Density of states prediction for materials discovery via contrastive learning from probabilistic embeddings.

Autor: Kong S; Department of Computer Science, Cornell University, Ithaca, NY, USA., Ricci F; Material Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA., Guevarra D; Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA., Neaton JB; Material Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA. jbneaton@lbl.gov.; Department of Physics, University of California, Berkeley, Berkeley, CA, USA. jbneaton@lbl.gov.; Kavli Energy NanoSciences Institute at Berkeley, Berkeley, CA, USA. jbneaton@lbl.gov., Gomes CP; Department of Computer Science, Cornell University, Ithaca, NY, USA. gomes@cs.cornell.edu., Gregoire JM; Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA. gregoire@caltech.edu.
Jazyk: angličtina
Zdroj: Nature communications [Nat Commun] 2022 Feb 17; Vol. 13 (1), pp. 949. Date of Electronic Publication: 2022 Feb 17.
DOI: 10.1038/s41467-022-28543-x
Abstrakt: Machine learning for materials discovery has largely focused on predicting an individual scalar rather than multiple related properties, where spectral properties are an important example. Fundamental spectral properties include the phonon density of states (phDOS) and the electronic density of states (eDOS), which individually or collectively are the origins of a breadth of materials observables and functions. Building upon the success of graph attention networks for encoding crystalline materials, we introduce a probabilistic embedding generator specifically tailored to the prediction of spectral properties. Coupled with supervised contrastive learning, our materials-to-spectrum (Mat2Spec) model outperforms state-of-the-art methods for predicting ab initio phDOS and eDOS for crystalline materials. We demonstrate Mat2Spec's ability to identify eDOS gaps below the Fermi energy, validating predictions with ab initio calculations and thereby discovering candidate thermoelectrics and transparent conductors. Mat2Spec is an exemplar framework for predicting spectral properties of materials via strategically incorporated machine learning techniques.
(© 2022. The Author(s).)
Databáze: MEDLINE
Externí odkaz: