Turkish Journal of Physics
Abstract
Machine learning interatomic potentials (MLIPs) provide an efficient approach to large-scale molecular dynamics simulations of high-entropy alloys. In this work, an MLIP trained on AIMD data was developed for the CrMnFeCoNi Cantor alloy and validated against DFT calculations. In addition, we compared MLIP results with those obtained using the classical modified embedded atom method (MEAM)-based interatomic potential for the CrMnFeCoNi Cantor alloy. The model achieved energy and force RMSEs of 7.39 x 10-6 meV/atom and 4.46 x 10-3 eV/A, respectively. Tensile simulations demonstrated that the MLIP accurately reproduced the stress–strain response, elastic constants, and deformation mechanisms observed in DFT calculations. The predicted maximum strain reached 0.104, compared with 0.098 obtained using the MEAM potential. This highlights the potential of MLIP to extend DFT accuracy to realistic large-scale simulations of complex HEA systems, enabling the efficient exploration of their mechanical performance and practical engineering applications.
Author ORCID Identifier
NGUYEN PHUONG: 0009-0008-9134-930X
LE NGHIA: 0009-0003-4505-1020
HUYNH VAN: 0009-0004-0232-5420
VO DIEN: 0000-0002-7974-9852
NGUYEN TIEN: 0000-0002-6678-3787
DOI
10.55730/1300-0101.2944
Keywords
DeePMD-kit, high-entropy alloys, AIMD, elastic constants
First Page
341
Last Page
352
Publisher
The Scientific and Technological Research Council of Türkiye (TÜBİTAK)
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
PHUONG, N. L, NGHIA, L. H, VAN, H. K, DIEN, V. K, & TIEN, N. T (2026). Stress–strain behavior of the high-entropy Cantor alloy CrMnFeCoNi using a deep neural network potential. Turkish Journal of Physics 50 (5): 341-352. https://doi.org/10.55730/1300-0101.2944
