High-precision measurement of Ag isotopes for silicate rocks by MC-ICP-MS

Abstract

Silver (Ag) isotopes can be useful geochemical and environmental tracers, but their applications are hindered by the challenges of Ag isotope analyses, especially for samples with extremely high matrix element and low Ag contents (<100 ng g⁻¹). In this study, we developed a new protocol by using four-step chromatography for high-precision Ag isotope measurements for silicate samples. The δ¹⁰⁹Ag values were measured using a multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) using dry plasma. The sensitivity was 100 V ppm⁻¹, reducing the consumption of Ag to 5 ng. Moreover, instrumental mass bias was corrected using the Pd-doping and standard-sample bracketing methods. The robustness of this method was assessed by analyzing the NIST SRM978a doped with different matrix element ratios. The average δ¹⁰⁹Ag_SRM978a value of synthetic solutions (0.00 ± 0.05‰; 2SD, n ≥ 4) is consistent with the recommended values (0). The external precision is 0.05‰ (2SD) based on the measurement of two in-house standards. We further analyzed seven USGS rocks, including basalt (BCR-2 and BIR-1), andesite (AGV-2), granite (GSP-2 and G-2), rhyolite (RGM-2), and ferromanganese crusts (NOD-P-1). Their δ¹⁰⁹Ag_SRM978a values vary from −0.24 ± 0.05‰ to 0.20 ± 0.05‰, suggesting that Ag isotopes can be significantly fractionated in silicate rocks. This study provides a great opportunity for the future application of Ag isotopes in geochemical and environmental studies.

---

Introduction

Silver (Ag) is a moderately incompatible, chalcophile, and siderophile element. The physicochemical characteristics of silver lead to numerous applications. As a noble metal, silver has been widely employed as a medium of international and domestic trade. In archeology, tracing the Ag origin can provide insights into historical development of human society.^1,2 Ag is enriched in the core relative to the mantle during core–mantle segregation of planets.³ During partial melting of the mantle, Ag prefers melts and sulfur-rich hydrothermal fluids relative to the residual peridotite. Therefore, Ag is more enriched in the continental crust³ (50 ng g⁻¹) than in the mantle (8 ng g⁻¹ in Bulk Silicate Earth³).

Silver has two stable isotopes: ¹⁰⁷Ag (51.839%) and ¹⁰⁹Ag (48.161%).⁴ The isotope composition is usually expressed as: δ¹⁰⁹Ag = [(¹⁰⁹Ag/¹⁰⁷Ag)_sample/(¹⁰⁹Ag/¹⁰⁷Ag)_standard − 1] × 1000 (‰), which is the per mil deviation of the ¹⁰⁹Ag/¹⁰⁷Ag of the sample relative to that of the reference material NIST SRM978a. Ag isotopes can be a new tool to study a number of important geochemical problems, such as the early stages of planetary evolution,⁵ the formation of mineral deposits,^6,7 and the historical development of civilizations.^1,2 The key to these applications is to precisely measure Ag isotope ratios. Although thermal ionization mass spectrometry (TIMS) was used to analyze radiogenic isotopes (¹⁰⁷Pd–¹⁰⁷Ag) of iron and CI chondrites,⁸ it cannot precisely correct mass fractionation which results in an error of 2‰ (2SD).⁸ Application of multi-collector inductively coupled plasma-mass spectrometry (MC-ICP-MS) significantly improved the precision of Ag stable isotope measurement,^9–12 but previous methods may not be appropriate for low Ag sample analyses such as silicate materials because of the following limitations:

(1) At least 20 ng (typically about 100 ng) of Ag is used for its isotope measurement.^7,13 However, this would require digestion of too many samples with low Ag content such as most terrestrial and extra-terrestrial samples.

(2) The purification procedure for samples with a low content of Ag (<50 ng g⁻¹) is complicated and very challenging because of the high matrix/Ag ratios.¹⁴

(3) Only two international silicate reference materials (BHVO-2 and AGV-2) were reported for Ag isotope ratios,^9,10,15 which hinders the comparison of data among different laboratories.

In order to solve the above-mentioned problems, we developed an analytical protocol to measure Ag isotope compositions of samples with low Ag content. The Ag was separated from the matrix with a four-step chromatographic procedure in silicate rocks. The Ag isotope ratios were determined using a MC-ICP-MS with dry plasma, which significantly reduces Ag consumption down to merely 5 ng. Furthermore, the instrumental mass-bias was calibrated by Pd-doping and standard-sample bracketing (SSB) methods. With this method, we report the first δ¹⁰⁹Ag data of five silicate reference materials. Because our new protocol can measure samples with various Ag contents, we can explore more geoscience problems using Ag isotope tracers.

