Sputter roughening of inhomogeneous surfaces: impurity pinning and nanostructure shape selection

A model is proposed for sputter roughening of inhomogeneous systems with slowly sputtered impurity particles randomly distributed in the bulk. Surface inhomogeneity, which develops as a result of coupling between the time evolution of the local surface impurity concentration and the local surface shape, is tuned by changing the dependence of the sputtering probability upon impurity concentration. In 1+1 dimensions, we find long-time scaling exponents that are consistent with Kardar-Parisi-Zhang (KPZ) values. However, for a range of surface inhomogeneity, impurity pinning results in a persistent growth regime where the surface roughens rapidly. We correlate this rapid roughening to fluctuations of the impurity concentration at the surface. Roughening in this regime leads to the formation of cones whose shape is determined by material property and sputtering flux, suggesting a unique method of nanostructure fabrication. In 2+1 dimensions, a similar variation of the roughening behavior with surface inhomogeneity is observed. For small surface inhomogeneity, there is an initial exponential roughening followed by power-law roughening with an effective growth exponent much smaller than KPZ. For larger surface inhomogeneity two power-law roughening regimes are observed, with an initial rapid roughening that crosses over to slower roughening; the effective exponent in each of these regimes increases with surface inhomogeneity. The surface morphology observed in the simulations is considerably noisier than experimental data for InP and GaSb. Our model shows noisy nonlinear pattern formation in contrast to the marked long-range hexagonal ordering seen in experiments. However, the scaling behavior is robust enough that roughening kinetics similar to that observed experimentally can be obtained depending upon the values of inhomogeneity and the strength of the nonlinear term in the model.