Purification scheme with AG1-X8 and TBP resins for Cd isotopic composition determination by double-spike thermal ionization mass spectrometry

Abstract

Sn is the main isobaric interferent for Cd isotopic composition determination. To eliminate Sn, two-step separation methods using an anion resin combined with commercially available TRU and UTEVA extraction resins are usually required for high-precision Cd isotope ratio measurements. First, the AG1-X8 or AGMP-1M anion resin is utilized to eliminate the matrix and most Sn isobaric interferences. Subsequently, extraction resins are employed to eliminate residual Sn. However, these extraction resins are costly. In this study, we present a new approach using the commercially available TBP (tri-n-butyl phosphate) extraction resin as an alternative. The TBP extraction resin is highly cost-effective (∼5% of the cost of TRU or UTEVA extraction resins) and shows excellent performance in Sn removal. Our two-step procedure achieves high recovery (>99% Cd) with a low procedural blank (<30 pg), enabling the efficient isolation of Sn and Cd from bulk rock solutions. The validity of the proposed method was confirmed by analyzing six geological reference materials, including GSD-11 with a very high Sn/Cd ratio, yielding results in good agreement with the published values.

---

1. Introduction

Cadmium is a chalcophile and toxic element with an average abundance of approximately 0.13 mg kg⁻¹ in the Earth's crust. It has eight isotopes spanning a mass range of 10 amu (¹⁰⁶Cd to ¹¹⁶Cd). Similar to other metal stable isotopic systems, Cd exhibits natural variation in its isotopic composition owing to mass-dependent isotope fractionation, as a result of its isotopes participating in physical and biogeochemical processes to slightly different extents. Examining the variation in the Cd stable isotope ratio (typically expressed as the deviation of ¹¹⁴Cd/¹¹⁰Cd from the NIST 3108 standard solution with a ‰ unit) has been proven to be a powerful indicator for investigating geological processes,^1–12 such as biogeochemical cycles, paleo-environment reconstruction, anthropogenic environment pollution, cosmochemical evolution, and ore deposit evolution. A previous review^1,2 demonstrated that the δ^114/110Cd values in sulfide ores, shales, Fe–Mn nodules, phosphates, upper continental crust, sediments and river water are generally limited to 0 ± 3‰. Therefore, high-precision analytical methods are vital for cosmochemical, environmental, and geochemical research. With the advancement of mass spectrometry and chemical separation techniques,^13–33 the double-spike method combined with TIMS^7,15,21,31 or MC-ICP-MS^{8,18–20,22,23,26–30,33} can provide excellent analytical precision for most geological and environmental materials, with the long-term reproducibility of δ^114/110Cd being greater than ±0.06‰ (2SD). Even for 2–20 ng Cd,^{12,26,27,30,31} high accuracy and precision can be achieved using double-spike TIMS based on a highly sensitive MoSi₂ emitter or double-spike MC-ICP-MS.

Effective Cd sample preparation^{5–8,14–33} is a precondition for achieving accurate and precise Cd isotopic data for geological, biological, and environmental materials, as isobaric interferences give rise to wrong analytical data in both TIMS and MC-ICP-MS. Incomplete isolation of Cd from Pd, In, and Sn may lead to significant isobaric interferences, such as ¹⁰⁶Pd⁺ on ¹⁰⁶Cd⁺, ¹⁰⁸Pd⁺ on ¹⁰⁸Cd⁺, ¹¹⁰Pd⁺ on ¹¹⁰Cd⁺, ¹¹²Sn⁺ on ¹¹²Cd⁺, ¹¹³In⁺ on ¹¹³Cd⁺, ¹¹⁴Sn⁺ on ¹¹⁴Cd⁺, and ¹¹⁶Sn⁺ on ¹¹⁶Cd⁺. Single-column techniques based on AG1-X8 or AGMP-1M anion-exchange resins are very effective in removing matrix elements (K, Na, Ca, Mg, Al, Fe, Ti, and Mn) as well as In and Pd isobaric interferences, but they cannot completely eliminate Sn isobaric interference, with 2–5% of Sn co-eluting with the Cd fraction.^5,31 To completely eliminate Sn isobaric interference, current Cd purification strategies mainly adopt two-step separation schemes combining the anion-exchange resin and extraction resin techniques, which can be roughly classified into four categories (I) Using the TRU extraction resin^12,13,26 to eliminate Sn after purification by the AG1-X8 or AGMP-1M anion resin (II) Using the UTEVA extraction resin³⁰ to eliminate Sn after purification by the AG1-X8 or AGMP-1M anion resin (III) Employing the BPHA (N-benzoyl-N-phenylhydroxylamine) extraction resin²⁸ to eliminate Sn after purification by the AG1-X8 or AGMP-1M anion resin (IV) employing repeated anion exchange resin separation with AG1-X8 (ref. 31) or AGMP-1 M^20,32 to remove Sn.

