Critical evaluation of a seaFAST system for the analysis of trace metals in marine samples
Résumé
A seawater preconcentration system (sPAFAST) with offline sector-field inductively coupled plasma mass spectrometry (SF-ICP-MS) detection was critically evaluated for ultra-low trace elemental analysis of Southern Ocean samples over a four-year period (2015-2018). The commercially available system employs two Nobias PAl resin columns for buffer cleaning and sample preconcentration, allowing salt matrix removal with simultaneous extraction of a range of trace elements. With a primary focus on method simplicity and practicality, a range of experimental parameters relevant to oceanographic analysis were considered, including reduction of blank levels (over weeks and years), instrument conditioning, extraction efficiencies over different pH ranges (5.8-6.4), and preconcentration factors (similar to 10-70 times). Conditions were optimised for the analysis of ten important trace elements (Cd, Co, Cu, Fe, Ga, Mn, Ni, Pb, Ti and Zn) in open ocean seawater samples, and included initial pre-cleaning and conditioning of the seaFAST unit for one week before each separate analytical sequence; a controlled narrow buffer pH of 6.20 +/- 0.02 used for extraction; and a sample preconcentration factor of 10 for (relatively) concentrated rainwater or sea ice, 40 for typical seawater samples, and up to 67 times for seawater samples collected in the remote open ocean such as the Southern Ocean. Method accuracy (both short - days to weeks - and long term - months to years) were evaluated through extensive analysis of a range of oceanographic standard reference samples including SAFe D1 (n = 20), D2 (n = 3), S (n = 15), GEOTRACES GD (n = 6), GSC (n = 42) and GSP (n = 42), as well as NASS-6 (n = 6). Measured values for oceanographic samples were found to agree with consensus values to within +/- 6% for Cd, Cu, Fe, Ni, Pb and Zn. Offsets were noted for Co (labile fraction only; no UV oxidation), Mn (difference also noted in other recent studies) and Ti (limited reference values). No consensus values currently exist for Ga. Iron and Mn in Southern Ocean samples were also independently verified via flow injection analysis methods (R-2 = 0.95, n = 244 (Fe) and 0.92, n = 85 (Mn), paired t-test, p \textless\textless 0.05). Precisions over four years were evaluated through analysis of community seawater samples as well as a range of bulk in-house seawaters (3 sources, each n-100) and acid blanks (n = 250), and were typically found to be within 5-8%, depending on analyte and concentration. Values presented here represent one of the largest independent data sets for these reference samples, as well as the most documented comprehensive suite of GSP and GSC values currently available (consensus values have not yet been released). Samples covering a range of salinities (0-60) were investigated to demonstrate method versatility, with excellent recoveries noted using the seaFAST Nobias PAl column (\textgreater 98% for most elements, with 70-80% for Ga and Ti). By way of example, data is presented showing the application of the method to samples collected on the Kerguelen plateau in the Indian sector of the Southern Ocean (HEOBI voyage, January February 2016) and in land-fast ice and brine collected near Davis station, Antarctica, in austral summer 2015 (with a salinity range from 0 to 73 g kg(-1)). Finally, a range of recommendations for successful implementation of a seaFAST system are provided, along with considerations for future investigation.