en.Wedoany.com Report on Mar 26th, Researchers from BASF Germany, the Theory Department of the Fritz Haber Institute, and the BasCat – UniCat BASF Joint Lab have jointly developed a novel AI-driven method for Self-Driving Laboratories (SDLs). This method not only accelerates catalyst discovery but also explains the underlying chemical principles. It has been successfully validated in the industrial reaction of propane conversion to propylene.

Self-Driving Laboratories, which combine AI for experimental planning and automated operations, are typically known for their speed, enabling rapid testing and optimization of new materials. However, traditional AI methods are often seen as "black boxes," providing results without explanations, raising concerns about scientific progress and reliability. This research, published in ACS Catalysis, shows that by designing a "grey box" AI workflow, efficiency can be enhanced while maintaining explainability.

The new method achieves a balance between understanding and efficiency in catalyst discovery. It quickly identifies catalysts that outperform industrial standards and translates performance improvements into insights understandable by chemists, such as revealing synergistic interactions between promoters. The method efficiently searches through more than 10¹³ possible combinations, requiring fewer than 50 experiments.

The study demonstrates that the application of AI in chemistry can avoid sacrificing understanding, shifting material development from trial-and-error to genuine comprehension, and making AI a partner in scientific discovery. This explainable AI catalyst discovery method offers a new pathway for industrial catalysis research.