Sample Collection and Filtration
\nAll seawater samples analyzed here for dissolved iron (dFe) concentration and isotope ratios (\u03b456Fe) were collected between 21 November and 4 December 2013 aboard the R/V Thomas G. Thompson as part of the U.S. GEOTRACES GP16 East Pacific Zonal Transect (Resing et al. 2015, Fitzsimmons et al. 2017, John et al. 2018). The EPZT encompassed 35 stations along an 8000 km transect that traversed from Peru to Tahiti (Peters et al. 2017, Moffett & German 2018). All seawater samples were collected as part of the standard U.S. GEOTRACES trace metal casts, using a trace metal clean carousel and CTD (Seabird) with a Kevlar conducting cable, fitted with 24 x 12-L Teflon-coated GO-Flo bottles (General Oceanics), following established GEOTRACES protocols (Cutter & Bruland 2012). Rosette deployments began with GO-Flo bottles open, allowing them to flush during deployment to depth, and the GO-Flo bottle was subsequently tripped on ascent at ~3 m/min, to minimize chances of contamination from rosette hardware. Upon recovery, GO-Flo bottles were transferred immediately into a clean van for subsampling and were pressurized using HEPA-filtered air. Each GO-Flo bottle was fitted with acid-cleaned Teflon tubing connected to a 0.2 um Acropak-200 capsule filter (Pall), allowing collection of filtered seawater for bulk dissolved Fe analysis into 4 L acid-clean low-density polyethylene (LDPE) bottles, after three 10% volume seawater rinses of the caps and threads (Cutter et al. 2017). Filtered seawater from Station 26 at 2650 m and 2700 m depths were combined to achieve sufficient volume for ultrafiltration processing.
Ultrafiltration and Sample Processing
\nImmediately after sample collection and 0.2 um filtration, 1 L of the 4 L homogenized sample was poured into a clean 1 L LDPE bottle for dissolved Fe analysis, and the remaining sample (~ 3 L) was ultrafiltered. Ultrafiltration took place using a cross flow filtration system (CFF) with a 10 kDa (~0.003 um for globular proteins; Erickson et al. 2009) regenerated cellulose membrane (Pellicon XL PLCGC) in single-pass mode (see Fitzsimmons and Boyle 2014 for detailed methods). Briefly, before collection, the CFF system was conditioned using ~500 mL of sample seawater, and the sample bottles, caps, and threads were rinsed with a minimal amount of ultrafiltered seawater. After conditioning, ~1 L of permeate (soluble fraction) and ~1 L of retentate were collected into separate 1 L LDPE bottles. All samples were then acidified to 0.012 M hydrochloric acid (HCl, Optima, Fisher) and stored for more than a year before analysis. Soluble Fe (sFe) is defined as the concentration in the permeate solution.
Dissolved Fe Concentration Analysis
\nIron concentrations in the dissolved, permeate, and retentate phases were analyzed using a 15 mL aliquot of the 1 L samples. Samples were analyzed in the Sherrell lab at Rutgers University with a SeaFAST pico system (ESI, Omaha, NE, USA) and Thermo Element 1 HR-ICPMS following the SeaFAST methods of Lagerstrom et al. (2013). Samples were analyzed offline, as outlined by Jensen et al. (2020).
Dissolved Fe isotope ratio (\u03b456Fe) Analysis
\nIron stable isotope ratios (\u03b456Fe) were determined using a double-spike multi-collector MC-ICPMS technique, based on Conway et al. (2013). All sample processing and Fe isotope analyses were carried out at the University of South Florida (USF). Chemical processing of samples took place in ISO-5 laminar flow hoods within the Marine Metal Isotope and Trace Element ISO-6 clean laboratory at the USF College of Marine Science. All ultrapure water used was from a Thermo Scientific Barnstead Genpure Ultrapure Water System; all plasticware was cleaned using standard procedures (Conway et al. 2013); and all reagents were Optima grade from Fisher Scientific. Briefly, the ~1 L of dissolved, permeate, and retentate seawater samples were spiked with a double spike containing 57Fe and 58Fe at a ratio of ~1:1, to provide a ~1:2 ratio of natural:spike Fe in samples. Following spiking, samples were amended to 10 mM H\u2082O\u2082 and left for 24 h after spiking. Samples were then brought to pH ~6, and Fe was extracted from seawater onto Nobias PA-1 chelating resin beads using a batch extraction technique. The Fe was then eluted from the Nobias resin using 3 M HNO\u2083, evaporated to dryness, refluxed with a conc. HNO\u2083 and conc. H\u2082O\u2082 mixture, evaporated to dryness, taken up in 1 M HCl, and purified using an AGMP-1 (BioRad) anion-exchange column chemistry technique following Sieber et al. (2019). Following column purification, samples were evaporated to dryness before being re-dissolved in 0.5 mL of 2% HNO\u2083 (v/v) for analysis by MC-ICPMS.
