Presentation on theme: "Ana Maldonado, Gadi Rothenberg Van t Hoff Institute for Molecular Sciences, University of Amsterdam, The Netherlands. A world of."— Presentation transcript:
Ana Maldonado, Gadi Rothenberg Van t Hoff Institute for Molecular Sciences, University of Amsterdam, The Netherlands. A world of possibilities In homogeneous catalysis, the combination of laboratory automation and advanced modelling algorithms puts us on the brink of in silico catalyst design. To realise this goal, we must assemble and screen virtual libraries of ligand-metal complexes (Figure 1). Catalyst selection is the major problem. Generating millions of structures via computer is easy, but how should we choose the candidates for synthesis and testing? Backbone Diversity Analysis in Catalyst Design References. 1. Understanding ligand diversity. J. A. Westerhuis, J. A. Hageman, H. -W. Frühauf and G. Rothenberg, Chim. Oggi - Chemistry Today, 25, (2007). 2. Molecular Similarity and Diversity: Concepts and Applications. A. G. Maldonado, M. Petitjean, J.-P. Doucet and B.T. Fan., Molecular Diversity, 10, (2006). Scheme 1. Nickel Catalysed Hydrocyanation of butadiene in the Nylon process. Figure 1. Left: catalyst decomposition framework in building blocks. Right: a catalyst space generated by thousands of combinations. Backbone Diversity holds the key The ligand backbone often dictates the ligand size and flexibility, which are related to the reaction pocket environment and the catalytic performance. Choosing the right backbone is a crucial step in ligand design. By analysing the backbone diversity in the descriptor space (space B, Figure 2) we can generate a diverse spread of ligand- metal complexes over a given catalyst space A. Figure 2. Space A (the catalyst space), is a grid containing the metal-ligand complexes. Space B (the descriptor space) contains the values of the catalyst descriptors, and space C contains the figures of merit. CH 3 O P O Combinatorial Chemistry The backbone dataset is described using three main attributes: size, flexibility, and polarity, extracted from a list of 168 computed descriptors. PCA score plots (Figure 3) as well as descriptor maps (Figure 4) were built for the backbone dataset. Diversity analysis was done by computing the average distance of each catalyst to all other catalysts in Figure 4. A TON TOF selectivity polarity size flexibility ligating group backbone residue B C Application to hydrocyanation catalysis We examined a group of 42 backbones of a set of biphosphite and biphosphine ligand-nickel complexes used for catalysing the hydrocyanation of butadiene to adiponitrile (Scheme 1). Each catalyst was divided into two L ligating groups, a backbone B, and the residue groups R (Figure 1). This division gives a standard framework for backbone comparison. Figure 3. Backbone 3D descriptor space Figure 3. PCA score plot of the backbone dataset. Zero centred graph is shown in blue. Green circles shows clusters and red circles outliers. The combination of the original 42 backbones with 10 ligating groups and 20 residues, represents a catalyst space of 1.1 ×10 9 possible combinations. Sampling even 1% of this is practically impossible. Our selected subset of 24 representative backbones correspond to bidentate catalysts, which represents a reasonable 3.79% of the 2.6 ×10 5 total combinations. Whats next? This database serves as a basis for further QSAR and in silico catalyst design for the Ni-catalyzed hydrocyanation of butadiene.