Adatom pair distribution up to half coverage: O-Pd(100)

Abstract

The superposition approximation and the Born-Green-Yvon (BGY) equation allows calculation of pair distributions in thermal equilibrium from pair potentials. A 2-d variant can be used to derive adatom pair distributions from arbitrary analytical pair potentials. As practical example substrate mediated elastic interactions, fitted previously to first principles (FP) calculations, are used to derive adatom pair distributions of O-Pd(100). The evaluation method utilizes the particle-hole symmetry of the pair interaction lattice gas Hamiltonian. The nonlinear BGY-type integral equation is solved numerically up to half coverage using power series expansion of coverage.The resulting adatom pair-and three-body distributions are used to analyze the short-range order of adatoms. A Monte Carlo (MC) simulation was performed in parallel to compare the current model with established methods. MC derived pair distributions fit well to those found with the analytical model, except in the high coverage region. The Fourier transformed pair correlations are compared with experimental LEED spot data and provide additional insight.Due to significant attraction the model predicts formation of 3rd and 4th nearest neighbors (n.n.) at small 3rd n.n. chains. A long range order of adatoms like a p(2x2) lattice is not found with the current model but a glassy structure is proposed as also indicated by the equilibrium Monte Carlo configurations. The glassy structure is assumed to contain fluctuations explaining broad p(2x2) LEED spots in the 0.25 monolayer region. The assumptions and limitations of the model towards higher coverages are discussed and open questions are formulated. Keywords Adatom interactions Pair distributions Nonlinear integral equationsMonte Carlo simulation LEED Short-range adatom order 2 APD-04-2a.nb

Finite size scaling in the heterogeneous reaction 2A + B2 → 2AB

“…Diffusion processes of adsorbed species or interactions between adatoms have not been included for simplicity [16][17][18]. This allows us to reduce the number of adjustable parameters of the model, although it does not alter the main achievement of the work, which is predicting the general aspects of the solution in large domains from the exact solution in small domains.…”