PHY312: The Mössbauer Effect

Minh-Thi Nguyen
minhthin@princeton.edu

Abstract

The Mossbauer Effect, which describes the recoil-free emission and absorption of gamma rays in nuclear transitions, allows
us to study nuclear energy levels and transitions. Using Mossbauer spectroscopy for a Cobalt source and stainless-steel
and iron absorbers in different environments, we compute the lower limit for the lifetime of an excited 57 Fe state, examine
isomer shifts and observe the second Doppler Effect, and compute the effective magnetic field of the Fe nucleus.

1. Introduction

Analogous to electron transitions for specific energy

levels, the nucleus also has quantized energy levels and
can make energy level transitions through the emission or
absorption of photons. In nuclear transitions, the conser-
vation of momentum requires a recoil process, in which
the nucleus gains momentum equal and opposite to the Figure 1: Due to characteristics of nuclear lifetimes and energy nat-
emitted gamma ray. Therefore, taking into account this ural line widths compared to recoil energy, resonance can’t occur in
recoil process, absorption requires a gamma ray of energy nuclear transitions: Er < 2Γ (Right). However, if the recoil energy
is reduced relative to the natural line width, resonance can occur at
larger than the transition energy and similarly, emission the intersection of the two curves Er > 2Γ (Left).)
will result in a gamma ray with energy less than the tran-
sition energy. These energy shifts are large relative to the
width of the transition line and the overlap is minimal, recoil energy is less than the phonon energy, the gamma
so a gamma emitted by a nucleus in one transition can’t ray will produce a zero phonon transition, and the entire
be used in another transition process in another nucleus. solid lattice absorbs the recoil momentum, thus allowing
Thus, resonant absorption would not be expected to oc- the gamma ray to carry exactly the energy of transition
cur. The Mössbauer Effect is based on the phenomena and the emission/absorption lines (Figure 1) to overlap
of recoilless emission and resonance absorption of gamma and allow resonance.
rays.

The Mossbauer effect allows us to study nuclear transi-

tion energies and perturbations in the keV an MeV regimes.
1.1. Theory We measure resonance energies by Doppler shifting the
gamma rays around the resonance peak; the radiation
emitted by a source moving with velocity v has energy
The energy of an emitted γ ray will differ from the ∆E = vc Eγ . Varying the velocity and energy shifts, we
E0
transition energy by the recoil energy Er = 2mc 2 , where scan through a range of energies on small scales to study
m is the mass of the atom. For resonance to occur, the in- nuclear energy levels, such as hyperfine splitting and tem-
coming photon must provide energy equal to the transition perature shifts. The shifted gamma rays are absorbed in
energy. In 1957, Mössbauer discovered recoil-free emission proportion to the overlap of the shifted emission line with
and absorption of gamma rays from nuclear transitions: if the absorption line.
the nucleus is embedded in and bound to a solid crystal,
the recoil momentum is transferred to the lattice which in-
duces vibrations, which induces phonons in the solid. The In order to observe the resonance absorption, the gamma
minimum phonon energy that a single nucleus can emit ray energy is changed over a small range using Doppler
is related to the Debye temperature: E = kb ΘD . If the motion. If the source and the absorber are stationary,
Preprint submitted to supervisor March 24, 2020
and the nuclei of the source and absorber are in the same used to move the inner frame. The drive receives a con-
environment, resonance absorption and re-emission takes trol voltage that sets the position of the inner frame, and
place. When the absorber moves, the spectrum of gamma this servo system compares the value of the input control
radiation experiences a Dopper energy shift ∆E = vc Eγ . voltage to the measured voltage obtained from the LVDT
By measuring the resonant energies, we can determine the (linear variable differential transformer), thus generating
line width and mean lifetime of the emitting Fe nucleus, a voltage difference representative of a position error. By
internal magnetic field, and the magnetic moment of Fe. negative feedback, the voice coil aims to reduce the er-
ror voltage and the system outputs a triangular waveform
from −vmax to +vmax .
1.2. Experiment Setup
The γ ray photons that pass through the absorber
are detected with a proportional counter. A proportional
In this experiment, using the Mossbauer effect, we aim counter produces an output voltage proportional to the
to measure the effect of resonance absorption on the trans- incident photon energy, optimized in efficiency for detect-
mission of gamma rays through an absorber, thus measur- ing 14.4 keV photons. When radiation from the source
ing the Zeeman effect and environmental shifts in the Fe- is not absorbed, the counter gives a certain number of
57 isotope by taking the velocity spectra using Iron and counts/second. When Doppler shifting of the radiated en-
Stainless Steel absorbers. Radioactive Co57 serves as the ergy occurs, there is less absorption, and the counter reg-
gamma radiation source for the experiment. Co57 decays isters a decreased number of counts/second. The counts
by electron capture and populates the 136 keV nuclear level on the counter are accumulated repetitively in short time
of Fe57 , which decays after 10 ns and populates the 14.4 intervals, each corresponding to a particular velocity of the
keV level, which has a halflife of 100 ns. Most of these ex- absorber.
cited Fe nuclei decays to the ground state with the emission
of 14.4 keV gamma rays without recoil. Thus, absorbers
containing Fe57 can have recoilless resonance absorption. A Single Channel Analyzer (SCA) is used to filter and
select the 14.4 keV pulses by selecting pulses with peak
heights (Vpeak ) in a specific range (Vbase - Vbase +Vwindow ),
where the baseline voltage Vbase determines the lowest
peak height that is accepted set by the computer and
Vwindow is set by the SCA.

