Hydrosilylation of alkenes and alkynes with Pt(II)-based catalysts and application of Machine Learning algorithms to optimize the reduction of amides to amines.

Abstract

The purpose of this thesis was twofold: the Hydrosilylation of Alkenes and Alkynes with seven different Pt(II)-based catalysts, and the application of Machine Learning (ML) method in chemistry with the aim of optimizing reaction’s conditions and limiting the waste of time, chemicals and money. The Hydrosilylation of Alkenes and Alkynes, that is, the addition of silanes across carbon–carbon double or triple bonds, represents the ideal pathway to produce organosilicon compounds, thus, it is hardly surprising that this transformation constitutes the core of the organosilicon industry in order to produce various silicon compounds ranging from bulk commodities to fine chemicals and specialty products, for example, lubricating oils, paper release coatings, or grafting agents. Furthermore, the organosilane products serve as valuable foundational components for organic synthesis, by taking advantage of the richness and versatility of organosilicon chemistry. Concerning my project, first of all two simple [Pt(thioether)2Cl2] pre-catalysts were synthesized starting from PtCl2 or K2[PtCl4] and Tetrahydrothiophene (THT) or Dimethyl sulfide (DMS). Then, these two catalysts and other five were used for the Hydrosilylation in different loadings (mol%). The other five catalysts ([Pt(IPr)(DMS)Cl2], [Pt(SIPr)(DMS)Cl2], [Pt(IMes)(DMS)Cl2], [Pt(IPr*)(DMS)Cl2] and [Pt(ICy)(DMS)Cl2]) are all based on [Pt(DMS)Cl2] and the fourth ligand is an N-heterocyclic carbene. Their catalytic activity was studied with Gas Chromatography and 1H-NMR, by calculating the yield and the conversion of the substrate into the desired product. The catalysts that were used for the Hydrosilyaltion are the same that were employed in the second section of the project: the optimization of the reduction of amides to amines. The optimization of chemical reactions is often guided by empirical methods, it means that chemists engage in testing operational parameters that, according to their understanding of the reaction, are expected to yield optimal results. This approach can be time and resource-intensive, given that slight adjustments to various factors can significantly impact production. Machine Learning (ML) enables the creation of complex patterns between the different process variables and quickly pinpoints the optimal operation point while guiding a smart and efficient experimental campaign. Sunthetics ML is an easy-to-use machine learning (ML) platform that helps you develop new materials, processes, and formulations using very few data points to accurately predict your system's behavior. The chemical industry has become the third largest contributor of greenhouse gas emissions, with more than half of its resources ending up in waste streams. Sunthetics' mission is make the chemical industry more sustainable, one reaction at a time. The reaction that was tried to be optimized in this project by using the algorithms suggestions was the reduction of amide to amine and, from reading the results, it was possible to confirm that the algorithm had proposed the reaction with the most optimal conditions already in the second set of suggested reactions.