Journal of Alloys and Compounds 487 (2009) 550–555
Contents lists available at ScienceDirect
Journal of Alloys and Compounds
journal homepage: www.elsevier.com/locate/jallcom
Thermoelectric properties of Bi, Nb co-substituted CaMnO3 at high temperature
J.W. Park a,∗ , D.H. Kwak b , S.H. Yoon b , S.C. Choi b
a
b
Department of Materials Science and Engineering, Ajou University, San 5, Woncheon, Youngtong, Suwon, Gyeonggi 443-749, Republic of Korea
Division of Energy Systems Research, Graduate School, Ajou University, Republic of Korea
a r t i c l e
i n f o
Article history:
Received 27 April 2009
Received in revised form 4 August 2009
Accepted 4 August 2009
Available online 8 August 2009
Keywords:
Thermoelectrics
Oxide thermoelectric material
N-type semiconductor
Thermoelectric properties
Thermal conductivity
a b s t r a c t
Ca1−x Bix Mn1−y Nby O3 (0 ≤ x = y ≤ 0.1) was prepared using a conventional solid-state reaction method. The
microstructures and homogeneity were confirmed by scanning electron microscopy (SEM) and X-ray
diffraction (XRD), and its high temperature thermoelectric properties were also characterized. The lattice parameter of orthorhombic perovskite increased with additional amounts of Bi and Nb. The electrical
conductivity () increased, and Seebeck coefficient (˛) decreased, as the amount of Bi and Nb increased
up to x = y = 0.08. The electrical conductivity decreased as the temperature increased, indicating metallic
behavior, and the Seebeck coefficient increased with increasing temperature. As a result, thermoelectric properties were improved by substituting Bi and Nb into CaMnO3 . It showed maximum values at
x = y = 0.04, in which the power factor (PF) was 200 ␮W/m K2 and dimensionless figure of merit (ZT) was
approximately 0.1 at 873 K. The thermoelectric properties of Ca0.96 Bi0.04 Mn0.96 Nb0.04 O3 were improved
nearly twofold over un-doped CaMnO3 .
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
Thermoelectric power generation involving the direct conversion of waste heat and thermal energy into electrical energy has
become a recent topic of intense interest. The main issue in thermoelectric generation is to develop materials whose properties
remain highly stable at high temperatures. Due to the oxidation
and decomposition of metallic semiconductors such as SiGe, PbTe,
Bi2 Te3 at high temperatures, oxide materials have been recognized
as good candidates for application on thermoelectric generation at
high temperatures.
Thermoelectric materials are evaluated by the figure of merit,
Z, defined as ˛2 /k, or by ZT, where ˛, , k and T are the Seebeck
coefficient, electrical resistivity, thermal conductivity and temperature, respectively. Conventionally, oxide materials are not suitable
for use as thermoelectric materials due to their low carrier mobility. However, after Nax CoO2 was found to have low resistivity [1],
a significant amount of research has been directed toward finding
materials with improved thermoelectric performance and stability
at high temperatures. For example, several systems for oxide thermoelectric materials have been investigated, including Nax CoO2 ,
Ca3 Co4 O9 , NiO, (Zn1−x Alx )O, and (In2 O3 )m (ZnO)n [1–6]. Especially,
it was reported that the thermoelectric properties of p-type thermoelectric materials with Nax CoO2 [1,2] are as large as those of
conventional materials with SiGe. Recently, several efforts have
∗ Corresponding author. Tel.: +82 31 219 2471; fax: +82 31 219 2471.
E-mail address: won8864@naver.com (J.W. Park).
0925-8388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2009.08.012
been reported to produce textured thermoelectric oxide materials
by hot pressing and spark plasma sintering for elevating thermoelectric properties of Ca3 Co4 O9 [7,8]. But there are relatively a
few n-type oxide candidates, and especially no report about high
performance n-type oxide thermoelectric materials comparable to
conventional thermoelectric materials. However, some researchers
have reported new candidates for n-type oxide thermoelectric
materials, such as Sn1−x−y Tiy Sbx O2 [9] ceramics, SrTiO3 processed
by hot pressing [10], and CaMnO3 substituted with various elements [11–18].