1.2.2. Measurement Procedure

The nuclei of the Fe57 from different compounds expe-

riences different electric and magnetic fields in the lattice,
Figure 2: Schematic of Moossbauer spectrometer and apparatus. so these fields interact with the nuclei to produce different
The DAQ controls the voice coil drive and counts events from the splitting and shift behaviors.
proportional counter. The right diagram shows the bottle that con-
tains the Cobalt source.)
We perform three primary measurements. We first
measure the isomer shift using a stainless steel absorber.
1.2.1. Apparatus The Mossbauer spectrum, because steel is non-magnetic
and thus the Fe57 experiences no magnetic field, displays
a single absorption dip. Nevertheless, because the chemi-
The Cobalt source is housed in a brass vacuum vessel.
cal properties of the nuclei in the source and absorber are
The absorber (the atom/molecule absorbing the gamma
different, the dip is displaced and we aim to observe the
ray radiation) is placed between the source and the propor-
isomer shift as a result of electric monopole interaction.
tional counter. Gamma rays emitted by the Cobalt source
We perform the same measurement with the source cooled
pass through the absorber and the transmitted gamma
to liquid nitrogen temperature to observe the second-order
rays are detected by the proportional counter. The ab-
Dopper effect and another shift in the absorption dip.
sorber consists of metallic foils that are attached to a mov-
ing frame relative to the Cobalt source controlled by a
drive system. The inner frame of the drive system can In addition to the steel absorber, we also perform the
move in the direction of the γ rays and a voice coil is experiment using both natural and enriched Fe absorbers
to observe magnetic dipole interaction and thus the Zee-
2
man splitting of Mossbauer transitions. From selection
rules, the 14.4 keV level is split into four components and
the ground state is split into two components, resulting in
a total of six expected splits in the absorption line. To
avoid excessive interference from the 6.5 keV gamma rays,
we use a Chromium foil to select only the 14.4 keV radia-
tion.

2. Results and Analysis

2.1. Experimental Setup for Cobalt Spectrum

We first want to scan and record the nuclear spectrum

of the Mössbauer source, Cobalt 57, to identify its nuclear
transitions. Because we are interested in only the 14.4 keV
transition, we first identify the 14.4 keV line from the Co57
source to calibrate the SCA to maximize the number of
14.4 keV events observed. We do not want the Mössbauer
absorption overwhelmed by the detection of non-resonant
radiation energies. This setup of selecting the 14.4 keV γ
rays allows us too obtain higher count rates for the desired
experiments. Figure 3: Top: First initial SCA sweep of the Cobalt spectrum,
displaying both the observed first 6.5 keV peak and the second 14.4
keV peak. Middle: Applied Chromium absorber to absorb most
of the 6.5 keV gamma rays, leaving the observed 14.4 keV peak.
Leaving the path to the detector clear from our ra- Bottom: Enhanced and amplified peak after altering amplification
dioactive source, we use the SCA to sweep voltages from gain parameters from the Middle figure.
0V to 8V to obtain the pulse height spectrum for Co57 .

