Vij D.R. Handbook of Applied Solid State Spectroscopy

Подождите немного. Документ загружается.

A typical configuration would be a 50-mm diameter cylinder with

0.25-mm thick beryllium side entry and exit windows, 25 mm in diameter,

and a Xe+3% CO

fill at one atmosphere. The thin Be windows admit the

14.4-keV

J without loss and the 1 atm Xe fill has enough stopping power to

efficiently collect them without providing too much efficiency for the much

more intense 122-keV

J-photons, which mostly exit undetected through the

second Be window. A thin (~1 mm) plexiglass filter over the detector window

can also reduce detection of the Co and Fe K

and count rates on the 14.4-keV

J of 5 to 10 u 10

–1

can be obtained without saturating the electronics.

A Kr+3% CO

fill is also very popular as it yields even lower sensitivity to

the higher energy

J-photons and so reduces background rates. However, the

14.4-keV

J is above the Kr K-edge and a significant fraction of the Mössbauer

J-photons appear in the 1.8 keV “escape” peak, necessitating the use of an

additional single channel analyzer if good efficiency is to be obtained. The

fluorescence yield of Kr is also poor, with only about 50% of the absorbed

J-photons yielding a photoelectron that can be detected.

The increased high energy sensitivity of NaI(Tl) or HPGe detectors means

that their response is dominated by the 136-keV and 122-keV

J-photons.

Counting rates are more limited and increased backgrounds from Compton

scattering of higher energy

J lead to very poor performance in

Fe Mössbauer

spectroscopy.

Proportional counters with a 1 atm Ar+3% CO

fill are commonly used to

detect the 8.4-keV

169

Tm Mössbauer J, and can also be used for X-ray

backscatter (XBS, see below) work with the ~6.4-keV Fe-K

X-ray. In

principle, a 2-atm Xe+10% CO

fill can be used for either

151

Eu or

119

Mössbauer spectroscopy, but freedom from intense higher energy

contamination and the availability of more efficient detectors for the 21-keV

and 24-keV

J make NaI(Tl) or HPGe better choices for these Mössbauer

resonances.

5.2.2.2 NaI(Tl) Scintillation Detectors

For many applications in the 20–100 keV range, a thallium-activated sodium

iodide scintillation detector is a good choice. A 0.25-mm thick beryllium or

25-Pm thick aluminium entry window minimizes attenuation losses above

20 keV, and a crystal 6–8 mm thick is sufficient to give over 95% efficiency

out to 100 keV (there are no useful Mössbauer resonances beyond this

energy). These detectors actually give slightly poorer energy resolution than

proportional counters (~50% at 6 keV and ~20% at 90 keV) but can be used at

similar count rates. The presence of a photomultiplier tube behind the crystal

does make these detectors significantly heavier and more fragile than the gas-

filled proportional counter. The gain of the photomultiplier can also be

5.2 Methodology

217

5. Mössbauer Spectroscopy

affected by stray magnetic fields and while they are typically manufactured

with P-metal shielding, they should be protected from external magnetic

fields.

5.2.2.3 High Purity Germanium (HPGe) Detectors

HPGe solid state detectors exhibit energy resolutions of < 0.5%, with better

than 95% efficiency from 6 keV to 200 keV. When equipped with a pulsed

optical feedback network to stabilize the detector bias voltage, count rates in

excess of 10

–1

are readily achievable. This performance comes at a

significant price, with a good HPGe detector costing 10–20 times that of a

NaI(Tl) scintillation detector. The detector must also be continuously cooled

with liquid nitrogen during operation, adding a bulky Dewar assembly to the

unit. However, the detector can be allowed to warm up, and may be stored at

room temperature when not in use, unlike Si(Li) detectors, which would be

damaged by such treatment. The comparable performance and increased

convenience has led to HPGe detectors replacing Si(Li) units for most

applications.