Samples were analyzed for Fe isotope ratios by Thermo Neptune Plus MC-ICPMS at USF\u2019s Tampa Bay Plasma Facility at the USF College of Marine Science. Briefly, samples were introduced via a ~100 \u00b5L/min PFA nebulizer into an ESI Apex \u03a9 desolvator (with Ar but no N\u2082 add gas), using a Ni Jet sampler cone and an Al X skimmer cone. Fe isotope ratios were measured using high resolution mode, and instrumental mass bias was corrected using the double spike technique. \u03b456Fe were calculated following an iterative procedure based on Siebert et al. (2001) and are expressed relative to the IRMM-014 international standard.
\nProblem report: Filtered seawater from Station 26 at 2650 m and 2700 m depths were combined to achieve sufficient volume for ultrafiltration processing.
Dissolved Fe Concentration Data Processing
\nThe raw counts from the Element 1 analyses were first corrected for any instrument blank derived from a 10% nitric acid and In solution repeatedly analyzed during the instrumental session. Following this, each sample was corrected for machine drift and variation in instrumental sensitivity during analysis using the added 115In as well as instrument mass bias using an Fe elemental standard. Subsequently, all counts for isotope spiked elements were converted to concentration in nmol/kg based on the weight of the sample and known concentrations of the added isotope spike mixture. For every 40 seawater samples at least 24 \"Procedure blanks\" were included throughout the run. These process blanks consisted of freshly acidified ultrapure water spiked with a diluted version of the isotope spike mixture. Each blank was processed identically to a sample in an attempt to account for contamination coming from solutions (buffer, eluent) used or the SeaFAST itself. After conversion to concentration units, the most consistent procedure blank values for each metal are used to subtract an average procedure blank from all sample concentrations determined during that run. Each run, an internal standard was analyzed six times and an external standard (such as SAFe water) was analyzed two times as a \"check\" of our long-term precision and accuracy.
For the session in which these samples were analyzed, accuracy of the technique was assessed using analyses of SAFe consensus seawater solutions during the analytical session: SAFe S = 0.13 nmol/kg (consensus value 0.093\u00b10.008 nmol/kg), SAFe D1 = 0.73 nmol/kg (consensus value of 0.67\u00b10.04 nmol/kg), and SAFe D2 = 0.95 nmol/kg (compared to consensus value of 0.933\u00b10.023 nmol/kg). An error of 5% is reported on all concentration values, based on the long-term external reproducibility of an in-house seawater solution. Dissolved Fe concentrations compared well with published dissolved Fe concentrations (Fitzsimmons et al. 2017, John et al. 2018) from the GP16 section reported in the Intermediate Data Product (IDP) 2017 (Schlitzer et al. 2018).
\nDissolved Fe isotope ratio (\u03b456Fe) Data Processing
\nTo account for any instrumental drift during the analytical session, each group of five samples was expressed relative to the average \u03b456Fe of two bracketing IRMM-014 double spike mixtures. The validity of this method for generating precise and accurate seawater \u03b456Fe data has been shown previously by several inter-laboratory inter-comparison exercises, as well as repeated analysis of seawater and standard reference materials (Conway et al. 2013). At USF, precision and accuracy of \u03b456Fe measurements are based on the long-term analysis of a 100 ng g^-1 NIST 3126a Fe standard (+0.36\u00b10.04\u2030, n=190 over 2 years). Following that approach, we apply 0.04\u2030 as a conservative 2SD measure of uncertainty on all \u03b456Fe data presented here, except if the standard internal error (2SD) is larger. In those cases, which typically occur at lower sample Fe concentrations, the 2SD is taken as a more conservative estimate of error and is used instead. Dissolved \u03b456Fe measurements made during this study compared well with previously reported data measured on separate subsamples in another laboratory (John et al. 2018), and when direct depth comparisons were not available we compared to the mean of bracketed sampling depths. 100% of measurements taken at the same depth were within error of one another, while 83% of all \u03b456Fe measurements were within error across the two studies (including samples compared to bracketed depth means).
Quality Flags:
\nData were flagged using the SeaDataNet quality flag scheme. For more information on SeaDataNet flags, see: https://www.seadatanet.org/Standards/Data-Quality-Control
SeaDataNet quality flag definitions:
\n0 = No quality control;
\n1 = Good value;
\n2 = Probably good value;
\n3 = Probably bad value;
\n4 = Bad value;
\n5 = Changed value;
\n6 = Value below detection;
\n7 = Value in excess;
\n8 = Interpolated value;
\n9 = Missing value;
\nA = Value phenomenon uncertain.
BCO-DMO Processing:
\n- added ISO_DateTime_UTC column.