Problems in the deconvolution of SIMS depth profiles using delta-doped test structures

The notion that the experimentally observed sputter depth profile can be corrected for the distortions introduced by ion mixing, through a deconvolution process using the as-measured result for a delta-function like impurity distribution, is examined critically. Starting point is an existing simple, but conceptually transparent and basically not incorrect, model to describe ion mixing effects. First it is shown that the characteristic exponential rise and decay lengths can be expressed as the fourth power sum of intrinsic and mixing contributions. Then it is demonstrated that the observed variation of the full-width at half-maximum with primary beam parameters (ion type, impact energy and angle of incidence) for delta-like impurity distributions is incompatible with the notion that the mixing efficiency may be estimated from the slopes of the profile. Consequently, the outcome of a deconvolution step does not result in a unique solution, but depends (strongly) on the measurement conditions chosen.