Density of States Prediction for Materials Discovery via Contrastive Learning from Probabilistic Embeddings

TL;DR

This paper introduces Mat2Spec, a contrastive learning framework with probabilistic embeddings that significantly improves the prediction of spectral properties like phonon and electronic density of states in crystalline materials, aiding materials discovery.

Contribution

It presents a novel probabilistic embedding generator combined with supervised contrastive learning for spectral property prediction, outperforming existing methods.

Findings

Mat2Spec achieves state-of-the-art accuracy in predicting phDOS and eDOS.

It successfully identifies eDOS gaps below the Fermi energy.

The model aids in discovering candidate thermoelectrics and transparent conductors.

Abstract

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.