Analytical methods

Reagents and reference materials

Concentrated acids (HCl, HNO₃, and HF) used in this study were purified by a sub-boiling distillation system (Milestone Inc.; Shelton CT, USA) of reagent grade feedstock (Fisher Scientific; Ottawa ON, Canada). Hydrogen peroxide solution 31% (H₂O₂) was from ICPMSPURE, China. Ultrapure water (18.2 MΩ) (Milli-Q, Sigma-Aldrich; Oakville ON, Canada) was employed to dilute the reagents.

The Ag isotope primary standard was provided by the National Institute of Standards and Technology (NIST). The absolute silver isotope ratio (¹⁰⁹Ag/¹⁰⁷Ag) is 0.92904 ± 0.00022, and the nuclide relative masses of ¹⁰⁷Ag and ¹⁰⁹Ag are 106.90509 and 108.90475.¹⁶ NIST SRM978a was dissolved in 2% HNO₃ and stored in a 30 mL beaker, and was further diluted to 0.5 μg g⁻¹ for measurements. Two in-house laboratory standards, USTC-Ag (≥99.9999% silver nitrate standard solution, trace metal base from MERCK) and ICP-Ag (a domestic single-element silver standard solution), were regularly measured to monitor the instrumental stability and long-term external precision. To avoid potential light effects, the beakers were placed in a PET lightproof sealed box.

Sample preparation and Ag purification

Chemical digestion and Ag purification were performed in an ISO Class 6 clean room at the CAS Key Laboratory of Crust-Mantle Materials and Environments, USTC. 100 mg USGS petrological powder was digested in 6 mL of a 3 : 1 (v/v) mixture of concentrated HF–HNO₃ solution in a capped 15 mL Teflon PFA beaker (Savillex). The solutions were heated at 130 °C on a hotplate for 72 hours and subjected to ultrasonic treatment several times for complete dissolving. After evaporation, 6 mL of concentrated aqua regia (HNO₃–HCl ∼1 : 3, v/v) was added to beakers and heated at 125 °C on a hotplate for one day to ensure that residual fluoride was eliminated. The samples were dried down and converted to chloride form by adding 1 mL of concentrated HCl. After sample digestion, the solutions were evaporated and residues were re-dissolved in 0.5 mol L⁻¹ HCl for the following chromatographic procedures. Before chemical purification, the Ag contents of USGS rock standards were measured by ICP-MS. The Ag contents of solutions are consistent with the recommended values within error.

For high-precision Ag isotope measurements in natural samples, it is crucial to separate Ag completely from matrix elements such as Cu, Zn, Sr, Y, Zr, and Nb. These elements could interfere with Ag isotope analyses in various forms, as listed in Table 1. We improved the chemical procedure from the protocols of Schönbächler et al.¹⁰ and Luo et al.,¹⁷ and our method can significantly reduce acid consumption and the duration of purification. Since the residual Ag on the resin may increase the procedure blank,¹⁸ the resins were well cleaned before conditioning. The anion resin was cleaned with 30 mL of 9 N HCl, followed by 5 mL of Milli-Q H₂O. The cation resin was cleaned with 30 mL of 6 N HNO₃, followed by 5 mL of Milli-Q H₂O. Both resins were filled in Savillex microcolumns with an inner diameter of 6 mm and height of 7 cm. The first chromatographic procedure (columns 1 and 2) is based on anion resin Bio-Rad AG1-X8 (200–400 mesh). After loading the sample, most major elements were preferentially eluted by 0.5 mol L⁻¹ HCl with 0.001% H₂O₂, while Ag was well adsorbed on the resin. After that, Zn and Cd were removed in sequence by 0.05 mol L⁻¹ HCl with 0.001% H₂O₂ (ref. 19) and 0.001 mol L⁻¹ HCl (Fig. 1a). Then, the Ag elution was collected and dried down for the following chromatographic procedures (columns 3 and 4) with cation resin AG50 W-X8 (200–400 mesh) to further separate Ag from Nb, As, and other minor matrices (Fig. 1b).

Table 1 Isobaric and polyatomic interferences on Pd and Ag isotopes

	Isobaric interferences	Polyatomic interferences
^104Pd		^88Sr^16O
^105Pd		^65Cu^40Ar^+, ^89Y^16O^+
^106Pd	^108Cd^+	^66Zn^40Ar^+, ^90Zr^16OH^+
^107Ag		^73As^34S, ^67Zn^40Ar^+, ^71Ga^36Ar^+, ^91Zr^16O^+, ^91Zr^16OH^+, ^92Zr^16O^+
^108Pd	^108Cd^+	^68Zn^40Ar^+, ^92Mo^16O^+
^109Ag		^69Ga^40Ar^+, ^73Ge^36Ar^+, Nb^93O^16+
^110Pd	^110Cd^+

---

Fig. 1 Elution curves of RGM-2 for the Ag purification procedure by anion-exchange resin AG1-X8 (a) and cation-exchange resin AG50 W-X8 (b).

