In an ideal world corrosion potentials and currents would be rock steady, electrodes would corrode uniformly and all measurements would be easy. In the real world electrodes tend to corrode non-uniformly, corrosion potentials and currents vary with time.
Electrochemical noise is a term myself and Les Callow coined back in 1979. At the time we were playing with a.c. impedance measurements on corroding samples. At very low frequencies frustrating scatter would appear in the measurement and averaging would not get rid of it. Les suggested that we detect and measure the noise somehow. Electrochemical noise is 1/f noise, the power spectral density increases at low frequencies. The usual 'white' noise has a flat frequency distribution - equal power at all frequencies. The mean is zero, longer and longer averaging reduces the observed value to that mean. This does not work for 1/f noise, in fact the longer you average the more noise you get. 1/f noise is everywhere, from semiconductors to planetary orbits. Interesting links have been found between 1/f noise and the fractal nature of things.
Have a look at some of the original papers:
In electrochemistry noise can be measured either as potential noise - for example the fluctuations of the free corrosion potential of a corroding electrode, or as current noise - such as the fluctuations of the current required to polarise an electrode.
In corrosion, electrochemical noise is usually the result of non-uniform or localised dissolution. The classic example, one of the first we played with, is mild steel immersed in NaNO2 - NaCl mix solution. It is well known that at low chloride concentrations NaNO2 acts to passivate the steel surface. As the chloride concentration is increased pitting initiates, until at high chloride concentrations general corrosion takes over.
The corresponding potential noise traces look something like this (note that these figures are intended to be a typical representation only and are not real data) :
Passivity gives quite steady corrosion potential - the trace is boringly flat. Addition of chloride leads to some pit initiation, but very few of these will propagate for any length of time. The trace is that of an occasional glitch associated with a growing pit. Further increase in the chloride level causes more and more of the initiating pits to propagate and the noise level and character changes to a seemingly random trace. Finally at high chloride levels the steel surface depassivates and general corrosion occurs. This again gives a flat trace.
Current noise from the same system - measured using zero resistance ammetry on two 'identical' electrodes - gives similar outputs. If these are causally linked over a 'long' time scale, as is usually the case, then it may be possible to take a further step - the calculation of 'resistance noise'. This is taken as the rms value of the potential noise divided by the rms value of the current noise. It has been proposed that resistance noise may be used in certain cases to estimate the rate of the dissolution process, the rate being inversely proportional to the resistance noise magnitude.
Further statistical parameters may also be calculated. In particular, skewness and kurtoisis which describe the asymmetry of the fluctuations, may be used to identify specific localised corrosion phenomena and to estimate their rates.
Much of the technology was covered by a number of patents - my original patent on electrochemical potential noise (US 4575678) had now expired and since all subsequent ones were dependent on it, they are now of a dubious value. The other 'fundamental' patent, on electrochemical current noise and 'noise resistance' (US 5139627), had lapsed due to non-payment of fees. :) There are many other patents, some of which may be found here.
If you are new to this field of electrochemistry, beware. Misconceptions and fallacies abound in the published papers, most of which simply regurgitate and plagarise previous publications without any real understanding of the concepts involved. You have been warned :)