Determination of mass-dependent variations in nickel isotope compositions using double spiking and MC-ICPMS

Abstract

We present a new technique for the accurate and precise determination of mass-dependent variations in nickel isotope compositions in geological materials. Our method involves an ion-exchange procedure comprising three columns and utilising the ability of Ni to form strong complexes with both ammonia and dimethylglyoxime. The separation procedure is independent of sample pH and works even for samples with large matrix to analyte ratios. Processed Ni solutions are free of matrix elements and direct isobars of Ni, and the yield is normally 85–95%. The purified Ni solutions were analysed using a Nu Plasma, multi-collector inductively coupled plasma mass spectrometer (MC-ICPMS), where instrumental mass fractionation—together with potential isotopic fractionation during chemical separation due to incomplete yield—was corrected for by a double-spike technique, where samples were spiked prior to column chemistry. Tests performed on both mixtures of synthetic and natural terrestrial standards demonstrates that the method is accurate. Replicate measurements of USGS reference materials (peridotite PCC-1, basalt BHVO-2, and shale SCo-1) yield a long-term external reproducibility (2 s.d.) of typically ± 0.07‰, ± 0.1‰, and ± 0.14‰ for ⁶⁰Ni/⁵⁸Ni, ⁶¹Ni/⁵⁸Ni, and ⁶²Ni/⁵⁸Ni respectively.

---

1 Introduction

Isotopic variations of the first row transition metals are of particular interest due to their abundance in geological samples and their utilisation in many biological processes. Nickel (Z = 28) has five stable isotopes, ⁵⁸Ni, ⁶⁰Ni, ⁶¹Ni, ⁶²Ni, and ⁶⁴Ni, with respective natural abundances of 68.0769%, 26.2231%, 1.1399%, 3.6345%, and 0.9256%.¹ Relatively few studies have addressed mass-dependent variations in Ni isotopes, in contrast to other transition metal isotope systems, such as iron and molybdenum,² which in part reflects the difficulty of purifying Ni from geological samples. Recent research into Ni isotope geochemistry has almost exclusively focused on extraterrestrial samples, searching for nucleosynthetic anomalies and radiogenic ⁶⁰Ni—the decay product of the extinct nuclide⁶⁰Fe.^3–5

In this study, we present a method for Ni separation from geological matrices and isotopic analysis of mass-dependent Ni isotope variations by multi-collector inductively coupled plasma mass spectrometer (MC-ICPMS), employing a Ni double spike to correct for instrumental mass bias and possible fractionation during separation. Nickel is the seventh most abundant transition metal in the Earth's crust, incorporated in both sulphide and silicate minerals and present in most geological reservoirs. Although Ni can posses a range of different oxidation states (from 0 to +IV) it is essentially only present in the +II state in natural samples. This lack of change in oxidation state with redox reactions renders Ni distinct from most other first row transition metals. Instead, Ni isotope fractionation may be caused by changes in speciation or coordination, making it an interesting addition to the already established field of non-traditional stable isotopes. Nickel also plays a key role in several enzymes important for the carbon cycle, including methane production and uptake of nitrogen during primary productivity,⁶ which makes Ni a necessary element for geologically interesting biological reactions.

As in other stable isotope systems, Ni isotope ratios are expressed in the delta (δ) notation as permil (‰) deviations in isotope ratios (ⁿNi/⁵⁸Ni, where n = 60, 61, or 62) relative to those of a pure Ni standard. The isotopic reference material used is a pure Ni metal from NIST (SRM 986). ⁵⁸Ni is both the most abundant isotope and the isotope of lowest mass, which is why isotope ratios are expressed using ⁵⁸Ni in the denominator.

2 Materials and methods

2.1 Sample preparation and dissolution

All chemical work during sample preparation was performed under clean laboratory conditions at the Department of Earth Sciences in Oxford. All acids (HCl, HNO₃, HF) used in the procedure were purified by in-house sub-boiling distillation. Solutions of dimethylglyoxime (DMG), ammonium citrate ((NH₄)₂C₆H₆O₇) and sodium hydroxide (NaOH) were prepared from crystalline commercial reagents, ammonia solution (NH₄OH), acetone and ethanol were reagent grade solutions, and dichloromethane (DCM) used for solvent extraction was of trace element grade—all from Fischer Scientific. Any water added during chemistry was 18.2 MΩ-grade taken from a Milli-Q Element water purification system.

For testing and initial comparison of different separation methods we used a calibration solution made up of elemental ICP-standards (Alfa Aesar) of geologically important matrix elements (e.g.Mg, Al, Ca, Fe, Ti, Mn) and elements with single or doubly charged species with charge/mass ratios overlapping with those of the Ni-isotopes (e.g. Cd⁺⁺ and Sn⁺⁺), mixed in concentrations approaching the natural composition of silicate rocks. The final separation method was tested by processing a variety of geological samples of different matrices, mainly USGS reference materials.

