www.advmat.de www.MaterialsViews.

com

COMMUNICATION

Enhancement of the Thermoelectric Figure-of-Merit in a Wide Temperature Range in In4Se3xCl0.03 Bulk Crystals
Jong-Soo Rhyee,* Kyunghan Ahn, Kyu Hyoung Lee,* Hyo Seok Ji, and Ji-Hoon Shim
Because of the increasing awareness of renewable energy issues, much attention has been devoted to thermoelectric energy harvesting technology. The dimensionless thermoelectric gureof-merit is dened by ZT = S2T/, where S, , T, and are the Seebeck coefcient, electrical conductivity, absolute temperature, and thermal conductivity, respectively. Regarding a high ZT, it has long been sought both by lowering the thermal conductivity by exploiting the phonon-glass and electron-crystal concept[15] and by enhancing the power factor S2 by utilizing the quantum connement effect[6,7] in low-dimensional nanostructured materials. Recently, we proposed that the charge density wave represents a new direction for high thermoelectric performance in bulk crystalline materials.[8,9] Low-dimensional electronic transport with strong electronphonon coupling breaks the translational symmetry of the lattice because of the energy instability in a high-symmetry crystalline lattice.[10] The lattice dimerization along the electronic transport plane lowers the lattice thermal conductivity as a result of lowering of the phonon energy. Through quasi one-dimensional lattice distortion (Peierls distortion) in In4Se3x bulk single crystals, we achieved a high ZT of 1.48 at 705 K.[9] However, two challenges remain for practical applications. Firstly, the reported ZT could be increased further if we could increase the carrier concentration of the In4Se3x crystals because it is far from the carrier concentration of a heavily doped semiconductor (on the order of 1019 cm3) that is generally considered to be optimal for thermoelectric materials. Secondly, ZT decreases signicantly as the temperature decreases, which limits the operational temperature range to within 350 430 C.[9] Here, we report a signicant increase in ZT (maximum ZT (ZTmax) = 1.53) over a wide temperature range in chlorine-doped In4Se 3x Cl0.03 compounds mainly as a result of the increase in the electrical conductivity because of the increase of both the carrier density and Hall mobility. X-ray diffraction patterns of the cleaved surface of In4Se2.67Cl0.03 single crystals (left axis) and the crosssectional plane of In4Se2.35 and In4Se2.32Cl0.03 bulk crystal ingots (right axis) are presented in Figure 1. The easily cleavable In4Se2.67Cl0.03 single crystal shows very sharp (h20) peaks indicating a well-aligned single crystal, while the In4Se2.32Cl0.03 has other peaks, including (16l), (42l), and (h 20) peaks. The In4Se2.35 bulk crystal ingot, with a high ZT of 1.48 at 705 K,[9] has several preferred orientation planes of {h00}, {k00}, and other crystal planes. Figure 2 shows the thermoelectric properties of the chlorine-doped compounds of In4Se2.32Cl0.03, In4Se2.67Cl0.03 bulk single crystals, and the previously reported[9] In4Se2.35 crystal for comparison. Because high thermoelectric performance was revealed along the bc-plane of the crystal containing charge density wave lattice instability, we measured the thermoelectric properties along that plane. The thermal conductivity decreases as the temperature increases, mainly from acoustic phonon contribution, as shown in Figure 2a. The inherent nature of van der Waals interactions along the a-axis and Peierls distortion in the bc-plane render the low thermal conductivity. The of In4Se2.32Cl0.03 is comparable to that of In4Se2.35, while the of In4Se2.67Cl0.03 has a relatively low value. The thermal conductivity reduction mechanism by the quasi-one-dimensional Peierls distortion is intrinsically revealed in the measurements along that plane. The highly anisotropic crystal orientation of In4Se2.67Cl0.03 has lower thermal conductivity than other compounds. The thermal conductivity consists of the lattice ph and electronic thermal conductivity el. In usual cases, the electronic thermal conductivity el can be obtained by the Wiedemann Frantz (WF) law, el = L0T, where L0, , and T are the Lorenz factor, electrical conductivity, and absolute temperature, respectively. The lattice thermal conductivity of In4Se2.35 is obtained by subtracting the electronic thermal conductivity from the WF law, as shown in the open rectangular symbol in Figure 2a. However, the WF law is violated in a special case of bipolar transport.[11,12] When we apply the WF law using chlorinedoped samples, arbitrary negative lattice thermal conductivity is obtained, indicating the existence of bipolar transport. An estimation of the lattice thermal conductivity can be made in a plot of versus , which shows a linear relationship, as shown in the inset of Figure 2a, because the thermal and electrical conductivities follow 1/T temperature dependence in both cases. The extrapolation of the thermal conductivity to zero electrical conductivity gives the lattice thermal conductivity at a high temperature limit. A rough estimation of the high temperature lattice thermal conductivity of these compounds is presented in

