Grace N.
Manestar
*ac,
Hilary M. K.
Lewis
bc,
Alex
McCoy-West
c,
Nishen
Naidoo
d,
Stefan
Makart
e,
Ondrea
Thompson
e and
Brandon
Mahan
ac
aMelbourne Analytical Geochemistry, School of Geography, Earth and Atmospheric Sciences, University of Melbourne, Melbourne, Australia. E-mail: grace.manestar@student.unimelb.edu.au
bResearch School of Earth Science, College of Science, Australian National University, Canberra, Australian Capital Territory, Australia
cIsoTropics Geochemistry Lab, Earth and Environmental Science, James Cook University, Townsville, Queensland 4811, Australia
dThermo Fisher Scientific, Level 1/4 Talavera Rd, North Ryde, NSW 2113, Australia
eThermo Fisher Scientific, 5 Caribbean Dr, Scoresby, VIC 3179, Australia
First published on 6th June 2025
Radiogenic strontium (87Sr/86Sr) is a powerful tool for characterizing and differentiating water reservoirs (among many other applications). The development and improvement of high-precision analytical platforms (namely MC-ICP-MS) has enhanced throughput for isotope ratio determination. However, analyte purification—needed to remove isobaric interferences—continues to occur largely via conventional manual gravity-driven ion exchange chromatography (hereafter: manual IEC), which generally cannot match instrument throughput. This has created a persistent throughput gap that encumbers use and proliferation, emphasizing the need for rapid separation of Sr, and of comprehensive, end-to-end high throughput workflows and analytical approaches that are fit-for-purpose. Here we have developed a workflow-optimized protocol for sample preparation and separation of Sr from natural water samples using both workflow-optimized manual IEC and automated high pressure ion chromatography (HPIC), for subsequent analysis via MC-ICP-MS. These methods have been designed to seamlessly integrate with common international practice for water sample collection. The automated HPIC technique accommodates introduction of water samples filtered with standard 0.45 μm membranes and acidified with ultra-high purity nitric acid (HNO3, to pH of 1–2, approximated as 0.09 mol per L HNO3). Filtered and acidified samples are directly introduced into the HPIC system where Sr is separated from other cations (namely Ca) and collected as an isolate in a specific volume of ultrapure water. Strontium isolates, with no further preparation (e.g. dry-down and reflux), are then directly acidified to 0.5 mol per L HNO3 and analyzed by MC-ICP-MS. This technique can process 40–50 samples in a 24 hour period with mitigated potential for human error, matching current MC-ICP-MS analytical capacity, and achieving analytical precision sufficient to distinguish the variability observed in natural samples across many applications.
Applications of 87Sr/86Sr are myriad because this isotopic ratio is not significantly perturbed (fractionated) by equilibrium or kinetic processes—i.e. it is largely “isotopically conservative”—and is therefore a robust tracer of many materials back to their geological/geographical context.11–14 In brief, older and/or more Rb-rich geologic units will have higher 87Sr/86Sr due to the radioactive decay of 87Rb to 87Sr over time; this means that many geologic, and therefore geographic locations, have unique 87Sr/86Sr values relative to adjacent areas, allowing for discrimination between reservoirs. Importantly, 87Sr/86Sr is not significantly affected by secondary processes (weathering, biological uptake, etc.), therefore it is a powerful means to trace materials—solid and aqueous, inorganic and organic—back to their original source; in this respect, it is called a “conservative” isotope ratio. Lastly, due to efficient mixing on geological timescales, the 87Sr/86Sr of the modern global ocean (or indeed any point in time) is a fixed value (currently 0.709165 ± 0.000005 2σ), marking a robust point of difference to other natural waters.15
Water is the most critical resource on the planet, especially in the face of changing climate. As such, it is paramount to have the means and capacity to characterize water reservoirs, their interaction with each other and with their host environment, and to do this at scale. This is true across many applications, e.g. environmental source tracing and impact assessment, and is of particular importance with respect to groundwater-dependent environments, where it is critical not only in itself but in its impact on livestock, agriculture and the like, thereby also impacting food security.16,1787Sr/86Sr ratios are intensely useful for differentiating between aquifers and other water sources, and because the above-mentioned observation regarding a universal 87Sr/86Sr for modern seawater, this tool is immensely useful in applications such as constraining seawater ingress to freshwater systems, and differentiation freshwater reservoirs (e.g. aquifers) hosted in differing lithologies.18–22 Further, 87Sr/86Sr has shown utility in tracking brine migration related to carbon dioxide injection for carbon storage,23,24 as well as in being a reliable proxy for bioavailable Sr12,20 and therefore also viable for developing large-scale 87Sr/86Sr “isoscapes” for modern and paleo-forensics (the former notably including food security applications).8,25–29 In all, this highlights 87Sr/86Sr as incredibly useful across a wide variety of applications, and strongly aligned to many of the UN's Sustainability Development Goals (e.g. SDGs 2, 3, 6, 9, 11–15). This marks 87Sr/86Sr of great utility in the march towards a sustainable future, while underscoring a need for pan-applicable, highly adaptable/adoptable and high-throughput protocols for its determination by mass spectrometry.
