Optimal filtering of scanning probe microscope images for wear analysis of smooth surfaces

A procedure for removing cumulative drift and white noise from scanning probe microscope images has been constructed. Smooth amorphous carbon overcoats on superpolished hard disk media in particular were examined using a scanning probe microscope. The surfaces typically had a similar to 1 nm rms roughness over a scan length of 10 mu m. The low roughness yielded a relatively low signal to noise ratio in the unfiltered image. While a conventional filter removes a great deal of noise, an optimal Fourier (Wiener) filter that more selectively removes noise from the image is discussed. White noise and drift were modeled and their contributions to the power spectrum are estimated, resulting in an open clamshell-shaped two-dimensional filter. The effect of the filter was demonstrated by subjecting filtered images of unworn and worn areas to a smooth surface to second derivative calculations in different directions. Anisotropy in the wear process associated with the wear direction is apparent in the optimally filtered images. (C) 1996 American Vacuum Society.