Effect of titanium carbide addition on the thermoelectric properties of B4C ceramics

Article abstract of DOI:10.1016/S0038-1098(00)00245-3

TiB₂/B₄C composite ceramics are prepared via hot pressing, in which the TiB₂ particles are formed by introducing TiC_0.78 and through the reaction between B₄C and TiC_0.78. The electrical and thermal conductivities and Seebeck coefficient of samples containing 0, 12.5 and 25.4 vol% TiB₂ are measured from room temperature up to 1200 K. The results show that the transport properties of the samples vary with the TiB₂ content. The transport properties of a 12.5 vol% TiB₂/B₄C sample are dominated by the B₄C matrix as in undoped B₄C ceramics, whilst the transport properties of a 25.4 vol% TiB₂/B₄C sample are dominated by TiB₂ particles. The figure of merit of the 25.4 vol% TiB₂/B₄C sample is higher than that of the undoped B₄C ceramic sample between room temperature and about 700 K, which indicates that it is possible to improve the thermoelectric properties by selecting an optimum combination of different materials.

Full text of DOI:10.1016/S0038-1098(00)00245-3

PERGAMON
Solid State Communications 115 (2000) 523–526
www.elsevier.com/locate/ssc
Effect of titanium carbide addition on the thermoelectric properties
of B4C ceramics
b
d
c
K.-f. Caia,c, , C.-W. Nan , Y. Paderno , D.S. McLachlan
*
aState Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan 430070,
People’s Republic of China
bDepartment of Materials Science and Engineering, Tsinghua University, Beijing 100084, People’s Republic of China
cDepartment of Physics, University of the Witwatersrand, Wits 2050, Johannesburg, South Africa
dInstitute for Problems of Materials Science, National Academy of Sciences of Ukraine, 03142 Kiev, Ukraine
Received 17 August 1999; received in revised form 29 May 2000; accepted 1 June 2000 by M.F. Collins
Abstract
TiB2/B4C composite ceramics are prepared via hot pressing, in which the TiB2 particles are formed by introducing TiC0.78 and
through the reaction between B4C and TiC0.78. The electrical and thermal conductivities and Seebeck coefficient of samples
containing 0, 12.5 and 25.4 vol% TiB2 are measured from room temperature up to 1200 K. The results show that the transport
properties of the samples vary with the TiB2 content. The transport properties of a 12.5 vol% TiB2/B4C sample are dominated by
the B4C matrix as in undoped B4C ceramics, whilst the transport properties of a 25.4 vol% TiB2/B4C sample are dominated by
TiB2 particles. The figure of merit of the 25.4 vol% TiB2/B4C sample is higher than that of the undoped B4C ceramic sample
between room temperature and about 700 K, which indicates that it is possible to improve the thermoelectric properties by
selecting an optimum combination of different materials. ᭧ 2000 Elsevier Science Ltd. All rights reserved.
Keywords: A. Semiconductors; D. Electronic transport
1. Introduction
Much progress has been made [1,2]. But none of these
approaches have yet made a breakthrough.
Recently, there has been renewed interest in thermoelec-
tric materials [1,2]. The efficiency of a thermoelectric mate-
rial depends on its material properties through the
dimensionless figure of merit ZT ꢀZT ˆ sa2T=k; where
s, a, k, and T are the electrical conductivity, Seebeck
coefficient, thermal conductivity and temperature, respec-
tively). For both applications in refrigeration and power
generation, the efficiency increases monotonically with ZT
[3]. So far the highest ZT value of the thermoelectric mate-
rials has been about 1, which leads to rather low efficiencies.
Hence, until now thermoelectric materials have been only
used in applications where reliability or convenience is
more important than economy. In order to increase the
value of ZT, many approaches, such as novel alloys [4]
and low-dimensional structures [5], have been developed.
In the present work, an approach using composite thermo-
electric materials, namely B4C-based composites, is inves-
tigated experimentally. According to an effective-medium
theory (EMT) [6], the ZT of a matrix-based particulate
composite is always less than that the biggest ZT of each
component of the composite, which implies that there can be
no improvement in ZT for such particulate composites.
However, the EMT prediction may be invalid for the
nonmatrix-based composites, such as those with percolation
cluster structures. In these systems, if one phase has much
higher electrical conductivity than the matrix, the electrical
conductivity of the composite can change by several orders
of magnitude for a very small change in the volume fraction
of the high conducting phase near its percolation threshold
(fc) [7,8]. Therefore, if the Seebeck coefficient and thermal
conductivity for the composite do not change rapidly near
fc, it is possible to get a higher value of ZT for the compo-
site than that for the matrix.
* Corresponding author. Tel.: ϩ27-11-7176835; fax: ϩ27-11-
3398262.
Boron-rich carbides (e.g. B6.5C) are considered as
E-mail address: cai@physnet.phys.wits.ac.za (K. Cai).
0038-1098/00/$ - see front matter ᭧ 2000 Elsevier Science Ltd. All rights reserved.
PII: S0038-1098(00)00245-3

