The objective of this research project is to address the following issues involved in the predictive modeling and real-time monitoring of chemical mechanical planarization/polishing (CMP) process used in the finishing of semiconductor chips: the interaction of chemical and mechanical phenomena at the silicon wafer-pad interface, the effect of machine vibrations, forces, temperature profiles, and acoustic emission signals, and the modeling of nonlinear stochastic process-machine interactions that capture the dynamic relationships between the wafer-pad interactions and the response of the sensor signals. Both experimental and analytical investigations will be undertaken to address these issues. A production machine will be instrumented with an array of heterogeneous sensors, including force, temperature, vibration, and acoustic emission (both wired and wireless). A sensor fusion approach will be used to monitor various stages of the process. The complex relationships connecting machine-specific and material-specific parameters with performance variables, namely, removal rate and planarity will be delineated by the application of a suite of statistical analysis methods applied to the experimental data. The mechanical and chemical interactions at the wafer-pad interface at various temperatures will be determined by developing a thermal model using Jaeger's classical heat source theory, Gibbs free-energy minimization, and molecular dynamics modeling.
Productivity gains in the finishing of semiconductor devices depend on advances in chemical mechanical planarization/polishing. Wafer sizes, device density, feature dimensions, surface quality, and defect structure are posing serious challenges to the science and technology of chemical mechanical planarization/polishing . This investigation, if successful, will yield deeper insights into various chemo-mechanical interactions and will integrate a heterogeneous sensor network for improving productivity and integrated circuit quality.