The residuals of matrix elements may result in significant deviations on Ag isotope measurements. For synthetic solutions, purification using one column of anion resin and cation resin was enough to completely separate Ag. However, natural samples require repeating the elution of anion resin twice to remove most of the matrix. The Ag fraction was dried and then treated with 1 mL of 0.5 mol L⁻¹ HCl with 0.001% H₂O₂ before being loaded onto the cation resin column. Then, Ag samples were purified using the cation resin procedure twice to completely remove As, Sn, and Nb. Notably, although some elements such as Ti can be eluted in the Ag fraction with cation resin AG50 W-X8 (Fig. 1b), they have been effectively separated by anion exchange column AG1-X8 (Fig. 1a). The contents of potential matrix/Ag ratios after the four-step chromatography process are listed in Table S1.† The anion resin procedure took around 10 hours for separation after pre-cleaning, while the cation resin took about 6 hours. After purification, the Ag fraction was evaporated and then 1 mL of 2% HNO₃ was added for MC-ICP-MS measurement. Before and after the collection of the Ag fraction, 1 mL of solution was collected to assess the possible loss of Ag during the chemical purification. The overall chemical procedure yield of Ag was >95%. The detailed purification procedure is listed in Table 2.

Table 2 Purification procedure of Ag

Separation stage	Eluent	Vol. (mL)
Columns 1 and 2, Bio-Rad AGI-X8, 200–400 mesh, 2 mL resin bed
Cleaning	9 mol L^−1 HCl	30
Conditioning	0.5 mol L^−1 HCl + 0.001% H2O2	10
Sample loading	0.5 mol L^−1 HCl + 0.001% H2O2	1
Elution of matrix elements	0.5 mol L^−1 HCl + 0.001% H2O2	26
Elution of Zn	0.05 mol L^−1 HCl + 0.001% H2O2	10
Elution of Cd	0.001 mol L^−1 HCl	10
Elution of Ag	1 mol L^−1 HNO3	30
Columns 3 and 4 Bio-Rad AG50 W-X8, 200–400 mesh, 2 mL resin bed
Cleaning	6 mol L^−1 HNO3	30
Conditioning	0.5 mol L^−1 HNO3	10
Sample loading	0.5 mol L^−1 HNO3	1
Elution of matrix elements	0.5 mol L^−1 HNO3	40
Elution of Ag	0.5 mol L^−1 HNO3	24

---

Mass spectrometry

The Ag isotope compositions were determined using MC-ICP-MS (Thermo-Fisher Scientific Neptune Plus) in the CAS Key Laboratory of Crust-Mantle Materials and Environments, USTC. Because Ag only has two stable isotopes, SSB and Pd internal mass bias correction were combined to calibrate the instrumental mass bias. The Pd standard (NIST3138 solution) was doped into both samples and standards in this study.

The Ag content in rock samples is generally low (∼50 ng g⁻¹),^20,21 posing a challenge to provide enough Ag for Ag isotope measurement. To solve this problem, we used Aridus III membrane desolvation (CETAC) for dry plasma to increase the sensitivity. The MC-ICP-MS was equipped with a Jet sampler cone and H skimmer cone. Multiple Faraday cups were applied to simultaneously collect signals in low-resolution mode, including L3 (¹⁰⁴Pd), L2 (¹⁰⁵Pd), L1 (¹⁰⁶Pd), C (¹⁰⁷Ag), H1 (¹⁰⁸Pd), H2 (¹⁰⁹Ag), H3 (¹¹⁰Pd), and H4 (¹¹¹Cd). The potential Cd interferences on ¹⁰⁶Pd, ¹⁰⁸Pd, and ¹¹⁰Pd were corrected by monitoring ¹¹¹Cd. Each measurement cycle consisted of serial washes with 5% and 2% HNO_3, and 40 isotope ratio measurements in a single block with 4.194 s integration time. To eliminate the memory effect, it took almost 9.5 minutes to thoroughly clean the system before taking sample measurement.

Prior to MC-ICP-MS analysis, all solutions were spiked with Pd at the appropriate ratio to ensure equivalent signals between ¹⁰⁷Ag and ¹⁰⁸Pd. Instrumental mass bias correction was performed by internally normalizing to a known Pd isotope ratio, based on the exponential mass fractionation law. The ¹⁰⁷Ag and ¹⁰⁸Pd signals were approximately 0.5 V with 50 μL min⁻¹ took up rate for the solutions with an Ag content of 5 ng g⁻¹. Each sample was analyzed at least three times in parallel, and the reference material NIST SRM978a was used as the bracketing standard to correct the instrumental mass bias. Meanwhile, the ICP-Ag and USTC-Ag solutions were used as in-house standards to monitor the stability of the instrument after every 4 analyses of samples. The instrumental parameters for MC-ICP-MS are summarized in Table 3.

