice site. It might be anticipated that a daughter
57Fe atom will be formed in the most stable oxidation state for
the site occupied. This is probably what happens, but conflicting
results have been found in many compounds (particularly oxides).
This is believed to be because of the influences of non-stoichiometry
and lattice defects on the charge-stabilization. It is not unusual
to find that both Fe 2+ and Fe 3+ ions are formed in the same matrix,
and Fe+ or Fe 4+ ions are also sometimes observed. For example,
the decay of5 7Co in CaO (NaCl face-centred cubic lattice) has
been reported to result in Fe+, Fe 2+ and Fe 3+ charge states [10] .
The relative proportions in which different charge states are form-
ed are often temperature dependent, and it is likely that the mobi-
lity of lattice defects and the ease with which electrons can be
'trapped' in the lattice are the primary factors.
Data for doped anhydrous halides appear to be more consistent
than for the oxides. A typical spectrum from 57Co-doped MgF 2
at 295 K is shown in Fig. 10.6. The spectrum comprises two over-
lapping quadrupole doublets. That with a quadrupole splitting
of 1.56 mm s-1 and a chemical isomer shift of +0.49 mm s-1 (in
the absorption sense, relative to the Na4Fe(CNk 1OH 20 absorber)
corresponds to Fe 3+ (38% of the total area). The other doublet
with a splitting of 2.82 mm s-1 and a shift of + 1.41 mm s-1 cor-
responds to Fe2+. It has been suggested in the past that Fe 3+ will
tend to be formed if the cation size of the host is smaller than the
size of the Fe 2+ ion (the ionic radii for Fe 2+ and Fe 3+ are about 76
pm and 64 pm respectively). However, no Fe 3+ is formed in
57Co/MgC1 2 where the Mg2+ ionic radius is 65 pm. Recent results
for series of related compounds such as CoF 2, ~-Co(OHh, COCl2
CoBr2 and 57Co-doped MgF2, ZnF2, CoF 2, NiF2 suggest that the
Fe3+ charge state is preferentially stabilized in compounds with a
large lattice energy [11]. This can be ascertained by reference to
Table 10.1. In this context it is relevant to note the stabilization
of Fe4+as well as Fe 3+ in K3 57CoF6 for similar reasons [12]. This
compound has the cryolite structure which has a large value of
1500 ':"::~~';:::~~i.'''~''~~:;:':'::''':'' : i:\:-:~~:;;£;:::':";:':~:'

......;:,'
.r
"::' "/"{:.':-;:'::';';"~ .'
'. .'
'. .~:.
c
o
.~
.~ '485
e
...
~
".
1"70~L-__-L____~__~"____~__~____~__~__~
3 z o -I -2 -3 -4
Velocityf(mm 5- 1)
Fig. 10.6 The Mossbauer emission spectrum at 295 K for 57Co doped into
MgF2' 38% of the decays produce an Fe 3+ ion, while the rest give Fe 2+. The
spectrum has been reversed to resemble the equivalent absorption spectrum.
([ 11 ], Fig. O.
Table 10.1 The fraction of Fe3+ ions formed by electron-capture decay of

57Co in simple ionic compounds
Compound Structure Lattice % Fe 3+

energy
/(kJ mol-I)
NiF2 rutile 3080 59

MgF2 rutile 2930 38
CoF2 rutile 3060 32
ZnF2 rutile 2930 >0
FeF2 rutile 2840 0
MnF2 rutile 2760 0
l3-Co(OHh CdI2 2840 20
CoCl2 CdCl2 2610 0
CoBr2 CdI2 2590 0
MgCl2 CdCl2 2470 0
the Madelung constant and therefore a high lattice energy.
In co-ordination compounds with molecular ligands the effects
of the 57Co decay are more catastrophic. Hydrated C02+ salts
show some Fe 3+ in the Mossbauer spectrum, but conversely some
low-spin C03+ chelates generate a reduced Fe 2+ species. It is now
widely accepted that the electrons and X-rays released in the Auger
cascade often cause radiolysis of the surrounding ligands to produce
ions and free radicals. The final charge-stabilization of the daugh-
ter iron atom is determined by the oxidizing or reducing pro-
perties of the radicals, and by the ease with which electrons can
be 'trapped' nearby in the lattice. Thus the hydroxyl radical OH"
which will readily oxidize Fe 2+ to Fe 3+ is probably responsible
for the Fe3+ formed in hydrated compounds, while the organic
radicals produced in C03+ chelates are highly reducing and will
result in some·Fe 2+ ions.
Convincing evidence in support of the autoradiolysis mechanism
has been obtained by comparing the emission spectra from 57Co-
doped C03+ chelates with the conventional absorption spectra of
the isostructural iron compounds following intense irradiation with
electrons [121. Typical spectra for acetylacetone derivatives,
57Co:Co(acach at 77 K and Fe(acach irradiated at 80 K, are shown
in Fig. 10.7. The similarity between the emission and absorption
spectra in these and related complexes suggests that the species
stabilized by the 57Co decay are similar to those formed by extf'Tnal
radiolysis. In the case of 57Co: [CoIII(bipyridylh1 (C104h, the
pure compounds [Fe II(bipyridylh1 (CI04h and [FeIII (bipyridylh1-
(Cl04h are also known. Both the 57Co decay and the external
radiolysis of [Fe III (bipyridylh1 (C10 4h appear to generate the
[FeIl(bipyridylh]2+ species, but there may of course be significant
differences in the detailed mechanism for formation of the reduced
species in the two cases.
The radiation damage appears to be more severe in the more
ionic complexes. In the instance of 57Co:CoC12.2H20 it has been
demonstrated that the internal radio lysis produces a number of
different defect species in both the Fe 2+ and Fe 3+oxidation states
[ 131. This is particularly clear in the spectrum of the low-
temperature anti ferromagnetic phase where some of the defects
do not participate in the magnetic superexchange and a 'paramag-
netic' component is seen superimposed on the magnetic hyperfine
splitting. The EC-decay is probably less destructive in the bipyri-
4 2 0 -2 -4
J'.~ ••0::;' ".. . .... :. °0 .: ....

....
. ": .. .... . ..:.... ......;:
o •
.0
".
....
. .'
•••••• (0)
......
~
.'
....
°0 ••
~ .::: .. :... :.: .....

.~ .: .... :..
"e'" ':" I FeH
n
c:
c .'. ...
,=. ....: ,
.. ' ::
(b)
-4 -2 0 2 4
Velocity/(mm 5- 1)
Fig. 10.7 The 57Co emission spectrum from (a) 57Co:Co(acac» at 77 K

and (b) the 57Fe absorption spectrum of Fe(acac)3 at 80 K following irradia-
tion with electrons at the same temperature. The sense of the velocity in
the source experiment has been reversed. ([ 12], Fig. 4).
dyl derivatives because the ligand itself is more resistant to radioly-

sis, and it seems likely that none of the strong Fe-N bonds are ever
ruptured.
The effects of a nuclear decay by isomeric transition (IT) are
very similar to those of electron capture. A proportion of the de-
cays result in an internal ionization (in this case by internal con-
version of the 'Y-ray preceding the decay of the Mossbauer level)
and a subsequent Auger cascade. The 23.8-keV level of 119S n is
populated by a highly converted isomeric transition from the
89.54-keV level of 119mS n. A good example of after-effects is
given by the decay of 119mSn in tin(II) sulphate [14]. The
119mSn :S nS04 emission spectrum and the 119Sn absorption
spectrum of the same matrix are shown in Fig. 10.8. The
sign of the velocity scale for the source experiment has
been reversed to facilitate comparison. The tin atoms in the matrix
are all initially in the +2 oxidation state as expected for tin(II)
sulphate, but a significant proportion of the decays from the 119m Sn
state result in the production of 119Sn in the +4 oxidation state.
. . ..:.::..:.,,-'/
......'"\
...........
... ...
-.:
..' '.
......
:- .,- .
.,:I ••
575 f- ..:-.......... .:,: .

