A rapid method for separating magnesium, iron and calcium from low-Mg rocks for precise measurement via MC-ICP-MS

Abstract

Magnesium (Mg), iron (Fe) and calcium (Ca)—key rock-forming elements—play critical roles in numerous geological processes, making them invaluable tracers in geochemical studies. However, conventional methods for their separation often involve a series of individual purification protocols and repeated column procedures. This study introduces a rapid chemical separation scheme for Fe, Mg and Ca, suitable for diverse rock types, especially for high-Ca and low-Mg samples. This protocol begins with precipitation to remove alkali metals (K and Na), followed by sequential separation of Fe, Mg, and Ca under varying acidic conditions using a single elution protocol with 1.5 mL of AGMP-50 resin. For samples with the Ca/Sr ratio less than 100, additionally a TODGA resin column is incorporated to further separate Ca from Sr. The purified Fe, Mg and Ca fractions exhibited high purity and low procedural blanks, enabling precise isotopic analysis by multi-collector inductively coupled plasma mass-spectrometry (MC-ICP-MS) using a sample-standard bracketing (SSB) method. The method's validity was confirmed through analysis of ten international geological reference materials, with results accurately reproducing published reference values. Therefore, the protocol's efficiency, reproducibility, and adaptability demonstrate its suitability for high-precision isotopic studies for a wide range of geological samples.

---

1. Introduction

Iron (Fe), magnesium (Mg), and calcium (Ca) are key rock-forming and bioessential elements whose concentrations and isotopic compositions are shaped by diverse geochemical and biological processes. These elements possess multiple stable isotopes: Fe has four (⁵⁴Fe, ⁵⁶Fe, ⁵⁷Fe, and ⁵⁸Fe), Mg has three (²⁴Mg, ²⁵Mg, and ²⁶Mg), and Ca has six (⁴⁰Ca, ⁴²Ca, ⁴³Ca, ⁴⁴Ca, ⁴⁶Ca, and ⁴⁸Ca) isotopes.¹ Due to their high abundance and critical roles in both high- and low-temperature geological systems, Fe, Mg, and Ca isotopes have emerged as powerful tools for tracing a wide range of geological processes. These include magmatic activity,^2–7 epigenetic low-temperature weathering,^8–12 deposit genesis,^13–16 planetary formation,^17–20 fluctuations in ocean water composition,^21,22 and early biological evolution.^23–25 Consequently, precise measurement of their isotopic compositions is essential before applying these isotopic systems to reconstruct the recorded geological processes.

With the development of multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS), it has been possible to obtain precise Fe, Mg, and Ca isotope ratios. However, the chemical composition of geological samples is complex and it must be chemically separated using ion exchange columns before testing.^26–29 Furthermore, the measured elements must be purified as much as possible to reduce the interference of matrix effects and isobaric elements in the mass spectrometry process, so as to obtain high-precision isotopic composition. Currently, Fe, Mg, and Ca isotope separation involves distinct protocols, requiring at least three independent column procedures to analyze all three elements from a single sample.^26,27 These methods are mostly applicable to the magmatic rock samples with low calcium and high magnesium contents. The combination of individual approaches to acquire Fe, Mg and Ca together is time-consuming and increases sample blanks and the risk of incomplete sample recovery with associated isotopic fractionation. In recent years, many studies have been conducted by analyzing different isotopes of the same sample, which has been proved to offer obvious advantages in identifying different geological processes.^2,5,15 The development of a chemical process capable of separating Fe, Mg, and Ca would not only enhance experimental efficiency but also significantly advance the characterization of geochemical compositions in limited-quantity samples.

In this study, we introduced a single-column filled with AGMP-50 resin to separate Fe, Mg, and Ca from multiple sample matrices with the total time of less than 10 h. For samples with low Ca/Sr (< 100), further purification of calcium using TODGA resin was performed, where the collected strontium can also be used for subsequent Sr isotope determination. Rock standards for isotopic compositions from felsic rocks to ultramafic rocks, and carbonatite have been treated with loads varying for Mg from 10–15 μg, Fe from 2–75 μg and Ca from 1–500 μg to demonstrate the robustness of the method.

2. Samples and the analytical method

2.1 Reference materials

The reference materials, including rhyolite (RGM-2), granite (GSR-1), basalts (BCR-2 and BHVO-2), diabase (W-2A), andesite (AGV-2), granodiorite (GSP-2), carbonatite (COQ-1), limestone (GBW07120) and seawater were obtained from the United States Geological Survey (USGS), the Chinese National Research Center for Certified Reference Materials (CNRCCRM) and the National Research Centre of Geoanalysis (NRCG). These reference materials, which have an Fe₂O₃ content of 13.77 to 0.21 wt%, a MgO content of 7.23 to 0.29 wt%, and a CaO content of 51.1 to 1.55 wt%, are employed for whole-rock Fe–Mg–Ca isotopic determination. A detailed description of the reference materials can be found in Table 1.

