Unsupervised algorithms for the automatic classification of EWS maps: a comparison
University of Pavia, Ticinum, Lombardy, ItalyDOI: 10.1109/ISSM.2005.1513349 Conference: Semiconductor Manufacturing, 2005. ISSM 2005, IEEE International Symposium on
Source: IEEE Xplore
Recently, it has been shown that the classification of electrical wafer sorting failure maps can be performed by means of unsupervised methods. In this work four different unsupervised methods are compared: SOM, K-means, neural gas, and an expectation maximization. The algorithms are compared using a benchmark based on a probabilistic model. The performance of the classification is assessed by means of an new index, called index-F, based on the knowledge of the real classification. Moreover it is studied the correlation between the proposed index and the following indexes: CH-index, D-index, I-index and average likelihood.
Full-text previewDOI: · Available from: unipd.it
Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.
Conference Paper: Process diagnosis via electrical-wafer-sorting maps classification[Show abstract] [Hide abstract]
ABSTRACT: The commonality analysis is a proven tool for fault detection in semiconductor manufacturing. This methodology extracts subsets of production lots from all the available data. Then, data mining techniques are used only on the selected data. This approach loses part of the available information and does not discriminate among the lots. The new methodology performance the automatic classification of the electrical wafer test maps in order to identify the classes of failure present in the production lots. Subsequently, the proposed procedure uses the process history of each wafer to create a list of the root cause candidates. This methodology is the core of the software tool ACID which is currently used for process diagnosis at the Agrate site of the ST Microelectronics. A real analysis is presented.
- [Show abstract] [Hide abstract]
ABSTRACT: Process diagnosis is critical to the semiconductor industry. Early detection of faulty process steps can dramatically reduce business losses and accelerate the development cycle of new products. This article proposes a novel methodology called AC/ID (automatic classification/interactive diagnosis) that can isolate the root causes of yield loss by combining end-of-line tests (such as wafer sort test) with process history information. The authors have automated the AC/ID methodology in a software tool, ACID (Automatic Classification for Interactive Design), which is fully operational at industrial production sites.