THE MÖSSBAUER EFFECT

INTRODUCTION

In this experiment you will observe the Mössbauer effect (the recoil-free emission and
absorption of gamma rays from nuclear transitions) and use Mössbauer spectroscopy to examine
the properties of recoil-free radiation processes and investigate the effects of environmental
factors on the energy levels of the nucleus.

In quantum mechanics, the nucleus can be treated in much the same way as the atom.
Although the details of nuclear binding differ considerably from the Coulomb binding of the
atom, both the atom and the nucleus are multiparticle systems subject to the same sort of
quantum mechanical analysis. In particular, just like the atom, the nucleus has quantized energy
levels. Although the nucleus generally occupies the ground state energy level, it can be excited
to higher levels and make transitions back to lower levels, absorbing or emitting photons in the
process. While atomic (electronic) transitions have energies in the eV to keV range, nuclear
gamma rays have energies on the order of keV to MeV. Reminder: photons that originate from
atomic deexcitations are called x-rays, and those from nuclear deexcitations are γ-rays.

Transitions between nuclear states generally occur by radiation processes involving

recoil. In recoil emission, conservation of momentum requires the nucleus to carry away
momentum equal and opposite to that of the emitted gamma ray. The kinetic energy of the
nucleus necessarily takes up some of the energy available from the transition. As a result, there
is less energy left for the gamma, and its energy is lower than the energy released by the
transition. Conversely, absorption with recoil requires a gamma ray of energy greater than the
energy of the transition, as some of the energy is again taken up by the kinetic energy of the
nucleus. The energy shifts of the gammas in emission and absorption are large compared to the
width of the transition line (determined by the lifetime of the state according to the Heisenberg
relation, ΔEΔt ≥ h ). You should calculate the value of the energy shift for some typical
example(s). Because the overlap is minimal, a gamma emitted by a nucleus in a transition from
state 2 to state 1 generally cannot excite the transition from state 1 to state 2 in another nucleus
of the same element.

B-2-1
 In 1957 Rudolf Mössbauer, a graduate student at the Max-Planck Institut in Heidelberg,
discovered the recoil-free emission and absorption of gamma rays from nuclear transitions, a
discovery for which he was awarded the Nobel Prize three years later at the age of thirty-two.
The essence of the Mössbauer effect is the absorption of recoil momentum by the entire lattice of
the solid. Since the lattice is much more massive than the nucleus, its recoil kinetic energy is
effectively zero. In this case the gamma ray carries away exactly the energy of the transition,
and the emission and absorption lines overlap, centered about the transition energy. Suppose a
nucleus emits a gamma ray (without recoil) of energy Eγ in making the transition from state 2 to
state 1. The overlap between emission and absorption lines allows another nucleus to absorb that
gamma ray in going from state 1 to state 2. Such resonant absorption is the key to Mössbauer
spectroscopy. The probability of recoil free emission/absorption can be calculated using
quantum mechanics.

Given the possibility of recoil-free emission and absorption, one can use resonant
absorption to examine energy profiles of nuclear transitions. The range of energies incident on
an absorber can be precisely shifted by one part in 1013 or so (energies on the order of µeV or
neV) using the Doppler effect: radiation emitted by a source moving with velocity v has an
energy shift ∆E = (v/c)Eγ , where Eγ is the unshifted gamma-ray energy.1 By varying the
velocity, and thus the energy shift, it is possible to scan through a range of energies on the order
of those for natural line widths of nuclear processes, for temperature shifts in the nuclear
transition lines, and for hyperfine structure in the energy levels. The shifted gamma rays are
absorbed in proportion to the overlap of the shifted emission line with the absorption line. Thus
the relative absorption at different energy shifts reveals much information about the details of the
nuclear energy levels. Even more importantly, for solid-state physics, it serves to test lattice
vibrational models predicting the fraction of recoil-free processes, as well as to provide data on
the environment of the nucleus in the lattice -- e.g., internal magnetic fields, Coulomb
interactions with atomic electrons, and quadrupole interactions with field gradients, among many
other effects. The (1) depth of the absorption line(s), (2) width, (3) energy and, (4) the number
of lines observed provide the experimental information. This is summarized in Table 1.2
(Mössbauer parameter and effects) reproduced from Gonser [1975]. An outline of some of the
theory follows, but the references and original papers need to be consulted for explanations and
derivations.

1To gain an appreciation for the minuteness of this Doppler shift, R.V. Pound has given the following analogy.
Imagine a tuning fork attached to the end of the hour hand of a clock. The acoustic Doppler shift due to the motion
of the tuning fork on the hour hand is many orders of magnitude greater than the shifts encountered in the
Mössbauer effect.

B-2-2
B-2-3
THEORY

We begin by considering what determines the depth and width of the Mössbauer absorption line.
This is followed by brief discussions of the various phenomena affecting the absorption line(s).
The (1) isomer shift, (2) nuclear Zeeman effect, and (3) quadrupole splitting are due to hyperfine
interactions, while the (4) temperature shift is a relativistic effect.

