Jul 15, 2005 - Thermoelectric properties of non-doped amorphous zinc oxide (ZnO) thin film were ... semiconductor at room temperature.1) The ZnO thin film.
Materials Transactions, Vol. 46, No. 7 (2005) pp. 1470 to 1475 Special Issue on Thermoelectric Conversion Materials #2005 The Thermoelectrics Society of Japan
Thermoelectric Properties of Amorphous Zinc Oxide Thin Films Fabricated by Pulsed Laser Deposition Yoshihiro Inoue1; * , Masaki Okamoto1; *, Toshio Kawahara2 , Yoichi Okamoto1 and Jun Morimoto1 1 2
Department of Materials Science and Engineering, National Defense Academy, Yokosuka 239-8686, Japan The Institute of Scientific and Industrial Research, Osaka University, Ibaraki 567-0047, Japan
Thermoelectric properties of non-doped amorphous zinc oxide (ZnO) thin film were studied at between room temperature and around 673 K, where the samples were prepared by pulsed laser deposition at room temperature. The results of X-ray diffraction, transmittance spectroscopy and electron microscope observation indicate that both as-deposited film and heat treated film had amorphous structure with random hexagonal network of ZnO, where the latter included crystallized ZnO grains. The highest absolute value of Seebeck coefficient was 135 mV/K at 643 K, where the electrical resistivity was 1:01 104 m and the power factor P was 1:79 104 W/mK2 , and the was almost the same as those of other ZnO compounds such as (ZnO)5 In2 O3 and (Zn1x Alx )O. The heat treatment dependence of the and the indicated the decrease of free electrons originated in amorphous and the transition to the quasi-stable phase of amorphous. It was suggested that the amorphous structure with variable range hopping conduction relates to the thermoelectric properties of a-ZnO thin film. (Received October 20, 2004; Accepted May 18, 2005; Published July 15, 2005) Keywords: thermoelectric, zinc oxide, thin film, amorphous, pulsed laser deposition
1.
Introduction
Zinc oxide (ZnO) has attracted much attention as one of the promising materials of optical devices for ultra violet region because of direct wide-band-gap (about 3.2 eV) semiconductor at room temperature.1) The ZnO thin film has practical transparency and high conductivity required by solar cell transparent electrodes and the other practical devices in place of indium tin oxide. On the other hand, bulk materials of ZnO polycrystalline and mixed crystals have been used for various practical applications in the past, such as varistors, piezoelectric transducers, surface acoustic wave filters and so on. Recently, in thermoelectric materials of n-type ZnO compounds, (ZnO)5 In2 O3 and (Zn1x Alx )O have been investigated eagerly for utilized as p-type partners for thermoelectric module production,2,3) but n-type oxides still show rather low thermoelectric performance compared to p-type oxide such as NaCoO2 and Ca3 Co4 O9 .4,5) These thermoelectric oxides have the advantage of bulk materials for thermoelectric power generation, but for the sensor, film materials are advantageous over bulk because the response time is fast due to small heat capacity and fast heat transfer. For example, Shin et al. reported hydrogen gas sensors using thermoelectric with Nickel oxide film coated with platinum catalyst on half of its surface, where the Seebeck coefficient is 708 mV/K at 295 K and temperature difference is 0.12 K.6) We have reported the anomalous large thermoelectric power of the Si–Ge–Au amorphous alternately deposited thin film, and consider that the origin of huge thermoelectric power is related with amorphous structure.7,8) In this work, we have prepared non-doped amorphous ZnO (a-ZnO) thin films by pulsed laser deposition (PLD) to evaluate the thermoelectric properties. Then we examine the influence of the fatigue by thermal cycling with heat treatment on the amorphous structure in a-ZnO thin film. PLD is one of the deposition techniques to give nearly *Graduate
Student, National Defense Academy
stoichiometric films, and one can easily fabricate controlled oxide thin films because it can control oxygen gas pressure independent of other growth conditions.9) 2.
