BASIC THEORY OF THE MSSBAUER EFFECT

Takele Seda
Western Washington University
Department of Physics and Astronomy
sedat@physics.wwu.edu

July 2003

The Mssbauer effect is the recoilless resonant emission and absorption of rays (quanta of photon radiation) by nuclei of atoms. The energy levels of the
nuclei in the Mssbauer active atoms are sensitive to the electronic and
magnetic fields present at the nuclei which permit to explore various
hyperfine (magnetic dipole, electric quadrupole and electric monopole)
interactions. The unique characteristic feature of the technique is its high
relative energy resolution to probe fine changes in the hyperfine parameters
of the nucleus and its environment; one part in 10

13

(e.g. for the 14.41 keV of

57

Fe it is 510-13) which was difficult before the discovery of the Mssbauer

effect to measure with other techniques. Detailed elaboration of the basic

foundation and applications of the technique can be found in the books by
Greenwood and Gibbs (1971), Wertheim (1964), Gonser (1976), Trautwein et
al. (1978), Thosar et al. (1983), Nasu and Brand (unpublished). In the
following we base our discussions on these books.

1. Resonant emission and absorption of -rays

The Mssbauer system can be characterised as consisting of a -ray photon
and a nucleus. The properties of the nuclear transition and of the -ray
radiation are coupled through the conservation of energy and momentum.
We consider as a before process a nucleus of -ray source in an excited
state with energy Ee and atomic number A (proton number Z and neutron
number N). This nucleus in its excited state emits a -ray by transition to the
ground state (resonant emission). The emitted -ray can be absorbed by
another nucleus originally in the ground state (Fig. 1) with energy Eg if the
nucleus is of the same kind (same atomic number) as in the emitter by the
reverse process called resonant absorption (the after process).

before

Ee

nucleus 1

after
E0 = Ee Eg

-ray

Eg

emission

nucleus 2

absorption

Figure 1: Schematic of nuclear emission and absorption processes between

the excited and ground nuclear states.
The energy difference E0 between the excited Ee and the ground Eg states is
the energy of the -ray (photon) emitted or absorbed during the processes.
Practically, however, this can only happen in special cases. Resonant
absorption occurs only when the energy distributions (spectral lines) for the
emission and absorption processes overlap. The overlap of spectral lines is
determined by the magnitude of the natural line width of the nuclear transition
energy and the recoil energy imparted to the nucleus by the -ray during
emission or absorption.
By conservation of energy, a free nucleus of a solid (originally at rest) with
mass Mn emitting a -ray receives a recoil energy ER given by
ER =

1
M nv2
2

(1)

Where v is the velocity of the nucleus with which it moves in opposite

direction to the emitted photon. By conservation of momentum
Pn = P

Mnv =

(2)

where c is the speed of light and E is the energy of the emitted -ray. From
the above two equations, the recoil energy becomes

ER =

2
1 E
2 M nc2

(3)

A nucleus in the ground state absorbing the emitted -radiation will also
receive the same amount of recoil energy as the emitting nucleus and hence
will move with the same velocity v in opposite direction to the emitter as
shown in Fig. 2
ER

ER
-ray

emission

absorption

Figure 2: Recoil energy ER received by an isolated nucleus up on -ray

emission or absorption.
Therefore, the energy spectral lines for the emission and absorption will be
separated by 2ER and the energy of the -ray is not exactly E0. It is rather
given by E = E 0 E R for the before (emission) process and E = E 0 + E R for
the after (absorption) process. The energy spectral lines for the emission and
absorption do overlap only if the natural line width (determined by the mean
lifetime of the nuclear energy involved through Heisenbergs uncertainty
relation by =

is Plancks constant divided by 2) is greater

; where

than the recoil energy. However, the recoil energy is always greater than by
considerable orders of magnitude (e.g. it is six orders of magnitude greater
than for the 14.4 keV of

57

Fe). Therefore, resonance condition can not be

fulfilled and the spectral lines of the source and absorber can not overlap.
However, if the nucleus (atom) under consideration is not free, but embedded
in a solid of mass Ma, either of the following two may take places (e.g.
Wertheim 1964). The nucleus will dissipate its recoil energy by phonon

creation (heating the lattice) if the recoil energy is greater than the lattice
vibration, or the whole crystal (solid) will receive the recoil energy without
creation of phonons if the recoil energy is less than the phonon energy. For
such cases of systems where the nucleus is bound in a solid the mass of the
nucleus in (3) can be replaced by the mass of the whole solid Ma. Since this
mass is much greater than the corresponding mass of free nucleus (atom)
the recoil energy will be very small as compared to that for a free nucleus.
Hence, a certain fraction of the emitted -ray can be resonantly absorbed
without phonon creation. This zero phonon process of photon emission and
absorption is the core of the Mssbauer effect and makes it such a powerful
practical tool.