Few Mössbauer resonances demand the performance of a HPGe detector

and for the commonly used isotopes such a detector is either inappropriate

(

Fe) or expensive overkill (

151

Eu and

119

Sn). Where HPGe detectors are

essential is in work with noncommercial Mössbauer sources. Neutron

activation or proton irradiation, even of sources prepared with isotopically

separated starting materials, inevitably leads to the presence of one or more

unwanted radioisotopes. Even when the source is effectively pure, J-energies

tend to be higher than the 24-keV

119

Sn resonance, and copious amounts of

fluorescence X-rays can accompany the Mössbauer

J. In this situation, good

resolution is needed to isolate the Mössbauer energy from surrounding

X-rays, and excellent rate capability permits operation with high-activity

sources to compensate for low f-factors without saturating the detector. Count

rates of 5 u 10

–1

on the Mössbauer J in the presence of comparable rates in

non-Mössbauer X-rays are routinely encountered.

5.2.3 Data Collection

Collecting a Mössbauer spectrum requires the proper synchronization of two

sets of signals: the motion of the drive and the detection of

J-photons. The

charge pulses generated in the detector by the stopping of a

J are shaped and

amplified in a pre-amplifier and amplifier. The voltage pulses are then passed

through a single-channel analyzer (SCA), where those within a certain

amplitude range, chosen to correspond to the J-energy of interest, generate a

TTL pulse, while those outside the selected range are rejected. The pre-

amp/amp/SCA chain is available as a dedicated commercial Mössbauer

218

system, or can be assembled from individual nuclear instrumentation modules

(NIM). Costs are comparable and both approaches have advantages. A

dedicated unit is simpler to operate, but more difficult to troubleshoot when

problems develop, whereas a NIM-based system requires a larger initial

investment of time to set up, but it is more flexible and far simpler to

troubleshoot.

The stream of TTL pulses being generated by the SCA must now be

synchronized with the motion of the Mössbauer drive. A typical arrangement

is as follows. The velocity waveform to which the drive’s motion is slaved is

produced digitally by a function generator that also provides two additional

data streams: sweep START and channel advance (CHA). These are used to

control an array of counters in a multichannel scaler (MCS), typically an add-

on card for a PC. The START pulse resets the multichannel scaler and opens

the first data register to receive counts. Any TTL pulses that arrive from the

SCA while the register is open are added to the current contents. When each

successive CHA pulse arrives, the current register is closed, and the next one

is opened. In this way, the arriving TTL pulses are placed in a register that

corresponds to a specific velocity of the drive. When the complete waveform

has been generated (commonly 512 or 1024 channels), a new START is

issued and the cycle repeats. At velocities corresponding to absorption lines in

the sample, the count rate will be slightly lower and those data channels will

increase more slowly. Over time, a Mössbauer spectrum develops.

Once the spectrum is of sufficient quality (determined by visual inspection,

predetermined time, or by evaluation of intermediate fits carried out during

the data collection) collection is halted and the spectrum is transferred and

folded ready for analysis. Since the motion of the drive is necessarily cyclic,

every velocity occurs twice during the waveform, but the two copies are not

identical. The source is slightly closer to the detector during one half of the

cycle than the other, leading to a higher average count rate. Indeed, since the

source position changes continuously during the cycle, so does the count rate,

and the background in the absence of a sample exhibits an approximately

sinusoidal modulation with an amplitude that can exceed 1% for set ups with

short source-detector distances. A second problem arises through mechanical

lags in the drive motion, which can be as much as 1%, and lead to the center

of symmetry of the actual source motion being offset from the center of the

intended velocity waveform. Folding the data greatly reduces the effects of

position-induced count rate variations, but this must be done with care to

avoid distortions due to mechanical phase shifts. There are three common

strategies:

Auto-fold. It is possible to arrange for some multichannel scalers to count

down through their registers as well as up. Therefore, once the drive reaches

its maximum velocity and starts back down, the MCS can reverse its sweep

through the registers. This approach can mask any number of failures in the

instrumentation and runs the risk of destroying the spectrum as it is collected

5.2 Methodology

219

5. Mössbauer Spectroscopy

while only avoiding a simple arithmetic manipulation that is trivially carried

out after the collection is complete.

