Surface roughness causes increased light-scattering. This leads to decreased specular reflectivity and degradation of interference. The light scattering is more pronouced at a shorter weavelengths. For this reason, long Visible and Near-infrared (NIR) wavelength range (700-1700nm) is better suited for appication with surface roughness.  Simulation of the effect of surface roughness on reflectance spectra (Fig.1) shows that interference pattern is significantly degraded for RMS> 100nm. And it becomes challenging to measure thickness in these conditions.                                             

  Rough surface exhibits peaks and smooth areas on the microscopic level (Fig. 2). It is clear that there is significant roughness on 20um scale but, there are smooth areas on a < 5um scale. So, it should be possible to select a small enough measurement spot to reduce the roughness effect. Indeed, results on Fig.3, 4 shows that using 2um measurement spot surface roughness effect can be practically eliminated and coating thickness can be  reliably measured. Roughness topology is also leading to variation of  the coating thickness on microscopic level - a potential source of measurement degradation(as we will review below). But the small spot size helps to reduce this as well. The smallest possible measurement spot size does not always gives best results. The spot size needs to be optimized based on the particular application.