NOMAD CoE
A tailored AI Approach for Heterogeneous Catalysis

Work by NOMAD CoE researchers published in MRS Bulletin

NOMAD CoE researchers Lucas Foppa, Luca M. Ghiringhelli, Matthias Scheffler and other colleagues have developed a customized artificial intelligence approach for modeling of heterogeneous catalysis. This method takes into account the key physicochemical parameters that are correlated with catalytic performance to accelerate the discovery of improved or novel materials. The study was published in the prestigious high-ranking journal MRS Bulletin in November 2021.

The search for efficient heterogeneous catalysts is challenged by the intricate interplay of many, and not exactly known, underlying processes that govern the catalyst performance. Because of this intricacy, the explicit atomistic modelling of the full catalytic progression is unfeasible, if not inappropriate. Instead, a new study suggests the use of detailed experimental data and artificial intelligence for identifying exploitable correlations describing catalyst reactivity.

A team of researchers from the Novel Materials Discovery (NOMAD) lab and the Department of Inorganic Chemistry at the Fritz Haber Institute, together with researchers from the BasCat UniCat BASF JointLab at the TU Berlin, has proposed a novel AI approach that identifies so-called "material genes" of heterogeneous catalysis. These are parameters that, by analogy with biological genes, reflect the processes responsible for the materials performance. In the search for high-performance catalysts, these parameters are the crucial ones to be used for catalyst design.

In this AI study, nine vanadium-based oxidation catalysts were synthesized, fully characterized and systematically tested in the propane oxidation reaction. The consistent "clean" data thus generated were subjected to symbolic regression in order to identify nonlinear relationships between the key physicochemical parameters ( "material genes") and the catalyst reactivity, for instance its selectivity. The proposed AI approach is interpretable and can be applied even to the rather small number of materials that can be addressed by experiment. 
The results of the study give hope that the discovery of improved or novel materials can be accelerated by AI. At the same time, the physical understanding of catalytic materials will be improved.

As an MRS Bulletin Impact paper, the publication by Foppa et al. ranks among the most important original research papers of exceptional interest to the materials research community. 

The original publication you can find here:
Lucas Foppa, Luca M. Ghiringhelli, Frank Girgsdies, Maike Hashagen, Pierre Kube, Michael Hävecker, Spencer J. Carey, Andrey Tarasov, Peter Kraus, Frank Rosowski, Robert Schlögl, Annette Trunschke, and Matthias Scheffler (2021)
Materials genes of heterogeneous catalysis from clean experiments and artificial intelligence.
MRS Bulletin (2021). https://doi.org/10.1557/s43577-021-00165-6