High-efficiency and high-precision analysis of barium isotope ratios achieved through in-tandem column purification and ICP optimization

Abstract

Barium (Ba) isotopes have emerged as powerful tracers in geochemical, environmental, and cosmochemical studies. However, achieving high-precision Ba isotope measurements remains challenging due to matrix removal, procedural blanks, isotopic ratio measurement uncertainties, and accurate mass bias correction. Here, we develop a robust analytical protocol for δ^137/134Ba determination using a ¹³⁰Ba–¹³⁵Ba double spike on a Nu Plasma II MC-ICP-MS. Our method employs an in-tandem micro-column chromatography (AG50-X12 cation-exchange resin followed by Sr-Spec™ resin) to efficiently purify Ba from matrix elements with minimal acid consumption. By eliminating intermediate evaporation and re-dissolution steps, we achieve rapid purification of Ba with a procedural blank of only 278 pg, negligible for most geological samples. Both MATLAB simulations and experimental validation suggested an optimal of ∼20% double-spike proportion in the spike-sample mixture. Additionally, we found that a 200 ppb Ba concentration balances sample consumption, signal intensity and Faraday cup performance. To further refine sampling strategies and minimize isobaric interferences, we mapped the spatial distributions of Ba and Xenon (Xe) ion intensities, and isotope ratios in the ICP both in wet and dry plasma conditions, identifying a stable plasma region where Ba isotope ratios show minimal variability and Xe interference is low. We demonstrate that even trace matrix elements (a few millivolts in intensity) can significantly impact the precision in isotope ratio measurements. The method achieves a long-term external reproducibility better than 0.03‰ (2SD). Analyses of twelve geological reference materials (AGV-2, BCR-2, BHVO-2, BIR-1a, COQ-1, DTS-2B, GSO-2, GSP-2, GSR-8, JF-1, RGM-2, and SCo-1) yield δ^137/134Ba values consistent with published data except for three previously unreported materials (DTS-2B, JF-1, and SCo-1), confirming the reliability of the proposed method. This protocol provides a robust foundation for the mechanism of ion interaction in the ICP and contributes to high-precision Ba isotope applications across diverse geological processes.

---

1 Introduction

Barium (Ba) is a large ion lithophile element (LILE) known for its high chemical reactivity. Therefore, elemental Ba does not occur naturally; instead, it is primarily found in minerals such as hyalophane (K,Ba)[Al(AlSi)Si₂O₈], witherite (BaCO₃) and barite (BaSO₄).¹ During subduction, Ba (a fluid-mobile element) can be released from the subducted plate along with fluids and migrate into the mantle wedge.² It is also highly incompatible during mantle melting,³ causing most of the Ba in the mantle to partition into the melt.⁴ As a result, Ba concentrations in the Earth's crust and in sediments are significantly higher than those in the mantle.⁵ For instance, the average content of Ba in the primitive mantle is approximately 6.9 ppm,⁶ while the average Ba content in the continental crust is about 456 ppm. Within the crust, Ba contents are approximately 628 ppm in the upper crust, 532 ppm in the middle crust, and 259 ppm in the lower crust.⁷ On the contrary, Ba exhibits compatibility with potassium feldspar in highly fractionated granites, hence, the concentration of Ba significantly drops with the fractional crystallization of potassium feldspar in the granites.^8,9 Additionally, sediments typically contain ∼768 ppm.⁵ This pronounced difference between mantle and crustal Ba abundance makes Ba an important tracer for studying subduction-related recycling processes.^10–12 Furthermore, the accumulation rate of Ba bound to sulfate or carbonate in marine sediments correlates closely with the organic carbon flux, making Ba isotopes valuable for reconstructing marine paleoproductivity.^13–16

Barium has seven stable isotopes: ¹³⁸Ba, ¹³⁷Ba, ¹³⁶Ba, ¹³⁵Ba, ¹³⁴Ba, ¹³²Ba, and ¹³⁰Ba, with the abundances of 71.699%, 11.232%, 7.853%, 6.592%, 2.417%, 0.1012%, and 0.1058%, respectively.¹⁷ Early study by Nier, who employed a Thermal Ionization Mass Spectrometry (TIMS) to determine Ba isotopic compositions,¹⁸ however, the limited instrumental precision at that time resulted in relatively large analytical uncertainties. Eugster et al. later introduced ¹³⁴Ba–¹³⁷Ba double-spike method and reduced the analytical error to below 1‰, finding that the isotopic difference between meteorites and terrestrial Ba was about 1‰.¹⁷ Subsequent research extensively applied Ba isotopes to meteorite studies^19–22 and investigations of natural fission reactors,^23–25 although applications to stable Ba isotopic fractionation in geological samples were limited until the early 21st century.²⁶ With the rapid development and widespread application of Multi-Collector Inductively Coupled Plasma Mass Spectrometry (MC-ICP-MS), von Allmen et al. achieved high-precision Ba isotope measurements (±0.15‰, 2SD) by combining ¹³⁰Ba–¹³⁵Ba double spike (DS) technique with MC-ICP-MS.²⁷ Miyazaki et al. further refined the ¹³⁰Ba–¹³⁵Ba double spike method to achieve a reproducibility of ±0.032‰ (2SD) for δ^137/134Ba. These results demonstrate observable differences in δ^137/134Ba among the igneous reference rocks BHVO-2, JA-2, and JB-2.²⁶ High-precision Ba isotope measurements have also been made using a standard-sample bracketing (SSB) approach, with reproducibilities better than 0.07‰,²⁸ and 0.05‰ (2SD).²⁹ These analytical improvements have led to broader applications of Ba isotopes in geosciences. For instance, the Ba isotope data from Li et al. suggest that the mantle sources of initial melts for MORBs, OIBs, and carbonatites exhibit relatively homogeneous Ba isotope composition.³⁰ Moreover, the Ba isotope composition of igneous carbonatites can serve as a reliable proxy for the Ba isotope composition of their mantle sources.³⁰ Wei et al. demonstrated that Ba isotopes can serve as tracers of paleo-productivity in oxic marine environments.³¹ In addition, Li et al. indicate that the assimilation of a tin-enriched source, identified via Ba isotope analysis, played a crucial role in the formation of tin-rich granites and associated deposits.³²

With the extensive application of Ba isotope, it is urgent to develop an efficient and high-precision Ba isotope analytical method. Here, we established a high-efficiency and high-precision analytical method for determining δ^137/134Ba using a double spike approach that achieves reproducibility better than 0.03‰ (2SD). Because ¹³⁸Ba comprises ∼71.7% of total Ba, large voltage discrepancies can arise among Faraday cups. Furthermore, interference from ¹³⁸La and ¹³⁸Ce can affect measurements of ¹³⁸Ba.²⁷ Most of all, part of ¹³⁸Ba is the radioactive product of ¹³⁸La decay.^33,34 Therefore, we chose ¹³⁷Ba to measure and calculate Ba isotopic compositions (δ^137/134Ba), using the following equation: δ¹³⁷Ba (‰) = [(¹³⁷Ba/¹³⁴Ba)_sample/(¹³⁷Ba/¹³⁴Ba)_{NIST 3104a} − 1] × 10³. Moreover, δ¹³⁷Ba values can be converted to δ¹³⁸Ba via the equation of δ¹³⁸Ba = δ¹³⁷Ba × 1.33.¹ To achieve efficient separation of Ba, we employed two in-tandem micro-columns. We then corrected the instrumental mass fractionation using both SSB method and DS method. Finally, we measured δ¹³⁷Ba in twelve geological references (andesite, barite, basalt, carbonate, dunite, feldspars, granodiorite, rhyolite, shale, and trachyte) by MC-ICP-MS to validate our analytical procedure. In addition, we also investigated the spatial behavior of Ba and Xe ions within the inductively coupled plasma. By performing two-dimensional mapping of ion signal intensities and isotope ratios, we visualized how plasma conditions influence isobaric interferences and mass bias across the torch interface. These results allowed us to identify a spatially optimal sampling zone. The findings of this study offer new strategies for reducing interference in MC-ICP-MS analyses, leading to improved precision in isotope ratio determination. The interference/matrix effects revealed in this study contribute to higher-quality isotopic data.