range to cover 99.7% of the 14.4 keV radiation, thus set-

After our first initial sweep (Figure 3), we identify two ting the baseline voltage to be µ = 2.4 V with a total
peaks, with the second weaker peak corresponding too the window width of 6σ = 6 ∗ 0.4 = 2.4 V. With modeled
14.4 keV radiation. The first peak corresponds to the 6.4 channel count uncertainty using Poisson statistics, we ob-
keV K-shell X-rays from ionization of the electrons of the tain a reduced χ2 = 0.832 for our fitted Gaussian to the
Fe atoms by photoelectric absorption. In this case, the peak shape, indicating a relatively appropriate fit to the
atom is left in an excited state with a hole in the K-shell, curve.
thus the X-ray is produced (Auger electrons emitted). To
identify the 14.4 keV peak, we repeat the procedure with
an attached Chromium absorber to absorb most of the 6.5
keV gamma rays because this causes the electrons to be 2.2. Mössbauer Spectrum
knocked out of their orbital, which eliminates the 6.4 keV
X-rays, thus leaving a distinct 14.4 keV peak around 2.4 V.
We repeat the same procedure again with adjustments to 2.2.1. Stainless Steel - Room Temperature
the amplifier gain to maximize the 14.4 keV signal, which
does not change the peak center significantly, but does
We place a stainless steel absorber between the source
enhance the amplitude of the peak.
and the proportional counter to observe the Mössbauer ef-
fect. Because the stainless steel absorber is a paramagnetic
From this last spectral scan, we fit a Gaussian shape to material with no magnetic field, as well as a cubic lattice
the peak to select an appropriate window width and base- structure without an electric field gradient, and does not
line voltage that will maximize the signal-to-noise ratio for result in Zeeman splitting, we observe a a single absorption
our experiments to follow. Based on our fitted model for peak and a very simple Mossbauer spectrum.
the shape of the peak (Gaussian with µ = 2.38 ± 0.01,
σ = 0.389 ± 0.01), we select the total accepted voltage 57
The mass of the Co nucleus differs from that of the
3
 χ2G = 0.43 for the Gaussian fit than the χ2L = 0.60 for the
Lorentzian fit, we see that the Lorentzian profile is a more
accurate fit of our obtained data results.

We determine the width of the recoilless emission and

absorption lines of the 14.4 keV transition in 57 Fe. The
Lorentzian fit gives us a velocity dip center at v = 0.21 ±
0.003 mm/s and width Γv = 0.20 ± 0.004 mm/s, which
represents an energy profile width of ΓE = Eγ ( Γcv ) =
9.60 × 10−9 ± 1.92 × 10−10 eV, resulting in a lower limit of
the lifetime of the 3/2 Fe excited state: τ = 2Γ~E = 34.3
± 0.6 ns. We compare this to the typical lifetime of an
excited nuclear state of iron 100 ns.
Figure 4: Cobalt spectrum with Chromium absorber with fit for the
14.4 keV peak to select desired window width and baseline voltage.
Due to the thickness of the source and absorber, as
well as the limits of the velocity resolution of the drive,
57
Fe nucleus, so the resonant energies of their 14.4 keV impurities in the sample, and other environmental effects,
transitions are different. Thus, we expect that the reso- our resonance absorption line is broadened compared to
nant energy of the 57 Fe nucleus to be shifted compared the ideal, narrow resonance line of the the excited state of
to the same energy of the resonant emission of the 57 Co Fe-57, thus resulting in a shorter computed mean lifetime
nucleus. than the expected theoretical value. We mainly see that
we obtain a lower limit because the ideal case and theo-
retical value assumes the limit of a thin source and thick
absorber so that the observed line width is exactly twice
the natural line width of the source and absorber. Nev-
ertheless, physical processes can broaden the linewidth of
the transition. By the Heisenberg principle, the linewidth
of the radiation and the lifetime of the excited state is
related by ∆E∆t ≥ ~ = Γτ , so τ gives us the lower bound.

