High-throughput density functional perturbation theory and machine learning predictions of infrared, piezoelectric, and dielectric responses

Kamal Choudhary,Francesca Tavazza,Angela R Hight Walker,Kevin F Garrity,Vinit Sharma,Adam J Biacchi

doi:10.1038/s41524-020-0337-2

Abstract

Many technological applications depend on the response of materials to electric fields, but available databases of such responses are limited. Here, we explore the infrared, piezoelectric, and dielectric properties of inorganic materials by combining high-throughput density functional perturbation theory and machine learning approaches. We compute Γ-point phonons, infrared intensities, Born-effective charges, piezoelectric, and dielectric tensors for 5015 non-metallic materials in the JARVIS-DFT database. We find 3230 and 1943 materials with at least one far and mid-infrared mode, respectively. We identify 577 high-piezoelectric materials, using a threshold of 0.5 C/m2. Using a threshold of 20, we find 593 potential high-dielectric materials. Importantly, we analyze the chemistry, symmetry, dimensionality, and geometry of the materials to find features that help explain variations in our datasets. Finally, we develop high-accuracy regression models for the highest infrared frequency and maximum Born-effective charges, and classification models for maximum piezoelectric and average dielectric tensors to accelerate discovery.

Highlights

The Materials Genome Initiative (MGI)[1] has revolutionized the development of new technologically important materials, which historically has been a time-consuming task that was mainly dominated by trial and error strategies
Applications of these computational techniques in a highthroughput manner have led to several databases of computed geometries and many physicochemical properties, AFLOW4, Materials-project[3], Khazana[17], Open Quantum Materials Database (OQMD)[5], NOMAD7, Computational Materials Repository (CMR)[39], NIMS databases[40], and our NIST-JARVIS databases[6,8,21,22,23,41,42,43,44,45,46,47]
After the density functional perturbation theory (DFPT) calculations, we obtain the phonon-frequencies at Г-point as well

Summary

INTRODUCTION

The Materials Genome Initiative (MGI)[1] has revolutionized the development of new technologically important materials, which historically has been a time-consuming task that was mainly dominated by trial and error strategies. Subsequent targeted experimental synthesis and validation provide a means of rapid iteration to verify and improve computational models Applications of these computational techniques in a highthroughput manner have led to several databases of computed geometries and many physicochemical properties, AFLOW4, Materials-project[3], Khazana[17], Open Quantum Materials Database (OQMD)[5], NOMAD7, Computational Materials Repository (CMR)[39], NIMS databases[40], and our NIST-JARVIS databases[6,8,21,22,23,41,42,43,44,45,46,47]. Born effecƟve charges, Infrared intensity, Dielectric constant, Piezoelectric constant mance materials without performing additional first-principles Fig. 1 Flow-chart portraying different steps for the DFT and ML calculations. ML models can be considered as a screening-tool for the DFT calculations

AND DISCUSSION

METHODS

CODE AVAILABILITY