Determination of Total Dissolved Cobalt in UV-Irradiated Seawater Using Flow Injection with Chemiluminescence Detection

A sensitive flow-injection method with chemiluminescence detection (FI-CL) for the determination of dissolved cobalt in open ocean samples, suitable for shipboard use has been developed. To date, FI methods for dissolved cobalt have been used only in coastal and estuarine waters. Therefore, significant modifications to existing methods were required, including (1) the use of a commercially available iminodiacetate (IDA) resin (Toyopearl AF-chelate 650M) in place of resin immobilized 8-hydroxyquinoline for online preconcentration and matrix removal, (2) the introduction of acidified ammonium acetate (pH 4) as a column-conditioning step before sample loading and rinse steps, and most importantly, (3) UV irradiation of acidified seawater samples to determine total dissolved cobalt, rather than an operationally defined fraction. This method had a detection limit of 4.5 pM (3s of the blank). The accuracy of the method was evaluated by determining total dissolved cobalt in acidified North Pacific deep seawater (1000 m) samples from the Sampling and Analysis of Iron (SAFe) program and NASS-5. The method yields a mean (± SD) value of 40.9 ± 2.6 pM (n = 9), which is in excellent agreement with the SAFe consensus value of 43 ± 4 pM, and 208 ± 30 pM for NASS-5 (certified value 187 ± 51 pM). This study demonstrates that UV irradiation is an essential step for the determination of total dissolved cobalt in seawater by FI-CL. The method was applied to vertical profiles from the Sargasso Sea, indicating that total dissolved cobalt is influenced by both biological and physical processes. *Corresponding author: E-mail: *[email protected] Acknowledgments The authors thank Geoffrey Smith for providing the SAFe samples used in the method development and Simon Ussher and Christopher Marsay for assistance collecting samples during the FeAST-6 cruise. The authors also thank Mak Saito and an anonymous reviewer for their invaluable comments on an earlier version of this manuscript. This work was funded by a University of Plymouth, Marine Institute PhD scholarship to R. U. Shelley and US National Science Foundation grant OCE0550594 to P. N. Sedwick. DOI 10.4319/lom.2010.8.352 Limnol. Oceanogr.: Methods 8, 2010, 352–362 © 2010, by the American Society of Limnology and Oceanography, Inc. LIMNOLOGY and OCEANOGRAPHY: METHODS Shelley et al. FI method: Cobalt in seawater 353 (Sakamoto-Arnold and Johnson 1987; Cannizzaro et al. 2000; Sohrin et al. 2008; Milne et al. 2010). In recent years, the increased use of inductively coupled plasma mass spectrometry (ICP-MS) and flow injection (FI) techniques for trace metal analysis has resulted in widespread use of chelating ion exchange resins for separation and preconcentration. Traditionally, the preconcentration resin of choice for the FI analysis of trace metals has contained 8-hydroxyquinoline (8-HQ) as the chelating group. Resin-immobilized 8-HQ is attractive, as it has a strong affinity for binding a number of trace metals of interest in seawater (Landing et al. 1986). Because resin-immobilized 8-HQ is not commercially available, however, it needs to be synthesized (Landing et al. 1986; Dierssen et al. 2001), which produces resins of varying quality. Commercially available chelating resins are arguably preferable since the quality of the resins is reproducible, and such resins have been successfully used in FI systems (e.g., AguilarIslas et al. 2006; Lohan et al. 2006) and combined with ICPMS detection (Lohan et al. 2005; Sohrin et al. 2008; Milne et al. 2010). Commercially available resins that have been used for the analysis of trace metals in seawater include an NTAtype chelating resin for Fe and Cu (Lohan et al. 2005), an EDTA-type resin for Al, Cd, Co, Cu, Fe, Mn, Ni, Pb, and Zn (Sohrin et al. 2008), and Toyopearl AF-Chelate 650M resin, which contains tridentate iminodiacetate functional groups, for Al, Cd, Co, Cu, Mn, Ni, Pb, and Zn (Warnken et al. 2000; Ndung’u et al. 2003; Aguilar-Islas et al. 2006; Brown and Bruland 2008; Milne et al. 2010). Flow injection systems are particularly well suited to oceanographic analyses, since they are portable and robust, contain relatively