2 Experimental

2.1 Instrumentation

The Ba and Xe mapping measurements both in wet and dry conditions were conducted using a double-focusing MC-ICP-MS (ThermoFisher Scientific, Neptune Plus, Germany) at the National Research Center for Geoanalysis (NRCGA), Beijing, China. Two-dimensional ion signal and isotope ratio distributions were acquired by systematically adjusting the torch position along the axial (Z-axis) and radial (X-axis) directions with time-resolved acquisition (TRA) mode. In addition, all experimental procedures and instrumental analyses were conducted at the Ministry of Natural Resources Key Laboratory of Isotope Geology, Institute of Geology, Chinese Academy of Geological Sciences (CAGS). Barium isotopic compositions were measured using a double-focusing MC-ICP-MS (Nu Plasma II, Nu Instruments, UK) operated in low-resolution mode. An Agilent 5100 ICP-OES and an Agilent 7900 quadrupole ICP-MS (Agilent, USA) were used to monitor elution curves and prescan the mixtures of the sample and double spike.

The Nu Plasma II is equipped with 16 faraday cups and 6 ion counters, allowing for simultaneous collection of multiple isotope signals. Each Faraday cup is equipped with a 10¹¹ Ω resistor. For Ba isotope analyses, ¹³⁴Ba and ¹³⁷Ba were collected in the H2 and L1 Faraday cups, respectively. To eliminate interferences from ¹³⁰Xe, ¹³⁴Xe and ¹³⁶Xe on ¹³⁰Ba, ¹³⁴Ba, and ¹³⁶Ba, ¹³¹Xe was monitored in the L4 cup. Additionally, the interference of ¹³⁸La and ¹³⁸Ce on ¹³⁸Ba was eliminated by monitoring ¹³⁹La in the H4 cup and ¹⁴⁰Ce in the H5 cup. The sensitivity of ¹³⁸Ba on the H3 Faraday cup was ∼180 V per ppm under dry plasma conditions using an Aridus II desolvator. Zoom optics were applied to ensure that all Ba isotopes could be collected simultaneously on the Faraday cups. Sample and standard solutions were aspirated through a Desolvating Nebulizer System (DNS). Because the double spike method requires stable instrumental conditions for precise isotopic ratios, the nebulizer aspiration rate was carefully optimized to ∼65 μL min⁻¹. The same DNS and nebulizer were also used for Neptune Plus experiment for dry condition. Detailed operating parameters are listed in the Table 1.

Table 1 MC-ICP-MS (Nu Plasma II and Neptune Plus) Parameters

Instrument parameters	Nu plasma II	Neptune plus
RF forward power	1300 W	1300 W
RF reflected power	1 W	1 W
Cooling gas (Ar)	13.0 L min^−1	14.8 L min^−1
Auxiliary gas (Ar)	1.0 L min^−1	1.0 L min^−1
Mix gas (Ar)	1.2 L min^−1
Sample gas (Ar)		0.9 and 1.0 L min^−1
Resolution mode	Low resolution	Low resolution
Sample introducing type	Dry	Wet and dry
Sample uptake rate	65 μL min^−1	200 and 65 μL min^−1
Integration time	2 s	0.275 s
Blocks	1
Cycles	150
Wash-out time	170 s

---

Cup configuration (present setup) Nu plasma II
H5	H4	H3	H2	H1	C	L1	L2	L3	L4	L5
^140Ce	^139La	^138Ba	^137Ba	^136Ba	^135Ba	^134Ba	^133Cs	^132Ba	^131Xe	^130Ba
		^138La		^136Xe		^134Xe		^132Xe		^130Xe
		^138Ce		^136Ce

---

Cup configuration Nu plasma II
H5	H4	H3	H2	H1	C	L1	L2	L3	L4	L5
^139La	^138Ba	^137Ba	^136Ba	^135Ba	^134Ba	^133Cs	^132Ba	^131Xe	^130Ba	^129Xe
	^138La		^136Xe		^134Xe		^132Xe		^130Xe
	^138Ce		^136Ce

---

Cup configuration Neptune plus
H3	H2	H1	C	L1	L2	L3	L4
^138Ba	^137Ba	^136Ba	^135Ba	^134Ba	^131Xe	^130Ba	^129Xe
^138La		^136Xe		^134Xe
^138Ce		^136Ce

---

2.2 Reagents and solutions

Concentrated HCl, HNO₃, and HF (analytical grade, Beijing Chemical Works, Beijing, China) were purified using DST-4000 sub-boiling distillation systems (Savillex, Eden Prairie, USA). HCl acid underwent two purification cycles, while HNO₃ and HF acids were purified once. All acids, samples, and standards were prepared using 18.2 MΩ cm Milli-Q water (Millipore, Bedford, MA, USA).

The bracketing standard used in this study was SRM 3104a (National Institute of Standards and Technology, NIST), which contains 6.994 ± 0.017 mg g⁻¹ Ba in ∼10% (v/v) HNO₃. This stock solution was diluted to 8 μg g⁻¹ Ba and stored in 2% (v/v) HNO₃ in a pre-cleaned PTFE bottle. Two Ba-enriched carbonates from Oak Ridge National Laboratory (ORNL, Oak Ridge, TN) were used to produce a double-spike solution: one enriched in ¹³⁰Ba (35.8% enrichment) and the other in ¹³⁵Ba (93.5% enrichment). These spikes were initially dissolved in 2% HNO₃, and combined to create the double spike solution. To evaluate the accuracy and precision of the Ba isotope measurement, twelve geological reference materials (GRMs) were analyzed, including AGV-2 (andesite), BCR-2 (basalt), BHVO-2 (basalt), BIR-1a (basalt), COQ-1 (carbonate), DTS-2B (dunite), GSO-2 (barite), GSP-2 (granodiorite), GSR-8(trachyte), JF-1 (feldspars), RGM-2 (rhyolite), and SCo-1 (shale). All solutions and reference materials were prepared in a Class-100 clean laboratory at Ministry of Natural Resources Key Laboratory of Isotope Geology to minimize blank contamination.

2.3 Sample preparation

Approximately 50 mg of each geological reference material was digested in Teflon beakers using a mixture of 3 mL concentrated HF (∼28 M) and 1 mL concentrated HNO₃ (∼14 M),^35–37 except that GSO-2 barite was dissolved with Milli-Q water directly.³⁸ The beakers were sealed and placed on a hotplate at 120 °C for at least 48 hours to ensure complete digestion. The solutions were then evaporated to dryness, and the remaining residues were re-dissolved in 2 mL of concentrated HNO₃. The sealed beakers were then heated again at 120 °C for 48 hours to break down fluoride complexes. The solutions were evaporated a second time to dryness, and the residues were finally dissolved in 2 mL of 1 M HCl. From this solution, a 100 μL sample was taken for Ba concentration analysis using an Agilent 5100 ICP-OES. The remaining solution was used for isotope analysis. An aliquot containing 320 ng of Ba was combined with 80 ng of the double spike solution, and the mixture was heated at 100 °C for 2 hours to ensure thorough isotopic equilibration. Afterward, the mixture was evaporated to dryness and re-dissolved in 0.5 mL of 1 M HCl for further purification.

