DEDUCTIVE PREDICTION OF MEASUREMENT PRECISION FROM SIGNAL AND NOISE IN FLUOROMETRY

The aim of this paper is to examine the precision of fluorometry on th e basis of a recently proposed theory to predict the relative standard deviation (RSD) of measurements from signal and noise in an analytica l instrument. According to the theory, the instrumental baseline drift which is often formulated as l/f noise can well be approximated by a mixed random process of white noise and the Markov process. The standa rd deviations (SD), (w) over tilde, of the white noise and the SD, (m) over tilde, and auto-correlation parameter, rho, of the Markov proces s can be parametrized from the power spectral density of the blank lin e drift by the least squares curve fitting. All the parameters necessa ry for the uncertainty prediction ((w) over tilde, (m) over tilde, rho , signal domain, (k(c) + 1, k(f)) and signal area, A, over the domain) , can be determined directly and uniquely from experiments and experim ental design. No arbitrary constants are involved in the prediction th eory. The uncertainty prediction is shown to be superb over a wide con centration range of Rhodamin B in the fluorescence measurement. The in fluence of the integration domain on the precision is considered.