Chemical-shift referencing and resolution stability in methanol:water gradient LC-NMR

Solvent-gradient LC-NMR can generate wide ranges of solvent conditions during an experiment. This complicates the chemical-shift referencing of the resulting NMR data. This problem and other experimental issues are evaluated here for LC-NMR in methanol:water, using solvent mixtures running from 0% to 100% methanol. It is shown that the use of the methanol methyl signal is superior to the use of the water signal in any form (either the 1H or the 2H signal), either as a secondary reference, as a signal for shimming, or as a lock signal. Also shown are the limitations of the referencing methods and other experimental parameters, and the limitations of the solvent-gradient ramp parameters, primarily as they affect lineshapes.