Machine learning methods move self-driving labs closer to materials discovery at scale
Machine learning doesn't replace human intelligence, but it can outlast human endurance, which makes it a helpful tool for chemistry and materials discovery. Scientists know machine learning models can make predictions based on the vast reams of data they are trained on, but can they take it a step further and massively scale up testing those predictions?
Machine learning is aiding materials discovery at scale, complementing rather than replacing human intelligence. Zhiling Zheng, a chemistry professor at Washington University in St. Louis, argues that AI can predict new chemical structures and scale up testing of those predictions. Christopher Cooper, from Washington University's McKelvey School of Engineering, agrees.
He presented a paper in Matter detailing how to curate data for polymer synthesis. The first step is data curation, converting known chemistry rules into a digestible format for machine learning models. The next step involves translating chemical synthesis instructions, found in journals and textbooks, into a form the AI can understand.
With these elements in place, AI can run simulations to predict successful molecules. Zheng's research focuses on metal-organic frameworks (MOFs), which have endless design possibilities but are too complex for human testing. The team trained an AI on 4,000 linker transformations of MOFs, filtering the data to find 10 new materials with superior water-harvesting capabilities.
Similarly, Cooper is working on collecting community knowledge for polymer design. He and his team created the Dynamic Polymer Annotated Library (DPAL), which accelerates model prediction and accuracy in predicting polymer designs.
Written by urgent.news from Phys.org's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
