A PRECISION NOISE MEASUREMENT AND ANALYSIS METHOD USED TO ESTIMATE RELIABILITY OF SEMICONDUCTOR-DEVICES

In this paper, some problems with previous ultra-low noise measurement methods have been discussed, then a double-channel preamplifier cross -spectrum measurement method has been adopted, different from the prev ious cross-correlation method [A. van der Ziel, Noise: Sources, Charac terization, Measurement, p. 54. Prentice-Hall, Englewood Cliffs, NJ (1 970), L. Stor, Experimental techniques in noise measurement with speci al emphasis on precision measurement, Proc. 10th Int. Conf: on Noise i n Physical Systems, pp. 551-560. Budapest, Hungary (1989)] in that an average periodogram using a windowing procedure has been performed. Th e theoretical analysis shows that the expected value of the cross-spec trum is incoherent with background noise and zero-drift from the pream plifier and power supply system, the average periodogram can decrease the variance of the periodogram and the additional bias of the cross-s pectrum periodogram. Experimental results demonstrate that if the equi valent input noise of measuremental set-up is two orders of magnitude lower than the noise of each preamplifier, then an ultra low noise spe ctrum can be measured accurately, the low limit is about 0.1 nV/root H z at 1 kHz, which is 20 dB lower than the noise of each preamplifier. The thermal noise of a small resistance and the shot noise of a diode under forward conditions have been measured, the experimental results are in good agreement with the theoretical value, this means that this method is feasible and accurate for an ultra-low noise spectrum measu rement. Finally, the noise spectrum analysis procedure based on the cu rve fitting method has been presented, which ensures that we obtain an accuracy value of three noise components in the semiconductor, i.e. 1 /f noise, white noise and g-r noise. This noise spectrum analysis meth od is a useful tool for investigation of noise mechanism, the diagnosi s of defects in semiconductors and reliability estimation. (C) 1997 El sevier Science Ltd.