Quantitative analysis of sulfur functional groups in natural organic matter by XANES spectroscopy

Abstract

Two new approaches to quantify sulfur functionalities in natural organic matter from S K-edge XANES spectroscopy are presented. In the first, the K-edge spectrum is decomposed into Gaussian and two arctangent functions, as in the usual Gaussian curve fitting (GCF) method, but the applicability of the model is improved by a rigorous simulation procedure that constrains the model-fit to converge toward chemically and physically realistic values. Fractions of each type of functionality are obtained after spectral decomposition by correcting Gaussian areas for the change in X-ray absorption cross-section with increasing oxidation state. This correction is made using published calibration curves and a new curve obtained in this study. Calibration-induced errors, inherent to the choice of a particular curve, are typically lower than 5% of total sulfur for oxidized species (e.g., sulfate), may reach 10% for organic reduced sulfur, and may be as high as 30–40% for inorganic reduced sulfur. A generic curve, which reduces the calibration-induced uncertainty by a factor of two on data collected to avoid X-ray overabsorption, is derived.In the second analytical scheme, the K-edge spectrum is partitioned into a weighted sum of component species, as in the usual linear combination fitting (LCF) method, but is fit to an extended database of reference spectra under the constraint of non-negativity in the loadings (Combo fit). The fraction of each sulfur functionality is taken as the sum of all positive fractions of references with similar oxidation state of sulfur.The two proposed methods are applied to eight humic and fulvic acids from the International Humic Substances Society (IHSS). The nature and fractions of sulfur functionalities obtained by the two analytical approaches are consistent with each other. The accuracy of the derived values, expressed as the difference in values of a fraction obtained on the same material by the two independent methods, is on average 4.5±3.0% of total sulfur for exocyclic reduced sulfur, 4.1±2.1% for heterocyclic reduced sulfur, and 1.6±1.4% for sulfate. Total reduced sulfur has a better accuracy of estimation (2.4±1.6%) than either exocyclic and heterocyclic sulfur, because the errors on the two reduced pools have opposite sign. Experimental difficulties and uncertainties of the results associated with the analysis of concentrated and heterogeneous samples are discussed. The spectra of the IHSS materials and the reference compounds are made available as an open source for interlaboratory testing.