Simplex Algorithm for Deep-Level Transient Spectroscopy: Simplex-DLTS
Résumé
The Nelder–Mead simplex algorithm improved by Lagarias for low-dimension functions is introduced for the first time on transient capacitance signal analysis to decrease noise sensitivity and increase the resolution of deep-level transient spectroscopy method (DLTS). The application of the resolution figure of merit in predetermining the success of DLTS analyses' ability to resolve defects has shown that the performances of the simplex-DLTS developed in this work are significantly better than those of matrix pencil-DLTS method (MP-DLTS) published in 1998. Comparing experimentally the two methods in analyzing the signals generated by the same 150 keV germanium preamorphized p+n samples, we found that the simplex-DLTS method detects six defects in which some show very close activation energy values, while the MP-DLTS method found only two defects. Hence, the superiority of the simplex-DLTS method is proved; it reveals deep levels in which the emission process is carried out in the same temperature range.