Although existing methods achieved high Cd recoveries (>90%) and low Cd blanks (<0.15 ng), they typically require relatively large volumes of eluent and numerous evaporation/transfer steps, which prolong the separation process and increase analytical costs.

This study aims to develop a rapid and cost-effective Cd separation scheme based on the AG1-X8 and TBP (tri-n-butyl phosphate) extraction resin for determining Cd isotopes using DS-TIMS. The TBP resin exhibits a comparable capability to commercially available TRU and UTEVA Spec resins in separating Cd from Sn, but its cost is only ∼5% that of these extraction resins, amounting to only ∼0.15 U.S dollars per separation. The reliability and accuracy of the proposed method were validated using synthetic solutions with high Sn/Cd ratios and geological reference materials.

2. Experimental section

2.1 Reagents, chemical materials and geological standards

All AR grade acids (hydrochloric acid, perchloric acid, nitric acid and hydrofluoric acid) were obtained from Sinopharm Chemical Reagent Ltd., China. These acids were further purified using a Savillex™ DST-1000 sub-boiling distillation system (USA). Phosphoric acid was purified using a Bio-Rad AG50W-X12 resin column. Ultrapure water with a resistivity of 18.2 MΩ cm⁻¹ was prepared using a Milli-Q Element system. The molybdenum silicide powder (99.8%, <0.5 µm) was bought from Yaoge company (Shanghai, China). NIST SRM3108 Cd international standard of 1000 µg g⁻¹ was diluted to 10 µg g⁻¹ to monitor the stability of Triton Plus TIMS. Re ribbons are 0.025 mm thick, 0.77 mm wide and 99.98% pure (H. Cross Company). A ¹⁰⁸Cd-¹¹⁶Cd double spike was used to correct the fractionation from instrumental and chemical separation. The detailed preparation and calibration method of this double spike are described in our recent study.³¹

AG1-X8 anion resin column: 7 cm long × 6 mm i.d. With a 5 mL reservoir, packed with the Bio-Rad AG1-X8 resin (200–400 mesh), and 1 mL of resin bed volume.

TBP extraction resin column: 6 cm long × 4 mm i.d. With a 2 mL reservoir, packed with the TBP resin (80–120 mesh) purchased from Beijing Realkan Separation Technology Co., Ltd. (https://realkan.cn/Products_d/10.html), and 0.13 mL of resin bed volume. The TBP extraction resin was pre-cleaned by shaking it with Millipore water and 6 mol per L HCl in sequence. This was repeated 3–4 times (until no more foam appeared), and then the mixture was stored in 1 M HCl.

2.1.1 Labware. The labware used included 7.0 mL and 15.0 mL perfluoroalkoxy alkane (PFA) Teflon vials with screw-top lids (Savillex Corporation, USA). These vials were used for solution collection and evaporation and cleaned prior to use with a degreasing agent followed by sequential washing in AR grade HNO₃, HCl, and ultra-pure H₂O.

2.1.2 Standard solutions. The AAS standard solution of 1000 µg per g Cd (lot#1797541) was obtained from the Key Laboratory of Crust-Mantle Materials and Environments at the University of Science and Technology of China (USTC), Hefei, China.

¹⁰⁸Cd-¹¹⁶Cd double spike: The ¹⁰⁸Cd–¹¹⁶Cd method reported by Li et al.³¹ was employed to correct for isotopic fractionation during chemical purification and TIMS measurement.

2.1.3 Geological reference materials. Six geological reference materials were used to verify our method, including silicate, carbonate, and soil powders, which were purchased from the United States Geological Survey (USGS), the National Institute of Standards and Technology (NIST), USA, the Geological Survey of Japan (GSJ), and the Chinese National Research Centre for Certified Reference Materials (NRCCRM). These CRMs included USGS NOD P-1 (manganese nodule), NOD A-1 (manganese nodule), COQ-1 (carbonate), GSJ JDo-1 (carbonate), NRCCRM GSS-1a (GBW07401a, soil), and GSD-11 (GBW07311, sediment).

2.2 Sample digestion and chemical separation

In this work, sample digestion followed the procedure described by Li et al.³¹ For the silicate materials (GSD-11, NOD P-1, NOD A-1, and GSS-1a), approximately 40 ± 0.1 mg (GSD-11) and 10 ± 0.1 mg (NOD P-1, NOD A-1, and GSS-1a) of sample powders were weighed into screw-top PFA vials. The samples were then digested on a hotplate at 180 °C using an acid mixture of 2 mL of 29 M HF, 0.2 mL of 14 M HNO₃ and 0.2 mL of 11.8 M HClO₄ for four days.^29,30 After digestion, the solutions were then evaporated to dryness at 140 °C. Subsequently, the solution prepared in 2 mL of 6 M HCl was added to the PFA vials. With the vials sealed tightly, the samples were reheated to 180 °C for two hours to completely decompose the fluoride complexes.

