Determination of selenium isotopic ratios by continuous-hydride-generation dynamic-reaction-cell inductively coupled plasma-mass spectrometry

Abstract

The goal of this study was to develop an accessible method for the determination of selenium isotopes within selenium-rich geological samples and examine the influence of sample introduction, instrumental parameters, column separation and the application of standard-sample bracketing for mass bias and drift correction. Quantitative selenium separation and enrichment of samples was achieved by a column separation using 0.2 g thioglycollic impregnated cotton fiber prior to introduction of the sample to an ICP-DRC-MS. 0.6 ml min⁻¹ premixed Ar (95%) + H₂ (5%) was favored over CH₄ and NH₄ as a reaction cell gas and was used within the DRC coupled with optimized DRC rejection parameters RPa (0) and RPq (0.65) to effectively reduce the signal to background ratio of all measured selenium isotopes (m/z 76, 77, 78, 80 and 82). Ion signal intensity of all measured selenium isotopes were increased 100 fold over classic nebulization by mixing of 1% NaBH₄ and acidified sample digestions in a membrane-less computer-controlled continuous hydride generator. Transient hydride ion signals were time-averaged for five readings and three replicates to produce an in-run precision (2σ) of ±0.45‰ δ^82/76Se_Merck (relative to a Merck titrosol ICP-MS standard) and ±0.85‰ δ^82/76Se_Merck over an 18 month period. In the absence of a selenium isotopic standard, the accuracy of the method was determined using four interlaboratory solutions and five geological standard reference materials covering 0 to −4.5‰ δ^82/76Se_Merck. Our results indicate excellent reproducibility within method precision. The minimum mass of Se required for isotopic ratio determination was 3 μg (>100 000 cps at m/z 82 and 76).

---

Introduction

Traditional ICP-MS instruments are prone to spectroscopic interferences caused by isobaric interferences that preclude the routine analysis of selenium isotopes (i.e., m/z 76 and 80). With the advance of dynamic-reaction and gas-collisional cells coupled with conventional ICP-MS, spectroscopic interferences near m/z 80 and 76 can nearly be eliminated.^1–3 The low ionization efficiency of selenium within an argon ICP (∼30%) coupled with the reduced signal intensities resulting from the use of a DRC, have led to the application of hydride generation in this study to improve the signal to background ratio (SBR).

Selenium concentrations within most crustal reservoirs are low, generally less than 0.1 μg g⁻¹,⁴ and require isotopic analysis via double spike negative thermal ionization mass spectrometry (N-TIMS)^5–15 or standard bracketed hydride-generation multicollector inductively coupled mass spectrometry (HG-MC-ICP-MS)^16,17 in light of their superior detection limits, precision and accuracy. Traditional ICP-MS has been applied to isotopic research involving mass-dependent fractionation of light isotopes, such as B where larger relative standard deviation (RSD) of ion intensity result in large but acceptable isotope ratio error (e.g., B; ∼7‰ δ^11/10B_NASS−1)¹⁸ given the large potential mass-fractionation range (e.g., B; >40‰ δ^11/10B_NASS−1).¹⁸ Isotopic ratio analysis of heavier ‘non-traditional’ isotopes by ICP-MS (e.g., Fe, Cu and Zn) have a lower RSD of ion intensity at high masses by traditional ICP-MS and require more precise measurement than possible by traditional ICP-MS due to the relatively small mass-dependent fractionation range (i.e., ∼3‰ δ^57/54Fe_IRMM−014).¹⁹ However, recent data indicate a wide range of mass-dependent fractionation within the selenium isotopes^4,17,11 (+3 to −8‰ δ^82/76Se_Merck) and imply earth reservoirs as negative as −15‰ δ^82/76Se_Merck.¹⁶ Furthermore, geological materials that have the greatest potential for isotopic mass-dependent fractionation, such as hydrothermal ore deposits, coal and organic-rich shale, are also Se-rich^20–22 (>1 μg g⁻¹) and are amenable to analysis by more traditional ICP-MS techniques. It was therefore the aim of this study to develop and test a more accessible method (>300 Perkin Elmer DRC instruments in North America) for the routine determination of selenium isotope ratios within Se-rich geological samples.

Experimental

Instrumentation

Measurements were preformed on an ELAN 6100 dynamic reaction cell inductively coupled mass spectrometer (Perkin Elmer Life and Analytical Sciences, Inc., USA), equipped with an automated continuous hydride generator (CGH) system developed specifically for this study. Samples, standards and wash solutions were introduced to the instrument via a Perkin Elmer AS-91 autosampler. The experimental conditions for both the Elan 6100-DRC and the CHG system are given in Table 1.

Table 1 Instrumental operating conditions and signal measurement parameters

a Pure Ar equivalent
ICP parameters
RF-Power	1300 W
Auxiliary gas flow	1.3 ml min^−1
Plasma gas flow	13 ml min^−1
Dwell time	10 ms
Sweeps/reading	1
Quadrupole offset	–3
Autolens voltage	6.5 V
Interface cones	Platinum
Instrument resolution	0.5% at 10% peak
DRC parameters
Reaction gas	Ar/H2(95–5%) mixture
DRC gas flow	0.6 ml min^−1^a
RPa	0
RPq	0.65
CGH parameters
Carrier gas flow (nebulizer gas flow)	0.8 ml min^−1
Reagent	NaBH4
Reagent flow rate	0.89 ml min^−1
Sample acidity	5.3 wt% HCl; 5.8 wt% HNO3
Sample flow rate	0.44 ml min^−1
Wash time	300 s
Signal duration	650 s (300 replicates)

