Random-field effects on dynamical scaling in the domain growth of a chemisorbed overlayer

We report a time resolved, high-resolution low-energy electron diffraction study of the self-similar growth (scaling) of (2×1) oxygen domains chemisorbed on a W(112) surface at 0.5 ML coverage in the presence of random nitrogen impurities (≤5% ML). The initial stage of domain growth after the system was quenched from a low-temperature disordered (frozen) state to a high-temperature ordered state was shown to follow a power law R¯∝$t^n$, where R¯ is the average domain size at time t. The growth exponent n was measured to be less than (1/2 and decreases with increasing N impurities, an indication of slowing down of the growth kinetics. The scaling function extracted from the angular profiles of superlattice beams was found to be an isotropic, universal function independent of up-quenching temperature, and can be described well by either a Lorentzian squared plus a Lorentzian form or a power Lorentzian form. The breakdown of the scaling was observed after the termination of the power-law growth regime. This breakdown is shown by model calculations to be caused by the change of the domain-size distribution as a result of the pinning of domains by impurities. These results are compared with recent theoretical predictions and Monte Carlo simulations on the kinetics of random-field Ising models.