' .
.-:~. ~'.\;..:\ ..
. ....
. .:..:
570 ~
< .'
....
..~
:, .
.....
..... ...
",
:.
.....
.:
, -.
:. .
..
............
.' .. !
i
\
:.
220 f-
..
:
.."'.,-..
~
.
200
0:'
'.:.
I I I I I J I I
-6 -4 -2 0 2 4 6 8
Velocity I (mm 5- 1 )
Fig. 10.8 The 119mSn emission spectrum (upper) and 119Sn absorption
spectrum (lower) from a matrix of SnS04 doped with 119mSn. Note the
production of some Sn4+ species by the isomeric transition. The velocity
scale of the source experiment has been reversed to facilitate comparison.
([ 14], Fig. 1).
The concentration of these defects is too low to register in the
absorption spectrum. This oxidation is attributed to the conse-
quences of an autoradiolysis process which accompanies the Auger
cascade. However, decay after-effects are not very common in
tin because the covalent bonds are usually stronger than in iron
compounds.
If the primary decay is by t3-emission, then the effects on the
atomic environment are less drastic. The increase in the nuclear
charge by +e effectively increases the oxidation state of the atom.
If the 'daughter' environment is isoelectronic and isostructural
with a stable compound, then the co-ordinate geometry is un-
likely to be affected. The primary influence appears to be the
chemical stability of the daughter product in the environment of
the parent lattice. In some instances it is possible to characterize
previously unknown ionic species trapped in the host matrix.
Thus the {3-decay of 129Te in (NJ4hTeCI6 populates the 27.8-keV
Mossbauer transition of 129 1 [15]. A single-line 129 1 emission is
observed at -6.08 mm s-l relative to a Zn 129 Te source (Le. at
+6.08 mm s-l for an absorption). The regular octahedral TeCll-
anion appears to decay to an isostructural ICI6" anion. There is
also an alternative source precursor. 129mTe decays by an isomeric
transition to 129Te prior to the ensuing t3-decay. A source of
(NH4hTeCI 6 doped with 129m Te gives the same single-line spectrum
as from 129Te. In this instance neither the isomeric transition nor
the t3-decay cause molecular fragmentation, showing that the 106"
anion is comparatively stable.
Similar experiments have been used to prepare molecules of
unknown xenon compounds. The 39.58-keV Ig= 4- ~ Ie= f reso-
nance of 129Xe can be easily observed using a source of 1291 in a
matrix of NaI, K104 or Na3H2I06, all of which show a single-line
absorption with solid xenon or xenon hydroquinone clathrate as
the absorber. In contrast, absorbers of XeF 2, XeF 4, and Xe03
show well resolved quadrupole doublets with Il = 39,41, and 11
mm s-l respectively because withdrawal of electron density from
the xenon by the ligands causes an asymmetry in the valence
orbital occupation.
Unexpectedly, the three sources mentioned above show small
but significant chemical isomer shifts when compared with the
same clathrate absorber, and it can be argued that the t3-decay of
the source generates a xenon species with some degree of chemical
binding to the lattice. More dramatically, the decay of 1291°3 in a
matrix of Nal03 gives a spectrum with a clathrate absorber show-
ing a quadrupole splitting identical to that from an Nal source and
an Xe03 absorber, so that it is an obvious inference that the i3-decay
has created an {Xe03} molecule trapped in the iodate lattice.
Similar experiments [16, 17] with KICI4.H20, KICI2.H20 and
KIBr2 sources show quadrupole split spectra attributable to trap-
ped {XeC14}, {XeCI2} and {XeBr2} molecules, none of which are
known as stable compounds but which may be presumed to be
isoelectronic with IC14", ICI 2 and IBr2 respectively. The observed
splittings are f1 = 25.6, 28.2 and 22.2 mm s-1 respectively. Com-
parison of these values with those of XeF 2 and XeF4 (see Fig. 10.9)
show consistency with the predicted bonding in these molecules.
{XeCI4} is square-planar, and {XeCI2} and {XeBr2} are linear mole-
cules. In the tetrahalides, fluorine is more electron withdrawing
than chlorine, and in the dihalides the order is fluorine> chlorine
> bromine as predicted.
However, not all 1291 i3-decay sources give a chemically bonded
xenon atom. Both 12912 and Cs129IBr2 for example appear to break
down to atomic xenon within the time scale of the Mossbauer
event (10- 9 s). It should be emphasized that the {XeBr2} mole-
cule in KIBr2 may in fact have a lifetime of less than a microsecond.
An example of a change in oxidation state following i3-decay is
given by the decay of 1930s in K2[OsCI 6] and (NH4h[OsCI6] in
which the Os4+ has a low-spin 5d 4 configuration. The decay popu-
lates the 73-keV level of 193Ir, and the oxidation state of the Ir
daughter nucleus proves to be +4 (Sd 5) [18]. This configuration
is known to be stable in K2[IrCI 6] and (NH4h[IrCI6] , and pre-
sumably the [IrCI 6] - ion which is initially formed is rapidly reduc-
ed to the more stable [IrCI6]2- configuration.
The effects on the Mossbauer spectrum of an energetic nuclear
reaction such as a-decay, Coulombic excitation by 2S-Me V oxygen
ions, or an (n, 'Y) reaction are often surprisingly small despite the
fact that the nucleus is displaced from its original site. In all cases
the nucleus seems to reach its final lattice site in a time much less
than the Mossbauer lifetime (10- 8 s), and in metallic matrices
the spectrum recorded usually appears identical to that from a con-
ventional absorption experiment. Few data are available for non-
metallic compounds, and it is in these that after-effects are likely
to be severe. A reduction in the apparent recoilless fraction is
4·0
-4,0
(a)
-8,0
-12'0
4'0
-4·0 (b)
{Xeel z }
_ -8'0
~
4·0
"
0
.~
0
...
~
.Q
-4·0
-8,0
-12'0
-16·0
4'0
0
...
-4·0
XeF z (d)
-8'0
-50 -40 -30 -20 -10 0 10 20 30 40 50

Velocity/(mm s-')
Fig. 10.9 The 1291 Mossbauer emission spectra of (a) {XeCI4} trapped in
K1C4, (b) {XeCI2} trapped in K1C12, and the 129Xe absorption spectra of
(c) XeF4and(d) XeF2. ([161, Fig.I).
sometimes found, although whether this is to be attributed to a

residual kinetic heating or to lattice defects is not clear.
Nevertheless there is considerable scope for the study of radia-
tion damage provided that the experiment is designed appropriate-
ly. A good example is given by data for thermal neutron capture
by 192()s impurities in a matrix of a-iron metal [19]. The 1930s
daughter nucleus j3-decays with a half-life of 31 h to the 73.1-keV
state of 193Ir. Observation of the 193Ir Mossbauer emission spec-
trum thus records the environment of the 1930s atom long after
any thermal disturbance has subsided. The irradiation was carried
out at 4.6 K, and the Mossbauer spectrum recorded subsequently
without any intervening rise in temperature showed an internal
magnetic flux density of -138.6 T produced by magnetic alignment
of the 193Ir within the iron lattice. The matrix was then annealed
at successively higher temperatures by a heating/cooling cycle, and
it was found that between 90 and 140 K the flux density (measur-
ed at 4.6 K) increased irreversibly to -140.8 T with a simultaneous
narrowing of the resonance linewidth. Furthermore, these effects
were independent of any change in the total neutron dosage, and
were therefore related to the 1920s(n, r) capture event and not to
general radiation damage of the iron lattice.
The recoil energy of the 1930s is estimated to be about 30 eV,
which can be compared with estimates for the minimum energies
required to produce displacement in iron along the (100), (110)
and (Ill) directions of 17, 34 and 38 eV respectively. The observ-
ed reduction in the magnetic flux density in the unannealed state
is believed to be due to a vacancy in the nearest neighbour iron site
(following the general arguments of Chapter 8). The recoil of the
(n, r) reaction causes the 1930s atom to displace an iron-atom from
a near-neighbour site, and thereby creates a vacancy at its original
site. The displaced iron atom occupies a nearby interstitial posi-
tion. This vacancy lin terstitial defect can thermally anneal once
the activation energy for migration of the atom of about 0.3 2 eV
has been exceeded.
References
[1] Steger, J. and Kostiner, E. (1973) 1. Chern. Phys., 58,3389.
[2] Abeledo, C. R., Frankel, R. B., Misetich, A. and Blum, N. A. (1971)
J. Appl. Phys., 42,1723.
[3] Wagner, F. E., Wortmann, G. and Kalvius, G. M. (1973) Phys. Letters,
42A,483.
[4] Liddell, P. R. and Street, R. (1973) J. Phys. F, 3, 1648.
[5] McNab, T. K., Micklitz, H. and Barrett, P. H. (1971) Phys. Rev. (B),
4,3787.
[6] Micklitz, H. and Barrett, P. H. (1972) Phys. Rev. (B), 5, 1704.
[7] Bos, A., Howe, A. T., Dale, B. W. and Becker, L. W. (1972) Chern.
Cornrn .• 730.
[8] Micklitz, H. and Barrett, P. H. (1972) Phys. Rev. Letters, 28, 1547.
[9] Data presented by Micklitz, H. and Litterst, F. J. at the International
Conference on the Applications of the Mbssbauer Effect at Bendor,
September 1974.
[10] Regnard, J. R. (1973) Solid State Cornrn., 12,207.
(11] Cruset, A. and Friedt, J. M. (1971) Phys. Stat. Sol (B), 47,655.
[12] Baggio-Saitovitch, E., Friedt, J. M. and Danon, J. (1972) J. Chern.
Phys., 56, 1269.
[13] Friedt, J. M., Shenoy, G. K., Abstreiter, G. and Poinsot, R.(1973)
J. Chern. Phys., 59,3831.
[14] Llabador, Y. and Friedt, J. M. (1971) Chern. Phys. Letters, 8, 592.
(15] Jones, C. H. W. and Warren, J. L. (1970)J. Chern. Phys., 53, 1740.
[16] Perlow, G. J. and Perlow, M. R. (1968) J. Chern. Phys., 48,955.
(17] Perlow, G. J. and Yoshida, H. (1968) 1. Chern. Phys., 49,1474.
(18] Zahn, U., Potzel, W. and Wagner, F. E. (1973) Perspectives in Mbssbauer
Spectroscopy, ed. Cohen, S. G. and Pasternak, M. (1973) Plenum Press,
New York, p. 55.
[19] Vogl, G., Schaeffer, A., Mansel, W., Prechtel, J. and VogI, W. (1973)
Phys. Stat. Sol. (B), 59, 107.
CHAPTER ELEVEN
Biological Systems
Some of the most interesting and stimulating applications of Mbss-