Characterization of 87Sr/86Sr in natural water samples is typically done via conventional time- and resource-intensive methodologies thus hindering widespread deployment of this tool, making large-scale projects very workload-intensive and therefore scarce and/or susceptible to failed follow-through (long-term abandonment). Several studies30–33 in the past have addressed this throughput gap, with most focusing on accelerating and/or automating the Sr separation aspect of the overall workflow. Of note, to the best of our knowledge all such studies have dried down collected water samples and refluxed in dilute acid prior to Sr separation chemistry.3,23,34–36 One exception is Meynadier et al. (2006) who directly injected filtered and acidified samples, however their automated separation protocol utilized an isocratic elution scheme (same eluent concentration throughout), and subsequent analyses via TIMS.
In addition, other studies have conducted direct measurements of 87Sr/86Sr ratio in bulk samples, i.e. with no Sr separation by direct analysis using MC-ICP-MS37,38 and TIMS (e.g. Li et al., 2015).39 However, these methods (Ehrlich et al., 2001, Yang et al., 2011, Li et al., 2015) without Sr separation are not suitable for water sample with high Ca/Sr (>700), and further, Ehrlich et al., (2001) and Yang et al., (2011) note that the samples are introduced through the use of a nebulizer, which could have potential long-term issues and/or during long analytical sequences, due to matrix build-up on introduction system parts over time. Furthermore, it is generally not advisable to introduce bulk water samples into MC-ICP-MS, as the presence of major cations (e.g. Na, Ca, Mg, K), major anions (e.g. Cl, F), and residual organics (and S), in appreciable abundances in natural waters all pose significant risks of contaminating or otherwise degrading the instrument's introduction system and peripherals (autosampler systems, tubing, torch assembly, cones, etc.), and cause accelerated degradation of the detector arrays.40
Karasinski et al. (2016) introduces an innovative on-line isotopic determination of 87Sr/86Sr by coupling ion chromatography (IC, Thermo Scientific Dionex ICS-5000+) and MC-IC-PMS (Nu Plasma II), which eliminates matrix effects and achieves precision comparable to traditional offline approaches. However, the paper discusses that the required introduction system configuration leads to significantly increased analysis time and significant peak broadening (entire analysis time is about 14 min; as opposed to 5–8 minutes herein), potentially counter to a key goal of increasing overall throughput.
In summary, 87Sr/86Sr is a powerful geo-analytical tool across many applications and is of acute utility in the characterization of water reservoirs and their co-interaction. While current analytical platforms (e.g. MC-ICP-MS) allow for the generation of large datasets via relatively high throughput, laboratory methods have lagged and are still largely conducted by manual IEC. Prior research that has explored accelerated and/or automated protocols for Sr (and other elements) separation31,32 have done so typically with specific applications and/or academic pursuits in mind, leading to non-comprehensive and non-generalized streamlining of methodological workflow(s).
Here, we have drawn on the literature to incorporate and innovate upon these advances to define a comprehensive protocol and workflow for Sr separation from natural water samples (ranging from fresh to seawater) and improved upon methodologies for both conventional and automated Sr separation chemistry. These methods have been validated by analyses of international Certified Reference Materials (CRMs) as well as a cohort of natural groundwater (GW) samples, all with comparative 87Sr/86Sr analytics determined via manual IEC. Moreover, the techniques developed herein have been developed within the context of international water collection protocols to be able to widely accept samples to the automated workflow by common international water sampling practices,41 thereby ensuring the viability of seamless integration into industry, government and commercial/consultancy protocols and guidelines for water sampling, “future-proofing” these technique for the expanded utilization of 87Sr/86Sr on the horizon.
Lastly, an in-house 7-cation standard was developed from single element standards from High Purity Standards™ 1000 μg mL−1 (Li, Na, K, Mg, Rb, Ca, and Sr) for use in refining the automated chromatography procedure for optimal Sr separation.
Ultrapure Milli-Q® IQ Element system water, HNO3 and ultra-pure double-distilled hydrochloric acid (HCl) were the only non-commercial reagents used in this work. All Polypropylene (PP) labware including vials, beakers, and pipette tips, were thrice rinsed with Milli-Q water before immersion in covered vessels containing 2 mol per L (M) HNO3 and leached for 2–4 weeks. Ultra-pure perfluoroalkoxy (PFA) vessels (Savillex®) were cleaned carefully with PP brushes with Decon-90™ lab detergent using heated Milli-Q, then batch cleaned in 6 M HNO3 for 24 hours on a hotplate at 120 °C. Batch cleaned PFA vessels were then individually filled with 16 M HNO3 (1/4 full) and heated at 120 °C for 24 hours, followed by 6 M HCl heating at 120 °C for 48–72 hours. All PFA vessels were rinsed with Milli-Q three times in between cleaning steps. Finally, prior to use, all labware (PP and PFA) undergo thrice rinsing with Milli-Q water and subsequent drying under positive pressure HEPA-filtered air (Thermo HeraGuard® Biobench™) inside an ISO-7 clean laboratory.
