Concerning the Nature of the 1/f Noise.
A.Ya. Shul'man
Zh. Exp. Theor. Fiz. 81, 784-797 (1981)
[Sov. Phys. JETP, 54(2), 420-426 (1981)]
Abstract
It is shown that the spectra with frequency dependence of the form 1/f, which
are observed in measurements of low-frequency noise, can be the result of
action on the spectrum analyzer of the stationary infralow-frequency noise (ILF
noise), whose correlation time is much longer that the analyzer reciprocal
bandwidth. Expressions are obtained for the instrumental function of the
spectrum analyzer used in analog and digital methods of spectral measurements,
and for the output signal when IFL noise acts on the analyzer. The measured
signal is proportional in this case to the mean square and not to the spectral
density of the noise, and does not reflect in any way the dependence of the
latter on the frequency. The reliability of the available experimental data on
the 1/f noise are discussed. It is concluded that it is impossible at present
to draw a definite conclusion that fluctuations with a spectrum of the form 1/f
exist really in nature. The experimental results of Voss and Clarke [Phys. Rev.
v.B13, 556 (1976)] on 1/f noise of metallic films are analyzed and a
quantitative and qualitative explanation, connected with equilibrium
fluctuations of the temperature, is proposed within framework of the theory of
ILF noise. A theoretical expression is also obtained for the Hooge constant and
it is shown that it depends not only on the parametes of the material and on
the temperature, but also on the instrumental function of the analyzer.
