AI helps design new materials that work in the real world

Researchers at the Massachusetts Institute of Technology (MIT) in the United States, in collaboration with Ohio State University, Michigan State University, and Oak Ridge National Laboratory, developed a new AI-based material design framework called “CrysVCD”. This research was published in *Nature Computational Science*. CrysVCD improves the stability rate of computationally generated materials by applying chemical valence electron rules during the initial stage of material generation, achieving nearly 70% stability. This method ensures that the designed materials comply with key chemical rules regarding the electrons surrounding atoms, enabling common models to more frequently adhere to valence layer constraints and achieve high lattice dynamics stability. CrysVCD supports the generation of materials with specific properties such as high thermal conductivity or high dielectric constants, making it suitable for use in chip and data center applications. Compared to traditional methods that require extensive post-generation screening, which accounts for approximately 90% of the total computational cost of available materials, CrysVCD significantly reduces resource consumption and helps small and medium-sized institutions break through innovation barriers.