Materials Transactions, Vol. 53, No. 4 (2012) pp. 588 to 591
Special Issue on Growth of Ecomaterials as a Key to Eco-Society V
© 2011 The Japan Institute of Metals
Effects of Low Rotational Speed on Crystal Orientation of Bi2Te3-Based
Thermoelectric Semiconductors Deformed by High-Pressure Torsion
Maki Ashida1,+1, Natsuki Sumida2,+2, Kazuhiro Hasezaki2,
Hirotaka Matsunoshita1,+1 and Zenji Horita1
1
Department of Materials Science and Engineering, Faculty of Engineering, Kyushu University, Fukuoka 819-0395, Japan
Department of Materials Science, Faculty of Science and Engineering, Shimane University, Matsue 690-8504, Japan
2
Bi2Te3-based thermoelectric semiconductors were deformed by high-pressure torsion (HPT) using a low rotational speed of 0.1 rpm, which
is less than the speed of 1 rpm used in our previous studies. The effects of different rotational speeds were investigated by metallographic and
thermoelectric studies. Sample disks of p-type Bi0.5Sb1.5Te3.0 were cut from sintered compacts made by mechanical alloying (MA) followed by
hot-pressing. The disks were deformed by HPT with 1, 3, and 5 turns at 473 K under 6.0 GPa of pressure at a rotational speed of 0.1 rpm. The
preferred orientation was investigated using X-ray diffraction. The orientation factors of the disks changed from 0.054 for pre-rotation up to
0.653 for post-rotation samples. The maximum power factor of the disk using 5 turns and a speed of 0.1 rpm was 6.6 © 10¹3 W m¹1 K¹2 at
363 K, which was larger than the reported power factors of 4.3 © 10¹3 W m¹1 K¹2 for a disk using 5 turns and a speed of 1 rpm, and
4.6 © 10¹3 W m¹1 K¹2 for melt-grown materials. Slow deformation by HPT was found to enhance the electrical conductivities and Seebeck
coefficients of Bi2Te3-based thermoelectric semiconductors by producing a preferred orientation and grain refinement.
[doi:10.2320/matertrans.ME201101]
(Received June 9, 2011; Accepted November 14, 2011; Published December 28, 2011)
Keywords: high-pressure torsion, Bi2Te3, rotational speed, preferred orientation, electrical conductivity
1.
Introduction
Intermetallic compounds such as Bi2Te3-based materials,
FeSi2, PbTe, and ZnSb are known to be suitable for use as
thermoelectric materials.1­3) These materials are eco-materials because they recover exhaust heat. Bi2Te3-based materials
have the highest thermoelectric performance at around room
temperature and are widely used in electronic cooling
devices. The performance of thermoelectric materials is
estimated using the figure of merit, Z, which is defined as
follows:1)
Z ¼ ¡2 ·¬1 ¼ P¬1
ð1Þ
where ¡ is the Seebeck coefficient (V K¹1); · is the electrical
conductivity (³¹1 m¹1); ¬ is the thermal conductivity
(W m¹1 K¹1); and P is the power factor (W m¹1 K¹2). To
enhance the Z-value, it is necessary to increase the Seebeck
coefficient and the electrical conductivity, as well as decrease
the thermal conductivity.
The crystal structure of Bi2Te3-based materials is rhombohedral with the R3m space group, where the hexagonal
lattice (shown by the a- and c-axes) is expressed by a
sequence of atomic layer units, ­Te(1)­Bi­Te(2)­Bi­Te(1)­,
along the c-axis.4) The Te(1)­Bi and Te(2)­Bi bonds are
strong iono­covalent, and the Te(1)­Te(1) bonds are weak
van der Waals.5) The c-plane acts as a preferential slip-plane
caused by the Te(1)­Te(1) bonds. The thermoelectric
properties in the c-plane are superior to those along the caxis. The anisotropy of electrical conductivity is estimated to
be · k =· ? 3 for values parallel and perpendicular to the
c-axis,6) but the Seebeck coefficient is almost isotropic.7)
Bi2Te3-based materials with a high degree of the preferred
orientation have been reported elsewhere.8) Isotropic fine
+1
Graduate Student, Kyushu University
Graduate Student, Shimane University
+2
grains result in an enhanced Seebeck coefficient.9) The
decreased thermal conductivity is a result of phonon
scattering at grain boundaries. Control of crystal orientation
and fine-grain size is known to enhance thermoelectric
performance.