For the carbonate materials (COQ-1 and JDo-1), approximately 40 ± 0.1 mg of the sample powder was weighed into screw-top PFA vials. The samples were dissolved on a hotplate at 110 °C using 2 mL of 3 M HCl for 4 hours.

These samples contained 23.2–226 ng of Cd, as calculated according to the previously published Cd content in literature.^19–22 Then, an appropriate amount of the ¹⁰⁸Cd–¹¹⁶Cd double spike was added to each sample solution with a spike/sample mass ratio (3.5) for all samples in this study. These PFA vials were then tightly capped and heated on a hotplate overnight at 120 °C to ensure complete homogenization of the mixed solutions. Finally, the vials were opened, and the sample solutions were evaporated to dryness and then re-dissolved in 1 mL of 6 M HCl.

In this study, as shown in Fig. 1 and Table 1, a two-step resin column was employed to purify Cd. First, the AG1-X8 anion resin chromatography method was employed to remove the matrix and most isobaric interference elements (Table 1), as reported by Li et al.³¹ After the AG1-X8 column, the collected Cd fractions, containing ∼99.5% of Cd along with ∼2.7% of Sn, were dried down on a hotplate at 120 °C. Subsequently, a small column filled with the TBP extraction resin (0.13 mL, TBP resin, 80–120 mesh) was used to purify Cd to eliminate residual Sn. Before performing TBP column separation, a synthetic mixed standard solution containing 5 µg of Zr, Mo, Cd, Sn, and In was employed to calibrate and optimize the column procedure in order to obtain satisfactory Cd purity. The elemental distribution yield was measured using an iCAP Q-ICP-MS (Thermo Fisher Scientific). The Cd fraction was re-dissolved with 0.3 mL of 6 M HCl for TBP column loading. As shown in Table 1b, the TBP resin column was washed with 2 mL of H₂O and conditioned with 2 mL of 6 M HCl. The 0.3 mL Cd sample solution was then loaded onto the TBP column. In 6 M HCl medium, 99.5% of Sn was absorbed on the TBP column, while all Cd was eluted using 3 mL of 6 M HCl. The final Cd fraction was dried down on a hotplate at 150 °C.

Fig. 1 Flow chart of Cd chemical separation using AG1-X8 and TBP resin columns.

Table 1 Two-step Cd purification scheme using AG1-X8 anion and TBP extraction resin columns^a

Procedure step	Reagent	Volume (mL)
a The AG1-X8 and TBP columns were filled with 1 mL of the AG1-X8 resin and 0.13 mL of the TBP resin, respectively. After the two-step column purification, the yield of Cd is approximately 99.4%, and the residual Sn is lower than 0.03%.
Step Ⅰ (AG1 column)
Cleaning column	2 M HNO3	10
Cleaning column	Milli-Q water	10
Cleaning column	3 M HCl	3
Loading sample	6 M HCl	1
Rinsing matrix	6 M HCl	3
Rinsing Fe, In, and Mo	0.4 M HCl	12
Rinsing Zn, Sn	0.5 M HNO3 + 0.1 M HBr	12
Collecting Cd	2 M HNO3	8
Step II (TBP column)
Cleaning column	Milli-Q water	2
Cleaning column	6 M HCl	2
Loading sample	6 M HCl	0.3
Collecting Cd	6 M HCl	3

---

Organic compounds released from resins are known to have a detrimental impact on the sensitivity of isotopic analyses by TIMS.^34–37 Minor organics released from the TBP resin, which strongly inhibited the ionization of Cd, were removed by evaporating the Cd fraction on a hotplate at 160 °C, and then, it was digested using 0.5 mL of 14 M HNO₃ and 0.1 mL of 11.8 M HClO₄ on a 180 °C hotplate for two hours in tightly sealed PFA vials. Subsequently, the sample solution was evaporated to dryness again on a 180 °C hotplate and then treated with 1 mL of 6 M HCl in closed PFA vials on a 160 °C hotplate for two hours for expelling residual HClO₄. Finally, the digested Cd sample was dried at 180 °C for TIMS analysis.

The whole procedural blank, encompassing all steps from sample digestion to the final Cd isotopic measurement by TIMS, was 25 ± 5 pg (n = 5), which is negligible compared with the amount of Cd (104–1017 ng, including spike) processed through the columns.

2.3 Thermal ionization mass spectrometry measurements

Single Re filaments were used to generate Cd ion beams. A mixture of MoSi₂ and phosphoric acid was used as the emitter to enhance the Cd ionization. The detailed loading method is described in our recent study.³¹ Cd isotope compositions were measured using a Triton Plus TIMS (Thermo Fisher Scientific) at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS) in Beijing, China. All the Cd isotope data were acquired in the static multi-collection mode. The collector array is shown in Table 2. ¹⁰⁸Cd,¹¹⁰Cd, ¹¹¹Cd, ¹¹²Cd, ¹¹³Cd, ¹¹⁴Cd, ¹¹⁶Cd and ¹¹⁷Sn were measured by a Faraday cup connected with 10¹¹Ω resistors simultaneously.

