-
The Effect of Severe Plastic Deformation on Thermoelectric
Performance ofSkutterudites, Half-Heuslers and Bi-Tellurides
Gerda Rogl1,2,3, Michael J. Zehetbauer4 and Peter F. Rogl1,3
1Institute of Materials Chemistry, University of Vienna,
Waehringerstr. 42, A-1090 Wien, Austria2Institute of Solid State
Physics, TU Wien, Wiedner Hauptstr., 8-10, A-1040 Wien,
Austria3Christian Doppler Laboratory for Thermoelectricity, Wien,
Austria4Physics of Nanostructured Materials, University of Vienna,
Boltzmanngasse 5, A-1090 Wien, Austria
Although thermoelectric materials with a figure of merit, ZT,
higher than 1.2 have gained considerable interest in electric power
generation,little is hitherto known on the influence of severe
plastic deformation on thermoelectric performance. Severe plastic
deformation is one of theelegant techniques to furnish ultrafine
grained microstructure with a high level of point, linear and
surface defects. Particularly the phononscattering on these defects
is used as a main route to reduce thermal conductivity in
thermoelectric materials and - as a consequence - increasingZT.
The present article provides an overview on the achievements of
the various techniques of severe plastic deformation to gain high
figures ofmerit in the thermoelectric materials, commonly used in
energy conversion devices, such as skutterudites, clathrates,
Heusler phases, andbismuth tellurides.
For all skutterudites, high pressure torsion, as one of the
major techniques of severe plastic deformation processes, is a
great tool to eitherenhance ZT of hot pressed samples or to
directly produce fast and easily high ZT thermoelectric bulks. A
still unsurpassed highlight is theenhancement of ZT from 1.6 to
almost 2 (Sr0.09Ba0.11Yb0.06Co4Sb12) at 825K.
Whilst for thermoelectric clathrates so far little success was
reported, high pressure torsion treatment of Heusler and
Half-Heusler phases insome cases was able to boost ZT
(VTa0.05Fe2Al0.95, rising ZT from 0.22 to 0.3) although the
absolute ZT increases are still disappointing. Forp- and n-type
Half-Heusler alloys (NbFeSb- and TiNiSn-based) at least three
temperature cycles are necessary to gain a thermally stable
state,indicating that obviously the introduced defects and
structure changes are more resistant against heat treatments than
in case of skutterudites.
Bismuth tellurides of type V1(VI)1 and/or V2(VI)3 (V, VI denote
the group elements) have been already deformed by high
temperaturepressure or high temperature extrusion before the SPD
methods were known, for the sake of improving ZT at least with
temperatures 300500K,at most fighting with the condition to achieve
a high electrical conductivity because of the strong anisotropy in
these materials. By starting withball milling followed by high
temperature pressing at not too high temperatures, not only the
electrical conductivity could be kept large but alsothe lattice
thermal conductivity was diminished such that figures of merit up
to ZT = 1.4 at T = 373K were achieved. This value could bereached
by many of the ECAP experiments published so far, although only
across the sample long axis because of lattice anisotropy.
Firstapplications of HPT did not reach that ZT level as either the
conditions of texture could be not fulfilled, or, above all, the
processing rates and/ortemperatures were too high. Recent
investigations not having involved SPD found the importance of the
lattice defects’ specific phononscattering efficiencies, especially
that of dislocations, and by introducing them in sufficiently high
densities, enhancements of p-type Bi-Tellurides up to ZT = 1.9 were
possible. These findings recommend the use of SPD methods here, not
at least as they have been already appliedvery successfully by the
authors of this review to both p- and n-type Skutterudites
increasing the figure of merit up to ZT ³ 2.
As concerns mechanical properties, the application of SPD
methods significantly raises the strength while leaving the elastic
moduliunchanged unless new phases have been formed.
[doi:10.2320/matertrans.MF201941]
(Received March 12, 2019; Accepted June 14, 2019; Published
August 30, 2019)
Keywords: severe plastic deformation, high-pressure torsion,
thermoelectrics, skutterudites, Half Heusler, clathrates,
tellurides, mechanicalproperties
1. Introduction
Thermoelectric (TE) materials are able to directly
convertthermal energy into electrical energy and vice versa.
Todetermine the quality of a TE material, the dimensionlessfigure
of merit ZT = S2T/(μ) is used, with S the Seebeckcoefficient or
thermopower (S ³ ¦V/¦T), T the temperature,μ the electrical
resistivity and = e + ph the total thermalconductivity, consisting
of the electronic part e and thephonon part ph. The electronic part
of the thermalconductivity is linked to the electrical resistivity
via theWiedemann-Franz law, e ³ L0T/μ, with the Lorenz numberL0.
Therefore one way to increase ZT is to decrease ph,which is
possible by enhancing the scattering of the heatcarrying phonons,
on various lattice defects like point defectsincluding vacancies,
dislocations, and grain boundaries. Notonly a high ZT value at a
certain temperature is important,but also a high average ZT, (ZT)a,
over a wide temperaturerange, to maximize the so-called
thermo-electric conversionefficiency
© ¼ Th � TcTh
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1þ ðZTÞap �
1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1þ ðZTÞap þ Tc
Th
ð1Þ
including the Carnot efficiency, where Th and Tc are
thetemperatures on the hot and cold side respectively.
“Bottom-up” methods like ball-milling (BM) followed
byhot-pressing (HP) achieve materials with a grain size of theorder
of micrometers and (especially with high energy ball-milling, HBM)
in the range of 100 nanometers and less. Itwas found that the
smaller the grain size the higher is thethermoelectric performance
due to a reduction in the phononpart of the thermal conductivity.1)
Another method to reducethe grain and/or crystallite size and
concomitantly to increasethe density of lattice defects in general
and, thereby, furtherincrease the scattering of heat carrying
phonons, is to applysevere plastic deformation (SPD) in its various
forms.217)
SPD methods produce materials with grains in sub-micrometer or
even nanometer range.1820) Because of thepresence of high
hydrostatic pressure, SPD methods provide
Materials Transactions, Vol. 60, No. 10 (2019) pp. 2071 to
2085©2019 The Japan Institute of Metals and Materials OVERVIEW
https://doi.org/10.2320/matertrans.MF201941
-
extremely large strains, almost without changing the
sample’sgeometry. Therefore SPD not only achieves very fine
grainswith high and small angle boundaries, but also
vacancies,dislocations and other defects.5,9,2125)
There exist many techniques to sustain plastic strains
asdescribed in an overview article by Valiev.18) Of course, theyall
have advantages and disadvantages.
A very effective method is high-pressure torsion(HPT).2629)
Using HPT, the sample’s shape remains almostunchanged by the
deformation, a finer grain structure isachieved but damage and
fracture are suppressed. Thetechnique is essentially based on the
use of a Bridgmananvil-type device: the thin sample (disk shaped)
is subjectedto a torsional strain under a high pressure between two
anvils.Additionally it is possible to process under various
temper-atures, using cooling or heating with an inductive coil.
Asin some cases it is necessary to prevent the sample
fromoxidation, a modified version of the HPT equipment with atight
“cage” for inert gas can be applied.
Using HPT one has to take into account that the processedsamples
are not completely homogeneous as the shear strain,£, is increasing
from the center of the sample to the rimaccording to £ = (2³nr)/h,
where n is the number ofrevolutions, r is the radius and h is the
thickness of thesample; however, as Pippan et al.30) demonstrated,
even inthe center (r = 0) the strain is not zero.
For measuring the electrical resistivity as well as theSeebeck
coefficient, the equipment requires samples withcuboid shape with
the dimensions of at least 8 © 2 © 1mm3.The thermal conductivity is
measured perpendicularly to theresistivity and the Seebeck
coefficient. As the HPT-processedsamples are discs with 10mm in
diameter and about 1mmin thickness, some inhomogeneities of the
samples must beaccepted. More details of the measurement techniques
can befound in Refs. 5, 6, 9, 15).
The influence of the processing parameters (number
ofrevolutions, applied pressure, temperature) as well as of
thestarting conditions (grain size of the powder before HP)7)
andvarious annealing processes5) on the structural and
physicalproperties have been studied systematically for a
p-typeskutterudite (DD0.60Fe3CoSb12, DD stands for didymium)by Rogl
et al.10) It was found that processing at roomtemperature produced
very brittle samples, almost unservice-able, therefore all further
experiments were performed at600670K. Independent of all
above-mentioned conditions,after HPT the lattice parameter was
enlarged, the relativedensity decreased, the crystallite size
became much smaller(up to fourty times), and the dislocation
density increasedby about ten times. Independent of the applied
pressure(28GPa) and the number of revolutions (15), the
HPT-processed skutterudite samples were not completely homoge-neous
and more or less strewn with microcracks.10) Theseobservations are
in contrast to those of Masuda et al.,17)
claiming that HPT-processing HP Heusler alloys at
roomtemperature with 5GPa and 10 revolutions yielded homoge-neous
and crack-free samples.
Whilst the Seebeck coefficient, independent of thecomposition
and processing parameters, in almost all casesis in the range of
the materials’ state before HPT, theelectrical resistivity, due to
introduced defects such as
dislocations, vacancies and/or micro cracks, is dependenton the
processing parameters and more or less enhanced. Thethermal
conductivity, out of the above-mentioned reasons, isdecreased. The
net effect usually is positive, i.e. as a result,ZT is higher for
all HPT-processed samples.
