As integrated circuit (IC) technology progresses to 7-nm nodes and beyond, cobalt (Co) has emerged as a promising substitute for copper (Cu) in interconnects. This shift is driven by Co's shorter electron mean free path, excellent electromigration resistance, and superior deposition characteristics. This study addresses the challenges associated with the selective removal of Co and titanium (Ti) barrier layers during the chemical mechanical polishing (CMP) process while also achieving global wafer planarization. By optimizing the functional groups in the slurry, the process enhances the selectivity of the removal process.

This research integrates CMP experiments with electrochemical tests, static etch experiments, and nanoscratch tests to analyze the removal behaviors of Co and Ti within Co interconnect heterostructures. Electrochemical tests are used to assess the impact of chemical reactions on material removal, while nanoscratch tests evaluate the mechanical strength of wafer surfaces after chemical exposure. The study focuses on the role of complexing agents containing amino (—NH₂) and carboxyl (—COOH) functional groups in improving removal efficiency. By correlating electrochemical data with removal rates, the action mechanisms of these functional groups in the slurry are explored. Additionally, the study analyzes how adjusting the abrasive concentration in the slurry affects Ti removal efficiency.

Experimental results demonstrate significant differences in the removal mechanisms of Co and Ti. Co removal is predominantly driven by chemical corrosion, significantly accelerated by amino functional groups. Mechanical action also plays a role, contributing to rapid Co removal. This behavior is attributed to strong complexation reactions between Co ions and amino groups, facilitating Co dissolution and enhancing the chemical corrosion process. Conversely, carboxyl functional groups have a relatively minor impact on Co removal. Ti removal is primarily led by mechanical action, with chemical corrosion playing a minor role. Increasing the abrasive concentration in the slurry significantly enhances the Ti removal rate. The study confirms that by optimizing the types and concentrations of functional groups in the slurry, selective removal of Co and Ti can be effectively controlled. A comprehensive database has been developed documenting the specific effects of different amino and carboxyl groups under various conditions. Furthermore, the study validates a proposed strategy for controlling the removal rates of heterogeneous Co/Ti structures through patterned wafer experiments. These experiments explored time thresholds for the bulk and barrier CMP processes. The process parameters for the two-step polishing of Co interconnect wafers were optimized, achieving a defect-free Co interconnect structure. During the bulk CMP step, —NH₂ group-based agents were employed, with polishing times controlled between 1.5 and 2.5 min to prevent excessive dishing. Simultaneously, reducing the abrasive concentration lowered the Ti removal rate, further optimizing selectivity between different materials and ensuring superior surface quality. For the barrier CMP step, complexing agents containing an appropriate amount of —COOH groups were used, with polishing times around one minute. An increase in abrasive concentration enhanced the mechanical action of Ti removal.

This optimization strategy not only reduces the Co removal rate but also increases the Ti removal rate, effectively minimizing height differences between material interfaces and achieving the desired planarization effect during the polishing process. These findings provide valuable insights into the removal mechanisms of Co and Ti in the CMP process. They also establish effective strategies for enhancing selectivity and overall process efficiency, offering a theoretical framework for tailoring CMP slurry formulations to meet the specific requirements of advanced IC manufacturing.