Molecular features of uranium-binding natural organic matter in a riparian wetland determined by ultrahigh resolution mass spectrometry

Abstract

Tims Branch riparian wetland located in South Carolina, USA has immobilized 94 % of the U released >50 years ago from a nuclear fuel fabrication facility. Sediment organic matter (OM) has been shown to play an important role in immobilizing U. Yet, uranium-OM-mineral interactions at the molecular scale have never been investigated at ambient concentrations. The objectives of this study were to extract, purify, and concentrate U-bound sediment OM along the stream water pathway and perform molecular characterization using Fourier transform ion cyclotron resonance mass spectrometry (FTICRMS). Out of 9614 identified formulas, 715 contained U. These U-containing formulas were enriched with Fe, N, and/or S compared to the total OM. Lignin-like and protein-like molecules accounted for 40 % and 19 % of the U-containing formulas, respectively. Phosphorus-containing formulas were found to exert an insignificant influence on complexing U. U-containing formulas in the ‘mobile’ (groundwater extractable) OM fraction had lower (reduced) nominal oxidation states of carbon (NOSC); and less aromatic moieties than OM recovered from the ‘immobile’ (sodium pyrophosphate extractable) OM fraction. U-containing formulas in the redox interfacial zones (stream banks) compared to those in nearby up-slope zones tended to have smaller molecular weights; lower NOSC; higher contents of COO and/or CONO functional groups; and higher abundance of Fe-containing formulas. Fe was present in 38 % of the U-containing formulas but only 20 % of the total OM formulas. It is postulated that Fe played an important role in stabilizing the structure of sedimentary OM, especially U-containing compounds. The identification for the first time of hundreds of Fe-U-OM formulas demonstrates the complexity of such system is much greater than commonly believed and numerically predicting U binding behavior in OM-rich systems may require greater use of statistical or artificial intelligence approaches rather than deterministic approaches limited to measuring metal complexation with well-defined individual analogue organic ligands.