For a short introduction into the solid state physics
ofthermoelectrics, into the various methods of severe
plasticdeformation as well for a comprehensive summary on
HPT-processed skutterudites the reader is referred to an
earlierreview article by G. Rogl et al.9)
2. Skutterudites
Since 2010 the influence of HPT on structural,
physical,mechanical and magnetic properties of p- and
n-typeskutterudites has been investigated, and the results
werepublished by the authors as a review article9) and in
variousjournals.38,10,11,1316) It could be shown that after
HPT-processing skutterudites do not exhibit any secondary
phasesand/or impurities but microcracks and pores, an
enhancednumber of dislocations and of other defects as well as
smallergrains. These observations were confirmed via
transitionelectron microscopy (TEM): as depicted in Figs. 1 and
2,where grain boundaries and dislocations are clearly visible.
The lattice parameter is larger, the density lower.
Duringmeasurement-induced heating or deliberate annealing,
thegrains grow, the major part of the cracks fuse to pores
ordisappear, the dislocation density becomes smaller, but allthese
temperature induced changes do not restore the stateas it has been
before HPT. These changes are reflected in the
Fig. 1 Superimposed energy-filtered TEM image of
HP+HPT-processedSr0.07Ba0.07Yb0.07Co4Sb12, showing grain boundaries
and dislocationstructure.
Fig. 2 TEM image of a selected area of CP+HPT-processed
DD0.7Fe3-CoSb12, exhibiting a close up of dislocations and grain
boundaries.
G. Rogl, M.J. Zehetbauer and P.F. Rogl2072
-
changes of the physical properties, mainly in the
electricalresistivity and thermal conductivity right after
HPT-process-ing and during the annealing process. Although the
electricalresistivity even after annealing is higher for the HPT
sample,the thermal conductivity is lower, resulting in a
positivenet effect in respect of ZT, generally with enhancements
of1328%.
As the scheme is the same for all p-type (mainly DD filled,Fe/Co
or Fe/Ni substituted) and n-type (multifilled)skutterudites, one of
each was exemplarily selected fordemonstration.
To evaluate the X-ray diffraction data in terms of lineprofile
analysis, models for the crystallite size and itsdistribution as
well as of the dislocations density wereapplied, using the CMWP-fit
software.31,32) For example forp-type, DD0.6Fe3CoSb12, BM, the
crystallite size was reducedby means of HPT from 152 nm to 53 nm,
and the dislocationdensity of 2.8 © 1013m¹3 was ten times larger.5)
Althoughfor the Sb/Sn substituted BM DD0.7Fe2.7Co1.3Sb11.8Sn0.2,
thecrystallite size through HPT was reduced from 336 nm to95 nm, it
grew back to 256 nm during the annealing process,but the residual
strain was about 2.5 times higher afterHPT.16) For n-type
Sr0.07Ba0.07Yb0.07Co4Sb12 the HPTinduced structural changes were
less spectacular, showinga crystallite size reduction from 36.5 nm
to 30 nm, and anenhancement of the dislocation density from 1.4 ©
1014m¹2
to 1.8 © 1014m¹2.5) Multiple filled p- and n-type
skutter-udites, (Sr,Ba,DD,Yb)y(Fe1¹xNix)4Sb12, with ZTs ³
1attracted interest because tuning the band structure and
thelocation of the Fermi level adjusted both p- and n-typematerials
to exhibit the same thermal expansion coefficientand mechanical
properties. After HPT-processing, ZT wasenhanced by 20% for the p-
and 38% for the n-typeskutterudite.8)
For DD0.44Fe2.1Co1.9Sb12 the changes after HPT andannealing were
investigated employing not only transitionelectron microscopy but
also Raman spectroscopy andtexture measurements.15) Raman spectra
of HP, HPT andHPT-annealed DD0.44Fe2.1Co1.9Sb12 showed almost the
samepeak position, with a move to the lower energy side inthe
higher frequency range, indicating a softening of thevibration
modes. The spectrum for the HPT sample exhibitedadditional
softening in the high-energy region as well as peakbroadening,
denoting that the vibration modes related to theshorter SbSb bonds
in the Sb4 rings are more affected thanthose with longer SbSb
bonds. After annealing, thesefeatures were not visible anymore and
the HPT-annealedskutterudite spectrum is undistinguishable from the
HPspectrum. Anbalagan et al.14) gained the same results forRaman
spectroscopy investigations of the double-element-substituted
unfilled skutterudite Fe0.05Co0.95Sb2.875Te0.125.
Texture measurements of DD0.44Fe2.1Co1.9Sb12, comparinga HP
sample with a HPT-processed (for the latter performedon a plane
perpendicular to the HPT pressing direction)did not show any
changes in the orientation distribution ofthe crystallites.
However, for Fe0.05Co0.95Sb2.875Te0.125Anbalagan et al.14) found
changes in the crystallographictexture, which indicated
strengthening of the (112), (102)poles and weakening of the 28
(123) pole of the HPT-processed sample.
The temperature and phase stability of p-type
skutteruditesDD0.7Fe3.1Co0.9Sb12, HP and HPT-processed, have
beenstudied by means of thermal analysis (TA) and Knudseneffusion
mass spectrometry (KEMS).33) The difference inthe mass loss per day
at 800K, due to antimony evaporation(into vacuum), of the HP (34%)
and the HPT-processedsample (56%) is marginal. For the sample,
DD0.7Fe3CoSb12,for which commercial powder (TIAG, Austria) was
simplyhot-pressed (labeled as HP), the lowest mass loss (
-
μ(T) but metallic behavior without anomalies. The temper-ature
dependent thermal conductivity, (T), as displayed inFig. 4, due to
the smaller grains and the defects discussedabove, introduced
during HPT, is much lower for the HPTsample and hardly changes
after annealing. The latticethermal conductivity in all cases is
very low.
Comparing the p- and n-type skutterudite in Figs. 3 and 4,one
can also see for the HP samples that the lower μ(T), thehigher is
(T). After HPT and after annealing, the relativeincrease in μ(T) is
much higher for Sr0.09Ba0.11Yb0.05Co4Sb12than for DD0.6Fe3CoSb12
and this behavior is reflected in(T), with a much bigger downsizing
for Sr0.09Ba0.11Yb0.05-Co4Sb12 than for DD0.6Fe3CoSb12. The
illustration (Fig. 5)defines the temperature dependent Seebeck
coefficient, S(T),which is positive for DD0.6Fe3CoSb12, indicating
holes asmain carriers, but is negative for
Sr0.09Ba0.11Yb0.05Co4Sb12, aselectrons are the main carriers.
Figure 5 informs that neitherHPT-processing nor annealing for the
p- as well as for then-type samples has an influence on S(T), as
all S(T) curvesare, within the error bar, alike. With a more or
less completely
unchanged Seebeck coefficient after HPT and annealing, buta
higher electrical resistivity in both cases, the power factor,pf =
S2T/μ becomes lower, therefore both HP samplesexhibit the highest
values (Fig. 6).
Figure 7 summarizes the figure of merit, ZT. The increaseof ZT
is higher for DD0.6Fe3CoSb12 by 41% for the HPTsample (from ZT =
1.1 to ZT = 1.6 at 823K) and 30%(to ZT = 1.5) than for
Sr0.09Ba0.11Yb0.05Co4Sb12 with 19%for the HPT sample (from ZT = 1.6
to ZT = 1.9 at 838K)and 12% (to ZT = 1.8). These ZT enhancements
can varyindividually: dependent on the starting material and
thecomposition of the skutterudite, so far for all HPskutterudites,
ZT was higher after HPT-processing. Thethermo-electric conversion
efficiency increased for DD0.6Fe3-CoSb12 from © = 13.9% (HP) to © =
14.6% (HP+HPT), forSr0.09Ba0.11Yb0.05Co4Sb12 from © = 15.0% (HP) to
© =16.0% (HP+HPT).
Enhancing ZT via HPT-processing does not only work forfilled p-
and n-type skutterudites, but also for unfilledskutterudites and
filled skutterudites substituted at the Sb-site.
Mallik et al.11) reported for Fe0.2Co3.8Sb11.5Te0.5 that
ZTincreased from ZT = 1.06 to ZT = 1.3, which equals anenhancement
of 23%.
Fig. 4 Temperature dependent thermal conductivity, (T), (big
symbols),and lattice thermal conductivity, ph(T), (small symbols)
of HP, HP+HPTand annealed skutterudites, DD0.6Fe3CoSb12 and
Sr0.09Ba0.11Yb0.05-Co4Sb12.
Fig. 5 Temperature dependent Seebeck coefficient, S(T), of HP,
HP+HPTand annealed skutterudites, DD0.6Fe3CoSb12 (right scale)
andSr0.09Ba0.11Yb0.05Co4Sb12 (left scale).
Fig. 6 Temperature dependent power factor, pf(T), of HP, HP+HPT
andannealed skutterudites, DD0.6Fe3CoSb12 and
Sr0.09Ba0.11Yb0.05Co4Sb12.
Fig. 7 Temperature dependent ZT of HP, HP+HPT and
annealedskutterudites, DD0.6Fe3CoSb12 and
Sr0.09Ba0.11Yb0.05Co4Sb12.
