Peak distortion, data sampling errors and the integrator in the measurement of very narrow chromatographic peaks

As more chromatographers consider using the techniques of fast chromatograp hy or capillary electrochromatography, the instrumentation problems of gene rating and measuring these very narrow peaks is re-appraised. In particular the general view that 20-30 samples per peak is sufficient data to measure a peak is shown to be true under limited circumstances. A reworking of num erical integration theory and calculation of the errors of Newton Cotes rul es when applied to chromatographic peaks, shows that asymmetry creates more subtle mischief: as many as 350 samples/peak may be needed to achieve 0.1% accuracy. Specialist, low time constant units are required to generate nar row peaks and a new breed of fast sampling data processor is required to me asure them. As peaks narrow, it is increasingly important that data process ors help analysts to identify data under-sampling by reporting peak asymmet ry, actual sampling frequency las opposed to that initially programmed) and number of measured samples/peak as a combined validation diagnostic. Final ly the article considers the lack of development of new deconvolution based data processors and points towards the lack of information inside flame io nisation and ultraviolet absorbance detectors. If new data processors were to become available for information rich detectors, their benefits: improve d accuracy, precision and greater confidence in results, would have to be w eighed against the costs of adopting them, re-working of analysis methods a nd retraining staff. Many production laboratories would find it uneconomica l and would stay with the old methods. (C) 1999 Published by Elsevier Scien ce B.V. All rights reserved.