Weighed in samples of silicate reference materials (approximately 0.1 g) were dissolved by conventional digestion methods, using concentrated HF–HNO₃ mixtures, in a volume ratio of 5 : 1, in pre-cleaned Savillex Teflon beakers. The samples were left on a hotplate at 130 °C overnight. The sample solution was evaporated to dryness and the residue dissolved in small amounts (0.2 ml) of concentrated HNO₃ and left to evaporate. This procedure was repeated three times to completely drive off fluorides formed during the first digestion step. The last stage of digestion comprised refluxing the sample in approximately 5 ml 6 M HCl. This step was repeated twice if complete dissolution was not achieved in the first reflux cycle. During the whole digestion procedure the sample was treated several times in an ultrasonic bath for approximately 15 min at a time.

Due to the high carbon content, USGS rock standard SCo-1 was digested in an Anton Paar Multiwave 3000 microwave. Approximately 0.1 g of sample was added to a pre-cleaned Teflon XF-100 vessel, to which were added concentrated acids as follows: 4 ml HF, 2 ml HCl, and 2 ml HNO₃. The samples were digested for 1 h at 800 W, during which time the maximum pressure and temperature reached was 60 bar and 220 °C respectively. After full digestion the samples were transferred to pre-cleaned Savillex Teflon beakers and treated with multiple cycles of HNO₃ and ultimately 6 M HCl, in a similar manner to other silicate samples.

2.2 Chemical separation

Quantitative separation and purification of Ni from geological matrices is particularly challenging due to the similarity in the chemical behaviour of Ni compared with a number of other elements that are abundant in geological samples, such as Na, Mg, and Ca. This difficulty is further exacerbated by the abundance of the latter elements in geological samples relative to Ni, which is most often present in trace quantities.

As discussed above, previous Ni isotope studies have predominately focused on mass-independent isotope effects in iron meteorites,^3,4,7,8 from which separation of Ni is straightforward. The procedure usually comprises a relatively simple anion-exchange chemistry in HCl media, and quantitative yields of Ni are not required as any isotopic fractionation during chemical processing is corrected using internal normalization to an assumed Ni isotope ratio. So far, only a few published studies^7,9 have concerned analyses of mass-dependent Ni isotope variations. These studies employed a range of methods for the extraction of Ni from geological samples, which are mainly based on complexing Ni with DiMethylGlyoxime (DMG), an organic chelator highly selective for Ni under alkaline conditions. Quitte and Oberli⁷ proposed a solvent extraction method using DMG-in-ethanol and chloroform as the organic phase. Regelous et al.⁸ developed a three column set-up using a DMG-in-acetone mixture (originally based on Victor¹⁰), and Cameron et al.⁹ used a commercially available Nickel-Specific Resin by Eichrom Inc. (also suggested as an alternative in Quitte and Oberli⁷), which binds Ni to DMG in the resin polymer.

After testing the methods discussed above we found them not suitable for high precision analysis of mass-dependent Ni isotope fractionation. The DMG-in-acetone column procedure developed by Regelous et al.⁸ requires significant volumes of DMG per sample in order to fully separate Ni from Na, Mg and Ca, which display similar chemical behaviour to Ni. Such large quantities of DMG are potentially problematic as the Ni(DMG)₂-complex must be completely broken down prior to further separation as Ni otherwise may be lost. Methods of Ni purification employing Ni-specific resin^7,9 also involve large quantities of DMG, although in this case DMG is bound to the resin beads. In this technique, Ni is eluted in a mixture of DMG and resin polymer, which is very difficult to fully decompose. A further disadvantage of this method is that samples need to be adjusted to a particular pH before loading on the resin, and the success of the extraction procedure is highly dependent on sample pH.

Our method of Ni purification involves three steps of column chromatography and minimises the volumes of DMG to simplify the post-purification treatment of the sample. The first step involves separation of Ni from the s-block elements (e.g.Na, K, Mg, and Ca), and other geologically abundant cations (e.g.Al and Ti) by using a weak solution of di-ammonium citrate ((NH₄)₂C₆H₆O₇) in combination with excess NH₄OH. The chemical basis for this extraction method is that the excess ammonia binds to the more electronegative divalent ions present in the sample (e.g.Ni, Cu, Zn) by forming an ammonia-complex (e.g.Ni(NH₃)²⁺₄), which is strongly bound to the resin. Other elements (e.g.Mg, Al, Fe) prefer forming citrate complexes, which do not attach to the resin under these conditions and are eluted.¹¹ The second step of our separation procedure removes remaining Mg, Ca, NH⁺₄, Na, Al, Ti, and Cr and assures the complete removal of other remaining divalent elements. This step is adapted from Victor¹⁰ and Strewlow et al.¹² and employs firstly oxalic acid-HCl solutions which separate elements with valences > + II (e.g.Fe, Al, Cr, and Ti) which elute as oxalic complexes, and subsequently HCl-acetone mixtures which elute divalent elements (e.g.Mn, Cu, Zn, Cd, Sn). Nickel is finally eluted as a Ni-DMG complex in HCl-acetone media. The final chromatography step is a miniature clean up anion-exchange column, which completely separates Fe from Ni using HCl–H₂O₂ mixtures. These procedures are described in detail below and summarised in Table 1.

