Frequency limits of 1/f noise

The concept that universal 1/fnoise is caused by defect migration is shown to be supported by experimental results and further theoretical development. Migrating defects are bombarded by conduction electrons and are driven against adjacent lattice atoms forming an impacted mass. It is predicted that a time constant, _s , defined as the average time taken for a single non-impacted defect to impact against an adjacent atom, would consequently vary with the externally applied current, I₀ . This would then, in turn, cause the high-frequency cut-off point of 1/fnoise, f_s = _s ^-1 , to vary with I₀ . For the case of a carbon conductor not only was this observation experimentally verified but also numerous new properties of 1/fnoise were discovered. Among these were the following. (1) The curve of f_s versus I₀exhibited resonances. (2) It was found that the more rigid the lattice was the closer the noise was to an exact 1/fresponse with a value fitting the Hooge empirical formula. (3) The noise spectrum above f_s was found to behave as f^{2( -1)}with an attenuating parameter = I /(I₀ +I ) where I ² R₀is the thermal energy flowing in the direction of the current as given by the Stefan-Boltzmann law. (4) A zero-current (I₀= 0) noise source was found with an amplitude given by S ( ) = (m_- /m₊ )^1/2 (I ² /N₀ )f^-1 . (5) A definite low-frequency end to the observable 1/fdependence was also found as indicated by a change from a 1/fto a 1/f²frequency dependence as predicted by the theory.