Machine Learning-Enhanced Design of Homogeneous Bifunctional Catalysts for CO2 Hydrogenation

ongoing

As a global society, we face the urgent challenge of reducing CO2 emissions. In response, governmental organisations, such as the European Union, have introduced policies promoting sustainable practices and renewable energy sources. One initiative is the conversion of CO2 into valuable products, with CO2-based methanol synthesis emerging as a promising approach. Methanol serves both as a low-density fuel and a feedstock for essential chemicals. While heterogeneous catalysts are commonly used in this reaction, they necessitate harsh conditions and exhibit low selectivity. Homogeneous catalysts, in contrast, operate at milder conditions and allow for fine-tuned active sites, potentially enhancing performance. Nevertheless, conventional methods for discovering new efficient catalysts are time-consuming and costly. BIFUCCO2 aims to overcome these limitations by leveraging computational techniques to pinpoint the most promising homogeneous bifunctional catalysts for this reaction from over a million in silico designed candidates. By implementing a machine learning (ML) workflow, the identification of the most efficient catalysts for this process will be achieved, enabling our experimental collaborators to validate the findings. The project merges the applicant’s knowledge in data-driven techniques, the proficiency of Res. Prof. Nova’s group in catalytic mechanisms (University of Oslo), the experience of Res. Prof. Balcells in ML applications (University of Oslo), and the expertise of Prof. Dr. Reiher (ETH Zürich, secondment) in chemical reaction networks, alongside contributions from experimental collaborators (Prof. Beller, LIKAT). BIFUCCO2 provides a framework for enhancing my existing research skills and acquiring novel insights from domain experts across various disciplines. The training activities during the grant period will significantly advance my professional career, consolidating my ability to lead a research group in the field of computational chemistry.