sign in sign up
<<
Home , Isotopes of osmium

NRC Publications Archive Archives des publications du CNRC

Determination of the isotopic composition of using MC-ICPMS Zhu, Zuhao; Meija, Juris; Tong, Shuoyun; Zheng, Airong; Zhou, Lian; Yang, Lu

This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur. For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien DOI ci-dessous.

Publisher’s version / Version de l'éditeur: https://doi.org/10.1021/acs.analchem.8b01859 Analytical Chemistry, 90, 15, pp. 9281-9288, 2018-06-21

NRC Publications Archive Record / Notice des Archives des publications du CNRC : https://nrc-publications.canada.ca/eng/view/object/?id=45709299-174b-4f06-91d3-411ee9765217 https://publications-cnrc.canada.ca/fra/voir/objet/?id=45709299-174b-4f06-91d3-411ee9765217

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at https://nrc-publications.canada.ca/eng/copyright READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE.

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site https://publications-cnrc.canada.ca/fra/droits LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

Questions? Contact the NRC Publications Archive team at [email protected]. If you wish to email the authors directly, please see the first page of the publication for their contact information.

Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à [email protected]. Article

Cite This: Anal. Chem. 2018, 90, 9281−9288 pubs.acs.org/ac

Determination of the Isotopic Composition of Osmium Using MC- ICPMS Zuhao Zhu,†,‡ Juris Meija,‡ Shuoyun Tong,∥ Airong Zheng,§ Lian Zhou,∥ and Lu Yang*,‡ †Fourth Institute of Oceanography, State Oceanic Administration (SOA), Beihai 536000, China ‡National Research Council Canada, 1200 Montreal Road, Ottawa, Ontario K1A 0R6, Canada §College of Ocean and Earth Sciences, Xiamen University, Xiamen 361102, China ∥State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences, Wuhan 430074, China

*S Supporting Information

ABSTRACT: Despite its widespread applications in geology, all osmium ratio measurements are either uncalibrated or rely on the veracity of the uncalibrated 1937 Nier values by adopting them as normalizing constants typically in conjunction with an exponential mass bias correction model. In this study, isotope ratios of osmium were determined in six commercial osmium materials, including the DROsS standard and a new NRC isotopic osmium reference material OSIS-1, by MC-ICPMS. We use a previously optimized and validated regression mass bias correction model to correct instrumental isotope fractionation effects which does not rely either on Nier’s values or on a strictly mass-dependent behavior of the . Deviations from mass-dependent fractionation (mass independent fractionation) were observed for osmium isotopes in MC-ICPMS with the most dramatic effect occurring for 187Os, wherein, on average, close to half-percent bias in the isotope ratio 187Os/188Os was observed as a result of imposing Russell’s law.

1. INTRODUCTION Union of Pure and Applied Chemistry is that of Völkening et 3 ’ In 1970, Riley and Delong noted that “very few comparisons of al. measured by N-TIMS which relies on Nier s values to the natural isotopic composition of osmium have ever been correct for instrumental fractionation of isotope ratios. In fact, ffi most measurements of osmium isotope ratios adopt fixed made because of analytical di culties and limited availability of ’ osmium rich samples”.1 Since then, significant improvements canonical values of nonradiogenic isotope ratios, such as Nier s value 192Os/188Os = 3.083,28,29 for mass bias correction of the in negative thermal ionization mass spectrometry (N-TIMS) ’ methods2,3 have led to numerous applications of osmium radiogenic isotope ratio. It is remarkable that Nier s measure- isotopes in geological sciences. These include the study of low- ments have not been superseded after eight decades of Downloaded via NATL RESEARCH COUNCIL CANADA on April 11, 2019 at 16:54:32 (UTC). progress in mass spectrometry. See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. and high-temperature processes such as mantle−crust interaction, sedimentation, cosmic evolution, ocean circulation, TIMS has been the traditional instrument of choice to and even industrial waste migration.4−20 The obtain high-accuracy osmium isotope ratio measurements, fi despite the investment of extensive sample preparation and of -187 to osmium-187 to signi cant variations 26,30 of this radiogenic osmium isotope. As an example, isotope ratio long measurement times. In the last several decades, 187Os/188Os varies by several orders of magnitude from 1:10 in multicollector inductively coupled plasma mass spectrometry ’ 7 21 (MC-ICPMS) has become a powerful tool for the determi- the Earth s mantle to 1:1 in seawater and 10:1 in some 26,31 sulfide ores.22 Owing to these variations, isotope ratios nation of isotope ratios in a variety of applications, largely involving osmium-187 are of key interest.23 due to its advantages in sample introduction and high ffi 32 Despite these advances, however, the measurement ionization e ciency. It has been noted that MC-ICPMS is technique is still not mature enough to allow for new better suited to measure osmium isotope ratios compared to 33 independent measurements of isotope ratios of osmium, and TIMS owing to the fact that ICP measures elemental ions 24−27 + − in turn, isotopic abundances or atomic weight. To this (Os ) directly whereas TIMS detects oxide ions (OsO3 ) day, most if not all osmium isotope ratio measurements are traceable (directly or indirectly) to the values published by Received: April 25, 2018 Nier in 1937.28,29 As an example, the best osmium isotope Accepted: June 21, 2018 ratio measurement currently recognized by the International Published: June 21, 2018