Table 1 Description of reference materials analyzed in the study

Sample	Sample weight (mg)	Description	MgO concentration (% (m/m))	CaO concentration (% (m/m))	Fe2O3 concentration (% (m/m))	Prepared by
RGM-2	6.00	Rhyolite	0.29%	4.68%	1.90%	USGS
GSR-1	4.00	Granite	0.42%	1.55%	2.14%	CNRCCRM
BHVO-2	0.30	Basalt	7.23%	11.40%	12.30%	USGS
W-2A	0.40	Diabase	6.37%	11.05%	10.76%	USGS
BCR-2	0.50	Basalt	3.59%	7.18%	13.77%	USGS
AGV-2	1.00	Andesite	1.79%	5.20%	6.69%	USGS
GSP-2	2.00	Granodiorite	0.96%	2.10%	4.90%	USGS
COQ-1	1.50	Carbonatite	1.20%	48.55%	2.94%	USGS
GBW07120	3.50	Limestone	0.71%	51.1%	0.21%	NRCG
Seawater	—	—	—	—	—	—

---

Commercially available hydrofluoric acid (HF, 40% (v/v), UP grade), nitric acid (HNO₃, 68% (v/v), UP grade) and hydrochloric acid (HCl, 36% (v/v), UP grade) were further purified using a Savillex sub-boiling distillation system. Ultrapure water with a resistivity of 18.2 MΩ cm was obtained from a Milli-Q water purification system (Millipore, Bedford, MA, USA). The sodium hydroxide (purity > 99.99%, LOT: M17D042) was manufactured by the Kelong company. During analyses, the sample-standard bracketing (SSB) method was applied to account for instrumental mass bias. A pure magnesium solution GSB-Mg and another purified concentrated Mg solution IGG were used to calculate the Mg isotope value of the reference materials and monitor instrument stability and data reproducibility. Similarly, two pure iron solutions IRM524 and GSB-Fe were also used for Fe isotope ratio measurements.^26–28 For the Ca isotope analysis, Alfa Ca and Merck Ca solutions were used for validation purposes.²⁹

2.2 Sample digestion procedure and ion-exchange chromatographic separation

All chemical procedures were performed in an ultra-clean lab at the State Key Laboratory of Lithospheric and Environmental Coevolution (SKLLEC) in the University of Science and Technology of China (USTC), Hefei, China. The reference materials used in this study have covered various types of samples from high-K₂O and low-MgO rocks to high-CaO and low-MgO rocks. About 0.3–6.0 mg of sample powders (containing about 10–15 μg Mg) were weighed and fully digested for chemical purification based on the MgO content. The weight of each sample is listed in Table 1. For most reference materials, samples containing ∼10 μg magnesium also contain enough iron and calcium for isotope measurements.

Weighed powders were dissolved in 7 mL Savillex Teflon® beakers treated with a mixture of concentrated HF–HNO₃ (3 : 1, v/v). The capped beakers were heated on a 120 °C hotplate for 24 hours and evaporated to dryness after complete dissolution. A second digestion step was performed by adding concentrated HCl–HNO₃ (3 : 1, v/v) to the dried residue, reheating at 120 °C for 24 hours, and evaporating to dryness. The samples were then redissolved in concentrated HNO₃, evaporated again, and finally reconstituted in 0.5 mol per L HNO₃ with a volume of 1 mL, which was transferred into 5 mL centrifuge tubes for precipitation.

Prior to placing the samples, centrifuge tubes were rigorously cleaned to prevent sample contamination by sequential rinsing with 4 mol per L HCl for 48 hours and Milli-Q water for 48 hours. For hydroxide precipitation, 1 mL of 4 mol per L NaOH was added to a 5 mL centrifuge tube. The mixture was ultrasonicated in a heated bath (80 °C) for 15 minutes to precipitate Mg, Fe, and Ca hydroxides, thereby isolating soluble K⁺ and Na⁺. The sample was centrifuged at 4500 rpm for 15 minutes, after which the pellet was resuspended in 3 mL of Milli-Q water, centrifuged again under identical conditions, and the supernatant discarded. To further remove residual Na⁺ and K⁺ while maintaining alkaline conditions, the precipitate was washed twice with 5 mL Milli-Q water, with each cycle consisting of centrifugation at 4500 rpm for 15 minutes. Finally, the purified precipitate was dissolved in 1 mL of 6 mol per L HCl, dried under controlled conditions on a hotplate, and reconstituted in 0.2 mL of 0.2 mol per L HCl for chromatographic chemistry. The whole precipitation process has been arranged into a diagram and shown as Fig. 1.

Fig. 1 Schematic diagram of the element precipitation process.

The dissolved rock solutions containing about 10–15 μg Mg were passed through a column to eliminate matrix elements (Table 2). Fe–Mg–Ca purification was performed in a customized column (4 mm × 10.5 mm) filled with 1.5 mL of Bio-Rad 100–200 mesh AGMP-50 cation exchange resin. Prior to separation, the resin was pre-treated sequentially with 10 mL of 6 mol per L HCl, 7 mol per L HNO₃, and Milli-Q water (18.2 MΩ cm), followed by conditioning with 3 mL of 0.2 mol per L HCl. The samples dissolved in 0.2 mol per L HCl solution were loaded onto the column. Subsequently, 5 mL of 0.2 mol per L HCl + 0.05 mol per L HF were loaded to remove Al and Ti. Fe was eluted with 12 mL of 0.2 mol per L HCl + 0.5 mol per L HF, with pre- and post-elution fractions (1 mL each) collected to monitor Fe recovery efficiency. As the precipitation process has removed most Na and K, the next 6 mL of 1 mol per L HCl was added to remove the remaining sodium and potassium. Finally, Mg and Ca were selectively eluted with 12 mL of 2 mol per L HCl and 20 mL of 9 mol per L HCl, respectively (Fig. 2).

