Influence of SnO2 addition on the thermoelectric properties of Zn1 - xSnxO (0.01 ≤ x ≤ 0.05)

Article abstract of DOI:10.1016/j.materresbull.2007.02.018

The grain size and the density of the Zn_{1 - x}Sn_xO (0 ≤ x ≤ 0.05) samples decreased with increasing SnO₂ content. The addition of a small amount of SnO₂ (x ≤ 0.01) to ZnO led to an increase in both the electrical conductivity and the absolute value of the Seebeck coefficient, resulting in a significant increase in the power factor. The thermoelectric power factor was maximized to a value of 1.25 × 10^-3 Wm^-1 K^-2 at 1073 K for the Zn_0.99Sn_0.01O sample.

Full text of DOI:10.1016/j.materresbull.2007.02.018

Materials Research Bulletin 43 (2008) 54–61
www.elsevier.com/locate/matresbu
Influence of SnO2 addition on the thermoelectric properties
of Zn1 ꢀ xSnxO (0.01 ꢁ x ꢁ 0.05)
a
a
b
c
K. Park a, , J.K. Seong , Y. Kwon , S. Nahm , W.-S. Cho
*
a Department of Advanced Materials Engineering, Sejong University, Seoul 143-747, Republic of Korea
b Department of Materials Science and Engineering, Korea University, Seoul 136-701, Republic of Korea
c School of Materials Science and Engineering, Inha University, Inchon 402-751, Republic of Korea
Received 17 November 2006; received in revised form 24 January 2007; accepted 8 February 2007
Available online 13 February 2007
Abstract
The grain size and the density of the Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples decreased with increasing SnO2 content. The addition
of a small amount of SnO2 (x ꢁ 0.01) to ZnO led to an increase in both the electrical conductivity and the absolute value of the
Seebeck coefficient, resulting in a significant increase in the power factor. The thermoelectric power factor was maximized to a
value of 1.25 ꢂ 10ꢀ3 Wmꢀ1 Kꢀ2 at 1073 K for the Zn0.99Sn0.01O sample.
# 2007 Elsevier Ltd. All rights reserved.
Keywords: A. Ceramics; C. X-ray diffraction; C. Electron microscopy; D. Electrical properties; D. Microstructure
1. Introduction
Thermoelectric materials with high-energy conversion efficiency are strongly required for both the generation of
electric power in terms of waste heat recovery as well as the refrigeration of electronic devices. The evaluation of the
properties of the thermoelectric materials can be expressed by the figure-of-merit (Z) as follows [1]: Z = sa2/k, where
s, a, and k are the electrical conductivity, Seebeck coefficient, and thermal conductivity, respectively. The electrical
properties are determined by the power factor (P) defined as [1]: P = sa2. To be a good thermoelectric material, it is
essential to have a large power factor and a low thermal conductivity.
In recent years, there has been great interest in developing novel oxide thermoelectric materials for energy
conversion. This is because, unlike conventional materials such as FeSi2 [2–4], PbTe [5], and Si–Ge alloys [6,7], they
are chemically and thermally stable and thus can be used at high temperatures without deterioration of their
performance due to oxidation. Moreover, their production costs are comparatively low. It has been reported that for the
n-type oxides, In2O3-, SrPbO3-, and SrLa0.1TiO3-based oxides show promising thermoelectric properties with good
durability at high temperatures [8–10]. However, the thermoelectric properties alone are not enough to be applicable to
the power generators. The development of new oxide materials is still the most important issue for practical
applications of thermoelectric power generation. In the present work, we have selected zinc oxide (ZnO) as a candidate
host material.
* Corresponding author. Tel.: +82 2 3408 3777; fax: +82 2 3408 3664.
E-mail address: kspark@sejong.ac.kr (K. Park).
0025-5408/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2007.02.018

