Learning organo-transition metal catalyzed reactions by graph neural networks.

Chemical reaction outcome prediction presents a fundamental challenge in synthetic chemistry. Most existing machine learning (ML) approaches focus on chemical reactions of typical elements. We developed a simple ML model focused on organo-transition metal-catalyzed reactions (OMCRs). Instead of overall reactions observed in experiments, we let the ML model learn the sequence of simplified elementary reactions. This drastically reduced the complexity of the model and helped it find common patterns from distinct reactions. We let a graph neural network learn the reactivity index of a pair of atoms. The model was able to learn a wide variety of OMCRs, and the accuracy of reaction prediction reached 97%, even though the model has extremely fewer learnable parameters than other standards. The learned reactivity indices of bonds nicely summarize the knowledge of reactions in the dataset.

Synthesis of 1-(Difluoromethyl)alk-1-enes via Palladium-Catalyzed SN2'-Type Substitution Reaction of Difluoromethylated Allylic Phosphates with 1,3-Dicarbonyl Compounds and Imides.

Herein, we report the synthesis of 1-(difluoromethyl)alkenes via a palladium-catalyzed reaction of difluoromethyl-substituted allylic phosphates with 1,3-dicarbonyl compounds using PdCl2(PPh3)2 as a precatalyst. 1,3-Dicarbonyl compounds attacked the γ-carbon with respect to the difluoromethyl group to afford their corresponding SN2'-type substitution products irrespective of the substitution pattern in the allylic phosphates. This regioselectivity has been ascribed to the electronic environment of the unsymmetrical π-allylpalladium intermediate using density functional theory (DFT) calculations. The reaction of difluoromethyl-substituted allylic phosphates with imides was also carried out using a different catalyst system composed of [PdCl(η3-allyl)]2 and di(diphenylphosphino)butane (dppb).