Fold before analysis. This is the simplest method. The center of symmetry (of

the spectrum) can be determined either by fitting the two halves, seeking the

maximum correlation between them, or by simply observing the folding point

at which the maximum signal is obtained. Some care needs to be exercised

when folding at noninteger points as some form of interpolation of the data

will be needed, and this will make the actual data distribution non-Poisson

and can affect subsequent data analysis. Folding the data after collection

greatly reduces the background variations due to source motion and has the

added benefit that it is nondestructive: the unfolded data are still available for

refolding. Unexpected variations in folding point or poor overlap between the

two halves are immediate indications that all is not well with the

spectrometer.

Fit the unfolded halves together with a background term. A final option is to

not fold at all, but simply fit the duplicated half and include a symmetry-

center variable explicitly in the fit. Some form of background function must

also be included in the fit. The data can be folded for presentation and

interpolation-included distortions of the counting statistics are irrelevant as

they do not affect the analysis.

5.2.4 Calibration

Accurate calibration of the spectrometer is critical and it should provide three

pieces of information: (i) velocity-zero (ii) the velocity scale, either V

max

the velocity step per channel, and (iii) confirmation of drive linearity and

performance. Ideally, calibration is achieved by measuring a standard sample,

but various forms of laser interferometers are also popular.

A thin foil (< 25 Pm) of iron metal provides an excellent calibration

standard for

Fe Mössbauer spectroscopy as the line positions have been

determined absolutely (see Section 5.1) and the material is extremely stable.

The 6-line magnetic pattern also provides a convenient drive linearity test. All

Fe isomer shifts are quoted relative to the center of the iron metal spectrum

at room temperature. Hematite (D-Fe

) can be used to extend the velocity

range out to r10 mm/s, and with confirmed linearity, velocity ranges two or

three times larger than this can be safely used. While calibrating with a

standard sample does use up counting time that could be spent measuring

“real” samples, it is an essential step and also provides a complete check of

the spectrometer performance in the actual mode that it used to acquire data.

Every component of the spectrometer is tested together, and problems that

might be missed on isolated tests can be caught on a clean standard long

before they distort the results on a more complex material.

220

For those not working with

Fe, calibration standards are less clear. The

simplest approach is to mount a

Co source on the back of the drive and use

iron metal as a calibrant, with a second standard being used to set velocity-

zero for the isotope being studied (e.g., BaSnO

is a common choice for

119

work). This procedure works well if

Fe Mössbauer is also being used in the

same laboratory and if the velocity scales being used are compatible. Other

standards that have been used include ErFe

(r 60 mm/s, for

166

Er), Dy metal

(r 200 mm/s, for

161

Dy) and TmFe

(r 800 mm/s, for

169

Tm; see Figure 5.18).

Laser interferometers are a popular and versatile alternative to calibration

that uses a standard sample. They can be run in parallel and so do not cost

counting time, and they have a much wider velocity range. However, they

only provide information on the motion of the reflecting surface mounted on

the back of the drive, and strictly speaking do not measure the actual relative

motion of the source-sample pair. Furthermore, laser interferometers cannot

be used to check source linewidths, nor do they provide any information on

the performance of the nuclear counting electronics. Laser interferometers

employ a Michelson interferometer to count the number of fringes that pass

during the opening of each velocity channel, and knowing the time each

channel is open, the average velocity of each channel is readily calculated.

However, for almost all Mössbauer applications, the motion of the drive

during any given step is comparable to the wavelength of the laser light. For

example, at a V

max

of 10 mm/s working at 30 Hz and 512 data channels, the

drive moves a maximum of about 650 nm over one channel. Coupled with the

highly stable, deterministic motion of the drive, a He-Ne laser (O = 633 nm)

signal shows large blocks of channels with zero, one, and two counts. Various

schemes for introducing position noise have been employed to smooth out this

behavior, but low velocity calibrations are not ideal, and they are best at

higher velocities (> 50 mm/s).

