Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Inline Wafer Identification Using Optical Character Recognition (OCR)

: Al-Hajjawi, S.; Hammer, T.; Haunschild, J.

Fulltext urn:nbn:de:0011-n-5486114 (575 KByte PDF)
MD5 Fingerprint: 151dffbf07722f6477c589d85fd6e435
Created on: 18.6.2019

Verlinden, P. ; WIP - Renewable Energies, München:
35th European Photovoltaic Solar Energy Conference and Exhibition 2018 : Proceedings of the international conference held in Brussels, Belgium, 24 September-28 September 2018; DVD-ROM
München: WIP, 2018
ISBN: 978-3-936338-50-8
ISBN: 3-936338-50-7
European Photovoltaic Solar Energy Conference and Exhibition (EU PVSEC) <35, 2018, Brussels>
Conference Paper, Electronic Publication
Fraunhofer ISE ()
Inline-Wafer-/Prozessanalytik und Bildverarbeitung; Photovoltaik; Silicium-Photovoltaik; Charakterisierung von Prozess- und Silicium-Materialien; characterization; software; tracking

In order to connect the information from incoming inspection and IV-characterization or even crystallization information, wafers need to be identified and tracked. Identifying each wafer helps to provide a better analysis of the influence or failures of production processing steps on each individual wafer rather than a general statistical analysis approach of broader number of wafers. Nowadays, several wafer identification methods are available to simplify the tracking of wafers. In this work, we present an approach of exploiting the already integrated cameras within the inline characterisation equipment for automatic detection and recognition of wafer IDs. To do so, Optical Character Recognition (OCR) is applied through adapting Tesseract open source software with optimised preprocessed images. Our approach presents an identifier of wafers which don’t have BSCs or DMCs but alphanumerical wafer IDs. Before using our algorithm, (alpha)numerical wafer labeling is required. For numerical wafer IDs in best cases, an average ID recognition rate of 99% was achieved. For alphanumerical wafer IDs, automatic software correction is highly relevant to solve the confusion of software when switching from numbers to letters and vice versa. A comparison between wafer ID reading techniques revealed that the best average wafer ID reading accuracy of 98% was obtained using brick slice code reader. The data matrix code reader revealed an average wafer ID accuracy of 93%. Finally, our OCR-based ID reader software had an average wafer ID reading accuracy of 85% for alphanumeric IDs.