Experiments
The a-ZnO thin film was deposited by various techniques.10–12) We prepared the non-doped a-ZnO thin films deposited on R-face (011 2) sapphire substrate by PLD with an ArF (193 nm) excimer laser. The ablating stoichiometric target was the ZnO ceramics (99.99% purity, JAPAN PURE CHEMICAL), and the substrate was positioned at the opposite side to the target, and the distance between the substrate and the target was 35 mm. The pulsed laser with a repetition rate of 10 Hz was focused onto the target with an energy density of about 1 J/cm2 . The deposition rate of the films was about 10 nm/min. The substrate temperature was kept at room temperature during deposition. Usually one can obtain crystalline films on substrates already heated to the optimal crystallization temperature. Thus, it could be expected that abrupt quenching onto cold substrate for evaporation of target material by laser heating forms the amorphous structure reflected to an extent the composition of target. The pressure was kept on the order of 105 Pa during the film deposition. Film thickness was measured by a surface profile measurement system (DEKTAK 3, Veeco), and the crystal structure was evaluated by X-ray diffraction (XRD; X’pert MRD, PHILIPS). The transmission of the films was measured in the wavelength range from 300 nm to 800 nm using a spectrophotometer (V-570, JASCO), and the surface morphology of the films was observed by scanning electron microscopy (SEM; S-4500, HITACHI). We simultaneously measured both the electrical resistivity and Seebeck coefficient in temperature range between room temperature and around 673 K, and estimated the power factor P (¼ 2 =) with these parameters. In order to obtain ohmic contact with small contact resistivity, the aluminum rectangular contacts were made on the four corners of a-ZnO
Thermoelectric Properties of Amorphous Zinc Oxide Thin Films Fabricated by Pulsed Laser Deposition
Intensity (a. u.)
sapphire (012)
1471
sapphire (024)
as-deposited
ZnO (110) after 10th cycle
30 40 50 60 70 Diffraction Angle, 2θ /deg.
80
Fig. 1 X-ray diffraction patterns of a-ZnO thin films for (a) as-deposited and (b) after 10th cycle.
thin film surface by vacuum evaporation. The and of a-ZnO thin films were measured by the four-terminal method and the conventional DC method in an infrared furnace under nitrogen gas with a small flow rate (about 30 cm3 /min), where the sample was heated and cooled at the rate of 10 K/ min. This heat treatment as one cycle was operated to total 10 cycles with temperature difference T of substrate both ends over 10 K. 3.
Results and Discussion
The a-ZnO thin film sizes on sapphire substrate (10 mm 10 mm) was almost 6 mm 10 mm, and film thickness was about 150 nm for as-deposited and about 135 nm after 10th cycle of heat treatment. We show the XRD patterns of as-deposited film and after 10th cycle in Figs. 1(a) and (b), respectively. The broad diffraction peak is shown at around 56 in addition to substrate peaks of sapphire (012) and (024) on pattern in Fig. 1(a) for as-deposited film. On the other hand, the pattern in (b) for after the 10th cycle shows ZnO(110) peak at the same angle as the broad diffraction peak on the pattern in Fig. 1(a). This indicates that a part of aZnO changed into the ZnO(110) grain due to heat treatment. We show the SEM image of a-ZnO thin film after 10th cycle in Fig. 2. The ZnO crystal grains with a diameter of range between about 50 and 100 nm were observed. As for as-deposited films, the micrograph shows the almost smooth surface without rough of grains although the grain boundaries-like can be observed in places. This result agrees with the result of XRD and indicates that the partial crystallization of amorphous was promoted due to heat treatment up to about 673 K. Figure 3 indicates the change of UV–vis transmittance for the a-ZnO thin films with as-deposited and after 10th cycle, where each spectrum are represented by solid lines and broken lines, respectively. The transmittance below 360 nm for as-deposited film exhibits about 30% and this is caused by the partial porous-like morphology. For both spectra, the transmittances increase from 380 nm and were ascribed to absorption edge of hexagonal ZnO crystal. Then, compared with the gradient of increase from 380 to 400 nm, that of the
Fig. 2
SEM micrograph of a-ZnO thin film after 10th cycle.