Table 2 Details of the Fe, Mg and Ca separation procedure using precipitation and chromatographic exchange resin AGMP-50

	Reagent/reaction	Amount/time	Comments
Stage1
	4 mol per L NaOH	1 mL	Precipitation
	Ultrasonic	15 min
	Centrifugation	4500 rpm for 15 min
	3 mL Milli-Q H2O	—	Discard supernatant, remove K
	5 mL Milli-Q H2O	4500 rpm for 15 min	Discard supernatant, dilution Na
	5 mL Milli-Q H2O	4500 rpm for 15 min	Discard supernatant, dilution Na
	6 mol per L HCl	1 mL	Dissolve precipitate
	0.2 mol per L HCl	0.2 mL	Dissolved precipitate for chromatograph
Stage2
AGMP-50	6 mol per L HCl + 7 mol per L HNO3 + Milli-Q H2O	10 mL	Resin cleaning
	0.2 mol per L HCl	5 mL	Resin conditioning
	Sample in 0.2 mol per L HCl	0.2 mL	Sample loading
	0.2 mol per L HCl + 0.05 mol per L HF	5 mL	Remove Al, Ti
	0.2 mol per L HCl + 0.5 mol per L HF	1 mL	Cut
	0.2 mol per L HCl + 0.5 mol per L HF	12 mL	Fe fraction collection
	0.2 mol per L HCl + 0.5 mol per L HF	1 mL	Cut
	1 mol per L HCl	6 mL	Resin washing, remove Na
	2 mol per L HCl	12 mL	Mg fraction collection
	9 mol per L HCl	20 mL	Ca fraction collection
	Milli-Q H2O	10 mL	Resin washing

---

Fig. 2 Elution curves for various sample matrices. (a) RGM-2, (b) COQ-1, (c) GSR-1, (d) BHVO-2, (e) W-2A, (f) BCR-2, (g) AGV-2, (h) GSP-2. The first 5 mL was eluted with 0.2 mol per L HCl + 0.05 mol per L HF, 6–19 mL with 0.2 mol per L HCl + 0.5 mol per L HF, 20–25 mL with 1 mol per L HCl, 26–37 mL with 2 mol per L HCl, and 38–57 mL with 9 mol per L HCl. M refers to mol L⁻¹.

Except for BHVO-2, W-2A, BCR-2, and GBW07120 with Ca/Sr > 100, a TODGA column modified from Gu et al. (2024) was used to further separate Ca and Sr for samples with a low Ca/Sr ratio such as RGM-2, AGV-2, and COQ-1. About 0.3 mL of TrisKem 100–200 mesh TODGA resin was filled in a customized column (an inner diameter of 4 mm) and sequentially cleaned with 0.2 mol per L HCl and Milli-Q water. Then it was conditioned with 2 mL of 4 mol per L HNO₃ before the chemical separation. The Ca eluted from the first column with 20 mL of 9 mol per L HCl was evaporated to dryness, and redissolved in 4 mol per L HNO₃ before being added to the column. Following sample loading, 2 mL of 4 mol per L HNO₃ were passed through the column to remove residual matrix elements. Sr was subsequently eluted with 5 mL of 4 mol per L HNO₃, while the Ca was collected separately with 8 mL of Milli-Q H₂O.

All collected solutions were then evaporated at 80 °C to dryness, and the residues were dissolved in 2% (v/v) HNO₃ for isotope ratio measurements.

2.3 Mass spectrometry

Fe–Mg–Ca isotope ratios were measured using a Thermo Scientific Neptune Plus MC-ICP-MS at the SKLLEC in USTC following the methodologies outlined by An et al. (2014) and Gu et al. (2024). All solutions were diluted with 2% (v/v) HNO₃ at room temperature to achieve a Mg–Fe–Ca concentration of 0.5 ppm, 2 ppm and 10 ppm, respectively, in the final solution. For Fe–Mg–Ca isotopic analyses, the sample-standard bracketing (SSB) method was employed, with IRM524 as the bracketing standard of Fe, GSB-Mg as the bracketing standard of Mg and Alfa Ca as the bracketing standard of Ca. Measurements were conducted in high-resolution mode for Fe and low-resolution mode for Mg and Ca. Fe–Mg isotopes of each sample were analyzed three times; each measurement involved one block with 30 cycles and the integration time for each cycle was 4.194 s. Similarly, Ca isotope measurements were performed in triplicate using blocks of 40 cycles each, employing a shorter integration time of 2.097 seconds per cycle to optimize signal acquisition. Uncertainties for δ⁵⁶Fe–δ⁵⁷Fe, δ²⁵Mg–δ²⁶Mg and δ⁴⁴Ca–δ⁴²Ca of standards and samples are given as two standard deviations (2SD) based on repeated measurements. Each measurement commences with the full process of sample digestion followed by column chemistry, and concludes with instrumental analysis. The analyses were separated by two successive washes during Fe–Mg isotope measurement and three successive washes during Ca isotope measurement, using 5% (v/v) HNO₃ in clean 1 and 2% (v/v) HNO₃ in clean 2 and clean 3 (10 cycles each) to avoid cross-contamination. Detailed instrumental parameters for the Neptune Plus MC-ICP-MS are provided in Table 3.

Table 3 Instrument operating parameters for Mg, Fe and Ca isotope measurements

Instrument parameters	Mg	Fe	Ca
Instrument model	Thermo Scientific Neptune Plus	Thermo Scientific Neptune Plus	Thermo Fisher Neptune Plus
RF powder	1200 W	1200 W	550–600 W
Cooling gas	16 L min^−1	16 L min^−1	16 L min^−1
Auxiliary gas	0.8 L min^−1	0.8 L min^−1	0.8 L min^−1
Sample gas	∼1 L min^−1	∼1 L min^−1	∼0.8 L min^−1
Cones	Ni orifice	Ni orifice	Ni orifice
Resolution mode	Low resolution	High resolution	Low resolution
Sample uptake	50 μL min^−1	50 μL min^−1	50 μL min^−1
Integration time	∼4.194 s	∼4.194 s	∼2.097 s
Block number	1	1	1
Cycles per block	30	30	40

---

Isotopic results are reported in δ-notation as per mil (‰) deviations relative to their respective standards, calculated automatically using the Isolution software.³⁰ The formula is expressed as:

δ^XFe (‰) = 1000 × ((^XFe/⁵⁴Fe)_sample/(^XFe/⁵⁴Fe)_IRM524 − 1), where X refers to mass 56 or 57.