G. Rogl, M.J. Zehetbauer and P.F. Rogl2074
-
Rogl et al.16) HPT-processed Sb-substituted
DD0.54Fe2.7-Co1.3Sb11.9Ge0.1 and DD0.7Fe2.7Co1.3Sb11.8Sn0.2. Whilst
ZT ofDD0.54Fe2.7Co1.3Sb11.9Ge0.1 after HPT was not much higher,for
DD0.7Fe2.7Co1.3Sb11.8Sn0.2 ZT = 1.3 at 780K of the HPsample could
be topped with ZT = 1.45 at 773813K of theHPT-processed sample.
Both ZT values, for a processed anda HP p-type skutterudite are, to
our knowledge, so far thehighest ZTs. The values of © ³ 14.6%
before and after HPTare about the same.
Recently, SPD via HPT at elevated temperatures and inprotective
gas atmosphere was used to directly consolidateand plastically
deform commercial p-type (DD0.7Fe3-CoSb12)34) and n-type
((Mm,Sm)yCo4Sb12)35) skutteruditepowder from TIAG, Austria, into a
dense thermoelectricsolid. Applying this method, not only time and
energyconsuming steps like ball milling and especially hot
pressingcould be eliminated, but in addition this method is
muchfaster as it takes about 15 to 25 minutes, dependent on
thenumber of revolutions during HPT, instead of several hours(about
10). For both, the p- and n-type skutterudite, thestructural,
physical and mechanical properties (see chapter 6)of the cold
pressed (CP) and HPT-processed sample (referredto as CP+HPT 1r or
CP+HPT 5r, dependent on thenumber of revolutions) were compared
with those of areference sample (referred to as HP), which was
traditionallyconsolidated in the hot press. It should be noted that
coldcompacting (CP) of the powder is only necessary in order
toeasily place the right amount of powder between the anvils ofthe
HPT equipment.
Whilst the calculated lattice parameters (Fig. 8, top) forboth,
p- and n-type, are lower for the homogeneous HPsample than for the
starting powder, they are higher afterCP+HPT. After
measurement-induced annealing they“shrink” to the size of those of
the starting powders.Interestingly, the lattice parameters for the
n-type skutter-udite, HPT-processed by five revolutions are
slightly lowerthan those processed with one revolution.
Figure 8 (bottom) compares the measured relative densityin %
with the calculated X-ray density (dX = (ZM)/(NV),where M is the
molar mass, Z is the number of formula unitsper cell, N is
Loschmidt’s number, and V is the volume of theunit cell). For the
p- as well as for the n-type skutterudite theCP+HPT sample has a
much lower density than the HPsample. The reason is that defects
like vacancies and small
cracks are introduced during HPT-processing. After anneal-ing
the density further decreases for DD0.7Fe3CoSb12,whereas for
(Mm,Sm)yCo4Sb12 a slight increase occurs.
The crystallite size (Fig. 9) was evaluated from the
X-raypattern via line profile analysis. The crystallite size of
bothHP samples (p-type: 76 nm, n-type: 78 nm) is practically
thesame and much smaller after HPT-processing (44 nm and45 nm,
respectively, for one revolution, 39 nm for fiverevolutions).
During measurement-induced annealing thegrains grow, but remain
still smaller in comparison to theHP samples. The results of the
profile analysis for thedislocation density is depicted in Fig. 9.
Especially for thesample processed with five revolutions the
dislocationdensity is ten times higher than for the HP sample,
forthe other two samples about four times (3.6 © 1014m¹2 and3.3 ©
1014m¹2). During the physical properties’ measure-ments many
defects anneal out, but not completely so thatfinally for the
annealed samples the dislocation density is stillabout three times
higher in comparison to that of the HPsample.
The temperature dependent electrical resistivity μ(T) forboth,
p- and n-type CP+HPT (see Fig. 10), is much higherthan for the HP
counterpart, descending from various defectsand cracks. For the
p-type, after a maximum, μ(T) decreases,undergoes a low minimum and
slightly increases.
Fig. 8 Lattice parameter, a (top) and relative density, drel
(bottom) ofDD0.7Fe3CoSb12 (left column) and (Mm,Sm)yCo4Sb12 (right
column).
Fig. 9 Crystallite size, cs (top) and dislocation density, dd
(bottom) ofDD0.7Fe3CoSb12 (left column) and (Mm,Sm)yCo4Sb12 (right
column).
Fig. 10 Electrical resistivity, μ, of DD0.7Fe3CoSb12 and
(Mm,Sm)yCo4Sb12prepared via HP and CP+HPT vs. temperature, T.
The Effect of Severe Plastic Deformation on Thermoelectric
Performance of Skutterudites, Half-Heuslers and Bi-Tellurides
2075
-
This behavior was backed by in situ synchrotron measure-ments
for DD0.7Fe3CoSb12, which were performed in atemperature range of
300 to 825K in steps of 50K, in orderto evaluate the changes in
grain size and dislocation densityof the CP+HPT samples before,
during and after anneal-ing.34) With increasing temperature the
size of the crystallites(³44 nm) decreases slightly (³38 nm) but
above about 500Kit increases (³60 nm), but even at 800K the
crystallite size issmaller than that of the HP sample (³78 nm). It
wasconfirmed that at room temperature the grains at the rimare
smaller than in the center, and that this proportion isindependent
of the heat treatments and inversely proportionalto the dislocation
density. Below about 600K the dislocationdensity (3.5 © 10144.2 ©
1014m¹2) is not affected by theheat treatment but is decreasing
almost linearly above 600K.Beyond that temperature, it is then
bisected but still two timeslarger than the dislocation density of
the HP sample. Furtherheat treatments do neither affect the
crystallite size nor thedislocation density. These changes are in
parallel with thechanges in the electrical resistivity for the
first measurementwith increasing temperature.
The thermally stable μ(T)-curve is higher than for the HPsample
but has the same curvature. For the n-type sample forthe first
measurement with increasing temperature, about thesame behavior
occurs, but for decreasing temperature, μ(T)increases, instead of
decreasing, especially for the sampleprocessed with five
revolutions. This performance indicatesa change from metallic to
semiconducting behavior duringmeasurement-induced annealing, which
could be explainedvia the change of lattice distortion and its
influence on theband gap.35) All further resistivity measurements
confirmedsemiconducting behavior.
Thermal conductivity (Fig. 11) for all three CP+HPTsamples is
much lower than for the HP reference sample.After annealing, the
thermal conductivity does not changemuch, but whilst the thermal
conductivity becomes slightlylower for the annealed CP+HPT n-type
sample it becomesslightly higher for the p-type. This feature is
not surprising,as thermal conductivity acts reciprocal to the
electricalresistivity. Lattice thermal conductivities are very
low,
indicating that the lower limit of the possible
thermalconductivity reduction is reached.
The temperature dependent Seebeck coefficient is dis-played in
Fig. 12. For DD0.7Fe3CoSb12 and for(Mm,Sm)yCo4Sb12, the difference
between HP and CP+HPTis marginal; also annealing hardly influences
the Seebeckcoefficient. The power factor (Fig. 13) reaches the
highestvalues for the HP samples, due to high and very
highelectrical resistivities and because of almost no change inthe
Seebeck coefficient.
All ZT values (Fig. 14) are remarkable. With CP+HPT thegoal is,
of course, to get high ZTs, but they should notcompete with the ZTs
of the HP+HPT samples as one alwayshas to consider the
sustainability and energy-savingpreparation procedure of the CP+HPT
samples. The highestZT with ZT ³ 1.44 at 823K could be achieved for
thethermally stable (Mm,Sm)yCo4Sb12 (5 revolutions).DD0.7Fe3CoSb12
has ZT ³ 1.3 at 783K. In addition thethermo-electric conversion
efficiencies for the CP+HPTsamples are in the range of © =
12.215.7%.
L. Zhang et al.36) compared the magnetic behavior
ofPr0.67Fe3CoSb12 before and after HPT. Despite the fact that
Fig. 11 Thermal conductivity, (big symbols) and lattice
thermalconductivity, ph (small symbols) DD0.7Fe3CoSb12 and
(Mm,Sm)yCo4Sb12prepared via HP and CP+HPT vs. temperature, T.
Fig. 12 Seebeck coefficient, S, of DD0.7Fe3CoSb12 (right scale)
and(Mm,Sm)yCo4Sb12 (left scale) prepared via HP and CP+HPT
vs.temperature, T.
Fig. 13 Power factor, pf, of DD0.7Fe3CoSb12 (right scale)
and(Mm,Sm)yCo4Sb12 (left scale) prepared via HP and CP+HPT
vs.temperature, T.
G. Rogl, M.J. Zehetbauer and P.F. Rogl2076
-
after HPT a long-range magnetic order for T < 5.6K couldnot
be detected in the temperature dependent electricalresistivity
curve, the susceptibility indicates anti-ferromag-netism, even
though diminished, revealing that the magneticorder at T < 5.6K
is just superposed by high residualresistivity. Before HPT the
effective magnetic moment is4.18®B, after HPT it amounts to 4.07®B,
accompanied by ametamagnetic transition in the isothermal
magnetizationcurves. The field dependent transitions considerably
smearout as a consequence of the nanograined structure.
For all skutterudites, HPT-processing is a great tool toeither
enhance ZT of HP samples or to directly produce fastand easily
high-ZT thermoelectric materials.
