High Resolution Determination of Nanomolar Concentrations of Dissolved Reactive Phosphate in Ocean Surface Waters Using Long Path Liquid Waveguide Capillary Cells (LWCC) and Spectrometric Detection

In the last decade, long path length, low volume, liquid waveguide capillary cells (LWCC) in conjunction with conventional nutrient auto-analyzers have been applied to determinations of nanomolar levels of phosphate, nitrate, and nitrite in oligotrophic waters. This article reports a high resolution, real-time, continuous method for nanomolar dissolved reactive phosphate measurements in ocean surface waters with data logging every 30 seconds for up to 16 consecutive hours. Surface seawater is pumped continuously from a shipboard underway tow-fish unit to a helium gas-segmented, continuous-flow, nutrient auto-analyzer modified with a 250 cm LWCC. To circumvent baseline instability due to reagents, a parallel channel with deionized water (DI) and reagents is run and later subtracted from the sample absorbances. The detection limit is 0.8 nmol/L. The precision (as relative standard deviation) at 5 nmol/L phosphate is 6.1% (n = 5) and 0.8% (n = 5) at 50 nmol/L. We also report an optimized method for discrete samples using a 200 cm LWCC. To minimize any background phosphate concentration in low nutrient seawater used as wash water solution, we use DI water, but increase sample and wash times to achieve plateau-shaped peaks after the transient wash/sample mixing period. The detection limit is 0.5 nmol/L. The precision at 10 nmol/L phosphate is 1.8% (n = 8) and 0.9% (n = 9) at 60 nmol/L. The two systems have successfully been deployed on the U.S. GEOTRACES 2010 cruise, transecting the upwelling area northwest of Africa and the highly stratified, oligotrophic, subtropical North Atlantic gyre. *Corresponding author: E-mail: [email protected] or [email protected] Acknowledgments We thank Geoffrey J. Smith, University of California, Santa Cruz, for an infinite supply of trace metal clean surface seawater for the continuous system, maintenance of the shipboard underway tow-fish unit, and for writing a Visual Basic program for data processing. We also thank Doug Bell, Bermuda Institute of Ocean Sciences, for processing one set of MAGIC results and Brandon Gipson, Old Dominion University, for LWCC phosphate determinations on U.S. GEOTRACES 2009 Intercalibration Cruise. This paper is part of the Intercalibration in Chemical Oceanography special issue of Limnology and Oceanography: Methods. The work was financially supported by the U.S. National Science Foundation (Grants OCE 0648408 and OCE 0926092 to G. Cutter). DOI 10.4319/lom.2012.10.568 Limnol. Oceanogr.: Methods 10, 2012, 568–580 © 2012, by the American Society of Limnology and Oceanography, Inc. LIMNOLOGY and OCEANOGRAPHY: METHODS The spectrometric method based on the formation of the blue form of reduced phosphomolybdate (molybdenum blue) is widely used for determining dissolved (or soluble) reactive phosphate at micromolar concentrations (Murphy and Riley 1962; Mee 1986; Alvarez-Salgado et al. 1992; Zhang and Chi 2002; Estela and Cerdà 2005; Paytan and McLaughlin 2007). For determining phosphate at nanomolar concentrations, preconcentration by magnesium-induced co-precipitation (MAGIC) has proven useful (Karl and Tien 1992; Rimmelin and Moutin 2005). After addition of NaOH, phosphate is scavenged into the Mg(OH)2 precipitate. The precipitate is centrifuged, then dissolved in dilute HCl and subsequently quantified via the molybdenum blue method. During the last decade, long path length, low volume, liquid waveguide capillary cells (LWCC), with spectrophotometric detection, have gained foothold as a simple, easily automated, and reliable technique for nanomolar phosphate measurements (Zhang 2000; Zhang and Chi 2002; Gimbert et al. 2007; Li et al. 2008; Patey et al. 2008; Ma et al. 2009). The standard wavelength for the molybdenum blue method is 880 nm, but due to the large absorption of far-red wavelengths by water, light transmitted in a LWCC is negligible at this wavelength (Patey et al. 2008). For this reason, the second absorption maximum of molybdenum blue at 700 nm (Murphy and Riley 1962) is used for determining phosphate when working with the LWCC technique. The molybdenum blue method does suffer from a number of interferences, which potentially can result in an overestimation of phosphate (Karl and Tien 1992; Anagnostou and Sherrell 2008; Ma et al. 2009). This is more likely to be significant when phosphate is present at nanomolar