CaMnO3 has attracted attention as a possible n-type oxide
thermoelectric material for use in generators. Ohtaki et al. and
Funahashi et al. reported the thermoelectric properties of CaMnO3
substituting Bi, Y, La, Ce and RE (Nd, Tb, Ho, Yb, Lu) at the Ca site
[11,12]. They found that when Bi, Y, La, Ce and RE (Nd, Tb, Ho,
Yb, Lu) were substituted in CaMnO3 at the Ca site, the electrical
conductivity increased considerably, while the Seebeck coefficient
decreased slightly. It was also found that when substituting Pr, Sr,
Mo and Bi at the Ca site of CaMnO3 , the thermoelectric properties improved [13–15]. The substitution effects on thermoelectric
properties of substituting Nb, Ta, Ru and V at the Mn site were also
investigated [16–18]. It was indicated that thermoelectric properties were improved owing to the increased electrical conductivity
that resulted from the substitutions at the Ca site and Mn site.
In the present study, Bi and Nb were simultaneously substituted
at the Ca site and Mn site, respectively, in an effort to improve
the thermoelectric properties of CaMnO3 . The effect of the substitution was investigated by evaluating the compound’s electrical
conductivity, Seebeck coefficient and thermal conductivity.
J.W. Park et al. / Journal of Alloys and Compounds 487 (2009) 550–555
2. Experimental procedures
2.1. Sample preparation
The Ca1−x Bix Mn1−y Nby O3 (0 ≤ x = y ≤ 0.1, by step 0.02) was prepared by a conventional solid-state reaction method. First, CaMnO3−ı powder was pre-synthesized
551
with CaCO3 and MnO2 powders at 1373 K, and then the calcined powder was pulverized into fine powder with particle size of approximately 40 ␮m. For the second
time, the prepared CaMnO3−ı powder, Nb2 O5 and Bi2 O3 were mixed in the stoichiometric chemical ratio. The mixed powder samples were calcined at 1373 K, ground
and pressed into pellets using a CIP (Cold Isostatic Press) at 250 MPa. The pressed
pellets were sintered at 1573 K for 12 h in air.
Fig. 1. SEM image of Ca1−x Bix Mn1−y Nby O3 . (a) CaMnO3 , (b) x = y = 0.02, (c) x = y = 0.03, (d) x = y = 0.04, (e) x = y = 0.05, (f) x = y = 0.06, (g) x = y = 0.08, and (h) x = y = 0.1.
552
J.W. Park et al. / Journal of Alloys and Compounds 487 (2009) 550–555
2.2. Characterization
The microstructures of the sintered samples were confirmed by scanning electron microscopy (SEM; JEOL JSM-6380). The homogeneity of the samples was
estimated by X-ray diffraction (XRD; Rigaku D/max-2500V/PC) using Cu K␣ radiation. Their electrical conductivity was measured by the DC four-probe method
at 300–1000 K in air. The Seebeck coefficient, representing the thermoelectric
property, was measured by the least-squares method of the plot of V (thermoelectromotive force) versus T (temperature difference), and the contribution of
the lead wires (Pt) was subtracted. The thermal conductivity was calculated from
the equation k = Cp d, where Cp , and d are the specific heat measured by means
of differential scanning calorimeter (DSC; Netzsch 20-F3) at 300–1000 K, thermal
diffusivity by a laser flash technique using a Netzsch LFA 457 at 300–1000 K and
density by the Archimedes method.
3. Results and discussion
3.1. The microstructure of Ca1−x Bix Mn1−y Nby O3
The
microstructure
of
sintered
Ca1−x Bix Mn1−y Nby O3
(0.02 ≤ x = y ≤ 0.1) is shown in Fig. 1. It shows coarse grain
appearing as the amount of Bi and Nb increased. It also shows that
the pore density of the prepared sample gets decreased gradually.