Table 3 Instrument parameters for Ag isotope measurements by MC-ICP-MS

Solution MC-ICP-MS	Thermo Fisher Scientific, Neptune Plus
Cooling gas (Ar)	15 L min^−1
Auxiliary gas (Ar)	0.60 L min^−1
Mass resolution	Low resolution
Typical sensitivity	100 V/(μg g^−1) for ^107Ag
Interface cones	Jet sample cone + H skimmer cone
Ar sweep	5.45 L min^−1
Solution uptake	50 μL min^−1
Integration time	4.194 s per cycle
Aridus gas (N2)	∼1.5 L min^−1
Faraday cup configuration	L3 ^104Pd	L2 ^105Pd	L1 ^106Pd	Central ^107Ag	H1 ^108Pd	H2 ^109Ag	H3 ^110Pd	H4 ^111Cd

---

Tests of method performance

To evaluate the effect of matrix and the acidity of HNO₃ on Ag isotope measurements, the doping experiments were conducted by adding certain amounts of Fe, Cu, Zn, Zr, Nb, and Cd to the standard solution with 5 ng g⁻¹ Ag. The molar ratios of Fe : Ag ranged from 1 : 10 to 5 : 1, Cu : Ag from 1 : 2 to 2 : 1, Zn : Ag from 1 : 100 to 1 : 2, Nb : Ag from 1 : 100 to 1 : 2, and Cd : Ag from 1 : 1 to 10 : 1, respectively. Additionally, different concentrations of HNO₃ were diluted to test the impact on Ag isotope measurements. A solution of SRM978a was diluted to a concentration of 5 ng g⁻¹ by adding 2% HNO₃ (m/m) as a standard, while another solution of SRM978a was diluted using varying concentrations of HNO₃ ranging from 1% to 3%. These diluted solutions were then treated as samples for bracketed measurements.

We further determined the concentration effect on Ag isotope composition when there was a discrepancy between the sample and the standard. The sample/standard ranged from 0.5 to 2. These solutions were measured by MC-ICP-MS with the same method as described above.

Results and discussion

Establishment of elution curves

In order to ensure reliable and robust measurements, the elution curves for Ag should be consistent for different samples. In this study, a series of experiments were carried out using synthetic solutions and natural samples. We tested whether the leaching curves for Ag are uniform in the solutions with different matrixes (Table S2†). Then we conducted experiments on both synthetic solutions and rock standards with the same procedure. The data show that the Ag isotope compositions of the synthetic solutions with different Ag contents are consistent with the recommended values (Table S2†). The elution curves of the synthetic solutions were the same as that of the geological reference material RGM-2 (Fig. 1), while rhyolite (RGM-2), basalt (BCR-2), and andesite (AGV-2) samples with an equivalent mass of Ag have almost the same elution position for Ag (Fig. 1a, S1a and b†), indicating that the chemical separation effect could be the same for different types of geological samples. From this, we can conclude that changes in the Ag or matrix content have little effect on the Ag elution during the chemical purification.

As depicted in Fig. 1, some minor elements still exist in the Ag elution. The potential spectral interferences can compromise the accuracy and precision of Ag isotopes in MC-ICP-MS analyses. Specifically, matrix components (e.g., ⁹³Nb¹⁶O, ⁷³As³⁴S) have isobaric effect on high precision Ag isotope measurement. Because geological samples such as basalt and ferromanganese nodules exhibit elevated Nb content and Nb/Ag ranging from 100 to 500, cation resin was used to mitigate Nb and As in purified Ag solution. We evaluated the elution efficiency at different concentrations of HNO₃, specifically 0.5 mol L⁻¹ (Fig. 1b), 1 mol L⁻¹, and 2 mol L⁻¹ (Fig. S1(c) and (d)†). The results show that the 0.5 mol L⁻¹ HNO₃ on the cation resin can well separate Nb, Ti, and As from Ag. For samples with low Nb/Ag (<0.01), the fourth column with AG50 W-X8 is unnecessary. We suggest that for silver-containing sulfide minerals, especially for samples with high Ag content, two steps may be enough for a good purification.

