Experimental testing, using both commercially available polishing machines and a specially built test platform, demonstrates that material removal rates (MRRs) observed during polishing of fused silica are strongly affected by nanometer-scale vibration amplitudes. Specifically, a nanometer level increase in system vibrations can produce MRRs approximately 150% higher than on an inherently smoother running machine. Moreover the higher spatial frequency surface roughness values are little-effected by the spectral content of the polishing machine. Polishing under controlled conditions, using the test platform, shows that for vibration amplitudes, A ≲ 1.6 μm, and over a fairly wide range of vibration frequencies, MRR increases almost linearly with increasing input power. By contrast, for A ≳ 10 μm, MRR exhibits a rapid decay with increasing A. Order of magnitude analyses and physical arguments are presented in order to explain the qualitatively distinct MRR trends observed. In the small-amplitude limit, A ≲ 1.6 μm, two arguments are presented which suggest that the total observed removal rate, MRRtot, reflects the superposed action of chemical-mechanical removal, MRRcm, and vibration-driven, flow-induced removal, MRRflow, i.e., MRRtot=MRRcm+MRRflow. The analyses further indicate that MRRflow primarily reflects cyclic viscous shears and pressure gradients extant within the thin, non-Newtonian slurry film that exists between the polishing tool and workpiece. Shears and pressure gradients, and corresponding flow-induced MRRs, are, in turn, found to scale as A/doω, where A is the vibration amplitude, do is the characteristic gap thickness between the tool and workpiece, and ω is the vibration frequency. In the large-amplitude limit, A ≳ 5 μm, experimental measurements and a simple scaling argument show that the polishing slurry film becomes thick enough that the workpiece and polishing tool lose direct contact. In this limit, observed MRRs thus reflect strictly vibration-driven, flow-induced removal. Comparisons of theoretical material removal rates, derived from the scale analyses presented, with experimentally measured removal rates, as observed in both the small- and large-amplitude limits, show that the models proposed provide reasonable predictions of observed removal rates.
Read full abstract