bauer spectroscopy have been in the field of biological science.
However, the practical and theoretical pro blems involved are pro-
bably more severe than in any of the topics covered in the preceed-
ing chapters.
It is now recognized that many of the large protein molecules
which control biological functions utilize the oxidation-reduction
properties of a transition-metal atom, and that the complicated
protein moiety functions largely in the role of ensuring steric
specificity of the reaction. In very few instances has the detailed
X-ray structure of a metallo-protein been determined. For this
reason, any technique which can exclusively monitor the active
metal-centre provides valuable infonnation about the chemical role
of a particular protein. For example, if the metal atom has a para-
magnetic configuration, the electron spin resonance (E.S.R.) spec-
trum records the behaviour of the metal site to the exclusion of
the protein bulk.
In a similar way, any protein containing iron can be studied
using the 57 Fe Mbssbauer resonance which is also specific to the
active site, but with the additional advantage that diamagnetic
compounds, which are excluded from E.S.R. measurements, can
also be studied. One of the major experimental difficulties lies
in the relatively low iron content of proteins and the low natural
abundance of 57Fe (2.17%). Fortunately the elements forming
the amino-acid chains do not give a strong non-resonant scattering,
234
Biological Systems 235
but nevertheless with na tural concentrations of iron the Moss-
bauer resonance is always comparatively weak. However, in many
instances it is possible to cultivate organisms with a diet enriched
in 57Fe, and the 40-fold improvement in the resonant cross-section
which can be achieved results in an adequate absorption intensity.
Alternatively, for some proteins it is possible to remove the iron
and then to re-constitute the protein with enriched 57Fe without
affecting the bio logical activity.
The preparation of separated protein material is often a major
task in itself, and particular care must be taken to ensure that
accidental denaturing of the protein does not take place. Pure
compounds in crystalline form are rarely obtainable, but experi-
ence has shown that satisfactory results may be obtained on frozen
solutions of protein concentrates. Freeze-drying or lyophilization
is also often used. Nevertheless, there remains the possibility of a
conforma tional change in the solid so that the 'active' solution
form is not observed.
The Mossbauer spectrum of the iron site reflects the ligand-field
of the surrounding organic groups, and therefore the principles
described in Chapters 4 and 5 are directly applicable. The major
innovation is that individual iron atoms are now separated by
distances of the order of 25 A (2500 pm), and as shown in Chapter
6 this has an important influence on the relaxation properties of
paramagnetic configurations. In such cases the Mossbauer spectra
at low temperatures usually show a paramagnetic hyperfine split-
ting which slowly collapses with rising temperature. The spin-
spin mechanism of electronic relaxation between neighbouring iron
nuclei is comparatively ineffective, and in some instances is domi-
nated by the very weak interactions between the electrons on the
resonant atom and the nuclear spins of the ligand atoms. When
this occurs, the zero-field Mossbauer spectra are extremely diffi-
cult to interpret. The application of an external magnetic field
then results in a considerable simplification of the observed spec-
trum, from which many of the interaction parameters appropriate
to the zero-field spectra may be obtained.
Some biologically active compounds such as vitamin B12 (cyano-
cobalamin) contain cobalt, and although it is possible to dope with
57Co, this technique has not been widely exploited because of two
major disadvantages. It is unlikely that the electronic configuration
of the 57Fe daughter nucleus will be related to that of the cobalt
parent, and furthermore there is also a danger of local damage by
autoradiolysis (see Chapter 10).
More detailed examples to illustrate these general principles are
taken from the haemoproteins and ferredoxins.
11.1 Haemoproteins
The haemoproteins are well known as the oxygen carriers in blood.
The oxygen is transported by chemical binding to an iron atom,
which is itself in the centre of a planar porphyrin structure known
as protoporphyrin IX (illustrated in Fig. 11.1). In the protein
haemoglobin, the porphyrin is attached by co-ordination of the
iron to a fifth nitrogen in a histidine unit of the protein chain
below the plane of the four nitrogen ligands. The sixth co-ordina-
tion position of the iron on the opposite side of the porphyrin is
thus left vacant, and forms a site for labile bond formation with
small molecules.
It is possible to isolate compounds of iron with protoporphyrin
IX, and although it might be expected that these would be model
compounds for study of the haemoproteins themselves, this pro-
mise is not entirely fulfilled. The iron(lII) chloride derivative of
CH z
II
CH
Me
Me
Fig. 11.1 The structure of protoporphyrin IX.

protoporphyrin IX is commonly referred to as haemin, and the
reduced iron(II) derivative as haeme.
Data for haemin and other ring-substituted porphyrins with a
selection of ligands in the fifth co-ordinate position show that the
iron is normally present in the S = 1high-spin Fe 3+ state [ 1]. The
unusually low chemical isomer shift and large quadrupole splitting
(see Table 11.1) are not unexpected because of the substantial
covalency in the bonds to nitrogen and the distorted symmetry of
the square-based pyramidal geometry. The degree of distortion
of the iron environment as monitored by an increase in the quadru-
pole splitting follows the sequence fluoro < acetato < azido
< chi oro < bromo.
These haemin derivatives are peculiar in that they show an un-
expected temperature dependent effect. The spectrum at 4.2 K
is a sharp quadrupole doublet, but as the temperature is raised
the component at more positive velocity (which is the m z = ±t
-+m z = ±t transition because e 2qQ is positive) broadens rapidly to
give a very asymmetric spectrum. This effect is caused by a para-
magnetic relaxation mechanism which is described in detail in
Chapter 6 and illustrated there with data for Fe(acachCI which
has a similar co-ordinate geometry.
The bispyridyl haeme iron(II) derivative is an S = 0 low-spin
compound, and the M6ssbauer parameters are almost identical to
those of the iron(lII) derivatives because of the covalency effects.
Table ILl Mossbauer parameters for haemoprotein derivatives
Compound* S T/K Ll 8(Fe)