2. Probability of resonant transitions

Recoilless emission and absorption of -ray is only observed for elastic
transitions in which the lattice does not change its vibrational state. The
probability of this recoil free emission or absorption is determined by the
Lamb-Mssbauer factor f, and is related to the general features of the lattice
dynamics of the solid under consideration. In the harmonic limit, this recoil
free fraction is given by (Mssbauer and Clauser 1967, Maradudin 1964)

f = exp k 2 x 2

where

(4)

k is the wave vector of the -ray and x 2

is the mean square

displacement of the resonant nucleus in the direction of the -ray

propagation. For a Debye model (e.g. Freeman and Watson 1965), f will take
the following form:
6 ER 1
T
f = exp
+
kB D 4
2D

QD
T

x
dx
e 1

(5)

where kB is Boltzmann constant, T is the temperature in Kelvin and D is the

Debye temperature. Practically the high or low temperature limit (relative to
D) is important. In such cases

ER 3 2T 2
f = exp
+
k B D 2 2D

for T << D

6E RT
kBD

for T

f = exp

D
2

(6a)

(6b)

From (6) it is clear that the higher (lower) is the temperature, the lower
(higher) is f. At absolute zero,

f (T = 0) = exp

3E R
2k B D

(7)

Which means that even at absolute zero, f doesnt reach unity. Comparison
of (4) with (6) implies that the reason for f not to be unity at T = 0 is that x 2
is not exactly zero at T = 0 because of the quantum mechanical zero-point
motion of the nuclei.
A satisfactory Mssbauer resonance (effect) can be expected if the value of
the product nfa is high enough. Here n is the number of resonant nuclei per
unit area of the absorber, a is the relative abundance of the resonant nuclide
and

= 0

2 + 4( E E 0 )

(8)

is the total cross-section of the resonant absorption centred at E0 for a

Lorentzian line form. The maximum cross-section is given by

2 2 I e + 1 1
0 =
2 2I g + 1 + 1

(9)

where Ie and Ig are the nuclear spins for the excited and ground states
respectively. is the coefficient of internal conversion while one photon is
emitted and is the wave length of the -ray.

57

Co

270 days

E.C
56

Fe (d,n)

57

Co

I
5/2

136.5 keV (89 ns)

57

Fe + 0.6 MeV

9%

91%

3/2
1/2

99%

14.4 kev (98 ns)

57

Fe (stable)

(a)

(b)

Figure 3 (a) Production of the 57Co radioactive source for 57Fe Mssbauer
spectroscopy and (b) its decay scheme.
From (6) we can see that ER should be small to get high recoil free fraction
and ER is small if the resonant -ray energy is small. This implies that
resonant effect is hardly observed for high-energy -rays. It is also difficult to
obtain resonance effect for -rays with very small energies as atomic
absorption increases with decreasing -ray energies (e.g. Long et al. 1983).
The internal conversion in (9) should not be high to get maximum resonance
cross-section. All these conditions are relatively best fulfilled for the 14.4 keV
of

57

Fe Mssbauer transition (Fig. 3) than all other Mssbauer active nuclei

studied so far. The radioactive source for this transition is

is produced by (d,n) reaction (Figure 3 (a)) on

57

Co isotope which

56

Fe (Greenwood and Gibbs

1971). Some important parameters for this transition are also given in table 1.

Table 1: Nuclear parameters of the

57

Fe Mssbauer isotope for the 14.4 keV,

3/2 1/2 -ray transitions (Stevens and Stevens 1976).

