Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792494
C. Edwards, Aditya Kumar, Alex Vaske, Nathan McDaniel, Dipali Pradhan, Debashis Panda
Our test tools pick and place units into sockets for electrical testing. Defects or loose debris accumulated inside the test sockets will likely damage each subsequent unit being tested until the issue is detected and the defective socket is repaired or replaced. To resolve this critical issue, we equipped each pick-and-place arm with a new machine vision system designed to fit inside the existing tool. The limited footprint constraints required a highly compact imaging system which resulted in a variety of image artifacts, creating several unique challenges for the inspection system. We developed an inspection algorithm that utilizes a variety of advanced computer vision and machine learning techniques to normalize and match the images, remove artifacts, and detect defects. The flagged socket images can be manually dispositioned by the user and the socket can be sent for repair or cleaning as needed.
{"title":"Real-Time Automated Socket Inspection using Advanced Computer Vision and Machine Learning : DI: Defect Inspection and Reduction","authors":"C. Edwards, Aditya Kumar, Alex Vaske, Nathan McDaniel, Dipali Pradhan, Debashis Panda","doi":"10.1109/asmc54647.2022.9792494","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792494","url":null,"abstract":"Our test tools pick and place units into sockets for electrical testing. Defects or loose debris accumulated inside the test sockets will likely damage each subsequent unit being tested until the issue is detected and the defective socket is repaired or replaced. To resolve this critical issue, we equipped each pick-and-place arm with a new machine vision system designed to fit inside the existing tool. The limited footprint constraints required a highly compact imaging system which resulted in a variety of image artifacts, creating several unique challenges for the inspection system. We developed an inspection algorithm that utilizes a variety of advanced computer vision and machine learning techniques to normalize and match the images, remove artifacts, and detect defects. The flagged socket images can be manually dispositioned by the user and the socket can be sent for repair or cleaning as needed.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"251 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115846617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792528
O. Patterson, MD Golam Faruk, Datong Zhang, Guanchen He, B. Sheumaker
Contact CMP is one of the most popular layers for application of e-beam inspection. Using voltage contrast, contact opens and shorts may uniquely be detected. Generally positive surface charging (positive mode) is used, but negative surface charging (negative mode) provides a number of noteworthy advantages, including throughput, which are highlighted in this paper. Usually, adjustment of the landing energy or electric field is used to shift a wafer image into negative mode. This paper introduces and demonstrates a new control knob, beam density.
{"title":"Negative Mode E-Beam Inspection of the Contact Layer","authors":"O. Patterson, MD Golam Faruk, Datong Zhang, Guanchen He, B. Sheumaker","doi":"10.1109/asmc54647.2022.9792528","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792528","url":null,"abstract":"Contact CMP is one of the most popular layers for application of e-beam inspection. Using voltage contrast, contact opens and shorts may uniquely be detected. Generally positive surface charging (positive mode) is used, but negative surface charging (negative mode) provides a number of noteworthy advantages, including throughput, which are highlighted in this paper. Usually, adjustment of the landing energy or electric field is used to shift a wafer image into negative mode. This paper introduces and demonstrates a new control knob, beam density.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116001970","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792514
Rebecca Clain, Ikram Azzizi, Valeria Borodin, A. Roussy
Manufacturing processes are often subject to drifts over production cycles. This study applies the paradigm of Transfer Learning (TL) to support the updating of a Virtual Metrology (VM) model based on a Convolutional Neural Network (CNN). The VM is applied to a Chemical Mechanical Planarization (CMP) to predict the average material removal rate. Through the prism of a benchmark case study, this paper empirically investigates how transfer learning can improve the updatability of a VM CNN-based model.
{"title":"On Updating a Virtual Metrology Model in Semiconductor Manufacturing via Transfer Learning","authors":"Rebecca Clain, Ikram Azzizi, Valeria Borodin, A. Roussy","doi":"10.1109/asmc54647.2022.9792514","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792514","url":null,"abstract":"Manufacturing processes are often subject to drifts over production cycles. This study applies the paradigm of Transfer Learning (TL) to support the updating of a Virtual Metrology (VM) model based on a Convolutional Neural Network (CNN). The VM is applied to a Chemical Mechanical Planarization (CMP) to predict the average material removal rate. Through the prism of a benchmark case study, this paper empirically investigates how transfer learning can improve the updatability of a VM CNN-based model.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"1706 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127451732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792532
Qiong Luo, Huang Jing Yan, Zhou Jianbo, Ong Shiang Yang
This work presents a robust process development of deep-trench isolation (DTI). The instability of DTI junction break down is largely attributed to the process-induced stress. Through TCAD and SEM analysis, the high stress point and stress-induced dislocation are found to be located at sharp DTI corners. Based on this observation, we aim to reduce the thermal/mechanical stress by optimizing the liner oxidation and oxide hard mask for DTI trench. The combined effect will eliminate the dislocation defect that is usually associated with high aspect ratio DTI and offer a significant yield gain for power devices.