Table 2 Collector configuration for the static multi-collection mode

Element	L3	L2	L1	CC	H1	H2	H3	H4
Cd	^108Cd	^110Cd	^111Cd	^112Cd	^113Cd	^114Cd	^116Cd	^117Sn

---

For Cd isotope measurements, the Re filament was heated at a rate of 500 mA min⁻¹ until the ¹¹⁴Cd signal reached approximately 10 mV. The ion beam was then roughly focused, and the filament current was gradually increased to produce a typical signal of approximately 500–1500 mV for ¹¹⁴Cd, depending on the sample loading size. Each measurement run consisted of 200 cycles, yielding an internal precision of 0.03–0.05‰ (2SE) for most δ^114/110Cd values. The integration time per cycle was 4 seconds. ¹¹⁷Sn was monitored to correct for possible isobaric interference on ¹¹⁴Cd (from ¹¹⁴Sn) and ¹¹⁶Cd (from ¹¹⁶Sn). The ¹¹⁷Sn/¹¹²Cd ratios obtained during sample analysis were all <3 × 10⁻⁵, indicating that Sn isobaric interferences were negligible. Additionally, ¹⁰⁵Pd and ¹¹⁵In peaks were checked in the peak-scan mode prior to data acquisition, and no detectable Pd or In signals were observed, confirming the absence of such interferences.

All data for the geological samples doped with the ¹⁰⁸Cd–¹¹⁶Cd double spike were calculated offline following the procedures described by Li et al.^31,34,35 for Cd and Cr isotope analysis and by Heuser et al.³⁸ for Ca isotope analysis. The NIST SRM 3108 Cd standard was analyzed to monitor the instrument status. The ¹¹⁴Cd/¹¹⁰Cd ratios for un-spiked NIST SRM 3108 were normalized to ¹¹⁰Cd/¹¹²Cd = 0.517 330 for mass fractionation correction using the exponential law.^21,31 In this study, 100 ng NIST SRM 3108 standard loads gave a mean ¹¹⁴Cd/¹¹⁰Cd of 2.30429 ± 8 (2SD, n = 8), which is in good agreement with the previously reported value (2.30425 ± 20).^21,31

3. Results and discussion

3.1 Sorption efficiency of Cd and Sn on the TBP resin at different HCl concentrations

To obtain the best sorption efficiency for Sn, three experiments were conducted. Two artificial mixed standard solutions were prepared, each containing 5 µg of Sn, Cd, In, Mo and Zr, to evaluate the sorption efficiency of the TBP resin for Sn, Cd and potential interfering elements such as In, Mo and Zr in different hydrochloric acid concentrations. After the mixed standard solutions were dried, they were dissolved in 0.3 mL of 1 M HCl, 0.3 mL of 3 M HCl, and 0.3 mL of 6 M HCl, respectively. These samples were then loaded onto the TBP columns and eluted using 1 M, 3 M, and 6 M HCl, respectively, following the procedures listed in Table 3. Nine fractions were collected and analyzed using ICP-MS. As shown in Fig. 2 and Table 3, Cd is not retained in hydrochloric acid. 3 mL of 1 M, 3 M or 6 M HCl could elute more than 99.5% of Cd. For the 3 M and 6 M HCl medium, the yield of Mo and Sn was below 1%, indicating excellent sorption efficiency (>99%) of Mo and Sn. In the 6 M HCl medium, the sorption efficiency of In reached ∼100%, significantly higher than that in the 3 M HCl medium. Therefore, 6 M HCl was selected as the eluent, and fractions 8 and 9 were co-collected for the Cd TIMS measurement. These observations demonstrate that Cd, Sn, In, Zr and Mo species exhibit markedly different sorption behaviors depending on the HCl concentration used for TBP resin elution.

Table 3 Yields of Zr, Mo, Cd, In and Sn in Fractions F1–F9 using different HCl concentrations as eluents^a

Fraction	Zr (%)	Mo (%)	Cd (%)	In (%)	Sn (%)	Eluting volume (mL)
a F1–F3 was eluted with 1 M HCl, F4–F6 was eluted with 3 M HCl, and F7–F9 was eluted with 6 M HCl.
F-1	75.1	42.3	61.8	51.5	0.30	0.3
F-2	16.7	43.7	36.2	43.4	0.30	1.5
F-3	0.90	2.5	1.32	2.00	1.60	1.5
Yield%	92.6	88.6	99.3	96.9	2.20
F-4	35.5	0.07	0.02	0	0.20	0.3
F-5	5.40	0.04	70.9	0.50	0.17	1.5
F-6	0.60	0.03	28.6	29.7	0.14	1.5
Yield%	41.5	0.1	99.6	30.2	0.51
F-7	28.0	0.04	0.03	0.01	0.41	0.3
F-8	1.97	0.02	72.4	0	0.32	1.5
F-9	0.29	0.01	27.4	0.02	0.21	1.5
Yield%	30.2	0.08	99.8	0.03	0.93

---

Fig. 2 Sorption behavior of Sn, Cd, Mo, In and Zr using the TBP resin column in 1 M, 3 M and 6 M HCl media.

