(121l) Gocia: A Grand Canonical Global Optimizer for Clusters, Interfaces, and Adsorbates

Many seemingly simple catalytic systems have been found to undergo significant surface restructuring in reaction conditions, often coupled with adsorption and/or incorporation of reagents and reaction intermediates. Hence it is crucial to explore the chemical space of the off-stoichiometric reconstructions, within the grand canonical ensemble, which is a global optimization challenge.

Here we report a highly versatile python code, GOCIA (Global Optimizer for Clusters, Interfaces, and Adsorbates), which features grand canonical genetic algorithm (GCGA) to efficiently explore the aforementioned scenarios. It has been applied to a wide range of systems, ranging from thermal to electro-catalysis, and from sub nano cluster to extended surface and overlayer, successfully yielding atomistic insights into their structure, reactivity, and spectroscopy, with fine resolution into the metastable states. We also demonstrate the latest development to constrain or bias the global optimizer towards a specific metastable regime, and to combine with machine learning interatomic potentials for accelerated sampling. This update enables investigation of kinetic trapping into metastable regimes and non-equilibrium evolution.

This talk will be a mix of both implementation and applications of the method, and we hope GOCIA can be useful to the communities of catalysis, material, and AI4Science.