Application of near-infrared spectroscopy to the measurement of inorganic acids and ions in aqueous solution

During the past ten years, near-infrared spectroscopy (NIRS) has become a popular means for analyzing chemical systems. However, relatively few studies have demonstrated the ability of NIRS to characterize and quantify aqueous inorganic analytes. This research evaluates short-wavelength NIRS (700-1100 nm) for such analyses, using the test systems of (1) hydrofluoric acid and related species in solution and (2) ions in simple salt mixtures.;In the first study, the simultaneous measurement of hydrofluoric acid, fluoride, bifluoride, hydronium, and hydroxide ions is demonstrated over broad ranges of concentration and pH. Equilibrium calculations show that intercorrelations among the species' concentrations are reduced by employing a titration approach to characterize the hydrofluoric acid system. The model-based regression method of chemical regression provides a means for determining optimum values of the systems' equilibrium constants, as well as estimates of the pure-component concentrations and spectra. However, isolation and interpretation of the pure-component spectra is complicated by the bulk water's contribution to the spectra, was well as the charge-balance constraint. Calibration and prediction errors below 10 mM were obtained for all five species.;The second application illustrates the determination of ions in pure and mixed salt solutions. Here, the ions are sensed entirely by their perturbation of the water spectrum. Although it is impossible to obtain the spectrum of an individual ion, the difference spectra of selected alkali-halide salts demonstrate that each of the ions studied has a unique signature on the water spectrum. Quantitative determinations of ions in pure and mixed solutions of sodium chloride and sodium iodide using classical least squares calibration indicate that a two-component description of a salt in water fails to account for the solutions' spectral behavior. Superior results are obtained by employing chemical regression and ion-pairing equilibrium models to rationalize the unexplained spectral variance. For pure solutions, the relative quantitation errors were less than 0.5%, yet the corresponding values for the binary solutions were approximately five to ten times higher due to the severe overlap of the hydrated ions' spectra.