Shevchenko V. Construction of color scales, metrological tools for total and separate determination of analytes with the use of solid phase group reagents.

The comparative analysis of suitability of the two methods of assessing the quality of color scales for visual colorimetry was carried out in the thesis, those are the method of colorimetry and the statistics of observations. For this purpose the color scales with the coefficient of concentration geometric progression q=1.5, 1.62, 2, 3 were prepared for the seven test systems. It was found that the criterion of visual color differentiation Е*=10 is not applicable for the modified PPU and RIP, it cannot be used for colored transparent gelatin films. The visual method should be given preference to when choosing the best color scales for visual colorimetry. It has been experimentally proved in the work that statistical criteria (3s, t), that had been applied earlier only in spectrophotometry of solutions, can be used to identify deviations starting from additivity of total analytical signal in determining the sum of metals using solid reagents. No significant differences from additivity were detected for the analytical signals from sum of metals in equal molar concentrations. The hi2-criterion was proposed to be used in order to test additivity of total color difference of the mixture of analytes, immobilized on non-transparent sorbents (PPU, RIP). The possibility to reveal deviations from additivity of the summary signal, using the empirical criterion, which is based on the visual estimation of defining measures of the separate analytes or their sum, was demonstrated in visual colorimetry. It is demonstrated that the method of the diffuse reflectance spectroscopy with the use of indicated reagent paper RIP-Metal-Test-III using the chemometric algorithms (the methods MLR or the method PLS) for differentiation of overlapping signals, allows to determine the content of the three pairs of ions of metals (Cu(II)+Zn(II), Co(II)+Ni(II), Рb(II)+Mn(II)) or their mixtures with a relative error less than 30%.