Energy Scales for Noise Processes in Metals

Abstract

We ascribe the excess noise in metals to random fluctuations with thermally activated characteristic times. From the known temperature dependence of the magnitude of the noise in Ag we infer a narrow distribution of activation energies that peaks in the vicinity of 1 eV. Frequency power laws close to ${f}^{\ensuremath{-}1}$ are deduced and the small variations with temperature are consistent with observations. The predicted spectrum, like the observed one, has an apparent power-law nature throughout the frequency region accessible to experiment.