(a) Recoil-free fraction: The depth of the Mössbauer resonance is determined by the
fraction of recoil-free emissions in the source and the resonant absorption cross section in the
absorber (as well as photoelectric and Compton scattering). The recoil-free fraction of the
source is defined to be the fraction of gamma rays that is emitted without recoil. Similarly, the
recoil-free fraction of the absorber is the fraction that is absorbed without recoil. Since recoil-
free processes occur when the recoil momentum is taken up by the entire lattice, the fraction
depends on the structure of the lattice. Solid-state physics tells us that lattice vibrations are
quantized, coming only in packets called phonons. Particular theories provide different spectra
of lattice vibrations. Accordingly, the recoil-free fraction depends of the phonon spectrum that a
given theory predicts. The expression for the recoil-free fraction in the Debye theory is called
the Debye-Waller factor, f:
 E 2γ 2
f = exp − 2 x 
 ( hc) 
2
where Eγ is the gamma energy, and x is the mean square displacement of the atom in its
thermal vibrations about its origin (see Frauenfelder, for example); f will only be large for small
values of gamma energy and for tightly bound atoms. The Debye theory of lattice vibrations
gives an estimate for the mean square displacement of the atom and thus f can be evaluated.
However, the finite thickness of the absorber modifies the recoil-free fraction. On one hand, a
thin absorber cannot absorb all the recoil-free gamma rays because there are not enough nuclei in
their path to ensure an interaction. On the other hand, a thick absorber offers more chances for
an incident gamma ray to be absorbed with recoil. The thickness of the absorber thus changes
the observed absorption so that it no longer equals the recoil-free fraction of the source.
Corrections for the thickness can be applied to the measured absorption and then compared to the
recoil-free fraction, which is predicted by the Debye theory. If the values are consistent, one can
consider the Debye theory an adequate description of the lattice for this purpose.

(b) Natural linewidth: Assuming the nuclei in the source are decaying exponentially, the
energy distribution of the emitted gamma rays can be shown to be Lorentzian:

B-2-4
 No
N(E) = .
(E − E γ )2 + (Γ / 2)2
Here Eγ is the central energy of the distribution and Γ = h /τ is the full width at half-maximum
(FWHM) of the emission line (see Greenwood & Gibb). The absorption line has the same
Lorentzian shape with the same linewidth Γ. The shapes of the emission and absorption lines
determine the observed absorption curve. In particular, it is Lorentzian with an observed
linewidth of Γ obs. If the source is infinitely thin, all the recoil-free emitted gamma rays escape
from the source, and no unwanted interactions broaden the emission line. Similarly, if the
absorber is infinitely thin, the least number of incident gamma rays are lost to non-resonant
absorption. In this ideal case, the observed linewidth is simply the sum of the emission and
absorption linewidths, giving Γ obs = Γ + Γ = 2 Γ . However, the ideal case is impractical for
reasons of signal-to-noise ratio. In more realistic situations in which both the source and
absorber thickness is finite, it no longer holds that Γ obs/ Γ = 2. Instead, non-resonant processes
broaden the observed absorption curve, and corrections have been developed by workers in
Mössbauer spectroscopy. Using these correction formulas one can calculate the linewidth Γ of
the absorption and emission processes from the observed FWHM Γ obs of the absorption curve.
By the Heisenberg principle, the linewidth of the radiation and the lifetime of the excited state
are related by ∆E∆t ≥ Γ /τ = h . The linewidth thus allows one to calculate the lifetime of the
state for comparison with the accepted value, 1.4x10-7 sec.

(c) Isomer shift (chemical shift): In many cases it is sufficient to treat the Coulomb
interaction between the atomic electrons and the nucleus as though the nucleus were a point
charge. With this assumption, the electronic wavefunctions would not overlap at all with the
nucleus. However, this is not the case and the s-electrons in particular have a finite probability
of being found within the nuclear volume. The overlap between the electrons and the nucleus
produces a change in energy due to Coulomb interaction, which in turn changes the total energy
of the system.

Suppose that the ground and

excited states of the nucleus have slightly
different radii. Then the overlap between
nucleus and electrons will be different for
the two states. Consequently, the change
in energy will also be different. This
difference shows up as a shift of the peak
away from the resonance energy Eγ. The isomer shift appears in some degree in all samples

B-2-5
because the electrons always overlap to some extent with the nucleus. Thus Mössbauer spectra
are never exactly centered about zero energy shift, or zero velocity.

(d) Zeeman effect (magnetic hyperfine splitting): A nucleus can have a characteristic
angular momentum, J, which is accompanied by a magnetic dipole moment. When the nucleus
is placed in a magnetic field, the energy states are split, with energies depending on the
orientation of the nuclear moment with respect to the field. A state with quantum number J can
split into 2J+1 components characterized by the quantum number mJ. The splitting of the energy
levels is called the nuclear Zeeman effect and is exactly analogous to the Zeeman effect of
atomic physics.

The splitting of the energy states produces a splitting of the line corresponding to the
transition between two states. In Fe-57, we are looking at the transition between the J=1/2 and
J=3/2 levels. The presence of a magnetic field splits the levels as shown.

There are nominally eight possible transitions. Since these are magnetic dipole transitions, the
selection rule ΔmJ = -1, 0, +1 becomes operative. That rules out two transitions, leaving six lines
that are observed. The separations between the levels are given just as in the Zeeman effect:

Δo = goµNH = µoH/Jo
Δ1 = g1µNH = µ1H/J1

B-2-6
where µN is the nuclear magneton and go and g1 are called the g-factors for the J=1/2 and J=3/2
levels. If we measure the energy splittings and know the g-factors, we can calculate the strength
of the magnetic field at the nucleus, a quantity of interest in solid-state physics.

