Multiscale dynamics of the semi-catalytic Rochow-Müller process.

Take a look around and you'll quickly notice silicone products are everywhere—from the spatula in the kitchen or the sealant around the sink to your contact lenses. Silicone has a combination of properties that makes it unique: thermally stable (e.g. kitchen oven), water-repellent, non-toxic, low conductivity, low reactivity, mechanically durable, resists microbial growth, and many more. It is therefore no surprise that silicone production is one of the major markets in process chemistry, and that silicone will continue to be an important product in the future. Silicones are long chains of molecules assembled somewhat like beads on a string or Lego blocks connected together. Our bead or Lego block is the molecule dichlorodimethylsilane, Si(CH3)2Cl2, and is produced in the Rochow-Müller process discovered in the 1940s. It is based on solid silicon reacting with methyl chloride (CH3Cl). An atom is picked from the surface of a silicon lump and bonds with Cl and CH3 from two CH3Cl molecules to form the building block dichlorodimethylsilane. Despite the fact that this process was discovered 80 years ago there is still much we don't know about how the reaction takes place. It is knows silicon must be mixed with copper to react, and there is extensive knowledge about how to do this optimally. But what happens at the atomic level? How do the copper and silicon atoms mix? And going further, how does this system behave differently in the lab versus in a large industrial reactor? In DYNCAT, we will build a modelling toolbox and knowledge database focused on the reactivity of molecules and relevant silicon mixtures. Using machine learning and simulations we will study the Rochow-Müller process down to the atomic level. By understanding what affects the reaction at the atomic scale, we position ourselves to make knowledge-based decisions that can improve the reaction. Such improvements could include higher selectivity, lower waste production, or reduced energy consumption.

DYNCAT aims to develop a highly predictive, physics-based model of the Rochow-Müller process. The project focuses on understanding the dynamic nature of this process starting from the atomic level. The primary objectives of DYNCAT are to elucidate the evolution and behavior of Cu-Si dynamic solid phases at the atomic scale, determine pathways and mechanisms for solid-gas phase reactions at the electronic scale, identify key variables that influence the process at various levels of granularity and how they propagate to the reactor scale, and build a comprehensive knowledge base for understanding and predicting the Rochow-Müller process. A major challenge is understanding the formation and evolution of mixed Cu-Si phases, particularly Cu3Si, which is crucial for the production of dimethyldichlorosilane (M2). Another challenge is investigating how the formation of mixed Cu-Si phases affects the reactivity of gas-phase CH3Cl and surface intermediates, making the activity of both solid and gas reactants highly dynamic and heterogeneous. A key to process improvement is to identify (rate-determining) reactions and energetics for M2 production. Additionally, DYNCAT addresses deactivation by formation of metallic Cu and coke, and how to alliviate this. The project employs interdisciplinary approaches, combining molecular dynamics (MD), density functional theory (DFT), machine learning (ML), microkinetic models and process simulation to model the complex reaction network and predict chemical compositions, structures, reaction pathways, and ultimately provide conversion and selectivity profiles. The project emphasizes open science practices, ensuring that data and tools generated are accessible to both academic and industrial research communities. DYNCAT aims to bridge the gap between computational chemistry and industrial applications, providing a foundation for future studies and practical improvements in large-scale chemical processes.