AI-Guided Design and Property Prediction for Zeolites and Nanoporous Materials

15 Overview of AI in the Understanding and Design of Nanoporous Materials

Seyed Mohamad Moosavi¹, Frits Daeyaert², Michael W. Deem³, and German Sastre⁴

¹ Department of Mathematics and Computer Science/Mathematics, Artificial Intelligence for the Sciences, Freie Universität Berlin, Berlin, Germany² Synopsisdenovodesign, Beerse, Belgium³ Certus LLC, Houston, Texas, USA⁴ Instituto de Tecnología Química UPV-CSIC, Universidad Politécnica de Valencia, Valencia, Spain

15.1 Introduction

What would have happened if Fritz Haber and Carl Bosch had approached the optimization of their catalyst [1] for ammonia synthesis using machine learning? Certainly history cannot be rewritten, but we thought this could be a case that illustrates one of the ways artificial intelligence (AI) could substitute systematic work using parametric methods. Industrial application is a realm in which AI should find almost unanimous appraisal, as “reaching a solution” is more important than “understanding the problem.” However, others may argue that AI can also guide us in learning new chemistry and rationalizing trends in vast amounts of data. This does in fact play a central role in our view and in the design of many future applications of AI [2].

Nowadays, materials research is routinely generating (big-)data, be it experimental or computational [3]. AI and machine learning algorithms can extract patterns, group similarities, and make predictions from these data. This is the so-called fourth paradigm of science ...