The preferred method of fabricating copper metal interconnects is damascene patterning using chemical mechanical polishing (CMP). This work examines the direct CMP of two low dielectric constant (low-κ) polymer materials, bis-benzocyclobutene (BCB) and “silicon-application low-κ material” (SiLK), and three organosilicate (OSG) materials.; The goals of this work are to (1) gain fundamental insight into the mechanical and chemical processes that occur during the direct CMP of low-κ thin films; (2) to develop a successful CMP process for the direct polishing of low-κ dielectrics that results in physically, chemically, and electrically stable films following CMP; and (3) to discern the mechanisms that control low-κ CMP through slurry chemistry and abrasive interactions.; A successful CMP process has been obtained for each low-κ material investigated, with surfaces having a root mean squared (RMS) roughness typically ranging from 0.6–2.0 nm after CMP. BCB and SiLK CMP rates are independent of nitric acid concentration in slurries with 0.05 and 0.30 μm Al 2O3 abrasive particles. However, when a surface-weakening chemical reaction occurs at the polymer surface, the CMP rate becomes dependent on the reactant concentration in the slurry, as is the case of SiLK CMP with a novel potassium-hydrogen-phthalate slurry.; OSG removal rates vary from 40–80 nm/min in slurries commonly used to polish silicon dioxide, with removal rate increasing as the film carbon content decreases and the slurry pH increases. OSG surface roughness after CMP is as low as 0.15 nm at a slurry pH of 10.8 and 0.41 nm at a slurry pH of 6.0. Surface and bulk chemical measurements show that chemical reactions with the slurry during CMP do not result in a composition gradient normal to the OSG surface, and that these reactions do not penetrate into the bulk of the films.; We have developed modified Langmuir-Hinshelwood surface kinetics to describe a five step CMP process for SiLK polymer: (i) mass transport of reactant to polymer surface, (ii) adsorption of reactant, (iii) reaction at the polymer surface, (iv) shear-enhanced desorption, and (v) mass transport of abraded product away from polymer surface. We have compared model results to experimental data to discover which mechanistic steps determine the overall rate of SiLK removal. The experimental data indicate that SiLK CMP is reaction-rate limited for low slurry reactant concentrations and desorption-rate limited at high slurry reactant concentrations. Within the desorption-rate limited regime, the rate constant for removal has a power-law relation to the shear stress developed during CMP. The new mechanistic approach to CMP surface reaction kinetics shows promise as a robust model for CMP removal of other materials such as copper, tungsten, aluminum and SiO2, independent of the slurry chemistry and abrasive particle. (Abstract shortened by UMI.)
展开▼