K. Park et al. / Materials Research Bulletin 43 (2008) 54–61
55
˚
ZnO forms in the wurtzite structure (a = 3.249 and c = 5.206 A) and is a direct wide band gap (Eg = 3.3eV) n-type
semiconductor. Its semiconducting behavior is caused by the deviation from the stoichiometry due to the presence of
intrinsic defects such as oxygen vacancies and zinc interstitial [11]. Doped ZnO has recently gained much attention
due to many interesting potential applications such as a transparent electrode [12], piezoelectric device [13], gas
sensor [14], and window layer in solar cells [15]. Current studies are devoted to investigate the thermoelectric
properties of the Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05), especially with regard to adding SnO2, and we have demonstrated that
the Zn1 ꢀ xSnxO is strongly desirable as a novel high-temperature thermoelectric material. The added Sn acts as a
doubly ionized donor impurity, providing electrons, which lead to a good quality ZnO-based semiconductor [11].
2. Experimental
High-purity ZnO and SnO2 powders (Kojundo Chemical Lab Co., Japan) were used in this study. Zn1 ꢀ xSnxO
(x = 0, 0.005, 0.01, 0.02, 0.03, 0.04, and 0.05) samples were prepared by solid-state reaction. ZnO and SnO2 powders
were weighed in appropriate proportions. A mixture of the ZnO and SnO2 powders and ethyl alcohol was milled for 6 h
using a planetary mill (Fritsch pulverisette 6) and ZrO2 ball as a grinding media. The mixed powders were calcined in a
mullite crucible at 1273 K for 5 h. The dried powders were pressed at a pressure of 98 MPa to prepare pellets of 5 mm
thick and 20 mm diameter. The green compacts were heated at 1673 K for 20 h in air and then furnace-cooled to room
temperature.
The crystalline structure and the surface microstructure of the as-sintered Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples at
room temperature were investigated with an X-ray diffractometer (XRD; Rigaku RINT2000) and a scanning electron
microscope (SEM; Hitachi S4700), respectively. For microstructural observations, the polished samples were thermal-
etched at 1523 K for 30 min. The porosity of the as-sintered samples was measured by the Archimedes principle. The
average grain sizes of the etched samples were estimated by the line-intersecting method. In order to identify the
distribution of constituent elements in the as-sintered ceramics, a secondary electron image (SEI) and elemental maps
were investigated using electron probe microanalysis (EPMA).
The samples for the measurements of thermoelectric properties were cut out of the sintered bodies in the form of
rectangular bars of 2 mm ꢂ 2 mm ꢂ 15 mm. Electrical conductivity (s) was measured by the direct current (dc) four-
probe method. For thermopower measurements, a temperature difference in the sample was generated by passing cool
Ar gas over one end of the sample placed inside a quartz protection tube. These measurements were carried out at
temperatures from 723 to 1073 K. Further details on thermoelectric property measurements may be found elsewhere
[16].
3. Results and discussion
The SEM images from the surface of the as-sintered Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples are shown in Fig. 1. Fine
Zn2SnO4 particles of 1–3 mm in size for the Sn-added Zn1 ꢀ xSnxO samples are frequently dispersed on the grain
boundaries. The result, shown in Fig. 1, strongly suggests that the added SnO2 did not fully dissolve in the ZnO crystal
lattice owing to its rather low solubility (x ꢃ 0.001) [11]. The formation of Zn2SnO4 is expressed as follows [17]:
2ZnO + SnO2 ! Zn2SnO4. The solid reaction occurs at 973 K and continues at 1173 K [17]. A similar behavior was
found in TiO2-added ZnO thick films [18]. Zhu et al. [18] detected Zn2TiO4 phase, which was formed by the following
reaction: 2ZnO + TiO2 ! Zn2TiO4, in the TiO2-added ZnO films.
We found that the average grain size of the Zn1 ꢀ xSnxO samples became smaller with higher SnO2 content, ranging
from 13.9 to 3.5 mm. This is indeed attributed to the pinning effect caused by the Zn2SnO4 particles on grain
boundaries as well as to the dragging effect between the added SnO2 and grain boundaries, resulting in a reduction in
the mobility of the grain boundaries [19,20]. In addition, it can be seen that increasing SnO2 content results in an
increase of the porosity. This is attributed to the fact that the added SnO2 lowers the grain-boundary mobility during
sintering and enables the pores to stay attached to the moving grain boundaries. The porosity ranged from 0.8 to 4.8%.
The average grain size and the porosity of all the prepared Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples are given in Table 1.
The XRD patterns of the as-sintered ZnO, Zn0.99Sn0.01O, Zn0.97Sn0.03O, and Zn0.95Sn0.05O at room temperature are
shown in Fig. 2(a–d), respectively. The as-sintered Sn-added Zn1 ꢀ xSnxO (0.01 ꢁ x ꢁ 0.05) samples consist of the
majority Zn1 ꢀ xSnxO with a wurtzite structure and a secondary phase Zn2SnO4. The Zn2SnO4 belongs to the space
˚
group Fd3m (2 2 7) and has the cubic spinel structure (a = 8.657 A) [21]. The basic spinel structure of Zn2SnO4 is

