Limit of detection in analysis with ion-selective electrodes

The limit of detection in analysis with ion-selective electrodes is discussed and definitions that are based only on the deviation of an electrode's calibration from the theoretical, and take no account of the random errors of measurement, are shown to be inadequate. Equations are derived that express the limit of detection in terms of the random error of measurement and the factors determining the deviation of the electrode response from the Nernstian value, i.e., reagent blanks, solubility products and interferences. The equations enable one to predict (a) the degree of precision with which the e.m.f. has to be measured if an electrode is to attain a desired limit of detection in specified conditions or (b) whether changing the conditions might bring the desired limit of detection within reach of a given precision of measurement. Practical examples with ion-selective electrodes justify the proposed statistical treatment of limit of detection and demonstrate that the errors for electrodes operating in the non-Nernstian region are normally distributed.