Surface Statistics Determined From IR Scatter

Abstract

ABSTRACTThis paper examines the requirement for reflector that is imposedwhen BRDF data is used to calculate surface statistics. There are applications foruse of these calculations at surfaces that appear rough at visible wavelengths(greater than about 300 angstroms). Wavelength extension into the mid IR andincreasing the angle of incidence are the two techniques investigated. Experimentalresults are compared to two widely used diffraction calculations by using knownsinusoidal gratings and the advantages and limitations of each are discussed. Extension of the calculations to surfaces that have surface features as large as5000 angstroms has been demonstrated at a wavelength of 10.6 micrometers.2. INTRODUCTIONThe conversion of the bidirectional reflectance distribution function (BRDF)into the reflective surface power spectral density (PSD) may be accomplished viaseveral diffraction theories [1 -7] providing that the sample has the followingproperties. It must be a front surface reflector, which is to say that the scatteris due only to the surface and not to sub -surface or bulk scattering. The samplemust also be clean, implying that the scatter is not due to surface contaminants.The third requirement, and the one that is the subject of this paper, is that thesurface must be smooth. The first two criteria are necessary to assure that thescatter signal is dominated by surface topography. The third requirement is aresult of the mathematical relationships that are used to convert the BRDF tosurface topography information (ie: the PSD, rms roughness, slope, etc.). A smoothsurface may be loosely defined as one for which the rms roughness is much less thana wavelength of the incident light (v « A). This means that for visible light,surfaces are restricted to those of mirror like quality. As indicated in thefollowing paper of these proceedings, there are industrial processes, involving muchrougher surfaces, for which a non -contact roughness measurement is desirable [8].The work reported here examines ways of pushing the smoothness requirement as faras possible in order to accomplish measurement of rougher surfaces.