Structural Investigation and Enriched Catalysis of Cu-complex Encapsulated Microporous Catalyst with Pragmatic Modelling for Prediction of Activity by Using Machine Learning.

Abstract

A facile and efficient post-synthetic method has been systematically employed to prepare a hybrid silicoaluminophosphate by immobilizing a mixed ligand copper (׀׀) complex comprising 2,9-dimethyl-1,10-phenanthroline and copper nitrate on base-functionalized material. The effective activity of this catalyst in ring opening reaction of epoxide was achieved when this complex was anchored on amine functionalized SAPO materials. Structure, phase integrity, thermal stability and the existence of functional groups are identified by using different analytical techniques like powder XRD, N2 adsorption-desorption, FT-IR, nuclear magnetic resolution, etc. 90% and 88% conversion were achieved using copper complex immobilized SAPO-34, and SAPO-5 materials, respectively which demonstrate the effectiveness of the materials for this reaction. Machine learnings are employed to predict product conversion and selectivity for scaling up the reaction for further industrial applications. Linear regression (LR), support vector machine (SVM), and k-nearest neighbors (kNN), have been utilized in this study, whereas SVM and kNN both exhibited good performance in predicting catalyst conversion and selectivity, by analysing mean squared error (MSE), mean absolute percentage error (MAPE), and R score. This study showcases the innovative synthesis and performance of the hybrid silicoaluminophosphate catalyst, illustrating the accurate predictive machine learning algorithms to find quality of catalyst for specific reactions.