---

The CHG system was designed and built at the University of Texas at Dallas and is shown in Fig. 1. The gas–liquid separator was assembled using an inverted 500 ml fluorinated ethylenepropylene (FEP) Nalgene™ bottle with a modified perfluoroalkoxy copolymer (PFA) cap. A peristaltic pump (Gilson Minipuls3 Peristaltic, Middleton, WI) was used to deliver the acidified sample and reducing agent to the CHG chamber. Hydride gas was swept using an argon carrier gas into the ICP torch via 4 mm internal diameter FEP tubing. A modified Scott double-pass spray chamber was used between the CHG and ICP torch to dampen pulsation (induced by using a peristaltic pump) of the hydride gas flow to the ICP. Stability of the hydride gas was greatly improved by simultaneously combining sample and reagent within the CHG chamber via a mixing manifold rather than using a traditional mixing coil (cf., Rouxel et al.¹⁷). Sample-reagent mixing minimized the pulsation of hydride gas into the ICP torch. Hydride stability was improved by removing waste liquid as a batch at the end of the sample analysis and minimized analyte hydride gas loss via continuous peristaltic pumping, similar to Wolnik et al.²³ Memory effects and hydride stability were also improved by not incorporating a microporous polytetrafluoroethylene (PTFE) membrane at the CHG-ICP interface as in previous studies.^{16,17,24–27}

Fig. 1 Schematic diagram of continuous hydride generator (CHG).

Standard solutions and reagents

Ar (99.999% purity, Air Liquide, TX) was used as a plasma gas and Ar/H₂ mixture (99.999% purity, 5% H₂ 95% Ar, Air Liquide, TX), CH₄ (99.999% purity, Scott Speciality Gases, TX), and NH₄ (99.999% purity, Scott Speciality Gases, TX) were tested in the DRC as potential reaction gases.

During digestion of standard reference material (SRM) and preparation of solutions only high purity reagents were used. 18.2 MΩ cm⁻¹ deionized water (Millipore™, Bedford, MA) and high purity quartz/PFA distilled 12.1 M HCl, 15.8 M HNO₃, 28.9 M HF, 11.6 M HClO₄, 17.4 M HC₂H₃O₂, 18 M H₂SO₄ (Baseline grade, Seastar Chemicals Inc, Sidney, BC) were used in this study. PFA beakers, centrifuge tubes, and bottles were cleaned by soaking in 2% (v/v) citronox solution (ALCONOX Inc., USA), using a hot (60 °C) ultrasonic bath for 90 min. Labware were then placed in successive baths of 6 M HNO₃ and 2 M HNO₃ for 24 h, with thorough rinses using 18.2 MΩ cm⁻¹ water between each step. All sample and reagent preparation was conducted in a Class 100 total exhaust fume hood and SRM powder digestions were carried out in a 99.97% HEPA filtered digestion enclosure to minimize contamination.

In generating the hydride gas, a 1% sodium borohydride was used and prepared daily dissolving 3 g of powdered NaBH₄ (EMD Chemicals Inc., Gibbstown, NJ) in 250 ml of 0.05% NaOH solution to stabilize the reagent. The NaBH₄ solution was then filtered through a microfibre filter (Whatman 934-AH) to remove any undissolved material and transferred to a clean PFA bottle and refrigerated at (4 °C) during use to minimize the decomposition of NaBH₄.

In the absence of an international isotopic reference standard for selenium, an ICP-MS standard solution obtained from EMD Chemicals Inc. (Merck Titrosol SeO₂ in HNO₃) was adopted as a running lab standard as in other studies.^15,17 Aliquots of reference materials and lab standards used by Rouxel et al.¹⁷ were also used in this study to quantify external precision.

Procedure

Sample preparation

Pulverized samples and standard reference materials (SRM) were digested in PFA digestion vessels (Savillex, Minnetonka, MN) using 28.9 M HF (10 ml), 15.8 M HNO₃ (10 ml), and 11.6 M HClO₄ (1 ml). Digestion vessels were sealed and placed in an oven at 80 °C for 76 h and taken to incipient dryness at 70 °C on a hotplate. The temperature of the hotplate was monitored via a Pt-thermocouple mounted within an empty PFA digestion vessel. Se degassing¹⁷ and loss by evaporation²⁸ was avoided by using a HCl-free decomposition and maintaining the temperature <80 °C. Residue obtained after decomposition was instantaneously sealed after addition of 6 M HCl (10 ml) and placed in an agitating boiling bath for 60 min to quantitatively reduce all selenate(VI) to selenite(IV).²⁹

Thioglycollic cotton fiber (TCF) column separation was used to simplify the sample matrix and quantitatively concentrate selenium. TCF was prepared in a similar manner to that described by others.^17,30,31 Reduced samples were allowed to cool and were filtered through 25 mm 0.45 μm hydrophobic polytetrafluoroetheylene (PTFE) syringe filters (Aerodisc, Gelman Science, Northampton, UK) that were initially primed using isopropyl alcohol (Omnisolv grade, EMD Chemicals Inc., Gibbstown, NJ). Filtered samples were then diluted to 1 M HCl before being loaded onto 10 ml polypropylene columns (BIORAD, Hercules, CA) previously filled with 0.2 g of TCF. Prior to sample loading, TCF was washed with 18.2 MΩ cm⁻¹ water (5 ml) and conditioned with 6 M HCl (1 ml) followed by 1 M HCl (1 ml)^17,30 to bring the TCF to the appropriate pH (<3) for optimum selenium adsorption. Filtered sample solutions were passed through the TCF columns a total of three times at a flow rate of 1.0 ml min⁻¹. Final column eluent was then collected and analyzed via classical nebulization ICP-MS to ensure complete selenium TCF adsorption. Sample-loaded TCF was then eluted with 6 M HCl (2 ml) to remove any residual higher oxidation state species such as Sb.

