Optimization of electroplating processes using Kriging-based response surface method (K-RSM) and genetic algorithm (GA) for enhanced surface uniformity of copper

Non-uniform copper surface roughness in semiconductor manufacturing causes false Automated Optical Inspection (AOI) signals, necessitating manual re-inspection and reducing production efficiency. This study aims to optimize electroplating processes to improve surface quality and reduce labor dependence by systematically investigating the effects of critical process parameters. Uniform Design (UD) was employed to establish the experimental framework, examining the influence of current density, production time limitation (Q-Time), and plating thickness on electroplating quality. A Kriging-based Response Surface Method (K-RSM) integrated with Genetic Algorithm (GA) was applied for process optimization, followed by experimental validation. Analysis of Variance (ANOVA) assessed the contribution of each factor to surface uniformity. Results demonstrated that optimal parameters—current density of 9 A/dm², Q-Time of 1 hour, and plating thickness of 5.7 μm—achieved 17.48% improvement in surface roughness uniformity and 23.33% reduction in manual re-inspection rates after AOI. Current density exhibited the most significant influence on surface quality. The proposed methodology provides a systematic and reproducible approach for electroplating process optimization, effectively enhancing manufacturing reliability while minimizing labor costs in semiconductor production.