We compare our obtained value with the given lifetime

of the excited 57 Fe state of τ0 = 97.8 ns. Thus, we would
expect a natural line width of 6.73 × 10−9 eV, which cor-
responds to ∆vγ = 0.1401 mm/s. Thus, the line width of
a resonant event is satisfied by the Uncertainty Principle:
∆E ≥ Γ2 = 0.0701 mm/s (compare this to our obtained
Figure 5: Mössbauer Spectrum data collected for Stainless Steel ab-
value of the resonant linewidth).
sorber fitted against a Lorentzian profile. The red dashed line in-
dicates the center of the profile, indicating the shift from 0 velocity
(blue line). The error in counts are obtained from Poisson counting Furthermore, we observe a shift from zero velocity of
statistics. δv = 0.21 ± 0.003 mm/s and thus a corresponding energy
shift of δ = 1.01 × 10−8 ± 1.8 × 10−10 eV due to an isomer
Our data collected for the stainless steel absorber is shift in the different chemical environments of the absorber
shown in Figure 5. We fit our data against a Lorentzian and source; the iron nuclei created in the decayed Co-57
profile. Although the absorption spectrum is represented source is different than the Fe-57 nuclei in stainless steel.
by a Lorentzian fit, if there exists errors in the equipment
and apparatus that are explicitly accounted for that would
result in a wider spread of the absorption spectrum, the
data would be better be represented and fitted by a Gaus- 2.2.2. Stainless Steel Absorber at Low-Temperature
sian curve. A wider spread of the absorption spectrum can
be a result of the thickness of the absorber that can affect
We wish to study and show the temperature depen-
the fraction of non-recoilless emission. Thus, we fit both a
dence on the center shift of the Mossbauer spectrum (as a
Lorentzian and Gaussian profile to our obtained spectral
result of the second-order Doppler shift) in the case of the
data to determine the accuracy of our experimental equip-
stainless steel absorber for simplicity. In this experiment,
ment and procedure; because we obtain a higher reduced
the source was cooled by liquid-nitrogen to a temperature
4
of 77.6 ± 0.1 K. By the second order Doppler shift, we tial shift in the velocity peak to be δv0 = − 3k
M c (∆T ) =
B

know that by time dilation, the frequency of a wave ra- 0.158 ± 0.01 mm/s with ∆T = (294.3 − 77.6) ± 0.2
diated from the source (moving relative to the absorber K. Our observed velocity shift from our data results is
in the center of mass frame) p moving at a speed v is al- δv = 0.127 ± 0.005 mm/s. This yields a −19.6% error
tered by a factor of γ = 1 − v 2 /c2 . Thus, as a result from the theoretical upper limit.
of the Second Doppler Shift, knowing that the velocity
increases with increasing temprature (the thermal mean
square temperature < v 2 >T ≈ kT 2 From our results, we can obtain the fractions of 14.4
m ; < v >T = 0 for the
first order Doppler Effect) and decreases with decreasing keV photons emitted without recoil at each of the two tem-
temperature, we see that at a lower source temperature, peratures. The recoilless fraction f describes the fraction
the observed frequency shifts downward less, which results of gamma rays emitted without recoil, which describes the
in a larger velocity registered by the detector, which results probability that the γ ray will be emitted with no energy
in a larger velocity and thus energy shift in the resonant loss due to phonons. Taking the model of a Debye solid,
absorption line with respect to zero velocity. Therefore, we see that we can describe f as a function of temperature:
we anticipate a larger shift in the velocity peak. Eγ 2 mv2
<x2 > − 2k
f = e−( ~c ) =e B T
(1)

The observed (not completely corrected) recoilless frac-

tion can be computed from our spectrum can be deter-
mined by:
βI∞ − I0
f= (2)
βI∞ − δIB
where I0 is the resonance count rate at emission-absorption
line overlap, I∞ is the off-resonance count rate, IB is the
background count rate and the corrections β = 1.008 and
δ = 1.02 obtained by Nussbaum et al.