/(mm s-l) /(mm s-l)
5 4 0.83
Haemin 2" 0.36
bispyridyl haeme 0 4 1.14 0.36
HbCO 0 4 0.36 0.26
Hb02 0 1.2 2.24 0.24
77 2.19 0.26
195 1.89 0.20
Hb (reduced) 2 4 2.40 0.91
5
HiF 2" 1.2-195 broadened
HiCN 1 195 1.39 0.17
2"
* The abbreviation Hb is used for an Fe(I1) haemoglobin compound and Hi for an Fe(lII)
haemoglobin compound.
However, this diamagnetic configuration cannot show relaxation
and therefore gives a sharp symmetrical spectrum at all
temperatures.
The majority of the data on haemoglobin derivatives have been
obtained using the blood from rats injected with ferric citrate
solution containing iron enriched to 80% in 57Fe [2]. It is not
always easy to isolate pure compounds, and in this case spectra
have usually been recorded in frozen solutions of red-cell concen-
trates or haemoglobin extracts. The oxidation state of the iron
is very dependent on the nature of the sixth ligand. The carbon
monoxide and oxygen derivatives of haemoglobin (HbeO and
Hb02) are both S = 0 low-spin iron(II) compounds. The spectra
comprise a simple quadrupole doublet, and typical parameters
are given in Table 11.1. The diamagnetic S = 0 state may be
distinguished from paramagnetic configurations by the absence
of any induced hyperfine splitting when a small external magnetic
field is applied.
The applied-field method of determining the sign of e 2 qQ has
been used to show that e2qQ is negative in Hb02, but unfortunately
the magnitude of f/ is not known. The large quadrupole splitting
and low chemical isomer shift of Hb02 (compared for example
to the bispyridyl haeme) show that the oxygen-iron bond is strong-
ly covalent. A possible geometrical arrangement and bonding
scheme [2] are shown in Fig. 11.2. The oxygen molecule is
assumed to lie parallel to the porphyrin plane to give a strong
overlap between the d yz metal orbital and the 7T*-antibonding or-
bital on the oxygen. The resultant separation of the molecular
orbitals causes spin-pairing of the otherwise paramagnetic oxygen
electrons to give the diamagnetic configuration observed. The
initially non-bonding dyz electrons thus become intimately in-
volved in the bonding. The molecular orbital (dyz + 7TiJ/..j2 has
only half the character of a dyz orbital, and this results in an effec-
tive decrease in the occupation of this orbital. As a result, the
principal axis of the electric field gradient tensor should lie along
the x axis of Fig. 11.2, and e2qQ should be large and negative (as
observed experimentally). The quadrupole splitting is unusually
temperature dependent for a diamagnetic configuration (see
Table 11.1). This has been suggested to indicate a possible libra-
tional motion of the oxygen molecule at higher temperatures.
The paramagnetic derivatives show various configurations; e.g.
0, Complex Fe
Fig. 11.2 A possible bonding scheme in Hb02 showing how the interaction
of the1r:molecular orbital on the oxygen with the dyz orbital on the iron can
remove the degeneracy and cause spin pairing in the bound state.
reduced haemoglobin is probably an S = 2 high-spin Fe 2+ com-

pound, the fluoride (HiF) is an S = i high-spin Fe 3+ compound,
and the cyanide (HiCN) is an S = 4- low-spin Fe 3+ compound. In
all three the quadrupo1e-split spectra seen at high temperatures
become grossly broadened on cooling by paramagnetic hyperfine
splitting as the relaxation time increases. Equally complex be-
haviour results when an external magnetic field is applied. The
resultant spectra are characteristic of each particular compound,
and are a sensitive monitor of the iron environment. Successful
solution of the appropriate theory can give highly detailed infor-
mation. A good example is provided by the zero-field Mossbauer
spectrum of haemoglobin fluoride at 1.2 K [3] which is shown in
Fig. 11.3. The S = has the Sz = ±} Kramers' doublet
~ Fe 3+ ion
lying lowest (see Chapter 6), and measurements in an applied
magnetic field confirm this. However, in zero-field conditions
the observed spectrum is clearly inconsistent with the prediction
for an Sz = ±} electronic ground-state shown in curve (a) of Fig.
11.3. The discrepancy can be explained by invoking additional
weak spin-interactions with the ligand nuclei which also induce
electronic relaxation. Curve (b) was calculated including the
effects of the fluorine nucleus, and is clearly a substantial improve-
ment. A calculation including the fluorine and all four nitrogens
is formidable, but curve (c) represents the effect of including only
one additional 1= 1 nitrogen nucleus in the calculation. The
result is good evidence in support of these nearest-neighbour nuclear
spin interactions. Future developments may lead to the determina-
-10 o 5 10
Velocity I (mm S-I)
Fig. 11.3 The Mossbauer spectrum of haemoglobin fluoride at 1.2 K with-

out an applied magnetic field. Curve (a) shows the predicted spectrum
neglecting ligand-nucleus interactions. Curve (b) shows the predicted effect
of including the transferred huperfine interaction with fluorine, and in curve
(c) an additional I = I nucleus is introduced to represent nitrogen. ({3], Fig. I ).
tion of the co-ordinate geometry in uncharacterized proteins.
The general principles described briefly here are also applicable
to other haeme derivatives; e.g. the myoglobins involved in oxygen
storage in muscle tissue, the cytochromes associated with the oxi-
dation of nutrients in cells, and the peroxidase enzymes which acti-
vate hydrogen peroxide.
11.2 Ferred0 xins

An equally important class of proteins occurring in bacteria and
plants are the iron-sulphur proteins such as the mbredoxins and
ferredoxins. They are associated with electron-transfer processes,
and in particular with photosynthesis. Ferredoxins from the
higher plants have a molecular weight in the region of 12000-
24000 and contain two atoms each of Fe and labile sulphide (i.e.
not attached to the amino-acid cysteine), and accept one electron
upon reduction. The ferredoxins in photosynthetic bacteria contain
eight atoms each of Fe and labile sulphide, accept two electrons
per molecule, and are also more deeply involved in other chemical
processes such as fixation of atmospheric nitrogen.
The Mossbauer spectra at 195 K of the oxidized and reduced
forms of the plant ferredoxin from the green alga Scenedesmus
are shown in Fig. 11.4 [4]. Similar spectra have been recorded in
plant ferredoxins from the green alga Euglena and from spinach.
The four resonance lines in the spectrum of the reduced form can
be interpreted as two quadrupole doublets arising from S = i high-
spin Fe 3+ (inner doublet) and S = 2 high-spin Fe 2+ (outer doublet)
in equal proportions. The parameters are given in Table 11.2.
The chemical isomer shift of +0.56 mm s-1 for Fe 2+ (with respect
to Fe metal) is considerably lower than is usually observed
because of the high covalency in a tetrahedral co-ordination to
sulphur. The complicated spectra measured at low temperatures
in an external magnetic field establish that the two iron atoms
couple antiferromagnetically to give a net spin of S =}. The oxi-
dized form of the ferredoxin contains two almost equivalent
i
S = high-spin Fe 3+ cations which couple antiferromagnetically
to give a diamagnetic ground-state. The iron atoms must therefore
be very close together in the protein. Data from electron spin
resonance measurements show that the co-ordination of the iron
is tetrahedral. This active centre which is illustrated in Fig. 11.5
·::/~~::i;::·:1~.\. /.;-:~~~~:~.:t~~:
(a) .'
/:.
.:"~'.
".
-3 -2 -1 0 1 2 3 4
Fig. 11.4 Mossbauer spectra at 195 K of (a) the oxidized and (b) the reduc-
ed forms of Scenedesmus ferredoxin enriched with 57Fe. ([41, Fig. 6).
Table 11.2 Mossbauer parameters in ferredoxins

------_.
Compound T/K S ~ 8(Fe)
/(mm s-l) /(mm s-l)
Scenedesmus ferredoxin
reduced form 195
{~ 5
0.59
2.75
+0.22
+0.56
oxidized form 195 2" 0.60 +0.20
Chromatium high-potential
iron protein
reduced form 77 ? l.l2 +0.42
oxidized form 77 ? 0.79 +0.32
(Et4Nh[Fe4S4(SCH2Ph)41 ? 1.26 +0.32
r
z
CYS CYS'l
"JJ
W 0
f-
-<
~ ~
z
YCys s Cysr-J
Fig. 11.5 Representation of the active site in the reduced form of a plant
t
ferredoxin. The S = and S = 2 spin-states couple antiferromagnetically
to give a resultant of;:} = -}. In the oxidized form, both spin-states are S = ~
and couple antiferromagnetically to give S = o.
appears to be characteristic of all plant ferredoxins.

