Volume versus surface origin of1/fnoise in metals

We have examined the question of whether excess low-frequency (1/f) noise in metal films arises in the bulk or at the sample surface. We measured the noise in continuous Cr films with 80 A{}2600 A{}, where t is the film thickness. Establishing that the relative noise intensity S_ρ(f)/ρ²{∝}1/t supports the hypothesis that the noise is produced throughout the bulk; here, f is the frequency, {ρ} is the resistivity, and S_ρ(f) is the power spectral density of fluctuations in {ρ}. By considering a model of noise from the surface shunted by an underlying noise-free conductor volume, we devised a quantitative test which we used to refute the possibility that the noise arises only at the surface of our metal films.