Reactivity and fluxes of antimony in a macrotidal estuarine salinity gradient: Insights from single and triple quadrupole ICP-MS performances

The goal of this study is to evaluate and compare the performance of single quadrupole ICP-MS and triple quadrupole ICP-MS (QQQ-ICP-MS) for quantifying dissolved antimony (Sb) in estuarine waters, specifically along the salinity gradient of the Gironde Estuary, France. By analyzing previously collected samples under different hydrological conditions, the researchers demonstrate that the QQQ-ICP-MS method provides more robust and accurate results, especially by eliminating isotopic interferences.

This improved method allows for the first precise quantification of Sb net fluxes from the Gironde Estuary to the Atlantic Ocean and refines the value of the seawater endmember. The study highlights the importance of advanced analytical techniques for trace element analysis in complex brackish environments and sets a foundation for routine Sb monitoring in estuarine systems.

The original article

Reactivity and fluxes of antimony in a macrotidal estuarine salinity gradient: Insights from single and triple quadrupole ICP-MS performances

Teba Gil-Díaz, Frédérique Pougnet, Lionel Dutruch, Jörg Schäfer, Alexandra Coynel 

Marine Chemistry, Volume 267, November 2024, 104465

https://doi.org/10.1016/j.marchem.2024.104465

licensed under CC-BY 4.0

Selected sections from the article follow. Formats and hyperlinks were adapted from the original.

Unless combined with pre-injection techniques (e.g., HG, HPLC, GC, etc.), analyses of water samples via ICP-MS are assumed to provide “total” or bulk trace element dissolved concentrations. That is, out of all Sb species (i.e., inorganic or organic, in the truly dissolved phase or in colloidal form, Andreae et al., 1981, Filella et al., 2002b, Caplette and Mestrot, 2021, Filella and Rodríguez-Murillo, 2021), when quantifying water at the ICP-MS one will obtain a concentration equivalent to the total sum of Sb species/forms. Furthermore, depending on the sampling approach, water samples can contain both “truly” dissolved and colloidal phases (i.e., determined by the filter mesh-size in field studies). This means that, apart from the dissolved species, Sb complexed with natural organic matter (NOM) can also be present in the sample, for which little is yet known regarding their relevance in Sb natural cycling and the associated analytical artifacts (Filella and Rodríguez-Murillo, 2021). However, to the best of our knowledge, nobody has verified if the presence of organic components in water samples causes any analyte loss during classical ICP-MS analyses, as observed for other elements like Sn (Pougnet et al., not published). Therefore, pre-treatments like UV-digestions (i.e., commonly used in voltammetry studies to avoid the presence of organic matter during the measurements, Abdou et al., 2020, van den Berg et al., 1991), can provide a simple insight to the proportion and influence of organic vs inorganic element species when analyzing aliquots of the same water sample (UV-treated and non-UV treated) via ICP-MS, even though the efficiency of the UV-digestion has not yet been demonstrated for the case of Sb (Filella and Rodríguez-Murillo, 2021).

This study aims at (i) comparing the analytical performance of the ICP-MS vs the QQQ-ICP-MS for dissolved Sb (Sb_d) concentrations in brackish and seawater matrices at the case study area of the Gironde Estuary (France) to reliably quantify Sb_d concentrations in challenging matrices, (ii) identifying the potential influence of organometallic Sb or Sb-NOM related species on total Sb_d quantification along the salinity gradient via direct vs UV-irradiated replicates of water samples, and (iii) quantifying for the first time for the Gironde Estuary the daily export (net) fluxes of Sb_d to the Atlantic Ocean. To achieve this, we compare water samples collected in the Gironde Estuary during contrasting hydrological conditions with nearby areas, independent from the estuarine influence. This approach will allow to establish reliable routine analyses to better understand the biogeochemical processes concerning Sb_d reactivity, advancing the knowledge of Sb dynamics in estuarine systems and providing a more reliable tool for risk assessment.

