Efficient Deep Learning Approach for Fault Detection in the Semiconductor Industry

International audienceThe semiconductor industry is a very cost sensitive industry and yield is key to profitability. The ability to analyse and detect the faulty parts at several 1nanufacturing steps is also very in1portant to ensure the quality of the delivered integrated circuits. Severa! factors as aligninents, shifts or 1nasks rotations can lead to errors during the front-end step (wafer fabrication), or others causes such as fingerprints, scratches and stains can cause cosmetic da.nage during the back-end step (Silicon packaging). Therefore , an automatic visual inspection is required to ensure that the parts are free of any defects. In this chapter, we focus specifically on classifying wafer maps according to predefined defaults. We propose a platform which ain1s al n1aking the classification process n1ore energy efficient, by n1eans of the interconnection of two hardware parts. The first one, the microprocessor STM32MP 1, is responsible for image pre-processing and for offloading inference to a dedicated hardware accelerator. The second one, the hardware accelerator, is i1nple1nented in a Xilinx Zybo Z7-20 FPGA and uses a quantized neural network model. Preliminary results show that, for this low throughput applications that has a limited number of classes, the solution presented in this article can classify in real-time with accuracy above 80% using limited resources.