An Evaluation of an Automated Detection Algorithm to Count Defects Present in X-Ray Topographical Images of SiC Wafers
- PDF / 404,712 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 73 Downloads / 193 Views
0994-F11-13
An Evaluation of an Automated Detection Algorithm to Count Defects Present in X-Ray Topographical Images of SiC Wafers Ian C. Brazil1, Patrick J. McNally1, Lisa O'Reilly1, Andreas Danilewsky2, Turkka O. Tuomi3, Aapo Lankinen3, Antti S‰yn‰tjoki3, Rolf Simon4, Stanislav Soloviev5, Larry B. Rowland5, and Peter M. Sandvik5 1 RINCE, Dublin City University, Dublin, 9, Ireland 2 Kristallographisches Institute, Universit‰t Freiburg, Freiburg, Germany 3 Helsinki University of Technology, Helsinki, Finland 4 Institut f¸r Synchrotronstrahlung (ISS), Karlsruhe, Germany 5 GE Global Research, Niskayuna, NY, 12309
ABSTRACT Full semiconductor wafer defect/dislocation characterization is difficult to implement manually. We present an analysis of an automated algorithm used to extract Threading Screw Dislocation defect data from Synchrotron White Beam X-Ray Topographical images of SiC wafers. This extraction involves a two-fold process; firstly the algorithm highlights the appropriate defect and secondly updates the counter to provide a final result of defect count. The result of the automated algorithm is compared to hand counts in all cases, thus allowing a critical analysis of the technique. Improvements to this algorithm have been made since last reported by the same authors [1], which are discussed. The analysis herein was also performed on a much larger sample of SiC wafer images than previously used by the same authors [1] allowing a better judgment of performance and critical evaluation. The algorithm is also compared with the original previous algorithm that was used [1]. The success of this methodology paves the way for a complete analysis of whole SiC wafers, which previously was extremely difficult due to image analysis inaccuracy or the bottleneck presented by manual counting. INTRODUCTION Silicon Carbide is a promising material for a variety of electronic applications. Properties [3, 8] include a large bandgap, high electrical breakdown, low chemical reactivity and a high operating temperature. Growth however is prone to defects, common defects include Micropipes, Basal Plane Dislocations (BPD), Stacking Faults (SF) and Threading Screw Dislocations (TSD) [2-4]. Synchrotron White Beam X-Ray Topography (SWXRT) can be used to analyze SiC wafers [4, 9, 10] allowing a large image to be created of a wafer by joining smaller images. In SWXRT TSDs and BPDs [6] show up as dots and lines [5]. Being non-destructive its advantages over other techniques, e.g. KOH etching [7], are obvious. However quantitative analysis (e.g. TSD count) is extremely time consuming and tedious making SWXRT unsuited to examination of large quantities/areas of wafers. It is therefore a massive advantage if an automated method can be established to perform quantitative analysis of SWXRT images. An algorithm to classify Basal Plane Dislocations was discussed in a previous paper [1]. However the TSD algorithm developed in [1] was only semi-automatic. The complete automation of this algorithm and a demonstration of its usefulness is outlined in
Data Loading...
