Organic complexation of rare earth elements in natural waters: evaluating model calculations from ultrafiltration data

International audienceThe Stockholm Humic Model (SHM) and Humic Ion-Binding Models V and VI were compared for their ability to predict the role of dissolved organic matter (DOM) in the speciation of rare earth elements (REE) in natural waters. Unlike Models V and VI, SHM is part of a speciation code that also allows us to consider dissolution/precipitation, sorption/desorption and oxidation/reduction reactions. In this context, it is particularly interesting to test the performance of SHM. The REE specific equilibrium constants required by the speciation models were estimated using linear free-energy relationships (LFER) between the first hydrolysis constants and the stability constants for REE complexation with lactic and acetic acid. Three datasets were used for the purpose of comparison: (i) World Average River Water (Dissolved Organic Carbon (DOC) = 5 mg L−1), previously investigated using Model V, was reinvestigated using SHM and Model VI; (ii) two natural organic-rich waters (DOC = 18–24 mg L−1), whose REE speciation has already been determined with both Model V and ultrafiltration studies, were also reinvestigated using SHM and Model VI; finally, (iii) new ultrafiltration experiments were carried out on samples of circumneutral-pH (pH 6.2–7.1), organic-rich (DOC = 7–20 mg L−1) groundwaters from the Kervidy-Naizin and Petit-Hermitage catchments, western France. The results were then compared with speciation predictions provided by Model VI and SHM, successively. When applied to World Average River Water, both Model VI and SHM yield comparable results, confirming the earlier finding that a large fraction of the dissolved REE in rivers occurs as organic complexes This implies that the two models are equally valid for calculating REE speciation in low-DOC waters at circumneutral-pH. The two models also successfully reproduced ultrafiltration results obtained for DOC-rich acidic groundwaters and river waters. By contrast, the two models yielded different results when compared to newly obtained ultrafiltration results for DOC-rich (DOC > 7 mg L−1) groundwaters at circumneutral-pH, with Model VI predictions being closer to the ultrafiltration data than SHM. Sensitivity analysis indicates that the “active DOM parameter” (i.e., the proportion of DOC that can effectively complex with REE) is a key parameter for both Model VI and SHM. However, a survey of ultrafiltration results allows the “active DOM parameter” to be precisely determined for the newly ultrafiltered waters studied here. Thus, the observed discrepancy between SHM predictions and ultrafiltration results cannot be explained by the use of inappropriate “active DOM parameter” values in this model. Save this unexplained discrepancy, the results presented in this study demonstrate that both Model VI and SHM can provide reliable estimates of REE speciation in organic-rich waters. However, it is essential to know the proportion of DOM that can actively complex REE before running these two speciation models