A Deep-learning Model for Fast Prediction of Vacancy Formation in Diverse Materials

TL;DR

This paper presents a graph neural network model trained on perfect materials that can rapidly predict vacancy formation energies in diverse materials, offering a faster alternative to traditional DFT calculations with reasonable accuracy.

Contribution

The study introduces a GNN-based approach capable of predicting vacancy formation energies without additional training on defect structures, enabling fast and accurate energy predictions across various materials.

Findings

GNN predictions are significantly faster than DFT calculations.

The model achieves reasonable accuracy in vacancy energy predictions.

Applied to over 55,000 materials, predicting nearly 200,000 vacancy energies.

Abstract

The presence of point defects such as vacancies plays an important role in material design. Here, we demonstrate that a graph neural network (GNN) model trained only on perfect materials can also be used to predict vacancy formation energies ($E_{vac}$) of defect structures without the need for additional training data. Such GNN-based predictions are considerably faster than density functional theory (DFT) calculations with reasonable accuracy and show the potential that GNNs are able to capture a functional form for energy predictions. To test this strategy, we developed a DFT dataset of 508 $E_{vac}$ consisting of 3D elemental solids, alloys, oxides, nitrides, and 2D monolayer materials. We analyzed and discussed the applicability of such direct and fast predictions. We applied the model to predict 192494 $E_{vac}$ for 55723 materials in the JARVIS-DFT database.