2.4 Purification procedure

A two-column ion-exchange protocol was employed to purify Ba (Table 2). Both columns had identical internal diameters of ∼4.5 mm but were packed with different resins. The first column was packed with 0.3 mL of Bio-Rad® AG 50W-X12 cation exchange resin (100 to 200 mesh) and then pre-conditioned with 1 M HCl prior to sample loading. The second column was packed with 0.6 mL Triskem's Sr-Spec™ resin (100 to 150 mesh) and connected in series with the first column during sample loading. After mixing 320 ng of Ba (from the sample) with 80 ng of the double spike, the solution was evaporated to dryness and re-dissolved in 0.5 mL of 1 M HCl. This 0.5 mL aliquot was then loaded onto the first column. The first column was rinsed with 4.5 mL (0.5 mL × 9) of 1.5 M HCl, which removed elements such as Al, Mg, Fe, and K, as well as high-field strength elements (e.g., Zr, Hf, Ta, Ti). Then, Ba and Sr (along with some K, Mg, and Ca) were eluted from the first column by adding 4 mL (0.5 mL × 8) of 3 M HNO₃. This eluent simultaneously passed through the second column, thus completing the sample loading onto Sr-Spec™ resin without evaporation. After this step, the two columns were separated. The first column was eluted with 5 mL of 7 M HNO₃ to collect rare earth elements (Ce, Nd).³⁹ The second column was rinsed with 2 mL of 3 M HNO₃ to remove residual alkali and alkaline earth elements (K, Mg, Ca). Ba was then collected with 8 mL of 7 M HNO₃, followed by collection of Sr with 8 mL of 0.05 M HNO₃. Finally, the columns were cleaned for re-use. The first column was washed with 5 mL each of 6 M HCl and Milli-Q water (repeated twice) to ensure complete removal of Fe and Ca. The second column was washed alternately with 7 M HNO₃, 0.05 M HNO₃, and 3 M HNO₃ (repeated twice) to remove all residual contaminants.

Table 2 Column separation procedure for Ba

Column 1: AG50-X12	Eluent	Volume (mL)
Column cleaning	6 M HCl	5
	Milli-Q	5
Conditioning	1 M HCl	5
Sample loading	1 M HCl	0.5
Remove matrix	1.5 M HCl	4.5 (0.5 mL × 9)
Two columns in series
Collect Ba, Sr	3 M HNO3	4
Separate two columns
Collect Ce, Nd	7 M HNO3	5
Column cleaning	6 M HCl	5
	Milli-Q	5
	6 M HCl	5
	Milli-Q	5

---

Column 2: Triskem's Sr-Spec™	Eluent	Volume (mL)
Column cleaning	7 M HNO3	3
	0.05 M HNO3	3
	3 M HNO3	3
Conditioning	3 M HNO3	2
Sample loading	Use column1	4
Remove matrix	3 M HNO3	2
Ba elution	7 M HNO3	8 (2 mL × 4)
Sr elution	0.05 M HNO3	8 (2 mL × 4)
Column cleaning	7 M HNO3	3
	0.05 M HNO3	3
	3 M HNO3	3
	7 M HNO3	3
	0.05 M HNO3	3
	3 M HNO3	3

---

3 Results and discussion

3.1 Theoretical optimized double spike composition

The double spike technique is a well-established method that can be applied to any element with four or more isotopes.^40,41 Recently, this method has garnered renewed interest due to its application in non-traditional stable isotope studies (such as Ca,⁴¹ Fe,⁴² Ba,⁴³ Zn,⁴⁴ Ti,⁴⁵ Cd,^46,47 Ni,⁴⁸ and Cr⁴⁹). The double spike method offers several key advantages compared with the SSB and element doping methods. It provides enhanced precision and accuracy in correcting for instrumental isotope fractionation, making it especially suitable for detecting subtle isotopic variations in nature.^1,50 An additional advantage of the double spike method is the ability to simultaneously correct for mass fractionation introduced during both chemical purification and instrumental analysis while preserving naturally occurring, mass-dependent isotope variations.⁵¹ This approach also enables precise determination of Ba isotopic compositions in samples with low Ba concentrations.

Barium has seven stable isotopes (¹³⁸Ba, ¹³⁷Ba, ¹³⁶Ba, ¹³⁵Ba, ¹³⁴Ba, ¹³²Ba, and ¹³⁰Ba), which offers multiple options for selecting an optimal double spike pair. Eugster et al. pioneered the use of a ¹³⁴Ba–¹³⁷Ba double spike.¹⁷ Since then, various double spike combinations, such as ¹³⁵Ba–¹³⁷Ba,^52,53135Ba–¹³⁶Ba,^1,54,55 and ¹³⁰Ba–¹³⁵Ba,^26,27,43,56 have been developed and tested. The specific composition of the double spike is crucial for achieving high-precision isotope analysis.

In this study, two considerations guided our choice of spike isotopes. First, we excluded ¹³⁸Ba due to it is affected by signal interference from ¹³⁸La, and a portion of it originates from the electron capture decay of ¹³⁸La,^33,34 which complicates its utility as a spike. Second, we avoided ¹³²Ba because of its significant interference from ¹³²Xe (26.89% abundance). We then used Rudge's “Double Spike Toolbox”⁴⁰ in MATLAB and Feng's independent free-installed software “Isolution”⁴¹ to evaluate different combinations of ¹³⁷Ba, ¹³⁶Ba, ¹³⁵Ba, ¹³⁴Ba, and ¹³⁰Ba. Both softwares gave consistent recommendations. The results indicate that the theoretical standard deviation (1SD) for the Ba isotope ratio is minimized when the double spike constitutes ∼20% of the total Ba in the spike-sample mixture for both ¹³⁰Ba–¹³⁴Ba and ¹³⁰Ba–¹³⁵Ba pairs. However, at higher spike-sample mixing proportions, the ¹³⁰Ba–¹³⁵Ba pair consistently yields lower theoretical 1SD values of alfa than ¹³⁰Ba–¹³⁴Ba pair, making ¹³⁰Ba–¹³⁵Ba pair the most effective choice (Fig. 1). To further optimize the spike composition, we utilized the “Double Spike Toolbox” (MATLAB-based programme) to determine the optimal mixing ratio of the two single spikes enriched in ¹³⁰Ba (Spike1) and ¹³⁵Ba (Spike2) is approximately 2.656, corresponding to a composition of 72.65% Spike1 and 27.35% Spike2. Repeated calibration (n = 14) against NIST SRM 3104a yielded a ¹³⁷Ba/¹³⁴Ba ratio of 2.73428368 with a standard error of 0.055387‰ (Table 3), demonstrating that the double spike is well suitable for high-precision Ba isotope analysis.

Fig. 1 The theoretical effects of double spike selections and their proportions in the double spike-sample mixture on the error of alfa. Alfa denotes the mass fractionation factor of the natural sample relative to the reference standard (NIST SRM 3104a), reflecting isotope fractionation in natural process before spiked runs.

Table 3 The isotope composition of the ¹³⁰Ba–¹³⁵Ba double spike

Spike composition
	^130Ba	^132Ba	^134Ba	^135Ba	^136Ba	^137Ba	^138Ba
72.65% of Spike1 ^130Ba	35.8%	0.4%	2.6%	5.6%	5.6%	7.2%	42.9%
27.35% of Spike2 ^135Ba	0	0	0.5%	93.5%	1.6%	0.9%	3.5%

---

Double spike ratio
	^130Ba/^134Ba	^132Ba/^134Ba	^135Ba/^134Ba	^136Ba/^134Ba	^137Ba/^134Ba	^138Ba/^134Ba
Ratio	12.9752115	0.1506210	14.7709502	2.2420272	2.73428368	16.0844596
SE (n = 14)	0.303910‰	0.006516‰	0.167726‰	0.029684‰	0.055387‰	0.40109‰

---

3.2 Experimentally validated proportion of double spike in double spike-sample mix

In addition to the isotopic composition of the double spike, the proportion of spike in the spike-sample mixture is crucial for achieving high precision in Ba isotope measurements. As shown in Fig. 1, a MATLAB-based optimization algorithm indicates that the optimal spike proportion ranges from 18% to 35%. To validate these computational results, we conducted a series of tests using the double spike method, preparing mixtures with varying spike percentages (e.g., 10% DS + 90% SRM 3104a, 20% DS + 80% SRM 3104a, … , 90% DS + 10% SRM 3104a). The deviation pattern in Fig. 2a matches in shape with the simulation results presented in Fig. 1, revealing that the smallest 2SE for δ^137/134Ba occurs when the double spike (¹³⁰Ba–¹³⁵Ba) content ranges from approximately 10% to 40% of the total spike-sample mixture. Fig. 2b further illustrates the accuracy of δ^137/134Ba at various spike percentages. In particular, δ^137/134Ba remains within analytical uncertainty when the spike proportion is between 20% and 30%. This range simplifies spike-sample preparation. Notably, at a 20% double spike proportion, δ^137/134Ba demonstrates the highest accuracy (as indicated by the lowest 2SD). Consequently, we selected a 20% (DS) to 80% (sample) ratio to optimize analytical precision throughout this study.