High-pressure torsion (HPT) introduced by Bridgman10)
is known to be an outstanding technique for producing
ultrafine-grain materials and textures with preferred orientations by shear deformation.11­14) The HPT process is
characterized by torsion of a disk around its cylindrical axis,
with collapse of the disk being prevented by a large
hydrostatic pressure.14) The textures of Bi2Te3-based materials deformed by HPT have more preferred orientations and
finer grains than those of materials produced by mechanical
alloying (MA) followed by hot-pressing.15) The power factor
for a post-HPT disk obtained using a rotational speed of
1 rpm has been reported to be 4.3 © 10¹3 W m¹1 K¹2.8,15)
Over-HPT using 10 turns was found to cause a decrease in
preferred orientations, leading to a low power factor.15) In
contrast, deformation at a lower strain speed is known to
control grain-boundary sliding and to suppress increases in
defect densities in grains to a greater extent than is seen at
higher strain speeds.16) The strain speed is equivalent to the
rotational speed in the HPT process. Furthermore, Todaka
et al. showed that the sample temperature during HPT
processing increases with increasing rotation and rotational
speed.17) These facts suggest that the decrease in the preferred
orientation in over-HPT Bi2Te3-based materials is caused by
recrystallization, and that HPT using low rotational speeds
leads to a high degree of preferred orientation.
In the present study, the effects of HPT deformation, using
a lower rotational speed, on the crystal orientation in Bi2Te3based materials were investigated. The sample disks were
cut from sintered compacts with a nominal composition of
Bi0.5Sb1.5Te3.0 prepared by MA followed by hot-pressing
(pre-HPT). After the HPT deformation, the preferred
Effects of Low Rotational Speed on Crystal Orientation of Bi2Te3-Based Thermoelectric Semiconductors Deformed by High-Pressure Torsion
D D0
;
1 D0
ð3Þ
where D0 is the degree of orientation for a random
orientation. In the present study, D0 is 0.043 in the 2ª range
from 10 to 70°, as determined using the Rietveld method.19)
According to eq. (3), F is equal to 1 for a perfect c-plane
preferred orientation structure.
The thermoelectric properties were measured using a
ResiTest8340 (Toyo Corporation, Tokyo, Japan). The Hall
coefficient RH and the electrical conductivity · of the in-plane
direction of the disks were measured by the van der Pauw
method, under a vacuum, using square wave bipolar current I
of 50 mA with the pulse duration of 16.7 msec and a magnetic
field B of 0.5 T in the temperature range 300­473 K.
The van der Pauw method is assumed the isotropic sample.
However, the electrical conductivities were measured in
whole samples and not taken account of anisotropic. The
Hall coefficient RH was evaluated using the equation RH =
VHdB¹1I¹1, where VH is the Hall voltage in a direction
All the disks had p-type conduction. XRD patterns in the
2ª range from 10 to 70° for the pre- and post-HPT disks are
shown in Fig. 1. All the observed diffraction peaks are
identified as being from a single phase of a Bi2Te3­Sb2Te3
solid solution. The patterns for the post-HPT disks show
strong (00l) peaks, indicating that the c-plane was preferentially oriented. Figure 2 shows the orientation factor F of
the (00l) planes plotted against the number of turns (N). The
orientation factor of 0.054 for the pre-HPT (N = 0) sample
corresponds to isotropic orientation. The orientation factors
for the post-HPT (N = 1, 3, 5) samples reached around 0.6.
These results indicate that after only 1 turn the HPT process
had resulted in almost complete preferred orientation of the cplane by shear deformation.8,15) The present maximum value
of 0.653 at N = 3 using a rotational speed of 0.1 rpm was
large compared to the previously reported value of 0.619
at N = 5 with a rotational speed of 1 rpm.15) Furthermore,
the post-HPT (N = 1, 3, 5) samples are expected to have
the same level of grain refinement as in previous reports.15)
0021
F¼
Results and Discussion
0018
where ­I00l and ­Ihkl are the sums of the diffraction
intensities of the (00l) and (hkl) reflections in the range of
the observed Bragg angle 2ª. The orientation factor was
estimated using eq. (3):
3.
