(492c) Modeling and Multi-Objective Optimization for Robust Design of Sulfuric Acid Purification Process in Semiconductor Industry

Authors: 
Song, K., School of Chemical and Biological Engineering, Seoul National University
Jung, J., School of Chemical and Biological Engineering, Seoul National University
Park, S., Seoul National University
Han, C., Seoul National University



Sulfuric acid is a widely used solvent in semiconductor manufacturing for cleaning the wafer. Pure and highly concentrated (usually higher than 80 wt%) sulfuric acid is consumed to effectively eliminate remaining particles in the wafer. Waste acid is discarded after dilution or repeatedly used after purification treatment. However, tighter environmental regulations promote on-site acid recovery system.

 Main design issues of the system in the semiconductor industry are the following. First, semiconductor grade materials should be used for stable repetition of acid recovery. This results in high capital investment requirement. Second, the equipment is recommended to be designed to be compact and in module type for on-site installation. Third, the recovery system is required to handle various waste acid feed conditions resulting from different end-users. This means that the device should be able to deal with consumed sulfuric acid in various temperature and concentration.

 In this work, authors present modeling and optimal design of sulfuric acid purification process. Vacuum distillation in batch process consisting of tanks, a condenser and a vacuum pump is designed to produce concentrated sulfuric acid and segregate waste water. Vacuum condition is essential to meet operating temperature limit posed by material selection. Dynamic modeling of batch process is performed to effectively calculate operating time and steam generation rate, which directly defines the condenser sizing. An optimal design in terms of minimizing total capital cost and operating cost is deduced first. Economic evaluation and full discussion of sensitivity analysis in main variables is performed. Then, robust design with maximizing the operating range to cope with various feed conditions as the objective function is solved in a multi-objective formulation.