AI-Driven Strategy Accelerates Catalyst Discovery and Explains Performance

A study published in ACS Catalysis outlines an advanced AI-driven strategy designed to accelerate catalyst discovery while simultaneously explaining why identified materials perform better.

Self-Driving Laboratories (SDLs), integrating AI with lab automation and robotics, are frequently recognized for their speed in testing and optimizing new materials. However, traditional AI approaches in materials science often function as "black boxes," delivering optimized results without providing explanations. This limitation has led to questions regarding the extent of true scientific progress and reliability control when fundamental understanding is limited.

Introducing the "Gray-Box" AI Approach

The new research, conducted by the Institute's Theory Department in collaboration with BASF and BasCat – UniCat BASF JointLab, introduces a "gray-box" AI approach. This strategy aims to improve performance and provide meaningful insights by carefully designing how AI explores potential material combinations.

Successful Demonstration: Propane to Propylene Conversion

This approach was successfully demonstrated on the industrial reaction of converting propane into propylene, a key building block for the chemical industry. The method identified a catalyst superior to the current industry reference and translated the improved performance into chemical understanding. It highlighted the effects of individual promoters and synergistic interactions that had not been identified in prior traditional studies.

Efficiency and Enhanced Understanding

The method remained highly efficient, requiring less than 50 experiments to search a design space containing over 10^13 possible promoter combinations.

The study suggests that AI and automation in chemistry can provide both efficiency and enhanced understanding, positioning AI as a more integrated partner in scientific discovery.