ABSTRACT
Determination of deuterium (D) concentration in tap water and urine is demonstrated to average error approximately 0.5% (coefficient of variation) using a 400-MHz nuclear magnetic resonance (NMR) instrument. Time domain data are obtained using 0.75-ml samples in a broadband probe. Peak areas derived from absorption and magnitude mode Fourier transforms and least-squares fitting of the time domain free induction decays (FIDs) are all investigated as means to derive D concentrations from raw data. Least-squares fits using a sum of exponentially damped sinusoids, which yields estimates for the amplitude, damping constant (relaxation time), wavelength (resulting from mixing of precession and reference frequencies), and phase for each of the two components, are shown to provide the best precision for unfiltered FID. Amplitudes are proportional to the number of spins at each frequency, as analysis of untreated urine from doubly labeled water experiments yield highly linear washout data (r2 > 0.99998) for baseline-corrected log-transformed data. The procedure is general and should extend to other body fluids with minimal modifications. These data show that least-squares curve fitting is the most precise method of quantitative NMR data reduction for a wide range of experimental conditions.
