| Title |
Comparative Analysis of Machine Learning-based Optimization and DOE for Performance Enhancement in Photolithography Process |
| Authors |
김채린(Charin Kim) ; 정유진(Yujin Jung) ; 최규철(Gyucheol Choi) ; 이승철(Seungcheol Lee) ; 이종욱(Jongwook Lee) |
| DOI |
https://doi.org/10.5573/ieie.2026.63.1.19 |
| Keywords |
Photolithography; Process optimization; Design of experiments; Machine learning; Semiconductor manufacturing |
| Abstract |
To overcome the limitations of conventional DOE-based optimization in enhancing photolithography precision, this study proposes a Gradient Boosting Regression (GBR) method that effectively captures nonlinear interactions among process variables. Experiments using RPM, exposure time, and develop time compared DOE and machine learning techniques on collected data. The machine learning approach yielded superior performance across major quality indicators versus DOE. However, for fine patterns, high sensitivity to process variations necessitates improved model precision and expanded data. This study demonstrates that machine learning-based optimization can accurately model complex process interactions and suggests directions for further enhancement in semiconductor mass production. |