Accurate determination of material parameters, such as carrier lifetimes and defect activation energy, is a significant problem in the technology of infrared detectors. Among many different techniques, using the time resolved photoluminescence spectroscopy allows to determine the narrow energy gap materials, as well as their time dynamics. In this technique, it is possible to observe time dynamics of all processes in the measured sample as in a streak camera. In this article, the signal processing for the above technique for Hg1-xCdxTe with a composition x of about 0.3 which plays an extremely important role in the mid-infrared is presented. Machine learning algorithms based on the independent components analysis were used to determine components of the analyzed data series. Two different filtering techniques were investigated. In the article, it is shown how to reduce noise using the independent components analysis and what are the advantages, as well as disadvantages, of selected methods of the independent components analysis filtering. The proposed method might allow to distinguish, based on the analysis of photoluminescence spectra, the location of typical defect levels in HgCdTe described in the literature.
The temperature dependence of photoluminescence spectra has been studied for the HgCdTe epilayer. At low temperatures, the signal has plenty of band-tail states and shallow/deep defects which makes it difficult to evaluate the material bandgap. In most of the published reports, the photoluminescence spectrum containing multiple peaks is analyzed using a Gaussian fit to a particular peak. However, the determination of the peak position deviates from the energy gap value. Consequently, it may seem that a blue shift with increasing temperature becomes apparent. In our approach, the main peak was fitted with the expression proportional to the product of the joint density of states and the Boltzmann distribution function. The energy gap determined on this basis coincides in the entire temperature range with the theoretical Hansen dependence for the assumed Cd molar composition of the active layer. In addition, the result coincides well with the bandgap energy determined on the basis of the cut-off wavelength at which the detector response drops to 50% of the peak value.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.