100 80
Transmittance (%)
20
60
as-deposited after 10th cycle
40 20 0 300
400
500
600
700
800
Wavelength, λ /nm Fig. 3 UV–vis transmittance spectra of a-ZnO thin films for as-deposited and after 10th cycle designated by solid line and dotted line, respectively.
spectrum for after 10th cycle is sharper than as-deposited film. In visible region between 400 and 700 nm, the transmittance of film after 10th cycle is higher than the asdeposited film. Actually, we could initially observe the asdeposited film surface as transparent dark brown color, and the color changed to more transparent light brown after 10th cycle. Hence, considering XRD and micrograph results, we suggest that a-ZnO deposited by PLD has an incomplete and defective (or degenerated) band structure of ZnO single crystal, and the heat treatment caused the almost amorphous state with including ZnO crystals partly. Here, we redefine that the term of amorphous designates the state with formation of random hexagonal network including a part of ZnO single crystals, that is, mixed state of both almost crystal-like phase and a fraction of crystallized phase of the ZnO grain. The temperature dependence of , and P in the 1st, 5th and 10th cycle are shown in Figs. 4(a), (b) and (c), respectively, where we plot the values at T > 10 K. In Fig. 4(a), the in every cycle decrease as temperature decrease, which implies that a-ZnO incline to be characteristic of semiconductor, and the lowest and highest values is
1472
Y. Inoue, M. Okamoto, T. Kawahara, Y. Okamoto and J. Morimoto
-4 1st cycle 5th cycle 10th cycle
0 (b)
-60 10 -4 -90
ρ /Ωm
10
-5
α /µVK-1
α ρ
-30
10
P /10 -4 Wm-1K-2
10 -3
0
(a)
α /µVK-1
ρ /Ω m
10-3
-50 -120
-100 -150 2.0 (c) 1.5 1.0
1st cycle 5th cycle 10th cycle
-150
1
2
3
4
5
6
7
8
9
10
10 -5
Heat treatment (cycle) Fig. 5 Heat treatment dependence of and are plotted by open triangles and open circles, respectively. The each abosolute value is highest and lowest at each cycle.
1st cycle 5th cycle 10th cycle
5.0 0.0 300
400
500
600
700
Temperature, T /K Fig. 4 Temperature dependence of (a) electrical resistivity , (b) Seebeck coefficient and (c) power factor P for a-ZnO thin film in cooling process with heat treatment. The values in the 1st cycle, the 5th cycle and 10th cycle are plotted by symbols of open circles, open triangles and closed circles, respectively.
about 1:01 104 m and 4:36 104 m. In Fig. 4(b), the absolute value of in every cycle increases as temperature increase, and the highest and lowest value is about 135 mV/K and 30 mV/K, respectively. In Fig. 4(c), the highest and lowest value is about 1:79 104 W/mK2 and 2:10 106 W/mK2 . Consequently, the lowest value of , the highest absolute value of and the highest value of P were obtained at 643 K in the 1st cycle, and the highest value of , the lowest absolute value of and the lowest value of P were obtained at 334 K in 10th cycle. Now, compared with the of remarkable n-type ZnO compounds for thermoelectric bulk materials, Malochlin et al. reported that the absolute value of (ZnO)5 In2 O3 single crystal and textured ceramics were about 17 mV/K and 65 mV/K at about 700 K, respectively (Fig. 3 in ref. 2). Ohtaki et al. reported that the of (Zn0:98 Al0:02 )O ceramics was about 150 mV/K at about 673 K (Fig. 1 in ref. 3). For the films of (ZnO)5 In2 O3 with two kinds of orientation prepared by radio frequency sputtering method, Hiramatsu et al. reported that the absolute value of was about 110 mV/K at about 650 K.13) In our case, it is found that the a-ZnO deposited by PLD has also the almost same absolute value of for about 135 mV/K at 643 K. This experimental fact suggests that non-doped a-ZnO films fabricated by PLD have an advantage over ZnO-related thermoelectric materials such as above in view of complexed process in fabricating.2,3,13) Because this sample is not require the accurate control of temperature and/or composition of precursor with dopant. On the other hand, in Fig. 4(a), the decrease for with temperature increase seems almost same in each cycle and the shift from the 5th cycle to 10th cycle is smaller than that from the 1st cycle to the 5th cycle. This shift, that is, the heat