Prof. J.-S. Rhyee Department of Applied Physics College of Applied Science Kyung Hee University 1 Seocheon-dong, Giheung-gu, Yongin-si, Gyeonggi-do 446-701 Republic of Korea E-mail: jsrhyee@khu.ac.kr; kyuhyoung.lee@samsung.com Dr. K. Ahn, Dr. K. H. Lee Advanced Materials Research Center Samsung Advanced Institute of Technology Nongseo-dong 14-1, Giheung-gu, Yongin-si, Gyeonggi-do 446-712, Republic of Korea E-mail: kyuhyoung.lee@samsung.com H. S. Ji, Prof. J.-H. Shim Department of Chemistry Pohang University of Science and Technology San 31, Hyojadong, Namgu, Pohang, Gyungbuk 790-784, Republic of Korea

DOI: 10.1002/adma.201004739

Adv. Mater. 2011, XX, 14

2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

wileyonlinelibrary.com

www.advmat.de www.MaterialsViews.com

COMMUNICATION
Figure 1. X-ray diffraction patterns of the cleaved surface of an In4Se2.67Cl0.03 single crystal (left axis) and the cross-sectional plane of an In4Se2.35 from Rhyee et al.[9] and In4Se2.32Cl0.03 bulk crystal ingot (right axis).

Table 1. The estimated lattice thermal conductivities are not signicantly changed by chlorine doping. The Seebeck coefcients decrease with chlorine doping, as shown in Figure 2b, because of the metallic behavior of electrical resistivity for chlorine-doped compounds, as shown in Figure 2c. When chlorine is doped into In4Se3x, the electrical resistivity decreases signicantly (e.g., 0.92 m cm at 50C for In4Se2.67Cl0.03) to two to three orders of magnitude lower than those of the undoped compound of In4Se2.35 (0.88 cm at 50 C). The chlorine-doped compounds exhibit metallic behavior of (T) rather than semiconducting behavior. The signicant decrease of (T) is mainly a result of the increase in the Hall mobility as well as the increase in the carrier concentration, as shown in Table 1. The improvement of the Hall mobility by chlorine doping can be attributed to the good alignment of the crystal along the bc-plane parallel to the direction of the transport property measurements, as these compounds have highly anisotropic transport properties. In compounds of In4Se3x, the selenium deciency gives rise to an increase in the carrier concentration.[13] However, in chlorine-doped In4Se3x Cl0.03 compounds, there is no straightforward relationship between selenium deciency and the carrier concentration because of the bipolar diffusion effect, which is expected from the violation of the WF law. The coexistence of electrons and holes in In4Se3x Cl0.03 complicates the behavior of the carrier density with selenium deciency. The higher carrier concentration of In4Se2.67Cl0.03 compared to that of In4Se2.32Cl0.03 indicates a heavily doped compound of In4Se2.67Cl0.03. It was found that lightly doped bismuth telluride

Figure 2. Temperature-dependent thermoelectric properties of the thermal conductivity (a), Seebeck coefcient (b), electrical resistivity (c), and Hall carrier concentration (inset of (c)) for In4Se2.35 (from Rhyee et al.[9]), In4Se2.32Cl0.03, and In4Se2.67Cl0.03 bulk crystals. The inset of (a) represents the thermal conductivity versus the electrical conductivity of the chlorinedoped samples, as indicated. The open rectangular symbol in (a) is the lattice thermal conductivity of In4Se2.35 obtained from the WF law. The pink inverse triangles with the solid line in (b) and (c) represent the Boltzmann transport calculation result for a xed chemical potential of = 0.36 eV.

showed higher thermal conductivity and lower electrical conductivity than that of a heavily doped compound.[14] The lightly doped In4Se2.32Cl0.03 exhibits higher thermal conductivity and lower electrical conductivity compared to that of In4Se2.67Cl0.03, as shown in the inset of Figure 2a.