If we assume that the magnetization is randomly directed, the intensities of the lines,
given by a detailed quantum mechanical calculation, come in the ratio 3:2:1:1:2:3. Random
magnetization can sometimes be obtained with thin foils of iron, which are often slightly
magnetized. On the other hand, if a uniformly directed magnetic field is applied, the intensity
ratio is 3:x:1:1:x:3, where x = 4 sin2θ / (1+ cos2 θ). Here θ is the angle between the direction of
magnetization and the direction of radiation. When observing magnetic hyperfine splitting, it is
best to have only one of the source or absorber be magnetic. Otherwise there would be six
emission and six absorption lines and the spectrum would be quite complex. Since we cannot
change the source, we use an absorber of metallic Fe, preferably enriched in Fe-57.

(e) Quadrupole splitting: The nucleus is held in

position in its environment by Coulomb forces. The fact
that it is stationary (neglecting thermal vibrations)
implies that there can be no electric field at the nucleus;
if there were, the nucleus would move to a point of no
field. However, there can be a gradient in the electric
field. A field gradient could arise, for example, from a
non-spherically symmetric distribution of atoms in the
lattice, as shown in the figure. [Calculate the field
gradients at the center of a square (the site of the Fe atom in the figure) with charges +q at the
corners, distance a from the center.]

If a nucleus in a particular spin state has a slight spheroidal deformation, its distribution
of charge is not perfectly spherical, and the nuclear quadrupole moment, Q, is a measure of the
deviation of the nuclear charge from spherical symmetry. If a nucleus has a quadrupole moment,
its energy depends on its orientation with respect to the electric field gradient. The energy of a
level with quantum numbers J and mJ is shifted from its zero-gradient level by

2 3m 2J − J(J + 1)
ΔE = e qQ
4J(2J − 1)

B-2-7
 1 dE
if the field gradient is axially symmetric, where q = measures the field gradient. Note that
e dz
the quadrupole splitting, unlike the magnetic splitting, does not depend on the sign of mJ but
only on its magnitude. This happens because the spheroidally deformed nucleus, unlike a spin or
magnetic moment, has no preference between up and down.

In Fe-57, the J=3/2 state is split into two

components. Application of the formula shows the
two levels are symmetric about the unsplit energy.
The J=1/2 level, however, is spherically symmetric
and shows no splitting. With an absorber containing
Fe-57 in a suitable material, one sees two lines of
equal intensity, equally spaced about the resonance
energy.
If we measure the energy difference between the two
lines in the spectrum, we can use the formula for the energy shift to find Q dE/dz, the product of
the quadrupole moment for J=3/2 and the electric field gradient. If we further know the
quadrupole moment from nuclear experiments, we can estimate the field gradient at the nucleus.

(f) Temperature shift (2nd-order Doppler shift): The thermal vibrations of nuclei can shift
the frequency of their recoil free radiation by means of the relativistic Doppler effect. One can
understand this shift by remembering that moving clocks, according to special relativity, run
slow. Since the nucleus is moving, an observer in the lab frame perceives a nuclear-frame clock
to run slow. For instance, suppose a lab-frame observer finds that 1 second passes in the lab
frame while the nuclear-frame clock ticks off 0.5 sec. If the nucleus radiates with a frequency νo
= 2 Hz in accordance with its own clock, it would emit one wavelength in 0.5 sec as read on its
clock. However, during this time the observer sees 1 sec pass on the lab-frame clock, yet only
one wavelength has emerged. The lab-frame observer will say that the frequency is not 2 Hz but
1 Hz: the observed frequency has been shifted downwards by the factor by which moving clocks
2
run slow, namely 1-(v/c)2 . Since v increases with temperature, the energy of the gamma ray
shifts lower as the temperature is increased.

(g) Antiferromagnetic transition in ferrous fluoride: If the iron atom is in a lattice that
undergoes a transition to ferromagnetic or antiferromagnetic (AF) order, then the lattice of iron
atoms is magnetized. We focus on the AF case. The iron nuclear spin will interact with the
sublattice magnetization M via the hyperfine interaction; hyperfine effective magnetic fields can
B-2-8
be enormous, splitting the ground and excited states, enabling the internal hyperfine field to be
determined. In the paramagnetic state the average of the magnetization is zero, whereas the
ordering breaks the symmetry and the sublattice magnetization or the hyperfine field is non-zero
as in the figure below for the antiferromagnet, FeF2 .

The hyperfine effective field as a

function of temperature in FeF2.

The sublattice magnetization M is the order parameter for an antiferromagnet; thus the
Mössbauer effect can be used to determine the order parameter. Above the Néel temperature TN
the M is zero whereas below it rises as a power law
M (T ) = M 0 (1 − T / TN )β ,
valid in the vicinity of the transition. The exponent β is called the critical exponent and
extensive theory shows that it has a value of around 1/3.

APPARATUS

In this section we give a general description of the various components. Detailed

information, especially instruction manuals and spec sheets can be found with the Bench Notes.
Be sure to read these, but they are to remain in the lab, i.e. do not take them home to read.
Copies of the Bench Notes are available on the course web site.

