Abstract

NMR FID instabilities due to the most common types of magnetic field and/or RF offset fluctuations are investigated. The study explains quantitatively some of the artifacts encountered in practice and indicates methods for overcoming them.

The effect of a random, normally distributed field noise on averaged FID's is shown to be their multiplication by a well-defined weighing function whose shape is intermediate between Lorentzian and Gaussian. While integral sensitivity of the instrument is not affected, there is a line shape distortion which might be easily mistaken for field inhomogeneity.

The effects of periodic/quasi-periodic field perturbations on FID's and spectra are also investigated in detail. The results are directly applicable to the effects of type A instabilities such as mains-related field brum and ripple (including environmental pick-up). Type B instabilities whose magnitudes have a non-uniform spatial distribution require an additional averaging over individual sample voxels. The study establishes the respective averaging procedures and shows that when the normalized self-correlation functions of the field fluctuations are the same for all voxels, the final effect on averaged FID's is still a simple weighting by a particular function. The results are applicable to effective field fluctuations induced by sample spinning (HR-NMR), sample vibrations (all branches of NMR) and field gradients noise (MRI, PFG self-diffusion measurements).

Exploiting the new insights, possible methods of suppressing some of the related artifacts are proposed and tested. This includes alternative methods of signal averaging as well as post-acquisition correction techniques for traditionally averaged data.