The performance of a optical component is greatly dependent on its surfaces. There are many ways to measure and
evaluate a surface so far, but actually no one knows exactly what a "real" surface likes, and thus makes some
measurement unexplainable. This paper develops a way to construct various synthetic surfaces. Referred to actual AFM
measurement, regular or irregular surfaces can be created and studied. Bumps, scratches, granules, profile errors and
other construction elements can be added to form a very complicated virtual surface. The number, size and distribution
of these elements can be changed, and if necessary, the surface roughness can be controlled in a specialized range. As
these surfaces are created synthetically, we know exactly what construction these surface contain. The height data of
created virtual surfaces can be transferred into the file that AFM instrument can read and handle. As a typical example, a
very complicated surface is created step by step in this paper, and its PSD for each step are calculated. This process
mikes it very clear how the surface component element affect its PSD function, and is benefit to our better understanding
of real surface construction.
This paper presents a comparative studied of surface roughness. Samples were first measured by atomic force microscope (AFM). The acquired height data was then processed to calculate their power spectral density (PSD) and multi-fractal spectrum (MFS). The calculation results indicate that MFS of different samples with same sampling length differs significantly from each other, while the MFS of the same sample with different sampling length or different sampling position is quite similar. The calculation also shows that MFS is very sensitive to particles or scratches appeared on the surface. The PSD of the same data acquired from these samples are also presented for comparison. It is clear from the calculation results that the PSD curves vary with the sampling position and sampling length, thus makes the evaluation uncertain. No quantitative index available from PSD, only qualitative information obtained. Comparatively, MFS is better in description of a surface roughness.