2. Experimental

2.3. Analyses

2.3.1. Dissolved Sb

In this work, dissolved Sb concentrations from the four MGTS sampling campaigns and from Lacanau were firstly analyzed directly, here labelled as non-UV-irradiated samples (Sb_x). That is, samples were analyzed by standard additions using monoelemental solutions (Sb SPEX CertiPrep 0.6 % Tart. Acid/Tr. HNO₃) and minimal sample dilution (i.e., factor 1.1). Quantification was performed with high matrix cones, argon gas dilution and kinetic energy dispersion (KED, He-based) mode at the QQQ-ICP-MS (iCAP TQ Thermo®). In the lack of estuarine or seawater certified reference materials (CRM) for Sb, a freshwater CRM (TMRAIN 23.4; N = 2) was used to verify the instrument methodology, obtaining good recoveries of 105 %. Analytical repeatability was followed-up at each analytical series (N = 8) with the coastal seawater sample from Lacanau, showing good precision with 9 % relative standard deviation (RSD). In addition, these analyses of Lacanau water were used to evaluate the instrument stability concerning direct matrix injection, obtaining immediate performing responses from the first analysis for Sb (10 % RSD). Field blanks (N = 3, Table 1) showed low onsite contamination and/or low filter influence (i.e., 0.2 μm Minisart® cellulose acetate) on dissolved Sb concentrations, showing average concentrations of 0.07 ± 0.08 nM (8.50 ± 9.20 ng L⁻¹). The complete raw data of the field blanks is shown in Table S1 (Supplementary data).

In addition to Sb_x, aliquots of the previously filtered and stored samples from Lacanau, MGTS I and III were subjected to UV-irradiation pretreatment, thus here called UV-irradiated samples (Sb_tot). For this, volumes (< 10 mL) from the PP bottles were transferred to acid-washed Teflon bottles, after rinsing three times with the sample. These replicates were left over night in a UV box (i.e., UV lamp of 254 nm and 40 W) and analyzed the day after following the QQQ-ICP-MS procedure for dissolved Sb quantification via standard additions. The duration of the UV irradiation on the samples is considered appropriate given that van den Berg et al. (1991) also applied successfully UV–photolysis to 50 mL sample volume for 4 h to analyze Sb in the Tamar Estuary. Blanks to account for potential contaminations from the UV-irradiation procedure (N = 6, Table 1) were prepared with ∼10 mL of deionized water (Milli-Q® Millipore) and subjected to the UV-treatment in the Teflon bottles. Results showed relatively high variations of Sb concentrations, i.e., 0.15 ± 0.13 nM (18.8 ± 16.4 ng L⁻¹). The full raw dataset from the UV-treatment blanks is also shown in Table S1.

The obtained results from Sb_x and Sb_tot at the QQQ-ICP-MS will be compared and discussed to previously quantified Sb_x aliquots of the same filtered and stored samples of MGTS I-III, Lacanau and Comprian (Table 1), analyzed and published in Gil-Díaz et al. (2016). The latter study quantified Sb from isohaline diluted samples through isotopic dilution (ID, Eq. (1) adapted from Castelle, 2008) with ¹²³Sb solution (99.43 % purity, Oakridge, USA) and Ar-gas dilution from external PC3 ESI system coupled to ICP-MS (X-Series Thermo®). To provide further insights on the comparison between the QQQ-ICP-MS and PC3-ICP-MS techniques, aliquots from MGTS I were further quantified via ID (as in Gil-Díaz et al., 2016) with the QQQ-ICP-MS (Table 1).