δ^XMg (‰) = 1000 × ((^XMg/²⁴Mg)_sample/(^XMg/²⁴Mg)_DSM3 − 1), where X refers to mass 25 or 26.

δ^XCa (‰) = 1000 × ((⁴⁴Ca/^XCa)_sample/(⁴⁴Ca/^XCa)_SRM915a − 1), where X refers to mass 40 or 42.

Mg and Ca were calculated using GSB-Mg and Alfa-Ca as the standard samples respectively, and converted according to δ²⁶Mg_DSM3 values of GSB-Mg = 2.049‰ (ref. 26) and δ^44/40Ca_SRM915a values of Alfa-Ca = 1.14‰.²⁹

3. Results and discussion

3.1 The efficiency of the AGMP-50 cation resin in purifying Fe–Mg–Ca

The calibration of the elution curves in this experiment was completed using an ICP-OES instrument at SKLLEC in USTC. In previous studies, Fe–Mg–Ca separation protocols rely on high volumes of hydrochloric or nitric acid paired with ion-exchange resins (e.g., AG50W-X8 and AG50W-X12).^31–33 These methods are resource-intensive due to prolonged processing times and excessive acid consumption. Critically, at equivalent acid concentrations, the partition coefficients of K, Mg, Fe, and Ca on these resins exhibit insufficiently distinct separation efficiencies, particularly for systems requiring concurrent isolation of all four elements.^34–36 Furthermore, existing approaches are optimized for mid- to high-Mg magmatic rocks, with limited applicability to high-Ca and low-Mg matrices due to resin incompatibility. To address these limitations, AGMP-50 resin was selected for its superior Ca adsorption capacity, enabling effective separation of high-Ca samples. In this study, we selected the sample GBW07120 with a Ca/Mg ratio of up to 50 for testing to demonstrate the applicability of this method, which is applicable to most natural rock samples. For some high-K samples, K interference was identified as a critical factor compromising Fe–Mg–Ca purification efficiency and subsequent isotopic analysis precision. According to the study of Meulen et al. (2003), the partition coefficient of K and Mg in citric/nitric acid media on AGMP-50 resin is quite different, which makes it possible to effectively separate K and Mg. At the same time, due to the strong Ca adsorption characteristics of AGMP-50 resin, Ca and Fe can also be separated. However, after many experiments, we found that this method is not effective in separating K.

NaOH-induced precipitation has been found to effectively isolate K and Mg, enabling precise Mg isotope analysis in low-Mg samples.³² In this study, we adapted the NaOH precipitation protocol of Bao et al. (2019) to remove potassium from samples. At the same time, under alkaline conditions, co-precipitation of Ca, Mn, Ti, and Fe was observed. This prompted further exploration of simultaneous Fe–Ca separation using this method. Since precipitation requires ensuring that the acid is neutralized under alkaline conditions, an excessive amount of NaOH is added, resulting in a significant amount of Na remaining in the solution after the supernatant is removed. Therefore, a certain quantity of water was added after precipitation for repeated dilution and removal of Na under alkaline conditions. Upon removal of the supernatant, approximately 0.1 mL of the liquid remained with the precipitate, and the addition of water diluted the K and Na in the solution by a factor of 75 000, therefore the influence of K and Na on Mg isotope was eliminated. Considering that the introduction of NaOH may cause contamination, the Fe–Mg–Ca contents in the sodium hydroxide were determined using ICP-MS. The results showed that the concentrations of Fe, Mg, and Ca in 4 mol per L NaOH were approximately 1 ng mL⁻¹, 0.35 ng mL⁻¹, and 0.25 ng mL⁻¹, which are significantly lower than those in natural samples. Therefore, we conclude that the introduction of sodium hydroxide does not affect the analysis of Fe–Mg–Ca isotopes.

Zhu et al. (2020) found that Fe can be effectively purified from an AGMP-50 cation-exchange resin with a mixture of 0.2 mol per L HCl and 0.5 mol per L HF. We therefore eluted Al and Ti using the first 5 mL of a mixture of 0.2 mol per L HCl and 0.05 mol per L HF, and then collected Fe using 12 mL of a mixture of 0.2 mol per L HCl and 0.5 mol per L HF (Fig. 2). To minimize the Na concentration in subsequently collected Mg solution, an additional 6 mL of 1 mol per L HCl was employed to ensure that the molar ratio of Na/Mg was less than 0.05, effectively eliminating potential interference during Mg purification. This approach is also supported by Bao et al.'s (2019) work which indicated that the Na/Mg ratio below 1 and K/Mg ratio below 2 would not induce a shift in the isotopic ratio of Mg. Concurrently, the residual matrix elements, including Mg, Mn, and Ca were retained on the resin. To achieve rapid Mg elution while selectively retaining Ca, we chose to separate Mg using 12 mL of 2 mol per L HCl. Though most Mn was leached during Mg collection, the influence of Mn on Mg isotope analysis is negligible due to its inherently low abundance in natural samples (except Mn nodules).