56
K. Park et al. / Materials Research Bulletin 43 (2008) 54–61
Fig. 1. SEM images from the surface of the thermal-etched (a) ZnO, (b) Zn0.99Sn0.01O, (c) Zn0.98Sn0.02O, (d) Zn0.97Sn0.03O, (e) Zn0.96Sn0.04O, and
(f) Zn0.95Sn0.05O samples.
Table 1
Grain size and porosity of the prepared Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples
Grain size (mm)
Porosity (%)
ZnO
13.9
11.3
10.3
7.0
0.8
1.5
2.0
3.4
4.3
4.6
4.8
Zn0.995Sn0.005
O
Zn0.99Sn0.01
Zn0.98Sn0.02
Zn0.97Sn0.03
Zn0.96Sn0.04
Zn0.95Sn0.05
O
O
O
O
O
4.1
3.9
3.5

K. Park et al. / Materials Research Bulletin 43 (2008) 54–61
57
Fig. 2. XRD patterns of the as-sintered (a) ZnO, (b) Zn0.99Sn0.01O, (c) Zn0.97Sn0.03O, and (d) Zn0.95Sn0.05O samples.
considered to be a cubic close-packed array of anions with cations occupying 1/8 of the tetrahedral and 1/2 of the
octahedral interstitial sites [22,23]. The spinel unit cell is face-centered cubic with 32 anions and 24 cations [23]. No
orthorhombic ZnSnO3 phase was ever detected. The ZnSnO3 exhibits a limited stability to decompose into ZnSnO4
and SnO2 at 873 K [24]. The amount of the Zn2SnO4 phase became greater with higher SnO2 content. The presence of
the Zn2SnO4 phase was confirmed by elemental maps. For example, the SEI and elemental mapping of the as-sintered
Zn0.96Sn0.04O are shown in Fig. 3. We found that the Zn distribution was relatively homogeneous, while the O and Sn
distributions were heterogeneous. Both ZnO and SnO2 do not form extensive solid solutions, but rather the Zn2SnO4
phase at the grain boundaries.
Fig. 3. SEI and elemental maps of as-sintered Zn0.96Sn0.04O sample.

58
K. Park et al. / Materials Research Bulletin 43 (2008) 54–61
Fig. 4. Electrical conductivity as a function of temperature for the Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples.
The electrical conductivity (s) of the Sn-added Zn1 ꢀ xSnxO (0.005 ꢁ x ꢁ 0.05) samples as a function of
temperature is shown in Fig. 4, together with that of the Sn-free ZnO sample. The electrical conductivity of the Sn-
free ZnO slightly increased with increasing temperature, indicating a semiconducting behavior. On the other hand,
the electrical conductivity of the Sn-added Zn1 ꢀ xSnxO (0.005 ꢁ x ꢁ 0.05) slightly decreased with increasing
temperature, indicating a metallic behavior. It is evident that the addition of SnO2 affects significantly the electrical
conductivity. The incorporation of a small amount of SnO2 (x ꢁ 0.01) to ZnO leads to a marked increase in the
electrical conductivity, whereas the conductivity for higher SnO2 content (0.02 ꢁ x ꢁ 0.05) is gradually reduced by
further additions of SnO2 content. This can be explained by considering the following competing factors
influencing the electrical conductivity as follows. (1) The addition of a small amount of SnO2 leads to a strong
donor effect [11]. The substitution of Sn4+ for Zn2+ may increase the electron concentration of the system to
compensate for the electric charge balance, originating from the extra electrons generated. The electron serves as a
donor, leading to an increase in the electrical conductivity. (2) The higher the SnO2 content, the higher the second
phase Zn2SnO4 having a low electrical conductivity in the sintered bodies, thereby decreasing the electrical
conductivity. The electrical conductivity of Zn2SnO4 as a function of temperature is shown in Fig. 5. The electrical
Fig. 5. Electrical conductivity as a function of temperature for Zn2SnO4.