Chromatographic columns were made from standard 1 mL pipette tips (Eppendorf e.p. Tips® or Thermo Scientific™ FinnTip™), wherein approximately 0.5 cm of the tip was cut off diagonally (to promote flow and reduce “dead volume” below frit), and a pre-fabricated 4 mm circular polyethylene (PE) frit (30 μm pore-size, 3 mm thickness; BioComma™) was fitted into the tapered end. After assembly, columns are sonicated to remove debris and rinsed thrice in Milli-Q® water, then placed in 2 M HNO3 for leaching and storage until use. Sr-spec resin (50–100 μm particle size) was batch cleaned thoroughly before use, by loading approximately 10 mL resin (enough for >100 columns) as a water slurry into 20 mL Bio-Rad Poly-Prep® column with a 225 mL Econo-Pac reservoir attached. Full column volumes of 0.05 M HNO3, 6 M HCl, 3 M HNO3 and Milli-Q water were passed in succession (at least thrice) to batch clean the resin. This step removes labile organic compounds and minimizes Sr blank contribution from the resin. The cleaned resin is stored as a slurry in dilute HNO3 (ca. 0.01 M) until use. Leached and rinsed (3×) pipette tip columns are then filled with approximately 75–100 μL of Sr-Spec™ resin. The resin is pre-cleaned in the columns by passing of two column volumes (approx. 2 mL) each of Milli-Q water, 3 M HNO3, Milli-Q water, then conditioned with 1 mL of 3 M HNO3. Strontium separation is achieved by loading the column with samples of pre-defined volumes (to achieve ∼600 ng Sr for groundwater samples and 1000 ng Sr for water standards), after which three column volumes of 3 M HNO3 is passed to remove matrix elements, notably Rb and Ca, which are crucial elements to be removed prior to analysis. Following matrix elution, clean PFA vessels (7 mL, Savillex®) are placed under the columns and Sr is eluted from the column in two column volumes of Milli-Q water. The Sr fraction is then dried and refluxed in 1.5 mL of 0.5 mol per L HNO3 for subsequent analysis by MC-ICP-MS.
![]() | ||
Fig. 1 Schematic illustration of the Dionex ICS-6000 automated HPIC system. Arrows represent the flow path. |
The detector and chromatography module are split into two units, the detector upper unit houses the Dionex™ DRS 600 (2 mm) dynamically regenerated suppressor and the conductivity detector. The auxiliary pump (AXP pump) delivers Milli-Q water to the suppressor to help hydrate and assist the removal of MSA from the water sample, i.e. after passage through the eluent suppressor, only single element isolates in pure water pass downstream to the conductivity detector, subsequently collected by the fraction collection. The lower chromatography unit houses the two columns (guard column, analytical column) and a 6-port valve with the 100 μL sample loop; the sample loop is where the eluent and the sample are mixed before injected into the columns. In the configuration selected herein, a Dionex IonPac CG16 guard column (3 × 50 mm) is placed before the Dionex IonPac CS16 separation column (3 × 250 mm); the guard column prevents sample contaminants from eluting into the main separation column, and the internal volume of the separation column is selected based on injection volume (here, the maximum 100 μL loop volume was utilized). While not within the scope of the present work, prior research has shown that, depending on the specific application or sample matrices (high matrix may increase degradation), these cation columns can last for several hundreds to a few thousand runs.31 Additionally, the DC module is individually temperature controlled by separate heating units to maintain peak stability. Overall, individual modules are connected by inert polyether ether ketone (PEEK) tubing and the whole system is controlled by the Chromeleon™ CDS version 7.2 software.
Briefly, each water sample is introduced as a 100 μL injection into the Dionex ICS-6000. The sample is acidified with MSA and then moves through the columns where the cations are separated, then into the suppressor where H+ ions from the sample matrix are electrolytically removed (this step removes MSA for conductivity detection), delivering MSA-free sample cation solutions to the conductivity detector according to their time-dependent retention on the chromatographic column. Individual cations (i.e. Ca or Sr) will pass through the conductivity detector one at a time (due to different charge and element radii volume, and therefore retention time on the analytical column), wherein the detector sends a signal to the Chromeleon™ software, generating a live read-out of conductivity; this provides real-time elution profile monitoring. With the addition of the fraction collector, once the time window is defined for the selected ion—in this case Sr—the system can be programmed for optimal timing of isolate collection; this is essentially a fixed volume collection, as flow rate is a set constant of the methodology.
Because Sr isolates from fraction collection are devoid of MSA, and of a well-constrained volume (defined by time-window of collection and constant flow rate), in this case producing 0.36–0.54 mL of purified Sr in water. Samples can be directly acidified to 0.5 mol per L HNO3 using high-purity 16 M HNO3 with no dry-down/reflux and are thereafter ready for isotopic analysis by MC-ICP-MS following vortex agitation after approximately five minutes (allowing for equilibration of dilute HNO3 solution).
To monitor long-term instrument stability, the High Purity Standard (HPS) Sr was analyzed in every analytical session, yielding an average 87Sr/86Sr of 0.707618 ± 0.000012 (2SD, n = 9). 7 cation standards reported an average of 0.708736 ± 0.000002 (2SD, n = 2).
SRM987 through the automated system, yielding 0.710244 ± 0.000033 (2SD, n = 6), in excellent agreement with previous data via multi-collector using similar automated system. For reference, Meynadier et al. (2006) reported 0.710277 ± 0.000028 (2SD) and Karasinski et al. (2016) reported 0.71025 ± 0.00003 (2SD).