concentrations. Hydrolysis of pyrophosphates and organic phosphorus compounds due to the reaction’s acidic conditions (pH < 1) also can form phosphomolybdate complexes, and hence “soluble reactive phosphate” is a common operational term used to describe the dissolved (<0.2–0.7 μm) phosphate measured with the molybdenum blue method (Benitez-Nelson 2000; Paytan and McLaughlin 2007). Li and Hansell (2008b) investigated dissolved organic phosphate (DOP) interference for a LWCC method and found DOP interference to be low and similar to the DOP interference for the MAGIC method (Karl and Tien 1992). Arsenate [As(V)], typically ranging from 12 to 21 nmol/L in the open ocean (Cutter et al. 2001; Cutter and Cutter 2006), forms a similar blue complex of an equivalent molar absorptivity to that of phosphate, but reacts more slowly with the reagents forming the arsenomolybdate complex (Johnson 1971; Downes 1978; Karl and Björkman 2002). The reduction of arsenate to arsenite [As(III)] with thiosulphate can be employed as a preliminary step for phosphate measurements because arsenite is nonreactive to the molybdate reagent, but is rarely applied for automated nanomolar phosphate measurements (Karl and Björkman 2002). Silicate [Si(OH4)] is another potential interferent with the molybdenum blue method, but can be minimized with optimized reaction conditions. Reducing pH < 1 in the final solution, reacting at room temperature, and the addition of antimony as a catalyst reduces silicate molybdenum blue formation (Zhang et al. 1999). Gimbert et al. (2007) reported that interference from silicate can be effectively masked by addition of tartaric acid. These finding are supported by Ma et al. (2009) who used similar conditions for LWCC phosphate experiments adding silicate in concentrations between 0 and 240 μmol/L to samples containing 0 or 83 nmol/L phosphate. The results did not show any significant difference in absorbance between samples with or without silicate (Ma et al. 2009). Variations in baseline over time due to instability of reagents and coating of the walls of the LWCC and sample lines with colloidal molybdenum blue are also well known (Li et al. 2008; Ma et al. 2009). These baseline variations can become a challenge for long-term continuous measurements (hours), since an increase in the baseline can mask the analyte concentration signal. Low nutrient seawater (LNSW) has routinely been used in oceanography for preparation of standards, blanks, and wash solution to match the salinity of the samples, and thereby, improve accuracy (Zhang 2000). However, in oligotrophic waters, phosphate concentrations are of the same order of magnitude as that in LNSW, and its determination is therefore inherently blank-limited. Zhang and Chi (2002) recommended “phosphate-free” seawater prepared from the supernatant from the MAGIC procedure for wash solution and for nanomolar level phosphate standards. However, by adding NaOH large amounts of Mg are removed, increasing the alkalinity and changing the matrix of the seawater. Alternatively, Li et al. (2008) eliminated phosphate from seawater by adding ferric chloride and subsequent co-precipitation of phosphate with ferric hydroxide, producing a clear solution at final pH of ~ 6. Hence, it is difficult to prepare a blank for nanomolar phosphate measurements with the same physical properties as seawater samples (Froelich and Pilson 1978). In addition to the chemistry of the water for blanks and standards, differences in the physical properties of high and low salinity seawater samples, combined with non-ideal optical characteristics of flow cell design in continuous-flow analysis, change how light is transmitted in the flow cell. This causes changes in the refractive index (Schlieren effect) when the wash water and sample mixing zone passes through the flow cell, with resultant absorbance errors (Froelich and Pilson 1978; Alvarez-Salgado et al. 1992; Motomizu and Li 2005; Ma et al. 2009). Such differences can be a problem when using deionized water (DI) as wash solution in between seawater samples in continuous-flow analysis, resulting in an analytical error up to 13 nmol/L in North Sea water for phosphate (Ormaza-González and Statham 1991). Thus, the choice of water type for wash solution in discrete sample analysis, or for standards and blanks is a critical aspect of nanomolar phosphate determinations. In automated, gas-segmented, continuous-flow analysis (SCFA), deliberate introduction, and later removal, of gas bubZimmer and Cutter Determination of nanomolar phosphate