Simulation tools for predicting the atomic configuration of bimetallic catalytic surfaces

Transition metal alloys are an important class of materials in heterogeneous catalysis due in no small part to the often greatly enhanced activity and selectivity they exhibit compared to their monometallic constituents. A host of experimental and theoretical studies have demonstrated that, in many cases, these synergistic effects can be attributed to atomic-scale features of the catalyst surface. Realizing the goal of designing -- rather than serendipitously discovering -- new alloy catalysts thus depends on our ability to predict their atomic configuration under technologically relevant conditions. This dissertation presents original research into the development and use of computational tools to accomplish this objective. These tools are all based on a similar strategy: For each of the alloy systems examined, cluster expansion (CE) Hamiltonians were constructed from the results of density functional theory (DFT) calculations, and then used in Metropolis Monte Carlo (MC) simulations to predict properties of interest. Following a detailed description of the DFT+CE+MC simulation scheme, results for the AuPd/Pd(111) and AuPt/Pt(111) surface alloys are presented. These two systems exhibit considerably different trends in their atomic arrangement, which are explicable in terms of their interatomic interactions. In AuPd, a preference for heteronuclear, Au-Pd interactions results in the preferential formation of Pd monomers and other small ensembles, while in AuPt, a preference for homonuclear interactions results in the opposite. AuPd/Pd(100) and AuPt/Pt(100) were similarly examined, revealing not only the effects of the same heteronuclear/homonuclear preferences in this facet, but also a propensity for the formation of second nearest-neighbor pairs of Pd monomers, in close agreement with experiment. Subsequent simulations of the AuPd/Pd(100) surface suggest the application of biaxial compressive strain as a means increasing the population of this catalytically important ensemble of atoms. A method to incorporate the effects of subsurface atomic configuration is also presented, using AuPd as an example. This method represents several improvements over others previously reported in the literature, especially in terms of its simplicity. Finally, we introduce the dimensionless scaled pair interaction, whereby the finite-temperature atomic configuration of any bimetallic surface alloy may be predicted from a small number of relatively inexpensive calculations.