In each case, the relative density obtained by measuring the
density with the Archimedes method was 90.9–94.2%, showing
the increase due to the additional amount of Bi and Nb. The grain
growth was likely promoted owing to the addition of Bi and Nb,
and the density was increased owing to the liquid-phase sintering
by Bi2 O3 . The result of XRD analysis of Ca1−x Bix Mn1−y Nby O3 is
shown in Fig. 2. Many researchers have reported the XRD data of
CaMnO3 , including the crystal lattice, atomic position, and other
parameters [19–22]. From the reported data, the lattice parameter
was calculated using least-squares refinement. According to the
XRD peak in Fig. 2, the diffraction peak matches the previously
reported data in the case of CaMnO3 , and there is no secondary
phase caused by the additional amount of Bi and Nb. Thus, it
has the orthorhombic perovskite structure of CaMnO3 , indicating
that the Ca2+ site and Mn4+ site of CaMnO3 were substituted
with Bi3+ and Nb5+ respectively. In addition, the lattice parameter
depending on the additional amount of Bi, Nb and Mn valence in
Ca1−x Bix Mn1−y Nby O3 (0.0 ≤ x = y ≤ 0.1) are shown in Fig. 3. The
relationships between the Mn valence, lattice parameter and
lattice volume is an important factor that determines the thermoelectric properties of the material. Xu et al. reported that Nb, with
electron configurations of [Kr]4d4 5s1 , has a greater possibility of
replacing the Mn site through the d–p hybridization of Nb and
Fig. 2. XRD patterns of Ca1−x Bix Mn1−y Nby O3 . (a) CaMnO3 , (b) x = y = 0.02, (c)
x = y = 0.03, (d) x = y = 0.04, (e) x = y = 0.05, (f) x = y = 0.06, (g) x = y = 0.08, and (h)
x = y = 0.1.
O [16]. This is similar to the d–p hybridization of Ru–O in the
superconductive material Sr2 RuO4 . Moreover, Pi et al. reported
that Mn site was substituted with Ru5+ in CaMn1−x Rux O3 [23].
From these reported results, the Mn valence was calculated from
the charge neutrality in the formula by assuming an Nb valence
of Nb5+ . Fig. 3 clearly shows that the lattice parameter increased
with additional amounts of Bi and Nb. This occurred due to the fact
that the ionic radius of Bi3+ (1.17 Å) and that of Nb5+ (0.64 Å) are
greater than that of Ca2+ (1.12 Å), Mn3+ (0.58 Å) or Mn4+ (0.54 Å)
ions.
3.2. The thermoelectric properties of Ca1−x Bix Mn1−y Nby O3
The electrical conductivity () of Ca1−x Bix Mn1−y Nby O3
(0 ≤ x = y ≤ 0.1) is shown in Fig. 4. Un-doped CaMnO3 shows
the typical behavior of a semiconductor, whereas when Bi3+ and
Nb5+ are simultaneously substituted, it shows typical metallic
conductivity, in which the electrical conductivity decreases with
increasing temperature. It also increases continuously with the
amount of substitution up to x = y = 0.08 at Ca1−x Bix Mn1−y Nby O3 .
Fig. 3. Lattice parameter and volume of Ca1−x Bix Mn1−y Nby O3 as a function of the Mn valence and x = y.
J.W. Park et al. / Journal of Alloys and Compounds 487 (2009) 550–555
553
Fig. 4. Electrical conductivities of Ca1−x Bix Mn1−y Nby O3 as a function of temperature.
Fig. 6. Activation energy of Ca1−x Bix Mn1−y Nby O3 as a function of doping level (x = y).
Generally, electrical conductivity by hopping conduction can be
expressed as presented in Eq. (1) [24]
= nea2
A
T
E a
exp −
kT
(1)
where n, e, a, A, T, Ea and k are the carrier concentration, the
electrical charge of carrier, the intersite distance of hopping, a
pre-exponential term related to the carrier scattering mechanism,
the absolute temperature, the activation energy of hopping and
the Boltzmann constant, respectively. As described in Eq. (1), the
increase in electrical conductivity is most likely caused by the
increase of carrier concentration when Bi3+ and Nb5+ are substituted, and by the increase of carrier mobility according to the
increase of the lattice parameter. However, Ca1−x Bix Mn1−y Nby O3
in a range of 0.08 < x = y shows a decrease. According to the study
of Huang et al., lattice distortion seems to become more serious
as the amount of Nb5+ into MnO6 octahedra increases, thus leading to a reduction in carrier mobility, and therefore decreasing the
electrical conductivity [25]. The activation energy Ea could be estimated from the Arrhenius plot of log T versus 1/T as shown in
Fig. 5. As shown in Fig. 5, the relationship between log ıT and 1/T
does not have complete linearity within the range of the measured
temperature, but it shows a series of linearity at temperatures
that exceed 500 K. The result of calculating the activation energy
Fig. 5. Arrhenius plots of the electrical conductivities of Ca1−x Bix Mn1−y Nby O3 .
