The spike is caused by a lot of charge generated by the undiffracted, and very intense electron beam, which is focussed on just a few pixels. The charge simply overflows into neighbouring pixels, much like water overflows into neighbouring buckets - here buckets are synonymous to pixels. Without special anti-blooming electrodes, the spike can become a real problem particularily when electron diffraction data is being collected from large unit cell-crystals. Due to the smaller Ig/I0-ratio for these crystals - a direct consequence of the larger unit cell - many more electrons are required to obtain the necessary counting statistics for each reflection. Because the program that determines the lattice vectors does this using a real-space auto-correlation function, the spike must be removed computationally. Starting from the center of the pattern, all pixels above 100 counts, are replaced with zero. By comparing pairs of pixels - first 1 pixel apart and later more - for this procedure, the program decides that all pixels have been reset, when the final number of reset pixels remains stable on a next iteration.
Before AUTOZERO |
After AUTOZERO |