Se-impregnated TCF was quantitatively removed from the columns and placed in sealed 15 ml PFA centrifuge tubes. 500 μl of 15.8 M HNO₃ and 500 μl 18 MΩ cm ⁻¹ water were added to the TCF before boiling for 20 min. During this oxidating process, complete desorption of selenium was achieved (see discussion below). Centrifuge tubes were removed and cooled in an ice bath for 10 min before the addition of 3.3 ml of 18.2 MΩ cm⁻¹ water. Samples were centrifuged for 10 min at 3500 rpm. Residual liquid was pipetted into 15 ml polypropylene centrifuge tubes with the addition of 500 μl of 12.1 M HCl.

Continuous-gas-hydride analysis

Reagent and acidified sample were conveyed with a peristaltic pump at 0.89 and 0.44 ml min⁻¹ respectively, into the HG chamber and were simultaneously mixed via a reaction manifold. Carrier Ar (0.8 ml min⁻¹) was introduced over the reaction manifold where reaction hydride and Ar were mixed to fill the HG chamber. Upon mixing of the acidified sample with reagent in the CHG, large quantities of H₂ enter the plasma. Preconcentration of the sample, as described above, maintains constant acid strength. Additionally, this serves to avoid mass bias problems that result from changes in sample viscosity and variable matrix conditions.

Ten isotopes were monitored in DRC-mode by using the shortest dwell time per AMU (10 ms), resulting in ‘near-simultaneous’ measurement. 300 replicates were collected over a 12 min analysis period (Fig. 2). After analysis, the computer-controlled waste valve was opened and spent reagent, sample and hydride–argon gas was vented into a waste container. Subsequently, the HG chamber and mixing manifold was washed with 10% HNO₃ before a second venting of waste.

Fig. 2 CGH response of 10 replicate analyses of 800 ppb Merck standard solution. (A) Raw ion signal response of 10 replicate analyses. Ion signal can be divided into three distinct signal responses, including: gas build-up, stabilization and stable. (B) Time-averaged integration for 350 to 650 s for each replicate over a typical unknown sample run. Error bars indicate 2σ.

The ion signal response of the CGH can be divided into three distinct time-dependent intervals (Fig. 2), namely: hydride build-up, stabilization delay and, stable response. As such, the ion signal for a given sample via CGH is in many respects analogous to the transient signal of LA-ICP-MS analysis.^32–34 An initial interval of hydride build-up is required to fill peristaltic tubing and fill the HG with the reacted gases. Following this, a period was required for the ICP to stabilize and for hydride production to become constant. This was followed by an interval of constant and stable ion response in which a time-averaged ion signal was attained for calculation of relative isotopic measurements. Waste drainage at the end of each analysis (rather than continuous waste pumping), resulted in a less pulsed and more stable ion signal. However, continued reagent reaction with sample acid results in production of excess hydrogen^23,26,30,35 and suppression of the ion signal with time, which is most notable at the highest intensity m/z 80. Precise computer-controlled timing (±500 ms) of reagent and acidified sample mixing within the CHG allowed accurate replication of samples (Fig. 2b).

For a given sample run (two blanks, five replicate samples, eight Merck standards), a time-stacked integration of the stable ion signal response (flat response between 350 s and 640 s; (Fig. 2a) was used to select the timing of ion signal use.

The isotopic ratios are calculated by standard bracketing time averaged ion response, that involves the measurement of a sample between standards of known composition.^36,37 This process/calculation corrects for mass discrimination and instrument drift, assuming sample and standard are closely matrix-matched.³⁸ For samples reported in this study, matrix simplification and Se concentration on TCF ensures matrix-matching of samples and standards. The Se isotopic nomenclature in this paper is reported in ‘delta notation’³⁹ where the isotopic composition is calculated as deviation relative to the Merck standard by:

where R_std is the measured mean ⁸²Se/⁷⁶Se ratio for the bracketing Merck solution and R_spl is the measured ⁸²Se/⁷⁶Se ratio of the unknown sample. Unless otherwise stated, each sample run within this study represent a mean of five readings and a minimum mean of three sample run replicates are reported.

Eight procedural blanks measured in this study resulted in low signal intensities (in the hundreds of counts per second m/z 80). These are considered negligible and therefore no procedural blank correction was considered necessary for calculation of selenium isotopic ratios. Instrumental background correction was performed by subtraction of a matrix-matched ultra-low level acid blank signal at m/z of interest that allowed the correction for any fluctuation in the background signal present at m/z 60 (to monitor potential NiO isobaric interferences at m/z 74 and 76), 72, 74, 75, 76, 77, 78, 79, 80, 81, 82, and 83. All isotope ratio data in this study are calculated from background corrected values. Calculations were performed off line in Excel™ spreadsheets using macros for the integration of transient hydride signals. All the uncertainties reported in this paper are expressed as standard deviation (2σ).

Holding capacity and yield rates of TCF

TCF has been shown to have a strong adsorption affinity for selenite(IV) with quantitative isotopic recoveries following digestion in concentrated nitric acid.^16,17 The theoretical holding capacity of TCF has been reported by other workers to be in the order of 7–9 mg g⁻¹, with a maximum operational value of 1/3 employed in routine trace element analysis.^30,31 Previous workers have only semi-quantitatively documented the required TCF mass and acid strength/volume to provide 100% Se yield. Given the lack of quantification in previous studies and the fact that the TCF was generated in-house, we quantitatively tested both optimal holding capacities and acid strengths to ensure complete Se adsorption and removal and to prevent Se isotope fractionation during sample processing. (Fig. 3).

Fig. 3 Thioglycollic cotton fiber yield as a function of acid strength and selenium solution concentration. (A) Below 0.1 g of TCF selenium is not quantitatively adsorbed. (B) At elevated concentrations (>3000 ppm Se) and at recommended nitric desorption volumes and strengths (0.1 ml; ref. 17) selenium is not quantitatively removed.

