Effect of Colloids on the Calculation of Distribution Coefficients in Studies of Metal Sorption on Organic Matter

For proper calculation of distribution coefficients in metal sorption studies it is essential to fully separate dissolved from particulate metal. This is typically done via membrane filtration whereby the cutoff between dissolved and particulate fractions is somewhat arbitrarily set at 0.22 μm, dictated by available pore sizes. However, the pH-dependent formation of colloid-bound metal, able to bypass this procedure, can lead to analytical artifacts by adding an unknown and variable amount of particulate metal to the mechanically defined ‘dissolved’ pool, especially for organic substrates. We investigated this phenomenon in the context of yttrium and rare earth element (YREE) sorption on the marine macroalga Ulva lactuca (sea lettuce). U. lactuca is a suitable model for marine organic matter as it has a simple morphology, is ubiquitous throughout the world’s oceans, and readily sorbs a great variety of trace metals. Solutions containing all YREEs were equilibrated for 6-12 hours with dehydrated, powdered U. lactuca tissue over a wide pH range (3.0-8.5) at three ionic strengths (0.05, 0.5 and 5.0 M NaCl), after which aliquots were filtered through 0.22 μm membranes. The resulting filtrates were further separated into >30 kDa and >3 kDa colloidal fractions by sequential centrifugation in Amicon® ultrafiltration tubes. In all three experiments, YREEs are truly dissolved (<3 kDa) at low pH but almost entirely colloidal (>30 kDa) at high pH with a sharp transition in between, suggesting pH-dependent YREE complexation with large organic ligands released by the algal cells. The fraction of small colloids (3-30 kDa) is generally negligible. The same sorption edge emerged for fresh algal tissue, implying that the release of organic ligands is not caused by pervasive cell rupture. In 0.5 and 5.0 M NaCl solutions the sorption edge is centered around pH 6-8, but in 0.05 M NaCl it occurs around pH 4-6 whence more than 80% of dissolved YREEs is actually bound to colloids at pH 6. On the other hand, the amount of organic ligand released, as monitored by DOC concentrations, is nearly identical in each case. All sorption edges were satisfactorily modeled with an appropriate pH-dependent complexation model. At 0.05 M ionic strength, distribution coefficients calculated from the 0.22 μm filtrates show an unexpected inversion around pH 5 to decreasing sorption with increasing pH, most prominently for the MREEs, clearly reflecting the erroneous omission of YREE colloids from the particulate fraction. This feature could be completely reversed by using the modeled colloidal contributions to correct the measurements. A similar correction of the 0.5 and 5.0 M data revealed a more insidious effect on both absolute and relative distribution coefficients that was not otherwise apparent from their pH-dependent behavior. Failure to account for colloid-bound metals can ostensibly lead to a substantial misinterpretation of experimental results, an issue that is not sufficiently acknowledged or addressed in many metal sorption studies.