(Invited) Computation-Driven Materials Search for Thermoelectric Applications
Along this line, nanostructured bulk materials, which are synthesized by hot pressing nanopowder into a bulk material, have been widely studied as an effective approach to improve ZTs of existing or novel materials . By introducing nanostructured interfaces within a bulk material, the interdependent electron and phonon transport can be decoupled to dramatically reduce kL but still maintain S2σ, resulting in enhanced ZT. Unrestricted to conventional materials, this approach may also lead to high ZTs in unconventional TE materials with both a high S2σ and a high kL, such as bulk Si [2,3]. This will reach beyond conventional materials that heavily depend on toxic, rare, and expensive elements, e.g. Te. Based on the nanostructuring approach, broad material search is in urgent need for novel TE materials using nontoxic, cheap, and abundant elements.
For computation-driven TE material search, challenge generally exists in accurately predicting the energy-dependent electron and phonon mean free paths (MFPs) for TE property predictions. For phonon transport, there exist first-principles-based studies for standard materials such as Si and Ge. However, such calculations are intrinsically very time-consuming and are limited to relatively simple material structures. On the other hand, first-principles electrical property predictions are also restricted to well-studied materials and are thus unsuitable for high-throughput material search, in which thousands of complicated materials will be screened based on the computed ZTs. As the result, highthroughput TE material search is mostly based on S2σ/τ or ZTs further estimated with guessed kL/τ values [4-6], in which τ is the averaged relaxation time of all charge carriers. For materials screening, experimental inputs of σ and kLare often required and the guidance from computations becomes very limited.
For nanograined bulk materials synthesized by hot press, however, the need of computing electron and phonon MFPs can be eliminated when these MFPs are largely restricted by the grain size. In such materials, the optimum grain size should be larger than electron MFPs but shorter than phonon MFPs to reduce kL without deteriorating σ. The maximum ZT is anticipated when the grain size is reduced to the majority electron MFPs, called the small-grain-size (SGS) limit for nanograined bulk materials . This simple treatment has been used to compare the power factor of new TE materials  but no ZTs are computed without phonon calculations. In this work, we propose a ZT formulation of general materials under the SGS limit, which is solely based on phonon dispersions and electronic band structures predicted by first-principles calculations. The materials search is demonstrated with representative oxides that can be potentially used for high-temperature energy harvesting.
 A. J. Minnich, M. S. Dresselhaus, Z. F. Ren, and G. Chen, Energy & Environmental Science 2, 466 (2009).
 S. K. Bux, R. G. Blair, P. K. Gogna, H. Lee, G. Chen, M. S. Dresselhaus, R. B. Kaner, and J.-P. Fleurial, Advanced Functional Materials 19, 2445 (2009).
 Q. Hao, G. Zhu, G. Joshi, X. Wang, A. Minnich, Z. Ren, and G. Chen, Applied Physics Letters 97, 063109 (2010).
 I. Opahle, G. K. H. Madsen, and R. Drautz, Physical Chemistry Chemical Physics 14, 16197 (2012).
 O. Ingo, P. Alessandro, J. M. Eunan, D. Ralf, and H. M. Georg K New Journal of Physics 15, 105010 (2013).
 G. K. Madsen, Journal of the American Chemical Society 128, 12140 (2006).
 C. Bera, M. Soulier, C. Navone, G. Roux, J. Simon, S. Volz, and N. Mingo, Journal of Applied Physics 108, 124306 (2010).
 S. Wang, Z. Wang, W. Setyawan, N. Mingo, and S. Curtarolo, Physical Review X 1, 021012 (2011).