Mechanical recoil can also affect calibrations. While the mass of the

moving coil and source assembly is generally kept as small as possible, it

cannot be made zero and its motion necessarily induces some recoil of the

drive. For conventional transmission work, the effects of this recoil can be

minimised by rigid coupling of the drive to a heavy mount; however when

one works with a cold-source system, or with the source and sample in a

magnet cryostat, the driven mass increases and rigid mounting becomes more

problematic. Elasticity in the long coupling rod between the drive and source

can also affect the actual motion of the source. In such situations, an

interferometric measure of the drive motion may not provide an accurate

measure of the source-sample relative motion. Deviations of 5% are not

uncommon and cross-checking with one or more standard samples is

essential.

5.2 Methodology

221

5. Mössbauer Spectroscopy

5.2.5 Sources

Commercial sources are currently available for three of the most commonly

used Mössbauer resonances:

Fe,

119

Sn, and

151

Eu.

Fe Mössbauer sources

are prepared by diffusing

Co into a cubic metal matrix (commonly rhodium)

to give a single-line source with a good recoil-free fraction. The half-life is

271 days and initial activities of 10–50 mCi (300–2000 MBq) provide useful

count rates for two or more years (depending on the application). One side

benefit is that the source can be retired to an undergraduate teaching lab and

be used for teaching experiments for another year of use. Observed linewidths

for a thin iron metal absorber are around 0.113 mm/s (half-width at half

maximum–HWHM).

119

Sn is unusual in that the parent for the decay is a long-

lived excited state of

119

Sn, usually designated

119m

Sn, as shown in

Figure 5.10.

Figure 5.10 Decay scheme associated with

119

Sn Mössbauer spectroscopy.

As the parent and daughter are chemically identical they cannot be

separated. It is essential therefore that sources be prepared from fresh

119m

Sn,

as a build-up of

119

Sn in the source leads to resonant self-absorption, which

can cause a significant reduction in source f-factor (recoil-free

J are

preferentially reabsorbed in the source, while those from recoil events escape)

and an increase in linewidth (

Jfurther from the center of the Mössbauer line

are less likely to be reabsorbed in the source and so are over-represented in

the emission spectrum). A source that starts out bad, can only get worse as

ground state

119

Sn continues to accumulate. The standard matrices are CaSnO

and BaSnO

. The parent half-life is ~290 days so the working lifetime of a

source is comparable to that of

Co sources. The J is emitted quite efficiently,

so activities in the 2–10 mCi (70–400 MBq) range yield acceptable to high

counting rates. Typical source linewidths observed with a BaSnO

absorber

are ~0.42 mm/s (HWHM).

Sources for

151

Eu Mössbauer spectroscopy usually employ

151

Sm in a SmF

matrix (Figure 5.11). Linewidths of 0.12 mm/s are typical. The parent half-life

is 87 years, so apart from the inevitable line broadening and decline in f-factor

associated with the build-up of

151

Eu in the source, a good

151

Eu Mössbauer

source will last a researcher’s career. Unfortunately, the

J emission is

extremely inefficient. Only about 1% of the electron capture decays of the

parent

151

Sm actually lead to a

151

Eu nucleus in the 21.5-keV Mössbauer state.

119

23.9 keV, 18 ns

0 keV 1/2

3/2

11/2

222

89.5 keV, 293 d

Internal conversion dominates the subsequent decay process so that only

about 1% of the transitions are radiative. The inefficiency of the decay

process means that significant source activities are essential to attain useful

count rates. Even with 100 mCi (4 GBq) counting rates are of order 200 s

–1

High activity sources are not currently available, with 30 mCi (1 GBq) being

the limit. Such a source would not lend itself to high throughput experiments.