Parameter
Unit
Value

-20

10

cm

257

__

mm/s

8.2

ER
-3

10 eV

0.192

1.957

t1/2

Ns

97.8

-0.155

0.0902

2.19

3. Modulation of -ray energy

In the foregoing sections we have considered the absorption line to be
Lorentzian centred at Eo. In most cases, however, this does not happen. Due
to hyperfine interactions (see next section) the nuclear energy levels can
shift their centre and/or split into sub-levels. In order to have resonance
conditions fulfilled between the source and absorber, a -ray with variable
energy (within the limit of the shift and splitting of the levels) is required. This
can be achieved by changing the energy of the emitted -ray along the
propagation direction of the -ray by the Doppler effect.
If the source (-ray) is moving with a velocity v relative to the absorber, then
the mean energy of the -ray emitted towards the absorber is
v
E (v ) = E 0 (1 + )
c
The

E v
=
E0 c

(10)

-ray thus emitted by the source and passing through the absorber is

detected behind the absorber as a function of the velocity of the source. The
principle arrangement of a Mssbauer spectrum measurement is shown in
Figure 4 for horizontal transmission geometry. The number of counts N( v )
detected during a certain time interval at a Doppler velocity v is the
Mssbauer spectrum. A normalized experimental spectrum

source
drive

source

absorber

detector
N(v)

(a)
(a)

Rel. transmission

Control
electronics

(b)

-v

+v

Velocity (mm/s)
Figure 4 (a) Schematic arrangement for a Mssbauer spectrum measurement
and (b) an example of the resulting spectrum.
S (v ) =

N ( ) N (v )
N ( )

(11)

is usually used to calculate the magnitude of the Mssbauer resonance effect

(e.g. Spijkirman 1971), where N( ) is the count rate far off resonance.

4. Optimum transmission and absorber thickness

The count rate and resonance effect in a Mssbauer spectrum are functions
of the absorber thickness. Therefore, for reliable analysis of Mssbauer
spectra, determination of the thickness of the absorber is a very important
factor. The usual way of analysing Mssbauer spectra is the approximation of
a thin absorber. This results in an effective line width given by the sum of the
source and absorber contributions (eff = S + A ). For high pressure
Mssbauer measurements in a DAC (see chap. 5), due to the minute sample
area and the attenuation of the resonant -ray flux by the diamonds, the
absorbers are made thick or enriched with

57

Fe resonant atoms. Therefore,

appropriate absorber thickness need to be calculated to avoid line distortions

as a result of saturation effects (e.g. Viergers, 1976) and to obtain reasonable
resonance effect. This has been briefly discussed by Hearne et al. (1994).
The resonance effect of the Mssbauer spectrum depends on the amount of
-rays attenuated by (or transmitted through) the absorber. The function that

determines the attenuation of the -ray is given by the ratio of the intensities
of transmitted radiation I and the incident radiation I 0 as

I
= exp( e x )
I0

(12)

where e is the mass absorption coefficient for the resonant -ray (in units of
cm2/g); is density of the absorber and

is the physical thickness of the

absorber in cm. A transmission value of 30% is normally used as a starting

guideline with background contribution suppressed (see chapter 5) so that
the pulse height spectrum of the resonant -ray is at least 20% above the
background. This depends on the sample composition and isotopic
enrichment of the absorber. The concentration of the natural abundance of
iron in the absorber is determined by de = (mFe/Fe), where mFe is the mass
fraction of Fe in the absorber.

The principal factor that determines the absorber thickness is ta (the effective
absorber thickness). It is given by (e.g. Hearne et al. 1994)

ta =

N0da
A 10 3

where

0 faa

N0da

A 10 3

(13)

is the number of Mssbauer atoms per unit area; N 0 is

Avogadros number; A is atomic mass; d a is absorber thickness in mg/cm

of the Mssbauer element; a is isotopic abundance of the resonant isotope

0 is the cross-section and f a is recoil-free fraction in the absorber. With a

desired value of t a and estimated value of f a from appropriate lattice

dynamics, the amount of absorber required, d a in a given area can be
calculated. Thus, d e and d a are important factors that could be used as
guideline parameters during enrichment to increase the isotopic abundance
of the absorber up to the desired value of transmission.
For a thin absorber approximation t a is taken to be less than unity. In this
case the resonance intensities are rather small, but the lines can be fitted to
Lorentzian curves. At increasing t a , line distortions occur both in relative line
area and line form. Thus fitting spectral lines to Lorentzian curves may not be
appropriate. Experimental spectra are usually analysed by way of corrections
made to the line width with S = A such that the effective line width is
(Gutlich et al. 1978) given as
eff
0

for t a 4

1 + 0.135t a

4
for 04 < t aa 10

1.01 + 0.145t a 0.0025t a2

10

(14)

where 0 is the natural line width.