0015
The samples of Bi2Te3-based materials had a nominal
composition of Bi0.5Sb1.5Te3.0; an excess of 0.07 mass% Te
was added to control the carrier concentration. The
constituent elements Bi (5N), Sb (6N), and Te (6N) were
weighed according to target compositions in a glove box,
placed in stainless steel vessels, and milled with silicon
nitride balls to prevent metal contaminations. The MA were
used a Fritsch P-5. The capacity of ball milling vessel was
approximately 500 cm3. The weight ratio of the balls to raw
materials was 20 : 1. MA was carried out dry milling for 30 h
at a maximum speed of 180 rpm under dry argon atmosphere.
The resultant MA powder was set in a graphite die and
sintered by hot-pressing at 673 K under a mechanical
pressure of 39 MPa in an argon atmosphere. The sintered
compacts had a diameter of 10 mm and a thickness of 10 mm.
The disks for HPT were cut from sintered compacts and had
a diameter of 10 mm and a thickness of 0.8 mm. The HPT
deformation was carried out in air at 473 K under a pressure
of 6.0 GPa using a rotational speed of 0.1 rpm. The turning
numbers imposed on the disks were 1, 3, and 5.
The preferred orientations of the pre- and post-HPT disks
were characterized by XRD using CuK¡ radiation in the 2ª
range 10 to 70°, with the disks set perpendicular to the
diffraction vector. The preferred orientation was estimated by
the Lotgering method.18) The degree of orientation D was
evaluated using eq. (2):
X
I00l
D¼X
;
ð2Þ
Ihkl
009
015
Experimental
006
2.
perpendicular to square wave bipolar current I, d is thickness
of the disks. The Hall voltage VH was measured the pulse
amplitude by delta voltage method. The influences of the
Peltier effect were eliminated from the measurement of the
Hall voltage. If the Bi2Te3 compounds have preferred
orientations, a more complicated band model, which takes
account of the anisotropic parameters for carrier concentrations n, with Hall tensors and resistivity tensors, must be
used instead of n = 1/eRH, where e is the elementary electron
charge.20) It was difficult to measure the Hall and resistivity
tensor components for the HPT disks, so anisotropic
parameters were not adopted in the present study. The
Seebeck coefficient ¡ of the in-plane direction of the disks
was estimated using ¡ = E/¦T, the linear relationship
between the thermoelectromotive force (E) and the temperature difference (¦T) within 3 K, and was measured under
a vacuum in the temperature range 300­473 K. The power
factor P of thermoelectric performance was evaluated using
P = ¡2·.
Intensity (arb.unit)
orientations were characterized by X-ray diffraction (XRD),
and the thermoelectric properties of the resultant disks were
measured.
589
HPT
5 turns
HPT
3 turns
HPT
1 turn
pre-HPT
0 turn
10
20
30
40
50
60
70
Diffraction angle, 2θ / degree
Fig. 1 XRD patterns of p-type Bi0.5Sb1.5Te3.0 processed by pre-HPT
(N = 0) and post-HPT (N = 1, 3, 5); N is the number of turns. The main
(015) and (00l) peaks are shown.
590
M. Ashida, N. Sumida, K. Hasezaki, H. Matsunoshita and Z. Horita
Hall coefficient R H / 10-7m3 C-1
Orientation factor, F
1
0.5
0
0
1
3
5
Number of Turns, N
10
HPT:5 turns
HPT:3 turns
HPT:1 turns
pre-HPT:0 turn
8
6
4
2
0
400
300
Fig. 4 Hall coefficient versus temperature for p-type Bi0.5Sb1.5Te3.0
processed by pre-HPT, and by post-HPT with N = 1, 3, and 5; N is the
number of turns.
6
4
2
0
500
Temperature, T / K
HPT:5 turns
HPT:3 turns
HPT:1 turns
pre-HPT 0 turn
300
400
500
Temperature, T / K
Fig. 3 Electrical conductivity versus temperature for p-type Bi0.5Sb1.5Te3.0
processed by pre-HPT, and by post-HPT with N = 1, 3, and 5; N is the
number of turns.
Figure 3 show plots of electrical conductivity versus
temperature for the pre- and post-HPT disks. The electrical
conductivities were measured in whole samples and not taken
account of anisotropic. The electric conductivity did not
depend on the turn number of the HPT. The electrical
conductivities of the post-HPT disks increased, as did that of
the pre-HPT disk. The gradients of the electrical conductivity
curves for the post-HPT disks were slightly larger than those
previously reported for pre-HPT and post-HPT disks
deformed at 1 rpm.15) The increasing of the electrical
conductivity for post-HPT disks is consistent with the
preferred orientation.6) The different results for those postHPT disks indicate that the anisotropic thermoelectric
properties were caused by different preferred orientations.
