Chinese Researchers Build AI Framework for Perovskite Interfacial Materials

Xinhua News Agency reported that a research team led by Cai Fengchun at the University of Science and Technology of China, collaborating with Southern University of Science and Technology and City University of Hong Kong, built an intelligent screening framework for self-assembled monolayers. The system combines artificial intelligence, high-throughput theoretical calculations, and expert knowledge to discover perovskite interfacial materials with synthetic feasibility and interface stability. To connect computational designs with experimental requirements, the researchers selected 24 functional fragments from 308 fragments extracted from carbazole derivatives in the PubChem database. They built 300 candidate molecules through modular assembly and filtered them down to 124 candidates by energy levels before scoring them across eight indicators, including synthetic feasibility, energy levels, hole reorganization energy, aromaticity, and heteroatom content. Using this framework, the team synthesized three high-ranking candidate molecules whose device performance aligned with the eight-descriptor scores. The top-ranked molecule achieved a photoelectric conversion efficiency of 27.2%, retaining about 95% of its initial efficiency after operating for 2000 hours at 65 degrees Celsius under continuous illumination and maximum power point tracking.

According to the research team, the modular screening framework can be expanded to other interfacial material systems, providing a new approach for discovering efficient and stable optoelectronic materials.