Deductive Prediction of Measurement Precision from Signal and Noise in Fluorometry

Abstract

The aim of this paper is to examine the precision of fluorometry on the basis of a recently proposed theory to predict the relative standard deviation (RSD) of measurements from signal and noise in an analytical instrument. According to the theory, the instrumental baseline drift which is often formulated as 1/f noise can well be approximated by a mixed random process of white noise and the Markov process. The standard deviations (SD), w, of the white noise and the SD, m, and auto-correlation parameter, ρ, of the Markov process can be parametrized from the power spectral density of the blank line drift by the least squares curve fitting. All the parameters necessary for the uncertainty prediction (w, m, ρ, signal domain, (kc+1, kf) and signal area, A, over the domain), can be determined directly and uniquely from experiments and experimental design. No arbitrary constants are involved in the prediction theory. The uncertainty prediction is shown to be superb over a wide concentration range of Rhodamin B in the fluorescence measurement. The influence of the integration domain on the precision is considered.