Optimizing Double-Sided Glass Polishing Using Rationally
Designed Experiments


Patrick G. Murray, Abigail Hooper, Harry W. Sarkas and Nathan Hoffmann

Nanophase Technologies Corporation


Modern processes for simultaneous double-sided polishing of glass parts are complex operations with a number of dependent and interdependent variables which impact removal rate, surface finish and defectivity.  As a practical consequence, and for lack of insights into how ceria slurry, pad and process related parameters interact, commercial polishing operations are often conducted under less-than-optimal conditions, which can translate into longer processes, poorer yields, more re-work and ultimately higher costs of operation and reduced profitability.

In this paper, the authors examine various approaches to optimizing a double-sided polishing process using a plasma-derived cerium oxide slurry to polish fused silica, including the application of statistically designed experiments (DOE).  The effect of particle and slurry chemistry, along with process elements and considerations related to tool operation, are explored throughout several experiments with a view towards improving polishing results (removal rate, surface finish and flatness) and reducing costs (polishing time and consumable consumption).  The predictive capability associated with the robust design of experiments to improve the efficiency and effectiveness of double-sided polishing is emphasized.