Figure 5.11 Decay scheme associated with

151

Eu Mössbauer spectroscopy.

5.2.5.1 Nonstandard Resonances

Many Mössbauer resonances are underused because sources are not readily

available. The parent isotope is often short-lived and some investment in

source materials preparation is needed and isotopically separated starting

materials must be used. In addition, higher

J energies mean that useful

recoil-free fractions are only obtained with both the source and the sample at

low temperatures (typically below 50 K). We consider several examples here.

197

Au: The principal problems with this transition are the short half-life of

the parent

197

Pt (19 hours) and interference between the Mössbauer J at

77.35 keV and the K

E

X-rays from Au (77.984 keV). The former demands

nearby access to a nuclear reactor while the latter must be accepted. Source

preparation is relatively easy as

196

Pt is available as a metal and can be

irradiated directly. Neutron fluxes of 3 u 10

–2

–1

are sufficient to obtain

useful source activities.

170

Yb: The radioactive parent (

170

Tm) is obtained by neutron activation of

169

Tm which is, conveniently, the only naturally occurring isotope of thulium

(Figure 5.12). A 10 wt.% alloy of thulium in aluminium (a mixture of TmAl

in an Al matrix) makes a convenient single-line source. The 130-day half-life

makes it a relatively convenient source to work with, but it demands higher

neutron fluxes so it can be activated in a reasonable time (15 h at 3 u 10

–2

–1

works well). The 84.2-keV Mössbauer J is well separated from the

many X-rays emitted by the source.

151

87 y

151

21.5 keV, 9.7 ns

0 keV

5/2

7/2

99%

5.2 Methodology

223

5. Mössbauer Spectroscopy

Figure 5.12 Decay scheme associated with

170

Yb Mössbauer spectroscopy.

166

Er: This resonance is also accessed by activation of a single-isotope

precursor, in this case

165

capture cross-section make high activity sources possible. Overnight

irradiation at 3 u 10

–2

–1

works well, but the reactor facility needs to be

fairly close by. A standard source matrix is Ho

0.4

0.6

which remains unsplit

down to about 5 K. At lower temperatures, the source needs to be heated to

prevent magnetic relaxation broadening the emission line.

Figure 5.13 Decay scheme associated with

166

Er Mössbauer spectroscopy.

169

Tm: There are many problems limiting access to this very useful

Mössbauer resonance. The source,

169

Er, must be prepared by neutron

activation of isotopically pure

168

Er in order to avoid the presence of

interfering radiations (Figure 5.14). As with

170

Yb, the best source matrix is

Al (+10 wt.%

168

Er). Unfortunately

168

Er is only available as an oxide, and the

reduction of

168

and subsequent alloying with aluminium is an

interesting challenge. The capture cross-section is small and the source must

be extremely thin to permit the 8.4-keV Mössbauer

J to escape, thus

necessitating long irradiations in order to attain useful source activities.

5.2.5.2 Environments

Mössbauer spectroscopy is a well-established nuclear technique in solid state

physics, therefore it is not the technique itself that is central to Mössbauer

work, but rather the samples and the environments to which they are exposed.

165

27 h

166

80.6 keV, 1.9 ns

0 keV

Ho (Figure 5.13). The short half-life (~27h) and large

224

Fe,

119

Sn,

151

Eu,

and

169

Figure 5.14 Decay scheme associated with

169

Tm Mössbauer spectroscopy.