Results confirm that there is a critical balance between the TCF mass used in the column separation, the enrichment of selenium and the volume and strength of nitric acid used to desorb selenium from the TCF. If TCF mass is insufficient, selenium fails to completely adsorb, remaining in the residual liquid passed through the column. If TCF mass is excessive, selenium remains adsorbed onto the TCF during the digestion of the cotton and if nitric acid concentration is too weak, selenium remains adsorbed at high TCF masses. However, at higher nitric volumes and concentrations, selenium is fully de-adsorbed, but stronger acid strength suppresses the ion signal in the ICP. Using 0.2 g TCF and 0.5 ml of 15.8 M HNO₃ during TCF digestion, selenium recovery is ∼98 to 101% with minimal signal suppression (∼14%) (see discussion below). By using this TCF mass and concentration of nitric acid volume in the digest, maximum TCF holding capacity and recovery is 6.7 mg g ⁻¹, which is lower than previously reported^30,31 (Fig. 3).

Influence of HCl and HNO₃ on the selenium ion signal

Changing TCF acid digestion volume from 0.1 ml (15.8 M HNO₃) to 0.5 ml results in a change in the final running solution from 1.1% to 5.8 wt% HNO₃. Previous workers^23,40–43 have observed that variation in the acid concentration of the running solution and calibration standards can radically alter the intensity of the ion signal (also known as a matrix effect) by classic nebulization ICP-MS. However, it was unclear how variation in acid concentration would affect analysis by CHG-ICP-DRC-MS. Therefore, an experiment was devised to investigate ion signal suppression of the Merck Se standard with increasing HNO₃ and HCl concentrations (Fig. 4). We found that variation in HNO₃ concentration from 0.4 to 11 wt% resulted in signal suppression from 3% to 26%, and signal suppression of samples of 14.5%, which given the intensity of the Se signal (∼1 × 10⁶ cps) was deemed acceptable. Similar signal suppression for HCl at 5.3 wt% (1.7 mol l⁻¹) was observed, where signal suppression is approximately 12%. It is also noteworthy that varying the HCl concentration had a minimal effect on the signal suppression (up to ∼10 wt%) when compared to the signal suppression for increasing HNO₃ concentration. These trends are consistent with those of other elements analyzed by standard nebulization ICP-MS.^40,41

Fig. 4 Dependence of the selenium signal at m/z 80 (1000 (μg Se) l⁻¹) normalized to 0.3 and 0.4 wt% on the concentrations of HCl and HNO₃, respectively.

Influence of reaction gas and DRC settings on the selenium signal

Considerable instrumental development and research has been devoted to the reduction of polyatomic interferences via collisional processes and ion–molecule reactions in ICP-MS.^44–48 Despite the numerous potential interferences, the Ar²⁺ interference at m/z 76, 78, and 80 are the most important for this study. The dominant interferences of interest that reside near m/z 80 are summarized in Table 2.

Table 2 Dominant interferences on selenium isotopes^a

Isotope	Natural concentration (at.%)	Potenial Interference
a Compiled from ref. 2 and 50.
74	0.89	^37Cl2^+,^40Ar^34S^+, ^74Ge, ^148Sm^2+, ^148Nd^2+
76	9.36	^36Ar^40Ar^+, ^38Ar^38Ar^+, ^40Ar^36S^+, ^31P2^14N^+, ^152Sm^2+, ^152Gd^2+
77	7.63	^40Ar^37Cl^+, ^40Ar^36Ar^1H^+, ^154Sm^2+,^154Gd^2+
78	23.78	^38Ar^40Ar^+, ^31P2^16O^+, ^156Gd^2+
80	49.61	^40Ar2^+, ^1H^79Br^+, ^80Kr^+, ^160Dy^2+, ^160Gd^2+
82	8.73	^12C^35Cl2^+, ^34S^16O3^+, ^82Kr^+, ^40Ar2^1H2^+, ^1H^81Br^+, ^165Ho^2+, ^164Dy^2+,^160Er^2+

---

It is generally agreed that in the ICP-DRC-MS the most important and abundant type of reactions that occur when using a reaction gas are those involving charge transfer:^47–49

where I⁺ is the charged interference and R is the reaction gas.

It was found that a premixed Ar (95%) and H₂ (5%) mixture was an effective reaction gas for the reduction of Ar dimer at m/z 76 (Ar⁴⁰ + Ar³⁶) and 80 (Ar⁴⁰ + Ar⁴⁰). Efficiency of the reduction of the Ar dimer to background levels was improved using Ar/H₂ over CH₄ and NH₄, and a flow rate of 0.5 ml min⁻¹† Ar/H₂versus 0.75 ml min⁻¹.† (Fig. 5). However, at elevated gas flow rate (>1 ml min⁻¹) for Ar/H₂, produced a greater decrease in the intensity of the ion signal when compared to CH₄.

Fig. 5 Ion signal response to different reaction gases and varied gas-flow rates. (A) Signal response at m/z 80 for CGH-ICP-DRC-MS analysis of 800 ng ml⁻¹ Se standard (Merck). (B) Response of total ion count over varied Ar/H₂ gas flow rates. Solid lines indicate non-acceptable isotopic ratios, whereas dashed line indicate acceptable isotopic ratios, within theoretical instrumental bias near m/z 80.