(Ea ) of Ca1−x Bix Mn1−y Nby O3 from the relation between log ıT and
1/T, indicates 0.17 eV when x = y = 0, 0.038 eV when x = y = 0.02,
0.022–0.024 eV when x = y = 0.04–0.08 and 0.041 eV when x = y = 0.1
(Fig. 6). According to Eq. (1), a considerable decrease of activation
energy due to the addition of Bi and Nb induced an increase in
electrical conductivity over that of CaMnO3 .
Fig. 7 shows the Seebeck coefficient (˛) of Ca1−x Bix Mn1−y Nby O3
(0 ≤ x = y ≤ 0.1). As the additional amount of Bi and Nb increase, the
Seebeck coefficient decreases. In particular, it shows a tendency to
increase continuously as the temperature increases. Considering
the carrier concentration, the Seebeck coefficient can be expressed
as shown below [23].
˛=−
2 k2 T
3e
N
n+c
(2)
where N and c are the density of state, and a constant related
to the materials. According to this equation, the increase of carrier concentration by the addition of Bi and Nb results in a
reduction of the thermo-electromotive force. That is to say, if
the carrier concentration of materials is increased by the substitution of elements, electrical conductivity increases but the
Seebeck coefficient decreases. This result is similar to other studies
in terms of the thermoelectric properties obtained upon substitution of CaMnO3 with other transition metal ions [11]. Fig. 8
Fig. 7. Seebeck coefficients of Ca1−x Bix Mn1−y Nby O3 as a function of temperature.
554
J.W. Park et al. / Journal of Alloys and Compounds 487 (2009) 550–555
Fig. 8. Power factors of Ca1−x Bix Mn1−y Nby O3 as a function of temperature.
shows the power factor calculated from the measured electrical
conductivity and the Seebeck coefficient of Ca1−x Bix Mn1−y Nby O3
(0 ≤ x = y ≤ 0.1). The electrical conductivity increases considerably
as the additional amount of Bi and Nb increase, and the Seebeck coefficient shows a slight decrease. The most appropriate
additional amount can be obtained by calculating the power factor. As shown in Fig. 8, the power factor reaches its maximum
value when x = y = 0.04 in Ca1−x Bix Mn1−y Nby O3 . The value at this
point is approximately 200 ␮W/m K2 at 873 K, and it generally
increases as the temperature increases. Fig. 9 shows the measured values of thermal conductivity when x = y = 0, 0.02, 0.04, and
0.06 in Ca1−x Bix Mn1−y Nby O3 . The thermal conductivity, k, can be
expressed by the sum total of the lattice vibration component (k␫ )
and electronic component (ke ). ke has a much lower value overall,
and increases as the electrical conductivity increases. However, ke is
very low in general, and its impact is insignificant. Accordingly, the
thermal conductivity depends on the lattice vibration component
(k␫ ). The thermal conductivity decreases slightly as the additional
amount of Bi and Nb increases, and it was considered that this was
caused by the increase of lattice parameters after the addition of
Bi and Nb, by irregularities due to the lattice distortion, and by
the change of phonon scattering owing to the added chemical element. Specifically, the increase in the added Bi and Nb promoted
the distortion of the MnO6 octahedron, which then resulted in the
Fig. 10. Dimensionless figure of merit (ZT) of Ca1−x Bix Mn1−y Nby O3 (x = y = 0, 0.02,
0.04, 0.06).
irregularity of the crystal lattice. This irregularity caused a reduction in the thermal conductivity. The dimensionless figure of merit
(ZT) when x = y = 0, 0.02, 0.04, and 0.06 in Ca1−x Bix Mn1−y Nby O3
is shown in Fig. 10. These results indicate that the value of ZT
is approximately 0.1 when x = y = 0.04 in Ca1−x Bix Mn1−y Nby O3 at
873 K, and it was improved nearly twofold over un-doped CaMnO3 .
4. Conclusions
The improvements in thermoelectric properties were evaluated by simultaneously substituting Bi and Nb into the Ca site and
Mn site, respectively, in CaMnO3 . A sample was synthesized in a
solid-state reaction and manufactured using uniaxial press and CIP.
