A Critical Evaluation of Tangential-Flow Ultrafiltration for Trace Metal Studies in Freshwater Systems. 1. Organic Carbon

Stephen Hoffmann, Martin M. Shafer, Christopher L. Babiarz, David E. Armstrong

Environmental Science & Technology · 2000 · 52 citations · 29 references

Concepts

Abstract

The utility of tangential flow ultrafiltration (TFUF) for size fractionation of natural organic matter (NOM) in freshwater streams was investigated, focusing on characterization or elimination of potential artifacts. Spiral-wound polyethersulfone (PES) and regenerated cellulose (RCL) membranes with nominal molecular weight limits of 10 kilodalton (kDa) and 100 kDa were compared as part of a large project assessing the utility of large volume (>5 L) ultrafiltration for determining trace metal speciation in freshwaters. With careful cleaning, reliable fractionations of carbon and trace metals in freshwater can be obtained, and a detailed protocol necessary to avoid potentially significant biases is presented. Both PES and RCL membranes can be cleaned efficiently to provide low carbon blanks (<0.03 mg C L - 1 ) and recleaned at least 10 times without compromising membrane integrity. For RCL, residual acid from the cleaning procedure can lower the permeate pH to below 4 in low-alkalinity river samples, but this acid can be removed prior to sample filtration with a simple treatment. Mass balance experiments with stream NOM show good recovery (99.7% ± 9.5%, n = 67). Sorption losses are very low (on average, <2%) to RCL membranes but are higher for PES. Replicate precision is typically better than ±5% for carbon. There are significant differences between PES and RCL membranes: permeate carbon levels are similar, but RCL predicts a larger >10 kDa fraction and carbon sorption is more significant on PES. These findings could be explained by greater hydrophobic interactions of dissolved organic carbon with PES and charge interactions with RCL. A strong negative correlation is observed between the percent of colloidal carbon (10 kDa − 0.4 μm) and the log specific conductance of streamwaters ( R 2 = 0.71).

References

29