simple and inexpensive components, and offer analytical determinations over a dynamic range of as much as three orders of magnitude (e.g., Xu et al. 2005). Modern FI systems are capable of high sensitivity, although the limits of detection vary with the method of detection and the analyte. Flow injection systems can also be coupled with towed, trace metal–clean sampling devices, enabling the near-realtime determination of trace metals in surface ocean waters (e.g., Bowie et al. 2002; Bruland et al. 2005). In addition, FI methods are ideal for kinetic catalytic methods that use the analyte of interest as a reaction catalyst, resulting in highly sensitive analytical techniques (Resing and Mottl 1992; Measures et al. 1995; Aguilar-Islas et al. 2006). Previously developed FI methods for the determination of dissolved cobalt in coastal and estuarine waters have used the Trautz-Schorigen reaction (TSR), which involves the oxidation of either gallic acid (3,4,5trihydroxybenzoic acid; Sakamoto-Arnold and Johnson, 1987) or pyrogallol (Cannizzaro et al. 2000) with hydrogen peroxide in the presence of cobalt as a catalyst to produce chemiluminescence emission in the visible region. The aim of the work described in this article was to develop an FI method that can determine total dissolved cobalt in open-ocean waters and to highlight the need to UV-irradiate seawater samples before determination of dissolved cobalt. The recent launch of the international GEOTRACES program, which aims to determine trace metal concentrations in diverse oceanic regions, requires accurate methods that are highly sensitive. The focus of the study described here was the analysis of dissolved cobalt in open-ocean seawater, for which no FI method has been developed to date. This method was then applied to water column samples collected from the oligotrophic Sargasso Sea in June 2008. Materials and procedures Apparatus—A schematic diagram of the flow-injection manifold used is shown in Fig. 1; it consists of an eight-channel peristaltic pump (Minipuls 3, Gilson); two micro-electronically actuated six-port, two-position injection valves (VICI, Valco Instruments); a photomultiplier tube (PMT, Hamamatsu H 6240-01) containing a quartz glass spiral flow cell (internal volume 130 μL; Baumbach and Co.); and a thermostatic water bath (Grant). The peristaltic pump tubing used was two-stop accu-ratedTM PVC (Elkay). All other manifold tubing was 0.8 mm i.d. PFA Teflon (Cole-Parmer). The FI manifold had one mixing coil (1.85 m) and one reaction coil (5 m). The reaction coil was constructed by French-knitting 5 m of 0.8 mm i.d. PFA Teflon tubing using a four-pronged knitting spool. Two acrylic columns (internal volume 70 μL), with porous HDPE frits (BioVion F, 0.75 mm thick), were incorporated in-line, one on the sample-buffer line to remove trace-metal impurities from the buffer solution, and a second for the preconcentration of cobalt and removal of the cations from the seawater sample matrix. Both columns were filled with Toyopearl AFChelate-650M resin (Tosohaas) (hereafter referred to as “IDAToyopearl”). The direction of flow through the cleanup column was one-way, so it was necessary to reverse the column at the end of each day to prevent the resin from becoming compacted. This was not required for the sample preconcentration column, which was loaded and eluted in opposing flow directions. The T-piece before the 1.85-m mixing coil, the mixing coil itself, and the 5-m reaction coil were maintained at 60°C by placing them inside a thermostatic water bath (Grant). The data acquisition module (Ruthern Instruments) and valve control software (LabVIEW v. 7.1) were operated using a laptop computer (Dell). To minimize contamination, all sample handling was carried out in a Class-100 clean bench. The FI system was flushed daily and after any configuration changes (such as pump tube or reagent replacement) with 1 M HCl solution. Reagents—Unless otherwise stated, all chemicals were obtained from Fisher Scientific, and used as received. All solutions were prepared inside a Class-100 clean bench using ultra-high-purity (UHP) water (≥ 18.2 MΩ cm, Elgastat Maxima). The eluent, 0.1 M hydrochloric acid (HCl), was prepared by diluting 8.8 mL of 10 M ultrapure subboiling distilled HCl (SpA, Romil) to 1 L in UHP water. The 0.05 M ammonium acetate column rinse and conditioning solution was prepared by dissolving 3.8 g of ammonium acetate crysShelley et al. FI method: Cobalt in seawater 354 tals in 1 L UHP water and adjusting to pH 4 with 3.8 mL of 10 M ultrapure HCl (SpA, Romil). The 0.3-M ammonium acetate sample buffer solution was also prepared in UHP water from ammonium acetate crystals (23.11 g L–1), and the acidified seawater samples were buffered online to between pH 5.2 and 5.5 by mixing with this solution. Because the detection limit of FI methods is often limited by the blank value rather than the sensitivity of the instrumental technique (Bowie and Lohan 2009), the potential contribution to the blank from the sample buffer solution was minimized by using a cleanup column (identical to the preconcentration column) on the sample buffer line. The 0.17-M sodium hydroxide (NaOH) reaction buffer solution was prepared by dissolving sodium hydroxide (NaOH) pellets (6.7 g L–1) in 1 L solution (20% vol/vol methanol [HPLC grade], 80% vol/vol UHP water). The 50 mM pyrogallol reagent was prepared by sonicating 6.30 g pyrogallol and 9.12 g acetyltrimethylammonium bromide (CTAB) in UHP water; when the pyrogallol and CTAB were fully dissolved, 58.4 mL of 35% hydrogen peroxide (H2O2) was added, and the solution was diluted to 1 L with UHP water. As the pyrogallol solution is reportedly stable for only 48 hours (Cannizzaro 2001), this reagent was prepared daily as required. Sample collection and pretreatment—Surface seawater (0–1 m depth) was collected using a trace metal–clean pole sampler: two 1-L widemouth low-density polyethylene (LDPE, Nalgene) bottles were secured in a Plexiglas frame at the end of a bamboo pole, which was extended from the ship’s stern for sample collection while backing slowly into the wind. Water column samples were collected in modified 5-L Teflon-lined external closure Niskin-X samplers (General Oceanics) suspended from a Kevlar line using a stainless-steel end weight and solid PVC messengers (Sedwick et al. 2005). Samples were immediately filtered at sea inside a shipboard Class-100 clean container laboratory, through a 0.4-μm pore-size Supor Acropak filter capsule (Pall) that was prerinsed with 5 L UHP water followed by several hundred milliliters of sample. Filtered samples were acidified to 0.024 M with ultrapure HCl (SpA, Romil) before analysis. Procedure—The peristaltic pump (Minipuls 3, Gilson) was set at 5.50 rpm to attain the flow rates shown in Fig. 1. After stabilization of the baseline (typically 30–45 min), with valve 1 and valve 2 in position B, acidified ammonium acetate was passed through the preconcentration column for 40 s. Valve 1 was then switched to position A, and a buffered sample was loaded onto the preconcentration column for 300 s. Valve 1 was then switched to position B, and the preconcentration column was rinsed for 40 s with the acidified ammonium acetate solution. After this rinse step, valve 2 was switched to position A for 90 s, and the eluent (0.1 M HCl) was passed over the chelating resin in the opposite direction to that of the loading phase, thus eluting the cobalt from the preconcentrating resin into the reagent stream that was then carried to the PMT detector. In total, one complete analytical cycle took 7.8 min. During the load and rinse phases, the eluting acid bypassed the column and mixed with the other reagents to produce the baseline signal. Standardization—The analytical system was calibrated daily by simple linear regression of standard curves. Stock standard solutions were prepared in acidified UHP water by serial dilution of a 17-mM cobalt atomic absorption standard solution (Spectrosol). Working standards (additions of 12.5–75 pM) were Fig. 1. FI-CL manifold configuration for the determination of total dissolved cobalt in seawater. Shelley et al. FI method: Cobalt in seawater 355 prepared in acidified low-trace-metal seawater (Atlantic Ocean surface seawater collected from 28°51’S, 4°41’W; total dissolved cobalt 13.7 ± 2.7 pM). Standards were run at the beginning and end of each program of analysis in triplicate, and concentrations were calculated from peak heights. Standard curves were linear (r2 > 0.99) up to concentrations of 2 nM.