A simple model for the prediction of coating thickness uniformity from limited measured data

The major problem with most thickness uniformity models is that they can only be applied retrospectively to measured distributions. A model has recently been developed to enable the prediction of coating thickness variation across a flat surface directly facing the vapour source. The model is based on an empirically observed inverse power law relationship between thickness and source-to-substrate distance, and the constants in the equation can be determined from a minimum of two measured thicknesses at different locations on the sample, and produce a good fit to the fully measured distribution.

The model has been shown to work for both electron beam evaporation based techniques (including vacuum and gas evaporation, and PAPVD) and also for sputtering systems, both balanced and unbalanced. The model is now being extended to enable the prediction of thickness profiles on the rear faces of the samples in order that an overall thickness distribution can be predicted. It is shown that the front-to-back thickness uniformity model reported previously is inappropriate for large substrates, and this is also discussed.