3. Clathrates
Yan et al. investigated the influence of HPT on type Iclathrate
Ba8Cu3.5Ge41In1.5, prepared via high frequencymelting followed by
BM and HP (for details see Ref. 12)).All features, typical of
HPT-processed samples (4GPa, 1revolution at 673K), like reduced
grain size, but enhancedresidual strain, dislocation density, point
defects and cracks,accompanied by a lower density, turned up. As
aconsequence, compared with the BM+HP sample, theHPT-processed
sample had a higher electrical resistivityand Seebeck coefficient
but a lower charge carrierconcentration, lower Hall mobility and
thermal conductivityso that finally no essential improvement of ZT
in theinvestigated temperature range occurred (Fig. 15).
Only one clathrate Ba8Cu3.5Ge41In1.5, was investigated.The
enhanced electrical resistivity was balanced by a reducedthermal
conductivity and slightly higher Seebeck coefficient,therefore ZTs
of the HP and HP+HPT sample were alike.
4. Heusler Alloys
Kourov et al.2) HPT-processed a rapidly quenchedHeusler-type
alloy Ni2.16Mn0.84Ga (2 and 5 revolutions,under a pressure of 3 and
5GPa), and studied the crystallinestructure and behavior of
electrical resistivity, thermoelectricpower, thermal expansion, and
magnetic properties. AfterHPT of the sub-microcrystalline alloy, a
mixture of
amorphous and nanocrystalline (90 vol%) phases wasformed,
however, after annealing at T > 600K the structurechanged first
to nanocrystalline and later back to a sub-microcrystalline one.
SPD was made responsible for the factsthat (i) the temperature
dependent electrical resistivityexhibits a negative temperature
coefficient (ii) the long-rangemagnetic order underwent a
transition to a magneticallyordered state accompanied with a
significant drop in themagnitude of magnetization (iii) the
temperature dependentSeebeck coefficient is proportional to the
temperature in themagnetically ordered state, but that (iv) the
absolute valueof S drops drastically when the magnetic order
vanishes andthat (V) the thermal expansion coefficient is
remarkablyincreased. All these observations indicate a
rearrangement ofthe electronic band structure near the Fermi level
caused bythe HPT treatment.
Whilst the thermoelectric properties helped to get insightinto
the electronic behavior of Ni2.16Mn0.84Ga, the inves-tigation of
Heusler alloys from the system VFeAl focusedon the thermoelectric
performance.
Masuda et al.17) HPT-processed Heusler alloys,VTa0.05Fe2Al0.95
and V1.05Fe2Al0.95, at room temperaturewith 5GPa and 10
revolutions, and observed neither cracksnor differences in hardness
between the center and the rimarea. The structural and physical
properties of the soobtained samples were compared with the
respective arc-melted samples before HPT and after it plus
annealing at873K. They observed some grain size reduction afterHPT,
which almost vanished after the annealing process.The lattice
parameters of both processed alloys measuredas a function of
temperature were enhanced right after HPTbut decreased with
increasing temperature during themeasurement.
The electrical resistivities after HPT are enhanced and showa
rather sharp peak with a decrease due to recovery and graingrowth
during measurement-induced heating, similar to thebehavior of
HPT-processed skutterudites. After recurrence,above 800K,
resistivities are in line with those of the arc-melted samples as
well as with those of annealed ones. Theabsolute values of the
Seebeck coefficient of the HPT sampleswere lower in comparison to
the arc-melted ones butincreased after annealing. Thermal
conductivity is almost
Fig. 15 ZT vs. temperature, T, of Ba8Cu3.5Ge41In1.5, adapted
fromRef. 12).
Fig. 14 ZT of DD0.7Fe3CoSb12 and (Mm,Sm)yCo4Sb12 prepared via
HPand CP+HPT vs. temperature, T.
The Effect of Severe Plastic Deformation on Thermoelectric
Performance of Skutterudites, Half-Heuslers and Bi-Tellurides
2077
-
bisected (about 5W/m) after processing and hardly changesduring
annealing. Element mapping revealed that Ta atomssegregated at the
grain boundaries, and it seems that they wereable to retard the
grain boundary migration and this waysuppress grain growth during
recrystallization. ThereforeVTa0.05Fe2Al0.95 even after annealing
exhibited a grain sizeof about 80 nm, whereas in non-doped
V1.05Fe2Al0.95 thegrains grew back to 270 nm, being a disadvantage
for thethermal conductivity. Therefore ZT (Fig. 16) of the
annealedV1.05Fe2Al0.95 is lower than that of the arc-melted
counterpart,however, ZT of the annealed, HPT-processed
VTa0.05Fe2Al0.95reaches ZT ³ 0.3 at around 500K because of the low
thermalconductivity even after annealing in parallel with
therestoration of the rather large power factor.
For the Heusler-type Ni2.16Mn0.84Ga the comparison of
thebehavior of the thermoelectric power, electrical resistivity,and
magnetic properties indicated a significant rearrangementof the
electronic band structure near the Fermi level due tothe HPT
treatment of the alloy. Whilst for VTa0.05Fe2Al0.95 ahigher ZT
after HPT was detected, this was not the case fornon-doped
V1.05Fe2Al0.95 because during annealing the grainsgrew back to
original size.
5. Half-Heusler Alloys
Although Half-Heusler (HH) alloys reveal increasinginterest as
thermoelectric materials, severe plastic deforma-tion has been
applied only recently. Particularly the group ofRogl et al.37)
investigated the influence of HPT on p-type(NbFeSb and
Ti0.15Nb0.85FeSb) and n-type
(Ti0.5Zr0.5NiSn,Ti0.5Zr0.5NiSn0.98Sb0.02 and Ti0.5Zr0.5NiSn + DA,
where DAstands for densification aid) Half-Heusler (HH) alloys.
AllHP samples were processed at 650K, applying 4GPa and 1revolution
under argon to prevent oxidation. X-ray diffractionpatterns
revealed peak broadening for all HPT samples,indicating finer
grains and/or an enhanced dislocationdensity. After
measurement-induced heating the peaksbecame slimmer. Lattice
parameters became larger afterHPT, whereas the relative density
became lower. Forskutterudites, the second measurement for
increasing temper-ature showed all physical properties being
similar to those
measured in the first measurement with decreasing temper-ature;
for HH alloys, however, at least three times cyclingis necessary to
gain a thermally stable, annealed sample,indicating that for HH
alloys the introduced defects andstructure changes seem to be more
resistant against heattreatments.
Both samples NbFeSb and Ti0.15Nb0.85FeSb were preparedas
described in detail in Ref. 38). ZTs of processed andannealed
samples were enhanced (Figs. 17 and 18,respectively).
Interestingly, the HPT-processed NbFeSb exhibits anelectrical
resistivity lower than that of the HP sample,already for the first
measurement; at room temperature theannealed sample has a
resistivity of about five times lowerthan that of the HP; but at
823K the values are about thesame. Seebeck coefficient of the HP
sample exhibits a cross-over from negative to positive values at
around 455K andafter undergoing a maximum (at 623K), it changes
back tonegative values at around 505K (insert in Fig. 17).
Thesechanges are also present after severe plastic deformation
viaHPT. As the Seebeck coefficient of the HPT annealed sampleis
slightly lower in comparison with the HP sample, and thethermal
conductivity is almost three times lower, ZT at 623Kreaches ZT =
0.00043. This is, as absolute value, very low,but as relative value
double the ZT-value of the HP sample.
Fig. 16 ZT vs. temperature, T, of VTa0.05Fe2Al0.95 and
V1.05Fe2Al0.95,adapted from Ref. 17).
Fig. 17 ZT vs. temperature, T, of NbFeSb. Insert: S(T) of
NbFeSb.
Fig. 18 ZT vs. temperature, T, of Ti0.15Nb0.85FeSb.
G. Rogl, M.J. Zehetbauer and P.F. Rogl2078
-
Ti0.15Nb0.85FeSb behaves like a typical HPT-processedalloy.
Electrical resistivity is - right after HPT - much higher,shows a
maximum and decreases. After annealing theresistivity is only
slightly higher than after hot-pressing.The Seebeck coefficient
does not change within the error bar,but thermal conductivity is
much lower, so that ZTs of theHPT sample at 823K are higher than
the ZT of the HP(ZT = 0.68) sample with ZT = 0.71 and ZT = 0.74
(for thefirst measurement and the annealed sample,
respectively).
Figures 1922 summarize the thermoelectric propertiesfor all
three investigated n-type HH alloys,
Ti0.5Zr0.5NiSn,Ti0.5Zr0.5NiSn0.98Sb0.02 and Ti0.5Zr0.5NiSn + DA,
however,not all measured temperature-dependent curves are
presented:only the graphs of the HP (the reference), of the
HPT-processed (first measurement, temperature increasing) and
theadditionally annealed sample (after cycling) are presented.All
samples, arc-melted, annealed, BM and HP, wereprepared as described
in detail in Ref. 39). For all threeHH alloys (Fig. 19), the
electrical resistivity after HPT isvery high, presents a maximum
between 500 and 600K andis furtheron decreasing with increasing
temperature. ForTi0.5Zr0.5NiSn0.98Sb0.02 and Ti0.5Zr0.5NiSn + DA
after threetimes cycling, μ(T) of the annealed and of the HP sample
areshowing the same curvature with higher values for theannealed
one. For these two compounds, the absolute values
of the Seebeck coefficient after HPT and cycling are lowerthan
those for the HP sample. The undoped sampleTi0.5Zr0.5NiSn without
DA, displays an extremely highresistivity right after
HPT-processing, but after measure-ment-induced heating it is even
lower than for the HP sample.The absolute values of the Seebeck
coefficient ofTi0.5Zr0.5NiSn after HPT and annealing are about 60%
lowerthan those for the HP sample (Fig. 20).