K. Park et al. / Materials Research Bulletin 43 (2008) 54–61
59
conductivity decreased with increasing temperature. It has been reported that the Zn2SnO4 possesses a wide
band gap (Eg = 3.35 eV) and a very low electrical conductivity, ranging from 3.9 to 101.7 Vꢀ1 cmꢀ1 at room
temperature [23,25]. (3) The addition of SnO2 leads to a significant decrease in the density and grain size, and to
an increase in Zn2SnO4. This is responsible for the decrease in the time between scattering events of charge
carriers, which thereby decreases the electrical conductivity. From the above competing factors, it is considered
that for low SnO2 content (x ꢁ 0.01), factor (1) dominantly affects the electrical conductivity, compared to the
factors (2) and (3). In contrast, for high SnO2 content (0.02 ꢁ x ꢁ 0.05), factors (2) and (3) dominantly affect the
electrical conductivity. A similar result was reported for the Sn-doped Zn1ꢀxSnxO (0.02 ꢁ x ꢁ 0.06) thin
films prepared by spray pyrolysis [26]. Bougrine et al. [26] found that the electrical conductivity of the films
increased sharply with the addition of SnO2, reached a maximum value (20 Vꢀ1 cmꢀ1) at 2 at.% SnO2, and then
decreased with further addition of SnO2. The electrical conductivity of the Zn1 ꢀ xSnxO samples studied in this
work was much higher than that of the films mainly owing to their low scattering events, originating from high
density.
Sn-added Zn1 ꢀ xSnxO shows a non-linear current (I)–voltage (V) behavior [11], which is similar to that of other
ZnO-based oxides. The non-linear behavior in ZnO-based materials results from the development of potential barriers
to electrical conduction at the ZnO grain boundaries [11]. Potential barriers strongly affect the carrier mobility. For a
degenerate semiconductor, according to the scattering model of grain boundary, mobility m varies with the potential
barrier of grain boundary DE as follows [27]: m = m0Tꢀ1/2 exp(ꢀDE/kT), where k is the Boltzmann constant and T is
the absolute temperature. Non-linear I–V characteristics in ZnO-based oxides depend on their microstructure,
especially grain size. Peiteado et al. [28] reported that the nominal breakdown voltage of ZnO-based oxides decreased
with an increase in grain size. As a result, because the grain size of the Sn-added Zn1 ꢀ xSnxO samples decreases with
an increase in SnO2 content, their nominal breakdown voltage increases with SnO2 content. In addition, the
incorporation of SnO2 does not affect the conduction mechanism in ZnO because the non-linearity of the Sn-added
Zn1 ꢀ xSnxO is quite similar to that of the Sn-free ZnO [11].
Fig. 6 shows the temperature dependence of the Seebeck coefficient (a) for the Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05)
samples. The sign of the Seebeck coefficient is negative over the whole temperature range for all the samples,
indicating that the major conductivity carriers are electrons. It is important to note that the absolute value of the
Seebeck coefficient for the Sn-added Zn1 ꢀ xSnxO (0.005 ꢁ x ꢁ 0.05) at 1073 K is higher than that of the Sn-free ZnO.
The observed temperature dependence of both the electrical conductivity and the Seebeck coefficient is unusual. In
general, the value of the Seebeck coefficient in common semiconductors decreases with an increase in carrier density
[1]. These results indicate that the conduction mechanism in the Zn1 ꢀ xSbxO samples is not explicable by a
conventional model based on the band theory, but rather by a strong electron-electron correlation effect [29–31]. We
are unsure how to explain this behavior at present. Furthermore, for high Sn-added samples (0.02 ꢁ x ꢁ 0.05), the
Fig. 6. Temperature dependence of the Seebeck coefficient for the Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples.

