CompuSem-AI: In Silico Design and Virtual Screening of Molecular Semiconductors

About the Project

This project develops a platform to discover and optimize molecular semiconductors for next-generation organic electronics. The core idea is to combine AI-driven molecular generation, physics-informed machine learning, and high-throughput quantum chemistry into a closed-loop in silico workflow that rapidly identifies promising semiconductor candidates without laboratory synthesis or device fabrication.

The platform will integrate curated molecular datasets, descriptor pipelines, and multi-objective optimization models to predict key semiconductor-relevant properties, including HOMO/LUMO energies, band-gap proxies, reorganization energy, charge-transport surrogates, and stability/processability proxies. Candidate molecules will be generated and ranked using uncertainty-aware models and Pareto optimization to balance performance, novelty, and synthetic plausibility.

A major outcome is a reproducible computational framework that shortens discovery cycles, improves hit rates compared with brute-force screening, and delivers prioritized candidate lists for future external experimental validation. The project also produces reusable data assets, benchmark protocols, and model documentation to support long-term digital R&D in molecular electronics.