For all three HH alloys, the thermal conductivity (Fig. 21)is
decreased although not by the same extent. The highestenhancement
of ZT (see Fig. 22), almost 20%, is exhibitedby Ti0.5Zr0.5NiSn with
ZT = 0.81 at 773K. For Ti0.5Zr0.5-NiSn + DA, however, the HP sample
has a higher ZT thanthe as-processed one, mainly because of a
rather low Seebeckcoefficient.
For p- and n-type HH alloys, at least three temperaturecycles
are necessary to gain a thermally stable state,indicating that
obviously the introduced defects and structurechanges are more
resistant against heat treatments than incase of skutterudites. For
both p-type HH alloys investigated,as well as for two of the three
n-type HH alloys, ZT wasenhanced after HPT in comparison to the HP
counterparts.
SEM images (Fig. 23), exemplarily shown for the HHalloy system
Ti0.15Nb0.85FeSb, compare the microstructure ofthe fracture
surfaces of the HP, HP+HPT and the HP+HPT
Fig. 19 Electrical resistivity, μ, vs. temperature, T, for
n-type HH alloys.
Fig. 22 ZT vs. temperature, T, for n-type HH alloys.Fig. 20
Seebeck coefficient, S, vs. temperature, T, for n-type HH
alloys.
Fig. 21 Thermal conductivity, , vs. temperature, T, for n-type
HH alloys.
The Effect of Severe Plastic Deformation on Thermoelectric
Performance of Skutterudites, Half-Heuslers and Bi-Tellurides
2079
-
annealed samples. Grain sizes of the HP sample ranging from0.1
to about 10 µm are reduced to less than 2 µm after HPT.However,
during annealing the grains grow and show a ratherhomogeneous size.
TEM images (Fig. 24) confirm thesefindings and show that the
dislocation density in the HPTsample is enhanced by almost two
orders of magnitude(for details on the thermoelectric HH phase
TixNb1¹xFeSb,please see Ref. 37)).
6. Bismuth Tellurides
This type of non-skutterudite alloys belongs to the groupof
semiconductors and shows its maximum of the figure ofmerit ZT
typically around T = 400500K, mainly due tothe maximum of
thermoelectric power and the minimum ofphonon thermal conductivity
at this temperature.
Attempts to improve ZT mainly concern macroscopicdeformation
processes. Already with elastic deformation,some tuning of the band
gap is possible especially when thedeformation induces a phase
transition. The paper ofOvsynnianikov et al.40) is an example the
effects of whichcan follow from applying elastic hydrostatic
pressuresbetween 225GPa in Bi2Te3 and Sb2Te3: whilst pressurestill
4GPa already enhance the power factor PF due to atransition from
semiconductor to metal characteristics, therhombohedral R�3m phase
of these compounds undergoes
a phase transition into a monoclinic C2/m lattice.
Thistransition, however, deteriorates PF in case of p-type
Bi-tellurides, while it increases it continuously till pressures
of25GPa in case of Se-doped n-type tellurides because of
acontinuous increase of electrical conductivity. Biswas et al.also
present a simple technical model with diamond anvils fora compact
thermoelectric high-pressure module, whichprovides both a permanent
pressure and a heat sink.40)
First ideas for applying plastic deformation for the sakeof
improvement of ZT in those alloys were to re-orient theTEs’ crystal
lattice by plastic deformation for the sake ofincreasing the
electrical conductivity i.e. achieve a texturealong (00ℓ) and thus
maximize conductivity. The first realexperiments with application
of plastic deformations goback to the nineteen-nineties when Seo et
al.41) applied hotextrusion (HE) to p-type Te-doped Bi0.5Sb1.5Te3
and n-typeSbI3-doped Bi2Te2.85Se0.15 thermoelectric compounds,
within573713K. Besides the improvement by doping with Te andSb, the
figure of merit could be improved by a factor 3, byapplying HE and
increasing the HE temperature, arriving atZT = 0.87 at RT, probably
mainly due to the increase ofelectrical conductivity.
One of the first works trying generation of lattice defects(i.e.
grain boundaries by nanocrystallization) in thesesemiconductor
materials was that by Yu and colleagues in200942) who reached a ZT
= 0.94 at 398K and/or at least
Fig. 23 SEM micrographs (images of secondary electrons) of the
fracturesurfaces of Ti0.15Nb0.85FeSb, HP (a), HPT (b) and HPT after
annealing (c).
Fig. 24 TEM micrographs of Ti0.15Nb0.85FeSb, HP (a), HPT (b) and
HPTafter annealing (c).
G. Rogl, M.J. Zehetbauer and P.F. Rogl2080
-
ZT = 0.7 between 325525K, through BM, cold pressingand sintering
between 473773K. This ZT was markedlyhigher than the former
record-figure-of-merit achieved byZone Melting (ZM)43) being ZT ³
0.75 only.
A significant increase of ZT in these materials waspresented
also in 2008 by Poudel et al.44) and Ma et al.45)
who reached a peak ZT = 1.4 at 373K in p-type BixSb2¹xTe3bulk
ternary alloys through reductions in thermal con-ductivity by means
of nanocrystallization and defectgeneration. The latter was
achieved by suitable ball millingand high pressure
consolidation.
The first real use of an SPD method has been done by Imet al.46)
in 2004 who chose a multi-pass ECAE (called alsoECAP) method at a
temperature of 773K. However, the ZTvalues reached by Seo et al.41)
could not be overcome. Thesame is true with the ECAE experiments
performed by Fanet al.47) in 2008, as well as by Hayashi et al.
carried out in200648) and 201049) although in the latter work they
tried tooptimize the ECAE route with regard to a (00ℓ) texture
forthe sake of maximizing the electrical conductivity. In
2008another treatise concerning ECAE processed
Bi0.5Sb1.5Te3compound was undertaken by Lim et al.50) at
temperatures653733K arriving at a still considerable grain size of
about10 µm which explains that the figure of merit reached thesame
values as did the previous ECAP works i.e. ZT = 0.9at RT.
A much better ZT value could be achieved by the groupof Sun et
al.,51) Fig. 25 reaching ZT = 1.16 at RT afterusing a modified ECAP
method also at 773K but with arotary die, after having sintered the
powder of (Bi,Sb)2Te3alloys. Although the Seebeck coefficient was
decreased byECAP, the electrical conductivity was strongly
increasedwhat led to an overall increase of ZT by 10% compared
withthe ball milled & pulse discharge sintered material.
The second SPD method applied to V-VI and/or V2-VI3alloys has
been High Pressure Torsion (HPT), by M. Ashidaet al.52,53,55) and
T. Hamachiyo et al.54) Investigations havebeen done with the
stoichiometric alloy Bi0.5Sb1.5Te3.0produced by ball milling (BM)
or Vertical Brigdman Method(VBM), sintered by hot pressing at 673K,
and then HPT-processed at 473K under a pressure of 6GPa and rotated
byspeeds of 0.11 rpm. A combination of VBM and HPT usinghigh HPT
rates yielded power factors (PFs) as low as 1 to4 © 10¹3Wm¹1
K¹2.52) However, modifying the processingroute by starting with BM
followed by a low-rate HPT of0.1 rpm provided a power factor being
around 6 © 10¹3
Wm¹1 K¹2 within a temperature interval from 320470K55)
(see also Fig. 26). This increase of PF could be
attributedpartially to the low-rate HPT’s texture evolution
whichproceeded along (00ℓ), and partially to the enhanced
Seebeckcoefficient,55) Fig. 26. Unfortunately, the works5255) did
notpresent the thermal conductivity, which is necessary toreliably
estimate the ZT-value resulting from these experi-ments. According
to the authors’ promising experiencesfrom Skutterudites (see this
article) and from the results fromother techniques of plastic
deformation applied to thetellurides mentioned above,44,45,51) one
can hope that HPTcould be very successful in further enhancing the
ZT valuethrough the generation of lattice defects acting as
additionalscattering centers for further decreases of the
thermal
Fig. 25 Increases of the electrical conductivity (upper graph)
and of thepower factor (lower graph) through increasing numbers x
of ECAP passesEPx, by measurements in transversal (T, full symbols)
and longitudinal(L, empty symbols), with increasing measuring
temperature, in a(Bi,Sb)2Te3 alloy (from Sun et al., Ref. 51)).
Fig. 26 BM and HPT-processed BiSbTe showing an increase in
electricconductivity (upper graph) and an enhancement of power
factor(lower graph) due to low-rate HPT-processing (from Ashida et
al.,Ref. 55)).
The Effect of Severe Plastic Deformation on Thermoelectric
Performance of Skutterudites, Half-Heuslers and Bi-Tellurides
2081
-
conductivity and thus increases of ZT. Among the numerousmethods
of plastic deformation, HPT seems to provide thehighest number of
advantages: (1) HPT is the most powerfulmethod achieving a maximum
of lattice defects, not at leastthanks to the enhanced pressure
which allows for very largestrains without developing cracks and
failures; (2) HPTcan achieve bulk nanocrystalline samples directly
frompowders; (3) HPT does not introduce a strong texture or under
certain conditions can achieve (00ℓ) textures atelevated processing
temperatures and low processing rateswhich maximize the electrical
conductivity in those Te-composites.
