Modeling Realistic Structures of Trimetallic Nanoalloy Catalysts Using Chemically Meaningful Descriptors

The computational design of alloy catalysts is hindered by the uncertainty in their structure and arrangement of constituent elements within the lattice of catalyst particles, i.e., their chemical ordering. Moreover, chemical ordering in alloy nanoparticles (nanoalloys) can be affected by the reaction temperature due to thermal disorder. In this study, we develop a method for realistic simulations of trimetallic alloy nanocrystallites with the lowest energy chemical ordering or chemical ordering taking into account thermal disorder in the nanoalloy. This method is based on Monte Carlo simulations using topological lattice Hamiltonian, whose parameters are fitted to the results of density functional (DFT) simulations of thoughtfully designed archetypal nanoalloy structures. The implementation of this method in Python code is freely available online. Using this method, we characterized chemical orderings in nanoparticles composed of 79 and 338 atoms of metals with known catalytic activity in CO2 hydrogenation, namely, Pd-Pt-Cu, Ni-Pd-Cu, and Co-Rh-Cu. Our simulations show that the thermal disorder in these alloys significantly affects the composition of surface sites. Such structural changes are demonstrated to affect the average binding energies of reaction intermediates to the catalyst surface by up to 1.1 eV, implying their critical effect on the alloy’s catalytic properties. Moreover, we demonstrate how the developed code can be used for brute-force evaluation of entropic contributions to mixing free energies in alloy nanoparticles. The demonstrated abilities of the proposed method to generate realistic models of trimetallic nanoalloys in a computationally efficient manner enable reliable simulations of catalytic properties of trimetallic catalysts for their in-depth understanding and computational design.