During the second TBP column stage, ∼99.5% of Sn, ∼69.8% of Zr and ∼100% of In and Mo were adsorbed by the TBP resin, while ∼99.8% of Cd was eluted using 6 M HCl. According to the AG1-X8 resin scheme reported previously,³¹ 99.6% of Cd was recovered and 97.6% of Sn was removed during the first AG1-X8 column. Therefore, the combined AG1-X8 and TBP column procedure developed in this study yielded a final Cd recovery of ∼99.4% and reduced residual Sn to ∼0.03%.

3.2 Evaluation of the interference of Sn

Proper correction of isobaric interferences on several Cd isotopes is crucial for precisely determining Cd isotope ratios using the DS-TIMS technique. The isobaric interferences include ¹⁰⁶Pd on ¹⁰⁶Cd, ¹⁰⁸Pd on ¹⁰⁸Cd, ¹¹⁰Pd on ¹¹⁰Cd, ¹¹²Sn on ¹¹²Cd, ¹¹³In on ¹¹³Cd, ¹¹⁴Sn on ¹¹⁴Cd, and ¹¹⁶Sn on ¹¹⁶Cd. Generally, Pd and In isobaric interferences are completely removed during the AG1-X8 resin stage, and thus, Pd and In signals are undetectable in TIMS measurements. Among the potential interferences, Sn may have a significant influence on Cd measurement due to incomplete separation, especially for samples with high Sn/Cd ratios (>50). Therefore, it is essential to assess the effect of Sn on δ^114/110Cd measurements in our proposed separation scheme.

To verify the Sn removal efficiency of our methods, two synthetic solutions with a high Sn/Cd ratio (200 : 1), which were much higher than those (5–20) of most natural silicates, were prepared by mixing NIST 3108-Cd or AAS-Cd with Sn. These artificial mixtures contained 200 ng of Cd doped with 40 000 ng of Sn, along with a suitable amount of the ¹⁰⁸Cd–¹¹⁶Cd double spike. They were purified using our two-step scheme and measured by TIMS.

As shown in Fig. 3 and Table 4, the external precision of the replicated purified samples (n = 8) is −0.673 ± 0.038‰ (2SD) for AAS and −0.014 ± 0.028‰ (2SD) for NIST 3108, both in good agreement with reference values^{5,19,25,31,32} within error. This demonstrates that our methods were robust even for samples containing up to 40 µg of Sn. No Sn isobaric interferences were detected, as ¹¹⁷Sn/¹¹²Cd was lower than 3 × 10⁻⁵ during the TIMS measurements.

Fig. 3 δ ^114/110Cd values of Sn-doped AAS and NIST 3108 standards with a high Sn/Cd ratio (200) after purification using AG1-X8 and TBP columns.

Table 4 Analytical results of δ^114/110Cd for Sn-doped standards (NIST3108 and AAS) with a high Sn/Cd ratio (200) after purification by AG1-X8 and TBP columns

CRMs	δ ^114/110Cd (‰)	2SE	CRMs	δ ^114/110Cd (‰)	2SE
AAS-M-1	−0.674	0.032	NIST3108-M-1	−0.030	0.022
AAS-M-2	−0.655	0.023	NIST3108-M-2	−0.014	0.031
AAS-M-3	−0.700	0.031	NIST3108-M-3	0.000	0.032
AAS-M-4	−0.677	0.023	NIST3108-M-4	−0.030	0.022
AAS-M-5	−0.654	0.034	NIST3108-M-5	−0.014	0.032
AAS-M-6	−0.698	0.033	NIST3108-M-6	0.011	0.031
AAS-M-7	−0.651	0.025	NIST3108-M-7	−0.019	0.033
AAS-M-8	−0.678	0.024	NIST3108-M-8	−0.013	0.025
Mean ± 2SD	−0.673	0.038	Mean ± 2SD	−0.014	0.028