There have been done also some but only few treatises toimprove
ZT in PbTe, e.g. by Biswas et al. in 2012,56) whodoped this alloy
with Na, or with SrTe and Na, and appliedprocessing by Spark Plasma
Sintering (SPS) where thesample is quickly molten and sintered
under pressure. Withthis processing technique, small scale second
phase particleswere formed with both phase boundaries for small
particleswith strain fields around them, as well as with
misfitboundary dislocations at the larger ones could be
generated.Such defects have been intentionally produced in order
toprovide additional scatterers for phonon scattering thusgiving a
minimum of lattice thermal conductivity. In thebest case, values of
figure of merit till ZT = 2.2 at 900Kwere reached.
In what follows, more recent developments are describedwhich
concern the specific scattering properties of the variouslattice
defects. A marked step herein was given by the paperof Kim et
al.57) which emphasizes the high scattering powerof dislocations
(especially closely neighboured ones): In
contrast to point defects as well as grain boundaries
whichpreferably scatter high and low frequency
phonons,respectively, the dislocations cover the broad
intermediaterange of phonons with medium frequency (see Fig. 27).
Byaimed introduction of such dislocations through liquid
phasecompacting process of melt-spun eutectic compound
ofstoichiometric Bi0.5Sb1.5Te3 with pure Te, Kim et al. showedthat
the ZT value can be increased up to ZT = 1.89 (Fig. 27)compared to
ZT = 0.91 of the ingot or BM initial materialand to ZT = 1.4
reported by the works of Poudel44) and Ma45)
using BM and Hot Pressing (HP) for these p-type BiSbTecompounds.
It has been shown by the authors of this article16)
that a nanostructure with a regular dislocation array in
thegrain boundaries indeed gives the lowest lattice
thermalconductivities possible (near to the theoretical minimum)
thusenabling records in ZT of 1.5 for p- and almost 2.0 for
n-typeskutterudites. That arrangement has been achieved by
hightemperature HPT. In a paper of Park & Lee,58) published
oneyear after that of Kim,57) they recommend to apply this
SPDtechnique also to BiSeTe compounds. Recently an extensivework on
n-type V2VI3 alloys was published by Hu et al.59)
who for the first time report the optimization of TE
propertiesby applying a special way of hot deformation (HD,
called“progressive”). By choosing a large number of HD countsbut
not too high HD temperatures (at maximum 723K) toaccount for a low
electrical conductivity as well as for asufficient number of
lattice defects including dislocations andgrain boundaries, at a
grain size of only 60 nm, they reachedrecord values of ZT = 1.3 in
Sb-doped n-type Bi, at T =470K (Fig. 28).
Fig. 27 A: Spectral lattice thermal conductivity ¬s(f ) with
different contributions of phonon scattering. Dislocation
scattering is operativeover the whole frequency range, in contrast
to boundaries and point defects which mainly scatter phonons with
low or high frequencies,respectively. B, C: Lattice thermal
conductivity ¬lat (left bottom, B) and figure of merit zT (right,
C) for melt-solidified (BM), solid-phasecompacted (S-MS), and
liquid-phase compacted (Te-MS) Bi0.5Sb1.5Te3 alloys. Only the
latter contains dislocations embedded in grainboundaries which
cause a large overall phonon scattering, thus minimizing the
lattice thermal conductivity and maximizing the figure ofmerit
(from Kim et al., Ref. 57)).
G. Rogl, M.J. Zehetbauer and P.F. Rogl2082
-
7. Mechanical Properties of Skutterudites and Half-Heusler
Alloys
So far for many HPT-processed materials, mechanicalproperties
have been measured. The details about theequipments, measurements
and evaluations of the data aredescribed in Refs. 3, 60, 61). Here
only room temperaturedata are discussed, and only examples, typical
of the generalbehavior, are presented here.
The values of the elastic moduli, independent of thepreparation
steps (HP+HPT, HP+HPT-annealed or CP+HPT, CP+HPT-annealed) of
HPT-processed skutteruditesare not much different from those of the
respective HP ones.Also a slight change in the samples’ density has
not muchinfluence. For example, all Young’s moduli (E) for
HP,CP+HPT and CP+HPT-annealed p-type DD0.7Fe3CoSb12are in the range
of 136GPa ¯ E ¯ 144GPa;34) for ball-milled and HP DD0.6Fe3CoSb12, E
= 150GPa, after HPT onegets E = 153GPa, which is practically the
same valueconsidering 3% uncertainty. Also for (Mm,Sm)yCo4Sb12there
is no change in E within the error bar, independentof the number of
revolutions, with all values 142GPa ¯E ¯ 145GPa.35)
The values of hardness of HP skutterudites and of HHalloys are,
besides the composition, strongly dependent onthe density. As shown
in Fig. 29, generally n-type HPskutterudites are harder than p-type
skutterudites for a largevariation of density. After
HPT-processing, the grains aremuch smaller and, as a consequence of
the Hall-Petchrelation, hardness must be higher, which is indeed
the casefor the p- and n-type skutterudites, independent of
thepreparation method HP+HPT or CP+HPT. Hardness of HPTsamples also
depends on the processing conditions, i.e.samples processed with
more than one revolution have muchhigher values, especially
prominent for (Mm,Sm)yCo4Sb12prepared by five revolutions (sample
CP+HPT-5r). As HPT-processed samples exhibit a gradient in strain
along theradius, the hardness depends on the distance from
thesample’s center as depicted in Fig. 30. For
example,DD0.6Fe3CoSb12 with an overall density of 97% after
HPT(4GPa, 1 revolution, 623K) exhibited a hardness of HV =498 in
the center section of the sample and HV = 522 at therim.5) It is
interesting to note that (i) the hardness of HH
alloys generally is much higher than that of skutterudites
and(ii) the difference in HV between center and rim area is
largerfor the HH alloys. These observations are in contrast to
thoseof Masuda et al.,17) as they claimed that no difference in
thehardness between rim and center occurred when theymeasured their
HP+HPT-processed (5GPa, 10 revolutions)Heusler alloys.
Besides all these facts, for HP+HPT samples the
startingconditions play a role. Hand milled (HM) and ball
milledsamples of DD0.44Fe2.1Co1.9Sb12 were HPT-processed underthe
same conditions (4GPa, 1 revolution, 623K). TheHM+HP sample had a
density of 95.2%, and HV = 393,which changed to 94% and a much
higher hardness HV =479 (rim) after HPT resulting in an enhancement
of 22%. TheBM+HP sample had a density of 98.3% and HV = 410.After
HPT, HV was increased to 562 (rim), which equals37%, Ref. 7).
Generally after HPT and HPT annealing, no mentionablechanges in
the elastic moduli occurred. Due to the changingstrain from the
center to the rim of a circular sample, hardnessvalues are higher
in the rim than in the center area but in anycase higher than those
of the HP sample although theirdensity is lower.
Fig. 29 Hardness, HV of various skutterudites vs. relative
density, drel, HPand HPT-processed. Most data of HP skutterudites
are adopted fromRef. 61).
Fig. 30 Hardness, HV (left scale) and in GPa (right scale) of HP
and HPT-processed (measured in the center, middle and rim area of
the circularsample) skutterudites and Half-Heusler alloys.
Fig. 28 Effect of Sb-alloying, and of progressive hot
deformation (HD) tothe zT-value in the n-type BiSbTe compounds
indicated (after Hu et al.,Ref. 59)).
The Effect of Severe Plastic Deformation on Thermoelectric
Performance of Skutterudites, Half-Heuslers and Bi-Tellurides
2083
-
8. Thermal Expansion of Skutterudites
Thermal expansion exhibits a quite unusual behavior
afterHPT-processing, related to the behavior of the
temperaturedependent electrical resistivity.6,7,9,15) As all
temperaturedependent curves of the thermal length change exhibit
thesame features, exemplarily the thermal expansion
ofDD0.6Fe3CoSb12 is shown (Fig. 31) and discussed.
In the low temperature range (4.2300K), no anomaliesfor the HPT
sample are visible, however, above 300K thecurve is not strictly
linear anymore, and at ³443K instead offurther expanding, the
sample contracts. At about 470K theexpansion continues. The anomaly
occurs in about the sametemperature window in which the electrical
resistivity startsto decrease after a maximum; obviously the
‘shrinking’ ofthe sample is connected with the annealing of
defectsand/or disappearance of cracks. As can be seen in Fig.
31,with decreasing temperature as well as for a second run
withincreasing temperature, no anomalies appear. It should beadded
that thermal expansion was not only measured in thepressing
direction as imaged in Fig. 31, but also perpendic-ular to it,6)
showing the same behavior. All thermal expansioncoefficients of
HPT-processed samples hardly differ fromthose of the HP samples and
are listed in Ref. 9).
9. Conclusions
This article is to show that SPD-processing of thermo-electrics
can provide values of figure of merit (ZT) higherthan 1.2, and thus
gets highly attractive for materials whichenable sustainable
generation of electric power. These highvalues arise from the
ultrafine grained microstructure incombination with a high level of
point, linear and surfacelattice defects which significantly
enhances the scatteringof the phonons, thus leading to a minimum of
thermalconductivity and therefore to exceptional values of ZT.
