Thermal Conductivity in Thermoelectric Materials

Michael W. Gaultois Department of Chemistry, University of Alberta, Edmonton, Alberta T6G 2G2, Canada [email protected] 20 November 2008 Abstract The recent cultural drive in Westernized countries toward sustainable societies has renewed interest in thermoelectric (TE) materials, which promise the generation of electricity from waste heat. A significant hurdle to TE materials is their low efficiency, which has hindered their potential applications and has limited their use to specialized markets. Higher efficiency materials are difficult to design and optimize, as many properties are intimately linked. Theoretical studies of low dimensional materials suggest that quantum confinement may separate some of these linked properties. This would allow a great leap in the design of efficient TE materials. Recent discoveries have unveiled promising nanostructured TE materials, whose thermal conductivities are greatly reduced due to increased boundary scattering of phonons.

History Thermoelectric materials generate a current from a temperature gradient; conversely, they can generate a temperature gradient when a current is applied. This property bestows the ability to generate power from heat, and the ability to cool (or heat) an area with an applied potential and current. Though these effects were discovered independently in the early 19th century (the Seebeck effect in 1823, and the Peltier effect in 1834, respectively), their complementarity was not realized until 1851 by Lord Kelvin.1 These discoveries had minimal impact in development of functional devices. It was not until the mid 20th century that research into semiconductor transistors led to the fortuitous discovery of materials with better thermoelectric properties. This began with research by Ioffe and co-workers,2,3 which was followed by an explosion of investigations due to potential military applications that were envisioned.1

Applications Thermoelectric generators or coolers have no moving parts and their operation depends only on a fundamental and intrinsic property. These characteristics allow TE devices to be simple, rugged, and scalable in size; they are not prone to mechanical failure and are able to operate in any condition under which the material is stable. This has led to their use in niche environments: power generators in space probes (such as the Voyager I probe, where, after 17 years, the TE power generator is still functional), solid-state coolers for optics and electronics, refrigerators for picnic coolers, and seat-heaters in cars.1,4 A typical device is shown in Figure 1,5 where many junctions (hundreds or thousands) are connected electrically in series and thermally in parallel.

Figure 1.5 A typical TE device. Many junctions (hundreds or thousands) are connected electrically in series and thermally in parallel.

The Seebeck effect arises due to a temperature gradient along a material. As shown in Figure 2,5 if the ends of a TE element are held at different temperatures, a potential is created, whose magnitude is related to the difference in temperature at either end. The potential drives the flow of an electric current. An electron/hole pair is created at the hot end; the pair flows away from the hot junction toward the cool end of the TE element, where the pair recombines. Electrons flow away from the junction through n-type materials, and holes flow away through p-type materials. The current generated can be used to charge a battery or power a device. (Figure 2, upper) The Peltier effect occurs when a potential is applied across the device. The potential drives the motion of electron/hole pairs. The formation of an electron/hole pair absorbs

energy, cooling one end of the device, while electron/hole pair recombination at the far end releases energy, heating the other end of the device. In this case, a temperature gradient is generated by an applied potential, which can actively cool or heat one side of the device. (Figure 2, lower)

Figure 2.5 Thermoelectric power generation (upper) is possible with a temperature gradient across a TE device. If the ends of a TE element are held at different temperatures, a potential is created, whose magnitude is related to the difference in temperature at either end. The potential drives the flow of an electric current. TE cooling is the reverse effect, where a temperature gradient is created by an applied potential.

Unfortunately, the efficiencies of current commercially available thermoelectrics are quite low. This has limited their applications to specialized markets.4 The original materials that were discovered in the 1950's are still used today, most notably bismuth telluride alloys.6 Although recent discoveries of more efficient thermoelectric materials have been made,7,8 industrial-processing requirements for large-scale device fabrication has prohibited their use to date.