These two results on the deformations produced using 1 and
0.1 rpm indicate that the damage done to the electrical
conductivity might not increase, even if the HPT deformation
was carried out using a low rotational speed, for N = 1, 3,
and 5.
Figure 4 shows plots of Hall coefficient versus temperature
for the pre- and post-HPT disks. In the Fig. 4, it was
illustrated the direction of the turn of HPT, magnetic field,
square wave bipolar current, and Hall voltage of disks,
respectively. The Hall voltage was measured in a direction
perpendicular to square wave bipolar current. The temperature-dependent behaviors indicate that the HPT disks had a
similar preferred orientation. For more rigorous discussions,
the anisotropic carrier concentrations and Hall mobilities
must be adopted in the case of materials with preferred
Seebeck coefficient, α / μVK -1
Electric conductivity, σ / 10 4 Ω -1m -1
Fig. 2 Orientation factors plotted against the number of turns (N) for
p-type Bi0.5Sb1.5Te3.0 processed by pre-HPT and post-HPT.
500
400
300
200
HPT:5 turns
HPT:3 turns
HPT:1 turns
pre-HPT 0 turn
100
0
300
400
500
Temperature, T / K
Fig. 5 Seebeck coefficient versus temperature for p-type Bi0.5Sb1.5Te3.0
processed by pre-HPT, and by post-HPT with N = 1, 3, and 5; N is the
number of turns.
orientations. It was difficult to measure the Hall and
resistivity tensor components of the HPT disks, so anisotropic parameters were not adopted in the present study.
Figure 5 shows plots of Seebeck coefficient versus
temperature for the pre- and post-HPT disks. The Seebeck
coefficients of all the post-HPT disks were larger than that of
the pre-HPT disk. In classical theory, the Seebeck coefficient
is expressed by eq. (4), obtained by Ioffe et al.:21)
(
)
3
k
2ð2³m kT Þ 2
r þ 2 þ ln
¡¼
;
ð4Þ
e
h3 n
where k is Boltzmann’s constant (J K¹1); r is the scattering
factor; m* is the carrier effective mass (kg); and h is Planck’s
constant (J s). According to eq. (4), the Seebeck coefficient
depends on the scattering factor r. The Bi2Te3-based materials
deformed by HPT were reported to have finer grains than
those of materials produced by MA followed by hotpressing.15) This indicates that the large Seebeck coefficients
of the post-HPT materials might be induced by enhancement
of the scattering factor as a result of formation of ultrafine
grains by HPT deformation.9,15) The dispersion of the
Seebeck coefficient might be depended on the scattering
factor for the temperature.
Figure 6 shows plots of power factor versus temperature.
The maximum power factors for pre- and post-HPT disks for
N = 1, 3, and 5 were approximately 4.1 © 10¹3 W m¹1 K¹2
Power factor, α 2σ / 10-3 Wm-1K-2
Effects of Low Rotational Speed on Crystal Orientation of Bi2Te3-Based Thermoelectric Semiconductors Deformed by High-Pressure Torsion
10
HPT:5 turns
HPT:3 turns
8
HPT:1 turns
pre-HPT 0 turn
6
4
2
0
300
400
500
Temperature, T / K
Fig. 6 Power factor versus temperature for p-type Bi0.5Sb1.5Te3.0 processed
by pre-HPT, and by post-HPT with N = 1, 3, and 5; N is the number of
turns.
at 313 K, 6.0 © 10¹3 W m¹1 K¹2 at 433 K, 6.1 © 10¹3
W m¹1 K¹2 at 323 K, and 6.6 © 10¹3 W m¹1 K¹2 at 363 K.
The power factors of all the post-HPT disks were found to
be higher than that of the pre-HPT disk. The available data
on power factors were 4.3 © 10¹3 W m¹1 K¹2 at 443 K for
a post-HPT disk using 1 rpm,15) and 4.6 © 10¹3 W m¹1 K¹2
at 300 K for melt-grown materials without plastic deformation.22) The high power factors of the post-HPT disks are
available to use a heat recovery thermoelectric generation.
These results indicate that deformation by HPT using a
low rotational speed can enhance electrical conductivity by
producing fine and preferentially oriented grains. It must be
noted that the enhanced Seebeck coefficients of the post-HPT
disks might have been induced by control of the scattering
factor by formation of ultrafine grains.9)
4.