Spectral interferences

Because the Ag contents of silicate rock samples are normally low,³ the matrix components may result in substantial interference with Ag isotope determination. For instance, the signals of ⁶⁵Cu⁴⁰Ar⁺, ⁶⁶Zn⁴⁰Ar⁺, ¹⁰⁶Cd⁺, ⁶⁸Zn⁴⁰Ar⁺, ⁹²Mo¹⁶O⁺, and ¹⁰⁸Cd⁺ potentially generate isobaric effect on those of ¹⁰⁴Pd, ¹⁰⁵Pd, ¹⁰⁶Pd, ¹⁰⁸Pd, and ¹¹⁰Pd, respectively, which may cause interference with the correction for Pd. Similarly, the polyatomic interferences from ⁶⁷Zn⁴⁰Ar⁺, ⁹¹Zr¹⁶O⁺, ⁹³Nb¹⁶O⁺, ⁶⁹Ga⁴⁰Ar⁺, and ⁷³Ge³⁶Ar on Ag isotopes may impact the precision of Ag isotope ratios. In particular, the presence of trace amounts of Nb and Zr can result in significant positive or negative deviations from the recommended δ¹⁰⁹Ag values (Fig. S2†). Therefore, we doped varying amounts of elements into the NIST SRM978a to obtain the window of allowed matrix contents during the Ag isotope determination. The results are presented in Table S3.† The δ¹⁰⁹Ag values of solutions approach the recommended value (0 ± 0.05‰) when the molar ratios of Fe/Ag < 1, Cu/Ag < 1.5, Zn/Ag < 1.5, Zr/Ag < 0.1, Nb/Ag < 0.2, and Cd/Ag < 10 (Fig. 2). Within the above range, the influence of these matrix elements on Ag isotope measurements is negligible.

Fig. 2 The δ¹⁰⁹Ag_SRM978a values of the doping solutions with different molar ratios of Fe, Cu, Zn, Zr, Nb, and Cd to NIST SRM978a were analyzed by MC-ICP-MS. Error bars represent the 2SD uncertainties for each doping solution. The gray band represents the 2SD uncertainties of 0.05‰ (n ≥ 3) for MC-ICP-MS analyses.

Instrumental mass fractionation correction

Measurements of Ag isotopes may possibly be subjected to instrumental drift as a result of various factors, including matrix effects, plasma systems, and environmental conditions (e.g., temperature and humidity). The instrumental fractionation can be corrected by adding Pd to the measured solutions to get high-quality data. Both the sample and standard solutions were doped with the same amount of NIST 3138. The Pd isotope signal ratios were determined to calculate the instrumental mass-dependent fractionation factors.

Assuming the measurement exhibits equivalent mass bias factor (f) for both Pd and Ag, the equation can be expressed as follows:

where R_T is the true isotope ratio of the sample, R_m is the measured isotope ratio, f is the mass bias factor, and m_i and m_j are the relative atomic masses of the isotopes. Since the Pd isotope ratio (R_T-Pd) is known, the corresponding measured ratio (R_m-Pd) is used in the equation to calculate the mass deviation factor (f) during the measurement. Based on the mass deviation factor (f), the real ratio of Ag isotopes can be obtained (R_T-Ag) from the measured ratio of Ag isotopes (R_m-Ag). Considering that there may be little difference between the mass biases of Pd and Ag, the SSB method is used to further deduce the instrumental effect. The Ag isotope composition is expressed as the deviations between the R_corrected of the sample and the adjacent standard.

δ^109/107Ag = (2 × R_{corrected-sample}/(R_{corrected-standard-last} + R_{corrected-standard-next}) − 1) × 1000‰ (2)

In order to maintain a consistent correlation between the mass factors of Ag and Pd during the measurement, it is important to identify an appropriate concentration ratio between Ag and Pd. Therefore, the quantity of doped Pd will exert a substantial influence on the calibration of Ag isotopes. Because deviations from the optimal Pd doping levels might result in a drift in the calibration results, the amount of Pd doped in both the sample and the standard solutions should be carefully evaluated. Pd and Ag signals between the sample and the standard are closely aligned to minimize the discrepancies. When ¹⁰⁸Pd/¹⁰⁷Ag is less than 0.1, the isotope composition of the corrected Ag will experience a significant shift. Similarly, excessive Pd doping may interfere with and even damage the MC-ICP-MS. Consequently, we recommend to maintain the Pd content within a specific range of ¹⁰⁸Pd/¹⁰⁷Ag from 0.1 to 10. This range promises stability in measurement and the ¹⁰⁹/¹⁰⁷Ag can be accurately corrected (Fig. 3a). We also noticed that the Pd-doped solution prepared too early may result in notable deviations in the Ag isotopes, which is probably due to the selective adsorption of Ag and Pd in the solution onto the inner surfaces of the tubes. As a result, we suggest to prepare the Pd-doped solution immediately before the measurement.

