Etch process optimization to improve line edge roughness using precise stochastic process metrology

Optimizing etch processes for Self-Aligned Double Patterning (SADP) is complicated by the metrology challenge of keeping the two populations (core and gap) separate. If that metrology challenge can be solved, the goal is to optimize the SADP process to achieve 1 nm Line-Edge Roughness (LER) for both core and gap while maintaining acceptable CD balance between the two. MetroLER was used to automatically determine SADP populations without relying on CD-SEM stage precision, and to provide unbiased roughness measurements and other metrics of stochastics behavior. The resulting analysis found that line wiggling at the hardmask etch step was one of the problematic steps in the SADP patterning. After full optimization of the SADP process, LER for both core and gap was reduced as much as 25%, to 1.1 nm. Separation of SADP populations and advanced analysis of unbiased roughness, including edge-edge correlations and line/space wiggling, are critical to process improvements and reduction of the impact of stochastics on line/space patterning in advanced DRAM manufacturing.