Evaluation and Modeling of Alternative Copper and Inter-Layer Dielectric Chemical Mechanical Planarization Technologies

Abstract

The novel consumables studied were abrasive-free copper CMP slurries and high-pressure micro jet technology as an alternative to diamond pad conditioning. Abrasive-free slurries were found to be effective in copper removal and were shown to demonstrate similar removal rate and coefficient of friction (COF) trends as conventional abrasive slurry CMP, while possibly decreasing wafer defects. Fundamental information from the friction spectrum indicated that the periodicity of the cyclic passivation layer formation and removal in copper CMP may be on the order of 10 milliseconds. HPMJ technology was found to be a possible alternative to diamond conditioning with some decrease in removal rate.A controlled atmosphere polishing (CAP) system was used and demonstrated that gaseous additives can feasibly be introduced real-time during a polish. Addition of complexing agents were found to increase removal rates, however it was found that direct etching of copper oxide on the copper surface was not the primary mechanism responsible for removal rate increases during CMP with low oxidant concentrations. Alternatively, it was found that direct etching of the copper oxide is significant in systems containing much higher oxidant concentrations, 1 wt% hydrogen peroxide for example. It was for this reason that a third removal step, chemical dissolution, was added to the two-step removal rate model.The remainder of the work in this dissertation was concerned with characterizing and modeling the copper oxidation and copper oxide dissolution steps of the three-step model separately and applying the appropriate expressions into the CMP removal rate model. The copper oxidation process was found to demonstrate oxide growth, or passivation behavior, at pH of 5 and higher. The oxide growth process was governed by oxidized copper migration through the oxide film. The copper oxide dissolution process was controlled by dissolution of the complexing agent through a dissolution byproduct film. These steps were characterized and applied to the three-step removal rate and predicted removal rate data quite well with one fitting parameter that varied within one order of magnitude. Two real-time experimental measurements, COF and leading pad temperature, can be input into the model to predict removal rates during a polish.