---

3.3 Cd isotope data for geological reference materials

To verify the robustness of the proposed separation method for real geological samples, six geological reference materials (GRMs), encompassing a wide range of matrix compositions, Cd contents and Sn/Cd ratios, were determined. Among these samples, GSD-11 (sediment) exhibited an extremely high Sn/Cd ratio of 500.²⁰ To the best of our knowledge, GSD-11 has the highest Sn/Cd ratio among the previously reported ∼60 natural CRMs.^5,7,13–33 Based on the Sn and Cd contents of GSD-11 reported by Tan et al.,²⁰ 40 mg of GSD-11 contains 14 800 ng of Sn and only 29.6 ng of Cd, making it an ideal test case and a sensitive indicator for assessing the Sn removal effect of our method. Throughout the GSD-11 measurements, no detectable ¹¹⁷Sn signal was observed. As shown in Table 5 and Fig. 4, fifteen replicated analyses of GSD-11 in this study yielded a value of −0.307 ± 0.052‰ (2SD), which is consistent with the average value (−0.304 ± 0.054‰) calculated from the previously published literature.^{17,20,23,25,32}

Table 5 Analytical results of δ^114/110Cd for geological reference materials relative to NIST SRM 3108^a

CRMs	δ ^114/110Cd (‰)	2SE	Cd ppm	δ ^114/110Cd (‰)
a The Cd sample size calculated according to the references is estimated to be 226 ng for NOD P-1 (10 mg digestion), 76 ng for NOD A-1 (10 mg digestion), 43 ng for GSS-1a (10 mg digestion), 29.6 ng for GSD-11 (40 mg digestion), 23.2 ng for JDo-1 (40 mg digestion), and 24 ng for COQ-1 (40 mg digestion).
GSD-11	−0.284	0.021	0.74 (ref. 20)	−0.274 ± 0.037 (ref. 20)
Sediment	−0.378	0.051		−0.305 ± 0.054 (ref. 17)
	−0.313	0.032		−0.34 ± 0.06 (ref. 23)
	−0.291	0.024		−0.281 ± 0.059 (ref. 32)
	−0.300	0.023		−0.32 ± 0.067 (ref. 25)
	−0.290	0.032
	−0.309	0.034
	−0.342	0.063
	−0.296	0.031
	−0.282	0.032
	−0.273	0.032
	−0.325	0.041
	−0.307	0.033
	−0.311	0.034
	−0.302	0.031
Mean ± 2SD	−0.307 ± 0.052			−0.304 ± 0.054
NOD A-1	0.146	0.042	7.6 (ref. 21)	0.12 ± 0.046(ref. 21)
Mn nodule	0.182	0.044		0.16 ± 0.10(ref. 18)
	0.185	0.025		0.127 ± 0.035(ref. 22)
				0.22 ± 0.02(ref. 5)
				0.184 ± 0.057(ref. 29)
				0.08 ± 0.02(ref. 23)
				0.124 ± 0.067(ref. 20)
				0.193 ± 0.047(ref. 28)
				−0.07 (ref.14)
				0.15 ± 0.036(ref. 31)
				0.112 ± 0.224(ref. 33)
Mean ± 2SD	0.171 ± 0.042			0.127 ± 0.154
GSS-1a	0.012	0.051	4.3 (ref. 19)	0.11 ± 0.05(ref. 19)
Asoil	0.068	0.082		0.14 ± 0.10(ref. 23)
	−0.068	0.032		0.11 ± 0.06(ref. 18)
				−0.078 ± 0.05 (ref. 29)
				0.098 ± 0.027(ref. 20)
				0.08 ± 0.23(ref. 14)
				0.099 ± 0.054(ref. 25)
				0.01 ± 0.058(ref. 31)
Mean ± 2SD	0.004 ± 0.138			0.071 ± 0.142
NOD P-1	0.147	0.042	22.6 (ref. 21)	0.14 ± 0.07(ref. 21)
Mn nodule	0.157	0.023		0.16 ± 0.03(ref. 30)
	0.182	0.025		0.16 ± 0.10(ref. 18)
				0.133 ± 0.038(ref. 20)
				0.135 ± 0.074(ref. 22)
				0.16 ± 0.04(ref. 26)
				0.185 ± 0.048(ref. 29)
				0.196 ± 0.073(ref. 28)
				0.12 ± 0.04(ref. 23)
				0.163 ± 0.04(ref. 19)
				0.13 (ref. 14)
				0.131 ± 0.169r(ref. 33)
				0.165 ± 0.028(ref. 31)
Mean ± 2SD	0.162 ± 0.036			0.152 ± 0.046
COQ-1	0.083	0.062	0.60 (ref. 19)	0.098 ± 0.052(ref. 19)
Carbonate	0.099	0.066		0.143 ± 0.053(ref. 29)
	0.152	0.063		0.230 ± 0.02(ref. 5)
				0.123 ± 0.048(ref. 28)
				0.113 ± 0.082(ref. 31)
Mean ± 2SD	0.111 ± 0.072			0.141 ± 0.104
JDo-1	0.119	0.062	0.58 (ref. 22)	0.002 ± 0.022(ref. 22)
Carbonate	0.125	0.076		0.080 ± 0.03(ref. 7)
	0.014	0.052		0.067 ± 0.06(ref. 31)
Mean ± 2SD	0.086 ± 0.132			0.050 ± 0.082

---

Fig. 4 Average δ^114/110Cd values of the geological reference materials in this study compared with previously published data. Comment: The green circle and orange diamond represent the average value of δ^114/110Cd in this study and the reported data from the literature, respectively.