Table 1 Summary of full chemical procedure for separating Ni from a silicate matrix by ion-exchange chromatography

Col. 1(AG50W-X4)	Reagents/mixtures	volume(mL)
Equilibration:	0.1 M NH4-citrate + 3 M NH4OH	3
Introduction:	Sample solution dissolved in:
	2 ml 0.5 M HCl
	+ 0.3 ml 1M NH4-citrate
	+ 0.6 ml 15M NH4OH	3
Cleaning step 1:	0.1 M NH4-citrate + 3 M NH4OH	1
Cleaning step 2:	0.4 M HCl	1
Elution of Ni:	3 M HCl	3
Col . 2(AG50W-X4)

---

Equilibration:	0.1 M HCl + 0.05 M oxalic acid	1
Introduction:	Sample solution dissolved in:
	0.1 ml 1 M HCl
	+ 0.8 ml 18 MΩ H2O
	+ 0.1 ml 0.5 M oxalic acid	1
Cleaning step 1:	0.1 M HCl + 0.05 M oxalic acid	2
Cleaning step 2:	0.5 M HCl + 95% acetone	3
Elution of Ni:	0.5 M HCl + 95% acetone + 0.1 M DMG	1
Col . 3 ( AG1-X8 )

---

Equilibration:	6 M HCl + 0.5% H2O2	0.4
Introduction:	Sample solution dissolved in
	0.2 ml 6 M HCl + 0.5% H2O2	0.2
Elution of Ni:	6 M HCl + 0.5% H2O2	0.2

---

2.2.1 Nickel purification, step 1. Dissolved and spiked samples were evaporated and converted to chloride form, then re-dissolved in 2 ml 0.5 M HCl, to which 0.3 ml 1 M diammonium citrate and 0.6 ml concentrated NH₄OH were added, creating a solution of approximately 3 M NH₄OH in 0.1M diammonium citrate. The sample solutions were subsequently loaded onto ion exchange columns filled with approximately 0.8 ml of cation resin (AG50W-4X, chloride form, 200–400 mesh; 1 ml Tteflon columns) preconditioned with 3 M NH₄OH–0.1 M diammonium citrate. When the loading fraction had passed through the resin a final 1 ml of the same ammonia solution was loaded on the column to elute remaining unwanted elements. The resin was subsequently cleaned of ammonia by loading an additional 1 ml of 0.4 M HCl, which was eluted to waste. The Ni-ammonium complex was subsequently eluted in 3 mL 3 M HCl. The collected Ni fractions were subsequently evaporated on a hotplate to dryness. Ten drops of concentrated HNO₃ was added to break the Ni-ammonium complex, after which the solution was evaporated and converted to the chloride form by addition of 1 drop of 6 M HCl, and again evaporated to dryness. As can be seen in Fig. 1a most alkaline earth metals, as well as geologically abundant elements such as Al, Ti, Fe, and Cr are washed out with the first phase, and the only elements that appear to elute together with Ni are Cu and Zn.

Fig. 1 Calibration of the first two columns in the chemical separation procedure performed using column sizes and solvents as specified in Table 1. Both calibrations were performed using the same concentration of the calibration solution as described in the text. The results of the column calibration show that if a solution is passed through both columns all cations in the figure apart from Ni would be completely removed from the sample.

2.2.2 Nickel purification, step 2. The ammonia complexation method employed above is an extremely effective means of separating Ni from the bulk of the matrix elements in geological samples (mainly alkali and alkaline earth metals), and the Ni(NH₃)²⁺₄ complex can easily be broken using concentrated HNO₃ after the eluent has been dried down. However, this procedure results in the presence of residual ammonium salts in the Ni fraction which require removal. In order to remove such ammonium salts and divalent elements such as Cu and Zn, which are not separated in the first step, we use a second column chemistry step, adapted from ref. 10 and, ref. 12 which is tailored to remove any remaining Mg, Ca, NH₄, Na, Al, Ti, and Cr as well as other divalent elements. Nickel fractions from the first step were re-dissolved in 0.1 ml 1 M HCl, to which 0.8 ml 18.2 MΩ H₂O and 0.1 ml 0.5 M oxalic acid were added, creating a solution of 0.1 M HCl in 0.05 M oxalic acid. The sample solution is loaded onto an ion exchange column filled with approximately 0.2 ml of preconditioned cation exchange resin (same resin as Step 1; 0.3 ml Teflon columns). The loading fraction was eluted to waste, following which an additional 2 ml of the same oxalic–HCl solution was loaded on the column to elute elements with valences higher than +II (e.g.Fe, Al, Cr, and Ti) as oxalic complexes. The elution reagent was then switched to 0.5 M HCl in 95% acetone, which extracts divalent elements (e.g.Mn, Cu, Zn, Cd, Sn) and preconditions the resin for final Ni elution, in 1 ml of 0.5 M HCl in 95% + 0.1M DMG. The eluted solution is evaporated to complete dryness on a hotplate, after which five drops of concentrated HNO₃ are added to break the Ni-DMG complex. Then the sample is evaporated again. As only small volumes of DMG are used in this column any remaining organics are easily broken down by treating the sample with ten drops of concentrated HNO₃ on a hotplate overnight. After evaporation the sample is converted to the chloride form by the addition of 1 drop of 6 M HCl, and again evaporated to dryness. This method effectively separates Ni from Fe, Al, Cr, Ti and divalent elements such as Mn, Cu, Zn, Cd, Sn (Fig. 1b).

