Grain boundary models

To be added.

---

Machine learning potentials

Machine Learning (ML) is well known for its successes in many fields, including in materials science. Among these applications, machine learning-based interatomic potentials have emerged as a new approach to achieve a balance between accuracy, generality, and efficiency in describing the energy and forces associated with different atomic structures in atomistic simulations. The machine-learning, Deep Potential (DP) method, is a relatively new and rapidly developing approach that has recently been applied to a wide range of materials systems. A particularly important application of atomistic simulation of materials is for defect structure and properties, since many important mechanical, electric, optical, etc. properties are determined by defects and their arrangements. We are collaborating with DP developers in Beijing and applying ML DPs for point defects, dislocation, and twinning properties of metals and alloys. Future opportunities lie in DPs for multi-component alloys and the expansion of DPs to a much broader range of materials.

References:
[1] Tongqi Wen, Linfeng Zhang, Han Wang, Weinan E, David J. Srolovitz, Deep potentials for materials science, Materials Futures 1, 022601 (2022).
[2] Tongqi Wen, Rui Wang, Lingyu Zhu, Linfeng Zhang, Han Wang, David J. Srolovitz, Zhaoxuan Wu, Specialising neural network potentials for accurate properties and application to the mechanical response of titanium, npj Computational Materials 7, 206 (2021).
[3] Rui Wang, Xiaoxiao Ma, Linfeng Zhang, Han Wang, David J. Srolovitz, Tongqi Wen, Zhaoxuan Wu, Classical and Machine Learning Interatomic Potentials for BCC Vanadium, arXiv:2209.12322 (2022).

---

Stress induced phase transformations and allotropy in ultra-strength materials

To be added.

---

Atomistic simulations of defect properties

To be added.

---

AI driven discovery of HEA

To be added.

---