Stochastic simulations of temperature programmed desorption kinetics

Abstract

In principle, temperature programmed desorption (TPD) can yield a wealth of information on the rates of surface reactions. In practice, however, it is often difficult or impossible to extract such information from the experimental data. We describe an accurate stochastic modelling technique that facilitates the determination of mechanisms and rate constants from TPD measurements. The model for a particular reaction or process is cast in terms of a detailed mechanism, which can be set up in a variety of ways depending on the primary characteristics of the system under study. Starting from experimental rate constants obtained under limiting conditions, a series of simulations are used to estimate and refine unknown rate constants. Three types of examples are given to illustrate the methodology used. First, simulations of desorption of Xe from Pt(111) are used to show how a very simple system can be modelled. By comparing experimental data and calculations performed with several different mechanisms, we explore the kinetic consequences of various modes of desorption from defect sites. Next, we investigate the competitive desorption and dissociation of CO on W(110). In this case the model successfully reproduces experimental TPD and surface coverage data, allowing experimental rate constants to be optimized. Finally, we show that stochastic simulations can be used to study systems with strong repulsive lateral interactions using a model which explicitly tracks changes in the environment of a desorbing species with temperature and coverage. Previous Monte Carlo simulations of desorption from a square lattice are reproduced using this approach. It is also applied to desorption of CO from Ru(001), yielding simulated TPD spectra in excellent agreement with experiment and allowing the magnitude of repulsive interactions to be estimated. In both the square and hexagonal lattice systems TPD data are simulated with coverage-independent pre-exponential factors: all features of the spectra are attributable to variations in populations of surface species alone. The stochastic method is shown to be valuable as an integral part of an experimental study, allowing detailed models to be proposed and refined, and yielding microscopic data which are amenable to theoretical analysis.