Therefore, assuming that IB = 0 (our background count

rate can be neglected), from our Lorentzian fits for both
Figure 6: Mössbauer Spectrum data collected for Stainless Steel ab- the room temperature (294.3 K) and low temperature (77.6
sorber cooled to low-temperature conditions at 77.6 K by Liquid K), we obtain:
Nitrogen, fitted against a Lorentzian profile. The red dashed line in-
dicates the center of the profile, indicating the shift from 0 velocity
(blue line). The error in counts are obtained from Poisson counting
f77.6 = 0.983 ± 0.01 (3)
statistics.
f294.3 = 0.973 ± 0.01 (4)
Figure 6 displays the data for the Mossbauer spec-
trum obtained from the low temperature Stainless Steel
absorber at 77.6 ± 0.1 K. While the velocity width (and The data given by Nussbaum et al. for the Cobalt
thus energy width for the transition) remain at ∆v = source diffused in Palladium suggests that at room temper-
0.20 ± 0.005 mm/s, the center of the peak was shifted ature the observed fraction of recoil-free transitions should
by δv = 0.342 ± 0.003 mm/s. We know that at room be f = 0.61 ± 0.01 and f = 0.830 at 78 K (low temper-
temperature of 294.3 ± 0.1 K, the spectrum was shifted by atures). We observe that these results are fairly incon-
v = 0.21 ± 0.003 mm/s. We see that this temperature shift sistent with our data, mainly because we assumed that
does not change the form of the absorption spectrum and there was no background intensity in our calculation. The
only provides a contribution to the shift in the spectrum fractional difference between the two observed ratios is:
in relation to zero velocity. f77.6 /f294.3 = 1.01, compared to the predicted value of
1.36.
Without the quantum effects on the atomic vibrations,
by the Debye approximation, we anticipate the upper limit We did not control the number of counts for both tri-
for the differential shift in the absorption minimum to be als, which influences the computation of the recoil-free
d ∆ET 3kB
a linear relationship: dT Eγ = − M c2 where ET is the fraction, as well as the standard error in the data ob-
Second-Doppler, temperature dependent shift, M is the tained for each sample; the low temperature sample had
molar mass of the isotope, and kB is the Boltzmann Con- a I∞ = 236000, nearly twice the number of counts ob-
stant. Therefore, we set an upper limit to the differen- tained for the room temperature sample I∞ = 150000.

5
Our source was also diffused in Palladium foil. We know
that the recoil-free processes happen when the recoil mo-
mentum is taken up by the entire lattice, so the fraction
depends on the structure of the lattice. Thus, the frac-
tion of recoil-free emission/absorption is larger for tightly-
bound atoms. Nevertheless, because the source has Pal-
ladium impurities, there is less binding of the Fe atoms,
thus making the fraction less accurate. From Nussbaum,
we see that these impurities can be corrected with various
correction factors that are based on characteristics of the
thickness of the absorber. We also observe that other im-
purities such as Cu would result in more tightly binding
Fe atoms, thus giving us a more accurate fraction.

2.2.3. Magnetic Natural and Enriched Iron Absorbers

When a nucleus is placed in a magnetic field, its energy

levels are split by hyperfine interactions from the Zeeman
effect. For the 57 Fe isotope from the magnetic iron ab-
sorber, as a result of the interaction of the nuclear mag-
netic dipole moment with the magnetic field due to the
absorber atom’s own electrons (naturally occuring strong
magnetic fields within the sample), we observe resonances
in the Mossbauer spectrum as a result of magnetic split-
ting of atomic levels. For the 14.4 keV transition, it is a
transition corresponding to the nuclear spin states I= 1/2
(ground) and I=3/2 (excited), and both states are present
in the 57 Fe nucleus. In the presence of the magnetic field,
both states experience splitting of the energy levels, and Figure 7: Mössbauer Spectrum data collected for Pure (top) and
we know that the particular energy of transition for a par- Enriched (bottom) Iron isotope samples absorber fitted against a
Lorentzian profile. The green dashed line indicates the center of the
ticular quantum number mj is shifted by: profile, indicating the shift from 0 velocity (blue line). The error in
counts are obtained from Poisson counting statistics. The fits have
∆E = −µHmj /I (5) a reduced χ2 value of 0.129 (top) and 0.220 (bottom)

∆E ∗ = −µ∗ Hm∗j /I ∗ (6)