Sample | ICP-OES | TQ-ICPMS | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Ca | SD | K | SD | Mg | SD | Na | SD | Sr | SD | Ca/Sra | |
a Ca/Sr ratio reported as a fractional percentage of the ppm values. b Uncertainties for elemental concentrations are reported as standard deviation (SD) and all data reported in ppm. | |||||||||||
GW1 | 100 | 0.1 | 11.4 | 0.1 | 82.7 | 0.1 | 649 | 0.9 | 1230 | 33 | 81 |
GW2 | 123 | 0.1 | 2.8 | 0.1 | 122.4 | 0.2 | 407 | 0.2 | 1179 | 20 | 104 |
GW3 | 63 | 0.2 | 6.3 | 0.1 | 56.6 | 0.2 | 222 | 0.4 | 508 | 18 | 123 |
GW5 | 119 | 0.3 | 1.0 | 0.1 | 64.1 | 0.2 | 456 | 1.3 | 1100 | 15 | 108 |
GW6 | 101 | 0.4 | 4.2 | 0.1 | 65.4 | 0.3 | 375 | 1.6 | 735 | 9 | 137 |
GW7 | 282 | 0.5 | 17.7 | 0.2 | 272.3 | 0.6 | 1384 | 2.2 | 3459 | 97 | 82 |
GW8 | 282 | 0.9 | 17.5 | 0.2 | 273.4 | 1.0 | 1385 | 5.8 | 3481 | 73 | 81 |
GW9 | 57 | 0.1 | 2.5 | 0.1 | 38.9 | 0.1 | 295 | 0.3 | 499 | 4 | 115 |
GW10 | 57 | 0.1 | 2.5 | 0.1 | 38.5 | 0.1 | 293 | 0.3 | 492 | 12 | 115 |
GW11 | 57 | 0.1 | 4.2 | 0.1 | 57.2 | 0.2 | 192 | 0.3 | 407 | 4 | 139 |
GW12 | 56 | 0.0 | 4.0 | 0.2 | 56.6 | 0.1 | 189 | 0.2 | 418 | 6 | 134 |
GW13 | 50 | 0.1 | 5.1 | 0.1 | 62.9 | 0.2 | 342 | 0.6 | 499 | 5 | 100 |
GW14 | 49 | 0.5 | 5.1 | 0.0 | 62.5 | 0.4 | 340 | 1.6 | 499 | 4 | 99 |
GW15 | 49 | 0.3 | 5.1 | 0.3 | 62.4 | 0.3 | 339 | 1.4 | 505 | 6 | 97 |
GW16 | 47 | 0.2 | 6.1 | 0.1 | 38.9 | 0.1 | 154 | 0.3 | 702 | 9 | 67 |
GW17 | 119 | 0.3 | 1.0 | 0.1 | 64.9 | 0.0 | 449 | 0.4 | 1669 | 40 | 71 |
GW19 | 82 | 0.2 | 1.1 | 0.1 | 69.0 | 0.2 | 183 | 0.3 | 712 | 9 | 116 |
GW20 | 119 | 0.2 | 5.8 | 0.1 | 104.6 | 0.3 | 736 | 2.8 | 1746 | 51 | 68 |
GW21 | 86 | 0.1 | 5.2 | 0.1 | 57.4 | 0.1 | 326 | 0.2 | 905 | 11 | 95 |
GW22 | 71 | 0.2 | 2.8 | 0.2 | 63.0 | 0.2 | 179 | 0.5 | 967 | 11 | 74 |
GW23 | 71 | 0.1 | 2.9 | 0.1 | 62.8 | 0.0 | 179 | 0.2 | 969 | 14 | 73 |
GW24 | 69 | 0.1 | 2.6 | 0.0 | 60.4 | 0.1 | 172 | 0.2 | 952 | 28 | 72 |
Q-ICP-MS | |||||||||||
CASS-6 | 373 | 0.89 | 6.8 | 0.46 | 55 | ||||||
NASS-7 | 386 | 2.07 | 7.0 | 0.86 | 56 |
Sample | IEC | HPIC | Difference | Relative | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
85Rb(V) | 86Sr(V) | 84Sr/86Sr | 2SE | 87Sr/86Sr | 2SE | 85Rb(V) | 86Sr(V) | 84Sr/86Sr | 2SE | 87Sr/86Sr | 2SE | (×106) | Offset (%) | |
a 87Sr/86Sr are reported as two times the internal standard error (2SE) from 50 cycles of sample signal collection, or the 2SD of the population in the case of standards used for constraining long-term reproducibility across multiple analytical sessions. 85Rb and 86Sr (volts) were monitored on the Neptune for each sample to constrain Rb/Sr in analyte solutions. | ||||||||||||||
GW1 | 2.36 × 10−4 | 5.35 × 10−1 | 0.056346 | 47 | 0.716068 | 26 | 1.78 × 10−4 | 8.29 × 10−1 | 0.05635 | 37 | 0.715905 | 21 | 164 | −2.28% |
GW2 | 9.49 × 10−5 | 7.93 × 10−1 | 0.055554 | 32 | 0.709534 | 17 | 7.56 × 10−5 | 1.45 × 100 | 0.05643 | 29 | 0.709454 | 22 | 80 | −1.13% |
GW3 | 1.97 × 10−4 | 1.47 × 100 | 0.054902 | 18 | 0.711298 | 17 | 5.23 × 10−4 | 7.94 × 10−1 | 0.05639 | 56 | 0.711309 | 41 | −12 | 0.17% |
GW5 | 3.17 × 10−4 | 7.58 × 10−1 | 0.055641 | 36 | 0.709942 | 20 | 7.16 × 10−5 | 1.05 × 100 | 0.05638 | 26 | 0.709865 | 18 | 77 | −1.08% |
GW6 | 1.36 × 10−4 | 1.91 × 100 | 0.056420 | 15 | 0.713436 | 16 | 1.66 × 10−4 | 6.93 × 10−1 | 0.05633 | 41 | 0.713558 | 20 | −123 | 1.72% |
GW7 | 2.40 × 10−4 | 9.50 × 10−1 | 0.055328 | 27 | 0.715823 | 12 | 2.31 × 10−4 | 2.60 × 100 | 0.05645 | 12 | 0.715743 | 9 | 80 | −1.12% |