Because of the high atomic percent and volume of hydrogen in CH₄ (25 at.% @ 99.9999% v/v) versus the Ar (95% v/v) H₂ (5% v/v) (2.5 at.% @ 5% v/v), SeH⁺ is created at a much higher efficiency within the DRC using CH₄ as the reaction gas. Because charge transfer is a molecular reaction, reducing the gas flow rate of CH₄ reduces the effective DRC reactions and thus increases BSR, but has little effect on the SeH⁺ production. Analyzing a 1000 ng ml⁻¹ Merck Se solution by DRC using CH₄ reaction gas results in ∼500 000 cps (m/z 80) and 70 000 CPS (m/z 81) and gives a 14% SeH⁺ yield, whereas using CH₄ as a reaction gas results in ∼2000 cps (m/z 81) resulting in 0.4% SeH⁺ generation using an Ar/H₂ mixture.

Residual ⁸⁰Se ion signal after running samples using CH₄ as a reaction gas was observed and required protracted wash times to lower ⁸⁰Se to background values. This residual ⁸⁰Se ion signal was not observed using Ar/H₂ or NH₃ as a DRC gas. Since other isotope intensities for Se (i.e., ⁷⁶Se and ⁸²Se) did not demonstrate this pattern of elevated SBR with CH₄, memory-effects do not appear to be the cause, but rather imply unresolved secondary ions as suggested by Hattendorf and Gunther.⁴⁸

The DRC is essentially a quadrupole ion guide and as such, the efficiency of charge transfer reactions can be optimized for the discrimination against unwanted reaction products by the bandpass and gas flow settings within the DRC.^46,47,49 The bandpass is adjusted by the rejection parameters RPa and RPq to achieve optimum transmission for the analyte ions in question.

The rejection parameter q (RPq) of the DRC quadrupole in this study was optimized for the Merck running standard prior to each analytical run, in order to achieve a close match of the measured selenium mass spectrum with Merck TIMS selenium isotope values¹⁵ (Fig. 6). At low RPq ∼0.3 the measured m/z intensities did not match the Merck selenium isotope values and only when the RPq value was increased to 0.6–0.7 did the measured signal intensities more closely match the Merck selenium isotope abundance. By using RPq ∼0.65 (variation of ±0.05 was recorded between runs), a favorable mass cut-off in the DRC is achieved and a reduction in the isobaric interferences followed the gas phase chemical reactions with the DRC gas Ar/H₂. However, the sensitivity decreased and thus increased the limit of detection. It was found that on adjusting the rejection parameter (RPa) by small increments (i.e., 0.01) the ion signal quickly approached 0, such that for this study RPa was kept at zero.

Fig. 6 The effect of the RPq DRC settings on the selenium m/z 82, 80, 78, and 76 for 1000 ng ml⁻¹ Merck solution.

Potential m/z interferences and correction

Isobaric interferences

Krypton. The measurement of Se isotope ratios by an argon ICP source and DRC must be corrected for small concentrations of Kr within the plasma and DRC gases. The measured ⁷⁸Se, ⁸⁰Se and ⁸²Se ion signal has direct isobaric interferences caused by ⁷⁸Kr, ⁸⁰Kr and ⁸²Kr. Blank subtraction of low measured signal intensities, a result of poor ionization of Kr in the ICP (<20%), reduced the m/z 78, 80 and 82 to background levels during the analysis of samples. However, during the analysis of unknown samples, the ion signal at m/z 83 was monitored for krypton.

Germanium. The measurement of ⁷⁶Se is interfered with the presence of ⁷⁶Ge within geological and standard reference materials. In most rock types, the concentration of Ge is 10⁻² to 10⁻³ times that of selenium⁵¹ and produces minimal interference. However, as demonstrated by Rouxel et al.,¹⁷ some SRM and geological samples that contain very low Se have Ge/Se as high as 273 (e.g., SRM IF-G: iron formation). As such, an experiment was conducted to quantify the affects of varying Ge/Se concentrations on the final TCF digest solution. Results of these experiments suggest that varied Ge/Se ratios have no consequence on the effective removal of Ge via column elution. These results are consistent with SRM’s analyzed where the Ge content is well established. After the loading of the sample on the column and elution of the TCF by 6 M HNO₃, and analyzing TCF eluent, we found that essentially >99.95% of germanium was effectively removed from the sample.

Despite the effective removal of Ge by TCF separation, m/z 72 was monitored during analysis and anomalous ⁷⁶Ge was corrected by using a ⁷²Ge/⁷⁶Ge natural isotopic value corrected for instrumental bias of 0.2763.

Polyatomic interferences

Hydrogen interferences within the DRC

Bromide. Under normal sample introduction by classic nebulization and using CH₄ as a reaction gas, ⁷⁹Br and ⁸¹Br ion signal response is excellent for solutions >100 pg ml⁻¹, with BrH⁺ generation in the DRC at background levels. Because only hydride forming elements are transported into the CGH-ICP-DRC-MS and because Br does form hydrogen interferences, we needed to consider whether ⁷⁹Br¹H and ⁸¹Br¹H would cause an interference with ⁸⁰Se and ⁸²Se, respectively. As such, an experiment was conducted to quantify the efficiency of BrH⁺ formation. Varying concentrations of Br standard (EM Science) were analyzed using both classic nebulization and CGH-ICP-DRC-MS. By CGH, background levels (<10² CPS) of ⁷⁹Br, ⁸¹Br, and ⁷⁹Br¹H, and ⁸¹Br¹H were detected at m/z 79, 80, 81, and 82, regardless of the concentration of Br standard analyzed (10 to 1000 ng ml⁻¹). Within the hydride generator, Br does not rapidly form BrH and is not swept into the ICP by the Ar carrier gas. Therefore, during isotopic analyses, neither Br nor BrH caused any interference.