Finally, Ca was quantitatively recovered from the sample matrix using 12 mL of 9 mol per L HCl. While An et al. (2024) showed that Ca can be eluted earlier, potentially overlapping with the purified magnesium solution in samples exhibiting a higher [Ca]/[Mg] mass ratio, our leaching curves indicate a significant disparity between the magnesium and calcium profiles, despite the sample's elevated [Ca]/[Mg] mass ratio. Furthermore, we can effectively collect iron, magnesium, and calcium within a defined interval (Fig. 3). The results demonstrate that our method can efficiently separate and purify Fe, Mg, and Ca within a single column, regardless of whether the samples are high in K, low in Mg, or rich in Ca (Fig. 2). The interfering elements of Ca, such as Al, Fe, Na, K, and Mg, have been removed during the previous precipitation and leaching processes. However, the research by Gu et al. (2024) indicates that for samples with low magnesium and low Ca/Sr (< 100), Sr still significantly affects the isotopic composition of Ca. To address this, we implemented TODGA resin which exhibits dual affinity for Sr and Ca. Sequential rinsing with 4 mol per L HNO₃ and Milli-Q water achieved effective Sr–Ca separation, yielding about 70% recovery of Sr, sufficient for radiogenic Sr analysis while retaining Ca in a purified fraction. Recently, Nie et al. (2024) demonstrated that ammonia-induced precipitation effectively partitions Fe, Ca, Mg, and Ti into the solid phase while retaining K in the supernatant, with simultaneous removal of Na interference. In this manner, the K in the supernatant may also be purified and analyzed.³⁷

Fig. 3 Juxtaposition of the elution curves of various sample matrices shows that Fe, Mg, and Ca can be extracted within the same interval range. M refers to mol L⁻¹.

3.2 High-precision Fe–Mg–Ca isotope measurements of geological standard materials

The collected cut and the results show that the recovery rate of Fe–Mg–Ca in this separation process is generally greater than 98.5%. To ensure stability during Fe–Mg–Ca isotope analysis, temperature fluctuations in the MC-ICP-MS laboratory were maintained within ±0.2 °C by optimizing the parameters of the air conditioning system. To ensure the consistency of acidity, the same barrel of acid was used for our samples and reference materials. The long-term external reproducibility for the standard solutions IRM524 and GSB-Mg over a year is better than 0.05‰ (2SD). The external precisions of Alfa-Ca are lower than 0.1‰ (2SD). The Fe–Mg–Ca isotope compositions of 10 reference materials were determined by the method presented in this paper. These results, along with those previously reported, are summarized in Tables 4–6 and plotted in Fig. 4. All samples were digested, chemically separated, and analyzed simultaneously under identical conditions and the reported data include six batches of independent experiments. The average δ⁵⁶Fe values of the eight rock standards are as follows: RGM-2 = 0.224 ± 0.029‰ (2SD; n = 6), GSR-1 = 0.128 ± 0.007‰ (2SD; n = 6), BHVO-2 = 0.168 ± 0.021‰ (2SD; n = 6), W-2A = 0.061 ± 0.029‰ (2SD; n = 6), BCR-2 = 0.123 ± 0.047‰ (2SD; n = 6), AGV-2 = 0.111 ± 0.010‰ (2SD; n = 6), GSP-2 = 0.153 ± 0.042‰ (2SD; n = 6) and COQ-1 = −0.052 ± 0.007‰ (2SD; n = 6), consistent with previously published data.^31,38–40

Table 4 Fe isotopic compositions of reference materials

Sample ID	δ ^56Fe (‰)	2SD (‰)	δ ^57Fe (‰)	2SD (‰)	References
RGM-2	0.224	0.029	0.326	0.030	This study
	0.18	0.025			40
	0.2	0.059			40
GSR-1	0.128	0.007	0.171	0.024	This study
	0.14	0.004	0.227	0.011	38
	0.148	0.047	0.214	0.084	39
BHVO-2	0.152	0.039	0.210	0.030	This study
	0.112	0.021	0.163	0.04	39
	0.109	0.028	0.16	0.058	39
W-2A	0.061	0.029	0.084	0.039	This study
	0.053	0.025	0.074	0.054	39
	0.054	0.038	0.071	0.07	39
BCR-2	0.123	0.047	0.186	0.079	This study
	0.084	0.029	0.13	0.048	39
	0.08	0.024	0.123	0.036	39
	0.062	0.038			40
	0.069	0.028			40
AGV-2	0.111	0.010	0.154	0.029	This study
	0.096	0.027	0.148	0.039	39
	0.102	0.022	0.151	0.036	39
	0.094	0.051			40
	0.105	0.034			40
GSP-2	0.153	0.042	0.236	0.117	This study
	0.156	0.021	0.213	0.01	38
	0.157	0.025	0.222	0.038	39
	0.154	0.012	0.23	0.032	39
COQ-1	−0.052	0.007	−0.106	0.004	This study
	−0.065	0.036	−0.094	0.055	39
	−0.065	0.028	−0.106	0.026	39

---

Table 5 Mg isotopic compositions of reference materials

Sample ID	δ ^26Mg (‰)	2SD (‰)	δ ^25Mg (‰)	2SD (‰)	References
RGM-2	−0.210	0.034	−0.107	0.013	This study
	−0.24	0.09	−0.12	0.03	32
	−0.22	0.06			42
	−0.182	0.041	−0.091	0.027	27
GSR-1	−0.221	0.018	−0.123	0.023	This study
	−0.2	0.04	−0.12	0.03	32
	−0.234	0.016	−0.123	0.011	27
BHVO-2	−0.256	0.028	−0.109	0.021	This study
	−0.19	0.09	−0.11	0.06	32
	−0.25	0.05			42
	−0.189	0.024			41
	−0.216	0.035	−0.102	0.03	27
W-2A	−0.234	0.039	−0.101	0.008	This study
W-2	−0.19	0.09			42
BCR-2	−0.199	0.038	−0.095	0.040	This study
	−0.20	0.10	−0.11	0.05	32
	−0.19	0.06			42
	−0.126	0.034			41
	−0.162	0.014	−0.082	0.021	27
AGV-2	−0.132	0.018	−0.057	0.032	This study
	−0.14	0.07	−0.07	0.03	32
	−0.16	0.06			42
	−0.09	0.029			41
	−0.124	0.033	−0.06	0.026	27
GSP-2	0.074	0.024	0.046	0.048	This study
	0.09	0.03	0.05	0.02	32
	0.02	0.07			42
	0.101	0.017			41
	0.042	0.02	0.03	0.011	27
COQ-1	−0.416	0.045	−0.214	0.046	This study
	−0.47	0.07			42
	−0.418	0.036			41
Seawater	−0.778	0.040	−0.403	0.056	This study
	−0.81	0.04	−0.42	0.02	32
GBW07120	−2.106	0.032	−1.094	0.037	This study
	−2.06	0.04	−1.07	0.03	33

