The project started as a breaking study of large molecular catalyst at carbon-water interfaces. We have been able to simulate reactions under highly realistic conditions with salts, electric fields, solvent molecules with full dynamic treatment. We revealed a new mechanism for the oxygen-oxygen bond formation in the water splitting process. One major finding was that this mechanism was unaffected by the environment of the carbon electrode, while others are blocked. With the current allocation we will develop models for exploring more cases where this mechanism is operating, including large catalytic oligomers. Our group is the only one in the world currently studying reactions with the full complexity of the realistic system. The established distinction is becoming blurred between homogeneous catalysis, where the catalyst is dissolved in a solvent, and heterogeneous catalysis, where the catalyst is a solid reacting with gaseous or liquid reactants. Molecular catalysts attached to surfaces have advantages from both worlds, bur are notoriously difficult to model. Neither periodic or molecular DFT methods are suitable. We have therefore developed classical models based on for simulation in Gromacs and the empirical valence bond (EVB) methods to be able to simulate such catalysts under realistic conditions, taking into account the environment of the interface. We will use Gromacs and the system sizes will be significant (20k-1M atoms), which is well suited for the Beskow computing environment. The calculations will be combined with quantum chemical computations and EVB simulations, but these will be run on systems better suited for this purpose. We have recently also developed our methods for simulating molecular catalysts to single atom catalysts, which are single metal atoms embeded in a material, e.g. iron atom in nitrogen doped graphite or nanotubes. We have indications that the effect of the environment on the reaction steps could be of decisive magnitude, enough to change the mechanism from one to another. This could lead to completely reversed product selectivity. In this project we will investigate the effect of different solvents, different salts in the electrolyte solution, and different ions on single atom catalyzed CO2 reduction. Finally we have developed a model for metal-organic frameworks, which will readily build the structure and generate force fields more or less automatically. We will run large simulations for these systems and extract a significant number of snapshots (ca 1000). From these structures we will cut out a subsystem containing several metal oxide nodes, several organic linkers and electrolyte molecules/ions (ca 500-2000 atoms). We will then use VeloxChem to estimate overlap integrals between a reduced linker and a non-reduced linker. This value is key to calculating electron transfer rates, which was found experimentally to be highly dependent on the choice of ions in the electrolyte.