{"title":"Robust Process Development for Dislocation Free DTI Formation","authors":"Qiong Luo, Huang Jing Yan, Zhou Jianbo, Ong Shiang Yang","doi":"10.1109/asmc54647.2022.9792532","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792532","url":null,"abstract":"This work presents a robust process development of deep-trench isolation (DTI). The instability of DTI junction break down is largely attributed to the process-induced stress. Through TCAD and SEM analysis, the high stress point and stress-induced dislocation are found to be located at sharp DTI corners. Based on this observation, we aim to reduce the thermal/mechanical stress by optimizing the liner oxidation and oxide hard mask for DTI trench. The combined effect will eliminate the dislocation defect that is usually associated with high aspect ratio DTI and offer a significant yield gain for power devices.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130828057","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792490
J. Trujillo-Sevilla, A. Roqué-Velasco, Miguel Jesús Sicilia, Ó. Casanova-González, José Manuel Ramos-Rodríguez, J. Gaudestad
In this paper we introduce a new optical metrology technique for measuring wafer geometry on full 300 mm blank and patterned silicon wafers. Wave Front Phase Imaging (WFPI) is presented that acquires 7.65 million data points in 5 seconds on a full 300mm silicon test wafer allowing for a lateral resolution of 96μm. The wafer geometry was measured 30 times on a blank 300mm silicon wafer front and backside while the wafer was resting on 3 pins close together and the deflection was found to be in close agreement to estimations for gravity pull. The system has repeatability with root-mean-square standard deviation (σRMS) of 4.75nm on the front side and 6.51nm on the backside of a 300mm silicon test wafer. Using a double Gaussian filtering technique with a 100μm lateral cutoff frequency, nanotopography was revealed on the full silicon wafer showing a repeatability of 0.265Å for the frontside and 0.204Å for the backside of the full 300mm silicon wafer.
{"title":"Wafer Geometry Technique for Blank 300mm Silicon Wafers","authors":"J. Trujillo-Sevilla, A. Roqué-Velasco, Miguel Jesús Sicilia, Ó. Casanova-González, José Manuel Ramos-Rodríguez, J. Gaudestad","doi":"10.1109/asmc54647.2022.9792490","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792490","url":null,"abstract":"In this paper we introduce a new optical metrology technique for measuring wafer geometry on full 300 mm blank and patterned silicon wafers. Wave Front Phase Imaging (WFPI) is presented that acquires 7.65 million data points in 5 seconds on a full 300mm silicon test wafer allowing for a lateral resolution of 96μm. The wafer geometry was measured 30 times on a blank 300mm silicon wafer front and backside while the wafer was resting on 3 pins close together and the deflection was found to be in close agreement to estimations for gravity pull. The system has repeatability with root-mean-square standard deviation (σRMS) of 4.75nm on the front side and 6.51nm on the backside of a 300mm silicon test wafer. Using a double Gaussian filtering technique with a 100μm lateral cutoff frequency, nanotopography was revealed on the full silicon wafer showing a repeatability of 0.265Å for the frontside and 0.204Å for the backside of the full 300mm silicon wafer.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122958775","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792533
J. Tan, R. Shick, Joseph A. Peri, I-fan Wang, Amarnauth Singh, R. Beera
Defect control is imperative to leading-edge semiconductor industry. Nano filtration has been effective at reducing defects caused by particles and metal contamination. In this presentation, we focus on filters that consist of polysulfone (PSF) membrane and ion exchange membrane (IEM) for effective removal of particles and metal ions. In the real application case, either one filter containing dual layer membranes (PSF + IEM) or two separate filters (PSF filter + IEM filter) can be installed to remove both particles and metal ions from chemicals or water used in semiconductor manufacturing, depending on the process conditions and requirements. Polysulfone polymer membrane has a highly asymmetric structure that allows superior flow rate and fine particle retention. With optimal filter design, the best filters can remove particles down to 1 nm size while exhibiting a low pressure drop. The membrane in IEM typically has negatively charged functional groups on the surface. The leading IEM product can effectively remove >90% trace metal ions from solutions. This manuscript will give an overview of the issues at end users and discuss how advanced filtration technology help solve the issues and enable semiconductor manufacturing.
