Memory effects and coverage dependence of surface diffusion in a model adsorption system

Abstract

We study the coverage dependence of surface diffusion coefficients for a strongly interacting adsorption system O/W͑110͒ via Monte Carlo simulations of a lattice-gas model. In particular, we consider the nature and emergence of memory effects as contained in the corresponding correlation factors in tracer and collective diffusion. We show that memory effects can be very pronounced deep inside the ordered phases and in regions close to first and second order phase transition boundaries. Particular attention is paid to the details of the time dependence of memory effects. The memory effect in tracer diffusion is found to decay following a power law after an initial transient period. This behavior persists until the hydrodynamic regime is reached, after which the memory effect decays exponentially. The time required to reach the hydrodynamical regime and the related exponential decay is strongly influenced by both the critical effects related to long-wavelength fluctuations and the local order in the overlayer. We also analyze the influence of the memory effects on the effective diffusion barriers extracted from the Arrhenius analysis. For tracer diffusion, we find that the contribution from memory effects can be as large as 50% to the total barrier. For collective diffusion, the role of memory effects is in general less pronounced. ͓S0163-1829͑99͒03011-8͔

Density profile evolution and nonequilibrium effects in partial and full spreading measurements of surface diffusion

“…The interaction parameters have been chosen 27,28 such that the resulting phase diagram shown in Fig. 1 is in close agreement with the experimental observations.…”

“…Thus, if the behavior of D C is dominated by particle number fluctuations, then the minima of D C provide information of phase boundaries. 38 In previous studies, we have shown that this situation is indeed realized in the present model system at Tϭ590 K. 28 Using this approach, results in Figs. 3͑c͒ and 4͑c͒ reveal that if one analyzes the Boltzmann-Matano data for full spreading at too early times, then the estimated locations of phase boundaries can deviate significantly from the actual equilibrium values.…”

“…1 is in close agreement with the experimental observations. 29 In the present study we concentrate on the coverage dependence of D C at a temperature of Tϭ590 K to allow direct comparison to the equilibrium results for D C ( ), 27,28 and to results obtained from Boltzmann-Matano analysis of full spreading. 26 At T ϭ590 K, starting from the coverage ϭ 1 2 corresponding to the ideal p(2ϫ1) phase, the system undergoes a continuous phase transition around Ϸ0.37 to a disordered ͑DO͒ phase at low coverages.…”

“…30 We wish to emphasize that the model has been extensively studied, and a complete set of equilibrium data for diffusion coefficients and other relevant quantities at various temperatures can be found elsewhere. 27,28 As is well known, a proper choice of the transition probabilities within the Monte Carlo method is important to guarantee realistic description of surface diffusion. [31][32][33] In particular, the thermally activated nature of surface diffusion processes must be described in a physically sensible fashion.…”

“…27,28 Thus D C ( ) in Eq. ͑2͒ is the collective diffusion coefficient calculated in equilibrium, and ⌫ is the average transition rate of all possible single-particle jump attempts.…”