Fig. 2 The experimental effects of various proportions of double spike in double spike-sample mixtures on precision (a) and accuracy (b). The error bars in the (b) represents the range of ±2 standard deviations (2SD). The hatched area represents the long-term 2SD reproducibility of the double-spike Ba isotope measurements, which is ≤0.03‰.

3.3 A novel in-tandem columns Ba purification procedure

Although previous work has advanced methods for Ba purification, many existing approaches involve using a first cation-exchange column to co-elute Ba, Sr, and minor amounts of Ca and Mg in the HCl medium. The Ba-bearing fraction is then typically evaporated to dryness and re-dissolved in HNO₃, followed by a second cation-exchange column to isolate Ba from remaining matrix components. However, these extra evaporation and dissolution steps can increase the risk of sample contamination. In this study, we address these limitations by employing two micro-columns connected in series (Fig. 3a). This setup minimizes the need for intermediate evaporation and re-dissolution steps. The chromatographic procedure has been adapted from Pin et al.^57,58 and Zhu et al.,^39,59 and further optimized to enhance efficiency and reduce contamination risk.

Fig. 3 (a) Schematic diagram of the purification progress and in-tandem ion-exchange column configuration. (b) Sample elution curve of the AG50 X-12 cation-exchange column. (c) Sample elution curve of the Sr-Spec™ exchange column. The long-term tests demonstrate that the yield of Ba is greater than 96%.

Specifically, two identical columns-each containing different types and volumes of resin-are used in series, resulting in lower acid consumption and shorter processing time. Only 19 mL of acid is required from sample loading to final Ba collection, and the entire elution process (excluding the column cleaning steps) can be completed within six hours. This streamlined approach reduces contamination risks, yielding a blank level of only 278 pg. Because this blank is negligible compared to the Ba content of most geological samples, its influence on Ba isotope analyses is negligible. To further improve separation efficiency, Sr-Spec™ resin is used in the second column. This resin has selective adsorption (or complexation) properties for Sr, allowing efficient separation from the sample matrix. When sulfur-bearing samples are dissolved in HCl or HNO₃, minor amount of SO₄²⁻ can form and potentially degrade Sr-Spec™ resin's performance through unwanted reactions. However, the first cation-exchange column in our procedure isolates Ba, Sr, and other matrix elements early on, thereby reducing the amount of SO₄²⁻ that reaches the Sr-Spec™ resin (Fig. 3b). This design prolongs resin lifespan and prevents matrix elements (e.g., Pb) from accumulating in the second column. Notably, the same procedure can be applied to purify Sr for isotope analysis. After Ba collection, Sr is eluted using only 8 mL of 0.05 M HNO₃ (Fig. 3c). Because this is a relatively low-concentration acid, the collected Sr solution can be directly analyzed for Sr isotopes, provided that the Sr concentration meets the requirements of the instruments.

3.4 The optimal zone for Ba isotope analysis in the plasma-insights from Xe interference

Elemental and isotopic distribution maps were generated by manipulating the torch via a computer-controlled X-Y-Z translation stage integrated into the sampling interface of the MC-ICP-MS. Here, the X-axis corresponds to the back/forward direction, the Y-axis represents the up/down position, and the Z-axis denotes the in/out direction, i.e., the sampling depth. After instrumental parameters were optimized, the vertical (Y-axis) position was fixed throughout the entire experiment. The ion intensity and isotope ratio maps were obtained by sequentially adjusting the torch position along the axial direction (Z-axis) and radial direction (X-axis). The map was constructed by combining multiple linear scans. For instance, the Z-axis was first fixed at −4.8 mm, and a cross-sectional line was acquired by moving the X-axis at a constant speed of ∼21 μm s⁻¹. This was followed by another linear scan at Z = −4.6 mm, with each subsequent line collected at incrementally changing Z values. The gridded dataset of the map was in the appendix. The lower Z-axis values indicate a position closer to the interface between the sampler cone and the ICP. Time-resolved acquisition (TRA) mode was employed to simultaneously monitor the signal intensities of ¹²⁹Xe⁺, ¹³⁷Ba⁺ and their corresponding isotope ratios across different spatial regions of the plasma. The Z-axis was incremented in steps of approximately 200 μm, while the X-axis was scanned from −1 mm to +5 mm. The integration time for each pixel was approximately 0.275 s, ensuring sufficient temporal resolution to capture fine spatial variations in ion intensities and isotopic ratios across the plasma.

The thermodynamic properties of the inductively coupled plasma (ICP), including gas temperature (T_g), electron temperature (T_e), ion temperature (T_i), and electron number density (n_e), play a critical role in controlling ionization efficiency and ion transport behavior.⁶⁰ Variations in these parameters directly influence the spatial distribution of ions within the plasma. Prior research has systematically investigated the spatial distribution of elemental ions within ICP under varying experimental conditions. These studies establish a robust framework for understanding ion transport and interference behavior in ICP-MS.^61–64 Building upon previous investigations in ICP-MS, subsequent studies have focused on the behavior of isotopic ratios within the plasma of MC-ICP-MS. Researchers have investigated the axial variation of Pb⁶⁵ and Li isotope ratios,⁶⁶ as well as the spatial distribution of B,⁶⁷ Li,⁵⁹ and C isotopes.⁶⁸ These studies consistently underscore the importance of prioritizing isotope ratio stability over signal intensity during MC-ICP-MS optimization.

3.4.1 Spatial variation of Ba and Xe ions in the ICP and implications for mass bias

The spatial distributions of Ba⁺ and Xe⁺ within the plasma were examined to understand the ionization process and the mass bias both in wet and dry conditions (Fig. 4a, c, e, and g are results of wet plasma; Fig. 4b, d, f, and h are results of dry plasma). As shown in Fig. 4a and b, the radial distribution of Ba⁺ follows a symmetric Gaussian profile, with the maximum intensity located at X = 2.13 and 1.85 mm for wet and dry plasma, respectively. Although the measured ion intensity does not represent the absolute ion concentration, it can be considered a reasonable approximation of the ion distribution within the plasma.^61,69Fig. 4c and e show the two-dimensional spatial distributions of ¹³⁷Ba⁺ and ¹³¹Xe⁺ ion intensities in wet plasma, respectively. Both ions exhibit a “flame-shaped” signal profile that closely resembles the actual morphology of the plasma. Moreover, the axial positions of their signal maximum differ significantly: the ¹³⁷Ba⁺ signal peaks at Z = −3.2 (wet plasma) and −3.8 (dry plasma) mm, while the ¹³¹Xe⁺ signal reaches its maximum at Z = −1.4 (wet plasma) and −3.2 (dry plasma) mm. The spatial distributions of ¹³⁷Ba⁺ (Fig. 4d) and ¹³¹Xe⁺ (Fig. 4f) ion intensities in the dry plasma are similar to those in the wet plasma. However, in wet plasma, a significant portion of the applied Radio Frequency (RF) power is consumed by solvent evaporation, resulting in a more dispersed distribution of ¹³⁷Ba⁺ (Fig. 4c) and ¹³¹Xe⁺ (Fig. 4e). In contrast, the RF power in dry plasma is primarily used for ionizing argon and the analyte, leading to a more focused distribution of Ba and Xe. Consequently, the cross-sectional profile of ¹³⁷Ba⁺ in dry plasma (Fig. 4b) is significantly narrower than that in wet plasma (Fig. 4a), which translates to a substantially higher analytical sensitivity. Therefore, we recommend using dry plasma for Ba isotope analysis to achieve higher sensitivity and minimize sample consumption. The spatial offset of Ba and Xe highlights the distinct formation or transport mechanisms of Ba⁺ and Xe⁺ ions both in wet and dry plasma and has important implications for the correction of isobaric interferences in Ba isotope measurements. The difference is that Ba and Xe are decoupled more thoroughly in the wet plasma, although Xe is ionized more close to the cone in both conditions. The two-dimensional spatial distribution of the ¹³⁷Ba/¹³⁵Ba ratio is relatively stable both in the central region of the wet and dry plasma (Fig. 4g and h), but significant mass bias are observed at three distinct locations in wet plasma: the bottom of the central axis (X = 2.13) and the two radial edges (X ≈ 0 mm and X ≈ 4 mm) (Fig. 4g). The enhanced mass bias in these regions is closely linked to the thermodynamic structure of the plasma. Key parameters such as gas temperature (T_g), electron temperature (T_e), ion temperature (T_i), and electron density (n_e) play critical roles in ionization efficiency and ion transport within the ICP.^59,60,68 The bottom central region lies near the interface between the plasma tail and the sampler cone, where relatively low gas temperatures and increased argon density likely enhance kinetic differences in ion trajectories, increasing the likelihood of mass-dependent separation during ion transport. The two radial edge regions influenced by localized high-temperature zones near the radio frequency (RF) coil. In these areas, elevated T_e and T_g make ion migration increasingly mass-dependent, this may lead to differential diffusion rates between lighter and heavier isotopes, thereby amplifying mass bias. However, the Ba signal intensity in these regions was too low to obtain reliable ¹³⁷Ba/¹³⁵Ba data, due to a better focused energy of dry plasma (Fig. 4g). Since the RF power in wet plasma is consumed by solvent evaporation, the position of peak temperature differs from that in dry plasma. This difference consequently alters the distribution of ions and isotope ratios. Despite these differences, selecting plasma regions with stable thermodynamic conditions as sampling points is crucial for high-precision isotopic analysis both in wet and dry plasma.