• Co57 source (in palladium matrix) mounted on velocity driver (loudspeaker)

• velocity transducer (Sanborn LVsyn 6LV2)
• driver control unit (laboratory-built)
• proportional counter (Reuter-Stokes RSG-61) with H.V. power supply (Ortec 478)
• proportional counter preamp (Canberra 806)
B-2-9
• spectroscopy amplifier (Ortec 570)
• single channel analyzer (Ortec 551)
• linear gate and pulse stretcher (Ortec 542)
• multichannel analyzer (computer-based)
• oscilloscope (Tektronix 2225)
• universal counter (Hewlett Packard 5315A)
• liquid nitrogen cryostat (Precision Cryogenic Systems PCS 4-06-102) with cold finger
• mechanical vacuum pump
• temperature controller (Lake Shore 330)
• Michelson interferometer for velocity calibration (5mW HeNe laser, beam splitter, mirrors
and photodetector)
• various absorbers, including Fe enriched in Fe57, Fe2O3 (ferric oxide), K4Fe(CN)6·3H2O
(potassium ferrocyanide), Na4Fe(CN)6·10H2O (sodium ferrocyanide), FeF2 (ferrous
fluoride), FeS2, FeS (ferrous sulfide), stainless steel (the stainless steel absorbers are
available in thicknesses from 0.2 to 3 mils), garnet, permalloy, and others.

B - 2 - 10
Radiation source: Co57 (half-life = 270 days) decays by K-electron capture to Fe57 in the
nuclear excited state with J=5/2 to J=3/2. The 14.41 keV line is emitted when the J=3/2 excited
state in Fe57 decays to the J=1/2 ground state with a lifetime of τ = 1.4x10-7 sec. The Co57 has
been electrodeposited on a Palladium matrix and has an active diameter of 6 mm; the foil is held
in a "Ranger" type mount, which, in turn is attached to the velocity transducer. A lead shield
surrounds it. The source strength was 5.0 mCi on 1 September 2006. Although the source has
been sprayed with an acrylic coating it is NOT a sealed source and, if necessary, should be
HANDLED WITH CARE. Make sure you review and follow the radiation safety procedures
outlined on pages 6-8 of the Introduction in this Laboratory Manual.

Velocity driver/transducer: To produce the variable overlap between emission and absorption
lines, the emission line from the Fe57 is Doppler shifted by moving the Co57 source. The source
is connected to a loudspeaker, which is driven harmonically by a sinusoidal voltage supplied by
the driver control unit. The connecting rod between the Co57 source and the loudspeaker is a
velocity transducer. It consists of two parts: a shielded cylindrical coil assembly and a
permanent magnet. The relative motion of the magnet and the coil generates a voltage whose
magnitude is proportional to the velocity, and whose polarity indicates the direction of motion.
The sensitivity is about 225 mV cm-1 sec (open circuit), but this is something you should
calibrate as described in the Experimental Procedure section. The output signal from the LVsyn
velocity transducer is available at the blue Pomona box mounted on the tripod support over the
speaker, and can be monitored on the scope. (The BNC jack and the coax cable coming out of
the box are wired in parallel.) Note that the position and velocity of the source are 90˚ out of
phase with each other.

Detector: The detector is a proportional counter featuring a 1" diameter, 0.010" thick (46
mg/cm2), beryllium side window -- USE EXTREME CAUTION NOT TO PUNCTURE THE
WINDOW. The counter is filled with a mixture of Xe-CO2 gas at one atmosphere pressure. The
(instrumental) line width for the 14 keV line is 1.3 keV, giving it about a 10% (FWHM)
resolution at that energy. Its efficiency is 67% at 14.4 keV (and 89% at 6.5 keV). The bias
voltage (+1850 V) is supplied by a H.V. power supply located in the NIM Bin Crate. It is fed
through the preamp to the counter; the cable (with MHV connectors) that connects the preamp to
the counter also carries the signal.

Driver control unit: The driver control unit is a precision oscillator/amplifier located in the
NIM Bin Crate. It provides the 10-100 Hz sinusoidal driving voltage for the loudspeaker. (The
universal counter is used to accurately determine and monitor the period of this sine wave; use
B - 2 - 11
the period mode of operation.) The controls on the upper half of the panel adjust the frequency
and amplitude. The controls on the lower half of the panel adjust the timing of the TTL-level
trigger signal. This trigger signal (5 µsec duration) is used to externally start the sweep of the
multichannel scaler. The phase of the trigger out can be set to 0˚ or 90˚ with respect to the amp
out, or any value in between with the delay adj . This makes it is possible to select at which
point one wants the MCS sweep to begin with respect to the movement of the Co57 source.
Simultaneous display of the LVsyn and trigger signals on the scope facilitates this adjustment.
For convenience, the rear-panel and front-panel signal connectors are wired in parallel -- using
both allows you to monitor the signals on the scope without using extra BNC "T" connectors.
Note that the LVsyn In signal connector is presently not in use (and is disconnected internally).

The proportional counter signal is first amplified (and buffered) by the preamp before it
goes to the amplifier (in NIM Bin). Note the HV/OPER switch on the preamp -- please make
sure that it is in the HV position whenever the high voltage bias is changed (or turned
off/on). The spectroscopy amplifier further amplifies (to a maximum of 10 volts) and shapes
(shortens) the signal. Use the scope to observe the signal at each point in this sequence. The
signal is typically split after the spectroscopy amplifier. It is routed to the multichannel analyzer
(MCA) and the single channel analyzer (SCA). The SCA is used to sort out those pulses
corresponding to the 14 keV gamma rays. The SCA window width and baseline voltage must be
set appropriately. This can be done with just a scope, but a more elegant solution is to use the
MCA in conjunction with the SCA as described below. Manuals for the preamp, amp, SCA, and
MCA are available with the Bench Notes and should be consulted for operation information.