524
K. Cai et al. / Solid State Communications 115 (2000) 523–526
103
102
101
100
10-1
20
25.4%
18
16
14
12
10
8
25.4%
12.5%
0%
12.5%
0%
200
400
600
800 1000 1200
200 400 600 800 1000 1200
T(K)
T(K)
Fig. 1. Electrical conductivity of the samples as a function of
temperature.
Fig. 3. Thermal conductivity of the samples as a function of
temperature.
potential materials for very high temperature thermoelectric
energy conversion [9]. Compared with boron-rich carbides,
B4C has a lower electrical conductivity, a little higher ther-
mal conductivity and almost the same Seebeck coefficient,
and thus its ZT is lower than that of the boron-rich carbides.
If the electrical conductivity of B4C could be increased
significantly, while its Seebeck coefficient and thermal
conductivity do not change very much, B4C would exhibit
a larger ZT value [10]. In this paper, we select B4C as the
thermoelectric (semiconductor) phase and more highly
conducting (metallic) TiB2 as the second phase. TiB2 is
introduced into the B4C by adding TiC0.78 and then through
the reaction between TiC0.78 and B4C [11]:
2. Experimental
B4C powder (average particle size 0.5 mm, Moudanjiang
boron carbide powder plant, China) was homogenized with
various amounts of nonstoichiometric TiC0.78 (average parti-
cle size 5 mm, Zhuzhou cemented carbide plant, China) by
milling the powders together for 10 h. Cylindrical samples
were prepared by hot pressing at 1900ЊC and 35 MPa for
30 min under argon. The pure boron carbide sample was hot
pressed at the same temperature for the same time, but at a
higher pressure (39 MPa) to get a lower porosity. The elec-
trical and thermal conductivities and Seebeck coefficient
were measured up to 1200 K in vacuum. A two-probe
method, using a chopped direct current, was employed to
measure the electrical conductivity. Seebeck coefficients
were determined from the linear relationship between the
thermal electromotive force and temperature difference
between the two ends of the sample. Thermal conductivity
was measured by using the common laser flash method. The
apparent porosity of the samples was determined by the
method given by the American Society of Testing and Mate-
rials Criterion (ASTMC375)[12].
TiC0:78 ϩ B4C ! TiB2 ϩ C;
ꢀ1†
which occurs at the high temperature, at which the compo-
site was sintered.
300
250
0%
200
150
100
3. Results and discussion
12.5%
50
From the XRD results, it can be seen that the TiC0.78 is
completely converted into TiB2. It also shows that the TiB2
content increases with the amount of TiC0.78 added and that
there is very little carbon content compared to the amounts
of B4C and TiB2. The apparent porosity of the samples
containing 0, 12.5, and 25.4 vol% TiB2 is 13.0, 14.0 and
0.82%, respectively. This indicates that adding TiC0.78 can
be an effective method of improving the densification of
B4C ceramics. From scanning electron microscopy
0
25.4%
-50
600
800
T(K)
1000
1200
Fig. 2. Seebeck coefficient of the samples as a function of
temperature.