5.2.6 Cryostats

Helium-flow cryostats have largely replaced older cold-finger and exchange-

transfer line to a large transport dewar, or have an integrated liquid helium

reservoir. The latter designs will generally also have a liquid nitrogen-cooled

radiation shield and can exhibit consumption rates as low as 100 cm

/hour for

operation from 5–300 K. Liquid helium is drawn through a capillary from the

storage vessel and vaporized as it passes through a small sintered copper plug

fitted with a resistive heater. Flow rates are metered using a needle valve and

temperature control is achieved by balancing cooling derived from the flow of

helium vapor with resistive heating applied to the copper plug. Silicon diodes

are popular as thermometers in this temperature range; however, for work

below 50 K, departures from standard curves can become significant and they

should either be individually calibrated, or replaced with a calibrated

ruthenium oxide “cernox” thermometer. Temperatures below 5 K can be

achieved by evacuating the sample space with a reasonably large roughing

pump and using the needle valve as a throttle valve. Work down to 1.3 K is

possible with some care. Much lower temperatures demand a dedicated

system or a

He/

He dilution refrigerator.

5.2.6.1 Closed-Cycle Refrigerators

Substantial improvements in closed-cycle refrigerator technology have led to

the widespread availability of simple two-stage units that can cool below

4 K without demanding liquid cryogens. While they represent a larger capital

investment than a simple helium-flow cryostat, they are extremely reliable

5.2 Methodology

225

magnetic fields. For many of the more popular resonances (

gas systems, as the former are far more efficient. Current designs use either a

the composition or pretreatment of the material under study; however, the

Tm) it is also possible to work at elevated temperatures (metallic iron

majority of Mössbauer studies involve reduced temperatures or applied

A significant amount of work is carried out under ambient conditions, varying

has been studied well above its Curie temperature).

9.4 d

5. Mössbauer Spectroscopy

and can quickly save on liquid helium costs. Unfortunately, the reciprocating

action of the expander head and the action of the large compressor combine to

make the cold stage a mechanically noisy environment. Making a Mössbauer

measurement while the sample is mounted on a vibrating stage is impossible.

Isolating the sample stage from the mechanical noise of the cold head, while

still coupling it thermally, adds some complexity to the system and usually

raises the base temperature of the sample. However, with proper designs, it is

possible to reduce the vibration levels to the point where they do not lead to a

measurable broadening of the Mössbauer lines of an

Fe spectrum. Current

best designs place the cold head under the sample where it is used to cool

helium gas at atmospheric pressure, which in turn cools the sample. The

sample mounting stage is mechanically isolated from the cold head, and only

coupled through soft (often rubber) bellows. The top-loading nature of this

design is a major advantage as samples are surrounded by cold helium gas and

so are uniformly cooled. They can be removed and remounted without

shutting down the system, Turn-around times of about an hour are normal.

Older, bottom-loading designs place the sample in vacuum, greatly increasing

the equilibration times and opening up the possibility of temperature gradients

between the sample and the thermometers. The refrigerator must be shut

down and warmed to ambient temperature before the sample can be changed.

Turn-around times for these systems can be as long as a day. With samples

being studied at many temperatures over several days, the sample change

delay is a nuisance but not a major problem.

5.2.6.2 Magnetic Fields

The magnetic ordering of materials can be strongly affected by an applied

magnetic field, and Mössbauer spectroscopy is widely used to study spin-flop

transitions, field-induced reorientations, and magnetostructural transitions.

Such work brings its own set of restrictions and problems. The large fields

inevitably demand a superconducting solenoid and so a liquid helium cryostat

is needed. High-field magnets are physically large so a large source-sample-

detector path is unavoidable. High-activity sources and efficient detection

become important. Detector choice becomes more limited as the stray fields

from a large solenoid are significant and these can affect the photomultiplier

used with NaI(Tl) scintillation detectors and also damage HPGe detectors

through penning discharges in the vacuum space around the biased detector.

Gas-filled proportional counters are popular choices, however it is possible to

use scintillation detectors if a long light pipe is placed behind the NaI(Tl)

crystal so that the photomultiplier can be located well away from the magnet.

The simplest cryostat geometry has a vertical field with the sample placed

in the center of a superconducting solenoid. This arrangement has the field

and J directions parallel to each other (axial geometry). The detector is placed

below the cryostat and the source is above the sample, located at a field-null

226