Modeling crystal defects using defect informed neural networks

Abstract Most AI-for-Materials research to date has focused on ideal crystals, whereas real-world materials inevitably contain defects that play a critical role in modern functional technologies. The defects break geometric symmetry and increase interaction complexity, posing particular challenges f...

Full description

Saved in:
Bibliographic Details
Main Authors: Ziduo Yang, Xiaoqing Liu, Xiuying Zhang, Pengru Huang, Kostya S. Novoselov, Lei Shen
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:npj Computational Materials
Online Access:https://doi.org/10.1038/s41524-025-01728-w
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849763955796344832
author Ziduo Yang
Xiaoqing Liu
Xiuying Zhang
Pengru Huang
Kostya S. Novoselov
Lei Shen
author_facet Ziduo Yang
Xiaoqing Liu
Xiuying Zhang
Pengru Huang
Kostya S. Novoselov
Lei Shen
author_sort Ziduo Yang
collection DOAJ
description Abstract Most AI-for-Materials research to date has focused on ideal crystals, whereas real-world materials inevitably contain defects that play a critical role in modern functional technologies. The defects break geometric symmetry and increase interaction complexity, posing particular challenges for traditional ML models. Here, we introduce Defect-Informed Equivariant Graph Neural Network (DefiNet), a model specifically designed to accurately capture defect-related interactions and geometric configurations in point-defect structures. DefiNet achieves near-DFT-level structural predictions in milliseconds using a single GPU. To validate its accuracy, we perform DFT relaxations using DefiNet-predicted structures as initial configurations and measure the residual ionic steps. For most defect structures, regardless of defect complexity or system size, only 3 ionic steps are required to reach the DFT-level ground state. Finally, comparisons with scanning transmission electron microscopy (STEM) images confirm DefiNet’s scalability and extrapolation beyond point defects, positioning it as a valuable tool for defect-focused materials research.
format Article
id doaj-art-6f17b29ad93449e7a2ae44ff034471d8
institution DOAJ
issn 2057-3960
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series npj Computational Materials
spelling doaj-art-6f17b29ad93449e7a2ae44ff034471d82025-08-20T03:05:15ZengNature Portfolionpj Computational Materials2057-39602025-07-0111111210.1038/s41524-025-01728-wModeling crystal defects using defect informed neural networksZiduo Yang0Xiaoqing Liu1Xiuying Zhang2Pengru Huang3Kostya S. Novoselov4Lei Shen5Department of Electronic Engineering, College of Information Science and Technology, Jinan UniversityDepartment of Mechanical Engineering, National University of SingaporeDepartment of Mechanical Engineering, National University of SingaporeInstitute for Functional Intelligent Materials, National University of SingaporeInstitute for Functional Intelligent Materials, National University of SingaporeDepartment of Mechanical Engineering, National University of SingaporeAbstract Most AI-for-Materials research to date has focused on ideal crystals, whereas real-world materials inevitably contain defects that play a critical role in modern functional technologies. The defects break geometric symmetry and increase interaction complexity, posing particular challenges for traditional ML models. Here, we introduce Defect-Informed Equivariant Graph Neural Network (DefiNet), a model specifically designed to accurately capture defect-related interactions and geometric configurations in point-defect structures. DefiNet achieves near-DFT-level structural predictions in milliseconds using a single GPU. To validate its accuracy, we perform DFT relaxations using DefiNet-predicted structures as initial configurations and measure the residual ionic steps. For most defect structures, regardless of defect complexity or system size, only 3 ionic steps are required to reach the DFT-level ground state. Finally, comparisons with scanning transmission electron microscopy (STEM) images confirm DefiNet’s scalability and extrapolation beyond point defects, positioning it as a valuable tool for defect-focused materials research.https://doi.org/10.1038/s41524-025-01728-w
spellingShingle Ziduo Yang
Xiaoqing Liu
Xiuying Zhang
Pengru Huang
Kostya S. Novoselov
Lei Shen
Modeling crystal defects using defect informed neural networks
npj Computational Materials
title Modeling crystal defects using defect informed neural networks
title_full Modeling crystal defects using defect informed neural networks
title_fullStr Modeling crystal defects using defect informed neural networks
title_full_unstemmed Modeling crystal defects using defect informed neural networks
title_short Modeling crystal defects using defect informed neural networks
title_sort modeling crystal defects using defect informed neural networks
url https://doi.org/10.1038/s41524-025-01728-w
work_keys_str_mv AT ziduoyang modelingcrystaldefectsusingdefectinformedneuralnetworks
AT xiaoqingliu modelingcrystaldefectsusingdefectinformedneuralnetworks
AT xiuyingzhang modelingcrystaldefectsusingdefectinformedneuralnetworks
AT pengruhuang modelingcrystaldefectsusingdefectinformedneuralnetworks
AT kostyasnovoselov modelingcrystaldefectsusingdefectinformedneuralnetworks
AT leishen modelingcrystaldefectsusingdefectinformedneuralnetworks