The paper cites a growing literature on determining chemical parameters in a solution by curve fitting, where makeup of the solution, and hence the governing functions, are known. The purpose here is to determine these quantities from titration data for an unknown solution. The method depends on an additive form of the titration function based on a device discovered by Simms. This eliminates protonicity, and makes number of components (terms) the sole determinant of functional form. The method, then, consists in finding the number of terms in the function needed to obtain a best curve-fit. Functions with one term, two terms, etc., are fitted until indicators show that a limit has been reached. This occurs when the fitted curve falls within the scatter range of the data. The components of pure compounds, such as citric acid, are cleanly separated, but mixtures for which pK values lie close together are not resolved completely. An analysis shows that the limit of resolution for two components is given by X2ΔpK2 ≃ 4σpH, where X2 is fraction of minor component, ΔpK the pK separation, and σpH the standard deviation. While component separation is thus sharply limited, a range of useful application exists. Illustrative results are given for different types of systems. The paper describes how the analysis is organized, and outlines important algorithms. © 1979 American Chemical Society.