Arsenic. Unlike Br, As is commonly elevated in geological materials enriched in Se.^52–56 Normally As, being mono-isotopic, causes little potential polyatomic interference in routine ICP-MS analysis. However, the use of DRC reaction gases with hydrogen and a sample-hydride introduction, introduces the likelihood of ⁷⁵As¹H formation and subsequent interference with ⁷⁶Se. Results obtained by Rouxel et al. using ArH₂ within a reaction cell imply AsH⁺ production within the collisional cell was significant and difficult to correct.¹⁷ As such, an experiment was conducted to quantify these effects when using a DRC over a collisional cell. Acid blank analysis indicates m/z 76 is free of interferences and permitted the analysis of varying concentrations of As standard (0.1 to 1000 ng kg⁻¹: EM Science) processed through TCF column separation by DRC-neb ICP-MS and using ArH₂ as a reaction gas. These results indicate that As has a weak affinity for TCF (1.7% to 2.3%) and that the m/z 76 remains at background signal intensities. Analysis of varied concentrations of As standard solutions (1 to 1000 ng kg⁻¹) by CGH indicate efficient AsH⁺ generation in the HG, which results in 1 × 10⁵ cps (1000 ng kg⁻¹) at m/z 75. However, even at these elevated intensities on m/z 75, no signal intensities above background at m/z 76 were observed. Furthermore, analysis of Merck in-house Se standard doped with varied concentrations of As, produced no discernible changes on ⁷⁶Se/⁸²Se. Therefore during isotopic analyses, AsH⁺ did not cause interference and no correction was made to unknown samples.

Internal precision

At the beginning, middle and end of each sample run the ‘in-house’ Merck standard was bracketed against itself to quantify within-run precision, but also to quantify internal reproducibility over an 18 month period. Within-run reproducibility of the Merck solution was ±0.45 per mille (n = 3; 95% confidence level) for the δ^82/76Se_Merck ratio. However, the greater scatter in the measured m/z 80 results in a larger uncertainty in the delta δ^80/76Se_Merck of ±0.83 per mille (n = 3; 95% confidence level). Because the precision of this technique will degrade with smaller signal strengths, as a direct product of counting statistics, the minimum mass of Se used was 3 μg (>100 000 cps at m/z 82 and 76 for 5% RSD), which is significantly higher than HG-MC-ICP-MS (>5 ng¹⁷) and N-TIMS (>500 ng¹²), but is two-orders of magnitude lower than traditional gas source mass spectroscopy (∼100 μg⁵⁷).

Analysis of standard solutions

External accuracy

To quantify the accuracy of the CHG-ICP-DRC-MS, four Se solutions were obtained from O. Rouxel (used in Johnson et al.¹⁴ and Rouxel et al.)¹⁷ and have been measured relative to the ‘in-house’ Merck standard (Table 3). These solutions have been analyzed by N-TIMS and MC-ICP-MS and cover a moderate range of δ^82/76 values (0.0 to −4.5 per mille). Results indicate, within analytical error, that Se isotope analysis by CHG-ICP-DRC-MS reproduces isotopic ratios measured by other techniques and that the ‘in-house’ Merck standard is isotopically identical to those used in other studies.^15–17,58

Table 3 Se isotopic composition of Standard Reference Materials (non-isotopic standards), provisional Se Standard Reference Materials (non-isotopic standards), Interlaboratory standards and geological material relative to Merck. N/A Not available

Sample	Description	Se/μg g^−1	Sample mass/mg	δ ^82/76SeMerck	δ ^80/76SeMerck	δ 82/78Merck
a Data from Rouxel et al.^17 b Data from Rouxel et al.^17 but calculated from δ^80/76 relative to MH495. c This study. d Calculated from linear mass fractionation law (e.g. Criss^59). e Calculated from five replicate digestions. f Provisional value (CCRMP).
SRM
SGR-1	Green River shale	3.5		−1.08^a		−0.68^a
			250	−1.0^c (n = 3)	−0.5^c (n = 3)	−0.4^c (n = 3)
GXR-4	Copper mill feed	5.6		−1.29 ^a		−0.88^a
			250	−1.4^c (n = 3)	−1.0^c (n = 3)	−1.0^c (n = 3)
MAG-1	Marine mud	1.16		−1.42^a		−0.76^a
			250	−1.2^ce (n = 2)	−0.9^c (n = 3)	−0.7^c (n = 3)
SCo-1	Shale	0.89		−1.57^a	N/A	−0.92^a
			250	−1.8^ce (n = 2)	−1.4^c (n = 2)	−0.9^c (n = 2)
SRM (provisional)
Canyon Diablo Triolite	Meteorite	10		−1.32^a	N/A	N/A
				−1.10^b	N/A	N/A
			250	−1.1^c (n = 2)	−0.6^c (n = 2)	−0.6^c (n = 2)
WMG-1	Mineralized gabbro	15 ± 3^f	250	−1.0^c (n = 3)	−0.5^c (n = 3)	−0.4^c (n = 3)
WMS-1	Massive magmatic sulfide	108 ± 13^f	250	−1.0^c (n = 3)	−0.6^c (n = 3)	−0.4^c (n = 3)
WPR-1	Altered peridotite	4 ± 1^f	250	−0.9^c (n = 3)	−0.5^c (n = 3)	−0.4^c (n = 3)
Interlaboratory standards
Merck–Rouxel	Nancy Merck split	1		0^a	N/A	0^a
				−0.2^c (n = 3)	−0.1^c (n = 3)	−0.1^c (n = 3)
CRPG	Nancy split	1		−3.15^a	N/A	−2.13^a
				−3.1^c (n = 3)	1.9^c (n = 3)	−2.0 (n = 3)
MH495	Illinois split	1		−4.46^a	N/A	−3.06^a
				−4.3^c (n = 3)	−2.6^c (n = 3)	−2.3^c (n = 3)
NIST 3149 split	NIST bulk Se standard	1		−1.4^a	N/A	−0.93^ad
				−1.5^c (n = 3)	−0.9^c (n = 3)	−0.8^c (n = 3)

---

Analysis of standard reference materials

In addition to interlab solutions, four geologically relevant SRM’s and four additional provisional SRM’s were analyzed to quantify the accuracy of the method as a whole (Se preconcentration and CGH-ICP-DRC-MS). The isotopic results of the SRM’s processed through the TCF columns, compared with those reported by others, indicate that, within analytical error, there was no mass fractionation within the Se preconcentration procedure, which is consistent with near 100% selenium yield and recovery.