Fig. 4 Two-dimensional spatial distributions of ion intensities and isotope ratios in the plasma. (a) and (b) Radial distribution of ¹³⁷Ba⁺ intensity in the ICP at different axial depths (Z-axis) for wet and dry plasma, respectively. (c) and (d) Two-dimensional distribution of ¹³⁷Ba⁺ intensity in the ICP for wet and dry plasma, respectively. (e) and (f) Two-dimensional distribution of ¹³¹Xe⁺ intensity in the ICP for wet and dry plasma, respectively. (g) and (h) Two-dimensional distribution of the ¹³⁷Ba/¹³⁵Ba isotope ratio in the ICP for wet and dry plasma, respectively.

3.4.2 Identifying the optimal sampling zone: minimizing the matrix effect of Xe

Fig. 4e and f investigate the interference of Xe on Ba isotopes for wet and dry plasma, respectively. As a trace contaminant in argon gas, Xe distribution is strongly affected by gas flow structure and the relative position of the sampler cone. This result provides valuable insight into the challenges of Xe isotope analysis using MC-ICP-MS, especially under trace-level conditions. Since isotope ratio precision is more critical than signal intensity in MC-ICP-MS, we further evaluated the reproducibility of Ba isotope ratios at various Z positions to confirm the affection of Xe interference and its matrix effect. Fig. 5 further presents the variation of ion intensities, isotope ratios, and corresponding RSDs along the Z-axis at X = 2.13 and 1.85 mm for wet (Fig. 5a and c) and dry (Fig. 5b and d) plasma, respectively. Fig. 5a and b confirms the signal distribution trend (Fig. 4c, d, e and f), with the peak intensities of ¹³⁷Ba⁺ at Z = −3.2 mm and −3.8 mm for wet and dry plasma, respectively. However, the peak intensity of ¹³¹Xe⁺ occurred at Z = −1.4 mm and −3.2 mm for wet and dry plasma, respectively. Fig. 5c and d show the depth-resolved variation in the ¹³⁷Ba/¹³⁵Ba, along with their relative standard deviations (RSDs). Notably, the isotope ratios exhibit their lowest RSD at Z = −2.6 mm and −3.6 mm for wet and dry plasma, respectively. Which indicating that this depth provides the highest stability for Ba isotope measurements. In contrast, the region spanning Z = −2.6 mm to −0.4 mm for wet plasma (Fig. 5c) and Z = −3.6 mm to 0 mm (Fig. 5d) for dry plasma exhibit significantly elevated RSDs. This zone overlaps with the maximum Xe signal intensity, indicating that Xe matrix effects not only cause isotopic bias but also degrade analytical precision, even for the ¹³⁷Ba/¹³⁵Ba, which is free of interferences. Interestingly, the optimal precision point at Z = −2.6 mm for wet plasma does not coincide with the maximum Ba signal (Z = −3.2 mm), but lies just slightly ahead of the Ba maximum-signal point (Fig. 5c), which is also true for dry plasma (Fig. 5d) consistent with previous studies.^59,68 These results indicate that the optimal sampling zone for high-precision Ba isotope analysis lies ahead of the Ba maximum-signal point, away from the Xe signal maximum, and coincides with the lowest RSDs for isotope ratio measurements. Sampling from this zone is therefore recommended to minimize isobaric interference from Xe and achieve both accurate and precise Ba isotopic data under MC-ICP-MS conditions.

Fig. 5 Axial profiles of ion intensities, Ba isotope ratios, and RSD values at radial position of X = 2.13 and 1.85 mm for wet and dry plasma, respectively. (a) and (b) Axial signal intensities of ¹³⁷Ba⁺ and ¹³¹Xe⁺ for wet and dry plasma, respectively. (c) and (d) Axial variation of the ¹³⁷Ba/¹³⁵Ba isotope ratio and its RSD for wet and dry plasma, respectively; dashed line represents ideal Xe-free condition. The dashed lines in (c and d) labeled “ideal curve” represent the theoretical isotope ratios expected under Xe-free conditions, serving as a baseline to illustrate deviations caused by Xe matrix interference.

3.5 Precision and accuracy

In this study, several strategies were employed to enhance the precision and accuracy of Ba isotope analysis:

(I) The precision and accuracy of the MC-ICP-MS were verified through repeated measurements of two in-house samples, GSB-Ba and GSO-2, over a prolonged period. Prior to each Ba isotope analysis, the δ^137/134Ba values of GSB-Ba and GSO-2 were measured to correct for instrument-related mass fractionation of Ba isotopes. The δ^137/134Ba values of GSB-Ba and GSO-2 were determined using the DS and SSB methods, respectively. The long-term values for GSB-Ba were 0.15 ± 0.030‰ (n = 98, DS) and 0.157 ± 0.079‰ (n = 10, SSB) (Fig. 6a), while the long-term value for GSO-2 was 0.24 ± 0.027‰ (n = 36, DS) and 0.208 ± 0.065‰ (n = 10, SSB) (Fig. 6b).

Fig. 6 (a)Long-term analysis results of GSO-2 using the SSB and DS methods. (b) Long-term analysis results of GSB-Ba using the SSB and DS methods.

(II) We evaluated the reliability of the chromatographic purification procedure by comparing our measured Ba isotope data for reference materials AGV-2 (0.056 ± 0.030‰, n = 11), BCR-2 (0.054 ± 0.028‰, n = 26), BHVO-2 (0.027 ± 0.021‰, n = 26), BIR-1a (0.180 ± 0.030‰, n = 10), COQ-1(0.069 ± 0.029‰, n = 11), DTS-2B (0.074 ± 0.028‰, n = 12), GSO-2 (0.24 ± 0.027‰, n = 36), GSP-2 (0.003 ± 0.026‰, n = 19), GSR-8 (−0.076 ± 0.011‰, n = 6), and RGM-2 (0.111 ± 0.030‰, n = 10) with previously published values. The close agreement with literature data attests to the robustness of our methods (Fig. 7).

Fig. 7 Comparison of δ^137/134Ba measurements for frequently reported reference materials, where JF-1, SCo-1, and DTS-2B are reported here for the first time.

