Uncertainty measurement for automated macro program-processed quantitative proton NMR spectra

Abstract

The evaluation of a fully automated quantitative proton nuclear magnetic resonance spectroscopy (qNMR) processing program, including the determination of its processing uncertainty, and the calculations of the combined uncertainty of the qNMR result, is presented with details on the use of a trimmed purity average. Quantitative NMR spectra (1359) were collected over a 4-month period on various concentrations of pseudoephedrine HCl dissolved in D₂O (0.0610 to 93.60 mg/mL) containing maleic acid (the internal standard) to yield signal-to-noise ratios ranging from 3 to 72,000 for analyte integral regions. The resulting 5436 purities exhibited a normal distribution about the best estimate of the true value. The median absolute deviation (MAD) statistical method was used to obtain a model of uncertainty relative to the signal-to-noise of the analyte’s integral peaks. The model was then tested using different concentrations of known purity chloroquine diphosphate. qNMR results of numerous illicit heroin HCl samples were compared to those obtained by capillary electrophoresis.