The quantification of SIMS depth profiles by Maximum Entropy reconstruction
The quantification procedures applied to raw SIMS data were devised on the basis of a simple model for the sputtering and ionisation that occur during measurement. The model and the associated quantification procedures have long been known to be inaccurate. If SIMS is to remain a useful analysis tool in the future, the quantification procedures must be adjusted such that current features of interest are accurately measured. This thesis describes the development of a more accurate (though empirical) model for the effects of the analysis, using the convolution integral. We propose a method for the quantification of SIMS depth profiles appropriate to this model, using Maximum Entropy (MaxEnt) reconstruction. SIMS depth profile data differ significantly from previous applications of the MaxEnt method: the very high signal to background ratio of the technique has lead users to plot the results on a logarithmic axis, giving much importance to extremely small signals. The noise on SIMS depth profiles has been characterised. A number of optimisation algorithms have been developed and tested, and the performance of the MaxEnt method on SIMS data has been assessed. A novel form of the entropy, particularly suited to SIMS depth profiles, has been suggested. This form has given excellent results.