60
K. Park et al. / Materials Research Bulletin 43 (2008) 54–61
Fig. 7. Temperature dependence of the Seebeck coefficient for Zn2SnO4.
Fig. 8. Temperature dependence of the power factor for the Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05) samples.
absolute value of the Seebeck coefficient became smaller with higher SnO2 content mainly due to an increase in the
Zn2SnO, which has a low absolute value of the Seebeck coefficient. The Seebeck coefficient of Zn2SnO4 as a function
of temperature is shown in Fig. 7. The absolute value of the Seebeck coefficient for Zn2SnO4 increased gradually with
increasing temperature.
The temperature dependence of the power factor (sa2) experimentally obtained for the Zn1 ꢀ xSnxO (0 ꢁ x ꢁ 0.05)
samples is shown in Fig. 8. The power factor of the Zn1 ꢀ xSnxO samples increases up to 1073 K. The power factors of
the Sn-added Zn1 ꢀ xSnxO samples are much higher than those of the Sn-free ZnO sample. The thermoelectric power
factor is maximized to a value of 1.25 ꢂ 10ꢀ3 Wmꢀ1 Kꢀ2 at 1073 K for the Zn0.99Sn0.01O sample. The power factor is
several times superior to that of In2O3ꢀ, SrPbO3ꢀ, SrLa0.1TiO3ꢀ, NiOꢀ, and HoCoO3-based oxides [8–10,32,33], and
is comparable to the values reported for the Zn1 ꢀ xAlxO [34] and for the conventional thermoelectric materials in
practical use such as FeSi2, PbTe, and Si–Ge alloys [2–7]. Moreover, Zn1 ꢀ xSnxO ceramics are chemically and
thermally stable in air, at least up to 1673 K without any surface protection.

