Catalysis in solution is central to the production of societally ubiquitous specialty chemicals, pharmaceuticals and polymers, as well as underpinning the development of sustainable technologies for Net Zero. Computational modelling can accelerate catalyst discovery, but to maximise its benefits a step-change in current capabilities is needed to achieve chemical accuracy, i.e. computed energetics (kinetic barriers, thermodynamics) to within 1 kcal/mol of experimental values. This requires highly accurate quantum mechanics (QM) calculations and great strides in method development continue to be made.^{1, 2} Likewise molecular dynamics (MD) simulations provide detailed descriptions of solute-solvent interactions,³ a prerequisite to achieving higher model accuracy. However, in the context of solution-phase chemical catalysis these two research communities remain largely separated. This workshop aims to bridge that gap.

Defining an accurate chemical model for reactivity in solution is extremely challenging. Continuum methods are commonly used to capture bulk solvent properties (dielectric; polarizability), but they fall short when specific solute-solvent interactions are crucial. A compromise involves including a few explicit solvent molecules to capture key interactions, but this raises fundamental questions: how many solvent molecules are needed, where should they be placed around the solute? How can this be done consistently across multi-step reactions where the solute may change significantly? An even-handed treatment requires effective explicit solvation with many solvent molecules and extensive sampling of the conformational space of both solvent and solute.

In principle, these problems can be tackled via ab-initio MD (AIMD) and QM/MM methods, and refined protocols for aqueous environments based on these approaches have been successfully applied to study (for example) protein structure and dynamics In contrast, applications in modelling chemical catalysis remain challenging and are often complicated by the treatment of transition metals and organic solvents. While ground-breaking work has been done, for example clarifying long-standing questions about Grignard chemistry,⁴ Pd catalysis⁵ and modelling catalysis at the liquid-surface interface,⁶ the computational cost remains prohibitive, making these studies the exception⁷ rather than the rule.

Against this backdrop, recent developments promise the step-change required to model solution-phase reactivity. Molecular cluster growth algorithms⁸ and microsolvation protocols⁹ bring explicit solvation modelling within reach. Machine learning (ML)-based potentials have also emerged as promising alternative for solution reactivity modeling^{10, 11 12} while GPUs permit the implementation of efficient workflows to generate data-sets for ML applications.^{13, 14} As these exciting approaches become available, collaboration between applied computational chemists and experimentalists remains essential to generate high quality empirical data for benchmarking studies.

This workshop will bring together experts in the development of computational protocols, molecular simulation, ML and data science. The participation of applied computational chemists with interests across homogeneous, heterogeneous and electrocatalysis will also ensure the connection between methodology developments exciting and topical experimental problems in these fields. To date, these sub-fields have pursued different solutions to the problem of explicit solvation. Through interactive discussions, gaps between the fields will be bridged, best practice identified and technologies exchanged. This workshop will therefore focus future developments on key problems in contemporary chemical catalysis.