In cases where it is important to accurately determine site populations or
phase fractions this correction is still unsatisfactory. Therefore, for
quantitative spectral analysis where split spectral overlap is pronounced
(which is the case for most mineral spectra), the so-called transmission
integral (Margulies and Ehrman 1961, Viegers 1976) is used in the analysing
program. This takes into account thickness effects and can be used to
determine the hyperfine parameters and relative abundance of different sites.
In this program it is important to be able to know (determine) a priori both the
recoil free fraction fs and the line width s of the source independent of the
experimental spectrum with unknown parameters. This is done usually using
a standard calibration spectrum or supplied with the source from the
manufacturer.

5. Hyperfine Interactions
With no hyperfine interactions at all, the spectral line would always be
centred at zero energy (velocity) without shift or split as in Figure 4 (b) in the
last section. The presence of nuclear charge and its moments (nuclear
monopole interaction, quadrupole moment and magnetic dipole moment)
bring the nucleus into interactions with its environment (electrons and other
ions). These interactions result in the hyperfine shift and splitting of nuclear
transitions. The energy of the hyperfine interactions generally does not
exceed 10-6 eV so that before Mssbauers discovery it was difficult to
observe directly the hyperfine splitting of nuclear transitions their energy
being ten orders of magnitude higher. The presence of nuclear hyperfine
interactions make the resonance emission and absorption of nuclear
radiation interesting for applications in physics and chemistry since they yield

11

many information regarding the magnetic state and electronic configuration of

the environment of the nucleus of the atom under consideration.
The total hyperfine interaction energy can be given as
H = He + Hm

(15)

where the subscripts denote the electrostatic (due to the surrounding

electrical charges) and the magnetic (due to magnetic moments) interactions
respectively.

These

interactions

can

be

described

by

classical

electrodynamics (e.g. Jackson 1965). The nucleus has an electric charge

distribution (r ) , the external charges (electrons, other ions etc.) surrounding
the nucleus in the solid are described by the potential V (r ) . The electrostatic
interaction energy between the nucleus and the surrounding charges is then
given by (the origin of the co-ordinate system is the centre of mass of the
nucleus):
H e = (r )V (r )d 3 r

(16)

V (r ) can be expanded in a Taylor series at r = 0 :

V ( r ) = V ( 0) +
i

V
X i

1
Xi +
2
r =0

i,k

2V
X i X k

XiXk +

(17)

r =0

Here the deviations from the position r are denoted by X i , i = 1,2,3 for the
three spatial directions. Using a simple notation
V ( 0) = V 0
V
X i

= Vi

(18)

r =0

V
X i X k
2

= Vik
r =0

and the integral over the nuclear charge distribution

12

 (r )d 3 r = Ze

(19)

we will arrive at

H e = ZeV 0 +

(r ) X i d 3r +

Vi
i

1
2

V ik

(r ) X i X k d 3r +

(20)

i,k

In this equation the first term can be neglected as it represents the Coulomb
interaction energy due to point charge at the centre of the nucleus which is
the same in all nuclear states. The second term consists two quantities; the
charge density (r ) and the spatial co-ordinate X i . The parity of the nuclei
is a constant of motion (the nuclear wave functions are either even or odd)
which means that the charge density (being the magnitude of the square of
the wave function) is always even. Since the integrand in the second term of
the above equation is the product of even (charge density) and odd ( X i ) it
vanishes when the integral is taken over the whole space. The only nonvanishing and very important term is the third term (all the other orders higher
than the third term are very small and can be ignored). Therefore, the
electrostatic interaction energy can be given by
1
2

Vik (r ) X i X k d 3 r

(21)

i,k

In (21) we add to and subtract from the quantity under the integral the

homogenous contribution given by the trace 1 r 2 ( r ) . This gives the

ik
3
expression
H

1
6

V ik
i,k

{r

(r )d 3 r +

(3 X

ik

r 2 (r )d 3 r

(22)

This has been separated into a part depending on the nuclear radius r 2 and
charge density (r ) , and another part depending on the nuclear shape
(form) respectively as
HI =

1
6

Vik r 2 (r )d 3 r

(23a)

i ,k

13

and
HQ =

1
6

Vik
i,k

(3 X

X k ik r 2 (r )d 3 r

(23b)

H I yields the expression for the isomer shift and is given by the trace of H e ,
while H Q yields the expression for the zero-trace electric field gradient (EFG)
tensor which is related to the electric quadrupole moment of the nucleus.

