Low frequency noise and DLTS as semiconductor device characterization tools

Abstract

The technique of low frequency noise vs temperature measurements is shown to be a powerful diagnostic technique for determining Generation Recombination (G.R.) trapping parameters in MOSFETS. From computer controlled measurements of low frequency noise vs temperature, the trapping parameters are extracted in a manner similar to that of Deep Level Transient Spectroscopy (DLTS). The trapping parameters are also extracted by curve fitting of the low frequency noise vs temperature curves. These noise-determined trapping parameters are compared with those measured by DLTS. The agreement between parameters determined by the spot frequency noise Arrhenius plot with those determined by DLTS is close, while the agreement between those determined by noise curve fitting is reasonable for noise peaks near room temperature, but becomes poor for low temperature noise peaks. We have found that the low frequency noise vs temperature technique appears to show higher sensitivity than our DLTS measurements. By the use of PECVD silicon nitride as a passivation material, the sensitivity to measure G.R. traps by the low frequency noise vs temperature technique can be increased. This improved sensitivity is due to the reduction of interface state noise. The PECVD silicon nitride passivation had no effect on reducing the G.R. noise peaks thus indicating that these traps are probably located in the bulk space-charge region.