Noise in Submicron Devices

Abstract

The essential problem in noise theory is relating local, microscopically produced fluctuations at an interior point to the observable, fluctuations in the macroscopic terminal currents and voltages. A variety of schemes for this have been developed, and a good review can be found in van Vliet et al. (1) The oldest scheme is the “salami” method, in which the device is subdivided into slices. Each slice possesses a local noise source, and a noise voltage appears across the slice as a result. The total mean square noise voltage is the sum of the mean square noise voltages of the individual slices. This summation process ignores correlations between slice noise voltages. In more sophisticated variants, a tensor Green’s function is defined, which relates fluctuations in field at point x to fluctuations in current or carrier density at point x’. The terminal noise voltage then is an integral over the product of the Green’s function and the noise source. The terminal mean square noise voltage is then the mean of the product of two such integrals. This mean involves a two-point spatial correlation function for the noise source. Although such spatial correlations of the microscopic fluctuations do exist, they are generally ignored.