Fig. 3 (a) The variation in δ¹⁰⁹Ag_SRM978a values of NIST SRM978a with different Pd/Ag molar ratios. (b) The variations in δ¹⁰⁹Ag_SRM978a during the analysis of Ag standard solutions diluted with varying concentrations of HNO₃. (c) The variations in δ¹⁰⁹Ag_SRM978a during the analysis of Ag standard solutions with different concentrations between the sample and standard. Error bars represent the 2SD uncertainties for each solution. The band represents the 2SD (n ≥ 3) uncertainties of 0.05‰ for MC-ICP-MS analyses.

The calibration process requires a careful selection of stable isotope pairs of Pd. Pd has six stable isotopes: ¹⁰²Pd (1.02%), ¹⁰⁴Pd (11.14%), ¹⁰⁵Pd (22.33%), ¹⁰⁶Pd (27.33%), ¹⁰⁸Pd (26.46%), and ¹¹⁰Pd (11.72%), and the specific choice of Pd correction varies across different laboratories. With excessive concentrations of Cu ions, such as the case of measuring natural silver samples, the presence of ¹⁰⁵Pd can lead to isobaric interference between ¹⁰⁶Pd and ⁶⁵Cu⁴⁰Ar⁺. Similarly, the isotope pair ¹⁰⁴Pd–¹⁰⁵Pd was selected to address the mass bias that may arise due to potential interference between ¹⁰⁶Cd and ¹⁰⁸Cd with ¹⁰⁶Pd and ¹⁰⁸Pd, particularly when the sample contains a high concentration of Cd¹⁷. To mitigate the influence of ²⁰⁸Pb on ¹⁰⁴Pd, the ¹⁰⁵Pd–¹⁰⁸Pd internal standard was used when the concentration of Pb ions in the sample was comparatively elevated.^13,22 We used the standard SRM978a, the laboratory internal standard ICP-Ag, and the USTC-Ag solution, to evaluate the correction effect of various Pd isotope ratios. The correction effect is presented in Fig. 4, reflecting the measurements over a 24 hour period. The normalized fits of ln(¹¹⁰Pd/¹⁰⁶Pd) and ln(¹⁰⁹Ag/¹⁰⁷Ag) demonstrated the best (R² = 0.987) among all possible Pd isotopes pairs (Fig. 4). This result corresponds well to the findings of Brügmann et al.¹² The procedure in this study demonstrates complete separation. The impact of matrix components such as Cu and Cd is almost negligible, but the ¹⁰⁶Pd–¹⁰⁸Pd for data calibration was notably biased due to the presence of ⁶⁸Zn⁴⁰Ar⁺. With dry plasma in high sensitivity, Zn content is on the same scale (ng) as Ag, which has significant effects on the signal of ¹⁰⁸Pd.

Fig. 4 Natural logarithms of Ag and Pd isotope ratios for the NIST SRM978a, ICP-Ag, and USTC-Ag solutions measured as bracketing standards (regression line) fall on the same or similar linear trends.

Precision and accuracy

Studies have shown that acid molarity and concentration mismatch may affect isotope composition.²³ In this study, a series of tests were carried out to determine the potential influence on Ag isotope measurements. The results suggest that acidity variation effect is limited when diluted HNO₃ is 1% to 3%. In addition, because instrumental mass fractionation between samples and standards should be consistent, it is critical to ensure that the samples and standards contain the same amount of Ag. As shown in Fig. 3, the mismatch has little effect on Ag isotope determination when sample/standard Ag concentration ratios are 0.6–1.4. Thus, sample-standard concentration differences were within 10% tolerance during analysis.

Long-term measurements of reference materials were conducted to assess the precision. Since there is currently only one international Ag isotope standard in common use (NIST SRM978a), we developed two new in-house standards, ICP-Ag and USTC-Ag. After three years of repeated measurements, their average values of ¹⁰⁹Ag_SRM978a were recommended to be 0.01 ± 0.05‰ and −0.02 ± 0.05‰, respectively (Fig. S3†). Furthermore, data quality was monitored using multiple internal standards, and the long-term external precision was better than 0.05‰ (2SD, n ≥ 4) at a 95% confidence level.

Ag isotope compositions of the USGS standards

Current Ag isotope studies focus on studies on Ag-ore deposit and cosmochemistry. More investigations for terrestrial and extra-terrestrial samples are needed to improve our understanding of the mechanisms governing Ag isotope fractionation. We analyzed 7 USGS standards: basalt (BCR-2 and BIR-1), andesite (AGV-2), granite (GSP-2 and G-2), rhyolite (RGM-1), and ferromanganese crusts (NOD-P-1) for an inter-laboratory comparison. The Ag isotope data of rock standards are still rare in the literature.^9,10,15 Our findings agree well with reported data within error for basalt BCR-2 (0.20 ± 0.04‰ vs. 0.24 ± 0.07‰) and andesite AGV-2 (0.07 ± 0.04‰ vs. 0.3 ± 0.2‰).

