Computer Simulation of Sputtering

Publisher Website
Google Scholar
Abstract
Present computers have neither enough memory capacity nor computation speed to completely simulate the sputtering process. The conclusions reported here were obtained with purely repulsive potential functions. Surface binding energies were included artificially. The simulations produce excellent spot patterns, but only approximately correct sputtering ratios. The Ar+–Cu system has been studied in detail and a few trials have been made on the Xe+–Cu system. The three Gibson potentials and the Anderson‐Sigmund potential were used for the Cu–Cu interaction. Changing these functions produces quantitatively, but not qualitatively, different results; so a majority of the investigation was conducted with the Gibson II potential. The Ar–Cu and Xe–Cu potentials were those obtained locally from secondary electron emission studies. Variation of parameters in these functions had little effect on the spot patterns. Sputtering ratios change appreciably as the parameters of the potentials are varied, but the unexplained systematic difference between simulation and experiment cannot be removed by parametric variation within a Born‐Mayer potential function model. The simulations indicate that sputtering is caused by two mechanisms: (a) In the more common situation the incoming ion strikes two lattice atoms almost simultaneously. The lower energy of these two primary knock‐on atoms then initiates the sputtering of several of its neighbors. The higher‐energy primary knock‐on may also cause some sputtering, but it usually is less efficient because a large fraction of its momentum is directed into the surface. (2) In a relatively small number of events the ion, after having made one or more collisions, is reflected from the second or third layer of the crystal. It then can initiate sputtering as it returns toward the surface. The second mechanism has never been seen when the ion is heavier than the lattice atoms. Primary knock‐on ions rarely sputter. Occasionally (after making several collisions) one escapes accidently. Almost every incident ion sputters at least one atom, even those which enter a (110) channel. Ions which hit the end of a (110) chain with zero impact parameter, or ions which penetrate normally into the absolute center of the trigonal lens of the (111) surface or the square lens of the (100) surface do not cause sputtering. Surface dynamics dominate the sputtering process. The probability of focuson sputtering appears to be vanishingly small. The relationships between the surface dynamics, transparency, and channeling models are discussed.