K. Cai et al. / Solid State Communications 115 (2000) 523–526
525
percolation path [7,8] through the whole sample and thus
dominates the thermoelectric properties of the composite.
Therefore, although the electrical conductivity of B4C
increases with temperature [9], the electrical conductivity
of TiB2 decreases with temperature [13] and the electrical
conductivity for the 25.4 vol% TiB2/B4C sample decreases
with temperature.
The Seebeck coefficient measurement also shows such
changes with TiB2 content (Fig. 2). The Seebeck coefficient
changes from positive to negative value as the content of
TiB2 increases from 12.5 to 25.4 vol%. That is, when the
distribution of TiB2 particles in the composite changes from
a dispersed to a continuous state, the dominant conducting
mechanism in the composites changes from small polaron
hopping to electronic transport. There is also the intriguing
possibility that, with a p-type conductor as the second phase,
one could obtain p-type B4C-based composites. Also
because B4C (or BxC) is a well-known p-type thermoelectric
material, the search for n-type B4C (or BxC)-based materials
has attracted much attention [14,15], but until now it has
been unsuccessful. The present work indicates that n-type
B4C (or BxC)-based materials may be obtained by combin-
ing a metallic-conducting phase (electronic conductor) with
B4C (or BxC) in a composite.
The dimensionless ZTs calculated from the measure-
ments of the electrical conductivity, thermal conductivity
and Seebeck coefficient for all the samples are given as a
function of temperature in Fig. 4, which shows that the
values of ZT for the 12.5 vol% TiB2/B4C sample are
lower than that of the undoped B4C sample at all tempera-
tures, as predicted by EMT [6]. However, for the 25.4 vol%
TiB2/B4C sample, the values of ZT are higher than that of
the undoped sample below about 700 K, contradicting the
EMT prediction [6] and showing that this prediction does
not hold necessarily for the percolation system. Note that all
the samples have different porosities and that the electrical
conductivity, as well as the thermal conductivity of the
samples, decreases with porosity, while the Seebeck coeffi-
cient of the sample is not very sensitive to porosity [16].
Therefore, the porosity of the present samples probably has
no pronounced effect on their ZT values.
0.010
0.008
0.006
0.004
0.002
0.000
0%
25.4%
12.5%
900 1050
450
600
750
T(K)
Fig. 4. Dimensionless ZT of the samples as a function of
temperature.
observations, it can be seen that the average grain size of the
TiB2 and B4C is nearly the same (about 1 mm).
Fig. 1 shows that the electrical conductivity of the
samples increases rapidly with the content of TiB2. Note
that the electrical conductivity for the 12.5 vol% TiB2/B4C
sample increases with temperature as does the undoped
semiconducting B4C sample. However, the electrical
conductivity for the 25.4 vol% TiB2/B4C decreases with
temperature, indicating metallic behavior.
The 12.5 vol% TiB2/B4C sample is a p-type semiconduc-
tor like the undoped B4C, since both Seebeck coefficients
remain positive and increase with temperature as shown in
Fig. 2. The Seebeck coefficient of the 12.5 vol% TiB2/B4C is
always lower than that of the undoped sample at the same
temperature. However, the 25.4 vol% TiB2/B4C sample
behaves like an n-type material with a negative Seebeck
coefficient, whose absolute value increases slowly with
temperature.
The thermal conductivity of the samples as a function of
temperature is shown in Fig. 3. The thermal conductivity
increases monotonically with the amount of TiB2, because
of the high thermal conductivity of TiB2 compared with
B4C. The thermal conductivity for the undoped B4C and
12.5 vol% TiB2/B4C samples both decreases with tempera-
ture, whereas the thermal conductivity of the 25.4 vol%
TiB2/B4C sample increases with the temperature and is
nearly constant above about 900 K.
Figs. 1–3 clearly show that the thermoelectric properties
change rapidly as the amount of TiB2 goes from 12.5 to
25.4 vol%, which suggests that conducting phase (TiB2) in
the samples becomes continuous in this range and that fc for
TiB2 is in the range of 0.125–0.254. The transport properties
of the 12.5 vol% TiB2/B4C sample are similar to the B4C,
with the dominant charge carrier being a small polaron
(B11C)ϩ [9], because the TiB2 particles and clusters are
isolated in a B4C matrix. For the 25.4 vol% TiB2/B4C
sample, the conducting phase (TiB2) has a continuous or
4. Conclusion
The amount of the second conducting phase (TiB2) affects
the thermoelectric properties of TiB2/B4C composite cera-
mics. For the 12.5 vol% TiB2/B4C sample, the value of ZT is
always lower than that of the undoped B4C sample, while for
25.4 vol% TiB2/B4C sample, the value of ZT is higher than
that of the undoped B4C sample at temperatures below
700 K. This implies that it is possible to get an improvement
in ZT using a continuum percolating composite with an
optimized microstructure. Further experiments should be
made on composites with volume fractions of TiB2 closer
to the percolation threshold, as well as on other similar