2.2.3 Nickel purification, step 3. In order to ensure complete separation of sample Ni from Fe, including any Fe that may have been added to the sample during the separation procedure, a final clean-up step was performed. Nickel fractions were dissolved in 0.2 ml 6 M HCl with approximately 0.05% H₂O₂. The anion exchange resin (0.1 ml AG1-X8, chloride form, 200–400 mesh; 0.15 ml Teflon columns) was preconditioned with the same reagent before loading the sample onto the resin bed. While Fe is retained on the resin bed at 6 M HCl, Ni immediately elutes and the loading volume was collected in addition to a further 0.2 ml of 6 M HCl + 0.05% H₂O₂ required for complete elution of Ni. After evaporation the Ni fractions were converted to the nitrate form by addition of a drop of concentrated HNO₃, allowed to evaporate to dryness, and then dissolved in 0.3 M HNO₃ for mass spectrometry analysis.

2.3 Nickel yields and procedural blank

While this separation procedure delivers a very pure Ni fraction for isotope analysis, the final yield of Ni for samples using the full separation procedure varies depending on sample matrix. A sample with a low matrix-to-Ni ratio has a near quantitative yield (95–100%), while a sample with a high matrix-to-Ni ratio has a slightly lower yield (normally 85–95%). For example, a peridotite sample containing relatively high Ni concentrations (approximately 2000 ppm) will show close to 100% yield, but a basalt or shale sample containing only small amounts of Ni (approximately 10–30 ppm) will instead produce a yield of only 85%.

To quantify the full procedural blank of the dissolution and complete chemical separation, a blank sample was processed together with every batch of geological samples. The blank ranged from 0.5 to 2.5 ng total Ni, which is negligible relative to the amounts of Ni processed for each sample (usually 5–10 μg). Both Ni yields and blank concentrations were determined by intensity measurements, relative to Ni standards of known concentrations, using MC-ICPMS.

3 Mass spectrometry

In this study we have focused on precise measurements of three stable isotope ratios: ⁶⁰Ni/⁵⁸Ni, ⁶¹Ni/⁵⁸Ni, and ⁶²Ni/⁵⁸Ni. We do not report ⁶⁴Ni/⁵⁸Ni ratios due to potential interferences from ⁶⁴Zn on this isotope. The Ni isotopic composition of all samples were analysed on a Nu Instruments Nu Plasma-HR, a multi-collector ICPMS, at the Department of Earth Sciences, Oxford. The instrument used for analyses is equipped with 12 Faraday collectors situated on a modified detector array to allow simultaneous measurement of all Ni isotopes of interest and monitoring of potential isobaric interferences. Table 2 shows both a normal and our modified detector array on the Nu Instruments Nu Plasma. Samples were usually run at a concentration of 100–200 ppb, introduced to the plasma via a microconcentric PFA nebulizer (ESI Scientific; 50μl min⁻¹) coupled to a desolvator (Nu Instruments, DSN-100) in order to enhance sensitivity and minimise oxide formation. The desolvating system was washed by aspirating with 0.3 M HNO₃ for 4 min between analyses. All isotopes were measured simultaneously in static collection mode. Each measurement consisted of a sequence of 40 cycles of 10 s integrations of the ion beam intensity. Electronic background noise was measured by ESA-deflection for 20 s prior to each analysis. Sample analyses were bracketed by analyses of NIST SRM 986 (pure Ni metal) at the same concentration, ±10%, as samples. Typical instrumental settings during analysis can be seen in Table 3.

Table 2 The normal detector array of a Nu Plasma multi collector mass spectrometer (top), and the modified detector array of the Nu plasma at the Department of Earth Sciences, Oxford (bottom). Numbers 1–11 denote the different Faraday collectors, — denotes a space in the collector array, IC denotes the place of a ion collector, and the numbers 57–62 show the placement of each mass ion beam. The comparison shows how the modified detector set-up allows us to simultaneously measure the four Ni isotope masses of interest as well as the ion beam on mass 57, which is necessary for correcting the ⁵⁸Fe-contribution on ⁵⁸Ni

0

—

1

—

2

3

4

5

6

7

8

IC

9

IC

10

IC

11

---

—	—	62	—	—	61	—	—	60	—	—	(59)	—	—	58	—	—
0	—	1	2	3	—	4	—	5	6	7	—	8	IC	9	10	11

---

62

—

—

61

—

—

60

—

—

(59)

—

—

58

—

—

57

—

---

Table 3 Typical instrument settings during Ni isotope ratio analysis on the Nu Plasma-HR

Parameters	Setting
RF power	1300 W
Acceleration voltage	5.85 kV
Sampler cone	Ni, B-type,
(Nu Instruments)	1.0 mm Ø
Skimmer cone	Ni, WA-type,
(Nu Instruments)	0.7 mm Ø
Ar gas flow rates (l min^−1):
Coolant	13
Auxiliary	0.8–1.0
Nebuliser	1.0–1.2
Uptake rate (μL min^−1)	50–75
Sample uptake time	60 s
Background measurement time	20 s
Cycle integration time	10 s
No. cycles per analysis	40
Washout time	300 s

---

3.1 Correction of natural and instrumental mass fractionation

The double-spike technique is proven to be a highly robust and accurate means of correcting for instrumental mass fractionation (e.g.refs. 13–15). The double-spike method also has the advantage that, if added to the sample before chemical separation, any isotope fractionation during chemical processing is corrected for during data reduction, which is important for a sample of difficult matrix. The method can be used for any isotope system where the element has at least four naturally occurring isotopes. We here used the isotope ratios 60/⁵⁸Ni, 61/⁵⁸Ni, and 62/⁵⁸Ni for data reduction.