(III) Owing to the high separation efficiency of the Sr-Spec™ resin, the final Ba solution was free from matrix elements and had a low blank level. This clean separation and minimal blank contribution further improved the precision and accuracy of our Ba isotope analysis.

(IV) We assessed BaO formation during Ba isotope measurements and obtained BaO/Ba ratios of 0.0135%, 0.0146%, and 0.0140% (Fig. 8). These values are significantly lower than the 0.2% oxide-generation threshold reported by Frères et al. for Nd isotope analysis under dry plasma conditions.⁷⁰ As oxide levels above 0.2% can adversely affect measurement accuracy and precision, our low BaO formation rates support high analytical reliability in this study.

Fig. 8 Oxide yield in the process of Ba isotope analysis.

(V) To eliminate the potential influence of acidity mismatch, all purified samples were evaporated to dryness and re-dissolved in 2% HNO₃ (v/v), following recommendations by Cheng et al.¹ This standardized acid matrix minimizes errors in mass spectrometric measurements.

(VI) We evaluated various solution concentrations during instrument analysis and determined that at 200 ppb Ba, δ^137/134Ba yielded the lowest 2SE and 2SD. Due to the voltage range limitations of the Faraday cup collecting ¹³⁸Ba (H3 position), we did not increase the concentration further. Thus, 200 ppb was selected as the optimal analytical concentration (Fig. 9). The long-term reproducibility analysis of this study shows that the precision of δ^137/134Ba values is ≤0.03‰ (2SD).

Fig. 9 The effect of varying Ba concentrations on the accuracy (a) and precision (b) of Ba isotope.

(VII) To address the potential impact of ¹³⁰Xe, ¹³⁴Xe, and ¹³⁶Xe on ¹³⁰Ba, ¹³⁴Ba, and ¹³⁶Ba, respectively, two distinct Faraday cup configurations were employed (Table 1). The signals of ¹²⁹Xe and ¹³¹Xe were used to deduce and correct interference from neighboring Xe isotopes. Fig. 10a and b illustrates three-isotope plots of Ba measured after interference correction via the SSB approach. The results confirm that the measured Ba isotopic compositions lie on the mass-dependent fractionation line (MDFL). The slopes of the fractionation lines are described by: Y = (1.5039 ± 0.0261)X − (0.5851 ± 0.0032) (R² = 0.9890) and Y = (1.5032 ± 0.0263)X − (0.0007 ± 0.0032) (R² = 0.9888), respectively (Fig. 10a and b). These results are consistent with theoretical predictions by Nan et al.²⁹ We corrected the interference using ¹²⁹Xe and ¹³¹Xe, respectively, and compared the δ^137/134Ba. The results demonstrated a high level of consistency, the fitted curve for both corrections closely followed the line Y = X (R² = 1) (Fig. 10c). In the case of δ^136/134Ba, the fitted curve for the corrections using ¹²⁹Xe and ¹³¹Xe were Y = (0.9995 ± 0.0025)X − (0.00007 ± 0.0003) (R² = 0.9998) (Fig. 10d). Even with minor discrepancies, the results remain significantly more precise than the overall analytical uncertainty.

Fig. 10 (a) Barium three isotope plot of the geological samples corrected with ¹³¹Xe. (b)Barium three isotope plot of the geological samples corrected with ¹²⁹Xe. (c) Comparison of δ^137/134Ba in geological samples corrected with ¹³¹Xe and ¹²⁹Xe, respectively. (d) Comparison of δ^136/134Ba in geological samples corrected with ¹³¹Xe and ¹²⁹Xe, respectively.

3.6 δ ^137/134Ba for natural materials

In this study, GSB-Ba (δ^137/134Ba = 0.150 ± 0.030‰, 2SD, n = 98) and GSO-2 (δ^137/134Ba = 0.240 ± 0.027‰, 2SD, n = 36) were used as long-term reference materials to assess instrumental stability and monitor analytical quality in barium isotope measurements. To evaluate the flexibility of the proposed procedure, including its ability to treat silicate samples, a group of geological reference materials (GRMs) were analyzed, including AGV-2, BCR-2, BHVO-2, BIR-1a, COQ-1, DTS-2B, GSO-2, GSP-2, GSR-8, JF-1, RGM-2, and SCo-1. Ba content differs greatly among these standards. The obtained δ^137/134Ba value of AGV-2 (0.056 ± 0.030‰, n = 11), BCR-2 (0.054 ± 0.028‰, n = 26), BHVO-2 (0.027 ± 0.021‰, n = 26), BIR-1a (0.180 ± 0.030‰, n = 10), COQ-1 (0.069 ± 0.029‰, n = 11), GSO-2 (0.24 ± 0.027‰, n = 36), GSP-2 (0.003 ± 0.026‰, n = 19), GSR-8 (−0.076 ± 0.011‰, n = 6), and RGM-2 (0.111 ± 0.030‰, n = 10), were similar to published data within uncertainty (Table 4).^{1,26,28,29,38,43,55,56,71–76} Furthermore, the δ^137/134Ba value of DTS-2B (0.074 ± 0.028‰, n = 12), JF-1 (0.002 ± 0.026‰, n = 12), and SCo-1 (−0.068 ± 0.030‰, n = 10) are presented herein for the first time. Although these materials are not frequently used, they provide useful reference values for understanding Ba isotope fractionation in the mantle, magmatic fractional crystallization processes, and sedimentary processes, and can serve as important reference materials for future studies.

Table 4 δ ^137/134Ba of reference materials in this study^a

Sample	Ba (μg g^−1)	δ ^137/134Ba (±2SD‰)	Reported in literature	Ref.
a Note: N = number of independent sample digestions; n = number of replicate measurements
BHVO-2 Basalt	130	0.027 ± 0.021‰, N = 8, n = 26	0.038 ± 0.048‰, n = 17	29
			0.058 ± 0.019‰, n = 5	26
			0.038 ± 0.015‰, n = 4	1
			0.05 ± 0.02‰, n = 6	43
			0.02 ± 0.03‰, n = 3	71
BCR-2 Basalt	683	0.054 ± 0.028‰, N = 8, n = 26	0.047 ± 0.028‰, n = 22	29
			0.05 ± 0.03, n = 6	72
			0.038 ± 0.038‰, n = 26	1
			0.06 ± 0.04‰, n = 2	71
			0.05 ± 0.03, n = 12	73
AGV-2 Andesite	1140	0.056 ± 0.030‰, N = 4, n = 11	0.06 ± 0.038‰, n = 11	1
			0.053 ± 0.068‰, n = 6	74
GSP-2 Granodiorite	1340	0.003 ± 0.026‰, n = 19, N = 5, n = 19	0.02 ± 0.03‰, n = 3	71
			0.008 ± 0.045‰, n = 6	1
			0.013 ± 0.046‰, n = 15	29
			0.02 ± 0.02‰, n = 4	72
			0.00 ± 0.03‰, n = 3	73
COQ-1 Carbonatite	1000	0.069 ± 0.029‰, N = 4, n = 11	0.06 ± 0.02‰, n = 5	43
			0.08 ± 0.04‰, n = 20	55
RGM-2 Rhyolite	842	0.111 ± 0.030‰, N = 4, n = 10	0.11 ± 0.04‰, n = 6	1
			0.11 ± 0.05‰, n = 4	28
GSR-8 Trachyte	1053	−0.076 ± 0.011‰, N = 3, n = 6	−0.10 ± 0.03‰, n = 4	72
GSO-2 Barite		0.24 ± 0.027‰, N = 4, n = 36	0.248 ± 0.023‰, n = 44	38
			0.233 ± 0.023‰, n = 20	75
			0.233 ± 0.038‰, n = 40	76
BIR-1a Basalt	7	0.180 ± 0.030‰, N = 4, n = 10	0.187 ± 0.039‰, n = 3	56
JF-1 Feldspars	1846	0.002 ± 0.026‰, N = 4, n = 12	Newly report
SCo-1 Shale	570	−0.068 ± 0.030‰, N = 3, n = 10	Newly report
DTS-2B Dunite	16	0.074 ± 0.028‰, N = 4, n = 12	Newly report