3. Results and discussion

3.3. Reactivity in the Gironde Estuary

Dissolved Sb along the salinity gradient of the Gironde Estuary showed variable concentrations ranging between 0.82 nM (100 ng L⁻¹) in the freshwater endmember to 3.61 nM (440 ng L⁻¹) in mid-salinities, obtaining 1.64 nM (200 ng L⁻¹) for the seawater endmember (Fig. 4a,b). The results on Sb endmembers are consistent with previous studies in the freshwater ranges of the fluvial Gironde Estuary. That is, the freshwater endmembers match known median concentrations of 1.31 ± 0.41 nM (160 ± 50 ng L⁻¹) at La Réole (2003–2016; Gil-Díaz et al., 2018) and, even more precisely, with the corresponding concentrations of the sampling month: ∼0.82 nM (∼100 ng L⁻¹) in MGTS I (Fig. 4a) and ∼ 1.15 nM (∼140 ng L⁻¹) in MGTS III (Fig. 4b). Likewise, the seawater endmembers at the estuary mouth, and those from the Arcachon Bay water used in repeatability tests, are also in accordance with literature average seawater concentrations of 1.51 ± 0.37 nM (184 ± 45 ng L⁻¹; Filella et al., 2002a).

Marine Chemistry, Volume 267, November 2024, 104465: Fig. 4. Geochemical profiles along the salinity and turbidity gradients of the Gironde Estuary. Distribution of antimony (Sb; a,b) dissolved concentrations, as well as suspended particular matter (SPM; c,d) along the salinity gradient of the Gironde Estuary during intermediate/high (MGTS I - 1203 m3 s−1 and MGTS II - 3450 m3 s−1; a,c) and low (MGTS III - 248 m3 s−1 and MGTS IV - 235 m3 s−1; b,d) freshwater discharges. Both non-UV-irradiated samples (Sbx; filled circles) and UV-irradiated replicates (Sbtot; filled triangles) are represented when applicable. Error bars (in grey) show standard deviations of total Sb concentrations from individual isotope standard additions. The theoretical range of 1.51 ± 0.37 nM (184 ± 45 ng L−1, cross; Filella et al., 2002a) for seawater is also included.

However, Sb trends along the salinity gradient suggest that the reactivity of Sb in the Gironde Estuary is complex. That is, Sb reactivity shows an additive behavior that varies in intensity with contrasting hydrological conditions (i.e., there are statistically significant differences between Sb concentrations in MGTS I and MGTS III, Fig. S2). During intermediate discharge conditions (1203 m³ s⁻¹, Fig. 4a), Sb_tot doubles its freshwater concentration between 0 < S < 10 and then remains stable at 1.64 nM (∼200 ng L⁻¹). A similar trend was observed in the Tamar River estuary, a macrotidal estuary with a defined MTZ of max. 3 g L⁻¹ and observed anoxic conditions (van den Berg et al., 1991), i.e., similar hydrological characteristics than the Gironde Estuary. In the Tamar River, the rapid Sb increase at S < 10 was explained as a particle release effect from the MTZ or a redox release from interstitial waters. Therefore, given that this Sb behavior appears at different salinity ranges, it seems that there is an additive release of Sb related to common intra-estuarine biogeochemical processes of macrotidal MTZ-dominated estuaries, other than pure effects from the salinity gradient. This is in accordance with the little evidence on Sb environmental chloro-complexation for both Sb(III) and Sb(V) species in seawater (Filella et al., 2002b).

During low water discharge conditions (248 m³ s⁻¹ and 235 m³ s⁻¹, Fig. 4b), Sb_tot showed a reproducible desorption/dilution bell-shape behavior along the salinity gradient of the Gironde Estuary. More precisely, Sb showed an M-shaped distribution along the salinity gradient, with maximum concentrations at S ∼ 10 and S ∼ 22 (Fig. 4b). Similar M-shaped patterns have been observed in the Scheldt Estuary and the Savannah Estuary. The Scheldt Estuary is also a macrotidal estuary with a MTZ of 400 mg L⁻¹ and an oxygen maximum zone at low S, i.e., approaching Gironde Estuary drought conditions where average 100–300 mg L⁻¹ Suspended Particulate Matter (SPM) are present along the salinity gradient. The Scheldt Estuary only presented M-shaped Sb_d distributions during a sampling campaign in September 1982 and not in other winter campaigns during February 1975 and October–November 1978 (van der Sloot et al., 1985). Likewise, the Savannah Estuary is a mesotidal continent-ocean transition system presenting reducing conditions, with also maximum Sb concentrations at S ∼ 10 and S ∼ 22 for a sampling campaign during June 1986 (Byrd, 1990). Therefore, the same estuary can present different Sb reactivity trends along the salinity and turbidity gradients according to intra-estuarine seasonal dynamics.