treatment cycle dependence of the is also seen in that of the in Fig. 4(b) and that of the P in Fig. 4(c). In Fig. 4(b), the absolute value of each cycle increase similarly up to about 625 K, but we can see that the absolute value of for 10th cycle is larger than that for the 5th in the region over about 625 K, and the slope of increase is larger. As for P in the higher temperature region in Fig. 4(c), the value for 10th cycle is also larger than that for the 5th such as as shown in Fig. 4(b), and where the sharp increase in the 1st cycle is shown. The lowest and highest seem to be dependent on the heat treatment. So, we show the heat treatment dependence of highest absolute value for and lowest value for at each cycle in Fig. 5, where and are designated by symbol of circles and triangles, respectively. The absolute value of decreases up to 6th cycle, and then is leveled off following the small increase after 6th cycle. The increases as the heat treatment increase, where the earlier stage of heat treatment cycle indicates the larger rate of increase than the later stage. These variations of and should relate with both the decrease of carrier and the change of the structure of a-ZnO. Generally, ZnO thin films deposited by physical and chemical vapor deposition have many free electrons caused by oxygen defects and so on, that show n-type conduction at room temperature.9,14) Hence, the as-deposited a-ZnO thin film should have so many multiple structural defects originated in the amorphous, such as dangling bonds and so on, that make free electrons in addition to native defects in ZnO grains and interface.15) As a result, the excess free electrons seem to cause the initial low as shown in Fig. 5. Here, the conduction of amorphous semiconductors such as amorphous-Silicon is known to suit the variable range hopping (VRH) conduction model,16) where this VRH conduction is often observed in low temperature range below the room temperature and the mechanism is frequently associated with Anderson transition in amorphous materials. Whereas, metal oxides such as perovskite-type oxide exhibit the metal-insulator transition in the relation between electrical resistivity and temperature,17) and then, Ag-doped La2 CuO4 thin films fabricated using PLD technique show the VRH behavior.18) Previously, we tried to analyze the
Thermoelectric Properties of Amorphous Zinc Oxide Thin Films Fabricated by Pulsed Laser Deposition
¼ A expðB=T 1=4 Þ;
B ¼ const½a3 =NðEF Þ1=4 :
9.5 1st cycle 5th cycle 10th cycle
ln (σ/Sm-1)
9.0
8.5
8.0
7.5
0.20
0.21 0.22 0.23 Temperature, T -1/4/K-1/4
Fig. 6 Temperature dependence of ln , and temperature axis is plotted as T 1=4 . The value in the 1st cycle, the 5th cycle and 10th cycle are plotted by symbols of open circles, open triangles and closed circles, respectively.
150 Absolute value of α/µVK-1
thermoelectric properties for YBa2 Cu3 O6þx by the theoretical expression of the thermoelectric power under the VRH conduction model in temperature range above the room temperature.19) Moreover, as for the other hopping conduction in perovskite-type oxide, Ohtaki et al. reported that small polaron hopping for CaMnO3 -based oxide, which was characterized by thermally activated hopping of a localized electronic carrier, would be advantageous to proceed beyond a theoretical limit predicted for the conventional broadband materials.20) For ZnO films, Tiwari et al. investigated electrical transport mechanism for c-axis oriented ZnO1 films grown on quartz substrates (923 K) by PLD at various oxygen partial pressure, and presented that the electrical transport in oxygen deficient Anderson localized ZnO1 films was dominated by the VRH conduction of charge carriers between the localized states.21) Furthermore, Schoenes et al. have measured the complex dielectric function of sputtered high-resistivity ZnO films over a temperature range 60 to 373 K, and the results revealed the contribution of band conduction and hopping conduction to the dielectric dispersion. In that regard, they pointed out that the trapped electron in an intrinsic defect not only conducted current in the conduction band when it is thermally activated but also hopped from one singly charged defect to the next with creating a dipole.22) Thus, we can consider that the conduction in a-ZnO film relates to the hopping conduction of some kind because of having the amorphous structure and the multiple structural defects as hopping sites originated in an amorphous phase and trap levels in ZnO grains and grain boudaries.9,14,15) Hence, we venture a hypothesis that a-ZnO film might show the VRH related conduction in higher temperature range above room temperature due to the above, although it should be expected that the VRH conduction is observed in lower temperature range below room temperature in addition to the other hypothesis base on conduction models of polaron hopping20) or nearest neighbor hopping.23) Here, the well known explanation for the carrier transport of VRH is expressed by T 1=4 law,24)
1473
1st cycle 5th cycle 10th cycle
100
50
0
19 20 21 22 23 24 25 Temperature, T 1/2/K1/2
Fig. 7 Temperature dependence of absolute value of , and temperature axis is plotted as T 1=2 . The value in the 1st cycle, the 5th cycle and 10th cycle are plotted by symbols of open circles, open triangles and closed circles, respectively.