---

Table 6 Ca isotopic compositions of reference materials

Sample ID	δ ^44/40Ca (‰)	2SD (‰)	δ ^44/42Ca (‰)	2SD (‰)	References
RGM-2	0.911	0.064	0.484	0.029	This study
	0.80	0.13			43
BHVO-2	0.860	0.033	0.453	0.010	This study
	0.77	0.10	0.36	0.04	29
	0.82	0.05	0.37	0.05	44
	0.80	0.12	0.40	0.07	45
	0.787	0.091	0.375	0.034	46
	0.86	0.10	0.47	0.14	47
W-2A	1.051	0.031	0.436	0.022	This study
	1.14	0.06			48
BCR-2	0.948	0.056	0.452	0.033	This study
	0.84	0.04	0.41	0.06	29
	1.00	0.07	0.45	0.07	44
	0.91	0.09	0.45	0.04	45
	0.798	0.049	0.383	0.058	46
	0.89	0.05	0.48	0.20	47
AGV-2	0.780	0.054	0.366	0.062	This study
	0.74	0.06	0.34	0.05	29
	0.75	0.08	0.33	0.09	49
	0.72	0.09	0.36	0.04	44
	0.751	0.124	0.349	0.044	46
COQ-1	0.781	0.025	0.399	0.088	This study
	0.70	0.08	0.31	0.04	29
	0.75	0.07	0.41	0.12	49
	0.659	0.121	0.335	0.039	46
GBW07120	0.986	0.028	0.451	0.042	This study
	0.98	0.12			50

---

Fig. 4 Mg isotope (a), Fe isotope (b) and Ca isotope (c) ratios in this study and references. The color solid symbols and hollow symbols represent the recommended values of previous studies and the test data of the same sample material in this study, respectively. The bars on the data points in the graph represent 2SD.

At the same time, the average δ²⁶Mg values of the nine standards are as follows: RGM-2 = −0.210 ± 0.034‰ (2SD; n = 6), GSR-1 = −0221 ± 0.018‰ (2SD; n = 6), BHVO-2 = −0.256 ± 0.028‰ (2SD; n = 6), W-2A = −0.234 ± 0.039‰ (2SD; n = 6), BCR-2 = −0.199 ± 0.038‰ (2SD; n = 6), GSP-2 = 0.074 ± 0.024‰ (2SD; n = 6), COQ-1 = −0.416 ± 0.045‰ (2SD; n = 6), seawater = −0.778 ± 0.040‰ (2SD; n = 6) and GBW07120 = −2.106 ± 0.032‰ (2SD; n = 6), which is also basically the same as the previously published reference material data.^{27,32,33,41,42} Furthermore, in the absence of a W-2 (diabase) reference material, W-2A was employed to address this lithological gap, and the test results were found to be consistent with those reported by Liu and Han (2021) for W-2 (−0.19 ± 0.09‰).

Ca isotopic compositions were analyzed following the chromatographic separation protocol of Gu et al. (2024), with instrumental mass bias and matrix effects corrected via sample-standard bracketing using MC-ICP-MS. All measured δ^44/40Ca values of geological standards agree with the compiled values, suggesting that the separation and purification process is reliable.^29,43–50 The measured average δ^44/40Ca values of the seven rock standards are as follows: RGM-2 = 0.911 ± 0.064‰ (2SD; n = 6), BHVO-2 = 0.860 ± 0.033‰ (2SD; n = 6), W-2A = 1.051 ± 0.031‰ (2SD; n = 6), BCR-2 = 0.948 ± 0.056‰ (2SD; n = 6), AGV-2 = 0.780 ± 0.054‰ (2SD; n = 6), COQ-1 = 0.781 ± 0.025‰ (2SD; n = 6) and GBW07120 = 0.986 ± 0.028‰ (2SD; n = 6). Notably, for ultramafic rocks which are rich in Mg but poor in Ca, the proposed method enables effective calcium separation. Conventional MC-ICP-MS analysis imposes relatively stringent requirements on calcium concentration, rendering direct measurement unfeasible on our instrument. However, with recent advancements in CC-MC-ICP-MS technology, particularly the Nu Sapphire system, Li et al. (2022) significantly lowered the required calcium concentration for analysis to approximately 100 ng g⁻¹, thereby expanding analytical capabilities.⁵¹ For samples with a MgO/CaO oxide ratio less than 90, when 15 μg of Mg is loaded, the content of Ca can also meet the test requirements of 100 ng.

Prior investigations have systematically evaluated matrix effects on multi-isotope determinations.^29,31,32 In our chromatographic separation scheme, Fe–Mg–Ca constitute the dominant analytes (>98.5% recovery) that are chromatographically resolved from Ti–Al–K–Na, effectively eliminating the interference from major elements in isotope measuring. For trace elements, when [Cu]/[Mg] > 1 and [Zn]/[Mg] > 6, this may interfere with the test results of Mg isotopes; when [Cr]/[Fe] > 1 and [Ni]/[Fe] > 1, it could influence the Fe isotope analysis; and when [Sr]/[Ca] > 0.01, this might affect the Ca isotope measurements. Notably, in natural samples, trace element concentrations typically do not reach these critical levels. During the precipitation process, most elements remain in the supernatant and are subsequently removed. Therefore, there is no need to be concerned about matrix element influence on Fe–Mg–Ca isotopic analysis, which significantly enhances the accuracy of Fe–Mg–Ca isotope measurements.