Table 1. Estimated high-temperature lattice thermal conductivity ph and room-temperature electrical resistivity , carrier density nH, Hall coefcient RH, and Hall mobility |H| of the indicated samples.
Samples In4Se2.35[9] In4Se2.32Cl0.03 In4Se2.67Cl0.03

ph [W m1 K1]
0.54[a] 0.43[b] 0.50[b]

[ m]
8840 32.24 9.24

nH [x 1018 cm3] 4.10 6.70 11.00

RH [cm3 C1] 1.52 0.93 0.57

|H| [cm2 V1 sec1] 0.017 2.889 6.137

a) at 450 C from the WF-law. [b] High-temperature limit.

wileyonlinelibrary.com

2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2011, XX, 14

www.advmat.de www.MaterialsViews.com

COMMUNICATION

The temperature-dependent behavior of the Seebeck coefcient and electrical resistivity of In4Se2.67Cl0.03 can be reproduced by the Boltzmann transport calculation with a xed chemical potential of = 0.36 eV, as shown in Figure 2b and c (line and inverse triangle symbol). In the supplementary information of Rhyee et al.[9] it is conrmed that the chemical potential of = 0.22 eV is comparable to the thermoelectric properties in In4Se2.35. The power factor with respect to the chemical potential from the Boltzmann transport calculation indicated that an increase of the chemical potential to = 0.8 eV can increase the power factor as a result of optimization of the carrier concentration. The increase in the chemical potential by chlorine doping (0.36 eV) has an effect on the electron doping, giving rise to an increase in the electronic carrier concentration (1019 cm3), as shown in the inset of Figure 2c and in Table 1. More chlorine doping does not play a signicant role in the electron doping because the solubility limit of the In4Se 3x system is lower than 3 at%, as conrmed by noting that the excess chlorine segregates to the surfaces of the crystal in depthprole measurements by secondary ion mass spectroscopy. The metallic behavior and relatively high Seebeck coefcients in the chlorine-doped In4Se3x Cl0.03 bulk single crystals lead to temperature-insensitive behavior of the power factor (S2), as shown in Figure 3a. Compared to the signicant decrease of S2 as the temperature decreased in a previously reported case (black-closed square), the high power factors for a wide temperature range from 50 to 450C of chlorine-doped compounds are very important ingredients for practical applications owing to

the possibility of stable power generation for such a wide temperature range. Because of a low thermal conductivity and high power factor, the ZT of chlorine-doped compounds exhibits a high value in a broad temperature range, as presented in Figure 3 b. The ZTmax value reaches 1.53 at 425C, which is the highest value ever reported in n-type materials, including the In4Se2.35 (ZT ~1.48) compounds previously discovered.[9] In addition, the room-temperature ZT value increases signicantly from 0.005 for In4Se2.35 to 0.4 for In4Se2.67Cl0.03. When it is taken into account that the required ZT for practical power generation is approximately 0.8, the operational temperature range (Top) can be widened remarkably from 350430 C (Top = 80 C) for In4Se2.35 to 150450C (Top = 300 C) for the In4Se2.67Cl0.03. In summary, a signicant ZT enhancement (ZTmax = 1.53 at 425 C) over the entire measured temperature range makes chlorinedoped compounds of indium selenides promising for practical applications over a wide temperature range for waste heat power generation. In particular, the temperature-insensitive power factor and high ZT (0.4) at room temperature are highly exceptional characteristics, as other known thermoelectric compounds exhibit a signicant decrease in ZT as the temperature decreases. Further optimization of the chemical potential can increase ZT.