In search of better materials The overall efficiency of the thermoelectric process is represented by ZT,9 a unitless thermoelectric figure of merit, shown in equation (1); S is the Seebeck coefficient, is the electrical conductivity, ph is the phonon thermal conductivity, e is the electronic thermal conductivity, and T is the temperature (in Kelvin).
S 2T ZT = ph + e

(1)

To maximize the efficiency of thermoelectric materials, ZT must be maximized. High electrical conductivity is necessary to allow unimpeded flow of electrons through the material and minimizing loss due to heat generation during electron flow. Thermal conductivity should be minimized to prevent backflow of heat. The Seebeck coefficient10 must be maximized, but this comes at the cost of decreased electrical conductivity. Optimization is challenging because properties are interrelated and affected by the electronic properties of the material, although recent theoretical developments hint at the possibility of decoupling some of these properties.9,11-13 To get an idea of the how efficiency corresponds to ZT, for solid-state Peltier coolers to compete with compressor-based refrigeration units, materials with ZT > 3 will need to be

developed.7 Bell has listed a wide number of applications of TE materials and the ZT required for a viable product.14

Thermal conductivity: phonons and electrons Thermoelectric materials were thought to be limited to ZT < 1 for a long time, until the discovery of thin film thermoelectrics with ZT = 2.4 by Venkatasubramanian in 2001.15 As previously mentioned, an ideal TE material will have a high electrical conductivity, allowing an unimpeded flow of electrons and thus minimal loss of energy due to resistance, and a low thermal conductivity, minimizing the backflow of heat. Recent discoveries of new high ZT materials (ZT > 1) have been made by efforts to decrease thermal conductivity.16 Thermal conductivity is the sum of the lattice thermal conductivity and electronic thermal conductivity, as shown in Equation 2. These conductivities represent the contributions from the flow of phonons and electron/hole pairs, respectively.

= ph + e

(2)

Acoustic phonons traveling through the lattice are the primary carrier for thermal energy; scattering phonons leads to a reduction in the overall thermal conductivity, as shown in Equation 2. There are many processes that can scatter electrons/holes and phonons, but in an ideal case for TE materials (high ZT), phonons are maximally scattered, lowering the thermal conductivity, while electrons are minimally scattered, maintaining a high electrical conductivity. Slack described this ideal material as a phonon-glass electron single-crystal (PGEC).17 A PEGC is a material with a minimum phonon mean-free path length, as seen in amorphous glasses,18 and a maximum electron

mean-free path length. The former implies a maximum scattering of phonons, while the latter implies a minimum scattering of electrons. In bulk materials (at ambient temperatures), phonon-phonon scattering is the dominant scattering mechanism. This limits the mean-free path length to a value well below the sample dimensions, so there is no change in thermal conductivity expected, nor seen, with change in size.19 However, as the size of the material decreases below the threshold for phonon-phonon scattering, other scattering processes, such as boundary scattering, become important. This phenomenon has led to the development of bulk materials with nanostructured domains,20 such as clathrate,21,22 skutterudite,23 or chalcogenide24,25 systems. The nanostructured domains in special instances of these systems cause selective scattering of phonons at grain boundaries, or at "rattlers" in clathrates. Alternate investigations into thin-film TE materials have yielded similar results,15,26 where phonons are scattered by layer boundaries.

Enter nanomaterials As the systems become increasingly small, new phenomena appear that may prove to be valuable in future development of TE materials.12 Theoretical studies have predicted high figures of merit for confined-carrier systems such as quantum wells (2-D confinement),9,11 quantum wires (1-D confinement),13,27 and quantum dots (3-D confinement).28 Empirical data has also shown promise in materials of lower dimensions. Recent investigations by two groups revealed that in addition to size-dependence of thermal conductivity, roughness on silicon nanowires leads to lower thermal conductivity; the nanowire ZT is two orders of magnitude greater than bulk silicon.29,30

The authors note that roughness increases phonon scattering far more than predicted by current theory. This unexpected gain in efficiency will, no doubt, lead to interesting findings throughout the next few years as theory evolves to deal with this discovery, and it will not be long before related studies on other materials take place to further explore this phenomenon.

Conclusion Thermoelectric materials have been known to the scientific community for quite some time, but their low efficiencies have prohibited widespread use of TE devices. Until recently, the artificial ceiling of efficiency of ZT = 1 was unbreakable; nanostructured materials have shown great promise, with high ZT due to reduced thermal conductivity and unchanged electrical conductivity. Materials that exhibit quantum confinement are especially interesting, as conflicting properties, which are linked in the bulk phase, are believed to become more easily an independently tunable. One area of current research involves the development of materials that scatter phonons effectively while allowing electrons to flow freely. An ideal material, a phonon-glass electron single-crystal, has the thermal conductivity of a glass and the electrical conductivity of a single crystal.

Acknowledgements I thank the developers of EndNote for producing an exceptional software tool. This has prevented many trials and tribulations that would otherwise have plagued the creation of this review, and has saved me many hours of tedium.

