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

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.Defects play an important role in our pursuit to engineer performance of a material. Vacancies are a type of defects which are ubiquitous and their presence can significantly alter 1

Defect graph neural networks for materials discovery in high-temperature clean-energy applications

“…This pat-383 tern of comparable accuracy but increased generalizability also 384 holds true for our method relative to other machine learning ef-385 forts for defect predictions, e.g., Frey et al's model for transition 386 metal dichalcogenides24 with MAE=0.67 eV and Cheng et al's 387 model for amorphous GeTe 70. Finally, in a concurrent preprint 388 with ours,71,72 Choudary et al used graph neural networks mod-389 els to predict total energy of a host structure and with an atom re-390 moved to estimate vacancy formation enthalpies, but this neglects 391 the relaxation of the host upon vacancy formation and yields an 392 MAE prediction of 1.5 eV for a single test set, including 2.3 eV for 393 oxides.…”

Materials discovery for high-temperature, clean-energy applications using graph neural network models of vacancy defects and free-energy calculations

“…This pat-383 tern of comparable accuracy but increased generalizability also 384 holds true for our method relative to other machine learning ef-385 forts for defect predictions, e.g., Frey et al's model for transition 386 metal dichalcogenides24 with MAE=0.67 eV and Cheng et al's 387 model for amorphous GeTe 70. Finally, in a concurrent preprint 388 with ours,71,72 Choudary et al used graph neural networks mod-389 els to predict total energy of a host structure and with an atom re-390 moved to estimate vacancy formation enthalpies, but this neglects 391 the relaxation of the host upon vacancy formation and yields an 392 MAE prediction of 1.5 eV for a single test set, including 2.3 eV for 393 oxides.…”

Unified graph neural network force-field for the periodic table: solid state applications

“…34,35 We developed an atomistic line graph neural network (ALIGNN) in our previous work 36 which can capture many body interactions in graph and successfully models more than 70 properties of materials, either scalar or vector quantities, such as formation energy, bandgap, elastic modulus, superconducting properties, adsorption isotherm, electron and density of states etc. [36][37][38][39][40][41] The same automatic differentiation capability that allows training these complex models allows for physically consistent prediction of quantities such as forces and energies; this enables GNNs to be used in quickly identifying relaxed or equilibrium states of complex systems. However, there is a need for a large and diverse amount of data to train unied force-elds.…”