Methodology for roughness measurement and contact analysis for optimization of interface roughness

Abstract

Surface roughness measurements are performed in the computer industry using various measuring instruments - contact type stylus profiler, noncontact optical profiler and atomic force microscope (AFM) with different scan sizes and sampling intervals. Methodology of choosing a suitable instrument, scan size and sampling interval for a given application, is developed. AFM is recommended for roughness measurement of commercial magnetic disks with a scan size of 20 /spl mu/m/spl times/20 /spl mu/m or greater and sampling interval of 0.1 /spl mu/m or less. For a surface with Gaussian height distribution, a surface can be adequately characterized by standard deviation of surface heights (Rq or /spl sigma/) and correlation distance (/spl beta/*). For ultralow flying head-disk interfaces, measurement of peak-to-mean distance (Rp) is also recommended. For contact analysis of dry and wet interfaces, a numerical computer model has been developed to predict real area of contact and meniscus forces. In this model, a large number of asperities can be modeled and no assumption on the shape and size of the asperities is made. The model can be used for measured or computer generated 3-D rough surfaces. The model has been used to study the effect of surface roughness and liquid film thickness on tribological performance, to rank various disk candidates, and to develop optimum roughness profiles. For computer generated Gaussian surfaces, contact area is proportional to (/spl beta/*/Rq). Nanoasperities present on the disk surface are found to go through plastic deformation whereas most of the deformation is elastic. Thus, high frequency roughness must be avoided to minimize plastic deformation and resulting wear. Stiction increases with liquid film thickness (h) and Rq but is independent of /spl beta/*. There is a critical film thickness for a surface with a given Rq (h/Rq/spl sim/0.5 to 1) above which stiction increases rapidly. Non-Gaussian surfaces with skewness of 0 to 0.2 and kurtosis of 5 or larger are found to exhibit lower real area of contact and meniscus force and are less sensitive to h/Rq than Gaussian surfaces. Identical asperities represent an optimum surface. A relationship between optimum number of asperities and asperity radii is presented. Finally, some design concepts to minimize stiction and wear are presented.