Conclusions
In the present study, the effects of using a low rotational
speed of 0.1 rpm in HPT processing of Bi2Te3-based
thermoelectric semiconductors were investigated. The results
are as follows:
(1) The orientation factors for all the post-HPT disks
(N = 1, 3, 5) were saturated at 0.5­0.7. The preferred
orientation obtained for the HPT process using one turn
was almost completely c-plane. The value of 0.653 at
N = 3 with a rotational speed of 0.1 rpm was slightly
larger than the orientation factor of 0.619 obtained for a
post-HPT disk using a rotational speed of 1 rpm.
(2) For N = 5, all the post-HPT disks had a maximum
power factor of approximately 6.6 © 10¹3 W m¹1 K¹2.
591
The values for all the post-HPT disks with a rotational
speed of 0.1 rpm were larger than the value of 4.3 ©
10¹3 W m¹1 K¹2 obtained for the post-HPT disk with
a rotational speed of 1 rpm, and the value of 4.6 ©
10¹3 W m¹1 K¹2 obtained for melt-grown materials.
(3) Deformation by HPT using a lower rotational speed
is effective in obtaining Bi2Te3-based thermoelectric
semiconductors with preferred orientations. The postHPT disks produced using a lower rotational speed
give a higher thermoelectric performance than that of
melt-grown materials.
REFERENCES
1) D. M. Rowe (ed.): Thermoelectrics handbook: macro to nano,
(CRC Press, London, New York, Tokyo, 2006) Chap. 1.
2) Y. Shinohara and Y. Isoda: J. Japan Inst. Metals 71 (2007) 869­875.
3) T. Ueda, C. Okamura and K. Hasezaki: Mater. Trans. 50 (2009) 2473­
2475.
4) D. M. Rowe (ed.): Thermoelectrics handbook: macro to nano,
(CRC Press, London, New York, Tokyo, 2006) Chap. 27-3.
5) S. Nakajima: J. Phys. Chem. Solids 24 (1963) 479­485.
6) J. Nagano, M. Ferhat, E. Hatta and K. Musaka: Phys. Status Solidi B
219 (2000) 347­349.
7) H. T. Kaibe, M. Sakata and I. A. Nishida: J. Phys. Chem. Solids 51
(1990) 1083­1087.
8) M. Ashida, T. Hamachiyo, K. Hasezaki, H. Matsunoshita, M. Kai and
Z. Horita: Mater. Sci. Forum 584­586 (2008) 1006­1011.
9) K. Hasezaki, T. Hamachiyo, M. Ashida, T. Ueda and Y. Noda: Mater.
Trans. 51 (2010) 863­867.
10) P. W. Bridgman: Phys. Rev. 48 (1935) 825­847.
11) G. Sakai, Z. Horita and T. G. Langdon: Mater. Sci. Eng. A 393 (2005)
344­351.
12) C. Xu, S. V. Dobatkin, Z. Horita and T. G. Langdon: Mater. Sci. Eng. A
500 (2009) 170­175.
13) M. Kai, Z. Horita and T. G. Langdon: Mater. Sci. Eng. A 488 (2008)
117­124.
14) Y. Harai, M. Kai, K. Kaneko, Z. Horita and T. G. Langdon: Mater.
Trans. 49 (2008) 76­83.
15) M. Ashida, T. Hamachiyo, K. Hasezaki, H. Matsunoshita and Z. Horita:
Adv. Mater. Res. 89­91 (2010) 41­46.
16) T. Sakai, H. Miura, A. Goloborodko and S. Sitdikov: Acta Mater. 57
(2009) 153­162.
17) Y. Todaka, M. Umemoto, A. Yamazaki, J. Sasaki and K. Tsuchiya:
Mater. Trans. 49 (2008) 7­14.
18) F. K. Lotgering: J. Inorg. Nucl. Chem. 9 (1959) 113­123.
19) F. Izumi: The Rietveld Method, ed. by R. A. Young (Oxford University
Press, Oxford, 1995) Chap. 13.
20) H. Kaibe, Y. Tanaka, M. Sakata and I. Nishida: J. Phys. Chem. Solids
50 (1989) 945­950.
21) K. Uemura and I. A. Nishida: Thermoelectric semiconductor and its
applications, (Nikkankougyou Shinbunsya, Tokyo, 1988) p. 150.
22) K. Uemura and I. A. Nishida: Thermoelectric semiconductor and its
applications, (Nikkankougyou Shinbunsya, Tokyo, 1988) p. 179.