Abstract
A low-field medium-resolution NMR spectrometer, with an operating frequency of 29 MHz for 1H, has been developed for use in process analysis. The information that is attainable at this field strength has been investigated with the use of samples from a wide range of applications including the qualitative monitoring of a benzene production process, the estimation of the average ethoxy chain length in nonyl phenol ethoxylates, and the determination of strong acid concentrations. At low field, the spectra of multi-component samples often exhibit overlapping peaks, so multivariate calibration methods are necessary for quantitative analysis. Simulated data have been used to illustrate the effects of signal-to-noise ratio (SNR), resolution, and line shape on the predictive ability of partial least-squares (PLS) calibration routines, with SNR shown to have a greater impact than resolution on accuracy and precision. PLS calibration models were applied successfully in the analysis of samples from a methacrylamide process and a simulated esterification reaction. The accuracy (<5%) and precision (<3%) were generally good for components at concentrations >1-5% w/w.
Original language | English |
---|---|
Pages (from-to) | 75-82 |
Number of pages | 8 |
Journal | Applied Spectroscopy |
Volume | 56 |
Issue number | 1 |
DOIs | |
Publication status | Published - 2002 |
Keywords
- NMR spectrometry
- process analysis
- partial least-squares