For the other five reference materials (NOD P-1, NOD A-1, COQ-1, JDo-1 and GSS-1a), three replicates were conducted for different digestions of the powder materials. All average values of δ^114/110Cd showed good agreement within uncertainty, indicating that the precision and accuracy of our separation method were robust and accurate. The Cd amounts shown in Table 5 were estimated according to the sample weights and previously published Cd concentration data.^18,20,21 The measured δ^114/110Cd values of these reference materials are listed in Table 5. The δ^114/110Cd values in these CRMs were obtained with a typical internal precision better than 0.050‰ (2 SE), which was consistent with the literature value^5,7,14–33 within error.

To evaluate the accuracy of δ^114/110Cd values obtained in this study, our results were compared with the published data. As shown in Fig. 4, the Cd isotope data show good agreement with the published datasets. Hence, these results demonstrate that the reproducibility and precision of the Cd isotope measurements achieved by the proposed method are satisfactory and fully meet the analytical requirements of environmental and geochemical research.

4. Conclusions

In this study, we present a two-stage column system combining the AG1-X8 anion resin and TBP extraction resin to purify Cd for the high-precision determination of Cd isotopic compositions of geological samples by double-spike TIMS. Our results demonstrate that the TBP extraction resin exhibits comparable separation performance to the widely used commercially available TRU and UTEVA Spec resins in removing Sn, while the cost of the TBP resin is only ∼5% that of these resins. The δ^114/110Cd values of the six reference materials relative to NIST SRM 3108 agreed well with the published values within analytical uncertainty. Our study thus offers a low-cost, high-recovery and low-blank chemical separation scheme for the high-precision determination of the Cd isotopic compositions of geological samples, even those with high Sn/Cd ratios.

Author contributions

C. F. Li: investigation, methodology, experiments, and writing-original draft. Z. Y. Chu: discussion and reviewing of the manuscript. P. Peng: discussion and reviewing of the manuscript. C. F. Li: project administration and funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

The authors confirm that all data supporting the findings of this study are presented in the Tables within the article.

Acknowledgements

This work was jointly supported by the State Key Laboratory of Lithospheric Evolution of IGGCAS (grant SKL-Z202401) and the National Natural Science Foundation of China (grants 42473036 and 42125206).

