Determination of the total content of similar substances using integrated indices or multivariate calibrations at a high intragroup selectivity of signals

Abstract

The total content (cs ) of similar organic compounds is usually determined without their separation by measuring their generalized spectral signals at one or more analytical wavelengths (AW). The resulted es­ timates of cs are approximately adequate if all the sensitivity coefficients of the substances (components of the desired group) are the values of the same order of magnitude. The possibility of a correct assessment of cs with a strongly pronounced intragroup selectivity of signals has not been previously tested. Model hexane solutions of the known composition simultaneously containing mono-, bi-, and tricyclic arenes at сs value about 10-4 mol/dm3were analyzed to verify this variant of the group analysis. At a fixed wavelength, the values of molar absorptivity of arenes with different numbers of rings differ by 2 - 3 orders of magni­tude. Two variants of group analysis were compared: 1) measurement of the generalized signal of arenes at 260 nm with subsequent calculation of the result using univariate calibration in terms of naphthalene or anthracene; 2) measurement of generalized signals for several m wavelengths in the spectral region of 250 - 290 nm with subsequent calculation of the result using the inverted multivariate calibration. The first method (calculation of the integrated index) led to large systematic errors, sometimes exceeding 100 % rel. (in modulus) which appeared insensitive both to the wavelength and to the nature of the stan­dard substance. The second method provided more correct results and even at m = 11 the errors of group analysis did not exceed 10 %. Nevertheless, the errors in the estimation of csdramatically increased if the sample contains components (arenes) of the desired group that were not taken into account when con­structing the inverted calibration. It is shown that with a strongly pronounced intragroup selectivity of signals, group analysis can and should be carried out using inverted multivariate calibrations.