Experimental Section
Sample preparation: Bulk crystal ingots of In4Se3x Cl0.03 (x = 0.33 and 0.68) were prepared by the Bridgman crystal growth method. Stoichiometric amounts of In, Se, and InCl3 were loaded into a quartz tube under an argon atmosphere in order to keep the indium chloride dry. The quartz tubes were vacuum sealed and heat treated at 590 C for 12 h in a rocking furnace for homogenization of the molten liquid. After heat treatment of the compounds, the quartz ampoules were loaded into a vertical Bridgman furnace. The compounds were melted at 590 C for 24 h and slowly pulled at a constant speed of 25 mm h1. Sample characterization and measurements: Powder X-ray diffraction was conducted to conrm the single phase of the compounds. The X-ray diffraction pattern of the cleaved crystals showed that the bulk ingots contained sizable crystal domains (25 mm). The thermal conductivity = sCp was estimated from the combined measurements of the sample density (s), the DulongPetit tting of the heat capacity (Cp), and the thermal diffusivity () by the laser ash method (TC-9000, ULVAC, Japan). The Seebeck coefcient (S) and electrical resistivity () were measured by a four-probe method using a thermoelectric measurement system (ZEM-3, ULVAC, Japan). Hall resistivity measurements were carried out by the ve-point contact method using a physical property measurement system (PPMS, Quantum Design, USA). Received: December 27, 2010 Revised: February 7, 2011 Published online:

Figure 3. Temperature-dependent power factor (a) and ZT value (b) for In4Se2.35 (from Rhyee et al.[9]), In4Se2.32Cl0.03, and In4Se2.67Cl0.03 bulk crystals. The pink inverse triangles with the solid line in (a) represent the Boltzmann transport calculation result for a xed chemical potential of = 0.36 eV.

[1] B. C. Sales, Mater. Res. Soc. Bull. 1998, 23, 15. [2] G. J. Snyder, M. Christensen, E. J. Nishibori, T. Caillat, B. B. Iversen, Nat. Mater. 2004, 3, 458. [3] B. Wlng, C. Kloc, J. Teubner, E. Bucher, Phys. Rev. Lett. 2001, 86, 4350. [4] A. I. Boukai, Y. Bunimovich, J. T.-Kheli, J.-K. Yu, W. A. Goddard III, J. R. Heath, Nature 2008, 451, 168. [5] R. Venkatasubramanian, E. Siivola, T. Colpitts, B. OQuinn, Nature 2001, 413, 597. [6] M. S. Dresselhaus, G. Chen, M. Y. Tang, R. Yang, H. Lee, D. Wang, Z. Ren, J.-P. Fleurial, P. Gogna, Adv. Mater. 2007, 19, 1043.

Adv. Mater. 2011, XX, 14

2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

wileyonlinelibrary.com

www.advmat.de www.MaterialsViews.com

COMMUNICATION

[7] Y.-M. Lin, M. S. Dresselhaus, Phys. Rev. B 2003, 68, 075 304. [8] J.-S. Rhyee, E. Cho, K. H. Lee, S. I. Kim, E. S. Lee, S. M. Lee, Y. S. Kwon, J. Appl. Phys. 2009, 105, 053 712. [9] J.-S. Rhyee, K. H. Lee, S. M. Lee, E. Cho, S. I. Kim, E. Lee, Y. S. Kwon, J. H. Shim, G. Kotliar, Nature 2009, 459, 965. [10] G. Grner, in Density Waves in Solids, (Ed: G. Grner) Addison-Wesley publishing company, New York, USA 1994, Ch. 1.

[11] J. W. Sharp, E. H. Volckmann, H. J. Goldsmid, Phys. Stat. Sol. 2001, 185, 257. [12] G. S. Nolas, H. J. Goldsmid, in Thermal Conductivity; Theory, Properties, and Applications, (Ed: T. M. Tritt) Springer, New York, USA 2004, Ch. 1.4. [13] J.-S. Rhyee, E. Cho, K. H. Lee, S. M. Lee, S. I. Kim, H.-S. Kim, Y. S. Kwon, S. J. Kim, Appl. Phys. Lett. 2009, 95, 212 109. [14] H. J. Goldsmid, Proc. Phys. Soc. B 1956, 69, 203.

wileyonlinelibrary.com

2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2011, XX, 14