Date of Submission


Document Type


Degree Name

Master of Science in Mechanical Engineering


Mechanical and Industrial Engineering


Dr. Joseph A. Levert

Committee Member

Dr. Sam B. Daniels

Committee Member

Dr. Eric Dieckman


Polishing rate, Pad-independent abrasive friction, CMP, Preston relation, Frictional measurements, Wear volumes


Chemical mechanical planarization, Grinding and polishing, Friction


As features on integrated circuits continue to grow smaller, they become more susceptible to damage from sequential planarization steps during fabrication. As planarization (known as chemical mechanical planarization, or simply CMP) is preformed multiple times and on every stage of fabrication, potential damage from it represents a significant financial risk, motivating a more fundamental understanding of the material removal process. CMP typically consists of a stiff polymer pad being used to bring a chemically active colloidal suspension of nanoscale particles into contact and relative motion with the substrate to be polished. For narrow sets of conditions, CMP is typically seen to follow the Preston relation: material removal as a function of normal force and polishing distance. Modifications to the Preston equation have been made to relate wear volume to polishing friction force. A model has been developed for oxide CMP friction that segregated the frictional force generation into three regions: non-polishing hare-pad fraction at the real pad-wafer contact sites, a pinned particle friction engaged in two-body particle abrasion at the real pad-wafer contact sites, and a swept region near the real contact engaging in three body particle abrasion. The aim of this thesis is to advance the modified Preston relation with the novel model of friction to present a novel model that relates the wear volume to the work done by the two- and three-body particle abrasion, removing confounding non polishing pad friction. The system studied was that of a fused silica wafer being polished with an alkaline silica particle suspension. A pin-on-disk tribometer was used for the frictional measurements and profilometry was used to determine the wear volumes. Independent wear factors were found for the pinned particles and the swept region of 7.78E-11(m3/N·m) and -9.94E-12(m3/N·m) respectively. It is concluded that the polishing occurs at the pinned particle contact sites, and that, while frictionally significant, the swept region does not meaningfully contribute to material removal.