GW8 | 1.75 × 10−4 | 7.57 × 10−1 | 0.055653 | 27 | 0.715869 | 22 | 2.56 × 10−4 | 3.15 × 100 | 0.05646 | 8 | 0.715735 | 10 | 134 | −1.87% |
GW9 | 1.10 × 10−4 | 6.36 × 10−1 | 0.055838 | 45 | 0.713999 | 22 | 8.36 × 10−5 | 3.24 × 10−1 | 0.05619 | 68 | 0.714040 | 37 | −41 | 0.57% |
GW10 | 1.31 × 10−4 | 6.57 × 10−1 | 0.055815 | 37 | 0.713982 | 20 | 6.39 × 10−5 | 4.41 × 10−1 | 0.05617 | 72 | 0.713999 | 32 | −17 | 0.23% |
GW11 | 4.70 × 10−4 | 6.58 × 10−1 | 0.055109 | 29 | 0.710439 | 19 | 5.24 × 10−5 | 2.89 × 10−1 | 0.05607 | 100 | 0.710622 | 40 | −183 | 2.58% |
GW12 | 1.13 × 10−3 | 7.24 × 10−1 | 0.055006 | 32 | 0.710465 | 23 | 2.36 × 10−5 | 3.09 × 10−1 | 0.05613 | 86 | 0.710583 | 37 | −119 | 1.67% |
GW13 | 1.49 × 10−4 | 1.76 × 100 | 0.055299 | 19 | 0.713417 | 13 | 1.90 × 10−4 | 3.45 × 10−1 | 0.05622 | 86 | 0.713632 | 39 | −215 | 3.01% |
GW14 | 1.74 × 10−4 | 1.35 × 100 | 0.055636 | 19 | 0.713451 | 15 | 1.83 × 10−4 | 3.24 × 10−1 | 0.05622 | 92 | 0.713659 | 36 | −209 | 2.93% |
GW15 | 1.45 × 10−4 | 1.12 × 100 | 0.055918 | 23 | 0.713483 | 14 | 4.90 × 10−4 | 3.13 × 10−1 | 0.05622 | 89 | 0.713724 | 31 | −241 | 3.38% |
GW16 | 1.97 × 10−4 | 1.81 × 100 | 0.055276 | 14 | 0.710495 | 13 | 1.38 × 10−4 | 5.18 × 10−1 | 0.05626 | 47 | 0.710666 | 28 | −171 | 2.41% |
GW17 | 8.88 × 10−5 | 1.77 × 100 | 0.055323 | 14 | 0.709783 | 10 | 6.73 × 10−4 | 1.27 × 100 | 0.05641 | 22 | 0.709874 | 13 | −90 | 1.27% |
GW19 | 1.03 × 10−4 | 1.71 × 100 | 0.055184 | 15 | 0.710320 | 12 | 1.36 × 10−4 | 3.67 × 10−1 | 0.05623 | 73 | 0.710530 | 32 | −210 | 2.96% |
GW20 | 3.63 × 10−4 | 2.64 × 100 | 0.055185 | 9 | 0.713835 | 9 | 3.34 × 10−4 | 1.15 × 100 | 0.05643 | 23 | 0.713925 | 17 | −90 | 1.25% |
GW21 | 2.59 × 10−4 | 1.94 × 100 | 0.055441 | 14 | 0.713288 | 12 | 1.71 × 10−4 | 4.65 × 10−1 | 0.05625 | 58 | 0.713477 | 29 | −189 | 2.65% |
GW22 | 2.49 × 10−4 | 2.13 × 100 | 0.055344 | 12 | 0.710003 | 13 | 8.12 × 10−5 | 6.32 × 10−1 | 0.05639 | 45 | 0.710121 | 26 | −118 | 1.66% |
GW23 | 1.18 × 10−4 | 2.17 × 100 | 0.055314 | 11 | 0.709990 | 12 | 1.57 × 10−4 | 5.89 × 10−1 | 0.05634 | 43 | 0.710117 | 25 | −127 | 1.79% |
GW24 | 9.39 × 10−5 | 2.27 × 100 | 0.055256 | 12 | 0.709972 | 9 | 6.58 × 10−5 | 5.72 × 10−1 | 0.05632 | 39 | 0.710128 | 26 | −156 | 2.20% |
Standards | 2SD (n) | 2SD (n) | ||||||||||||
SRM987 | 0.710244 | 33 (6) | ||||||||||||
7CS | 0.708736 | 2 (2) | ||||||||||||
CASS-6 | 0.709196 | 39 (2) | 0.709302 | 57 (4) | −106 | 1.49% | ||||||||
NASS-7 | 0.709186 | 70 (2) | 0.709351 | 44 (4) | −165 | 2.33% |
CASS-6 | NASS-7 | SRM987 | |
---|---|---|---|
a Manual IEC. b Automated HPIC. | |||
This study | 0.709196 ± 39a | 0.709186 ± 70a | |
0.709302 ± 57b | 0.709351 ± 44b | 0.710244 ± 33b | |
Phan et al. (2021)45 | 0.70918 ± 6 | ||
Plechacek et al. (2022)46 | 0.709165 ± 16 | ||
Zaky et al. (2018)47 | 0.709167 ± 9 | ||
De Muynck et al. (2009)48 | 0.710260 ± 67 | ||
Thirlwall (1991)49 | 0.710248 ± 11 | ||
Meynadier et al. (2006)32 | 0.710277 ± 28 | ||
0.710243 ± 2 | |||
McCoy-West et al. (2016)42 | 0.710242 ± 12 | ||
Balcaen et al. (2005)50 | 0.710251 ± 13 | ||
Galler et al. (2008)51 | 0.71030 ± 22 |
The Ca/Sr and Na (salinity) contents of samples within this study vary considerably, highlighting the spectrum of water matrices investigated herein. Strontium concentrations of groundwater samples ranged from 407 to 3481 ppb and show no relationship with Sr isotope compositions. Groundwater 87Sr/86Sr ratio from the manual IEC range from 0.709534 to 0.716068 while automated HPIC range from 0.709454 to 0.715905.
