Accurate mapping of dose variations on EUV scanners using dose-to-clear exposures and optical ellipsometry

Abstract

Over the last couple of years, EUV scanners have matured enough, prompting all major foundries to push for its introduction into high volume process development starting at the N7 node. In addition, extending EUV for single and multi-patterning for future nodes is challenging as pitches for both metal and gate contacts get tighter. At such small dimensions, dose control becomes an even more significant factor towards the final die yield. To achieve enough yield, the process must be optimized and tightly controlled. The challenge with process control is to separate out different effects (track, mask, scanner) and do it accurately and quickly.

In this work, we propose a method to actively monitor dose fluctuations on EUV scanners using a dose-to-clear wafer. This technique uses an open frame exposure of the resist in and around the E-0 energy for the resist of interest. A small set of fields are exposed with dose offsets to calibrate the resist sensitivity on the same wafer. This isolates any wafer-to-wafer and process variations that could arise from using a separate calibration wafer. The rest of the fields are exposed with a nominal energy that lies in the center of the calibration curve. Post exposure resist height measurements are done using an optical ellipsometry tool. By knowing the resist sensitivity, the variation in resist height can be translated to dose variations. Such a technique is insensitive to feature size, lens aberrations, mask write errors and reticle heating, thus allowing us to accurately map only dosed-driven variations on the scanner. Results using this technique to monitor the NXE:3400B at IMEC will be shared. Further, correlation to actual dense intra-field CD measurements and best energy estimates from FEM exposures measurements will also be discussed. We were also able to use this monitor to effectively isolate process and mask-driven effects, making a very powerful tool to do in-depth breakdown analysis.