Algorithms for clustering data
Algorithms for clustering data
| Hi-index | 2.88 |
The practicality of plasma etching, combined with low temperature and directional process capabilities make it an integral part of the IC manufacturing process. A significant cause of damage to wafers during plasma processing is arcing damage. Plasma arcing damage results in large pits and non-uniformities on the wafer surface which can lead to wafer breakage and high yield losses. Thus a non-destructive wafer damage metrology is crucial to the understanding of wafer failure mechanisms. We report on the successful use of a combined suite of non-destructive metrology techniques to locate the arc damage sites and examine the physical processes which have occurred as a result of the damage. These consist of 3D X-ray diffraction imaging (3D-XRDI), micro-Raman spectroscopy (@mRS), and scanning electron microscopy (SEM). In the case of the two examples examined in this study the plasma induced damage on the wafer surface appears as regions of damage ranging from 100@mm to 1000@mm in diameter. 3D-XRDI shows that the strain fields propagate out from the damage site in all directions, with the damage penetrating up to @? of the way through the substrate. K-means clustering and false colouring algorithms are used to highlight the regions of interest in 3D-XRDI, and to enhance the analysis process. Sectioning of the 3D images has enabled non-destructive imaging of the internal damage in the samples at any location. Micro-Raman spectroscopy results indicate the presence of both crystalline and amorphous silicon. Strain measurements at the damage site show tensile strains as high as 500MPa in certain situations, with strain levels increasing from the surface towards the bottom of the dislocation cell structures, which can be distinguished in the synchrotron X-ray topographs. 3D-XRDI and @mRS results are in close correlation, proving the potential for 3D-XRDI as non-contact, non-destructive metrology particularly suited to these problems.