Development and application of improved ligand descriptors and representations for inverse catalyst design

• Applicant: Professorin Dr. Viktoria H. Däschlein-Geßner, since 6/2023
• Subject Area: Organic Molecular Chemistry - Synthesis and Characterisation ; Inorganic Molecular Chemistry - Synthesis and Characterisation
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 497260357

The development of new transition metal catalysts and their corresponding ligands is of crucial importance for synthetic organic chemistry and a decisive factor for the development of new reactions and the improvement of existing synthesis protocols, for example to achieve higher selectivities or conversions. The development of tailor-made ligands for the specific requirements of different reactions is an extremely time and cost-intensive process. In order to optimize and accelerate the design of new, efficient ligands, ligand properties can be specifically adapted by elucidating structure-activity relationships. However, the simple use of individual parameters (descriptors) to quantify steric or electronic ligand properties was found to be too one-dimensional to reliably and quantitatively describe the complex structure-activity relationships. With the help of multivariate regression analyses and the application of machine learning methods, first advances in the prediction of ideal ligands have recently been made.The aim of this research project is the realization of an inverse catalyst design with the help of machine learning methods. This will initially be implemented by designing new phosphine ligands for palladium- and nickel-catalyzed cross-coupling reactions. To achieve this goal, new descriptors for an improved description of the ligands and their properties will at first be developed. On the one hand, these should better describe the conformational flexibility of ligands and, on the other hand, quantify additional secondary metal-ligand interactions. Both properties have repeatedly been proven to be decisive factors for the activity of different catalyst systems, but are not or only insufficiently represented by previous descriptors. The new descriptors will be easily accessible by means of quantum chemical calculations and thus enable broad screening of different structures. Their validity will be verified on the basis of experimental studies in order to subsequently enable a reliable prediction of the ligand properties that are decisive for the respective reaction. For this purpose, the descriptors for individual substituents will be generated in order to enable predictions of new ligand structures beyond the test and training set. The applicability of the developed descriptors for the prediction of new ligand structures will be demonstrated in the last step of this project on the basis of selected test reactions, such as the selective monoarylation of ammonia or the chemoselective coupling of chloroaryl triflates. Based on experimental data, machine learning methods will be used to determine ideal ligand substituents and hence to predict new catalysts.

• DFG Programme: Priority Programmes
• Subproject of: SPP 2363: Utilization and Development of Machine Learning for Molecular Applications – Molecular Machine Learning
• Ehemaliger Antragsteller: Dr. Tobias Gensch, until 5/2023