Multichannel analyzer: The multichannel analyzer (MCA) is used in either one of two modes
of operation: pulse height analysis (PHA) or multichannel scaling (MCS). In the PHA mode (as
described in the preceding paragraph) it stores a histogram of pulses its input sees versus the
voltage of those pulses, which, in this case, is equivalent to the energy of the gamma that
produced the pulse. The detector high voltage and spectroscopy amplifier gain together
determine the horizontal (energy) scale. The PHA is useful in viewing the total energy spectrum
of the radioactive source and selecting the 14 keV line. The SCA output (0.5 µsec long 5 Volt
pulses) can be used to enable the PHA in a coincidence mode so that, rather than seeing the
entire pulse height spectrum on the PHA, only the pulses corresponding to the 14 keV gammas
are recorded. In other words, the PHA will analyze only those input pulses coincident with the
logic pulses coming from the SCA. Since the SCA sends out a logic pulse only when a pulse at
its input falls within the SCA voltage window, the voltage window settings can be determined

B - 2 - 12
and set by viewing their effect on the PHA spectrum. Once the SCA window has been set, the
SCA output pulses can be counted in MCS mode.

MCS mode stores a histogram of pulses its input sees versus time. The MCS
consecutively advances through the channels of the selected memory size "dwelling" in each one
for a preset amount of time and stores the number of counts received during the dwell time in
that particular channel. Thus, by synchronously moving the Co57 source and monitoring the
counts from the proportional counter in the MCS mode, one obtains a histogram of counts versus
velocity (of the Co57 source). Furthermore, if the SCA output is used as the input signal to the
MCS, then the histogram will represent the number of 14 keV gammas detected versus velocity.
Once the velocity scale has been calibrated and converted to Doppler energy shift, then the
histogram represents the transmission (and absorption) of 14 keV gammas versus energy.

The pulse height analyzer is a LabVIEW2 virtual instrument named moss_pha.vi. A

multifunction data acquisition card (National Instruments PCI-6070E) interfaces the software
with the NIM modules. You can choose either continuous or timed execution. A breakout box
(BNC-2110) provides convenient connectors. The following discussion refers to names of BNC
connectors on the 2110. In the internal trigger mode, the analyzer processes any signal present at
the AI(zero) plug that crosses an adjustable threshold called the lower level discriminator. In
external trigger mode, the analyzer waits for a trigger pulse at AI7. The external mode is useful
for checking the SCA windows. The PCI-6070E is marginally fast enough to digitize the output
pulses from the Ortec 570 amplifier. If you choose a fast shaping time on the amp, you will get
strange results from the pha due to undersampling. If desired, the Ortec 542 pulse stretcher can
be used to improve the sampling.

The multichannel scaler is named moss_mcs.vi. The multichannel scaler has to sweep
synchronously with the Co-57 source modulation. Therefore the only trigger mode offered is
“external.” The trigger input is USER2, which in turn is connected to PFI8 (counter0 source) by
a hookup wire. The input for event counting is USER1, connected to PFI3 (counter1 source). A
typical number of channels is 1000, so the dwell time per channel has to be on the order of tens
of microseconds since the Co-57 modulation is on the order of tens of hertz. The onboard clock

2A few hints on LabVIEW: Click the ➪ button to run the VI. To stop execution before the timeout expires, use the
panel STOP button, not the  button. Both programs have adjustable “update times.” The display is updated, and
data is saved (If the SAVE? Button is switched to YES) once every update time. You will see nothing before the
first update time expires. The program only checks the SAVE? Button when it starts; if you switch it to YES while
the program is running, nothing will happen.

B - 2 - 13
of the PCI-6070E is used to time the channel advance. A control is provided for the clock speed.
Dwell time is the inverse of clock frequency. Set your desired clock frequency before running
the program since the clock frequency can’t be changed during execution. A hookup wire also
connects the clock output (FOUT) to PFI4 (counter1 gate) to define the length of the time bins.

Liquid nitrogen cryostat: In order to measure the Néel transition in iron fluoride, the sample
must be cooled a few degrees below the boiling point of nitrogen (77K). To avoid the difficulty
and expense of liquid helium, we use pumped nitrogen. The two-liter cryostat mounted behind
the Mössbauer apparatus contains the sample on a temperature-controlled copper rod. The iron
fluoride is in powder form, mixed with silicone grease to promote temperature uniformity and
ease of handling. The cryostat can be rotated 90˚ to place the iron fluoride sample between the
source and detector. The sample is located between the beryllium windows, which are nearly
transparent to gamma rays. The windows make it possible to keep the sample in vacuum for
thermal isolation and measure transmission through it simultaneously. CAUTION: TH E
BERYLLIUM WINDOWS ARE FRAGILE, POISONOUS, EXPENSIVE AND EASY TO PUNCTURE . DO NOT
TOUCH THEM. Once rotated, the cryostat jacket and liquid nitrogen space have to be connected to
the vacuum pump. A vacuum manifold with valves and tubing is in the blue cabinet next to the
experiment. Ask the faculty or staff for help with this procedure.
A copper cold finger is mounted in the bottom of the liquid container. It meets another
smaller copper bar at right angles, which holds the sample. This smaller copper bar is connected
to the cold finger through a calibrated copper wire, setting the time constant for cooling at about
25 minutes. A platinum resistance thermometer measures the temperature near the sample. The
Lake Shore 330 temperature controller supplies current to a heater (25Ω resistor) in order to
reach a setpoint temperature. You can see photographs on the course web site.
A thermocouple gauge measures the pressure in the vacuum jacket. This should always
be “low,” or less than 10mTorr while the cryostat is cold. A mechanical Bourdon gauge near the
top of the cryostat measures pressure in the nitrogen can. Hand-written numbers on the side of
the gauge give a rough idea of the temperature in Kelvin. “T” means triple point.
The cool-down procedure is as follows: Contact the staff at least one day before you
want to cool down. They will prepare the cryostat by pumping the vacuum jacket overnight. It
is best to fill the liquid nitrogen in the morning; the staff can do this as well. Before filling liquid
nitrogen, the vacuum jacket must be isolated from the mechanical pump as the pressure in the
cold cryostat will be lower than in the pump! If you leave the nitrogen at one atmosphere, the
measured temperature of the sample will reach about 83K, well above the Néel temperature. A
convenient operating pressure range for work below the Néel temperature is -20 inches Hg to -24
inches Hg. This takes the bath to 70K to 65K – cold enough to see splitting clearly, but still
B - 2 - 14
above the triple point. To reach this pressure, crack the small needle valve open about 1/8 to 1/4
turn. Use the large bellows valve only when you want to go to the lowest possible temperature
(about 60K).
The temperature controller is easy to operate. Simply enter a setpoint, and turn on the
heater (start with medium power). The controller will determine its PID parameters
automatically. For the ambitious, they can be entered manually. As of August 2006, the
thermometer has a calibration problem: the measured temperature is nearly 2K greater than the
actual temperature.