In summary, all analytical data are very consistent with the reported data and are within the analytical uncertainty range. Therefore, the method can be used for the determination of Fe, Mg and Ca in conventional, low magnesium, high potassium and high calcium samples.

4. Conclusions

A simple and rapid method for purification of Fe, Mg and Ca was established, and the Fe–Mg–Ca isotopes were measured by the SSB technique on a MC-ICP-MS, which effectively saved the experiment time and the amount of acid. The matrix elements are effectively removed after chromatographic separation, making the method suitable for conventional, low magnesium, high potassium and high calcium samples. Based on this method, the Fe–Mg–Ca isotopic composition of a set of geological reference materials was measured, and the analysis results were in good agreement with previously published data, demonstrating the accuracy of our method.

Author contributions

Zhao-Ya Li: conceptualization, investigation, methodology, data analyses, writing – original draft, writing – review & editing. Xing-Hao Zhang: data analyses, methodology. Guo-Chao Sun: conceptualization, methodology, writing – review & editing. Hai-Ou Gu: methodology. Qiong-Xia Xia: writing – review & editing. Li-Qun Dai: resources. Jin-Jing Huo: resources, data analyses. Zi-Fu Zhao: resources.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the main article.

Acknowledgements

This study was supported by funds from the National Natural Science Foundation of China (42303043, 42472069) and the Fundamental Research Programs for the Central Universities (WK2080250222). Thanks are due to Shuai Xu, Yu Chen, and Youshan Xia for their assistance in the ultra-clean lab and to Yao Zhou for assistance with the analyses of elution curves and isotopic ratios.