Where H is the magnetic field of the nucleus, µ = 0.0902 transitions relative differences and their uncertainties are
is the nuclear magneton’s magnetic moment of the ground shown in Table 1. From these relative separations, we can
state of 57 Fe and µ∗ = −0.1546 is the magnetic moment determine the splitting of the ground and excited states,
of the excited state. By selection rules ∆mj = 0, ±1, we thus allowing us to determine the effective magnetic field
know that mj ranges from −ItoI, so the possible values for of the nucleus.
mj are −3/2, −1/2, 1/2, 3/2 for I = 3/2 and the possible
values for mj are −1/2, 1/2 for I = 1/2. Thus, the tran- Table 1: Energy Shifts in Magnetic 57 Fe Sample. The peaks are
sition from I = 3/2 to I = 1/2 gets split into six possible identified as (format: mj ↔ m∗j ): Peak 1 (1/2 ↔ −3/2), Peak 2
transitions, which we observe in our Mossbauer spectrum (1/2 ↔ −1/2), Peak 3 (1/2 ↔ 1/2), Peak 4 (−1/2 ↔ −1/2), Peak 5
(−1/2 ↔ 1/2), Peak 6 (1/2 ↔ 3/2). We see that the ground state
obtained in Figure 8. We see as well that the separation splitting occurs between Peaks 2 and 4 and Peaks 3 and 5; and
between the ground state energy splitting is α = 2µH excited state splitting occurs between Peaks 1 and 2, Peaks 2 and
(I = 1/2) and the splitting separation between the excited 3, Peaks 4 and 5, and Peaks 5 and 6. We take the weighted average
states is β = 32 µ∗ H (I = 3/2). of these peak separations to determine the ground and excited state
splittings found in the table below.

Sample α (neV) β (neV) H (kG)

We approximate the spectrum with six Lorentzian pro-
Pure Fe 185.65 ± 0.5 105.73 ± 0.3 325.9 ± 0.1
files (using Poisson uncertainties with a reduced χ2 value
Enriched Fe 184.70 ± 0.3 106.18 ± 0.1 325.7 ± 0.1
of 0.129 for the Natural Iron sample and 0.220 for the En-
riched Iron sample) that allows us to obtain the full-width
of the peak to give the natural linewidth. The energy Therefore, we obtain α = 2µH with µ = 0.0902µN and

6
β = 23 µ∗ H with µ∗ = −0.1546µN , thus giving a weighted
average value for the effective magnetic field of the nucleus:
H = 325.8±0.1 kG (Table 1); compared to the expected
result of 330 kG, we obtain a 1.27% error to the theoretical
value.

3. Discussion and Conclusion

Therefore, using the Mossbauer spectrum of various

absorber samples from a 57 Co source, we perform experi-
mental techniques to compute the lifetime of the 57 Fe ex-
cited state, determine isomer energy shifts, observe the
second doppler effect, compute the recoilless emission frac-
tion at different temperatures, and observe the Zeeman
splitting of transitions to compute the effective magnetic
field of the Fe nucleus.

From the non-magnetic, stainless steel absorber at room

temperature, we obtain the lower limit for the lifetime of
the excited Iron state of τ = 34.3 ± 0.6 ns by the Uncer-
tainty relation. Furthermore, we observe an isomer shift
of δE = 1.01×10−8 ±1.8×10−10 eV. At low temperatures,
obtained by cooling the Cobalt source to liquid nitrogen
temperature, we obtain a differential shift in the resonance
velocity peak that yields a −19.6% error from the upper
limit when we ignore quantum effects. Nevertheless, due
to the lack of control and background counts in our data
collection, we were unable to reproduce the fraction of
photons emitted without recoil obtained by Nussaum et
al. From the pure and enriched Iron samples, by deter-
mining the energy separation between resonant peaks, we
determine the effective magnetic field at the nucleus to be
H = 325.8 ± 0.1 kG, giving us a 1.27% error from the
theoretical value.

4. Acknowledgements

I would like to thank Jason Rosenberg for being very

helpful in guiding me through the lab and understand-
ing the physics of the experiment. I would like to thank
Bianca Swindler for being a very cooperative and helpful
lab partner.