A Health Factor for Process Patterns Enhancing Semiconductor Manufacturing by Pattern Recognition in Analog Wafermaps

Schrunner, Stefan; Jenul, Anna; Scheiber, Michael; Zernig, Anja; Kaestner, Andre; Kern, Roman

doi:10.1109/smc.2019.8914309

Cited by 3 publications

(6 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They already introduced the usage of Rotational Local Binary Patterns (RLBP) and Histogram of oriented Gradients (HoG) features in their work which proved to extract sufficient information from the wafer maps to reliably predict process patterns on them. The proposed scheme enhances the initially proposed framework by Schrunner et al [10] by introducing Radon-transform-based features in combination with ensemble classification. The scheme depicted in Figure 2 is proposed, which enables wafer map classification by utilizing one-vs-rest (OVR) ensemble classification.…”

Section: Enhanced Machine Learning Pipeline For Pattern Recognitionmentioning

confidence: 96%

“…It is imperative to develop a classification framework that can produce robust models within low-data regimes which enable accurate predictions. Schrunner et al [10] introduced the Wafer Health Factor (WHF), which enables the assessment of wafer maps by combining domain knowledge with pattern recognition. They already introduced the usage of Rotational Local Binary Patterns (RLBP) and Histogram of oriented Gradients (HoG) features in their work which proved to extract sufficient information from the wafer maps to reliably predict process patterns on them.…”

Section: Enhanced Machine Learning Pipeline For Pattern Recognitionmentioning

confidence: 99%

“…Most of the existing research regarding ML within semiconductor manufacturing focuses on the development of algorithms for tasks such as predictive maintenance of equipment [9] as well as pattern recognition for wafers [10].…”

Section: Introductionmentioning

confidence: 99%

“…Due to the complexity involved in the semiconductor manufacturing process, pattern recognition is very challenging because some observed patterns tend to occur more often than others. Thus, in real-world applications and implementations, the lack of sufficient fully-descriptive wafer test data for less occurring patterns restricts the performance of predictive ML models [10].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Improving wafer map classification in Industry 4.0

Scheiber¹,

Nartey²,

Zernig³

et al. 2022

2022 IEEE 5th International Conference on Industrial Cyber-Physical Systems (ICPS)

Self Cite

View full text Add to dashboard Cite

At the heart of Industry 4.0. lies the automation of manufacturing processes and interoperability of corresponding applications, in most cases utilizing Cyber-Physical Systems and Internet of Things. Used within diverse industrial areas across the world, these concepts gained more importance in recent years. In the semiconductor industry, continuous improvement of all involved processes is achieved by utilizing these concepts in many different stages of the manufacturing process. This work proposes a machine learning framework that on one hand enhances wafer map classification in the testing stage of semiconductor devices, and on the other hand fulfills the requirements and demands of all involved stakeholders in adopted engineering processes. The core of the proposed system is the Wafer Health Factor, a machine learning framework that detects process deviations in analog wafer test data through pattern recognition. Additional integration of Radon-based features, as well as the use of an ensemble classification framework, boosts model performances in the presented real-world application scenario significantly. Moreover, we also show that the proposed framework performs well on two benchmark data sets from literature, even on small training data sets. We conclude, that the presented framework improves the performance of pattern recognition tasks in realworld applications and thus, enables the automatic and early detection of deviations within semiconductor manufacturing processes.

show abstract

Section: Enhanced Machine Learning Pipeline For Pattern Recognitionmentioning

confidence: 96%

Section: Enhanced Machine Learning Pipeline For Pattern Recognitionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Improving wafer map classification in Industry 4.0

Scheiber¹,

Nartey²,

Zernig³

et al. 2022

2022 IEEE 5th International Conference on Industrial Cyber-Physical Systems (ICPS)

Self Cite

View full text Add to dashboard Cite

show abstract

“…In the integrated circuits manufacturing industry, wafer fabrication is an essential procedure [1][2][3]. The manufacturing process of wafer is divided into two processes: front-end and back-end.…”

Section: Introductionmentioning

confidence: 99%

A class imbalanced wafer defect classification framework based on variational autoencoder generative adversarial network

Wang¹,

Wei²,

Wang³

2022

Meas. Sci. Technol.

View full text Add to dashboard Cite

Wafer Defect Classification (WDC) can be crucial to the wafer fabrication process. Engineers can quickly respond to improve the technological process, averting further defects through WDC. However, due to the complex fabrication steps, wafer defects are different in various types. This causes a severe data imbalance problem in WDC. To effectively solve the problem, this study introduces a class imbalanced wafer defect classification framework (CI-WDC) based on Variational Autoencoder Generative Adversarial Network (VAE-GAN). This framework consists of VAE-GAN and wafer defect classifier. Among them, VAE-GAN is responsible for creating new samples to solve the imbalance problem while the classifier is responsible for classifying wafer defect patterns. Specifically, VAE-GAN combines the advantage of a variational autoencoder (VAE) and generative adversarial network. VAE networks can produce subtle differences that do not affect the properties of the data when generating new images. At the same time, the proposed discriminator can help us constrain the generated images to be close to real samples and avoid irrational, feature-missing, and ambiguous samples. WM-811K dataset is utilized to verify the above method. The experimental results validate that the samples generated by VAE-GAN have a significant improvement in the performance of the WDC system.

show abstract