Jianghai Wang(王江海)

Jianghai Wang(王江海)

Graduate student in Computational Materials Science

CV

You can get the PDF directly.

Education

Projects

Crystal Structure Prediction of Ge-Sb-Te System Based on Genetic Algorithm

[ Aug 2023 - Aug 2024 ]

  1. Utilizing a combination of genetic algorithms and first-principles methods, I conducted crystal structure prediction studies on the Ge-Sb-Te system, a pivotal phase change material system, to uncover its inherent complexities and expand our knowledge of its configuration space. In this work, crossover and a variety of mutation operators are used to improve the sampling efficiency of the algorithm. The energy above hull is used as Fitness function for population iteration. After iterations, the candidates then further validated with a carefully considered screening criteria.

  2. Given the extensive crystal material database, we can search for new materials through element substitution based on structural prototypes. I adopted this strategy for the impressive GeTe-Sb2Te3 pseudo-binary line of the Ge-Sb-Te system, due to the large number of phase-change storage materials found along it. The overall process is similar to the above work, with a more stringent energy screening criterion due to the prevalence of metastable cubic phases in structures along pseudo-binary line.

The works above is expected to enhance comprehension of this intriguing system and potentially identify novel structures or compositions conducive to phase-change storage applications.

Machine learning potential accelerates crystal structure prediction

[ Jan 2024 – Sep 2024 ]

In this work, I have developed a framework combining genetic algorithm and pre-trained potential, which can effectively locate the ground-state or meta-stable states of the relatively large/complex systems. Utilizing machine learning potentials (e.g., general pre-training potentials CHGNet, MACE, MEG, etc.) as energy evaluators instead of first principles methods can significantly expedite the process of crystal structure prediction and more effectively probe the potential energy surface.

Skills