Electrical conductivity was increased owing to the increase of the
lattice parameter and the carrier concentration as the additional
amount of Bi and Nb up to 8 mol% increased. However, the Seebeck
coefficient tended to decrease. As a function of increasing temperature, the electrical conductivity tended to decrease and the Seebeck
coefficient tended to increase. The results show that the thermoelectric properties improved due to the substitution of Bi and Nb
into the CaMnO3 . This showed a maximum value at x = y = 0.04,
where the PF was 200 ␮W/m K2 , and the dimensionless figure of
merit was approximately 0.1 at 873 K. The thermoelectric properties of Ca0.96 Bi0.04 Mn0.96 Nb0.04 O3 were improved nearly twofold
over un-doped CaMnO3 .
References
Fig. 9. Thermal conductivities of Ca1−x Bix Mn1−y Nby O3 (x = y = 0, 0.02, 0.04, 0.06).
[1] I. Terasaki, Y. Sasago, K. Uchinokura, Phys. Rev. B 56 (1997) R12685.
[2] S. Tajima, T. Tani, S. Isobe, K. Koumoto, Mater. Sci. Eng. B 86 (2001) 20.
[3] T. Takeuchi, T. Kondo, K. Soda, U. Mizutani, R. Funahashi, M. Shikano, S. Tsuda,
T. Yokoya, S. Shin, T. Muro, J. Electron Spectrosc. Relat. Phenom. 137 (2004) 595.
[4] W. Shin, N. Murayama, Mater. Lett. 45 (2000) 302.
[5] T. Tsubota, M. Ohtaki, K. Eguchi, Mater. Chem. 7 (10) (1997) 85.
[6] M. Kazeoka, H. Hiramatsu, W.S. Seo, K. Koumoto, J. Mater. Res. 13 (3) (1998)
523.
[7] M. Prevel, O. Perez, J.G. Noudem, Solid State Sci. 9 (2007) 231.
[8] J.G. Noudem, J. Eur. Ceram. Soc. 29 (2009) 2659.
[9] T. Tsubota, T. Ohno, N. Shiraishi, Y. Miyazaki, J. Alloys Compd. 463 (2008) 288.
[10] M. Ito, T. Matsuda, J. Alloys Compd. 477 (2009) 473.
[11] M. Ohtaki, H. Koga, T. Tokunaga, K. Eguchi, H. Arai, J. Solid State Chem. 120
(1995) 105.
[12] R. Funahashi, A. Kosuga, N. Miyasou, S. Urata, K. Lee, H. Ohta, K. Koumoto, Proc.
26th Int. Conf. Thermoelectrics, 2007, p. 124.
[13] B.T. Cong, T. Tsuji, P.X. Thao, P.Q. Thanh, Y. Yamamura, Physica B 352 (2004) 18.
[14] G. Xu, R. Funahashi, I. Matsubara, M. Shiano, Y. Zhou, J. Mater. Res. 17 (2002)
1092.
[15] M. Miclau, S. Herbert, R. Retoux, C. Martin, J. Solid State Chem. 178 (2005) 1104.
[16] G. Xu, R. Funahashi, Q. Pu, B. Liu, R. Tao, G. Wang, Z. Ding, Solid State Ionics 171
(2004) 147.
J.W. Park et al. / Journal of Alloys and Compounds 487 (2009) 550–555
[17] Y. Zhou, I. Matsubara, R. Funahashi, G. Xu, M. Shikano, Mater. Res. Bull. 38 (2003)
341.
[18] R. Ang, Y.P. Sun, Y.Q. Ma, B.C. Chao, X.B. Zhu, W.H. Song, J. Appl. Phys. 100 (2006)
063902.
[19] M.E.J. Melo, A. Correia dos Santos, M.R. Nunes, Int. J. Inorg. Mater. 3 (2001) 915.
[20] K.R. Poeppelmeier, M.E. Leonowicz, J.C. Scanlon, J.M. Longo, J. Solid State Chem.
45 (1982) 71.
555
[21] H. Taguchi, Phys. Status Solidi (a) 88 (1985) K79.
[22] J.B. MacChesney, H.J. Williams, J.F. Potter, R.C. Sherwood, Phys. Rev. 164 (1967)
779.
[23] L. Pi, S. Herbert, C. Martin, A. Maignan, B. Raveau, Phys. Rev. B 67 (2003) 024430.
[24] H.L. Tuller, A.S. Nowick, J. Phys. Chem. Solids 38 (1977) 859.
[25] X.Y. Huang, Y. Miyazaki, T. Kajitani, Solid State Commun. 145 (2008)
132.