526
K. Cai et al. / Solid State Communications 115 (2000) 523–526
Engineering, Interscience, New York, 1961, pp. 1–6, 458–
517.
[4] B.C. Sales, D. Mandrus, R.K. Williams, Science 272 (1996)
135.
systems such as W2B5/B4C, or even systems with other
semiconducting and conducting phases.
[5] L.D. Hicks, M.S. Dresselhaus, Phys. Rev. B 47 (1993) 12727.
[6] D.J. Bergaman, O. Levy, J. Appl. Phys. 70 (11) (1991) 6821.
[7] D.S. McLachlan, M. Blaszkiewics, R.E. Newnham, J. Am.
Ceram. Soc. 73 (1990) 2187.
[8] C.-W. Nan, Prog. Mater. Sci. 37 (1993) 1.
[9] C. Wood, D. Emin, Phys. Rev. B 29 (8) (1984) 4582.
[10] K.-f. Cai, C.-W. Nan, X.-M. Min, Mater. Sci. Engng B 67
(1999) 102.
Acknowledgements
This work was supported by the National Nature Science
Foundation of China under grant 59825102. Authors thank
Dr N. Siman and Mr V. Filippov for their help in the
measurements of electrical properties.
[11] G.H. Tang, Symposium Proceeding of 8th Conference of
Composite of China, Aviation, 1994, p. 363.
[12] ASTM Specification, C 373, ASTM Standards part 13, Amer-
ican Society for Testing and Materials, Philadelphia, 1969.
[13] C.C. Wang, S.A. Akbar, W. Chen, V.D. Patton, J. Mater. Sci.
30 (1995) 1627.
[14] T. Aselage, Mater. Res. Soc. Proc. 24 (1991) 145.
[15] C. Wood, Symposium Proceeding of APS, Vol. 140, AIP,
New York, 1986, p. 362.
References
[1] T.M. Tritt, M. Kanatzidis, G. Mahan, H.B. Lyon Jr. (Eds.),
Thermoelectric Materials: New Approaches and Advances,
Material Research Society Proceedings, Vol. 478, 1997.
[2] T.M. Tritt, G. Mahan (Eds.), Thermoelectric Materials 1998:
the Next Generation for Small-scale Refrigeration and Power
Generation, Material Research Society Proceedings, Vol. 545,
1999.
[16] K.-f. Cai, J.-P. Liu, C.-W. Nan, X.-M. Min, J. Mater. Sci. Lett.
16 (1997) 1876.
[3] R.R. Heikes, R.W. Ure Jr., Thermoelectricity Science and