文摘
A procedure is introduced that can mitigate the deleterious effect of low-frequency noise-often termed drift-onthe precision of an analytical experiment. This procedureoffers several performance benefits over traditional designs based on the periodic measurement of standardsto diagnose and correct for variation in instrument response. Using repeated measurements of every sampleas a drift diagnostic, as opposed to requiring the periodicmeasurement of any given sample or standard, the analystcan better budget the measurement time to be devoted toeach sample, distributing it to optimize the uncertaintyof the analytical result. The drift is diagnosed from therepeated measurements, a model of the instrumentresponse drift is constructed, and the data are correctedto a "drift-free" condition. This drift-free condition allowsdata to be accumulated over long periods of time with littleor no loss in precision due to drift. More than 10-foldprecision enhancements of analytical atomic emissionresults have been observed, with no statistically significanteffects on the means. The procedure is described,performance data are presented, and matters regardingthe procedure are discussed.