References

1. C. W. Zhu, H. J. Wen, Y. X. Zhang, Y. Z. Liu and R. F. Wei, Acta Geological Sinica, English edn, 2015, vol. 89, pp. 2048–2057.
2. Q. H. Zhong, Y. C. Zhou, D. C. W. Tsang, J. Liu, X. Yang, M. L. Yin, S. J. Wu, J. Wang, T. F. Xiao and Z. F. Zhang, Sci. Total Environ., 2020, 736, 139585–139594.
3. M. Imseng, M. Wiggenhauser, A. Keller, M. Müller, M. Rehkamper, K. Murphy, K. Kreissig, E. Frossard, W. Wilcke and M. Bigalke, Environ. Sci. Technol., 2018, 52, 1919–1928.
4. T. M. Conway and S. G. John, Geochim. Cosmochim. Acta, 2015, 148, 269–283.
5. G. Devos, F. Moynier, J. Creech, D. Liu, I. Puchtel and M. Bizzarro, Chem. Geol., 2024, 650, 121996–122016.
6. W. Abouchami, S. J. G. Galer, B. De, H. J. W, A. C. Alderkamp, R. Middag, P. Laan, H. Feldmann and M. O. Andreae, Earth Planet. Sci. Lett., 2011, 305, 83–91.
7. R. Frei, L. Xu, J. A. Frederiksen and B. Lehmann, Chem. Geol., 2021, 563, 120061–120081.
8. H. Peng, P. Liu, H. T. Zheng, N. S. Belshaw, S. H. Hu and Z. L. Zhu, Chem. Geol., 2023, 620, 121341–121352.
9. T. J. Horner, R. B. Y. Lee, G. M. Henderson and R. E. M. Rickaby, Proc. Natl. Acad. Sci. U. S. A, 2013, 110, 2500–2505.
10. K. J. R. Rosman and J. R. De Laeter, Int. J. of Mass Spectrom. Ion Phys., 1975, 16, 385–394.
11. K. J. R. Rosman and J. R. De Laeter, Nature, 1976, 261, 216–218.
12. F. Wombacher, M. Rehkamper, K. Mezger, A. Bischoff and C. Munker, Geochim. Cosmochim. Acta, 2008, 72, 646–667.
13. F. Wombacher, M. Rehkamper, K. Mezger and C. Munker, Geochim. Cosmochim. Acta, 2003, 67, 4639–4654.
14. C. Cloquet, O. Rouxel, J. Carignan and G. Libourel, Geostand. Geoanal. Res., 2005, 29, 95–106.
15. A. D. Schmitt, S. J. G. Galer and W. Abouchami, J. Anal. At. Spectrom., 2009, 24, 1079–1088.
16. W. Abouchami, S. J. G. Galer, T. J. Horner, M. Rehkämper, F. Wombacher, Z. C. Xue, M. Lambelet, M. Gault-Ringold, C. H. Stirling, M. Schönbächler, A. E. Shiel, D. Weis and P. F. Holdship, Geostand. Geoanal. Res., 2013, 37, 5–17.
17. N. Pallavicini, E. Engstrom, D. C. Baxter, B. Ohlander, J. Ingri and I. Rodushkin, J. Anal. At. Spectrom., 2014, 29, 1570–1584.
18. D. D. Li, M. L. Li, W. R. Liu, Z. Z. Qin and S. A. Liu, Geostand. Geoanal. Res., 2018, 42, 593–605.
19. M. S. Liu, Q. Zhang, Y. Zhang, Z. Zhang, F. Huang and H. M. Yu, Geostand. Geoanal. Res., 2019, 44, 169–182.
20. D. S. Tan, J. M. Zhu, X. L. Wang, G. L. Han, Z. Lu, W. P. Xu and J. Anal, J. Anal. Atomic Spectrom., 2020, 35, 713–727.
21. J. Borovicka, L. Ackerman and J. Rejsek, Talanta, 2020, 221, 121389–121398.
22. Z. Lu, J. M. Zhu, D. Tan, X. Wang and Z. Zheng, Geostand. Geoanal. Res., 2021, 45, 565–581.
23. H. Peng, D. He, R. Guo, X. Liu, N. Belshaw, H. Zheng, S. Hu and Z. L. Zhu, J. Anal. Atomic Spectrom., 2021, 36, 390–398.
24. X. Liu, L. J. Chen, H. Peng, G. Wang, N. S. Belshaw, H. T. Zheng, S. H. Hu and Z. L. Zhu, Anal. Chim. Acta, 2022, 1215, 339980–339988.
25. Q. Dong, C. L. Xiao, W. H. Cheng, H. M. Yu, J. B. Shi, Y. G. Yin, Y. Liang and Y. Cai, Atom. Spectros, 2022, 43, 145–153.
26. C. Guo, T. Li, G. J. Li, T. Y. Chen, L. Li, L. Zhao and J. F. Ji, J. Anal. At. Spectrom., 2022, 37, 2470–2479.
27. H. Chang, J. M. Zhu, X. L. Wang, T. Gao and J. Anal, J. Anal. Atomic Spectrom., 2023, 38, 950–962.
28. Q. H. Zhong, J. Li, L. Yin, N. P. Shen, S. Yan, Z. Y. Wang and C. H. Zhu, J. Anal. At. Spectrom., 2023, 38, 939–949.
29. Q. H. Zhong, L. Yin, J. Li, Y. X. Feng, N. P. Shen, B. Y. Peng and Z. Y. Wang, J. Anal. At. Spectrom., 2023, 38, 2291–2301.
30. Z. Y. Zhang, T. Li, B. C. Li, Y. B. Lin, G. J. Li, T. Y. Chen and J. Anal, J. Anal. Atomic Spectrom., 2024, 39, 1142–1151.
31. C. F. Li, Z. Y. Chu and P. Peng, Anal. Chem., 2024, 96, 15436–15445.
32. X. Wu, Z. Y. Wang, G. Y. Sun, Y. Lin, X. W. Fu, Y. Tang and X. B. Feng, J. Anal. At. Spectrom., 2024, 39, 2298–2308.
33. J. Li, S. H. Tang, X. K. Zhu, J. X. Ma, Z. Y. Zhu and B. Yan, J. Anal. At. Spectrom., 2024, 39, 3106–3115.
34. C. F. Li, L. J. Feng, X. C. Wang, Z. Y. Chu, J. H. Guo and S. A. Wilde, J. Anal. At. Spectrom., 2016, 31, 2375–2384.
35. C. F. Li, L. J. Feng, X. C. Wang, S. A. Wilde, Z. Y. Chu and J. H. Guo, J. Anal. At. Spectrom., 2017, 32, 1938–1946.
36. L. J. Xu, K. Zhu, Q. Y. Man, J. Lewis, H. B. Hai and S. A. Liu, Atom. Spectros, 2023, 44, 142–152.
37. C. F. Li, Z. Y. Chu, X. C. Wang, J. H. Guo and S. A. Wilde, Anal. Chem., 2019, 91, 7288–7294.
38. A. Heuser, A. Eisenhauer, N. Gussone, B. Bock, B. T. Hansen and T. F. Nägler, Int. J. Mass Spectrom., 2002, 220, 385–397.

---