Abstract
Semiconductor manufacturing processes are reliant on physical measurements of metrology equipment to enable decision making related to, and control of, the process. These metrology tools are expensive and the measurement process is time consuming, motivating the need for alternative methods to acquire the same information. Virtual Metrology (VM) methods attempt to provide such alternatives by predicting the required characteristics based on data collected from the tool and work piece, prior to and during the manufacturing process. VM applications have been enabled by advancements in the sensing technology employed in this industry. In this paper, the high sampling rates that have recently become available are exploited to extract a set of features that capture dynamic effects in the tool behavior. The task of predicting defect levels, and hence product quality, is considered in this work by framing it as a classification problem. A Support Vector Machine (SVM) informed by the aforementioned features, in combination with traditional statistical features, is exploited, in this paper, to predict whether or not the defect levels will exceed a heuristically defined threshold. The proposed methodology is also evaluated using data obtained from a major 300mm semiconductor manufacturing fab, yielding prediction accuracy of over 90%.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.