---

4 Conclusions

In this study, we developed a streamlined and high precision method for the analysis of stable Ba isotopes (δ^137/134Ba), integrating a ¹³⁰Ba–¹³⁵Ba double spike approach with an innovative in-tandem two-column purification procedure. By eliminating extra evaporation and re-dissolution steps, this method greatly reduces acid consumption and blank levels. Our choice of Sr-Spec™ resin, combined with a preliminary cation-exchange column, not only prolongs the lifespan of the resin but also guarantees the effective removal of matrix elements, yielding a purified Ba fraction suitable for MC-ICP-MS analysis. Experimental verification of simulations showed that a 20% spike proportion in the spike-sample mixture yields optimal precision, while a final Ba concentration of 200 ppb offers the best analytical performance on our MC-ICP-MS. Additionally, through spatial mapping of Ba isotopes within the plasma torch, we identified the optimal sampling zone where isotope ratio stability and analytical precision reach their maximum. This study demonstrated that Xe is not only an interference element on ¹³⁰Ba, ¹³²Ba, ¹³⁴Ba, and ¹³⁶Ba, but also a matrix element for ¹³⁷Ba/¹³⁵Ba. The existence of matrix element (even for only several millivolts intensities) is proved to degrade the analytical precision, which should be also applicable for the other isotope systems. It highlights again the importance for a thorough purification of element before high precision analysis of isotope ratios. This targeted sampling minimizes the effects of plasma-related interferences and mass bias, further enhancing the reliability of Ba isotope measurements. Long-term reproducibility tests (2SD ≤ 0.03‰) confirm the robustness of this approach. Measurements of twelve geological reference materials (AGV-2, BCR-2, BHVO-2, BIR-1a, COQ-1, DTS-2B, GSO-2, GSP-2, GSR-8, JF-1, RGM-2, and SCo-1) yielded δ^137/134Ba values consistent with published data except for three previously unreported materials (DTS-2B, JF-1, and SCo-1). These results confirm the accuracy and precision of the analytical method (Table 4). Overall, this integrated protocol offers a powerful tool for advancing Ba isotope research. Its high precision, low blank, and cost-effective operation make it especially suitable for geochemical investigations that rely on Ba isotope tracers to gain detailed insights into various geological and environmental processes.

Author contributions

Hao-Ran Duan: experiments, data analysis, preparing the original draft. Zhi-Yong Zhu: conceptualization, experiments, code-editing, visualization, funding acquisition, project administration, supervision, review and editing of the manuscript. Suo-Han Tang, and Yi-Ming Huo: experiments, discussion, and reviewing the manuscript. Zheng-Yu Long, Kun-Feng Qiu, and Xiang-Kun Zhu: discussion, reviewing the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

All relevant data are within the manuscript.

Supplementary information is available. See DOI: https://doi.org/10.1039/d5ja00304k.

Acknowledgements

Dr Yi-Bo Lin was thanked for thoughtful discussions. We gratefully acknowledge Professor Chao Li for providing access to the Neptune Plus instrument. We extend our sincere gratitude to the three anonymous reviewers for their insightful comments and constructive suggestions, which have greatly improved the quality and rigor of this work. The inclusion of the wet versus dry plasma comparison is a direct result of the reviewers' insightful feedback, and it has substantially enriched our manuscript. This study is jointly supported by the National Science and Technology Major Project (2025ZD1005005 and 2022YFC2903502), National Natural Science Foundation of China (42573032, 42073015, 42273019 and 92462302), and the CAGS Research Fund (JKYZD202311, JKYZD202403, and J2312).