ð1Þ
Where, is the electrical conductivity, A is the numerical factor, a is the reciprocal of the localization length, NðEF Þ is the density of state near the Fermi energy. Thus, we plot the as T 1=4 of the temperature axis in Fig. 6, where the longitudinal axis is plotted as the ln . We can see T 1=4 dependence in each cycle below about 515 K (above about 0.21 K1=4 ), where the solid line are a guide for the eyes. This indicates that the conduction of a-ZnO in this temperature range is dominated by VRH. However, the temperature dependence of the of 1st cycle in particular in higher temperature range is clearly changed as represented by the broken line, and that might indicate overlap of the other carrier transport originated in band conduction with thermal activation. Hence, we apply this hopping conduction model to the thermoelectric power. In this model, temperature dependence of the is known to be / T 1=2 .25) We plot the T 1=2 dependence of in Fig. 7, where the longitudinal axis is plotted as absolute value of . We can see T 1=2 dependence between about 380 K and about 530 K (about 19.5 K1=2 and
about 23 K1=2 ) in each cycle, where the solid line are a guide for the eyes. However, the discrepancy of the T 1=2 dependence is seen in the higher temperature region, and the tendency seems more apparent in 10th cycle than 1st and 5th cycle. This indicates that the conduction based on VRH contribute in 1st cycle with strong dominance. Here, if this contribution of VRH for has an effect on the improvement of thermoelectric power, the d ln NðÞ=d, which is the gradient of the density of state, in the conventional T 1=2 law for of VRH should be effected by unexpected something originated in amorphous structure of a-ZnO film, where T 1=2 law is expressed by25) 1 2 1=2 d ln NðÞ ðT0 TÞ ¼ / T 1=2 : ð2Þ 3e d ¼0 Where e is unit charge, and both and T0 are a numerical factor defined in eq. (1). In this model, the conduction is based on only VRH between localized electronic states near the Fermi energy. Hence, as for the a-ZnO film in this study,
1474
Y. Inoue, M. Okamoto, T. Kawahara, Y. Okamoto and J. Morimoto
the based on the band conduction is not suited as the expression of thermoelectric power generally expressed by k 5 2ð2 mkTÞ3=2 ¼ þ þ ln : ð3Þ e 2 nh3 Where k is the Boltzmann constant, is the dðln Þ=dðln EÞ as the scattering relaxation time of carrier , m is the effective mass of carrier, n is the density of carrier, and h is Plank constant. In this case, a semiconductor without degeneration and VRH conduction is corresponded, and the is getting larger with lower density of carrier or heavier effective mass. If the a-ZnO in this study is a semiconductor without degeneration, the of a-ZnO should be smaller due to the excess free electrons caused by so many multiple structural defects originated in the amorphous. Then, when the heat energy was provided for a-ZnO, the crystallization of a-ZnO begins, that is, the multiple structural defects introduced in as-deposited a-ZnO thin film start to vanish due to the order reconstruction. However, in Fig. 7, the of 1st cycle is higher than that of 5th and 10th cycle, although the scattering relaxation time of carriers decreases due to lattice disorder in 1st cycle. And then, the heat treatment cycle dependence of highest absolute value for the as shown in Fig. 5 is also shown similarly. Hence, this origin of higher thermoelectric power should be in VRH conduction in a-ZnO film, and it is possible that the amorphous structure of a-ZnO effect