Bin2Vec: A Better Wafer Bin Map Coloring Scheme for Comprehensible Visualization and Effective Bad Wafer Classification

Abstract

A wafer bin map (WBM), which is the result of an electrical die-sorting test, provides information on which bins failed what tests, and plays an important role in finding defective wafer patterns in semiconductor manufacturing. Current wafer inspection based on WBM has two problems: good/bad WBM classification is performed by engineers and the bin code coloring scheme does not reflect the relationship between bin codes. To solve these problems, we propose a neural network-based bin coloring method called Bin2Vec to make similar bin codes are represented by similar colors. We also build a convolutional neural network-based WBM classification model to reduce the variations in the decisions made by engineers with different expertise by learning the company-wide historical WBM classification results. Based on a real dataset with a total of 27,701 WBMs, our WBM classification model significantly outperformed benchmarked machine learning models. In addition, the visualization results of the proposed Bin2Vec method makes it easier to discover meaningful WBM patterns compared with the random RGB coloring scheme. We expect the proposed framework to improve both efficiencies by automating the bad wafer classification process and effectiveness by assigning similar bin codes and their corresponding colors on the WBM.

A Systematic Review of Deep Learning for Silicon Wafer Defect Recognition

“…Although variations are good for a generalized model, synthetic images should be closer to the real images if data synthesis aims at addressing the imbalance. If similar colors were used to represent identical bin codes or defects, it will ease the process of pattern identification [30]. To prove their hypothesis, Kim et al [30] presented a neural network-based bin coloring method and built a four-layered CNN to distinguish good and bad wafers.…”

“…If similar colors were used to represent identical bin codes or defects, it will ease the process of pattern identification [30]. To prove their hypothesis, Kim et al [30] presented a neural network-based bin coloring method and built a four-layered CNN to distinguish good and bad wafers. However, further classification of bad wafers into respective defect types was not done, which is required for complete classification and the defect root cause analysis.…”

“…Instead of following conventional data sampling methods, Nakazawa and Kulkarni [25] produced an entirely new balanced dataset for CNN training. Kim et al [30] followed a batch training for the original imbalanced data, sampling an equal number of good and bad wafers in each batch.…”

A Deep Convolutional Neural Network-Based Multi-Class Image Classification for Automatic Wafer Map Failure Recognition in Semiconductor Manufacturing

“…A DL-based CNN for automatic wafer defect identification (CNN-WDI) was generated using a data augmentation technique to overcome the class imbalance problem [9]. A CNN-based wafer bin map classification model, as well as a neural network-based bin coloring method called Bin2Vec, have been proposed and designed [32]. Additionally, a deep convolutional encoder-decoder neural network architecture was proposed for detecting and segmenting the eight basic abnormal wafer map failure patterns [33].…”

“…CNN is a special type of DNN designed for image classification [35]. The basic structure of CNN consists of 5 layers: the input layer, the convolutional layer, the pooling layer, the fully connected layer, and the output layer; detailed explanations of these layers are provided in the following [1,12,32,36]:…”