Capturing Anharmonicity in a Lattice Thermal Conductivity Model for High-Throughput Predictions

Abstract

High-throughput, low-cost, and accurate predictions of thermal properties of new materials would be beneficial in fields ranging from thermal barrier coatings and thermoelectrics to integrated circuits. To date, computational efforts for predicting lattice thermal conductivity (κ L ) have been hampered by the complexity associated with computing multiple phonon interactions. In this work, we develop and validate a semiempirical model for κ L by fitting density functional theory calculations to experimental data. Experimental values for κ L come from new measurements on SrIn 2 O 4 , Ba 2 SnO 4 , Cu 2 ZnSiTe 4 , MoTe 2 , Ba 3 In 2 O 6 , Cu 3 TaTe 4 , SnO, and InI as well as 55 compounds from across the published literature. To capture the anharmonicity in phonon interactions, we incorporate a structural parameter that allows the model to predict κ L within a factor of 1.5 of the experimental value across 4 orders of magnitude in κ L values and over a diverse chemical and structural phase space, with accuracy similar to or better than that of computationally more expensive models.

Machine learning models for the lattice thermal conductivity prediction of inorganic materials

“…Pretty high R 2 coefficient (≥ 0.93) on the test set in both these cases suggest a good κ L model has indeed been developed. We compared the performance of our ML model with other semi-empirical models by computing the average factor difference (AFD) [8], using the definition AFD = 10 a , where a = 1 N N i=1 log(κ L ) expt. − log(κ L ) model , with N being the number of data points.…”

Metal phosphides as potential thermoelectric materials

“…[76][77][78] Therefore, the thermal conductivity was estimated with approximate and semi-empirical models enhancing the prediction of novel high-performance thermoelectric materials. 5,6,36 For example, the thermal conductivity has been tested with models for the amorphous limit of the thermal conductivity. The most established model to determine the amorphous limit is the Cahill-Pohl model 5,6 where the speed of sound was calculated from the bulk and shear moduli of the materials (see ESI †).…”

“…In addition to the amorphous limit, the thermal conductivity of the MPs was computed with a semi-empirical model as recently reported by Miller et al 36 In this approach, the thermal conductivity equation contains information of the lattice stiffness and crystal structure computed with DFT and was fitted to experimental data at 300 K. The Grüneisen parameter used for the prediction of the thermal conductivity is solely dependent on the coordination number. 36 A similar trend was observed by Zeier et al 80 While the amorphous limit describes the lower bound of the thermal conductivity, the semi-empirical approach provides an average thermal conductivity more suited for crystals limited by acoustic phonons. For a better comparison of the MPs, the semi-empirical approach at 300 K was compared to the amorphous limit using the Cahill-Pohl model (computed thermal conductivity data in Table S6, ESI †).…”

“…4 Thermal conductivity computed using the Cahill-Pohl model (k min ) 5,6 and a semi-empirical approach (k SE ) at 300 K for various MP classes. 36 Computed thermal conductivities are given in Table S6 (ESI †). However, although the choice of thermal conductivity model has only minor influence on the qualitative ranking of predicted zT calc of XYP compounds, the thermoelectric performance of GaP and BP decreased by a factor of five and 10, respectively, from zT calc (k min ) to zT calc (k SE ) (Fig.…”

“…Umklapp scattering. For more crystalline samples, k min underestimates k p and thus, a semi-empirical approach was used as described by Miller et al 36 While the semi-empirical approach shows more an average thermal conductivity of the material, k p approaches k min at high temperatures or small grain sizes and should be considered as a lower bound. The computed electrical and thermal properties revealed that several MPs indicate high TE performance.…”