Screening new thermoelectric intermetallic compounds using high-throughput computation and machine learning
Céline Barreteau
ICMPE, CNRS-UMR 7182, 2-4 rue Henri Dunant, 94320, Thiais, France
In the current economic and ecological context, the development of alternative energy production is a major challenge. With this in mind, thermoelectric materials, which convert heat flow into a temperature gradient (and vice versa), offer numerous advantages: no moving parts, high reliability, …. However, they are still confined to niche applications due to their high production costs and low yields. New high-performance materials are therefore needed to increase the use of this technology.
To facilitate and accelerate the search for new candidates, a dual approach, combining first-principles calculations and experiments, is of interest. For a wide range of compositions, our method combines high-throughput calculations to identify stable, non-metallic compounds, with experimental studies of the most promising screened materials.
Initially, we focused on ternary T-M-X intermetallic compounds, with T a transition metal, rare earth or alkaline earth metal, M an element from the first line of transition metals and X, a sp element [1,2]. Thus, for dozens of prototypes, all possible T-M-X combinations were investigated by DFT calculations. Following this theoretical screening, experimental investigations were carried out to confirm the theoretical results, particularly with regard to stability and thermoelectric properties [3].
Now, in our quest for more promising new materials, we are continuing to improve our screening method to increase the complexity and type of compounds, while reducing the number and duration of calculations. To this end, Machine Learning techniques have been applied to certain intermetallic prototypes, such as Heuslers, to optimize our screening [4].
______________________
[1] Barreteau, C. et al., Optimization of criteria for an efficient screening of new thermoelectric compounds: the TiNiSi structure-type as a case-study, ACS Combinatorial Sciences, 22, 813-820, (2020), https://doi.org/10.1021/acscombsci.0c00133
[2] Barreteau, C. et al, Looking for new thermoelectric materials among TMX intermetallics using high-throughput calculations, Computational Material Science, 156, 96-103 (2019), http://doi.org/10.1016/j.commatsci.2018.09.030
[3] Moll, A. et al, SrCuP and SrCuSb Zintl phases as potential thermoelectric materials, J. All. Comp. 924, 169123 (2023) https://doi.org/10.1016/j.jallcom.2023.169123
[4] Xie, R. et al, Screening new quaternary semiconductor Heusler compounds by machine-learning methods, Chem. Mater. 35, 7615-7627 (2023) https://doi.org/10.1021/acs.chemmater.3c01323