References

1. Y. Cheng, Y. Du, K. Chen, C. Zong, J. Zhao and L.-H. Chen, At. Spectrosc., 2022, 43, 236–245.
2. R. Kessel, M. W. Schmidt, P. Ulmer and T. Pettke, Nature, 2005, 437, 724–727.
3. D. B. Clarke, Geochim. Cosmochim. Acta, 1984, 48, 1575.
4. W. F. McDonough and S.-S. Sun, Chem. Geol., 1995, 120, 223–253.
5. T. Plank and C. H. Langmuir, Chem. Geol., 1998, 145, 325–394.
6. S.-S. Sun and W. F. McDonough, Chemical and Isotopic Systematics of Oceanic Basalts: Implications for Mantle Composition and Processes, Geological Society, London, Special Publications, 1989, vol. 42, pp. 313–345.
7. R. L. Rudnick and S. Gao, in Treatise on Geochemistry, ed. H. D. Holland and K. K. Turekian, Elsevier, Oxford, 2014, pp. 1–51,  DOI:10.1016/B978-0-08-095975-7.00301-6.
8. J.-L. Wang, H.-Z. Wei, M. Palmer, A. Williams-Jones, J. Ma, S.-Y. Jiang, S. V. Hohl, Y.-F. Zhu, C. Huan and M.-M. Zhang, Geochim. Cosmochim. Acta, 2023, 360, 138–150.
9. T. McCarthy and R. Hasty, Geochim. Cosmochim. Acta, 1976, 40, 1351–1358.
10. T. Elliott, T. Plank, A. Zindler, W. White and B. Bourdon, J. Geophys. Res. Solid Earth, 1997, 102, 14991–15019.
11. T. Kuritani, E. Ohtani and J.-I. Kimura, Nat. Geosci., 2011, 4, 713–716.
12. J. A. Pearce and R. J. Stern, Back-arc spreading systems: Geological, biological, chemical, and physical interactions, 2006, vol. 166, pp. 63–86.
13. F. Dehairs, R. Chesselet and J. Jedwab, Earth Planet. Sci. Lett., 1980, 49, 528–550.
14. P. W. Crockford, B. A. Wing, A. Paytan, M. S. Hodgskiss, K. K. Mayfield, J. A. Hayles, J. E. Middleton, A.-S. C. Ahm, D. T. Johnston and F. Caxito, Earth Planet. Sci. Lett., 2019, 519, 234–244.
15. S. G. Nielsen, Y. Shu, M. Auro, G. Yogodzinski, R. Shinjo, T. Plank, S. M. Kay and T. J. Horner, Geochim. Cosmochim. Acta, 2020, 275, 1–18.
16. A. Paytan and E. M. Griffith, Deep Sea Res. Part II Top. Stud. Oceanogr., 2007, 54, 687–705.
17. O. Eugster, F. Tera and G. Wasserburg, J. Geophys. Res., 1969, 74, 3897–3908.
18. A. O. Nier, Phys. Rev., 1938, 54, 275.
19. S. Umemoto, J. Geophys. Res., 1962, 67, 375–379.
20. D. Krummenacher, C. Merrihue, R. Pepin and J. Reynolds, Geochim. Cosmochim. Acta, 1962, 26, 231–249.
21. R. Andreasen and M. Sharma, Astrophys. J., 2007, 665, 874.
22. M. C. Ranen and S. B. Jacobsen, Science, 2006, 314, 809–812.
23. H. Hidaka and F. Gauthier-Lafaye, Geochim. Cosmochim. Acta, 2008, 72, 4123–4135.
24. H. Hidaka, P. Holliger and A. Masuda, Earth Planet. Sci. Lett., 1993, 114, 391–396.
25. M. Shmid, Y. Nir-El, G. Engler and S. Amiel, J. Inorg. Nucl. Chem., 1981, 43, 867–875.
26. T. Miyazaki, J.-I. Kimura and Q. Chang, J. Anal. At. Spectrom., 2014, 29, 483–490.
27. K. von Allmen, M. E. Böttcher, E. Samankassou and T. F. Nägler, Chem. Geol., 2010, 277, 70–77.
28. S. Ma, G. Han, Q. Zhang, J. Liu, Z. Ding and Y. Zhao, Spectrochim. Acta, Part B, 2025, 228, 107185.
29. X. Nan, F. Wu, Z. Zhang, Z. Hou, F. Huang and H. Yu, J. Anal. At. Spectrom., 2015, 30, 2307–2315.
30. W. Y. Li, H. M. Yu, J. Xu, R. Halama, K. Bell, X. Y. Nan and F. Huang, Geochim. Cosmochim. Acta, 2020, 278, 235–243.
31. W. Wei, Z. Zeng, J. Shen, L. L. Tian, G. Y. Wei, H. F. Ling and F. Huang, Geochim. Cosmochim. Acta, 2021, 299, 113–129.
32. Q. Li, K.-D. Zhao, Z.-Y. Zhu, M. R. Palmer, W. Chen and S.-Y. Jiang, Chem. Geol., 2024, 646, 121914.
33. J. R. de Laeter and N. Bukilic, Int. J. Mass Spectrom., 2005, 244, 91–96.
34. S. Nakai, A. Masuda and B. Lehmann, Am. Mineral., 1988, 73, 1111–1113.
35. K. F. Qiu, R. L. Romer, Z. Y. Long, A. E. Williams Jones, H. C. Yu, S. Turner, Q. F. Wang, S. S. Li, J. Y. Zhang and H. R. Duan, Sci. Adv., 2024, 10, eado6262.
36. H. R. Duan, Z. Y. Zhu, C. X. Pan, J. X. Ma, Z. Y. Long and K. F. Qiu, At. Spectrosc., 2023, 44, 198–206.
37. J. Y. Zhang, K. F. Qiu, R. Yin, Z. Y. Long, Y. C. Feng, H. C. Yu, Z. Y. Gao and J. Deng, Geology, 2024, 52, 423–428.
38. L.-L. Tian, Y.-Z. Gong, W. Wei, J.-T. Kang, H.-M. Yu and F. Huang, J. Anal. At. Spectrom., 2020, 35, 1566–1573.
39. Z. Y. Zhu, C. X. Pan and X. K. Zhu, Rock Miner. Anal., 2020, 39, 515–524.
40. J. F. Rudge, B. C. Reynolds and B. Bourdon, Chem. Geol., 2009, 265, 420–431.
41. L. Feng, L. Zhou, L. Yang, S. Tong, Z. Hu and S. Gao, J. Anal. At. Spectrom., 2015, 30, 2403–2411.
42. M.-A. Millet, J. A. Baker and C. E. Payne, Chem. Geol., 2012, 304, 18–25.
43. Y.-B. Lin, H.-Z. Wei, S.-Y. Jiang, S. Hohl, H.-L. Lei, X. Liu and G. Dong, Anal. Chem., 2019, 92, 2417–2424.
44. C. Wang, X.-k. Zhu, J. Li, Z. Zhu, J. Ma and Y. Chen, J. Anal. At. Spectrom., 2024, 39, 1529–1539.
45. J. Li, S. Tang, X. Zhu, J. Ma and X. Zhao, Acta Geol. Sin., 2022, 96, 517–524.
46. C. Zhu, Y. Wu, H. Wen, G. Wu, W. Song, Y. Zhang and B. Zhao, Acta Geochim., 2023, 42, 934–942.
47. J. Li, S.-h. Tang, X. Zhu, J.-x. Ma, Z. Zhu and B. Yan, J. Anal. At. Spectrom., 2024, 39, 3106–3115.
48. T. Yang, H. Yu, Z. Zhu, R. Ding, J. Wang and B. Zhu, J. Anal. At. Spectrom., 2024, 39, 1270–1276.
49. J.-M. Zhu, G. Wu, X. Wang, G. Han and L. Zhang, J. Anal. At. Spectrom., 2018, 33, 809–821.
50. D. TAN, J. ZHU and J. WANG, Bull. Mineral., Petrol. Geochem., 2016, 35, 138–145.
51. C. M. Johnson and B. L. Beard, Int. J. Mass Spectrom., 1999, 193, 87–99.
52. Y. T. Hsieh and G. M. Henderson, Earth Planet. Sci. Lett., 2017, 473, 269–278.
53. F. Hemsing, Y.-T. Hsieh, L. Bridgestock, P. T. Spooner, L. F. Robinson, N. Frank and G. M. Henderson, Earth Planet. Sci. Lett., 2018, 491, 183–192.
54. L.-L. Tian, Z. Zeng, X.-Y. Nan, H.-M. Yu and F. Huang, J. Anal. At. Spectrom., 2019, 34, 1459–1467.
55. Z. Zeng, X. Li, Y. Liu, F. Huang and H. M. Yu, Geostand. Geoanal. Res., 2019, 43, 291–300.
56. K. van Zuilen, T. F. Nägler and T. D. Bullen, Geostand. Geoanal. Res., 2016, 40, 543–558.
57. C. Pin, A. Gannoun and A. Dupont, J. Anal. At. Spectrom., 2014, 29, 1858–1870.
58. C. Pin, S. Joannon, C. Bosq, B. Le Fevre and P.-J. Gauthier, J. Anal. At. Spectrom., 2003, 18, 135–141.
59. Z.-Y. Zhu, T. Yang and X.-K. Zhu, J. Anal. At. Spectrom., 2019, 34, 1503–1513.
60. S. M. Nelms, Inductively Coupled Plasma Mass Spectrometry Handbook, Oxford, 2005.
61. Z. Wang, B. Hattendorf and D. Günther, J. Anal. At. Spectrom., 2006, 21, 1143–1151.
62. Z. Hu, S. Gao, Y. Liu, S. Hu, H. Chen and H. Yuan, J. Anal. At. Spectrom., 2008, 23, 1093–1101.
63. G. H. Vickers, D. A. Wilson and G. M. Hieftje, Spectrochim. Acta, Part B, 1990, 45, 499–509.
64. A. E. Holliday and D. Beauchemin, J. Anal. At. Spectrom., 2003, 18, 289–295.
65. J. Barling and D. Weis, J. Anal. At. Spectrom., 2012, 27, 653–662.
66. J. Lin, Y. Liu, X. Tong, L. Zhu, W. Zhang and Z. Hu, J. Anal. At. Spectrom., 2017, 32, 834–842.
67. L. Lin, Z. Hu, L. Yang, W. Zhang, Y. Liu, S. Gao and S. Hu, Chem. Geol., 2014, 386, 22–30.
68. Y.-M. Huo, Z.-Y. Zhu, C.-F. Fan, Y.-W. She, J.-L. Hao and X.-K. Zhu, J. Anal. At. Spectrom., 2025, 40, 700–714.
69. J. H. Macedone, A. A. Mills and P. B. Farnsworth, Appl. Spectrosc., 2004, 58, 463–467.
70. E. Frères, D. Weis, K. Newman, M. Amini and K. Gordon, Geostand. Geoanal. Res., 2021, 45, 501–523.
71. Y. Z. Gong, Z. Zeng, C. Zhou, X. Y. Nan, H. M. Yu, Y. Lu, W. Y. Li, W. X. Gou, W. H. Cheng and F. Huang, Sci. Total Environ., 2019, 687, 1295–1304.
72. Y. J. An, X. Li and Z. F. Zhang, Geostand. Geoanal. Res., 2020, 44, 183–199.
73. X. Y. Nan, H. M. Yu, R. L. Rudnick, R. M. Gaschnig, J. Xu, W. Y. Li, Q. Zhang, Z. D. Jin, X.-H. Li and F. Huang, Geochim. Cosmochim. Acta, 2018, 233, 33–49.
74. L. F. Gou and L. Deng, Anal. Sci., 2019, 35, 521–527.
75. F. Zhang, J. Frýda, M. Fakhraee, Y.-b. Lin, G.-Y. Wei, M. Cao, N. Li, J. Zhou, B. Frýdová and H. Wei, Earth Planet. Sci. Lett., 2022, 584, 117421.
76. Y.-B. Lin, H.-Z. Wei, F. Zhang, S. V. Hohl, G.-Y. Wei, T. Li, G.-L. Xiong, N. Li and S.-Y. Jiang, Palaeogeogr. Palaeoclimatol. Palaeoecol., 2022, 605, 111196.

---