Conclusions

The performance of CGH-ICP-DRC-MS for Se isotope ratio measurement has been evaluated and is shown to provide acceptable precision and accuracy between 0 and −4.5‰ δ^82/76Se_Merck. This allows the analysis of geological samples and the discrimination of samples that have undergone moderate to extreme mass-dependent fractionation (perhaps down to −15‰ δ^82/76Se_Merck). The accuracy and precision of the method using four interlaboratory solutions and the ‘in-house’ Merck standard analyzed over an 18 month period indicate a variation of ±0.85‰ δ^82/76Se_Merck (2σ).

Ion signal intensities of all measured selenium isotopes were increased 100-fold over classic nebulization. Premixed Ar (95%) + H₂ (5%) at a flow rate of 0.6 ml min⁻¹ was used in preference over CH₄ and NH₄ as a reaction cell gas to minimize the Ar₂⁺ interferences at m/z 76, 78 and 80, the formation of secondary ions at m/z 80, and the formation of hydrides within the DRC. Optimized DRC rejection parameters RPa (0) and RPq (0.65) were required to effectively reduce the SBR and to reduce the instrumental mass bias within range of the m/z 80 (∼3% per AMU) for the PE 6100 DRC instrument.

The throughput of this method is about 6–8 samples per day, similar to 4–5 by N-TIMS. Standard-bracketing of samples corrects for mass discrimination as well as instrument drift; however, selenium separation by TCF is always necessary. Once a laboratory has been established for TCF separation, the time required for sample preparation is minimal.

Future publication will include the application of this technique to selenium-enriched rocks, including volcanogenic massive sulfide ores and organic-rich black shale rocks.

Acknowledgements

We thank O. Rouxel and T. M. Johnson for providing aliquots of their Se solutions and for their interest in this project. We thank Jerry Beavers for his machining and aid in the CHG assembly. This work benefited from discussions with P. Pelchat (Geological Survey of Canada), O. Rouxel (Woods Hole Oceanographic Institute) and T. Johnson (University of Illinois at Urbana-Champaign). We are grateful for journal reviewer’s comments that greatly helped to improve the manuscript. Research support was gratefully received from Natural Sciences and Engineering Research Council, National Science Foundation, Geological Survey of Canada, Society of Economic Geologists and Mineralogical Association of Canada. This is the Geosciences Department at the University of Texas at Dallas contribution number 1073, and Geological Survey of Canada contribution number 2004309.

