Exploring tailored Ru-triphos catalysts for hydrogenation reactions by combination of experimental, computational, and machine learning techniques

用[Ru (E-triphosA3) (tmm)]配合物可以成功地催化CO2转化为甲醇、甲酸和甲醛衍生物。只有少数研究考虑了导致金属配合物类型[Ru(E-triphosAB2)(tmm)]或[Ru(E-TriphosABC)(tmm)]的非对称配体取代模式。由于对称破缺引起的化学空间的膨胀，包括非对映体配合物的引入，使用实验方法的研究很快变得不可行的。在这个项目中，使用现代统计和计算模型，结合实验证据，用来构建一个支架，捕捉这种对称破碎[Ru (E-triphos AB2) (tmm)]配合物的反应性，从而实现对称破碎配合物的反应性预测。这是通过使用基于经济高效的半经验电子结构理论计算的描述符的机器学习方法来实现的。结合生成的对称和非对称配合物的参考数据，可以对化学空间进行快速筛选。为了使计算方法与实验数据相结合，对[Ru (e -三磷酸AB2) (tmm)]配合物在CO2和乙酰丙酸加氢中的作用进行了详细的研究。因此，创建了一个广泛的数据集。在此基础上，开发了一个机器学习的结构-活性关系模型，目的是在特定底物的基础上预测以前未知的配合物的反应性。

The successful homogeneously catalyzed conversion of CO2 to methanol, formic acid and formaldehyde derivatives could already be demonstrated with [Ru (E-triphosA3) (tmm)] complexes. Only a few studies have taken into account non-symmetrical ligand substitution patterns that lead to metal complexes of the type [Ru(E-triphosAB2)(tmm)] or [Ru( E-TriphosABC)(tmm)]. Due to the expansion of the chemical space caused by symmetry breaking, including the introduction of diastereomeric complexes, an investigation using experimental methods quickly becomes infeasible. In this project, the use of modern statistical and computational models, together with experimental evidence, is used to construct a scaffold that captures the reactivity of such symmetry-broken [Ru (E-triphos AB2) (tmm)] complexes and thus a reactivity prediction for the symmetry-broken complexes enables. This is achieved through the use of machine learning methods based on descriptors from cost-efficient semi-empirical electronic structure theory calculations. Together with the generated reference data of symmetrical and non-symmetrical complexes, a quick screening of the chemical space should be made possible. In order to couple the computational methods with the experimental data, detailed investigations of [Ru (E-triphos AB2) (tmm)] complexes in the hydrogenation of CO2 and levulinic acid are carried out. Thus, an extensive data set is created. Based on this set, a machine-learned structure-activity relationship model is developed with the aim of predicting the reactivity of previously unknown complexes on a substrate-specific basis.