This page includes some useful code or data that we have used.
Jump to: molecular dynamics, Machine learning
To model multiprincipal elemental alloys using molecular dynamics, we need to know their interatomic interactions. In our previous work, as published in Acta Materialia, 219 (2021), pp. 117233 , we have parameterized the EAM potential for V based on the formalisms proposed by Zhou et al.[Phys. Rev. B. 69 144113 (2004)]. Combine with available parameters, we are able to simulate multicomponent alloys composed of different metals, including Fe, V, Cr, Mo, Ta, W. The complete potential file that we have used is provided here, or if it does not work, please go to our github site.
Ordered intermetallics represented by L12 Ni3Al is an important strengthening component in high-temepuare superalloys. When it is introduced into high entropy alloys, the site occupancy becomes disordered, leading to the formation of high entropy intermetallics. We have parameterized a set of MEAM potentials in order to model high entropy intermetallics with the composition of Ni-Co-Al-Ti. The details can be found in our publication, Journal of Materials Research and Technology, 30 (2024), pp. 9274–9284 . The potential file has been included in the NIST Potential Repository, which can be accessed from here.
We have developed Machine learning models based on ANN and SVM to predict the single-phase formation ability of different high-entropy carbide ceramics, as published in npj Computational Materials, 8 (2022), pp. 5 . The code is available on github from Ph.D. student Zhang Jun.
We have developed Machine learning models to predict the mechanical properties of high-entropy carbide ceramics with diverse compositions across the full phase space, as published in npj Computational Materials, 10 (2024), pp. 162 . The DFT data and code are available on github from Ph.D. student Zhang Jun.
We have developed Machine learning models to predict the mechanical properties of diverse high-entropy carbide ceramics with different compositions and carbon substoichiometry, as published in npj Computational Materials, 11 (2025), pp. 64 . The DFT data and code are available on github from Ph.D. student Lu Wenyu.
We have developed an Atomic Graph Attention Network (AGAT) Machine learning model to predict energy and force of different configurations with high accuracies, as published in Joule, 7 (2023), pp. 1832–1851 . The code is available on github from Ph.D. student Zhang Jun.