Molecular dynamics simulation of the transformation of Fe-Co alloy by machine learning force field based on atomic cluster expansion

Aug 4, 2026

Upload by Himbo

The accuracy of molecular dynamics (MD) simulations depends strongly on the interatomic force field used to model atomic interactions. Compared with traditional or semi-empirical force fields, machine learning force fields offer higher precision and faster evaluation. By combining atomic cluster expansion (ACE) with first-principles density functional theory (DFT), a machine learning force field for binary Fe-Co alloy is constructed. MD simulations using this ACE force field reproduce the correct Fe-Co phase transition range.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages17

Reading Time43 min

AccessFree to download

Introduction

Methods

Molecular dynamics simulation of the transformation of Fe-Co alloy by machine learning force field based on atomic cluster expansion Yongle Li 1, *, Feng Xu 1, Long Hou 1, Luchao Sun2, Haijun Su3, Xi Li 1,4, Wei Ren 1, * 1 Physics Department, State Key Laboratory ofAdvanced Special Steels, Materials Genome Institute, International Centre for Quantum and Molecular Structures, Shanghai University, Shanghai 200444, China 2 Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China 3 State Key Laboratory of Solidification Processing, Northwestern Polytechnical University, Xi'an 710072, China 4 Shanghai Key Lab ofAdvanced High-temperature Materials and Precision Forming, Shanghai Jiao Tong University, Shanghai 200240, PR China *[Emails: yongleli@shu.edu.cn](Emails: yongleli@shu.edu.cn); [renwei@shu.edu.cn](renwei@shu.edu.cn) Abstract The force field describing the calculated interaction between atoms or molecules is the key to the accuracy of many molecular dynamics (MD) simulation results. Compared with traditional or semi-empirical force fields, machine learning force fields have the advantages of faster speed and higher precision. We have employed the method of atomic cluster expansion (ACE) combined with first-principles density functional theory (DFT) calculations for machine learning, and successfully obtained the force field of the binary Fe-Co alloy. Molecular dynamics simulations ofFe-Co alloy carried out using this ACE force field predicted the correct phase transition range of Fe-Co alloy. Key words: Molecular dynamics, Atomic cluster expansion, Fe-Co Alloy, Density functional theory, Phase transition, Force field Introduction Alloy is usually a substance with metallic properties synthesized by two or more metal elements, or metal and non-metal elements through a specific method. According to the types of elements contained in the alloy, it can be divided into binary alloys, ternary alloys or multi-element alloys[1, 2] . The research on the phase transition of alloy materials has always been the focus of many scientific fields. Although numerous simulations of melting and solidification of alloys have been reported, progress in this area has been rather slow[3, 4] . Molecular dynamics (MD) is an effective means of simulating the phase transition of alloy materials, and molecular dynamics is widely applied in various fields such as physics, chemistry, biology and materials science[5-8] . At present, there are many methods for simulating the phase transformation of alloys by using molecular dynamics[9] . But the correctness of the molecular dynamics simulation results is largely limited by the precision of the chosen force field, which should benefit from the rapid development of empirical and semi-empirical many-body potentials describing metallic systems. At present, the force fields that can be selected in the simulation of alloy materials include the Lennard-Jones (LJ) potential[10], embedded atom method (EAM) potential[11] and modified embedded atom method (MEAM) potential[12, 13], etc. Given the diversity and complexity of elements contained in alloys, the use of machine learning to fit force field parameters is faster, more promising, and hopefully more widely applicable than traditional force field development. Therefore, this paper uses the method of atomic cluster expansion (ACE)[14, 15] combined with first-principles density functional theory (DFT) calculations for machine learning to fit a force field that can be used for binary Fe-Co alloys. The melting and solidification mechanism of the binary Fe-Co alloy is revealed through the ACE force field, which expands the application of the ACE force field. Methods Atomic cluster expansion is a complete descriptor that can describe the local atomic environment of multicomponent materials. Some expressions for multivariate systems and non-orthogonal basis functions

FAQs

Why is the force field crucial for molecular dynamics simulations of alloy phase transitions?

The correctness of MD results is largely limited by the precision of the chosen force field, since it governs how atomic interactions are represented.

How is the machine learning force field for Fe-Co alloy built in this work?

Atomic cluster expansion (ACE) is combined with first-principles DFT calculations to train and obtain the force field for the binary Fe-Co alloy.

What does the ACE-based force field enable in the MD simulations?

Using the ACE force field, MD simulations can predict the correct phase transition range of Fe-Co alloy.

Aug 4, 2026

Upload by Himbo