Application of cross-flow filtration for determining the solubility of iron species in open ocean seawater

Measurements of soluble iron species (organic and inorganic) are important for understanding the transport of iron within the ocean and its bioavailability. Recent developments in ultrafiltration equipment and analytical detection techniques for low level Fe determination has turned the spotlight on obtaining data on soluble iron species. However there have, until now, been few studies that have characterized the performance of an ultrafiltration system with respect to well described soluble iron complexes. In the present work, we describe a methodological study characterizing the behavior of soluble and colloidal iron species in seawater by combining a crossflow ultrafiltration system (Vivaflow 50TM) with a radioisotope (55Fe). During this study, we were able to maintain excellent mass balances by including all components: not only the solution phases (retentate and permeate) but wall-adsorbed and filter-adsorbed iron, which were recovered by an acid-rinsing step. Wall and filter adsorption were unavoidable when solutions were saturated with respect to Fe'. However in undersaturated solutions, such as with an excess of desferrioxamine B, wall and filter adsorption were minimized, indicating that these effects should be slight for natural samples where iron-binding ligands are in excess. Our results have important implications for the use of ultrafiltration membranes for open ocean iron biogeochemical studies. *Corresponding author: E-mail: [email protected] Acknowledgments We thank Uwe Rabsch and Petra Krischker for technical support with the radio-isotope work performed in the radio chemistry suite at IfMGeomar. Thanks also to Peter Streu (IfM-Geomar) for help in this work. Christina De La Rocha (AWI) is thanked for her comments on earlier versions of this manuscript. The helpful comments of two anonymous reviewers and the associate editor Steven Wilhelm greatly improved this manuscript. This study was supported by Deutsche Forschungsgemeinschaft (DFG) grants CR 145/5 and CR145/9 (to PLC). Limnol. Oceanogr.: Methods 6, 2008, 630–642 © 2008, by the American Society of Limnology and Oceanography, Inc. LIMNOLOGY and OCEANOGRAPHY: METHODS As shown in numerous works more than 99% of the dissolved Fe (<0.2 μm) is bound by organic ligands throughout the world oceans (Croot and Johansson 2000; Rue and Bruland 1997; Van Den Berg 1995; Witter and Luther 1998; Wu and Luther 1995). Hutchins et al. (1999b) concluded that the organic complexation of Fe increases the amount of dissolved Fe species and consequently its biological availability. However, the bioavailable differences of organically complexed and colloidal Fe is not well understood (Chen and Wang 2001; Hutchins et al. 1999a; Kuma et al. 2000). The presence of siderophores and other Fe complexing ligands produced, or released via grazing or viral lysis, from phytoplankton, and bacteria may stabilize soluble Fe, increasing both the residence time and total pool size of bioavailable Fe in the surface ocean (Barbeau et al. 2001, 2003). Recent studies have shown that a significant fraction of the dissolved Fe pool exists as colloidal Fe species (Bergquist et al. 2007; Nishioka et al. 2001; Wu et al. 2001) with dissolved Fe concentrations in the euphotic zone being dominated by the variability of the colloidal Fe fraction. The colloidal Fe variability in the NW Atlantic was suggested to be seasonally dependent with higher concentrations in the winter, than in the summer (Bergquist et al. 2007). Some fundamental issues that remain to be answered in the marine biogeochemistry of Fe are the extent to which organic binding agents increase Fe solubility and how those ligands prevent the formation of colloidal Fe in seawater. Until now only studies from Fe enrichment experiments have examined Fe ligand complexation and the formation of colloidal Fe (Boye et al. 2005; Wells 2003) with little attention paid to the overall solubility of Fe. Here we present a study that through the careful use of cross-flow ultrafiltration technique and theory outlines the impact of different ligands on Fe speciation and solubility. Materials and procedures Reagents—The impact of various different organic ligands on the speciation of Fe in seawater was carried out via crossflow filtration using the radioisotope, 55Fe (Hartmann Analytics). The 55Fe had a specific activity of 157.6 MBq/mg Fe, a total activity of 75 MBq, and was dissolved in 0.51 mL of 0.1 M HCl. 55Fe dilutions were produced with 18 MΩ deionized, ultrapure water and were acidified with quartz-distilled HCl (QD-HCl) to a pH below 2. The 7 ligands tested; desferrioxamine B (DFB), ethylenediaminetetraacetic acid (EDTA), 2-(2thiazolylazo)-p-cresol (TAC), phytic acid (IP6), protoporphyrin IX (PP IX), phytagel, and 2-keto-D-gluconic acid (2kDG) were obtained from Sigma-Aldrich. Ligand solutions were made up in 0.2 μm prefiltered Antarctic seawater (sampled during Eifex 2004 under trace metal clean conditions; total dissolved Fe concentration [Fed = 0.2 nM]). Prior to use, this seawater was irradiated with UV light (UV-Digester 705 from Metrohm) for 75 min to destroy any organic compounds present. All labware used was soaked in 10M HCl for at least 7 d and then rinsed with ultrapure water prior to use. Ultrafiltration setup and cleaning procedure—The ultrafiltration of Fe and ligand-containing solutions was carried out using a Masterfex(r) L/S(tm) system with a Vivaflow 50 membrane (10 kDa) constructed of PES (polyethersulfone) with an active membrane area of 50 cm2. The recirculation rate was set to approximately 300 mL/min, which typically gave a permeate flow rate of 5 mL/min. All ultrafiltration work was carried out using acid-cleaned polycarbonate (PC) container and polyethylene (PE) tubing. The Vivaflow 50 was precleaned by sequential rinsing with 100 mL Ultrapure water, 100 mL of a 1% solution of 6 M Q-HCl, a 100 mL EDTA wash solution (10 mM), and then finally a last rinse with 100 mL Ultrapure water to remove trace metal contamination. The ultrafilter could be reused several times following this procedure. 55Fe measurements—55Fe in the various ultrafiltration fractions was quantified using a liquid scintillation counter (TriCarb 2900TR) from Packard and the cocktail Lumagel Plus (Lumac LSC). The efficiency (55%-60%) of the instrument was obtained by several quench curve calibration measurements. The lowest measurable Fe concentration (detection limit) was at 8.1 ± 1.5 pM, equivalent to 1 count per minute (CPM). The ultrafiltration setup (PC container, PE tubing, and 10 kDa membrane filter) was cleaned prior to each experiment with a sequence of two short washes (0.1 M Q-HCl followed by 10 mM EDTA) and a rinse with Ultrapure water. Theory: cross-flow filtration—The theory for cross-flow filtration of solutes has been described in detail previously by other researchers in this field: Reitmeyer et al. (1996), Guo et al. (2001), and Hassellöv et al. (2007). In general, the separation of Fe species by ultrafiltration can be represented as a simple mass balance in a closed system with a fixed total solution volume: