Methodology of analysis of unseparated mixtures: Interval estimates of the total concentration of similar analytes

Abstract

Determination of the total concentration (c_Σ) of similar analytes recalculated to a standard substance X_st is a widely used but metrologically incorrect procedure leading to a high uncertainty of the results of analysis. An algorithm is proposed for the interval estimation of c_Σ without recalculating to the concentration of X_st. The algorithm takes into account different sensitivities of the determination of analytes of the studied type. No information on the nature and ratio of analytes in the sample is used. The width of the range of possible values of c_Σ objectively characterizes the systematic component of the error, it is independent of the choice of X_st, exceeds the width of traditional confidence intervals, and becomes zero at equal sensitivities of the determination of all analytes in a group. The accuracy of interval estimates is supported by the analysis of model mixtures of different types using several methods of measuring the total signal.