(a) Electric monopole interaction (Isomer shift)

To find exact expression for the interaction energy H I in (23a), we use
Poisson equation 2V = 4 (0) at the nuclear centre. For

Vik this will be

i ,k

Vik = 4 (0) = 4e (0)

(24)

i ,k

( 0) = e ( 0)

Here

is the density of negative charges (electrons)

responsible for Vii at the centre of the nucleus and (0)

is their probability

density.
The average square nuclear radius R 2 can be written in terms of (r ) as
R

(r ) r 2 d 3 r
(r )d r
3

1
(r )r 2 d 3 r
Ze

(25)

Combining (1.24) and (1.25), the electrostatic nuclear monopole interaction

energy due to finite nuclear size is
HI =

2 2
2
e Z ( 0) R 2
3

(26)

This interaction energy describes the electronic monopole (Coulomb)

interaction between the spherically expanded (distributed) nuclear charges
and the electrons surrounding it, which deviates (shifts) the nuclear energy
from the level of a point nucleus.

14

Generally the excited nuclear state has different R 2

from the ground state

and (0) for the source and absorber is also different. This difference (shift)
in energy is the isomer (chemical) shift and is given by (Vrtes etal.1979)

4
R
2
2
Ze 2
Reff2 (0) a (0) s
5
Reff

with R = R e R g and Reff =

Re2 Rg2
2 R

(27)

. Here Re and Rg are nuclear radii for

2

the excited and ground states and (0) a and (0) s are the probability
densities of electrons at the absorber and source respectively.
source
ES

E0

Eg

I = 1/2
absorber
I = 3/2

Ee

Rel. transmission

Ee

I = 3/2

ES +
E0

Eg

I = 1/2

-v

(a)

Velocity (mm/s)

+v

(b)

Figure 5 (a) Schematic of nuclear level shift due to electric monople

interactions in the source and absorber (b) example of the resulting
Mssbauer spectrum.
Classically only s electrons have non-vanishing density at the nuclear site.
However, in some cases (particularly for the 3-d transition metals) an

15

increasing number of 3-d electrons can lead to a shielding of the s electrons

and results in a decrease of the electron density at the nucleus. This in turn
results in an increase or decrease of isomer shift depending on the sign of
the nuclear factor (R/R). When (R/R) is positive, the electron density
increases in going from the source to the absorber if the isomer shift is
positive and when (R/R) is negative, the electron density at the absorber is
lower than at the source if the isomer shift is positive. For example, (R/R) is
negative for the 14.4 keV transition in

57

Fe and an increase in the electron

density is expected if the isomer shift is decreasing at the absorber. A

schematic of the energy level shift in the absorber as well as in the source is
shown in Figure 5(a) and an example of the resulting Mssbauer spectrum in
Figure 5(b)
Since Mssbauer spectroscopy is a relative method (source and absorber are
related by Doppler motion), the isomer shift is usually reported with respect to
some standard absorbers, e.g. metallic iron. From the measured value of the
isomer shift one can obtain information about the chemical environment of
the nucleus under investigation. Most important chemical information
tractable from the isomer shift are oxidation state and electronic spin-state
(high or low) of the Mssbauer active atom in the absorber.
In additional to the isomer shift, there is a contribution to the total centre shift
of a Mssbauer spectrum due to relativistic time dilatation, known as secondorder Doppler shift. The nuclear movement can be split up into the
mechanical motion of the source and thermal vibration of the atoms in the
lattice. The second-order Doppler shift or thermal shift is proportional to the
mean square velocity

v2

of the vibrating atoms and is given by (e.g.

Veigers 1976)

16

E =

v2

2c 2

(28)

The second-order Doppler shift is dependent on temperature and pressure,

but its contribution to the total centre shift is small as compared to the isomer
shift and is usually neglected.

(b) Electric quadrupole interaction (Quadrupole splitting)

In the foregoing section the electrostatic interaction between nuclear and
electric charges at the nuclear site was discussed assuming the nucleus is
spherical and the charge density is uniform where the interaction energy only
shifts the nuclear levels. However, if the nucleus has non-spherical
symmetry, higher order electrostatic interaction energy arises. This
interaction was given in (23b) and it splits the nuclear levels rather than
shifting them.
In (23b), the quantity Vik represents the electric field gradient (EFG) tensor
E because

(3 X

2
i

E i
2V
and the remaining term under the integral will be
= 2
X ii
X2

r 2 (r )d 3 r = eQii

(29)

Here E i is the electric field at the nucleus due to other charges at a distance
X i (i = 1,2,3) from the nuclear centre and Qii is nuclear quadrupole moment.