The analytical uncertainties reported in this study for samples processed through HPIC are, on average, only ∼10 ppm (i.e. 0.000010 for 87Sr/86Sr) higher than those obtained using conventional manual IEC, despite omitting several steps such as sample dry-down and reflux, which offer mitigation to analytical artefacts caused e.g. by residual organics from the sample or resin. This demonstrates the efficacy of the automated HPIC protocol in maintaining reproducibility while significantly streamlining sample preparation and Sr separation chemistry. Given that external reproducibility for natural samples typically exceeds 20 ppm13,23,37,52,53 and that Sr isotope values for applications such as aquifer differentiation and freshwater-seawater mixing commonly vary at the 3rd or 4th decimal place (0.00× to 0.000×, i.e. hundreds to thousands of ppm), the uncertainties observed herein are well within acceptable limits for the intended applications. While precision remains inherently constrained by sample limitations, including low analyte Sr concentrations in natural waters (a consequence of automated system injection only 100 microliters of sample per injection), the correlation between measurement offsets and 86Sr signal intensity using Pearson's correlation analysis (R2 = 0.5844) (excluding GW3, GW9, and GW10, analyses with exceptional poor reproducibility due to low analyte signal) suggests that low Sr intensities during measurement account for 58.4% of the difference in results obtained using manual IEC versus automated HPIC protocols. These findings reinforce that the automated HPIC method is a viable alternative to manual IEC for Sr separation for high throughput scalable applications, where rapid sample processing and differentiation are prioritized over lowest possible analytical uncertainties.
Herein, 85Rb voltages (as a proxy for 87Rb) in MC-ICP-MS analyte solutions were monitored for both manual IEC and HPIC methodologies, as 87Rb is a direct analytical interferent on 87Sr. Average 85Rb voltages for the HPIC method were on average slightly lower than that for manual IEC (0.197 mV and 0.235 mV, respectively), indicating excellent separation of Rb and Sr by HPIC; however, it is noted that 85Rb voltages of analyte solutions are cumulative and incorporate Rb not only from separation chemistry, but from subsequent preparation and dilution prior to isotopic analysis. In our HPIC procedure, Rb elutes approximately three minutes before Sr, and the fraction collector is only activated during Sr collection, minimizing potential for Rb carryover into Sr isolates. The lower Sr analyte concentrations with HPIC during isotopic analysis (e.g.87Sr on average 60% that by manual IEC) are due to the use 100 μL sample loop (largest possible). This artificially increases the Rb/Sr of MC-ICP-MS analyte solutions (relative to manual IEC). In the current work, this had no statistically significant effect on 87Sr/86Sr (R2 = 0.23 for 87Sr/86Sr difference in manual IEC or HPIC, vs. analyte solution Rb/Sr as 85Rb/86Sr). It is noted that while these factors did not affect 87Sr/86Sr measurements herein, active monitoring of analyte Rb/Sr is recommended and important in all 87Sr/86Sr analyses to ensure no systematic analytical bias is introduced, especially where MC-ICP-MS analyses are conducted at low analyte solution Sr concentrations.