Noteworthy, Byrd (1990) compared Sb concentrations to SPM, oxygen levels and inorganic nitrogen (i.e., nitrate + nitrite + ammonia concentrations) along the Savannah Estuary salinity gradient. He observed a decrease of Sb from the freshwater endmember to S ∼ 2.5, related to Sb removal by association with iron flocculating colloids or a source variation effect and then explains the observed M-shape as follows: (i) a first Sb increase at S ∼ 10 matching the lowest oxygen saturation and maximum nitrogen concentrations, i.e., related to organic matter remineralization or redox-conditioned Sb solubilization from SPM, (ii) a consequent decrease until S ∼ 17 related to water column mixing, and (iii) a second Sb increase at S ∼ 22 matching also a second inorganic nitrogen peak, i.e., suggesting a common link/source between Sb and the nitrogen cycle, before (iv) being quickly removed or diluted towards the seawater endmember. In fact, open ocean studies have also observed increased Sb concentrations in inorganic nitrogen rich areas like in upwelling systems, suggesting biotic uptake in surface waters and regeneration in depth (Cutter and Cutter, 1995). Thus, our results could suggest that the decrease in Sb at S ∼ 17 can correspond to Sb sorption onto SPM (Fig. 4d) and/or a transformation into organo-Sb species (see Section 3.2.), potentially from biofilm or periphyton microorganisms during low discharge conditions (Fig. 4b). In any case, additional parameters such as chlorophyll-a and phaeopigments (Fig. S1) do not provide any insights to clarify this hypothesis, thus further, complementary information (e.g., NO₃⁻ concentrations) should be collected in future field campaigns.

4. Conclusion

Overall, this work provides guidelines for good analytical quantification of Sb in challenging water matrices, improving current and future understanding of Sb biogeochemical cycles in continent-ocean transition systems.

Given the lack of estuarine/seawater CRM for dissolved Sb, results show a better analytical performance of the QQQ-ICP-MS technique, validating its use for direct routine analyses of dissolved Sb in brackish/seawater matrices. Direct quantification of acidified water samples (Sb_x) alone can be considered as a general good approach for total Sb concentrations (i.e., UV pre-treatment can introduce artifacts). Thus, we report for the first time since the year 2000, analytically sound values for the Atlantic seawater endmember, agreeing with the established range identified in Filella et al. (2002a). Furthermore, QQQ-ICP-MS efficiently removes site-dependent interferences, though their identity has not yet been identified for the Gironde Estuary mouth. This means that, other estuarine systems might show similar, yet unidentified, site-dependent effects, which can be easily verified/addressed by collecting coastal water samples nearby the study site, not influenced by the estuarine system.

Updated profiles of the non-conservative behavior of Sb in the Gironde Estuary allow to confidently calculate daily net/gross fluxes, highlighting a geogenic/natural predominance of Sb reactivity in the area, sensitive to complex redox and biological processes. This reactivity seems independent of major chloro-complexing effects, contrary to other trace elements like cadmium (Dabrin et al., 2009) or silver (Lanceleur et al., 2013), and shows a seasonal variability, characteristic of specific estuarine morphologies and hydrodynamics. This means that Sb reactivity is not stationary and that future works on estuarine systems should sample more than once in a lifetime (no more single campaigns, as evidenced in the literature), reporting several parameters including the month of sample collection, water discharges, SPM/nutrient load and residence times, whenever possible.