Baobin Xie , Yang Chen , Bin Liu , Qihong Fang , Peter K Liaw , Jia Li
{"title":"Automatic intelligent multiscale simulation to predict mechanical properties and deformation mechanism of complex concentrated alloys","authors":"Baobin Xie , Yang Chen , Bin Liu , Qihong Fang , Peter K Liaw , Jia Li","doi":"10.1016/j.actamat.2025.121359","DOIUrl":null,"url":null,"abstract":"<div><div>Due to differences in atomic interactions, chemical heterogeneous structures are inevitably formed, especially in complex concentrated alloys, which strongly influences their mechanical behaviors and properties. However, existing continuum mechanics models fail to consider key mechanical information on characteristics of atoms themselves and their interactions, and simultaneously bridge discrete atomic scale with continuum scale remains a great challenge. Here, we propose a general automatic multiscale simulation framework that integrates the approach of machine learning, atomic simulation, dislocation dynamics, crystal plasticity finite element method, and constitutive model for bridging from atomic discrete mechanics to continuum mechanics. The proposed framework mainly focuses on dual physical scales: (i) discrete atomic configuration-dependent nanoscale atomic-level strain tensor, and (ii) continuum theory-based deformation behavior and macroscopic mechanical response. To quantitatively describe atomic-strain-field constrained dislocation behavior, automatic workflow is developed for mesoscale strain field. The atomic discrete scale and dislocation continuum scale are seamlessly connected through simulation framework, enabling overall computational scheme of simultaneously dealing with atomistic effects and mesoscale dislocation movement. The atomic information-guided, machine learning-enabled automatic multiscale simulation is applied to complex concentrated alloys, which achieves an automated and intelligent prediction from atomic structure characteristics to macroscopic mechanical properties and plastic deformation mechanisms. The findings indicate that increasing chemical heterogeneous significantly enhances strain amplitude while broadening the spacing between strain peaks/valleys, resulting in a competition between hardening driven by dislocation cross-slip and softening caused by dislocation motion. The proposed automatic multiscale simulation framework not only creates a bridge between discrete mechanics and continuum mechanics, but also would be extended to other metal systems.</div></div>","PeriodicalId":238,"journal":{"name":"Acta Materialia","volume":"297 ","pages":"Article 121359"},"PeriodicalIF":8.3000,"publicationDate":"2025-07-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Acta Materialia","FirstCategoryId":"88","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1359645425006457","RegionNum":1,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"MATERIALS SCIENCE, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Due to differences in atomic interactions, chemical heterogeneous structures are inevitably formed, especially in complex concentrated alloys, which strongly influences their mechanical behaviors and properties. However, existing continuum mechanics models fail to consider key mechanical information on characteristics of atoms themselves and their interactions, and simultaneously bridge discrete atomic scale with continuum scale remains a great challenge. Here, we propose a general automatic multiscale simulation framework that integrates the approach of machine learning, atomic simulation, dislocation dynamics, crystal plasticity finite element method, and constitutive model for bridging from atomic discrete mechanics to continuum mechanics. The proposed framework mainly focuses on dual physical scales: (i) discrete atomic configuration-dependent nanoscale atomic-level strain tensor, and (ii) continuum theory-based deformation behavior and macroscopic mechanical response. To quantitatively describe atomic-strain-field constrained dislocation behavior, automatic workflow is developed for mesoscale strain field. The atomic discrete scale and dislocation continuum scale are seamlessly connected through simulation framework, enabling overall computational scheme of simultaneously dealing with atomistic effects and mesoscale dislocation movement. The atomic information-guided, machine learning-enabled automatic multiscale simulation is applied to complex concentrated alloys, which achieves an automated and intelligent prediction from atomic structure characteristics to macroscopic mechanical properties and plastic deformation mechanisms. The findings indicate that increasing chemical heterogeneous significantly enhances strain amplitude while broadening the spacing between strain peaks/valleys, resulting in a competition between hardening driven by dislocation cross-slip and softening caused by dislocation motion. The proposed automatic multiscale simulation framework not only creates a bridge between discrete mechanics and continuum mechanics, but also would be extended to other metal systems.
期刊介绍:
Acta Materialia serves as a platform for publishing full-length, original papers and commissioned overviews that contribute to a profound understanding of the correlation between the processing, structure, and properties of inorganic materials. The journal seeks papers with high impact potential or those that significantly propel the field forward. The scope includes the atomic and molecular arrangements, chemical and electronic structures, and microstructure of materials, focusing on their mechanical or functional behavior across all length scales, including nanostructures.