Therefore, the electrostatic interaction energy H Q =

e
6

Vii Qii =
i

( )

1
.E eQ
6

arises from the interaction of the nuclear quadrupole moment with the electric
field gradient at the nucleus. For a spherical charge distribution the EFG
tensor is zero. However, if the nuclear charge distribution deviates from
spherical symmetry the EFG tensor is a measure of this deviation. The value
of the EFG tensor depends upon the choice of the co-ordinate axes. For this

17

reason a standard form has been assigned such that the off diagonal
elements are zero and the diagonal elements are ordered in such away that

V XX VYY V ZZ and V XX + VYY + V ZZ = 0 . Another parameter (the asymmetry

parameter) is conventionally defined as =

V XX VYY
V ZZ

so that the only

independent parameters used to describe the diagonal elements of the EFG

tensor are VZZ and (0 1) .
The origin of the EFG at the nucleus may be the asymmetric distribution of its
own valence electrons outside the inner isotropic closed shells and/or
external charges of the crystal lattice (ligands). All the other electrons
contribute to the EFG indirectly by polarisation of the spherically symmetric
inner electrons. This effect is expressed by means of the Sternheimer
shielding

and

anti-shielding

factors

and

respectively

(Ingals

1962,1964,Travis 1971), thus the EFG and the asymmetry parameter may be
modified as V ZZ = (1 R )Vval + (1 )Vlat and = (1 R ) val + (1 ) lat . Where
the subscripts

and

val

lat

denote the contributions from the nucleus own

valence electrons and other charges from the crystal lattice (ligands)
respectively.
The interaction of the nuclear electric quadrupole moment with the principal
component of the diagonalized EFG tensor VZZ at the nuclear site splits the

(2 I + 1)

fold degenerate nuclear state with nuclear quantum number I >

into sub-levels. Quantum mechanically the eigen values of these sub-levels

are given by (e.g. Gutlich et al., Gonser 1975).

EQ =

eQV ZZ

3 m I I ( I + 1) 1 +
4 I ( 2 I 1)
3

18

1
2

(30)

where mI = I , I + 1,

, I 1 , and I is the nuclear magnetic spin quantum

number. A pure electric quadrupole interaction splits the excited state of the
14.4

keV

57

Fe

with

2
1
E Q = eQV ZZ 1 +
4
3

I = 3/ 2

into

sub-levels

of

eigen

values

The separation (splitting) between these two levels is the electric quadrupole
splitting and is given by

2
1
E Q = eQV ZZ 1 +
2
3

(31)

This splitting is shown in Fig. 6 along with an example of the resulting

Mssbauer spectrum. Note that (30) contains only the second power of the
magnetic quantum number mI, so states whose mI differs only in sign remain
degenerate (see the values of mI in Fig. 6(a))

mImI

I = 3/2

EQ(3/2)

EQ

EQ(1/2)

1/2

I = 1/2

a11

EQ

Rel. transmission

3/2

1/2

-v

+v

Velocity (mm/s)
(a)

(b)

(a)

Figure 6 (a) Schematic of nuclear level splitting in

57

Fe due to quadrupole

interaction (b) example of the resulting Mssbauer spectrum.

19

For randomly oriented polycrystalline materials, the probabilities for the

individual transitions (e.g. Wertheim 1964) described by the integrals

3
1 + cos 2 (sin )d and
2

5 3
cos 2 (sin )d have the same value,
2 2

thus the lines do not exhibit asymmetry. However, vibrational anisotropy of

the Mssbauer nucleus may give rise to directional anisotropy in the recoil
free fraction (Goldanskii et al. 1963). This in turn may result in relative
intensities of absorption lines that are different from those expected for a
randomly oriented polycrystalline sample. This is because the various nuclear
transition probabilities have an angular dependence on the relationship
between the propagation direction of the -radiation and the symmetry axis of
the electric field gradient.
From the quadrupole splitting one can obtain information about charge
distribution around the nucleus, the local symmetry of the nucleus itself and
the bond property of the solid.

(c) Magnetic dipole interaction (Magnetic hyperfine field)

An atomic nucleus with nuclear spin quantum number I > 0 has a magnetic
dipole moment and may interact with the effective magnetic field B at the
nucleus. This interaction is described by

H m = .B = g N m I .B

(32)

where g N is the nuclear splitting factor (Land factor), m is the nuclear

magneton. The magnetic dipole interaction (nuclear Zeeman effect) splits the
formerly degenerate nuclear state with spin I into (2I+1) non-degenerate substates with eigen values
Em =

Bm I
I

= g N m Bm I

(33)

20

Figure 7(a) shows the splitting of and transitions between the excited I = 3 / 2
and ground I = 1 / 2 states of the 14.4 keV

57

Fe and Figure 7(b) is an example

of the resulting Mssbauer spectrum for purely magnetic Zeeman interaction.