In summary, the manual IEC protocol developed herein reduces sample processing times by up to an entire day and makes it readily viable to process e.g. up to 30 water samples in less than half a day in the lab, with minimal use of consumables and reagents, laying open the possibility of both sample preparation/processing and isotopic analysis within the same day. Furthermore, while vacuum-assisted IEC was outside the scope of the current work, it is a near certainty that the manual IEC protocol developed herein could be easily converted for use with a vacuum manifold system. Lastly, our protocol could be adapted with limited to nil change to that detailed in Wall et al. (2013). Doing so would further reduce the sample processing times by the use of vacuum box and no post-IEC sample dry-down and reflux, making it potentially viable to receive, process and analyze 40–50 water samples for 87Sr/86Sr all within a 24 hour period.
Future work with manual IEC will focus on converting the above methodology to vacuum box systems, as well as to exploring modification of the workflow to allow for direct acidification and analysis of Sr separates after IEC to circumvent the additional rate-limiting step of post-IEC dry-down and reflux for MC-ICP-MS analysis. In particular, future conversion to a vacuum box system will focus on direct placement of pipette tip columns (or similar single-piece column) into the vacuum manifold, further and greatly reducing the number of consumables required. Once the workflow has been fully optimized for natural water samples, it can be readily adapted to other sample media as needed, e.g. in agricultural, geological and forensics applications.
Optimization of the automated technique focused on three major attributes – integration with international water collection protocols, minimization of cross-contamination, and optimized Sr separation efficiency from adjacently eluting and interferent matrix elements (i.e. Ca).
As detailed above, the majority of water collection protocols defined by international industry/government organization specify filtration of field water samples using 0.45 μm water filtration membranes, and acidification of water samples with dilute nitric acid (to pH of approximately 1–2) to arrest growth of organics and/or precipitation of solids.54 Here, all natural groundwater samples were filtered with 0.45 μm membranes and acidified according to volume (i.e. not exact/titrated molarity); this was done in order to simulate the least stringent conditions to develop the automated technique with minimal sensitivity to these likely variables. The two internationally recognized CRMs used (CASS-6 and NASS-7) were processed as is (each already filtered and acidified as part of developer production). Of important note, because of acidification in dilute HNO3, it follows that Sr separation in the protocol developed herein occurs in a mixed media of dilute HNO3 (0.09 mol L−1) and MSA.
Meynadier et al. (2006) noted significant so-called “memory effects” using a predecessor automated system (i.e. contamination of prior samples into current ones in the automated sequence). However, subsequent studies30–32 on more recent platforms like that herein, have shown that, due to advancements in valve construction, cross-contamination can be effectively mitigated or altogether avoided by the injection of blanks (i.e. here 0.09 mol per L HNO3) in between sample injections. This issue was investigated herein to ensure sufficiently low so-called “memory effects” between samples in the HPIC system (hereafter referred to as intra-sample blanks, quantified canonically as ng Sr), even for the most matrix-heavy samples. To address this, a small subset of 5 natural groundwater samples were processed through HPIC Sr separation where either one intra-sample blank (method 1: one blank injection wash between samples) or two intra-sample blanks (method 2: two blank injection washes between samples) 0.09 mol per L HNO3 were introduced in sequence between true samples. Statistical agreement with 87Sr/86Sr determined via manual IEC was better for sample injections preceded by two intra-sample blank injections (Table 4), and the use of two intra-sample blanks sufficiently cleans the HPIC tubing, valve and column circuit (see below) and therefore is the method used throughout the current work. Subsequently, intra-sample blanks were monitored for natural groundwater samples herein, with ng Sr for both the first and second blank injections quantified and reported in the ESI.†
Sample | Method 1 | Method 2 | ||||
---|---|---|---|---|---|---|
Injection 1 | Blank% | Injection 1 | Blank% | Injection 2 | Blank% | |
GW8 | 16.36 | 1.31% | 6.75 | 0.54% | 0.79 | 0.06% |
GW9 | 2.92 | 1.62% | 11.81 | 6.57% | 0.84 | 0.47% |
GW10 | 3.29 | 1.86% | 1.52 | 0.86% | 0.28 | 0.16% |
GW11 | 2.27 | 1.55% | 1.41 | 0.96% | 0.23 | 0.15% |
GW12 | 1.89 | 1.26% | 1.63 | 1.08% | 0.11 | 0.07% |
Table 4 presents Sr levels in intra-sample blanks (ng) (“memory effects”) between groundwater samples using the automated HPIC separation method. In method 1, where a single intra-sample blank was injected between samples, Sr levels were notably higher (e.g., 16.4 ng for GW8 and 3 for GW9), leading to potential carryover. In contrast, method 2, which included two blank injections to clean the chromatographic column, significantly reduces Sr levels by the second injection (e.g., down to 0.79 ng for GW8, and in general well below 0.5 ng Sr), demonstrating that a two-blank approach (method 2) effectively mitigates carryover contamination (“memory effects”). The residual Sr observed may stem from remnant sample solution in the capillary and valve circuitry, a known consideration when dealing with any type of automated system, and no different here (also see Meynadier et al., 2006). While carryover levels remain higher than procedural blanks reported for conventional methods, they are still much lower than typical Sr concentrations in groundwater samples by 100-fold or more, constituting on average only 0.13% of sample Sr, and are thus considered highly acceptable for the applications herein, constituting much less than 1% of overall Sr and therefore contribute negligibly to measurement results. It is noted that this may require further improvement for analytical methodologies using TIMS and/or requiring maximum analytical precision.