Michelson interferometer: The velocity of the source has to be calibrated in order to calculate
the Doppler shift of the gamma rays. An interferometer measures displacement in units of the
wavelength of light, so combined with an accurate measurement of time, you can measure the
velocity of the source accurately. The Michelson interferometer has two arms – one ending with
a fixed mirror and another ending with a moving mirror, which in this case is next to the Co-57
source. The two beams combine at a photodetector producing interference fringes. Both the
reflected beams, as well as the incident beam are split at a 50/50 beamsplitter cube. The cube is
not sensitive to polarization. The source of coherent light is a HeNe laser (wavelength 632.8nm).
SAFETY WARNING: Follow laser safety procedures at all times. Never look into the laser or
at specular reflections. Contain the beampath and all stray reflections above the footprint of the
optical table. Use black foamcore to block unwanted beams. The laser is a Class IIIa laser (less
than 5mW). Your blink response is enough to protect your eyes in case of accidental exposure.
If you prefer to wear laser safety glasses, they can be borrowed from other experiments. Consult
the staff.
In order for the beam to reach the moving mirror, remove the lead disk that serves as the
sample platform. Follow radiation safety procedures. Three additional mirrors steer the beam up
to the source area. Careful alignment is important. The retroreflected beam should retrace the
path of the incident beam. Use a white card with a hole in it to check this.
The velocity function is sinusoidal. You can measure its frequency with the HP counter.
You can measure vmax in two ways. The time required to pass through a few fringes near vmax
gives vmax directly. Otherwise you can count the total number of fringes in a half cycle to get the
amplitude of the motion, and then multiply by the frequency. It is best to trigger the scope on the
output of the Lvsyn pickup coil.

B - 2 - 15
EXPERIMENTAL PROCEDURE

Setting up the detector: After connecting the apparatus as described in the previous section, the
first thing you will need to do is to identify the 14.41 keV line from the Co57 source. Set the
proportional counter high voltage to +1850 V and look at the output of the amplifier with the
oscilloscope. Once you have made all the appropriate adjustments on the amp and scope, you
should be able to resolve two prominent energy levels in the midst of the other pulses -- they
show up as bright bands. The two levels should be the 14 and the 6 keV lines from the Co57
source; the height of the pulse you see is proportional to the energy of the line. You can also see
still higher levels clipped by the amplifier. Notice that by placing a thin plastic disk between the
source and detector you can cut out one of the lines without drastically altering the other. Now
connect the output of the amplifier to the PHA. The horizontal scale corresponds to the height of
the pulse and, therefore, to its energy. The main peaks (with the Co57 source in a Pd matrix, as it
is presently) are the 6 keV line from an Fe x-ray, the 14.41 keV line of interest, a 22 keV Pd x-
ray, and a 89 keV Pb x-ray, plus some lines above 100 keV from the Co57 decay. You may have
to adjust the amplifier gain to see all these peaks. Using the plastic disk or aluminum or copper
foils as x-ray filters, you can identify the energy of the displayed lines based on their relative
absorption. You can also compare the spectrum with that obtained from an Am241 source, which
has prominent lines in this energy range. You may wish to perform an "energy calibration" as a
double check in identifying the various lines in the pulse height spectrum because, if you end up
not using the 14 keV line in your setup there is no hope that the experiment will work. Finally,
adjust the gain on the amp until the 14 keV line falls in the upper half of the scale.

Set up the SCA: Now that you have an energy spectrum of the Co57 source, learn how to use
the SCA to select the detector pulses corresponding to the 14 keV peak in the spectrum. As
outlined above, this involves adjusting the voltage window appropriately. The numbers on the
dials of the SCA correspond to volts; inspection of the amplifier output on the scope should tell
you approximately where to set these dials. Fine adjustments are made by using the SCA output
to "gate" the PHA with its coincidence input (PFI0). Remember to select external trigger mode.