References

1. J. Hoefs, Stable Isotope Geochemistry, Springer Nature Switzerland AG, 2021.
2. M. Oeser, R. Dohmen, I. Horn, S. Schuth and S. Weyer, Geochim. Cosmochim. Acta, 2015, 154, 130–150.
3. J. G. Konter, A. J. Pietruszka, B. B. Hanan, V. A. Finlayson, P. R. Craddock, M. G. Jackson and N. Dauphas, Earth Planet. Sci. Lett., 2017, 450, 221–232.
4. Z.-Y. Hu, W.-X. Hu, X.-M. Wang, Y.-Z. Lu and L.-C. Wang, Geochim. Cosmochim. Acta, 2017, 208, 24–40.
5. S.-H. Tian, Z.-Q. Hou, X.-Y. Chen, H.-C. Tian, Y.-L. Gong, Z.-S. Yang, T.-Y. Huang, X.-F. Li and X.-X. Mo, Gondwana Res., 2020, 87, 23–40.
6. M. A. Antonelli, M. Schiller, E. A. Schauble, T. Mittal, D. J. dePaolo, T. Chacko, E. S. Grew and B. Tripoli, Earth Planet. Sci. Lett., 2019, 517, 71–82.
7. D. A. Ionov, Y.-H. Qi, J.-T. Kang, A. V. Golovin, O. B. Oleinikov, W. Zheng, A. D. Anbar, Z.-F. Zhang and F. Huang, Geochim. Cosmochim. Acta, 2019, 248, 1–13.
8. M. S. Fantle and D. J. de Paolo, Earth Planet. Sci. Lett., 2005, 237, 102–117.
9. T. Yesavage, M. S. Fantle, J. Vervoort, R. Mathur, L.-X. Jin, L. J. Liermann and S. L. Brantley, Geochim. Cosmochim. Acta, 2012, 99, 18–38.
10. X.-Y. Chen, F.-Z. Teng, K.-J. Huang and T. J. Algeo, Geochim. Cosmochim. Acta, 2020, 287, 263–276.
11. J. S. Ryu, N. Vigier, L. Derry and O. A. Chadwick, Geochim. Cosmochim. Acta, 2021, 292, 94–114.
12. U. Haldar, R. Chakrabarti and R. L. Rudnick, Chem. Geol., 2023, 637, 121666.
13. W.-Q. Li, S.-G. Zhao, X.-M. Wang, S.-L. Li, G.-G. Wang, T. Yang and Z.-D. Jin, Sci. China: Earth Sci., 2020, 63, 108–120.
14. W.-L. Song, C. Xu, M. P. Smith, J. Kynicky, K.-J. Huang, C.-W. Wei, L. Zhou and Q. Shu, Sci. Rep., 2016, 6, 37377.
15. Q.-Y. Tang, J. Bao, Y.-X. Dang, S. Ke and Y. Zhao, Earth Planet. Sci. Lett., 2018, 502, 221–230.
16. S. M. Elardo and A. Shahar, Nat. Geosci., 2017, 10, 317–321.
17. H. Zhu, M. Ling, Z. Zhang and W. Sun, Proceedings of the AGU Fall Meeting Abstracts, 2013, p. 2625.
18. F. Poitrasson, T. Zambardi, T. Magna and C. R. Neal, Geochim. Cosmochim. Acta, 2019, 267, 257–274.
19. P. Ni, N. L. Chabot, C. J. Ryan and A. Shahar, Nat. Geosci., 2020, 13, 611–615.
20. M. Schiller, M. Bizzarro and V. A. Fernandes, Nature, 2018, 555, 507–510.
21. O. Rouxel, B. M. Toner, S. J. Manganini and C. R. German, Chem. Geol., 2016, 441, 212–234.
22. C. Abadie, F. Lacan, A. Radic, C. Pradoux and F. Poitrasson, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, 858–863.
23. M. Guelke-Stelling and F. von Blanckenburg, Plant Soil, 2012, 352, 217–231.
24. A. Heuser, T. Tütken, N. Gussone and S. J. G. Galer, Geochim. Cosmochim. Acta, 2011, 75, 3419–3433.
25. J. L. Morgan, J. L. Skulan, G. W. Gordon, S. J. Romaniello, S. M. Smith and A. D. Anbar, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 9989–9994.
26. Z.-A. Bao, K.-J. Huang, J. Xu, L. Deng, S.-F. Yang, P. Zhang and H.-L. Yuan, J. Anal. At. Spectrom., 2020, 35, 1080–1086.
27. Y.-J. An, F. Wu, Y.-X. Xiang, X.-Y. Nan, X. Yu, J.-H. Yang, H.-M. Yu, L.-W. Xie and F. Huang, Chem. Geol., 2014, 390, 09–21.
28. C. Vega, S. M. Chernonozhkin, R. Grigoryan, M. Costas-Rodríguez and F. Vanhaecke, J. Anal. At. Spectrom., 2020, 35, 2517–2529.
29. H.-O. Gu, X.-Y. Liu, H. Sun, G.-C. Sun, D.-Y. Li, F.-Y. Wang, C. Ge, Y. Fan and T.-F. Zhou, Anal. Chem., 2024, 96, 6131–6138.
30. L. P. Feng, Goldschmidt Conference, Lyon, France, 2023.
31. G.-H. Zhu, J.-L. Ma, G.-J. Wei and Y.-J. An, J. Anal. At. Spectrom., 2020, 35, 873–877.
32. Z.-A. Bao, K.-J. Huang, T.-Z. Huang, B. Shen, C.-L. Zong, K.-Y. Chen and H.-L. Yuan, J. Anal. At. Spectrom., 2019, 34, 940–953.
33. Y.-J. An, X. Li, W.-N. Lu, J.-B. Xu, Y.-L. Xue, Q. Gong, Y. Peng, F. Liu, M.-X. Ling and Z.-F. Zhang, Geostand. Geoanal. Res., 2024, 48, 941–959.
34. F. W. E. Strelow, C. R. Van Zyl and C. J. C. Bothma, Anal. Chim. Acta, 1969, 45, 81–92.
35. F. W. E. Strelow, Anal. Chim. Acta, 1984, 160, 31–45.
36. N. V. D. Meulen, Ion exchange behaviour of 42 selected elements on AG MP-50 cation exchange resin in nitric acid and citric acid mixtures, Master's thesis, Stellenbosch University of Stellenbosch, 2003.
37. X.-J. Nie, Z.-A. Bao, C.-L. Zong, D.-Y. Peng, N. Lv and H.-L. Yuan, At. Spectrosc., 2024, 45, 298–308.
38. Y.-S. He, H.-J. Wu, S. Ke, S.-A. Liu and Q. Wang, Geochim. Cosmochim. Acta, 2017, 203, 89–102.
39. Y.-S. He, S. Ke, F.-Z. Teng, T.-T. Wang, H.-J. Wu, Y.-H. Lu and S.-G. Li, Geostand. Geoanal. Res., 2015, 39, 341–356.
40. X.-F. Hu, J.-L. Zhao, P.-F. Zhang, Y. Xue, B.-N. An, F. Huang, H.-M. Yu, G.-L. Zhang and X.-J. Liu, Chem. Geol., 2019, 524, 356–367.
41. Y.-S. He, A.-Y. Sun, Y.-C. Zhang, R.-Y. Yang, S. Ke, Y. Wang and F.-Z. Teng, Solid Earth Sci., 2022, 7, 188–199.
42. J.-K. Liu and G.-L. Han, An optimal method for magnesium purification and magnesium isotopic composition obtained by MC-ICP-MS, J. Anal. At. Spectrom., 2021, 36, 1986–1995.
43. L.-P. Feng, L. Zhou, L. Yang, D. J. DePaolo, S.-Y. Tong, Y.-S. Liu, T. L. Owens and S. Gao, Geostand. Geoanal. Res., 2017, 41, 93–106.
44. B.-Y. Gao, B.-X. Su, W.-J. Li, M. Yuan, J. Sun, Y. Zhao and X. Liu, J. Anal. At. Spectrom., 2022, 37, 2111–2121.
45. E. Amsellem, M. Schiller, M. Klausen, A. Bouyon, V. Rojas and M. Bizzarro, Chem. Geol., 2023, 636, 121663.
46. Y.-S. He, Y. Wang, C.-W. Zhu, S.-C. Huang and S.-G. Li, Geostand. Geoanal. Res., 2017, 41, 283–302.
47. S. Mondal and R. Chakrabarti, J. Anal. At. Spectrom., 2018, 33, 141–150.
48. M. A. Antonelli, M. Schiller, E. A. Schauble, T. Mittal, D. J. DePaolo, T. Chacko, E. S. Grew and B. Tripoli, Earth Planet. Sci. Lett., 2019, 517, 71–82.
49. W. Dai, F. Moynier, M. Paquet, J. Moureau, B. Debret, J. Siebert, Y. Gerard and Y. Zhao, Chem. Geol., 2022, 590, 120688.
50. F. Liu, X. Li, Z. Zhang, J. Bai and Y.-J. An, Geostand. Geoanal. Res., 2021, 46, 307–319.
51. W.-J. Li, C.-C. Cui, Q.-Q. Pan, J. Wang, B.-Y. Gao, S.-K. Liu, M. Yuan, B.-X. Su, Y. Zhao, F.-Z. Teng and G.-L. Han, Sci. China: Earth Sci., 2022, 65, 1510–1521.

---