to promote the absolute value of the d ln NðÞ=d near the Fermi energy in eq. (2). Moreover, as for progressed stage of crystallization in latter cycles between 6th and 10th in Fig. 5, the might to be related to the reconstruction of amorphous network which let effective mass of carrier in a-ZnO to get heavy, where the grains of ZnO crystal from a fraction of a-ZnO are much grown with the expansion and contraction of Zn–O network. And then, it is considered that these transformations in film result the heat treatment dependence of thermoelectric properties in Fig. 5. Where, the crystallization up to the 6th cycle affect mainly the decrease of electrons originated in amorphous with dependence on the decrease of the absolute value of the d ln NðÞ=d in view of the increase of the . Then, from 6th cycle to 10th cycle, the increase of the absolute value of the and the steady increase of the might indicate the contribution of the band conduction with the expansion and contraction of Zn–O network in addition to the hopping conduction in the amorphous layer. In this model of carrier transport between 6th cycle and 10th cycle, we consider that the transition to the quasi-stable phase in a-ZnO film is achieved in the latter cycle stage, and it is possible that this transition is caused by the grain boundary of ZnO crystals in amorphous network. However, this conclusion require the further analysis using an effective-medium approximation26) and theory of percolation on a continuum and the localization-delocalization transition.27) Thus, the heat treatment dependence of the in particular as shown in Fig. 5 cannot be explained easily and successfully by conventional theory of thermoelectric property for semiconductor without degeneration. We conclude that the origin of the thermoelectric power of a-ZnO is related to the amorphous structure with VRH conduction in addition to the crystallization with band conduction and the character of
ZnO itself. Recently, transparent flexible thin film transistors using amorphous oxide semiconductors consisted of indium– gallium–zinc–oxygen (IGZO) system fabricated by PLD at room temperature have been reported.28) In view of the application for thermoelectric device, although the is still insufficient for practical thermoelectric power, the a-ZnO film by PLD at room temperature is possibly promising material as the hydrogen gas sensor6) or infrared sensor under severe condition not as thermoelectric power generation. Because a-ZnO film under atmosphere is chemical stable due to oxide itself, and the thermoelectric power are obtained the same level as (Zn1x Alx )O and (ZnO)5 In2 O3 at high temperature, where a-ZnO film with a small size can be fabricated easily at room temperature without such difficulty in processing as required by other many thermoelectric oxides.2,3,13) 4.
Conclusions
The a-ZnO thin film was deposited by PLD at room temperature. The XRD patterns of as-deposited and after heat treatment indicated the amorphous structure and the crystallization of a fraction of a-ZnO into ZnO (110) crystal. We could also observe the rough surface with ZnO crystal grains for the heat treated film in the SEM image. The results of transmittance spectra of both films showed the band structure of ZnO like, and the heat treated film had more an absorption of the band edge. The thermoelectric properties of the , and P were studied at between room temperature and around 673 K, and the highest absolute value of was 135 mV/K at 643 K in the 1st cycle of heat