Velocity transducer: Begin by setting the frequency of the driver (amplifier) to some nominally
low value, like 40 Hz (To monitor the frequency more precisely, use the HP Universal Counter
in the period mode.). Calibrating the velocity with the Michelson interferometer takes some
time, and the lead sample platform has to be disassembled. This can be done either before or
after your data runs; you won’t want to do it in the middle. The amplitude of the output should
be low -- if you hear the speaker it is much too high! If you gently touch the speaker cone with
B - 2 - 16
your fingertips, it should feel like a purring pussycat. Typical output of the LVsyn will be in the
tens to hundreds of millivolts range, as monitored on the oscilloscope.

MCS setup: Having settled on a transducer frequency, the appropriate dwell time for the MCS
must be chosen. This further depends on the number of channels you wish to use. Once started
by the external trigger pulse (from the driver control unit), the MCS consecutively advances
through the channels of the selected memory size "dwelling" in each one for the preset amount of
time and stores the number of counts (pulses from the SCA) received during the dwell time in
that particular channel. Having reached the last channel (this is considered as one "sweep"
through the channels), the MCS waits for the next trigger pulse and repeats the cycle, adding the
new counts to the old. In this way each sweep of the MCS is synchronized to the velocity
transducer. However, you must make sure that the product of the dwell time and the number of
channels is slightly less than the period of the velocity transducer. It cannot be "exactly" equal to
the period and, if it is slightly larger than the period, every other trigger pulse will be missed by
the MCS. This would reduce your counting efficiency by a factor of two. The dead time
indicator of the mcs program will let you know if you made this mistake. Dead time should be in
the range of a few percent.

MEASUREMENTS

The following are suggestions of the kinds of measurements you may wish to undertake. You do
not have to do them all and you may wish to duplicate other experiments you find in the
literature. Discuss this with your instructor.

Background determination: As is true for any counting experiment, one should always
determine how many of the events are due to "background," which we usually consider to be
"noise." In the present experiment, a certain number of counts within the selected energy
window come from background -- from non-resonant gamma rays that just happen to have
energy around 14 keV. You will need to determine how much background there is within your
window: the important quantity is the fraction of the counts in the window that come from
background. In PHA mode one observes that there is a fairly constant level of background over
the full energy range of the spectrum. As an estimate, it is safe to take this as the background
level under the peak of interest.

B - 2 - 17
Recoil-free fraction and linewidth: The stainless steel absorber displays the simplest
Mössbauer spectrum. Select a fairly thin foil of stainless steel and place it in front of the
proportional counter. From the Mössbauer spectrum, measure the depth of the absorption dip.
The fractional absorption is defined by A = (N∞ - Ndip)/(N∞ - Nbg), where N∞ is the number of
counts per channel far from the dip, Ndip is the number of counts at the bottom of the dip, and
Nbg is the number of the N∞ counts that are due to background. Applying the Debye theory to
the properties of the absorber and calculating the expected correction to the absorption will yield
the recoil-free fraction of the source. Compare this value of the recoil-free fraction to the
theoretical value for the source. How good is the agreement? What does that tell you about the
applicability of the Debye theory?

Measure the full width at half maximum (FWHM) from the spectrum. Using the
correction formulas from the literature, determine the lifetime of the J=3/2 state. How well does
the value compare with the accepted lifetime of τ = 1.4x10-7 sec? Note that the dip does not
occur at zero velocity or zero energy shift. Which effects described above are responsible for
this?

Isomeric (chemical) shift: Measure the isomeric shift between the Fe57 source in Pd and Fe57 in
various absorbers such as iron, stainless steel, FeF2, Fe2O3, garnet3 (yttrium iron garnet
Y3Fe5O12 and dysprosium iron garnet Dy3Fe5O12), Na4[Fe(CN)6]·10H2O, K4[Fe(CN)6]·3H2O,
or FeS2.

Zeeman effect: To observe the Zeeman effect, use an absorber of iron enriched in the isotope
Fe57 (to make the effect more prominent). Because iron is ferromagnetic, it will probably have a
slight magnetization. Obtain a Mössbauer spectrum and determine the locations and depths of
the dips (correcting for background). Explain the origin of each dip in terms of the allowed
Zeeman transitions. Use the separations between the lines to determine the energy differences
Δo and Δ 1 between the Zeeman split levels of the J=1/2 and J=3/2 states. Calculate the ratio
|µo/µ1| of the magnetic moments. Take the magnitudes of the magnetons for the levels to be |µo|
= (0.0903 ± 0.0007)µN, |µ 1| = (0.1548 ± 0.0013)µN4. Do your results for |µo/µ1| agree? How
large is the magnetic field at the location of the iron nucleus? Calculate the relative intensities of
the lines. Are they in accord with expectations?

3Banminger, Cohen, Marinov, and Ofer, "Study of the Internal Fields Acting on Iron Nuclei in Iron Garnets," Phys
Rev 122, 743-748 (1961).
4 Greenwood and Gibb (1971) p. 102.

B - 2 - 18
Quadrupole splitting: To observe quadrupole splitting you need an absorber containing Fe in a
non-symmetric environment. Then the nuclear quadrupole moment interacts with the electric
field gradient, splitting the central transition line in two. Use a quadrupole-active absorber, such
as sodium nitroprusside, Na2[Fe(CN)5NO]·2H2O, and measure the locations and depths of the
dips. Demonstrate theoretically that the dips should be equally spaced about the resonance
energy. Are they equally spaced about the zero velocity? If not, explain why. Determine the
separation between the dips and evaluate QdE/dz, the product of the quadrupole moment for the
J=3/2 state and the field gradient. Given that Q is on the order of 0.1 to 1 barn, how large does
this make the field gradient? Is this reasonable?