{"title":"Nano Filtration Using Polysulfone Membrane : CFM: Contamination Free Manufacturing","authors":"J. Tan, R. Shick, Joseph A. Peri, I-fan Wang, Amarnauth Singh, R. Beera","doi":"10.1109/asmc54647.2022.9792533","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792533","url":null,"abstract":"Defect control is imperative to leading-edge semiconductor industry. Nano filtration has been effective at reducing defects caused by particles and metal contamination. In this presentation, we focus on filters that consist of polysulfone (PSF) membrane and ion exchange membrane (IEM) for effective removal of particles and metal ions. In the real application case, either one filter containing dual layer membranes (PSF + IEM) or two separate filters (PSF filter + IEM filter) can be installed to remove both particles and metal ions from chemicals or water used in semiconductor manufacturing, depending on the process conditions and requirements. Polysulfone polymer membrane has a highly asymmetric structure that allows superior flow rate and fine particle retention. With optimal filter design, the best filters can remove particles down to 1 nm size while exhibiting a low pressure drop. The membrane in IEM typically has negatively charged functional groups on the surface. The leading IEM product can effectively remove >90% trace metal ions from solutions. This manuscript will give an overview of the issues at end users and discuss how advanced filtration technology help solve the issues and enable semiconductor manufacturing.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"157 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116038705","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792508
Ki Dong Yang, Hanbit Park, Joonho Lee, E. Hwang, Jiwoo Jeong, S. Kwon, Kihyun Kim, Jaein Jeong, Eunyoung Han, Young Jeong Kim, Joong Jung Kim
In this work, the ability to remove post-etch residues by simply changing dechucking environment gas from Ar to O2 is presented. The O2 dechucking process was observed to enhance in-situ cleaning effect, reducing Cl and Br anion by 93% and 50%, respectively, on the wafer surface (HPIC) and 101 ~ 102 times within the Si substrate (tof-SIMS). FOUP ion analysis and optical emission spectroscopy results also revealed that the amount of process outgas can be significantly reduced. As the halogen atoms that are originally remaining inside the Si substrate was removed, the damaged Si layer was also reduced by O2 curing. By this simple method, within the etch process, the wafer can be patterned and additionally, in-situ cleaned. Furthermore, based on DFT calculation, we clarified that O radicals can remove polymeric residues like SiOxCly and SiOxBry by replacing halogen atoms on energy perspective. The O2 treated oxide has less number of interface trap (Nit), proving device performance also can be enhanced.
{"title":"In-situ Cleaning of Post-etch Byproducts by Manipulating Dechucking Environment Gas in Silicon Etch Process","authors":"Ki Dong Yang, Hanbit Park, Joonho Lee, E. Hwang, Jiwoo Jeong, S. Kwon, Kihyun Kim, Jaein Jeong, Eunyoung Han, Young Jeong Kim, Joong Jung Kim","doi":"10.1109/asmc54647.2022.9792508","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792508","url":null,"abstract":"In this work, the ability to remove post-etch residues by simply changing dechucking environment gas from Ar to O2 is presented. The O2 dechucking process was observed to enhance in-situ cleaning effect, reducing Cl and Br anion by 93% and 50%, respectively, on the wafer surface (HPIC) and 101 ~ 102 times within the Si substrate (tof-SIMS). FOUP ion analysis and optical emission spectroscopy results also revealed that the amount of process outgas can be significantly reduced. As the halogen atoms that are originally remaining inside the Si substrate was removed, the damaged Si layer was also reduced by O2 curing. By this simple method, within the etch process, the wafer can be patterned and additionally, in-situ cleaned. Furthermore, based on DFT calculation, we clarified that O radicals can remove polymeric residues like SiOxCly and SiOxBry by replacing halogen atoms on energy perspective. The O2 treated oxide has less number of interface trap (Nit), proving device performance also can be enhanced.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"2000 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116680570","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792521
Inimfon I. Akpabio, S. Savari
Prediction intervals which describe the reliability of the predictive performance of regression models are useful to influence decision making and to build trust in machine learning. Normalized conformal prediction is a rigorous and simple guideline to construct prediction intervals which has no distributional assumptions but requires other types of modeling to assess a regression model fit to training data, and quantile regression is a widely used technique in other fields to construct prediction intervals. We propose image denoising and other image processing techniques as a foundation to prediction interval construction procedures for line edge roughness (LER) estimation from noisy scanning electron microscope (SEM) images and show that these innovations offer significant improvements in efficiency over earlier approaches used to study the deep convolutional neural network EDGENet.