References
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) Rowe, D. M. CRC handbook of thermoelectrics; CRC Press: Boca Raton, FL, 1995. Ioffe, A. F. Zhurnal Sakharnoi Promyshlennosti 1953, 23, 1452-1459. Ioffe, A. F.; Airapetyants, S. V.; Ioffe, A. V.; Kolomoets, N. V.; Stil'bans, L. S. Doklady Akad. Nauk S.S.S.R. 1956, 106. DiSalvo, F. J. Science 1999, 285, 703-706. Rowe, D. M. Thermoelectrics handbook : macro to nano; CRC/Taylor & Francis: Boca Raton, 2006. Wright, D. A. Nature 1958, 181, 834-834. Majumdar, A. Science 2004, 303, 777-778. Vining, C. B. Nature 2001, 413, 577-578. Balandin, A.; Wang, K. L. Journal of Applied Physics 1998, 84, 6149-6153. Tritt, T. M.; Subramanian, M. A. Mrs Bull 2006, 31, 188-194. Balandin, A.; Wang, K. L. Physical Review B 1998, 58, 1544-1549. Dresselhaus, M. S.; Chen, G.; Tang, M. Y.; Yang, R. G.; Lee, H.; Wang, D. Z.; Ren, Z. F.; Fleurial, J. P.; Gogna, P. Advanced Materials 2007, 19, 1043-1053. Lin, Y. M.; Sun, X. Z.; Dresselhaus, M. S. Physical Review B 2000, 62, 4610-4623. Bell, L. E. Science 2008, 321, 1457-1461. Venkatasubramanian, R.; Siivola, E.; Colpitts, T.; O'Quinn, B. Nature 2001, 413, 597602. Snyder, G. J.; Toberer, E. S. Nature Materials 2008, 7, 105-114. Slack, G. A. Journal of Physics and Chemistry of Solids 1973, 34, 321-335. Kittel, C. Physical Review 1949, 75, 972-974. Kittel, C. Introduction to solid state physics; 7th ed.; Wiley: New York, 1996. Hogan, T. P.; Downey, A.; Short, J.; D'Angelo, J.; Wu, C. I.; Quarez, E.; Androulakis, J.; Poudeu, P. F. P.; Sootsman, J. R.; Chung, D. Y.; Kanatzidis, M. G.; Mahanti, S. D.; Timm, E. J.; Schock, H.; Ren, F.; Johnson, J.; Case, E. D. J Electron Mater 2007, 36, 704-710. Blake, N. P.; Mollnitz, L.; Kresse, G.; Metiu, H. Journal of Chemical Physics 1999, 111, 3133-3144. Iversen, B. B.; Palmqvist, A. E. C.; Cox, D. E.; Nolas, G. S.; Stucky, G. D.; Blake, N. P.; Metiu, H. Journal of Solid State Chemistry 2000, 149, 455-458. Sales, B. C.; Mandrus, D.; Williams, R. K. Science 1996, 272, 1325-1328. Chung, D. Y.; Hogan, T. P.; Rocci-Lane, M.; Brazis, P.; Ireland, J. R.; Kannewurf, C. R.; Bastea, M.; Uher, C.; Kanatzidis, M. G. Journal of the American Chemical Society 2004, 126, 6414-6428. Sootsman, J. R.; Pcionek, R. J.; Kong, H. J.; Uher, C.; Kanatzidis, M. G. Chemistry of Materials 2006, 18, 4993-4995. Venkatasubramanian, R. Abstracts of Papers of the American Chemical Society 2003, 225, U987-U987. Rao, A. M.; Ji, X. H.; Tritt, T. M. Mrs Bull 2006, 31, 218-223. Harman, T. C.; Taylor, P. J.; Spears, D. L.; Walsh, M. P. J Electron Mater 2000, 29, L1L4. Boukai, A. I.; Bunimovich, Y.; Tahir-Kheli, J.; Yu, J. K.; Goddard, W. A.; Heath, J. R. Nature 2008, 451, 168-171. Hochbaum, A. I.; Chen, R. K.; Delgado, R. D.; Liang, W. J.; Garnett, E. C.; Najarian, M.; Majumdar, A.; Yang, P. D. Nature 2008, 451, 163-U5.

(21) (22) (23) (24) (25) (26) (27) (28) (29) (30)