treatment, where the was 1:01 104 m and the P was 1:79 104 W/mK2 . The a-ZnO thin film had the comparable to those of ZnO compounds for thermoelectric materials. The heat treatment dependence of the based on the VRH conduction indicated the decrease of free electrons originated in amorphous structure at earlier stage of heat treatment cycle in particular, and between 6th cycle and 10th cycle, we consider that the transition to the quasi-stable phase in a-ZnO film is achieved, and it is possible that this transition is caused by the grain boundary of ZnO crystals in amorphous network. We can conclude that the thermoelectric properties for a-ZnO thin film are related to the amorphous structure with VRH conduction in addition to the crystallization with band conduction and the character of ZnO itself, and the a-ZnO thin film is ideal for thermoelectric device such as the hydrogen gas sensor or infrared sensor under severe condition not as thermoelectric power generation. REFERENCES 1) P. Zu, Z. K. Tang, G. K. L.Wong, M. Kawasaki, A. Ohtomo, H. Koinuma and Y. Segawa: Solid State Commun. 103 (1997) 459–463. 2) O. Malochkin, W.-S. Seo and K. Koumoto: Jpn. J. Appl. Phys. 43 (2004) L194–L196. 3) M. Ohtaki, T. Tsubota, K. Eguchi and H. Arai: J. Appl. Phys. 79 (1996) 1816–1818. 4) I. Terasaki, Y. Sasago and K. Uchinokura: Phys. Rev. B 56 (1997) R12685–R12687. 5) M. Shikano and R. Funahashi: Appl. Phys. Lett. 82 (2003) 1851–1853. 6) W. Shin, K. Imai, N. Izu and N. Murayama: Jpn. J. Appl. Phys. 40
Thermoelectric Properties of Amorphous Zinc Oxide Thin Films Fabricated by Pulsed Laser Deposition (2001) L1232–L1234. 7) Y. Okamoto, H. Uchino, T. Kawahara and J. Morimoto: Jpn. J. Appl. Phys. 38 (1999) L945–L947. 8) H. Uchino, Y. Okamoto, T. Kawahara and J. Morimoto: Jpn. J. Appl. Phys. 39 (2000) 1675–1677. 9) S. Hayamizu, H. Tabata, H. Tanaka and T. Kawai: J. Appl. Phys. 80 (1996) 787–791. 10) P. Pushparajah, A. K. Arof and S. Radhakrishna: J. Phys. D 27 (1994) 1518–1521. 11) N. Asakuma, H. Hirashima, T. Fukui, M. Toki, K. Awazu and H. Imai: Jpn. J. Appl. Phys. 41 (2002) 3909–3915. 12) Z. Wang, H. Zhang, Z. Wang, L. Zhang and J. Yuan: J. Mater. Res. 18 (2003) 151–155. 13) H. Hiramatsu, H. Ohta, W.-S. Seo and K. Koumoto: J. Jpn. Soc. Powder & Powder Metall. 44 (1997) 44–49. 14) F. D. Auret, S. A. Goodman, M. J. Legodi, W. E. Meyer and D. C. Look: Appl. Phys. Lett. 80 (2002) 1340–1343. 15) J. F. Cordaro and Y. Shim: J. Appl. Phys. 60 (1986) 4186–4190. 16) B. Chakrabarti, S. Chaudhuri and A. K. Pal: Jpn. J. Appl. Phys. 22
1475
(1983) 903–907. 17) P.-X. Thao, T. Tsuji, M. Hashida and Y. Yamamura: J. Cera. Soc. Jpn. 111 (2003) 544–547. 18) C. C. Wang and J. Zhu: Thin Solid Films 468 (2004) 75–78. 19) T. Kawahara, S. Tamura, H. Inai, Y. Okamoto and J. Morimoto: J. Mater. Res. 14 (1999) 1200–1203. 20) M. Ohtaki, T. Tokunaga, K. Eguchi and H. Arai: Proc. 16th Int. Conf. on Thermoelectrics, (IEEE, 1997) pp. 224–227. 21) A. Tiwari, C. Jin and J. Narayan: J. Appl. Phys. 96 (2004) 3827–3830. 22) J. Schoenes, K. Kanazawa and E. Kay: J. Appl. Phys. 48 (1977) 2537– 2542. 23) J. Han, M. Shen, W. Ceo, A. M. R. Senos and P. Q. Mantas: Appl. Phys. Lett. 82 (2003) 67–69. 24) N. F. Mott: J. Phys. C 13 (1980) 5433–5471. 25) I. P. Zvyagin: Phys. Status Solidi B 58 (1973) 443–449. 26) T. Suzuki and S. Adachi: J. Jpn. Appl. Phys. 32 (1993) 4900–4906. 27) R. Zallen and H. Scher: Phys. Rev. B 4 (1971) 4471–4479. 28) K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano and H. Hosono: Nature 432 (2004) 488–492.