Results from the HPIC protocol for Sr purification faithfully reproduce literature values for standards and reference materials and furthermore reproduce true-to-life sample results from conventional manual IEC. Comparative results between manual and automated methods demonstrate excellent agreement, with a Pearson's R2 greater than 0.997 (Fig. 3a).
Importantly, it is highly salient to note that HPIC Sr separation and post-chemistry acidification for MC-ICP-MS analyses herein were conducted in a high traffic mass spectrometry laboratory environment (i.e. ISO-8 or worse), not in an isotope geochemistry clean laboratory (ISO-7 or better). In this context, performance with respect to blank Sr contributions is considered excellent, and furthermore this greatly expands the viability of conducting HPIC Sr separation relative to manual IEC, as the HPIC methodology can be applied in a standard laboratory setting and does not require an ISO-7 or better clean laboratory environment (which are prohibitively expensive and relatively rare, especially outside academia).
Nevertheless, future work will focus on refining the method, such as optimizing gradient steps (e.g., increasing MSA concentration), and/or inserting sub-method cleaning step between samples, to further reduce intra-sample blanks and enhance trace-level Sr analysis, however this must be balanced against the overall protocol duration, to remain focused on overall efficiency of the methodology as has been done herein.
Karasinski et al. (2016) demonstrated the feasibility of on-line Sr separation and isotopic analysis using a fully hyphenated HPIC-MC-ICP-MS system, utilizing a Dionex ICS-5000+ system while this study uses the upgraded ICS-6000 system with automated EG and fraction collection, enabling high-purity Sr isolation. Furthermore, the method in Karasinski et al. (2016) was limited by broader elution peaks (when using dry sample introduction to the MC-ICP-MS plasma), whereas the workflow developed herein minimizes contamination and could improve recovery, delivering higher precision and reproducibility for Sr isotopic measurements. Despite its innovation, the limitations present opportunities for future projects to refine the IC-MC-ICPMS setup. For instance, gradient elution programs could enhance separation for multi-isotope analyses (e.g., Sr, Ca, and Rb), and system optimization for challenging matrices like brines or geothermal fluids could broaden its application.
In summary, the automated HPIC protocol developed herein provides significant innovations combining and building upon previous work,30–32 to develop a robust and highly reproducible technique tailored for natural water samples of various matrices, and one that is highly translational and adaptable with respect to current water sampling practices. On the latter specifically, this optimized protocol allows for the introduction of filtered and acidified water samples across many protocols developed internationally by industry, consultancy and government organizations, and introduces a step increase in the MSA concentrations to maximize Sr separation from Ca. This protocol can process 40–50 sample injections in a 24 hour period with minimal human interaction (e.g. 1–2 hours to set-up automated sequence), and the Dionex™ IonPac™ CS16 column can process hundreds of samples or more before replacement (depending on the instrument method parameters and sample matrices). Moreover, this can all be accomplished without the need or use of an ISO-7 (or better) geochemistry clean laboratory, as the current work has demonstrated this can be done in a standard mass spectrometry lab (ISO-8 or worse). Cumulatively, these attributes mark the HPIC protocol herein with very high translational capacity for integration into existing water collection and analytical practices, while also being time-efficient and parsimonious with respect to reagent and consumables use, thereby also working towards sustainable lab practices.
Future work will focus on exploring further time and reagent optimization for the blank injections to further reduce the per-sample processing time on the Dionex system, and further streamlining/simplifying the analytical approach for MC-ICP-MS analyses to optimally align with throughput and scalability needs, as even the analytical uncertainties constrained herein are likely beyond what is required for some applications. Once the automated protocol has been fully optimized, exploration of its adaptation to other aqueous (e.g. biofluids, wine) and solid (e.g. carbonate, bone apatite) media can be pursued.
The streamlined manual IEC protocol developed herein eradicates traditional time-consuming steps such as pre-IEC sample dry-down and reflux. This reduces contamination potential and processing times by nearly a day, plausibly allowing for preparation and analyses of up to 30 samples within the same day as collection.
The novel automated HPIC Sr separation protocol builds upon prior work by optimizing MSA concentrations to improve Sr separation efficiency, particularly between Sr and Ca peaks. Moreover, this method was developed in mixed-media (0.09 mol per L HNO3, MSA) such that it is fully compatible with standard international water collection procedures, supporting high-throughput analysis of 40–50 samples within 24 hours while minimizing operator intervention, reagents and consumables use, thus also promoting sustainable lab practices. An intra-sample cleaning protocol was optimized to reduce so-called “memory effects” to well below 1%. The protocol design is such that sample Sr separation and analyses by MC-ICP-MS can be accomplished within the same day as collection.
Both methodologies prioritize the needs of high-throughput and large-scale application requiring 87Sr/86Sr, significantly advancing the Technology Readiness Level (TRL) of 87Sr/86Sr as a geochemical tool in uses such as water reservoir characterization/differentiation, agricultural source-tracing, critical resource exploration, forensics and beyond, this services needs across academia, government, industry and commerce. Moreover, the HPIC methodology can function in a standard laboratory environment (i.e. not an ISO-7 or better clean lab), highlighting the broad adoptability relative to manual IEC.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ja00086f |
This journal is © The Royal Society of Chemistry 2025 |