Much less is known about the bacterial ferredoxins. Recent
X-ray studies have suggested the presence of a distorted cubic
Fe4S4 cluster with iron and sulphur at alternate vertices. Such a
cluster forms the active site of the high-potential iron protein
from Chromatium, and two similar clusters are found in the clostri-
dial ferredoxin from Micrococcus aerogenes (making a total of
8 Fe and 8 labile sulphide per molecule). Each iron is bonded to
the polypeptide chain by a mercaptide sulphur on a cysteinyl
residue.
The oxidized and reduced forms of Chromatium high-potential
iron protein both give a simple quadrupole doublet at 77 K (see
Table 11.2), and it appears that all four Fe atoms in the cluster are
equally affected by the reduction [5]. Only the oxidized form
shows paramagnetic relaxation effects at low temperature, although
there is some evidence suggesting that there might be two struc-
turally distinct pairs of iron atoms.
A similar Fe4S4 structural unit has recently been characterized
[6] in the inorganic compound (Et4Nh[Fe4S4(SCH2Ph)4]. The
Mossbauer spectrum between 4.2 K and 296 K is also a simple
quadrupole doublet with similar parameters. Polarographic stu-
dies revealed a reversible one-electron transfer process, and there
is good reason to believe that electron delocalization takes place
within the Fe4S4 core. The compound may therefore be taken
as a 'model' for the bacterial ferredoxins, the active centre of
which can be represented as shown in Fig. 11.6.
The clear distinction between the Mossbauer spectra of ferre-
/J! ,odor"
(cYS)S~
S---Fe
/ / -----.S(CYS)
Fe 5
(CYS)S/
Fig. 11.6 Representation of the active site in the bacterial ferredoxins.
doxins from bacteria and from plants has proved useful in the bio-
logical classication of the intermediate algae. The blue-green algae
and photosynthetic bacteria both lack a nucleus within the cell
envelope and carry out photosynthesis throughout the cell. In
contrast, the green algae and higher plants localize this function
within a chloroplast inside the cell. However in other ways the
biological activity of the ferredoxin in blue-green algae appears to
be more closely related to the higher plants. The M6ssbauer spec-
trum of a sample extracted from the planktonic blue-green alga
Microcystis flos-aquae and chemically reconstituted with 57Fe
enriched iron is identical with those described earlier from the
higher plant forms Scenedesmus and spinach ferredoxins [7].
The establishment of this rela tionship of the blue-green algae
ferredoxins to those of the higher plants provides useful evidence
in support of the hypothesis that the chloroplasts in the cells of
higher plants are the descendents of blue-green algae that lived
symbiotically within the early plant cells.
References
[1] Moss, T. H., Bearden, A. J. and Caughey, W. S. (1969) 1. Chem Phys.,
51,2624.
[2] Lang, G. and Marshall, W.(1966) Proc. Phys. Soc., 87,3.
[3] Lang, G. (1968) Phys. Letters, 26A,223.
[4] Rao, K. K., Cammack, R., Hill, D. O. and Johnson, C. E. (1971)
Biochem. J., 122,257.
[5] Evans, M. C. W., Hall, D. O. and Johnson, C. E. (1970) Biochem. J.,
119,289.
[6] Herskovitz, T., Averill, B. A., Holm, R. H., Ibers, J. A., Phillips, W. D.
and Weiher, J. F. (1972) Proc. Nat. Acad. Sci. U.S.A., 69, 2437.
[7] Rao, K. K., Smith, R. V., Cammack, R., Evans, M. C. W., Hall, D. O.
and Johnson, C. E. (1972) Biochem. 1., 129, 1159.
Bibliography
The following selection of more general references are intended

for the reader who wishes to pursue the subject further.
Books
Mossbauer spectroscopy - an introduction for inorganic chemists and geo-
chemists, Bancroft, G. M. (1973) McGraw-Hill, London, 252 pp.
Mossbauer effect and its applications, Bhide, V. G. (1973) New DeIhl,
Tata McGraw-Hill.
Mossbauerspectroscopy, Greenwood, N. N. and Gibb, T. C. (1971) Chapman
and Hall, London, 659 pp.
Chemical applications of Mossbauer spectroscopy, Goldanskii, V. I. and
Herber, R. H. (eds.) (1968) Academic Press, New York, 701 pp.
An introduction to Mossbauer spectroscopy, May, L. (ed.) (1971) Plenum
Press, New York, 202 pp.
Der Mossbauer-Effekt und seine Anwendung in Physik und Chemie,
Wegener, H. (1965) Bibliographisches Institut, Mannheim, 214 pp.
Mdssbauer effect: principles and applications, Wertheim, G. K. (1964)
Academic Press, New York, 116 pp.
Mossbauereffectdata index 1958-1965, Muir,A. H., Ando, K. J. and
Coogan, H. M. (1966) Interscience, New York, (a catalogue of references
up to early 1966).
Mossbauer effect data index, Stevens, J. G. and Stevens, V. E. (eds.) IFI/
Plenum Data Corp., New York, (an annual publication listing references
from 1969 onwards).
Mdssbauer effect methodology, Vols. 1-8, Gruverman, I. J. (ed.) Plenum
Press, New York, (an annual publication from 1965 containing a collection
of contributed papers and reviews biased towards instrumentation)
Perspectives in Mossbauer spectroscopy, Cohen, S. G. and Pasternak, M. (eds.)
(1973) Plenum Press, New York, 259 pp, (collected conference papers).
Proceedings of the conference on the application of the Mossbauer effect -
Tihany (Hungary), 1969, Dezsi, I. (ed.) (1971) Akademiai Kiado,
Budapest, 803 pp.
Mossbauer spectroscopy and its applications, a Panel Proceedings of the
International Atomic Energy Agency, Vienna (1972) 421 pp, (a collection
of 16 review articles).
Spectroscopic properties of inorganic and organometallic compounds, vols.
1-7, Greenwood, N. N. (ed.) The Chemical Society, London, (containing
an annual review of the literature from 1967 onwards).
245
Reviews
Bancroft, G. M. and Platt, R. H. (1972) Mossbauer spectra of inorganic com-
pounds: bonding and structure,Adv. Inorg. Chem. and Radioch em. , 15,
59-258.
Bearden, A. J. and Dunham, W. R. (1972) Iron electronic configurations in
proteins: studies by Mossbauer spectroscopy, Structure and Bonding, 8,
1-52.
Cohen, R. L. (l972) Mossbauer spectroscopy: recent developments, Science,
178,828-835.
Devoe, J. R. and Spijkerman, J. J. (1970) Mossbauer spectrometry, Analyt.
Chem., 42, 366R-388R.
Friedt, J. M. and Danon, J. (1972) Applications of the Mossbauer effect in
radiochemistry, Radiochimica Acta, 17, 173-190.
Gibb, T. C. Applications of Mossbauer spectroscopy to organometallic
chemistry. In Spectroscopic methods in organometallic chemistry, ed.
George, W.O., p. 33-60, Butterworths, London.
Go1danskii, V. I., Kbrapov, V. V. and Stukan, R. A. (1969) Application of
the Mossbauereffect in the study of organometallic compounds,
Organometallic Chem. Rev. A, 4, 225-261.
Herber, R. H. (1967) Chemical applications of Mossbauer spectroscopy,
Progr.lnorg. Chern., 8, 1--41.
Hobson, Jr., M. C. (l972) The Mossbauer effect in surface science, Progr.
Surface Membrane Sci., 5, 1-61.
Johnson, C. E. (1971) Applications of the Mossbauer effect in Biophysics,
J. Appl. Phys., 42, 1325-1331.
Kurkjian, C. J. (1970) Mossbauer spectroscopy in inorganic glasses, 1. Non-
cryst. Solids, 3, 157-194.
Lang, G. (1970) Mossbauer spectroscopy of haem proteins, Quart. Rev.
Biophys., 3, 1-60.
Maddock, A. G. (1972) Mossbauer spectroscopy in the study of the chemical
effects of nuclear reactions in solids, MTP International Review of
Science, Vol. 8, p. 213-250, Butterworths, London.
Parish, R. V. (1972) The interpretation of 119Sn-Mossbauer spectra,Progr.
Inorg. Chem., 15, 101-200.
Reiff, W. M. (1973) Magnetically perturbed Mossbauer spectra of iron and
tin co-ordination compounds, Coord. Chem. Rev., 10,37-77.
Stevens, J. G., Travis, J. C. and DeVoe, J. R. (1972) Mossbauer spectrometry,
Analyt. Chem., 44, 384R--406R.
Zuckerman, J. J. (1970) Applications of 119mSn Mossbauer spectroscopy to
the study of organotin compounds, Adv. Organometallic Chem., 9,
21-134.
Observed Mossbauer Resonances
A total of at least 100 Mossbauer resonances have been detected

