Surface diffusion measured by laser-induced desorption: Monte Carlo simulation of effects of surface defects on diffusion

Abstract

Recently, pulsed laser-induced desorption (LID) has been used to measure surface diffusion of adsorbed species. Often, however, pulsed-laser excitation leads to stress-induced deformations of the surface. We have found, in the case of a Ni(100) surface, that laser excitation causes a change in the surface morphology that results in an increase in the dissociative sticking coefficient of hydrogen by a factor of more than 3. At the same time, adsorption of other species is not appreciably affected. Based on low-energy electron diffraction (LEED) spot profile analysis, the laser-irradiated surface has <10% atomic-step defects. We attempt to estimate the effect that laser-induced surface defects have on the measurement of diffusion by means of a Monte Carlo simulation. Surface defects are introduced into the simulation as trap sites because such defects would have the most pronounced effect on diffusion. We find that 〈x(t)〉 and 〈x2(t)〉 of particles on the defected surface are characteristic of diffusion and obtain an effective diffusion coefficient D from the rate of increase in 〈x2〉 with time. For a p(2×2) array of trap sites and noninteracting particles, D decreases with increasing trap site residence time but the magnitude of the decrease is not large. For example, D is decreased only by a factor of roughly 3 when the trap residence time is 3000 times the normal site residence time. The number of particles N(t) that migrate onto the defected portion of the surface under these conditions is reduced by <10% relative to a defect-free surface. These variations are within the present experimental uncertainties of the LID technique. Hence, for noninteracting adsorbates such as hydrogen, the reported values of D are representative of the defect-free surface. Using the result that diffusion on the defected portion of the surface can be represented by a single effective diffusion coefficient, we derive an analytic expression for the effects of trap sites on N(t). This analytic expression accurately reproduces the Monte Carlo results.