Temperature shift: Place a sample in a liquid nitrogen bath and measure the Mössbauer
spectrum. Observe the shift of the peaks. Is it in the correct direction? How large is it? Is the
size of the shift in agreement with the predictions of the second-order Doppler treatment?
The Debye theory makes predictions about the change in fractional absorption with
2
temperature. Because the mean square displacement x depends in a specific way on
temperature, one can calculate the change in the recoil-free fraction of the absorber when it is
cooled. The recoil-free fraction of the source does not change, so you can predict how much the
observed absorption should change. Measure the absorption and compare with expectations.

Antiferromagnetic transition in ferrous fluoride: In this experiment you study the Mössbauer
effect in antiferromagnetic FeF2 which has a transition temperature near the boiling point of
liquid nitrogen at 1 atmosphere pressure. A powdered sample is mounted on a cold finger in a
cryostat using liquid nitrogen. By pumping on the liquid nitrogen you can lower its temperature
to around 60K. The cold finger can be heated and temperature controlled, thus enabling you to
probe the order parameter in the vicinity of the transitions and determine the magnetization and
the critical exponent, and the Néel temperature. In principle all eight transitions might be
observable because the wave functions of the excited state are mixed by the asymmetric electric
field gradient tensor.

REFERENCES

General Introduction

A.J. Bearden, P.L. Mattern, and P.S. Nobel, "Mössbauer-Effect Apparatus for an Advanced
Undergraduate Teaching Laboratory," Am J Phys 32, 109-119 (1964).

B - 2 - 19
J.R. Haskins, "Advanced Mössbauer-Effect Experiments," Am J Phys 33, 646-656 (1965).

H. Lustig, "The Mössbauer Effect," Am J Phys 29, 1 (1961). Highly recommended as a

readable, detailed introduction and a good place to start reading.

G.K. Wertheim, "Resource Letter ME-1 on the Mössbauer Effect," Am J Phys 31, 1 (1963). AJP
Resource Letters are always useful in starting your library research.

Debye-Waller Factor

J. Als-Nielsen, D. McMorrow, Elements of Modern X-ray Physics, (J. Wiley, New York, 2000).
See page 142.

Selected Reprints

R.L. Mössbauer, "Kernresonanzfluoreszenz von Gammastrahlung in Ir191," Zeitschrift für Physik

151, 124-143 (1958).

H. Frauenfelder, The Mössbauer Effect: a review - with a collection of reprints, (W.A. Benjamin,
1962). Cabot QC477.F7

Mössbauer Effect: Selected Reprints, published by AIP for the AAPT (1963). Cabot
QC491.A44

Antiferromagnetic transition in FeF2

G.K. Wertheim and D.N.E. Buchanan, Temperature Dependence of the Fe hfs in FeF2 below the
Néel temperature, Phys. Rev 161, 478 (1967).

Monographs

J. Danon, Lectures on the Mössbauer Effect, (Gordon & Greach, NY, 1968). Cabot QC490.L4

D.P.E. Dickson and F.J. Berry (editors), Mössbauer Spectroscopy, (Cambridge Univ Press,
1986). Cabot QC491.M6135

T.C. Gibb, Principles of Mössbauer Spectroscopy, (Chapman & Hall, London, 1976). Useful
introduction followed by in-depth treatment of chemical effects. Cabot QC491.G52

N.N. Greenwood and T.C. Gibb, Mössbauer Spectroscopy, (Chapman & Hall, London, 1971).
Cabot QC490.G74

U. Gonser (editor), Mössbauer Spectroscopy, (Springer-Verlag Topics in Applied Phys - Vol 5,

NY, 1975). The first chapter by Gonser is a good introduction. Phys Res QC491.M912

B - 2 - 20
G.K. Wertheim, Mössbauer Effect: Principles and Applications, (Academic Press, NY, 1964).
Cabot QC490.W4

The Exotic Side of the Mössbauer Effect

Once the extension of experimental work to lines as narrow as those of Fe57 was achieved, it
became possible to observe the gravitational red shift within an earth-bound laboratory. Robert
Pound and Glen Rebka performed their historic experiments here at Harvard in the Jefferson
Physical Laboratory. They were able to demonstrate that the frequency of gamma radiation
measured at its source (on the building's roof) differed from the frequency measured in the
basement. The incredibly small shift (about 1 part in a million billion) showed that clocks run
slower in the basement than the roof because, being closer to the Earth, the gravitational field,
and hence the space-time curvature, is larger there. The faster clock on the roof would read a
lower frequency of gamma radiation and thus the wavelength would be longer, or red-shifted.
The red-shift experiment was initially proposed in the first paper (below) and conclusive results
were given in the second. Later it was demonstrated that temperature effects (see page B-2-8 of
this write-up) could have completely obscured the gravitational shift. The third paper presents
the first conclusive experiments taking temperature effects into account

R.V. Pound and G.A. Rebka, Jr., "Gravitational Red Shift in Nuclear Resonance," Phys Rev
Letters 3, 439 (1959).

R.V. Pound and G.A. Rebka, Jr., "Apparent Weight of Photons," Phys Rev Letters 4, 337 (1960).

R.V. Pound and G.A. Rebka, Jr., "Variation with Temperature of the Energy of Recoil-Free
Gamma Rays from Solids," Phys Rev Letters 4, 274 (1960).

B - 2 - 21