Temporary storage allocation policy for lot delivery with automated material handling system in semiconductor FABs

Abstract

This thesis addresses the temporary storage allocation policy for lot delivery with automated material handling system (AMHS) in semiconductor fabrication facilities (FABs). In general, it is ideal to deliver the front opening unified pod (FOUP), which contains 25 wafers, directly to the machine for the next step. However, due to the limited space of the machine buffer, it may not be transported to the destination machine. In this case, the wafer FOUP is temporarily stored in the storage system such as side track buffer (STB). The selection of the STB directly impacts the overall performance of the semiconductor FABs and AMHS. In this research, we suggested factors to consider in the process of selecting storage. Based on this, we proposed an efficient storage allocation policy that considers specific scenarios occurring in the semiconductor FABs. This study used a reinforcement learning-based approach that applies to large-scale systems. Finally, we implemented the proposed policy in the testbed simulation, and results showed that the performance of the semiconductor FABs and AMHS were improved compared to the existing policies.