in 83 isotopes of 44 elements. These are (energies in kev):-
40J( (29), 57Fe (14,136), 61Ni (67), 67Zn (93), 73Ge (13, 68), 83Kr
(9)
99Tc (140), 99Ru (90), 101Ru (127), 117Sn (159), 119Sn (24), 121Sb
(37), 125Te (35), 1271 (58), 1291 (28), 129Xe (40), l3lXe (80)
133Cs (81), l33Ba (12)
139La (166), 141Pr (145), 145Nd (67, 72), 147pm (91), 147Sm (122),
149Sm (22), 151Sm (66), 152Sm (122), 153Sm (38), 154S m (82),
151Eu (22), 153Eu (83,97,103), 154Gd (123), 155Gd (60,86, 105),
156Gd (89), 157Gd (64), 158Gd (79), 160Gd (75), 159Tb (58), 160J)y
(87), 161Dy (26, 44, 75), 162Dy (81), 164Dy (73), 165Ho (95), 164Er
(92), 166Er (81), 167Er (79), 168Er (80), 170Er (79), 169Tm (8),
170Yb (84), 17lYb (67, 76), 172Yb (79), 174Yb (76), 176Yb (82),
175Lu (114)
176Hf (88), 177 Hf (113), 178Hf (93), 180Hf (93), 181Ta (6, 136),
1SOW (104), 18ZW (100), 183W (46,99), 184W (111), 186W (122),
187Re (134), 1860s (137), 1880s (153), 1890s (36, 70, 95), 1900s
(187), 1911r (82, 129), 193Ir (73, 139), 195Pt (99, 129), 197Au (97),
20lHg (32)
232Th (50), 231Pa (84), 234U (43), 236U (45), 238U (45), 237Np (60),
239pu (57),243Am (84)
247
Index
Absolute calibration of velocity, 15 Chemical isomer shift-cant.
Absorber thickness, 11-2 effect of pressure, 122
Absorption cross-section, 10 in europium compounds, 77,130
of s7Fe, 17 in ferredoxins, 241
Absorption intensity, II in ferrocene, 98-100
Adsorption, study of, 211 in gold compounds, 127-9
Angular dependence of absorption in haemoproteins, 237
intensity, 40-1 in iodine compounds, 26, 77-84
Anisotropy of recoilless fraction, in iridium compounds, 126-7
43,141-4 in iron carbonyls and derivatives,
Antiferromagnetic coupling (intra- 46-53,61-2
molecular), 154 in iron cyanide complexes,
Antiferromagnetic ordering (see also 58-9,93-6
magnetic hyperfine splitting) in iron halides, 99-100
in Au 2Mn, 191 in matrix-isolated iron atoms
in FeF 2, lIS 220-4
in FeF 3, 112 in neptunium compounds, 131-3
in MnF2' 216-7 in organotin compounds, 63-4,
in spinels, 168-74 89-93
Neel type, 168 in palladium-tin alloys, 183-4
spin canting, 168, 172-3 in ruthenium compounds, 124- 5
Yaffet-Kittel type, 168 in silicates, 201-6
Antimony (lUSb), 18,84-6 in spin cross-over compounds,
halides, 85-6 119-21
organoantimony compounds, in tellurium compounds, 84
85-7 in tin(II) compounds, 87-8
transferred hyperfine field in, in tin(IV) halides, 89
180 of 99Rh, 1930s, 197Pt impurities
Antiphase boundaries, 196 in the transition metals, 217-8
Asymmetry parameter, 31-6 partial chemical isomer shifts,
in (diphos)Fe(FOh, 50 96-8
in [(tr-C sH s )Fe(COhhSnCI 2, 91 second order Doppler shift,
in FeF 2, II7 26-8
sign of, 75
Bam (unit) defined, 31 Chi-squared distribution, 20
Beta decay, after-effects of, 229-32 Chromatium high-potential protein,
Binomial distribution, 170 243
Biological systems, 234-43 cis-/trans-geometrical isomerism,
Brillouin, function, 112 54-6
C1ebsch-Gordan coefficient, 39-41
Cation vacancies in 'Y-Fe203, 165 Cobalt (S7Co) impurity doping, 142
Centre shift (see chemical isomer diffusion of in copper, 147
shift) in biological systems, 235
Chemical isomer shift, 22-8 in CaO, 224
calibration of, 26,74-5 in coordination compounds,
effect of change in oxidation 226-7
state, 74-7 in inorganic halides, 224- 5
effect of electron correlation, 76 in Ni3Al, 217-9
248
Index 249
Cobalt (S7Co) impurity doping-cont. Electric field gradient tensor-cont.
in zinc metal, 142 in iodine compounds, 77-84
matrix isolation in xenon, 222-3 lattice term, 33-5
Combined magnetic and quadrupole point charge calculations,
interactions, 36-8 54,69-71
Compton scattering, 11 principal values, 31
Computation of data, 19 valence term, 33-5
Conduction-electron polarization, Electric monopole interaction, 22
183 Electric quadrupole interaction, 22
Conformations, determination of, Electric quadrupole (E2) transition,
60-3 40
Coordination number, determination Electromechanical drive, 13
of, 48-9, 64-9 Electron capture decay, effects of,
Core polarization, 182 222-7
Corrosion, study of, 206-11 Electron hopping, 145-7
Cosine-effect, 15 Electron spin resonance, 23
Coulombic interaction, 22 Electronic relaxation, effect on
Cross-section for resonant absorp- hyperfine field, III
tion, 10 Electrostatic potential, 31
Cryogenics, 15 Energies of nuclear and chemical
Cryostats, 15 interactions, 4
Curie temperature, 112, 169 Energy of Mossbauer 'Y-ray, 16
Curve-fitting, 20 Europium (lSIEu), 18, 130
electron hopping in EU3S4, 145
Debye model for lattice vibrations, impurity in ice, 148
8, 139-41 in europium iron garnets, 177-8
Debye temperature, 8, 17, 139 in inorganic compounds, 130
Decay schemes (for S7Co, 119Sn, range of chemical isomer shifts,
1291, 193Ir), 17-8 77,130
Detection system, 14 Exchange interactions, 168-79
Diffusion in solids, 145-8 and nonstoichiometry, 169-74
Dipolar contribution to internal in rare-earth iron garnets, 174-9
field, 110 in spinels, 168-74
Disorder in alloys, 183- 8, 195-6
Disorder caused by shock, 206 Fermi contact interaction, 110
Disproportionation in FeO, 167 Ferredoxins, 241-4
Distortion, determination of, from Focussing ~-spectrometer, 209
quadrupole splittings, 104-7
Doppler broadening, 5 - 7 Gaussian distribution,S
Doppler energy, 4 Geometrical isomerism, 54-65
Doppler scanning, 9 Gold (197 Au), 18
Doppler velocity, 9-12 impurity in the transition metals,
Dysprosium (lsIDy), 18 217-8
paramagnetic relaxation in, in gold-manganese alloys, 189-94
150-2 in inorganic compounds, 127-9
Goldanskii-Karyagin effect, 43,
Effective thickness, 11
143-4
Einstein solid, 6, 138-40
Einstein temperature, 138
Electric dipole (E I) transition, 40 Haemoproteins, 236-41
Electric field gradient tensor, 31-6 Half-life of excited state, 3, 16
asymmetry parameter, 31 Harmonic oscillator approximation, 8
250 Principles of Moss bauer Spectroscopy
Heisenberg uncertainty principle, 2 Iron (57Fe)-cont.
Hyperfine interactions, 22-43 high-spin iron(III) compounds,
109,112-3
impurity diffusion in ice, 147
Instrumentation, 13-6 impurity in CdF 2, 214-6
Intensity of absorption lines, impurity in MnF 2, 216-7
38-43
impurity in MnS2, 122
Interferometer for calibration, 15
impurity studies (see also under
Intermetallic compounds, 188-96
cobalt)
Internal conversion, 9-11
in FeF 2, 115-7
Internal conversion coefficient,
in FeF 3, 112-3
11, 16-18
Internal field, 30 in FeO, 166-7
in FeTi(S04h, 198
Internal standard, 15
Iodine (1271, 1291), 18,67-9,77-84 intermetallic compounds, 188-9,
. after-effects of {3-decay, 229 195-6
intramolecular antiferromagnetism,
calibration of chemical isomer
shift, 77-80 154
effect of covalency, 77-84 ion-exchange resins, 211
iron alloys, 184- 7
effect of i.-character, 83
iron(II) halides, 99-100
in frozen solutions of iodine, 80
iron(II) low-spin compounds, 97
in halide derivatives, 67-8
iron phthalocyanine, 109, 119
in inorganic iodides, 82-3
isocyanide complexes, 54
in iodomethanes, 82
lunar materials, 205-6
in Na3H2I06, 26 matrix isolation, 220-3
in solid iodine, 80
N, N-dialkyldithiocarbamate
Ion-exchange resins, 211
derivative, 118
Iridium (1 93Ir), 18
paramagnetic relaxation in,
after-effects of {3-decay, 230 148-9
after-effects of (n, 'Y) reaction, perovskites, 165,198-9
232 pigments, 199-200
impurity in the transition metals, ranges of chemical isomer shifts,
217-8
76
in inorganic compounds, 126- 7 rare-earth iron garnets, 174-8
Iron (57Fe), 18
relaxation in Fe(acac hCl, 152-3
adsorption of gases, 211 relaxation in FeCl3 . 6H20, 153
carbony1s and derivatives, relaxation in haemin derivatives,
46-53,60-3 153
cyanide complexes, 56-60, relaxation of impurity in Al20 3,
93-6,117,140-3 155
cyc100ctatetraene compound, silicates, 201-6
60-1 single crystal of Cr alloy, 186
effect of pressure on, 122-4 spin cross-over compounds,
electron hopping in Fe304, 118-21
145-6 spinel oxides, 160-74
ferredoxins, 241-4 superparamagnetism in fine
ferrocene, 98-100 particles, 156-7
frozen solutions, 147-8 surface layers, 206-11
haemin derivatives, 237 tetrahedral iron(II) compounds,
haemoglobin derivatives, 238-40 107-8
high-spin iron(II) compounds, Isomer shift (see chemical isomer
102-9 shift)
Index 251
Isomeric transition, effects of, 227-9 Magnetic hyperfine splitting- cant.
Isomerization, 60 origins of internal field, 109-11
Isotopic enrichment, 17-9 sign of internal field, 30
selective, 57 temperature dependence, 112-7
Magnetic moment, 28
lahn-Teller distortion in spinels, 163 Magnetic perturbation method for
sign of e2qQ, 38
Kramers' doublets, 32 in ferrocene, 98
in [(1T-C sH s )Fe(COhhSnCI 2,
Larmor precession frequency, ISO 91-3
Lattice vibrations, 5-8 in iron carbonyl derivatives,
Line-intensities, 38-43 49-50
angular dependence, 40-1 in iron cyanide derivatives, 54
effect of polarization, 43 in iron(I) nitrosyl complex, 118
effect of texture, 43 in organotin compounds, 64-7
effect of thickness, 43 in paramagnetic iron (II) com-