The splitting of the sub-states are determined by the magnitude and sign of

m I whereas the transitions between the levels and hence the number of lines
in the Mssbauer spectrum is determined by the selection rule m = 0,1 .

mI

+3/2
+1/2

I = 3/2
2

Rel. transmission

-1/2
-3/2

3 4

-v
-1/2

I = 1/2

+v

Velocity (mm/s)
(mm/s)
Velocity

+1/2

(a)

(b)

Figure 7 (a) Schematic of the Zeeman splitting for the excited (I = 3/2) and
ground (I = ) nuclear levels and transitions between theses levels in 57Fe (b)
example of the resulting Mssbauer spectrum.

6. Origin of the magnetic hyperfine field

The effective hyperfine field

at the nucleus arises from various

contributions (Thosar et al. 1983, Pelloth et al. 1995). With no external field
applied, B can be related to the major contributors as

B = Bd + Bo + Bc

(34)

21

Bd is the intra-ionic dipolar field, Bo is the orbital filed and Bc is the Fermi
contact field.
The dipolar field is due to the interaction between the spin magnetic moment
of the valence shell of the parent atom (spin part of the atomic magnetic
moment) and the nuclear magnetic moment. It is given by

B d = 2 B

( )

3r S.r r 2 S
r5

(35)

where r is the electron position from the centre of the nucleus, B =

eh
4mc

( m electron mass) is Bohr magneton and S is the spin operator. The angular
average (expectation value) is over the occupied atomic orbital. Here the
spin-orbit coupling is considered to be weak which is true for transition metal
ions. If the spin and orbital moments of the valence shell are not separable S
in (35) is replaced by J = S + L , where J is the total angular momentum and
L is the orbital angular momentum respectively. The magnitude of the dipolar

field depends on the distribution of electrons of the atom concerned. A

spherically symmetric charge distribution (whose angular average is zero) will
not cause dipolar field at the nucleus. Therefore, for an atom whose valence
shell is either completely filled or half-filled the dipolar field is zero (e.g.
Fe3+:3d5). Even for non-spherical charge distribution the dipolar contribution
to the hyperfine field is small.
The orbital motion of electrons in the valence shell produces orbital angular
momentum L that induces the orbital magnetic field Bo . It is given by (Pelloth
et al. 1995)

22

B o = 2 B

L
r3

(36)

Bo is zero for closed shells or half-field shells. For most transition series (Cox
1987) the orbital motion of d-electrons is also quenched by crystal field effect
and Bo is taken to be zero. Both Bd and Bo if different from zero are parallel
to the d-shell magnetic moment and hence contribute positive values to the
effective hyperfine field.
The most significant contribution to the effective hyperfine field comes from
the Fermi contact term, which is given by (Vrtes et al. 1983, Watson and
Freeman 1961a)
Bc =

(0)

2
2
8
B g (0) (0)
3

and (0)

(37)

are the spin-up (parallel to the d-shell moment) and

spin-down (anti-parallel to the 3d-shell moment) s-electron densities at the

nucleus respectively. The Fermi contact field results from the polarisation of
s-electrons by the magnetic 3d-ion itself and also by the neighbouring atoms
(Watson and Freeman 1961a). The polarisation effect includes both the core
electrons and conduction electrons. These have been separated as (Pelloth
et al. 1995) core contribution due to local moment of the 3d-shell and
transferred hyperfine field due to the neighbouring ions.
According to Paulis principle (anti-symmetrization of wave functions), the selectrons with their spins parallel to the net unpaired 3d-spin push way from
the 3d-electrons than those with anti-parallel spins due to the (negative)
exchange interaction. This causes the Coulomb repulsive interaction to be
stronger for electrons of anti-parallel spins (since they appear to be closer
together) than those with parallel spins. Therefore, the s-electrons with antiparallel spin are pushed more towards the nucleus. This means that when