3.1.1 Choice of spike composition. Optimising the isotopic composition of the double spike (and hence the isotopic compositions of the spike-sample mixtures) is critical for high-precision isotopic measurements. Using the mathematical description from ref. 16 the optimal double-spike composition was modelled, with respects to the angle θ and the error on the natural fractionation. In these calculations we chose to use ⁶¹Ni and ⁶²Ni as the double-spike components. The choice of isotopes was based on their natural abundances–as ⁶¹Ni and ⁶²Ni are the smallest isotopes of Ni (disregarding ⁶⁴Ni) they naturally also produce the smallest ion beams during mass spectrometry analysis, and will therefore have the highest errors associated with their measurement. Choosing these isotopes as the components in our double-spike will increase the proportion of these isotopes in our solution, producing beams of approximately equal size to ⁶⁰Ni and ⁵⁸Ni. This would drastically improve counting statistics and therefore produce an overall better isotope ratio measurement.

To theoretically evaluate the impact of different combinations of double-spike mixtures and sample/spike ratios on measurement precision, we used real Ni isotope analysis data (non-spiked), which was adapted for different spike additions and different spike–sample ratios. Our goal was to find a double-spike composition that resulted in high-precision measurements which were not highly sensitive to the spike–sample ratio. In our calculations we used the certified composition of the Ni isotope standard NIST SRM 986 as our true standard values, the Ni isotope composition of the enriched ⁶¹Ni and ⁶²Ni spikes— metal powder provided by Oak Ridge National Laboratory, of >99% and >96% isotopic enrichment respectively— as our original spike components, and assumed that Ni follows the exponential fractionation law during instrumental fractionation. The calculations followed the data reduction procedure described in Siebert et.al.,¹⁶ and we further assumed a value of the instrumental fractionation of 2 and that the sample is a fractionated standard with a natural fractionation of 0.2. The results gave an optimal double-spike composition as a mixture of >25 %⁶¹Ni and <75% ⁶²Ni and the optimal sample/spike ratio to be approximately 0.4 (i.e. 70% of the Ni should come from the double-spike).

3.1.2 Spike calibration. To accurately make up a double-spike solution of the composition determined above, the metal powders of the two isotopically enriched spikes were weighed in pre-cleaned and pre-weighed PEF-bottles. Afterwards each spike powder was dissolved in Teflon distilled 50% HNO₃ and diluted with the same acid to achieve solutions of 200 ppm ⁶¹Ni and 800ppm ⁶²Ni. The two spike solutions were then combined in the right proportions and diluted with 18 MΩ H₂O to make up the requested double-spike solution—25% ⁶¹Ni and 75% ⁶²Ni—with a total Ni concentration of 100 ppm and a final acid molarity of 1.5 M HNO₃.

In order to account for any impurities, both the reference standard and the double-spike solution were calibrated for all isotope ratios necessary for data reduction (60/⁵⁸Ni, 61/⁵⁸Ni, and 62/⁵⁸Ni). The calibrations of both solutions were done using an external, peltier cooled, cyclonic spray chamber from Elemental Scientific (PC³–SSI). This appliance replaces the Nu Plasma's built in spray chamber and is very useful for spike calibrations as it reduces oxide formation and enhances sensitivity without creating artificial isotope fractionation as often seen in membrane desolvators. To correct for instrumental mass bias during the calibration we used external normalisation to a Cu ICP-standard solution, assuming natural abundances for the Cu isotopes (69.17% for ⁶³Cu and 30.83% for ⁶⁵Cu). Both the reference standard and the double-spike solution had matched Cu : Ni concentrations (1 : 2) and instrumental mass fractionation was corrected assuming the exponential fractionation law (following e.g. Marechal et al.¹⁷). As calibration of the double-spike is relative to calibration of the standard used, this correction method gives accurate results, even if the absolute isotopic composition of NIST SRM 986 is not precisely known. Five analyses of both solutions were measured in the same session and the averages of these analyses assumed as the isotope ratios of each solution relative to each other. In order to test whether our given sample/spike ratio of 0.4 is suitable, solutions of NIST SRM 986 doped with the double-spike solution in different sample/spike ratios were analysed. The results of these analyses are plotted in Fig. 2–3, which shows how most measured sample/spike ratios (0.3–2.0) give accurate results, although a low ratio (0.4–0.6) is preferred. The mathematical treatment of raw data follows Siebert et al.,¹⁶ assuming Ni follows the experimental fractionation law and calculating the isotope ratios by iteration of the instrumental and natural fractionation factors. Results normally converge after three iterations.