Gold an skii-Karyagin effect, 43 pounds, 118
Lineshape, 10 in tin(II) compounds, 87
Linkage isomerism, 56 Magnetically induced distortion,
Lorentzian distribution, 10, 20 164
Matrix isolation, 220-1
Magnetic dipole interaction, 22 Mean-square vibrational amplitude,
Magnetic dipole (M I ) transition, 40 7,135-40
Magnetic flux density, 28 Mean-square vibrational displace-
Magnetic hyperfine splitting, 23, ment, 135-43
28-30,109-18 Molecular field theory, 172
combined with quadrupole Molecular orbital shceme for ferro-
interaction, 36-8 cene,99
dipolar contribution, 110-7 Mossbauer effect, 5-8
Fermi contact term, 110-7 Mossbauer spectrum, 9-12
in dysprosium compounds, 150-2 Multichannel analyser, 13
in erbium compounds, 152
in europium compounds, 130 Near-neighbour effects in alloys,
in FeF 2, 115-7 184-7,192-4
in FeF 3, 112-3 Neel temperature, 112, 169
in gold-manganese alloys, 189-93 Neptunium (237N p), 18, 131-3
in iridium compounds, 125-6 paramagnetic relaxation in,
in iron alloys, 184-7 131-3
in iron(II) halides, 117 range of chemical isomer shifts,
in iron oxides, 168-78 131
in K3Fe(CN)6, 117 Nickel (61 Ni), 18
in Laves phases, 189 in nickel-palladium alloys, 187-8
in neptunium compounds, 132-3 Non-resonant absorption, II
in nickel-palladium alloys, Non-stoichiometry, 159, 165-8
188-9 in FeO, 166-8
in samarium compounds, 152 in 'Y-Fe203, 165
in 119Sn , 29 in FeTi(S04h, 198
Influence of electronic relaxation in silicates, 201-6
time on, 111, 131-3, in SrRu03-SrFe03, 198-9
148-57 influence on magnetic exchange,
line intensities, 38-43 169-79
orbital contribution, 110-7 site occupancy in spinels, 169
252 Pr"mClp es 0
. 11M" ossbauer Spectroscopy
Nuclear magnet?n (unit) defined, 28 Quadrupole splitting-cant,
Nuclear magnetic resonance, 23 for a~-+2.transition 33-4
Nuclear parameters for selected 2 2 '
transitions, 18 in cis-/trans-isomers, 54-5
Nuclear quadrupole resonance in europium compounds, 130
23, 30 ' in Fe dimers in krypton, 220-1
Nuclear radius, change in 24-5 in ferredoxins, 241-2
Nuclear resonant absorption 2 in ferrocene, 98-100
Nuclear resonant fluorescen~e 2 in gold compounds, 127-9
Nuclear spin states, 1 7 - 8 ' in haemoproteins, 237-9
in iodine compounds, 67-9,
77-84
Orbital contribution to internal
field, 110 in iridium compounds, 126-7
Oxidation state, determination of in iron carbonyls and derivatives
46,56-60,74-7,85, 199 ' 46-53, 55,60-3 '
202-3 ' in iron(III) compounds 108-9
Oxygen deficiency in SrFe03, in iron cyanide comple~es,
165, 198-9 56-60,93-6
in iron(II) isocyanide complexes
54 '
Paramagnetic relaxation, 148-56
in matrix isolated iron atoms
in dysprosium compounds 150-2 220-4 '
in haemoproteins, 238-41'
in neptunium compounds, 131- 3
in iron compounds, 148-9
in organo-tin compounds,
in samarium compounds 152
Parity, 17-8 ' 54-5,63-7,89-93
in paramagnetic iron compounds
Partial quadrupole splitting, 96-8 102-9 '
Phonon energies, 6
in rare-earth iron garnets, 176
Photosynthesis, 240-4
in ruthenium compounds, 124-5
Pigment analysis, 199
in silicates, 201-6
Planktonic ferredoxins, 243-4
in spin cross-over compounds,
Poisson distribution, 20 119-21
Polarized absorber, 41 in tellurium compounds, 84
Pressure, effect of, 121-3 in tetrahedral iron(I1) com-
Pressure induced phase transitions
121-3 ' pounds, 107-8
in tin(II) compounds, 87-8
Principal values of electric field
in xenon compounds, 229-31
gradient tensor, 31
line intensities, 38-43
Protoporphyrin IX, 236
magnitude of q and 1'/, 35
Prussian blue, 58
partial quadrupole splitting,
Pyroxenes, 203-6
96-8
selection rules, 33
Quadrupole moment, 30
Quadrupole splitting, 30-6
at impurity sites in CdF 2 , 216 Rate ~f growth of surface layer, 209
combined with magnetic inter- Rayleigh scattering, 16
actions, 36-8 Recoil-energy, 3-4
determining sign of 35, 38-9 Recoil momentum 4
effect of distortions, 104-6 Recoil-free fractio~, 17, 135-143
effect of pressure, 122 Recoilless fraction, 17, 135-143
effect of spin-orbit coupling anisotropy in, 43, 141-4
104-6 ' in cyanide complexes of iron,
t
for a 1-+ transition, 33-4 140-1
Index 253
Recoilless fraction-cont. Spinel structure- cont.
in iron metal, 140 of 'Y-Fe203, 165
in sodium nitroprusside, 142 of Fe304, 145
low-temperature anharmonicity of oxides AB 20 4, 160-8
in, 141 of sulphides AB2S4, 164
zero-point motion and, 138 trigonal distortion in, 162-3
Recoiliess resonant absorption, 1 Statistics of gamma-ray counting,
Recoilless transitions, 7 13, 20
Relaxation, paramagnetic, 148-56 Sternheimer antishielding factor, 36
Ruthenium (99Ru), 18 Sternheimer shielding factor, 34
impurity in transition metals, Superconducting magnet, 15
217-8 Superparamagnetism, 156-7
in C02Ru04, 163
in inorganic compounds, 124-5 Tellurium (12STe), 18, 84
in SrRu03-SrFe03, 198-9 halides and oxides, 84
transferred hyperfine field in, 180
Samarium (l49Sm), 18 Thermodynamic equilibrium in
paramagnetic relaxation in, 152 pyroxenes, 204
Saturation with thickness, 11 Thickness broadening, 11
Scattering experiment, 16 Tin (119Sn), 18,54-5,63-7,77,
for corrosion studies, 206-11 86-93
for non-destructive analysis, decay after effects in, 227-9
199-200 Goldanskii-Karyagin effect in, 144
Scenedesmus ferredoxin, 241 in organic derivatives, 54-5,
Second-order Doppler shift, 63-7,89-93
26-8, 135-40 in tin(II) compounds, 87-8
high temperature limit of, 138 in tin(IV) halides, 89
in iron cyanide complexes, in palladium alloys, 183-4
140-1 in [(7T-C sH s )Fe(COhhSnCI2, 91-
zero-point motion in, 138-40 matrix isolation studies, 220-1
Selection rules, 28, 33 range of chemical isomer shifts, 77
Selective isotopic enrichment, 57 transferred hyperfine interactions
s-Electron density at the nucleus, in, 179-80
24-5,74-5 Transferred hyperfine interactions,
Shielding effects on chemical isomer 170, 179-80
shift,25 Transmission integral, 11
Shock, effect on disorder, 206 Transmitted intensity, 12
Silicates, 201-6 Turnbull's blue, 58
Source matrix, 17
Sources for Mossbauer experiments, Velocity calibration, 15
16-9 Velocity scanning, 9
Spectrometer, 13-6 Verwey transition in Fe304, 145-6
Spin canting, 168, 172-3
Spin cross-over, 119-21 Works of art, 199-200
Spin-lattice relaxation, 149
Spin-orbit coupling, 104-9 X-rays from internal conversion, 16-7
Spin-spin relaxation, 150 Xenon (129Xe), 18,229-30
Spinach ferredoxin, 241 after effects of {3-decay, 229-30
Spinel structure, 160-1
exchange interactions in, Yaffet-Kittelordering, 168
168-74
lahn-Teller distortion of, 163 Zero-phonon transitions, 7
Recoilless fraction-cant. Spinel structure- cant.
in iron metal, 140 of ,),-Fe203, 165
in sodium nitroprusside, 142 of Fe304, 145
low-temperature anharmonicity of oxides AB 20 4, 160-8
in, 141 of sulphides AB 2S4, 164
zero-point motion and, 138 trigonal distortion in, 162-3
Recoilless resonant absorption, 1 Statistics of gamma-ray counting,
Recoilless transitions, 7 13, 20
Relaxation, paramagnetic, 148-56 Sternheimer antishielding factor, 36
Ruthenium (99Ru), 18 Sternheimer shielding factor, 34
impurity in transition metals, Superconducting magnet, 15
217-8 Superparamagnetism, 156-7
in C02Ru04, 163
in inorganic compounds, 124-5 Tellurium (125Te), 18, 84
in SrRu03-SrFe03, 198-9 halides and oxides, 84
transferred hyperfine field in, 180
Samarium (149Sm), 18 Thermodynamic equilibrium in
paramagnetic relaxation in, 152 pyroxenes, 204
Saturation with thickness, 11 Thickness broadening, 11
Scattering experiment, 16 Tin (1l9Sn ), 18,54-5,63-7,77,
for corrosion studies, 206-11 86-93
for non-destructive analysis, decay after effects in, 227-9
199-200 Goldanskii-Karyagin effect in, 144
Scenedesmus ferredoxin, 241 in organic derivatives, 54-5,
Second-order Doppler shift, 63-7,89-93
26-8, 135-40 in tin(II) compounds, 87-8
high temperature limit of, 138 in tin(IV) halides, 89
in iron cyanide complexes, in palladium alloys, 183-4
140-1 in [(rr-C sH s )Fe(COhhSnCI2, 91-3
zero-point motion in, 138-40 matrix isolation studies, 220-1
Selection rules, 28, 33 range of chemical isomer shifts, 77
Selective isotopic enrichment, 57 transferred hyperfine interactions
s-Electron density at the nucleus, in, 179-80
24-5,74-5 Transferred hyperfine interactions,
Shielding effects on chemical isomer 170, 179-80
shift, 25 Transmission integral, 11
Shock, effect on disorder, 206 Transmitted intensity, 12
Silicates, 201-6 Turnbull's blue, 58
Source matrix, 17
Sources for Mossbauer experiments, Velocity calibration, 15
16-9 Velocity scanning, 9
Spectrometer, 13-6 Verwey transition in Fe304, 145-6
Spin canting, 168, 172-3
Spin cross-over, 119- 21 Works of art, 199-200
Spin-lattice relaxation, 149
Spin-orbit coupling, 104-9 X-rays from internal conversion, 16-7
Spin-spin relaxation, 150 Xenon (129Xe), 18,229-30
Spinach ferredoxin, 241 after effects of ~-decay, 229-30
Spinel structure, 160-1
exchange interactions in, Yaffet-Kittelordering, 168
168-74
Jahn-Teller distortion of, 163 Zero-phonon transitions, 7