K. Park et al. / Materials Research Bulletin 43 (2008) 54–61
61
4. Conclusions
The addition of SnO2 to ZnO, with the development of the cubic spinel Zn2SnO4, showed a dramatic effect on the
microstructure, yielding smaller grain size and lower density. The main mechanism for inhibiting grain growth was the
pinning of grain boundaries due to the presence of the Zn2SnO4 particles. The incorporation of a small amount of SnO2
(x ꢁ 0.01) led to a marked increase in both the electrical conductivity and the absolute value of the Seebeck
coefficient. The increase in electrical conductivity was mainly caused by an enhanced electron concentration. On the
other hand, for high SnO2 content (0.02 ꢁ x ꢁ 0.05), the addition of SnO2 lowered the electrical conductivity and the
absolute value of the Seebeck coefficient. The low electrical conductivity mainly originated from an increase in a
second phase Zn2SnO4 and from a decrease in the density and grain size. The thermoelectric power factor was
maximized to a value of 1.25 ꢂ 10ꢀ3 Wmꢀ1 Kꢀ2 at 1073 K for the Zn0.99Sn0.01O sample.
Acknowledgment
The authors would like to acknowledge the financial support provided for this research by the Korea Energy
Management Corporation (KEMCO).
Reference
[1] C.M. Bhandari, D.M. Rowe, in: D.M. Rowe (Ed.), CRC Handbook of Thermoelectrics, CRC Press, Boca Raton, 1995, pp. 43–46.
[2] T. Tokiai, T. Uesugi, J. Am. Ceram. Soc. 78 (1995) 1089.
[3] T. Kojima, Phys. Status Solidi A111 (1989) 233.
[4] W.-S. Cho, S.-W. Choi, K. Park, Mater. Sci. Eng. B68 (1999) 116.
[5] M. Ito, W.-S. Seo, K. Koumoto, J. Mater. Res. 14 (1999) 209.
[6] D.M. Rowe, V.S. Shukla, J. Appl. Phys. 52 (1981) 7421.
[7] C.B. Vining, W. Laskow, J.O. Hanson, R.R. Van der Beck, P.D. Gorsuck, J. Appl. Phys. 69 (1991) 4333.
[8] M. Ohtaki, D. Ogura, K. Eguchi, H. Arai, J. Mater. Chem. 4 (1994) 653.
[9] W. Shin, N. Murayama, Mater. Lett. 51 (2001) 351.
[10] H. Muta, A. Ieda, K. Kurosaki, S. Yamanaka, Mater. Lett. 58 (2004) 3868.
´
[11] M. Peiteado, Y. Iglesias, J.F. Fernandez, J. De Frutos, A.C. Caballero, Mater. Chem. Phys. 101 (2007) 1.
[12] S. Major, S. Kumar, M. Bhatnagar, K.L. Chopra, Appl. Phys. Lett. 49 (1986) 394.
[13] M.G. Ambia, M.N. Islam, M.O. Hakim, J. Mater. Sci. 27 (1992) 5169.
[14] C.H. Know, H.K. Hong, Sens. Actuators B 24–25 (1995) 610.
[15] J. Arnovitch, D. Golmaya, A.L. Fchrenbruch, R.H. Bube, J. Appl. Phys. 1 (1980) 4260.
[16] K. Park, K.U. Jang, H.-C. Kwon, J.-G. Kim, W.-S. Cho, J. Alloys Compounds 419 (2006) 213.
[17] C. Wang, X.M. Wang, B.Q. Xu, J.C. Zhao, B.X. Mai, P.A. Peng, G.Y. Sheng, J.M. Fu, J. Photochem. Photobiol. A: Chem. 168 (2004) 47.
[18] B.L. Zhu, C.S. Xie, W.Y. Wang, K.J. Huang, J.H. Hu, Mater. Lett. 58 (2004) 624.
ˇ
[19] S. Bernik, N. Daneu, A. Recnik, J. Eur. Ceram. Soc. 24 (2004) 3703.
[20] P. Cotterill, P.R. Mould, Recrystallization and Grain Growth in Metals, Wiley, New York, 1976, pp. 150–163.
[21] JCPDS file No. 24-1470.
[22] R.J. Hill, J.R. Craig, G.V. Gibbs, Phys. Chem. Miner. 4 (1979) 317.
[23] D.L. Young, D.L. Williamson, T.J. Coutts, J. Appl. Phys. 91 (2002) 1464.
[24] H.Q. Chiang, J.F. Wager, R.L. Hoffman, J. Jeong, D.A. Keszler, Appl. Phys. Lett. 86 (2005) 13503.
[25] D.L. Young, H. Moutinho, Y. Yan, T.J. Coutts, J. Appl. Phys. 92 (2002) 310.
´ `
[26] A. Bougrine, M. Addou, A. Kachouane, J.C. Bernede, M. Morsli, Mater. Chem. Phys. 91 (2005) 247.
[27] M. Jin, J. Feng, Z. De-heng, M. Hong-lei, L. Shu-ying, Thin Solid Films 357 (1999) 98.
´
[28] M. Peiteado, J.F. Fernandez, A.C. Caballero, J. Eur. Ceram. Soc. 25 (2005) 2999.
[29] I. Terasaki, Y. Sasago, K. Uchinokura, Phys. Rev. B56 (1997) R12685.
[30] I. Terasaki, Mater. Trans. 42 (2001) 951.
[31] W. Shin, N. Murayama, Mater. Lett. 49 (2001) 262.
[32] W. Shin, N. Murayama, Mater. Lett. 45 (2000) 302.
[33] J.-W. Moon, Y. Masuda, W.-S. SeO, K. Koumoto, Mater. Lett. 48 (2001) 225.
[34] M. Ohtaki, T. Tsubota, K. Eguchi, H. Arai, J. Appl. Phys. 79 (1996) 1816.