Fig. 2 Data collected for mixtures of the Ni reference standard (NIST SRM 986) and the ⁶¹Ni-⁶²Ni double-spike solution for a range of sample/spike ratios. Measured values for Ni solutions of all sample/spike ratios are within ± 0.07‰ of the accurate δ-value.

Fig. 3 Data collected for mixtures of the Ni reference standard (NIST SRM 986) and the ⁶¹Ni-⁶²Ni double-spike solution for a range of sample/spike ratios. The analyses show that a low sample/spike ratio provides lower errors.

3.2 Isobaric interferences

When using a double-spike approach to correct for instrumental mass bias, the data reduction procedure assumes the sample to be a mass-dependently fractionated standard. This approach will therefore only work if there are no isobaric interferences present, as any interferences on the Ni isotope masses potentially could compromise the analysis, resulting in apparent mass-independent isotope fractionation.

There are several possible isobaric interferences on the Ni isotope masses that need to be addressed, of which the most critical is the overlapping masses of ⁵⁸Fe and ⁵⁸Ni, especially since ⁵⁸Ni is the denominator isotope in all ratios. Unfortunately, it is not possible to separate the mass peaks of ⁵⁸Fe and ⁵⁸Ni by using high-resolution settings on the mass spectrometer. The mass resolution (m/Δm) needed to efficiently separate these peaks is approximately 28,000, which is not possible to achieve with current analytical techniques. Instead, we correct for any background Fe in the desolvator, mass spectrometer, or analysis solution by measuring the ⁵⁷Fe ion beam and calculating the theoretical contribution of ⁵⁸Fe. For this we assumed that Fe isotopes fractionate to the same extent as Ni isotopes in the mass spectrometer and that both elements follow the exponential fractionation law. Fig. 4 shows measurements of a standard Ni solution doped with different amounts of Fe, and the analysis results both with and without the applied correction for the ⁵⁸Fe contribution on ⁵⁸Ni. The figure demonstrates that the Ni isotope compositions of the mixed solutions can be accurately corrected regardless of Fe concentration if using our correction protocol.

Fig. 4 Analyses of a Ni standard solution doped with different amounts of Fe. The black circles are the measurements without correcting for the ⁵⁸Fe interference on ⁵⁸Ni, and the white circles are the same measurements after applying the correction. As the figure shows, any Fe present in the sample can be accurately corrected for.

In addition to Fe, other potential interferences on the Ni isotope masses are double-charged ions of Cd, Sn, Te, and Xe (Table 4). Our chemical separation procedure effectively removes most of these elements from the sample, so no signals were detected when searching these mass ranges. Mass 59 was also continually monitored as the presence of a detectable 59 beam would indicate presence of ¹¹⁸Cd⁺⁺ or ¹¹⁸Sn⁺⁺. However, no beam above detection limits was observed on mass 59 for any samples or standards measured. Therefore double-charged ions of Cd and Sn were considered insignificant for our Ni isotope measurements.

Table 4 Possible elemental and molecular isobaric interferences on the Ni isotope masses of interest

	^58Ni	^60Ni	^61Ni	^62Ni
Elemental:	^58Fe^+
Double-	^116Cd	^120Te ^++	^122Te ^++	^124Te ^++
charged:	^116Sn ^++	^120Sn ^++	^122Sn ^++	^124Sn ^++
				^124Xe ^++
Argides:	^22Ne^36Ar^+	^24Mg^36Ar^+	^25Mg^36Ar^+	^24Mg^38Ar^+
	^22Ne^38Ar^+	^20Ne^40Ar^+	^21Ne^40Ar^+	^26Mg^36Ar^+
		^22Ne^38Ar^+		^22Ne^40Ar^+
Oxides:	^40Ar^18O^+	^42Ca^18O^+	^43Ca^18O^+	^44Ca^18O^+
	^40Ca^18O^+	^44Ca^16O^+		^46Ti^16O^+
	^42Ca^16O^+
Hydroxides:	^40Ar^17OH^+

---

The most important polyatomic interference for Ni isotope measurements is ⁴⁰Ar¹⁸O⁺, potentially present on the high mass end of ⁵⁸Ni. Previous studies on Ni isotopes have used a higher resolution (m/Δm >2,500) setting to resolve this interference from ⁵⁸Ni. During careful investigation of possible interferences on mass 58 at a pseudo-resolution of m/Δm ≃ 4,000, we did not find any detectable peaks that could correspond to ⁴⁰Ar¹⁸O⁺ (or ⁴⁰Ar¹⁷OH⁺). Because of this all our measurements were made in the low resolution mode. However, when performing an analysis while only aspirating 0.3 M HNO₃ a measurable background in the range of 0.5–5mV on all Ni masses does show. This small signal was interpreted as the instrumental blank originating from the cones, as ICPMS sampler and skimmer cones typically are made of Ni. Tests using custom made in-house aluminium skimmer and sampler cones showed that Ni cones account for ∼50% of this background Ni, with the remaining background presumably originating from other stainless steel components of the mass spectrometer. As also reported by Quitte and Oberli⁷Al cones give lower sensitivity and poorer reproducibility which is why Ni cones were preferred here, although they do contribute towards the Ni background. The background is, however, constant and does not seem to affect the measurements as no bias was detected between analyses involving an on-peak-zero background measurement in blank acid compared to analyses where the background measurement was carried out by ESA deflection.

