Analysis of noisy spectra of electron paramagnetic resonance

The paper compares various methods for analyzing the signal of noisy EPR spectra, including those that allow increasing the signal-to-noise ratio. The EPR spectra were modeled (by computer calculation) or adjusted (in the case of an experiment) using the Tsallis function, which made it possible to smoothly adjust the shape of the line. It is shown that the method of minimizing the error function at low noise determines the parameters of the spectrum with high accuracy, but at high noise is inferior to the maximum likelihood method. It was found that the procedure of "smoothing" the spectra increases the signal-to-noise ratio, but practically does not improve the accuracy of determining the signal parameters. Measurements and analysis of the intrinsic noise of the Varian E-4 X-band EPR spectrometer, necessary for the application of the maximum likelihood method, revealed the Gaussian distribution for experimental noise.