Chips in semiconductor manufacturing are produced in circular wafers that are constantly monitored by inspection machines. These machines produce a wafer defect map, namely a list of defect locations which corresponds to a very large, sparse and binary image. While in these production processes it is normal to see defects that are randomly spread through the wafer, specific defect patterns might indicate problems in the production that have to be promptly identified. We cast wafer monitoring in a challenging image classification problem where traditional convolutional neural networks, that represent state-of-the-art solutions, cannot be straightforwardly employed due to the very large image size (say 20,000 × 20,000 pixels) and the extreme class imbalance. We successfully address these challenges by means of Submanifold Sparse Convolutional Networks, deep architectures that are specifically designed to handle sparse data, and through an ad-hoc data augmentation procedure designed for wafer defect maps. Our experiments show that the proposed solution is very successful over a dataset of almost 30,000 maps acquired and annotated by our industrial partner. In particular, the proposed solution achieves significantly high recall on normal wafer defect maps, that represent the large majority of the production. Moreover, our data augmentation procedure turns out to be beneficial also in smaller images, as it allows to outperform the state-of-the-art classifier on a public datasets of wafer defect maps.

Wafer defect map classification using sparse convolutional networks

Carrera D.;Fragneto P.;Boracchi G.
2019-01-01

Abstract

Chips in semiconductor manufacturing are produced in circular wafers that are constantly monitored by inspection machines. These machines produce a wafer defect map, namely a list of defect locations which corresponds to a very large, sparse and binary image. While in these production processes it is normal to see defects that are randomly spread through the wafer, specific defect patterns might indicate problems in the production that have to be promptly identified. We cast wafer monitoring in a challenging image classification problem where traditional convolutional neural networks, that represent state-of-the-art solutions, cannot be straightforwardly employed due to the very large image size (say 20,000 × 20,000 pixels) and the extreme class imbalance. We successfully address these challenges by means of Submanifold Sparse Convolutional Networks, deep architectures that are specifically designed to handle sparse data, and through an ad-hoc data augmentation procedure designed for wafer defect maps. Our experiments show that the proposed solution is very successful over a dataset of almost 30,000 maps acquired and annotated by our industrial partner. In particular, the proposed solution achieves significantly high recall on normal wafer defect maps, that represent the large majority of the production. Moreover, our data augmentation procedure turns out to be beneficial also in smaller images, as it allows to outperform the state-of-the-art classifier on a public datasets of wafer defect maps.
2019
IMAGE ANALYSIS AND PROCESSING - ICIAP 2019, PT II
978-3-030-30644-1
978-3-030-30645-8
Industrial monitoring; Pattern classification; Quality control; Sparse Convolutional Networks; Wafer Defect Map
File in questo prodotto:
File Dimensione Formato  
2019_ICIAP_Wafer_Classification_Sparse_CNN.pdf

Accesso riservato

Dimensione 1.4 MB
Formato Adobe PDF
1.4 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/1126167
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 12
  • ???jsp.display-item.citation.isi??? 9
social impact