Introducing Materials Funnel 2.0 - hierarchical search in infinite chemical space

Inventing new ways of exploring the materials phase space accelerates functional materials discovery. For new breakthroughs materials, faster exploration of bigger phase spaces is a key goal. High throughput computational screening (HTCS) is widely used to quickly search for materials with the right functional property. In this article we redefine HTCS methods to combine deep learning and physics based model to explore much large chemical spaces than possible by pure physics driven HTCS. Deep generative models are used to autonomously create materials libraries with high likelihood of desired properties, inverting the design paradigm. Additionally machine learnt surrogates enable next layer of screening to prune the set further such as high quality quantum mechanical simulations can be performed. With organic photovoltaic (OPV) molecules as test bench, we show the power of this redesigned HTCS approach in inverse design OPV molecules with very limited computational expense.