{"title":"On an Application of Denoising to the Uncertainty Quantification of Line Edge Roughness Estimation","authors":"Inimfon I. Akpabio, S. Savari","doi":"10.1109/asmc54647.2022.9792521","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792521","url":null,"abstract":"Prediction intervals which describe the reliability of the predictive performance of regression models are useful to influence decision making and to build trust in machine learning. Normalized conformal prediction is a rigorous and simple guideline to construct prediction intervals which has no distributional assumptions but requires other types of modeling to assess a regression model fit to training data, and quantile regression is a widely used technique in other fields to construct prediction intervals. We propose image denoising and other image processing techniques as a foundation to prediction interval construction procedures for line edge roughness (LER) estimation from noisy scanning electron microscope (SEM) images and show that these innovations offer significant improvements in efficiency over earlier approaches used to study the deep convolutional neural network EDGENet.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132644213","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792527
Oussama Djedidi, Rebecca Clain, Valeria Borodin, A. Roussy
This paper focuses on the feature selection problem in a virtual metrology task applied to a chemical mechanical polishing process. One of the main challenges specific to virtual metrology modeling is the relatively wide availability of measurements and traces (features) versus the scarcity of samples (entries), as they are usually costly to obtain. To overcome these challenges, we propose a hybrid feature selection algorithm, called Enhanced Hybrid Feature Selection (EHFS), that combines a filter approach and a genetic algorithm embedding a machine learning model. The filter starts by eliminating noisy and uninformative features. Then, in the wrapper stage, the genetic algorithm is augmented by a solution archive to favor exploration. This added feature avoids the reevaluation of duplicate candidate solutions and consequently decreases the computational time of EHFS.Numerical experiments, conducted on industrial and benchmark datasets, show that the proposed solution approach performs competitively in terms of both solution quality and computational time compared with two existing approaches: the general-purpose Forward Feature Selection (FFS) and virtual metrology-specific Evolutionary Repetitive Backward Elimination (ERBE).
{"title":"Feature Selection for Virtual Metrology Modeling: An application to Chemical Mechanical Polishing","authors":"Oussama Djedidi, Rebecca Clain, Valeria Borodin, A. Roussy","doi":"10.1109/asmc54647.2022.9792527","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792527","url":null,"abstract":"This paper focuses on the feature selection problem in a virtual metrology task applied to a chemical mechanical polishing process. One of the main challenges specific to virtual metrology modeling is the relatively wide availability of measurements and traces (features) versus the scarcity of samples (entries), as they are usually costly to obtain. To overcome these challenges, we propose a hybrid feature selection algorithm, called Enhanced Hybrid Feature Selection (EHFS), that combines a filter approach and a genetic algorithm embedding a machine learning model. The filter starts by eliminating noisy and uninformative features. Then, in the wrapper stage, the genetic algorithm is augmented by a solution archive to favor exploration. This added feature avoids the reevaluation of duplicate candidate solutions and consequently decreases the computational time of EHFS.Numerical experiments, conducted on industrial and benchmark datasets, show that the proposed solution approach performs competitively in terms of both solution quality and computational time compared with two existing approaches: the general-purpose Forward Feature Selection (FFS) and virtual metrology-specific Evolutionary Repetitive Backward Elimination (ERBE).","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122135581","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-02DOI: 10.1109/asmc54647.2022.9792530
Rabhi Ilham, Roussy Agnes, Pasqualini Francois
Considering the importance of detecting anomalies as soon as they occur in the semiconductor industry, we propose in this paper to study the effectiveness of a robust machine learning classification technique, which is the One-Class Support Vector Machine (OC-SVM), used for out-of-control detection in production line. An optimization of the OC-SVM is proposed to improve its performance with a brief overview of the different methods used in this purpose. Numerical results are then presented based on industrial data provided by STMicroelectronics Crolles.
{"title":"Optimization of OC-SVM engine used for out-of-control detection in semiconductor industry","authors":"Rabhi Ilham, Roussy Agnes, Pasqualini Francois","doi":"10.1109/asmc54647.2022.9792530","DOIUrl":"https://doi.org/10.1109/asmc54647.2022.9792530","url":null,"abstract":"Considering the importance of detecting anomalies as soon as they occur in the semiconductor industry, we propose in this paper to study the effectiveness of a robust machine learning classification technique, which is the One-Class Support Vector Machine (OC-SVM), used for out-of-control detection in production line. An optimization of the OC-SVM is proposed to improve its performance with a brief overview of the different methods used in this purpose. Numerical results are then presented based on industrial data provided by STMicroelectronics Crolles.","PeriodicalId":436890,"journal":{"name":"2022 33rd Annual SEMI Advanced Semiconductor Manufacturing Conference (ASMC)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125534316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：