Gibb

Uploaded by

Copyright:

Available Formats

Gibb

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gibb

Uploaded by

Copyright:

Available Formats

STUDIES IN CHEMICAL PHYSICS

Other titles in the series

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

Typeset by Mid-County Press

ISBN 978-0-412-13960-4 ISBN 978-1-4899-3023-1 (eBook)

All rights reserved. No part of

The emergence of M6ssbauer spectroscopy as an important experi-

Leeds, May 1975 T.e. Gibb

1 The Mossbauer Effect 1

4 Electronic Structure and Bonding:

5 Electronic Structure and Bonding:

6 Dynamic Effects 135

7 Oxides and Related Systems 159

9 Analytical Applications 197

10 Impurity and Decay After-effect Studies 213

11 Biological Systems 234

The Mossbauer Effect

The study of recoilless nuclear resonant absorption or fluorescence

where h (=2rrtt) is Planck's constant. T is related to the more fami-

r = 4.562 x 10-I6/t1/2 0.2)

F or a typical nuclear excited-state half-life of t 1/2 = 10- 7 s, r =

Mossbauer 'Y-ray energies (E'Y) 1OL I0 7 kJ mol- 1

1 eV = 1.60219 x 10- 19 J and is equivalent to 96.49 kJ mol- 1

so that the actual energy of the photon emitted is given by

The 'Y-ray is thus deficient in energy by a recoil kinetic energy

MV=M(V+v) +E'Y/c 0.5)

and the recoil momentum is Mv = -E'Y/c. Hence the recoil energy

where k is Boltzmann's constant and T is the absolute temperature.

1.2 The Mossbauer effect

vibrational energy, then the -y-ray energy is still degraded. How-

In a zero-phonon emission, the whole crystal rather than a

1= exp (_ _E-,-;_(x-:::2,--») (1.9)

where <x 2) is the mean-square vibrational amplitude of the nucleus

f= exp {_ 6E~ [~+ (~) 2

which still depends on ER and 00.

Fig. 1.2 A schematic representation of a Mossbauer transmission spectrum

where f is the recoilless fraction of the source and J o(x) is a zero-

Fig. 1.4 A schematic arrangement for a Mossbauer spectrometer.

1.S Mossbauer isotopes

57Fe 14.41 0.192 1/2- 3/2- 8.17 2.17 57Co(EC

*EC =electron capture, p- =beta-decay, IT = isomeric transition, a = alpha-decay

dium. However, these were chosen as the result of experience,

Thus 57Fe has an abundance of only 2.17% in natural iron, but

1.6 Computation of data

This expression can be used as a 'goodness-of-fit' function and

In Chapter I it was shown that the Mossbauer spectrum is a record

2.1 The chemical isomer shift

where EO is the permittivity of a vacuum, r is the radial distance,

where l'Ps(O) 12 is the non-relativistic Schr6dinger wave function at

The value of oR/R is characteristic of each transition, but is typi-

This can then be related quite simply to the measured shift in

where A and B are two different chemical environments of which

v = vo (1 _ ~ cos a) (1 _ ;~) - 1/2 (2.6)

where Vo is the frequency for a stationary nucleus, and v is the ap-

-10 -5 0·0 +5 +10

This gives rise to a shift in the Mossbauer line of

It is obvious that (v 2 ) increases with rise in temperature. Accord-

2.2 Magnetic hyperfme interactions

:J['= -p.' B = -gJiNI . B (2.9)

where JiN is the nuclear magneton (eh/4mnp = 5.04929 x 10- 27 A