According to the data in Fig. 5 and Table 4, most samples were analyzed with a precision better than 0.05‰. The δ¹⁰⁹Ag_SRM978a values of the silicate rocks ranged from −0.24 ± 0.05‰ to 0.20 ± 0.05‰, indicating significant Ag isotope fractionation during various geological processes. Because of its low Ag content, the analytical uncertainty in the granite G-2 analysis is higher (2SD, 0.08‰). δ¹⁰⁹Ag_SRM978a values of the standards measured in this study are not correlated with the content of Ag, SiO₂, or δ⁶⁵Cu values (Fig. 6a, b, and, f). In contrast, Fig. 6(c–e) depict the relationship between δ¹⁰⁹Ag_SRM978a and F, Cl, and S contents, showing an increasing trend followed by a decrease, suggesting that halogens in the hydrothermal fluid process may affect the Ag isotope composition. Besides, Fig. 5 shows that Mn-nodule NOD-P-1 has relatively lighter Ag isotopes, indicating that Ag isotope fractionation may occur during the seawater alteration. Furthermore, because Ag is a redox-sensitive element, electron migration from Ag⁰ to Ag⁺ may also fractionate Ag isotopes. In summary, our findings highlight the promising future of Ag isotope data in future studies.

Fig. 5 The δ¹⁰⁹Ag_SRM978a values of geological reference materials. Solid symbols indicate the δ¹⁰⁹Ag_SRM978a values cited from the literature.^9,10,15 Error bars represent the 2SD (n ≥ 3) uncertainties for each sample. The gray band represents the Ag isotope composition of BSE.¹⁰

Table 4 Ag isotope compositions of standards relative to NIST SRM978a

USGS standard	Rock types	Ag (μg g^−1)	δ^109AgSRM978a	2SD	N
a The reported Ag isotope data were cited from Schönbächler et al., 2007.^10 b Zhu et al., 2023,^15 and c Woodland et al., 2005.^9 d The Ag contents in USGS rock standards were obtained from Axelsson et al., 2002.^24 e Chen et al., 2020.^25 f Braukmüller et al., 2020.^20 g Wang et al., 2014,^18 and h Zou et al., 2020.^21
NOD-P-1	Ferromanganese crusts	0.21^d	−0.24	0.07	6
BHVO-2^a	Basalt	0.44^e	0.16	0.07	—
BHVO-2^b	Basalt	0.44^e	−0.04	0.06	≧4
BCR-2^a	Basalt	0.03^f	0.24	0.07	—
BCR-2	Basalt	0.03^f	0.20	0.04	5
BIR-1	Basalt	0.03^f	−0.24	0.04	3
AGV-2^c	Andesite	0.05^f	0.30	0.20	—
AGV-2	Andesite	0.05^f	0.07	0.04	5
SCO-1^c	Cody shale	0.13^g	0.08	0.05	10
GSP-2	Granite	0.09^h	−0.21	0.05	7
G-2	Granite	0.03^f	−0.07	0.09	5
RGM-2	Rhyolite	—	−0.04	0.05	6

---

Fig. 6 The δ¹⁰⁹Ag_SRM978a values of geological reference materials as a function of Ag content (a), SiO₂ content (b), S content (c), F content (d), Cl content (e), and δ⁶⁵Cu values (f).

Conclusions

(1) This study improved the chemical purification procedure by using AG1-X8 and AG50 W-X8 resins. The method can purify Ag from a complex matrix with recovery rate better than 95%.

(2) The MC-ICP-MS with dry plasma reduces the sample consumption to merely 5 ng. The SSB method with Pd addition was used to calibrate the instrumental fractionation. Two new in-house standards, ICP-Ag (δ¹⁰⁹Ag_SRM978a = −0.01 ± 0.05‰) and USTC-Ag (δ¹⁰⁹Ag_SRM978a = 0.02 ± 0.05‰) were developed to monitor data quality.

(3) The Ag isotope compositions of seven USGS geological standard were measured, and five of them were reported for the first time, ranging from −0.24 ± 0.05‰ to 0.20 ± 0.05‰. This study provides the analytical basis for further exploration of Ag isotopes.

Author contributions

Yuan Fang: methodology, formal analysis, investigation, writing – original draft, and writing – review & editing. Qiu-Yu Wen: formal analysis and investigation. Zi-Cong Xiao: formal analysis and writing – review & editing. Hui-Min Yu: methodology, validation and supervision. Fang Huang: conceptualization, validation, writing – review & editing and supervision.

