Effects of anisotropic diffusion and finite island sizes in homoepitaxial growth: Pt on Pt(100)-hex

Abstract

The diffusion, nucleation, and growth of Pt on the hexagonally reconstructed Pt(100)-hex surface are investigated. By means of Scanning Tunneling Microscopy (STM), the positions, sizes, and number densities of monoatomically high, rectangular, reconstructed Pt islands, formed in the submonolayer coverage regime, have been determined for substrate temperatures in the range T = 318-497 K and adatom deposition rates from R = 4 × 10 -5 to 7 × 10 -3 site -1 s -1 . The measurements are compared to the results of kinetic Monte Carlo (KMC) simulations and rate equation theory. The Pt(100)-hex surface exhibits a height modulation caused by the misfit between the topmost quasi-hexagonal layer and the quadratic substrate, resulting in a highly anisotropic large scale surface morphology with six-atom wide channels running along the [110] direction. From an autocorrelation analysis of the determined island positions, it is revealed that the islands are distributed with long/short correlation lengths along/perpendicular to the reconstruction channels. The autocorrelation analysis allows us to quantify the degree of anisotropy in adatom diffusion. Island size distributions obtained at different temperatures are found to collapse onto a single scaling curve also in the present case of strongly anisotropic diffusion. By comparison to similar curves derived from KMC simulations in a model incorporating anisotropic diffusion and finite island sizes, it is concluded that the critical island size is i = 1 and that the mobility of dimers is negligible. Furthermore, an early onset of island coalescence is revealed. From the scaling of the measured saturation island density, N x ∼ (R/h) χ , where h = v exp(-E d /k B T) is the adatom hopping rate, an effective fround to be influenced by the finite extent of the islands when diffusion is anisotropic. This is due to the increased ability of the islands to capture adatoms as they grow to cover more diffusion channels. Rate equations incorporating this effect are solved, and a scaling exponent of χ = 1/. is derived in contrast to the χ = 1/4 obtained for a 1-D point-island model.