23

the s-electrons are distributed inside the 3d-shell, the electron density
2

(0) of the anti-parallel s-electrons is larger than (0) for the parallel selectrons. Thus, the inner shell electrons produce a negative (relative to the
3d-moment) hyperfine field contribution. On the other hand s-electrons which
are outside the 3d-shell produce positive hyperfine field. The sign and
magnitude of the core electron contribution to the contact field depends on
whether the inner or the outer electrons produce greater contribution.
Table 2 shows the calculated hyperfine fields (Watson and Freeman 1961a)
produced by the s-electrons for Fe2+ (3d6) ion. It is clear from the table that
both 1s and 2s-electrons contribute negative hyperfine fields since they totally
lie inside the 3d-shell. The 2s-electrons contribute more hyperfine field than
1s-electrons because the 2s-electrons have greater overlap with the 3delectrons than 1s-electrons and therefore, are more polarised than 1selectrons. The 3s-electrons are even expected to be more polarised. But due
to the fact that they lie partly inside and partly outside the 3d-shell, the
resultant contribution for these electrons is smaller than the contribution by
the 2s-electrons. The total contribution to the hyperfine field by the core
electrons for this particular ion is negative implying that the electrons inside
the 3d-shell predominate.
Table 2: Core electron contribution to the hyperfine field of s-electrons for
Fe2+ (3d6) ion (Watson and Freeman 1961a).
Electron

1s

2s

3s

Total

B(T)

-3

-131

+79

-55

The transferred hyperfine field contribution comes from the spin polarisation
of the conduction electrons by the magnetic moment of the neighbouring
atoms. The polarisation can be through the Ruderman-Kittel-Kasuya-Yosida

24

(RKKY) or through the ordinary exchange interaction (Watson and Freeman

1961b). The RKKY interaction oscillates and decreases with nearest
neighbour atom distance. It produces positive hyperfine field while the
exchange interaction produces negative field. Hence, the transferred
hyperfine field can be positive or negative depending on the type of
interactions.

7. References
Cox P. A., The Electronic Structure and Chemistry of Solids (Oxford
University press, Oxford, 1987) p. 155.
Freeman A. J. and Watson R. E.: in Magnetism II A ed.: Rado G.T. and Suhl
H. (Academic Press, New York, 1965) p. 167
Goldanskii V. I., Makarov E.F. and Khrapov V.V, Phys. Lett. 3 (1963) 344.
Gonser U., Mssbauer Spectroscopy (Springer-Verlag, Berlin, 1975).
Greenwood N.N and Gibb T.C., Mssbauer Spectroscopy (Chapman and
Hall Ltd., London, 1971).
Gtlich P., Link R. and Trautwein A., Mssbauer Spectroscopy and Transition
Metal Chemistry (Springer-verlag, Berlin, 1978).
Hafmeister D.W.: in An Introduction to Mssbauer Spectroscopy ed.: May L.
(Adam Hilger, London, 1971) P. 60.
Ingals R., Phys. Rev. 128 (1962) 1155; Phys. Rev. 133A (1964) 787.
Jackson J. D., Classical Electrodynamics (Academic Press, New York, 1965)
P. 33.
Maradudin A.A., Rev. Mod. Phys. 36 (1964) 417.
Margulies S. and Ehrman J. R., Nucl. Instrum. Methods 12 (1961) 131.
Mssbauer R. L. and Clauser M. J.: in Hyperfine Interactions ed.: Freeman A.
J. and Frankel R.B. (Academic Press, New York, 1967) P. 497.
Pelloth J., Brand R. A., Takele S., Periera de Azevedo M. M., Kleeman W.,
Binek Ch., Kushauer J. and Bertrand D., Phys. Rev. B52 (1995) 15372.

25

Stevens J., G. and Stevens V., E., Mssbauer Effect Data Index (Plenum,
New York 1976) p. 51.
Thosar B. V., Srivastava J. K., Iyenger P. K. and Bhargava S. C., Advances
in Mssbauer Spectroscopy: Application to Physics, Chemistry and Biology
(Elsevier Scientific, Amsterdam, 1983).
Travis J. C.: in An Introduction to Mssbauer Spectroscopy ed.: May L.
(Adam Hilger, London, 1971) P. 75.
Vrtes A. and Nagy D. L., Mssbauer Spectroscopy of Frozen Solutions
(Akadmia Kiado, Budapest, 1990).
Vrtes A., Korecz L. and Burger K., Mssbauer Spectroscopy (Elsevier
Scientific, Amsterdam, 1979).
Viegres T., PhD thesis, (Nijmegen 1976), the Netherlands.
Watson R. E and Freeman A. J., Phys. Rev. 123 (1961a) 2027.
Watson R. E and Freeman A. J., Phys. Rev. Lett. 6 (1961b) 277.
Wertheim G. K., Mssbauer Effect: Principles and Applications (Academic
Press, New York, 1964).

26