4 Results and discussion

4.1 Isotope fractionation associated with the analytical procedure

In order to test our chemical separation technique aliquots of spiked Ni solution (NIST SRM 986) were processed and compared to the same non-processed spiked standard reference solution. Analyses of the isotopic composition of the chemically processed standards are shown in Fig. 5. Isotopic ratios of all processed reference solutions were identical to the untreated standard within errors, demonstrating that the chemical separation does not result in any additional instrumental matrix effects, or that any such fractionation effects are corrected for by our double-spike method.

Fig. 5 Measurements of the Ni standard NIST SRM 986 to test chemical separation procedure. Open symbols are a mixture of SRM 986 and the artificial calibration solution (plasma standards, Alfa Aesar, see text for more information) put through the separation procedure. Closed symbols are the SRM 986 on its own put through the separation procedure. Error bars are 2 s.e. of individual measurement. Dotted lines ±2 s.d. of all SRM 986 measurements.

In order to more thoroughly investigate the results of our chemical separation procedure we performed matrix tests, employing mixtures of ICP-standard solutions (Alfa Aesar) of geologically abundant matrix elements—Mg, Al, Ca, Fe, Ti, V, Cr, Mn, Co, Cu, Zn, Mo, Cd, Sn, and Pb—mixed together with the spiked Ni NIST standard solution to yield element abundances in similar proportions to silicate rocks. After processing the mixed solution through our column chemistry procedure the solutions were analysed, and yielded isotopic compositions identical with the untreated NIST standard reference solution within error (Fig. 5).

4.2 Isotope fractionation associated with matrix effects

The presence of other elements may depress the ionization and throughput of Ni in the plasma and interface region of the mass spectrometer. Although our chemical protocol effectively separates Ni from matrix elements, we carried out tests to evaluate the effects of residual matrix elements in sample solutions on isotopic ratio measurements. Nickel standard solutions were doped with a chosen set of individual elements (Ca, Cr, Fe, Mg, Na, Ti, and Zn) at different element-to-Ni concentration ratios. These elements were chosen as they are either of similar mass range (Fe, Zn), difficult to completely separate from Ni by ion-exchange chromatography (Ti, Cr), or ubiquitous in geological samples (Na, Ca, Mg). The results of these tests are shown in Fig. 6 where all measurements are plotted against their concentration ratio. At low concentrations, the effect the matrix elements have on the measurement is corrected for by the double-spike technique. But careful chemical separation of Ni from matrix elements is still required for high-precision isotopic ratio determinations.

Fig. 6 Analyses of a Ni standard solution doped with different elements in varying concentration ratios to Ni. The error bars are 2 s.e. of the individual measurement. The dotted lines show the 2 s.d. error on all (non-doped) Ni standard solution analyses during the measurement session.

Some remaining matrix elements may also combine to form argide or oxide molecules in the plasma and cause interferences on the masses of interest. For example ²⁴Mg³⁶Ar⁺ interferes on ⁶⁰Ni, and ⁴⁴Ca¹⁶O⁺ and ⁴⁶Ti¹⁶O⁺ on ⁶²Ni (see Table 4). Therefore it is particularly important to completely separate Ni from Mg, Ti, and Ca. However, as can be seen in Fig. 6 doping a Ni solution with Mg does not seem to have any effect on either the accuracy or the precision of the measurement. On the other hand, the addition of Ca or Ti significantly offsets the measurement accuracy. This shows that argides are not formed in the plasma to the same extent as oxides. Adding Ca to the solution pushes the measurement results towards heavier delta values, while Ti in the solution has the opposite effect, producing lighter delta values than the non-doped Ni standard, even at relatively low concentrations (Fig. 6). This is because the most abundant Ca isotope (⁴⁰Ca, natural abundance 96.9%) can form ⁴⁰Ca¹⁸O⁺ which interferes with ⁵⁸Ni, denominator in all isotope ratios measured, pushing the ratios higher. For Ti on the other hand only ⁴⁶Ti can form an oxide possible of interfering on any of the Ni masses analysed, ⁴⁶Ti¹⁶O⁺. This oxide would interfere with ⁶²Ni, an isotope used in the double-spike correction, and thus interfere with the double-spike data reduction procedure.

4.3 Reproducibility and accuracy of measurements

In order to determine the long-term reproducibility of Ni isotope ratio measurements we have taken two different approaches. Instrument reproducibility was tested by measurements of a pure NiO powder (from Sigma Aldrich, 98% Ni), which had been dissolved and spiked, but not treated with our the chemical separation procedure. The NiO solution was analysed during several measurement sessions over one year, giving an overall average value of δ⁶⁰Ni = −0.159 ± 0.034‰ (2 s.d., n = 310) (Fig. 7). Typically, 10 pure Ni standards were analysed per sample measurement session.