Theextent of SPD-induced enhancement depends on the
specialthermoelectric materials. So far, the following
thermo-electrics have been tried to get improved by SPD
processing:(1) From all thermoelectrics, n- and p-type
skutterudites
benefit most from SPD-processing especially if oneapplies high
pressure torsion (HPT) to already hot-
pressed or still powdered materials thus producingthermoelectric
bulks. Significant enhancements offigure-of-merit ZT for n-type
skutterudites from 1.6 toalmost 2, and for p-type skutterudites
from 1.15 toalmost 1.5 at 825K could be achieved.
(2) HPT-processing of Heusler and Half-Heusler phasesyields
about the same relative improvement of ZT (0.22to 0.3) as
skutterudites, although the absolute valuesstill appear limited.
However, SPD-induced increasesof ZT achieved in Clathrates were
found to be nearlynegligible.
(3) Low temperature thermoelectrics like Bismuth Tellur-ides of
type V1(VI)1 and/or V2(VI)3 (with V, VIrepresenting the group
elements) show a considerablepotential in SPD-induced increases of
ZT. Already withhigh temperature pressure after ball milling, ZT
could beenhanced from ZT ³ 1.0 up to ZT = 1.4 at T = 373K,at least
by ensuring a high electrical conductivity whileconsidering the
strong anisotropy in these materials.Therefore the choice of the
best SPD-processingtechnique includes the careful optimization of
texture,but also that of the processing rate and temperature.Recent
investigations underline the importance of thelattice defects’
specific phonon scattering efficiencies -especially those of
dislocations - and their generation insufficiently high densities,
thus reaching enhancementsof p-type Bi-Tellurides up to ZT =
1.9.
As concerns the mechanical properties of SPD-skutteruditesand
Half-Heusler-alloys, the application of SPD methodssignificantly
raises the strength while the elastic moduliremain unchanged unless
new phases have been formed.
Acknowledgements
The authors thank O. Eibl and J. Bursik for contributingthe SEM
and TEM images.
REFERENCES
1) G. Rogl, A. Grytsiv, P. Rogl, E. Bauer, M. Hochenhofer, R.
Anbalagan,R.C. Mallik and E. Schafler: Acta Mater. 76 (2014)
434448.
2) N.I. Kourov, V.G. Pushin, A.V. Korolev, V.A. Kazantsev,
E.B.Marchenkova and A.N. Uksusnikov: Phys. Met. Metallogr.
103(2007) 270277. Original text in Russian: Fiz. Metal.
Metalloved.108 (2007) 280287.
3) L. Zhang, G. Rogl, A. Grytsiv, S. Puchegger, J.
Koppensteiner, F.Spieckermann, H. Kabelka, M. Reinecker, P. Rogl,
W. Schranz, M.Zehetbauer and M.A. Carpenter: Mater. Sci. Eng. B 170
(2010) 2631.
4) G. Rogl, M. Zehetbauer, M. Kerber, P. Rogl and E. Bauer:
Mater. Sci.Forum 667669 (2011) 10891094.
5) G. Rogl, D. Setman, E. Schafler, J. Horky, M. Kerber, M.
Zehetbauer,M. Falmbigl, P. Rogl, E. Royanian and E. Bauer: Acta
Mater. 60 (2012)21462157.
6) G. Rogl, Z. Aabdin, E. Schafler, J. Horky, D. Setman, M.
Zehetbauer,M. Kriegisch, O. Eibl, A. Grytsiv, E. Bauer, M.
Reinecker, W. Schranzand P. Rogl: J. Alloys Compd. 537 (2012)
183189.
7) G. Rogl, D. Setman, E. Schafler, J. Horky, M. Kerber, M.
Zehetbauer,M. Falmbigl, P. Rogl and E. Bauer: in New Materials for
Thermo-electric Applications: Theory and Experiment, NATO Science
forPeace and Security Series B: Physics and Biophysics, ed. by V.
Zlatıcand A. Hewson, (Springer Science+Business Media Dordrecht,
NL,2013) 7, pp. 8091.
8) G. Rogl, A. Grytsiv, E. Royanian, P. Heinrich, E. Bauer, P.
Rogl, M.Zehetbauer, S. Puchegger, M. Reinecker and W. Schranz: Acta
Mater.
Fig. 31 Thermal expansion of DD0.6Fe3CoSb12 before and after
HPT,below and above room temperature. Note: the graphs of the 1st
run Tdecreasing and the 2nd run T increasing are shifted for better
visibility.
G. Rogl, M.J. Zehetbauer and P.F. Rogl2084
https://doi.org/10.1016/j.actamat.2014.05.051https://doi.org/10.1134/S0031918X07030076https://doi.org/10.1134/S0031918X07030076https://doi.org/10.1134/S0031918X07030076https://doi.org/10.1016/j.mseb.2010.02.022https://doi.org/10.4028/www.scientific.net/MSF.667-669.1089https://doi.org/10.4028/www.scientific.net/MSF.667-669.1089https://doi.org/10.1016/j.actamat.2011.12.023https://doi.org/10.1016/j.actamat.2011.12.023https://doi.org/10.1016/j.jallcom.2012.05.011https://doi.org/10.1016/j.actamat.2013.03.031
-
61 (2013) 40664079.9) G. Rogl, P. Rogl, E. Bauer and M.
Zehetbauer: in Thermoelectric
Nanomaterials, ed. by K. Kuomoto and T. Mori, Springer Series
inMater. Sci., 182, (2013) pp. 193254.
10) G. Rogl, A. Grytsiv, P. Rogl, E. Royanian, E. Bauer, J.
Horky, D.Setman, E. Schafler and M. Zehetbauer: Acta Mater. 61
(2013) 67786789.
11) R.C. Mallik, R. Anbalagan, G. Rogl, E. Royanian, P.
Heinrich, E.Bauer, P. Rogl and S. Suwas: Acta Mater. 61 (2013)
66986711.
12) X. Yan, M. Falmbigl, G. Rogl, A. Grytsiv, A. Prokofiev, E.
Bauer, P.Rogl, M. Zehetbauer and S. Paschen: J. Electron. Mater. 42
(2013)13301334.
13) G. Rogl, A. Grytsiv, P. Rogl, N. Peranio, E. Bauer, M.
Zehetbauer andO. Eibl: Acta Mater. 63 (2014) 3043.
14) R. Anbalagan, G. Rogl, M. Zehetbauer, A. Sharma, P. Rogl, S.
Suwasand R.C. Mallik: J. Electron. Mater. 43 (2014) 38173823.
15) G. Rogl, A. Grytsiv, J. Horky, R. Anbalagan, E. Bauer, R.C.
Mallik, P.Rogl and M. Zehetbauer: Phys. Chem. Chem. Phys. 17 (2015)
37153722.
16) G. Rogl, A. Grytsiv, P. Heinrich, E. Bauer, P. Kumar, N.
Peranio, O.Eibl, J. Horky, M. Zehetbauer and P. Rogl: Acta Mater.
91 (2015) 227238.
17) S. Masuda, K. Tsuchiya, J. Qiang, H. Miyazaki and Y.
Nishino: J. Appl.Phys. 124 (2018) 035106.
18) R.Z. Valiev, Y. Estrin, Z. Horita, T.G. Langdon, M.J.
Zehetbauer andY.T. Zhu: JOM 58 (2006) 3339.
19) M.J. Zehetbauer and Y. Estrin: in Bulk Nanostructured
Materials, ed.by M.J. Zehetbauer and Y.T. Zhu, (Wiley VCH,
Weinheim, 2009)pp. 109135.
20) R.Z. Valiev, Y. Estrin, Z. Horita, T.G. Langdon, M.J.
Zehetbauer andY.T. Zhu: JOM 68 (2016) 12161226.
21) M. Zehetbauer and V. Seumer: Acta Metall. Mater. 41 (1993)
577588.22) M. Mueller, M. Zehetbauer, A. Borbely and T. Ungar: Scr.
Mater. 35
(1996) 14611466.23) T. Ungar and M. Zehetbauer: Scr. Mater. 35
(1996) 14671473.24) M. Zehetbauer, H.P. Stuewe, A. Vorhauer, E.
Schafler and J. Kohout:
Adv. Eng. Mater. 5 (2003) 330337.25) M. Zehetbauer, J. Kohout,
E. Schafler, F. Sachslehner and A.
Dubravina: J. Alloys Compd. 378 (2004) 329334.26) P.W. Bridgman:
Studies in Large Scale Plastic Flow and Fracture, ed.
by P.W. Bridgman, (McGraw-Hill, New York, NY, USA, 1952).27)
R.Z. Valiev, A.V. Korznikov and R.R. Mulyukov: Mater. Sci. Eng.
A
168 (1993) 141148.28) A.P. Zhilyaev, G.V. Nurislamova, B.K. Kim,
M.D. Baro, J.A. Szpunar
and T.G. Langdon: Acta Mater. 51 (2003) 753765.29) N.A.
Smirnova, V.I. Levit, V.I. Pilyugin, R.I. Kuznetsov, L.S.
Davydova and V.A. Sazonova: Fiz. Met. Metalloved. 61
(1986)11701177.
30) R. Pippan: in Bulk Nanostructured Materials, ed. by M.J.
Zehetbauerand Y.T. Zhu, (Wiley VCH, Weinheim, 2009) pp. 217232.
31) M. Kerber, M. Zehetbauer, E. Schafler, F. Spieckermann, S.
Bernstorffand T. Ungar: JOM 63 (2011) 6170.
