Chemical Mechanical Planarization of Electro-Plated Copper Through-Wafer Interconnects

The demand for more functionality in a smaller amount of space has driven the microelectronics industry into developing 3-D chip stacking schemes to be used in next generation packaging. One possible solution is through the use of through-wafer interconnects (TWIs) in which vias in a silicon wafer are filled with electroplated Cu. The electroplating process results in a very thick layer of Cu being deposited on the surface of the wafer which can be up to 50μm thick with very poor uniformity. Typical plating schemes used in the semiconductor industry to provide other types of interconnects result in a Cu layer thickness of 0.5-2μm thick, much thinner than that used in TWIs. Chemical-mechanical planarization (CMP) is the process used to planarize the Cu and remove variation that may be present on the surface, however, very little is understood about the underlying physics and chemistry associated with thick Cu CMP behavior. For TWIs applications, optimization of the process is critical in achieving aggressive, uniform Cu removal, and up to the time of this thesis, had not been pursued. In this work, three designed experiments (DOEs) were performed to characterize the CMP process and deliver new methods for removing thick Cu. In the first DOE, a screening experiment was conducted identifying key variables associated with CMP. As a result slurry chemistry, certain machine parameters and the type of pad used all surfaced as having a primary influence on removal rate and uniformity, In the second DOE, slurry chemistry effects were investigated more in-depth with a strong interaction found between slurry pH levels and hydrogen peroxide concentration. Within the interaction space, a possible optimization point was discovered. In the third DOE, different types of pads were investigated to determine pad effects on the CMP process when combined with different slurry flow rates. Although pad effects alone did not surface as having a significant impact on the process, the interaction between the pad type and flow rate combination did provide insight into the mechanisms associated with pad selection and slurry flow rate. With limited information available with regards to thick Cu CMP, this study begins to understand the effects of process variables on the CMP process with conclusions made based on experimental findings