Application of Inverted Multivariate Calibrations to Determine the Total Content of Phenols

V. I. Vershinin,L. S. Bazhenova
DOI: https://doi.org/10.1134/s0020168524700109
2024-07-20
Inorganic Materials
Abstract:The total content ( c Σ ) of toxic phenols in waters is usually determined by methods involving the introduction of a group reagent, measurement of the generalized signal ( A Σ ) at a selected wavelength, and estimation of c Σ in terms of C 6 H 5 OH. The use of diazotized sulfanilic acid as a group reagent allows one to determine c Σ with errors not exceeding 30 rel %. Further reduction of errors is possible with transition to multiple-wavelength measurements and multivariate calibrations, but for the determination of phenolic toxicants, these techniques have not been used before. To test this possibility, model mixtures (colored aqueous solutions) were prepared, simultaneously containing up to five different phenols with their total concentration from 15 to 70 μmol/L. The generalized signals were measured at m wavelengths in the UV region of the spectrum 10 min after mixing the solutions. Inverted multivariate calibrations were constructed on the basis of the A Σ values of n mixtures of the same type that formed the training set. Under optimized conditions ( m = 7, n = 10), systematic errors (δ c ) in determining the amount of phenols in mixtures from the test sample did not exceed 13 rel %, which is half as much as when recalculating the signal to a standard substance. Obviously, it is advisable to use multivariate calibrations for a generalized assessment of phenolic pollution of water bodies instead of calculating total indices. However, in cases where the samples contained phenols that were not taken into account when constructing the calibration, an increase in systematic errors was observed, reaching as high as 80 rel % (in absolute value). Therefore, to use multivariate calibrations in hydrochemical analysis, a preliminary study of the qualitative composition of phenolic mixtures in waters of different types and taking into account the expected composition of samples when forming a training set are necessary.
materials science, multidisciplinary
What problem does this paper attempt to address?