G−R noise experiments in the semiconductor regime of silicon double injection diodes

Abstract

SILICON is an almost ideal material to produce double injection devices with a behaviour close to predictions from a highly idealized theory. The material, from which the devices reported here are made, is high resistive, so that thermal equilibrium carriers are almost negligible at medium injection levels, at which levels trapping is negligible as well. There are no basic problems in producing both an electron injecting and a hole injecting contact. Junction effects like the voltage drop across the junctions and diffusion tails nearby them can be made small compared to the gross bulk mode of operation. We produced a large number of typically 4-10 mm long devices with about 3-l 0 mm* cross section from silicon with resistivities between 5 and 200 kncm at room temperature. Most of them were able to reach the semiconductor regime at voltages of 10 V and at current levels between 2 FA and 100 PA. The only major experimental problem was to reduce any low frequency excess noise to low enough levels. This problem could not be solved completely. Our project was to produce devices with (in terms of theory) simple behaviour, both a.c. and d.c.-wise, to measure g-r noise spectra on them and to compare the results to recent theoretical predictions. Thus, a very short account of the theory is given, then a few typical measurements are presented and discussed. Theoretical considerations: A double injection device operating in the Lampert semiconductor regime [a LSD], neglecting junction effects, has the following I-V characteristic[ l]