Accelerated Material Discovery Through Machine Learning for Direct Air Capture

Project Lead:

Overview:

Direct Air Capture (DAC) requires a liquid or solid material that selectively adsorbs carbon dioxide (CO2) from air. Discovering next generation DAC materials will help us reduce the size and energy requirements of the mega-scale operation. The problem is that discovering new materials is slow. Can we teach chemical intuition to an Artificial Intelligence (AI) in order to develop novel composite porous materials for direct CO2 capture from air, faster than humans?

As we know material and material composite discovery and formulation is a time-consuming process. If human input can be minimised in the design of an experiment and analysis of its subsequent results, it is likely that optimised research outcomes can be reached faster and more efficiently. AI has traditionally been used for single compound discoveries. What is absent is how porous materials and composites can be formed from the existing pool of knowledge and how we can teach it chemical intuition for developing new porous and composite material systems.

This project will use AI and computerized simulations to predict, model and develop new composite materials for CO2 capture from air. With the aid of AI and simulations, this project attempts to use Machine Learning (ML) in workflows to produce new porous materials that are specifically tailored for CO2 capture from direct air and turn that into a work plan for a specialised chemist to produce. The project may result in discoveries of applications aligned in the development of new materials for CO2 capture.