References

1. E. H. Larsen, J. Sloth, M. Hansen and S. Moesgaard, J. Anal. At. Spectrom., 2003, 18, 310.
2. J. J. Sloth, E. H. Larsen, S. H. Bugel and S. Moesgaard, J. Anal. At. Spectrom., 2003, 18, 317.
3. J. J. Sloth and E. H. Larsen, J. Anal. At. Spectrom., 2000, 15, 669.
4. T. M. Johnson and T. D. Bullen, in Geochemistry of Non-Traditional Stable Isotopes, ed. C. M. Johnson, B. L. Beard and F. Albarede, Mineralogical Society of America, Washington, DC, 2004.
5. P. Van Dael, D. Barclay, K. Longet, S. Metairon and L. B. Fay, J. Chromatogr. B, 1998, 715, 341.
6. T. M. Johnson, Chem. Geol., 2004, 204, 201.
7. T. M. Johnson and T. D. Bullen, Geochim. Cosmochim. Acta, 2003, 67, 413.
8. M. J. Herbel, T. M. Johnson, K. K. Tanji, S. Gao and T. D. Bullen, J. Environ. Qual., 2002, 31, 1146.
9. M. J. Herbel, T. M. Johnson, K. K. Tanji, S. D. Gao and T. D. Bullen, J. Environ. Qual., 2002, 31, 1146.
10. P. Van Dael, L. Davidsson, E. E. Ziegler, L. B. Fay and D. Barclay, Pediatr. Res., 2002, 51, 71.
11. M. J. Herbel, T. M. Johnson, R. S. Oremland and T. D. Bullen, Geochim. Cosmochim. Acta, 2000, 64, 3701.
12. T. M. Johnson, M. J. Herbel, T. D. Bullen and P. T. Zawislanski, Geochim. Cosmochim. Acta, 1999, 63, 2775.
13. T. M. Johnson, M. J. Herbel, T. D. Bullen and R. S. Oremland, Geol. Soc. Am., Abstr. Prog., 1999, 31, 276.
14. T. M. Johnson, M. J. Herbel, T. D. Bullen and P. T. Zawislanski, Geochim. Cosmochim. Acta, 1999, 63, 2775.
15. M. Wachsmann and K. G. Heumann, Int. J. Mass Spectrom. Ion Processes, 1992, 114, 209.
16. O. Rouxel, Y. Fouquet and J. N. Ludden, Geochim. Cosmochim. Acta, 2004, 68, 2295.
17. O. Rouxel, J. Ludden, J. Carignan, L. Marin and Y. Fouquet, Geochim. Cosmochim. Acta, 2002, 66, 3191.
18. D. C. Gregoire, Anal. Chem., 1987, 59, 2479.
19. B. L. Beard and C. M. Johnson, Rev. Mineral. Geochem., 2004, 55, 319.
20. D. Mercone, J. Thomas, I. W. Croudace and S. R. Troelstra, Geochim. Cosmochim. Acta, 1999, 63, 1481.
21. G. M. Peters, W. A. Maher, D. Jolley, B. I. Carroll, V. G. Gomes, A. V. Jenkinson and G. D. McOrist, Org. Geochem., 1999, 30, 1287.
22. A. H. Clemens, L. F. Damiano, D. Gong and T. W. Matheson, Fuel, 1999, 78, 1379.
23. K. A. Wolnik, F. L. Fricke, M. H. Hahn and J. A. Caruso, Anal. Chem., 1981, 53, 1030.
24. R. M. Barnes and X. Wang, J. Anal. At. Spectrom., 1988, 3, 1083.
25. S. Branch, W. T. Corns, L. Ebdon, S. Hill and P. O’Neill, J. Anal. At. Spectrom., 1991, 6, 155.
26. E. J. McCurdy, J. D. Lange and P. M. Haygarth, Sci. Total Environ., 1991, 135, 131.
27. J. T. Creed, M. L. Magnuson, C. A. Brockhoff, I. Chamberlain and M. Sivaganesan, J. Anal. At. Spectrom., 1996, 7, 505.
28. Y. K. Chau and J. P. Riley, Anal. Chim. Acta, 1965, 33, 36.
29. S. P. Brimmer, W. R. Fawcett and K. A. Kulhavy, Anal. Chem., 1987, 59, 1470.
30. M.-Q. Yu, G.-Q. Liu and J. Q., Talanta, 1983, 30, 265.
31. S. Xiao-Quan and H. Kai-Jing, Talanta, 1985, 32, 23.
32. W. T. Perkins, R. Fuge and N. J. G. Pearce, J. Anal. At. Spectrom., 1991, 7, 445.
33. B. Hattendorf and D. Gunther, J. Anal. At. Spectrom., 2000, 15, 1125.
34. B. Hattendorf, C. Latkoczy and D. Gunther, Anal. Chem., 2003, 75, 341A.
35. R. F. Sanzolone and T. T. Chao, Geostand. Newsl., 1987, 11, 81.
36. N. S. Belshaw, X. K. Zhu, Y. Guo and R. K. O‘Nions, Int. J. Mass Spectrom. Ion Processes, 2000, 197, 191.
37. X. K. Zhu, R. K. O’Nions, Y. Guo, N. S. Belshaw and D. Rickard, Chem. Geol., 2000, 163, 139.
38. F. Albarede and B. Beard, in Geochemistry of Non-Traditional Stable Isotopes, ed. C. M. Johnson, B. L. Beard and F. Albarede, Mineralogical Society of America, Washington, DC, 2004.
39. J. R. O’Neil, Rev. Mineral., 1986, 16, 561.
40. E. G. Chudinov and G. V. Varvanina, J. Anal. Chem., 1989, 44, 662.
41. L. Jian, W. Goessler and K. J. Irgolic, Fresenius’ J. Anal. Chem., 2000, 366, 48.
42. II. Stewart and J. W. Olesik, J. Anal. At. Spectrom., 1998, 13, 1313.
43. P. Smichowski and J. Marrero, Anal. Chim. Acta, 1998, 376, 283.
44. J. T. Rowan and R. S. Houk, Appl. Spectrosc., 1989, 43, 976.
45. G. C. Eiden, C. J. Barinaga and D. W. Koppenaal, J. Anal. At. Spectrom., 1996, 11, 317.
46. P. Turner, J. Merren, C. Speakmen and C. Haines, in Plasma Source Mass Spectrometry, ed. G. Holland and S. D. Tanner, Royal Society of Chemistry, Cambridge, 1997.
47. S. D. Tanner, V. I. Baranov and D. R. Bandura, Spectrochim. Acta, Part B, 2002, 57, 1361.
48. B. Hattendorf and D. Gunther, J. Anal. At. Spectrom., 2004, 19, 600.
49. D. R. Bandura, V. I. Baranov and S. D. Tanner, Fresenius’ J. Anal. Chem., 2001, 370, 454.
50. B. Gammelgaard and O. Jons, J. Anal. At. Spectrom., 1999, 14, 867.
51. S. Gao, T. Luo, B. Zhang, H. Zhang, Y. Han, Z. Zhao and Y. Hu, Geochim. Cosmochim. Acta, 1998, 62, 1959.
52. D. Layton-Matthews, M. I. Leybourne, J. Peter and S. D. Scott, Geol. Soc. Am. Abstr. Progr., 2003, 35, 357.
53. G. E. M. Hall, A. I. MacLaurin, J. C. Pelchat and G. Gauthier, Chem. Geol., 1997, 137, 79.
54. G. E. M. Hall and J. C. Pelchat, J. Anal. At. Spectrom., 1997, 12, 97.
55. G. E. M. Hall and J. C. Pelchat, J. Anal. At. Spectrom., 1997, 12, 103.
56. J. A. Plant, D. G. Kinniburgh, P. L. Smedley, F. M. Fordyce and B. A. Klinck, in Treatise on Geochemistry, ed. H. D. Holland and K. K. Turekian, Elsevier, Amsterdam, 2003, vol. 9.
57. H. R. Krouse and H. G. Thode, Can. J. Chem., 1962, 40, 367.
58. O. Y. F. Rouxel and J. Ludden, unpublished work presented at the EUG XI meeting, Strasbourg, 8th–12th April, 2001.
59. R. E. Criss, Principles of Stable Isotope Distribution, Oxford University Press, 1999.

---

Footnote

† Ar-equivalent flow units: mass-flow controller calibrated for pure Ar.

---