Conflicts of interest

There are no conflicts of interest to declare.

Acknowledgements

We thank Lan-Lan Tian and Wei-Xin Lv for suggestions in early experiments. This study was supported by the National Natural Science Foundation of China (4233000014).

References

1. A. M. Desaulty, P. Telouk, E. Albalat and F. Albarède, Proc. Natl. Acad. Sci. U.S.A., 2011, 108(22), 9002–9007.
2. A. M. Desaulty and F. Albarede, Geology, 2013, 41(2), 135–138.
3. W. F. McDonough, Treatise Geochem., 2003, 2, 547–568.
4. K. Rosman and P. Taylor, Pure Appl. Chem., 1998, 70, 217–235.
5. R. W. Carlson and E. H. Hauri, Geochim. Cosmochim. Acta, 2001, 65(11), 1839–1848.
6. D. Wang, R. Mathur, Y. Y. Zheng, H. J. Wu, Y. W. Lv, G. Y. Zhang, Y. Huan, M. Yu and Y. J. Li, Miner. Deposita, 2022, 57, 701–724.
7. C. R. Voisey, R. Maas, A. G. Tomkins, M. Brauns and G. Brügmann, Econ. Geol., 2019, 114(2), 233–242.
8. J. H. Chen and G. J. Wasserburg, Geochim. Cosmochim. Acta, 1983, 47(10), 1725–1737.
9. S. J. Woodland, M. Rehkämper, A. N. Halliday, D. C. Lee, B. Hattendorf and D. Günther, Geochim. Cosmochim. Acta, 2005, 69(8), 2153–2163.
10. M. Schönbächler, R. W. Carlson, M. F. Horan, T. D. Mock and E. H. Hauri, Int. J. Mass Spectrom., 2007, 261(2–3), 183–191.
11. L. Yang, E. Dabek-Zlotorzynska and V. Celo, J. Anal. At. Spectrom., 2009, 24(11), 1564–1569.
12. G. Brügmann, M. Braunsa and R. Maas, Chem. Geol., 2019, 516, 59–67.
13. Q. Guo, H. Z. Wei, S. Y. Jiang, S. Hohl, Y. B. Lin, Y. J. Wang and Y. C. Li, Anal. Chem., 2017, 89(24), 13634–13641.
14. M. Matthes, M. Fischer-Gödde, T. S. Kruijer, I. Leya and T. Kleine, Geochim. Cosmochim. Acta, 2015, 169, 45–62.
15. Y. F. Zhu, H. Z. Wei, J. L. Wang, A. E. Williams-Jones, Z. C. Wang, S. Y. Jiang, S. V. Hohl, C. Huan and M. M. Zhang, ACS Earth Space Chem., 2023, 7, 2031–2041.
16. L. J. Powell, T. J. Murphy and J. W. Gramlich, J. Res. Natl. Bur. Stand., 1982, 87, 9–19.
17. Y. Luo, E. Dabek-Zlotorzynska, V. Celo, D. C. G. Muir and L. Yang, Anal. Chem., 2010, 82, 3922–3928.
18. Z. C. Wang, H. Becker and F. Wombacher, Geostand. Geoanal. Res., 2014, 39, 185–208.
19. S. H. Tang, B. Yan and J. Li, Geochim., 2013, 1, 46–52.
20. N. Braukmüller, F. Wombacher, A. Bragagni and C. Münker, Geostand. Geoanal. Res., 2020, 44(4), 733–752.
21. Z. Q. Zou, Z. C. Wang, H. Cheng, T. He, Y. N. Liu, K. Chen, Z. C. Hu and Y. S. Liu, Geostand. Geoanal. Res., 2020, 44(3), 501–521.
22. W. Argapadmi, E. R. Toth, M. A. Fehr, M. Schönbächler and C. A. Heinrich, Econ. Geol., 2018, 113(7), 1553–1570.
23. X. Y. Nan, F. Wu, Z. F. Zhang, Z. H. Hou, F. Huang and H. M. Yu, J. Anal. At. Spectrom., 2015, 30, 2307.
24. M. D. Axelsson, I. Rodushkin, J. Ingri and B. Öhlander, Analyst, 2002, 127, 76–78.
25. K. Chen, M. Tang, C. Lee, Z. C. Wang, Z. Q. Zou, Z. C. Hu and Y. S. Liu, Earth Planet. Sci. Lett., 2020, 531, 115971.

---

Footnote

† Electronic supplementary information (ESI) available: Additional tables and figures of the experimental details (DOCX). See DOI: https://doi.org/10.1039/d3ja00424d

---