Parts | Days | Selection | Search | Updates | Downloads | Help

MM: Fachverband Metall- und Materialphysik

MM 12: Poster I

MM 12.13: Poster

Monday, March 27, 2023, 18:15–20:00, P2/OG1+2

Deep learning for generation of optimal reaction environments — •Rhyan Barrett¹ and Julia Westermayr² — ¹Leipzig University — ²Leipzig University

The design of reaction environments to reduce activation energies holds enormous potential for advancing many areas of chemical engineering but remains a difficult task due to the high combinatorial complexity of different conditions that influence a reaction. Herein, we use a cutting-edge deep learning model to enable the optimization of reactions. Initially we use a model to generate an abstract electrostatic field that reduces the activation barrier of a given reaction. We then look to optimize the ratio of continuum solvents to match the influence of the optimal electrostatic field generated by the model. The advantage of our method is that it is not limited to the initial solvent selection since any designed mixture will be compared with the global optimum electrostatic field produced by our model. The potential of the method will be demonstrated by optimization of the Claisen rearrangement reaction of allyl-p-tolyl ether and construction of an optimal environment but is generally applicable to any organic reaction.