32) J. Lohmiller, R. Baumbusch, M. Kerber, A. Castrup, H. Hahn,
E.Schafler, M. Zehetbauer, O. Kraft and P.A. Gruber: Mech. Mater.
67(2013) 6573.
33) P. Brož, F. Zelenka, Z. Kohoutek, J. Vřešťál, V. Vykoukal,
J. Buršík, A.Zemanová, G. Rogl and P. Rogl: submitted to Phys.
Chem. Chem.Phys.
34) G. Rogl, A. Grytsiv, R. Anbalagan, J. Bursik, M. Kerber, E.
Schafler,M. Zehetbauer, E. Bauer and P. Rogl: Acta Mater. 159
(2018) 352363.
35) G. Rogl, K. Yubuta, M. Kerber, A. Grytsiv, M. Zehetbauer, E.
Bauerand P. Rogl: submitted to Acta Mater. (2019).
36) L. Zhang, A. Grytsiv, B. Bonarski, M. Kerber, D. Setman, E.
Schafler,P. Rogl, E. Bauer, G. Hilscher and M. Zehetbauer: J.
Alloys Compd.494 (2010) 7883.
37) G. Rogl, A. Grytsiv, J. Bursik, S. Gosh, M. Kerber, E.
Bauer, R.C.Mallik, M. Zehetbauer and P. Rogl: in submission
(2019).
38) A. Tavassoli, F. Failamani, A. Grytsiv, G. Rogl, P.
Heinrich, H. Müller,E. Bauer, M. Zehetbauer and P. Rogl: Acta
Mater. 135 (2017) 263276.
39) M. Guerth, G. Rogl, V.V. Romaka, A. Grytsiv, E. Bauer and P.
Rogl:Acta Mater. 104 (2016) 210222.
40) S.V. Ovsyannikov, M.V. Morozova, I.V. Korobeinikov,
L.N.Lukyanova, A.Y. Manakov, A.Y. Likhacheva, A.I. Ancharov,
A.P.Vokhmyanin, I.F. Berger, O.A. Usov, V.A. Kutasov,
V.A.Kulbachinskii, T. Okada and V.V. Shchennikov: Appl. Phys.
Lett.106 (2015) 143901.
41) J. Seo, C. Lee and K. Park: Mater. Sci. Eng. B 54 (1998)
135140.42) F. Yu, J. Zhang, D. Yu, J. He, Z. Liu, B. Xu and Y.
Tian: J. Appl. Phys.
105 (2009) 094303.43) T.M. Tritt: Science 283 (1999) 804805.44)
B. Poudel, Q. Hao, Y. Ma, Y.C. Lan, A. Minnich, B. Yu, X. Yan,
D.Z.
Wang, A. Muto, D. Vashaee, X.Y. Chen, J.M. Liu, M.S.
Dresselhaus,G. Chen and Z.F. Ren: Science 320 (2008) 634638.
45) Y. Ma, Q. Hao, B. Poudel, Y.C. Lan, B. Yu, D.Z. Wang, G.
Chen andZ.F. Ren: Nano Lett. 8 (2008) 25802584.
46) J.-T. Im, K.T. Hartwig and J. Sharp: Acta Mater. 52 (2004)
4955.47) X.A. Fan, J.Y. Yang, W. Zhu, S.Q. Bao, X.K. Duan, C.J.
Xiao and
K. Li: J. Alloys Compd. 461 (2008) 913.48) T. Hayashi, M.
Sekine, J. Suzuki and Y. Horio: Mater. Trans. 47 (2006)
19411944.49) T. Hayashi, Y. Horio and H. Takizawa: Mater. Trans.
51 (2010) 1914
1918.50) C.H. Lim, K.T. Kim, Y.H. Kim, C.H. Lee and C.H. Lee:
Mater. Trans.
49 (2008) 889891.51) Z.M. Sun, H. Hashimoto, N. Keawprak, A.B.
Ma, L.F. Li and M.W.
Barsoum: J. Mater. Res. 20 (2005) 895903.52) M. Ashida, T.
Hamachiyo, K. Hasezaki, H. Matsunoshita, M. Kai and
Z. Horita: J. Phys. Chem. Solids 70 (2009) 10891092.53) M.
Ashida, T. Hamachiyo, K. Hasezaki, H. Matsunoshita and Z.
Horita:
Adv. Mater. Res. 8991 (2010) 4146.54) T. Hamachiyo, M. Ashida,
K. Hasezaki, H. Matsunoshita, M. Kai and
Z. Horita: Mater. Trans. 50 (2009) 15921595.55) M. Ashida, N.
Sumida, K. Hasezaki, H. Matsunoshita and Z. Horita:
Mater. Trans. 53 (2012) 588591.56) K. Biswas, J. He, I.D. Blum,
C.-I. Wu, T.P. Hogan, D.N. Seidman, V.P.
Dravid and M.G. Kanatzidis: Nature 489 (2012) 414418.57) S.I.
Kim, K.H. Lee, H.A. Mun, H.S. Kim, S.W. Hwang, J.W. Roh, D.J.
Yang, W.H. Shin, X.S. Li, Y.H. Lee, G.J. Snyder and S.W.
Kim:Science 348 (2015) 109114.
58) J.G. Park and Y.H. Lee: Curr. Appl. Phys. 16 (2016)
12021215.59) L. Hu, Y. Zhang, H. Wu, Y. Liu, J. Li, J. He, W. Ao,
F. Liu, S.J.
Pennycook and X. Zeng: Adv. Funct. Mater. 28 (2018) 1803617.60)
G. Rogl, A. Grytsiv, M. Gürth, A. Tavassoli, C. Ebner, A.
Wuenschek,
S. Puchegger, V. Soprunyuk, W. Schranz, E. Bauer, H. Mueller,
M.Zehetbauer and P. Rogl: Acta Mater. 107 (2016) 178195.
61) G. Rogl and P. Rogl: Sci. Adv. Mater. 3 (2011) 517538.
The Effect of Severe Plastic Deformation on Thermoelectric
Performance of Skutterudites, Half-Heuslers and Bi-Tellurides
2085
https://doi.org/10.1016/j.actamat.2013.03.031https://doi.org/10.1016/j.actamat.2013.07.052https://doi.org/10.1016/j.actamat.2013.07.052https://doi.org/10.1016/j.actamat.2013.07.032https://doi.org/10.1007/s11664-012-2440-8https://doi.org/10.1007/s11664-012-2440-8https://doi.org/10.1016/j.actamat.2013.09.039https://doi.org/10.1007/s11664-014-3179-1https://doi.org/10.1039/C4CP05230Ghttps://doi.org/10.1039/C4CP05230Ghttps://doi.org/10.1016/j.actamat.2015.03.008https://doi.org/10.1016/j.actamat.2015.03.008https://doi.org/10.1063/1.5034390https://doi.org/10.1063/1.5034390https://doi.org/10.1007/s11837-006-0213-7https://doi.org/10.1007/s11837-016-1820-6https://doi.org/10.1016/0956-7151(93)90088-Ahttps://doi.org/10.1016/S1359-6462(96)00319-3https://doi.org/10.1016/S1359-6462(96)00319-3https://doi.org/10.1016/S1359-6462(96)00320-Xhttps://doi.org/10.1002/adem.200310090https://doi.org/10.1016/j.jallcom.2004.01.039https://doi.org/10.1016/0921-5093(93)90717-Shttps://doi.org/10.1016/0921-5093(93)90717-Shttps://doi.org/10.1016/S1359-6454(02)00466-4https://doi.org/10.1007/s11837-011-0115-1https://doi.org/10.1016/j.mechmat.2013.04.010https://doi.org/10.1016/j.mechmat.2013.04.010https://doi.org/10.1016/j.actamat.2018.08.020https://doi.org/10.1016/j.jallcom.2010.01.042https://doi.org/10.1016/j.jallcom.2010.01.042https://doi.org/10.1016/j.actamat.2017.06.011https://doi.org/10.1016/j.actamat.2015.11.022https://doi.org/10.1063/1.4916947https://doi.org/10.1063/1.4916947https://doi.org/10.1016/S0921-5107(97)00285-7https://doi.org/10.1063/1.3120865https://doi.org/10.1063/1.3120865https://doi.org/10.1126/science.283.5403.804https://doi.org/10.1126/science.1156446https://doi.org/10.1021/nl8009928https://doi.org/10.1016/j.actamat.2003.08.025https://doi.org/10.1016/j.jallcom.2007.07.007https://doi.org/10.2320/matertrans.47.1941https://doi.org/10.2320/matertrans.47.1941https://doi.org/10.2320/matertrans.M2010065https://doi.org/10.2320/matertrans.M2010065https://doi.org/10.2320/matertrans.MEP2007297https://doi.org/10.2320/matertrans.MEP2007297https://doi.org/10.1557/JMR.2005.0120https://doi.org/10.1016/j.jpcs.2009.06.002https://doi.org/10.4028/www.scientific.net/AMR.89-91.41https://doi.org/10.2320/matertrans.E-M2009807https://doi.org/10.2320/matertrans.ME201101https://doi.org/10.1038/nature11439https://doi.org/10.1126/science.aaa4166https://doi.org/10.1016/j.cap.2016.03.028https://doi.org/10.1002/adfm.201803617https://doi.org/10.1016/j.actamat.2016.01.031https://doi.org/10.1166/sam.2011.1181