AIML-QD Nanoreactors: Active Learning Force Fields for Quantum Dots and Perovskites

AI Technology: Deep Learning; Machine Learning

Metal halide perovskite and semiconductor quantum dots (QDs) are central to next-generation LEDs and photovoltaics, yet their stability is governed by surface processes that remain inaccessible to standard simulation tools. 

Building on a successfully funded EuroHPC project, AIML-QD Nanoreactors will use the AI Factory to complete a DFT-quality molecular dynamics library for II–VI, III–V, IV–VI and perovskite QDs, and to train robust machine learning force fields (MLFFs). 

The study will couple committee-based active learning with synthetic nanoreactor simulations that drive QD surfaces into reactive, high-uncertainty regimes and trigger on-the-fly DFT relabelling. 

This workflow will deliver production-ready MLFFs, a GPU-optimized active learning protocol, and an open QDSpace data and model hub that can be directly reused by the broader AI and materials communities. 

The outcome is a predictive, shareable AI infrastructure for nanosecond simulations of QD stability and synthesis under experimentally relevant conditions.