Fig. 7 Long term instrument reproducibility as tested by measurements of a Ni-oxide solution (NiO powder, Sigma Aldrich) analysed during several measurement sessions over one year. The overall average value for the NiO solution was δ⁶⁰Ni = −0.159 ± 0.034‰ (2 s.d.). Error bars in the figure are 2 s.e. of the individual measurements.

For actual samples however, the long-term reproducibility is expected to be larger than the one given by the instrument as the chemical separation also may have an effect on the analytical reproducibility. This was tested by measurements of three USGS reference materials of different silicates, processed and analysed during several measurement sessions over one year. The results of these analyses are shown in Fig. 8, where the top figure shows replicate analyses of each silicate sample, and the bottom figure the mean value and 2 s.d. error of each sample respectively. By these analyses, the long-term external δ⁶⁰Ni sample reproducibility for silicate samples is typically ± 0.07‰ (2 s.d.).

Fig. 8 Long term external sample reproducibility as tested by measurements of three USGS reference materials of different chemical compositions (one peridotite, one basalt, and one shale standard). The silicates were fully reprocessed and analysed during several measurement sessions over one year. The figure shows replicate analyses of each silicate sample, error bars are 2 s.e. of the individual measurements. By these analyses the external δ⁶⁰Ni sample reproducibility was determined to typically be ± 0.07‰ (2 s.d.) for silicate samples.

Small differences were noted between samples of different analyte-to-matrix ratios, as samples with higher ratios (e.g.ferromanganese nodules and sulphides) show lower values for the long term reproducibility (as low as ± 0.05‰ for some samples). Because of this, matrix matching with available rock standards is advisable to get best results for unknown samples. Finally, Fig. 9 shows a comparison for two USGS basalts (BHVO-2 and BCR-2) with data published by Cameron et al.⁹ As can be seen our data for these reference materials are in good agreement with already published values.

Fig. 9 Comparison of data between this study and Cameron et al.⁹ showing that our data for these USGS reference materials compare very well with already published values.

5 Summary

The new method presented in this paper allows for near quantitative extraction of Ni from a variety of geological sample matrices for high precision isotopic measurements. Our novel chemical separation procedure produces high yield results, our double-spike set-up provides a robust instrumental fractionation correction as well as correcting for possible isotope fractionation during the chemical separation, and our MC-ICP-MS analyses demonstrate the absence of interferences or matrix effects. Repeat analysis of USGS rocks standards and an inter-laboratory comparison confirms the excellent precision and reproducibility achieved by our analytical technique. This method will allow us to further investigate the Ni isotope variations for a variety of terrestrial samples, and possible fractionation between reservoirs.

6Acknowledgements

The authors would like to thank the Oxford Isotope Group, Nick Belshaw, and Derek Preston for invaluable technical assistance and Marcel Regelous for sharing of manuscript draft and method details. We would also like to thank the two anonymous reviewers for their helpful and insightful comments. This study was funded by Petrobras, the UK Science and Technology Facilities Council (STFC), and the Natural Environments Research Council (NERC). Oxford Isotope laboratories were funded by Petrobras, STFC and ERC grants to Prof. A.N. Halliday

References

1. J. Gramlish, L. Machlan, I. Barnes and P. Paulsen, J. Res. of the Natl. Inst. of Stand. and Tech., 1989, 94, 347–356.
2. C. Johnson, B. Beard and F. Albarède, Geochemistry of non-traditional stable isotopes, Reviews in Mineralogy & Geochemistry, Mineralogical Society of America, 2004, vol. 55.
3. G. Quitte, M. Meier, C. Latkoczy, A. Halliday and D. Günther, Earth Planet. Sci. Lett., 2006, 242, 16–25.
4. D. Cook, M. Wadhwa, P. Janney, N. Dauphas, R. Clayton and A. Davis, Anal. Chem., 2006, 78, 8477–8484.
5. J. Chen, D. Papanastassiou and G. Wasserburg, Geochim. Cosmochim. Acta, 2009, 73, 1461–1471.
6. S. Ragsdale, J. Inorg. Biochem., 2007, 101, 1657–1666.
7. G. Quitte and F. Oberli, J. Anal. At. Spectrom., 2006, 21, 1249–1255.
8. M. Regelous, T. Elliott and C. Coath, Earth Planet. Sci. Lett., 2008, 272, 330–338.
9. V. Cameron, D. Vance, C. Archer and C. House, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 10944–10948.
10. A. Victor, Anal. Chim. Acta, 1986, 183, 155–161.
11. F. Strelow, Anal. Chim. Acta, 1990, 233, 129–134.
12. F. Strelow, C. Weinert and C. Eloff, Anal. Chem., 1972, 44, 2352.
13. S. Galer, Chem. Geol., 1999, 157, 255–274.
14. C. Johnson and B. Beard, Int. J. Mass Spectrom., 1999, 193, 87–99.
15. M. Dodson, Geochim. Cosmochim. Acta, 1970, 34, 1241–1244.
16. C. Siebert, T. Nägler and J. Kramers, Geochem., Geophys., Geosyst., 2001, 2.
17. C. Maréchal, P. Télouk and F. Albarède, Chem. Geol., 1999, 156, 251–273.

---