© 2018 American Chemical Society 9281 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288 Analytical Chemistry Article which requires further signal deconvolution. Despite these approximately 5.6 min. Longer measurement time (17 min) advantages, however, MC-ICPMS suffers much larger was tested in our preliminary experiments, and it provided instrumental mass bias compared to that of TIMS.31 Although similar precision for osmium isotope ratios. Thus, the short the majority of the bias (if it is not caused by an interference measurement time was adopted for all subsequent work. effect) can be modeled using mass-dependent laws, mass- 2.2. Reagents and Solutions. Reagent grade nitric and independent fractionation (MIF) has been reported in MC- hydrochloric acids (Fisher Scientific, Ottawa ON, Canada) ICPMS for many elements including Nd,34−37 Ce,35 W,38 were purified before use via a sub-boiling distillation system Sr,35,39 Ge,40 Pb,40 Hg,40 Si,41 and Ba.42 Consequently, proper (Milestone Inc., Shelton CT, USA). A NanoPure mixed-bed mass bias correction is crucial when MC-ICPMS is used for the ion exchange system with reverse osmosis domestic feedwater determination of absolute isotope ratios. (Barnstead/Thermolyne Corp, Iowa USA) was used to A primary method to calibrate isotope ratio measurement prepare high-purity (18 MΩ cm) deionized water. Environ- results utilizes gravimetric mixtures of near-pure isotopes with mental grade aqueous ammonia (20−22% volume fraction) known chemical purity43 (not to be confused with the double was obtained from Anachemia Science (Montreal, Canada). spike calibration). For an element with N stable isotopes, this Aqueous ammonia (5%) and hydrochloric acid (1%) primary approach is based on measuring all (N − 1) isotope solutions were used to reduce the osmium memory effect ratios in all N isotopically enriched materials along with after each measurement session. A severe osmium memory isotopic composition measurements of (N − 1) independent effect was encountered with liquid introduction to MC-ICPMS gravimetric mixtures of any two enriched materials. This model during preliminary experiments. Commonly used rinsing has been successfully employed for many two-isotope systems solutions such as 3−5 M HCl26 and EtOH were tested (Li and C)44,45 and three-isotope systems (Si)46,47 and without much success. As suggested by Nowell et al.,26 it is recently adopted for more complex elements such as beneficial to keep osmium in a reduced or complexed state for 48 . While this approach provides the highest a better washout. Thus, Na2SO3,NH2OH·HCl, and NH3·H2O metrological quality of isotope ratio measurements, it requires were tested. It was found that alternate use of 5% ammonia weighable quantities of all enriched isotopes in a chemically solution and 1% HCl (blank solution) was sufficient to wash pure form. As the number of isotopes increases, so does the out osmium after a session of measurement to a blank level cost of enriched isotopes, efforts to characterize their chemical within 30 min, likely due to the ability of ammonia to complex purities, as well as the complexity of mathematical calculations. osmium ions. We also rinsed the spray chamber daily with 5% Hence, cost-effective alternative approaches are often sought. ammonia, 5% HCl, and water to reduce the memory effects. The aim of this study is to provide osmium isotopic Six osmium materials were purchased from different composition measurements which are fully independent from commercial vendors as summarized in Table 1. Two osmium any previous osmium isotope ratio measurements. The commonly used double spike calibration31 cannot help in Table 1. Osmium Standards Used in This Work this regard as it relies on a material (the double spike) whose isotope ratios have to be calibrated against a known Os isotope Symbol Compound Source reference. To obtain calibrated osmium isotope ratios, the Os-1 Os standard solution from (NH4)2OsCl6 SCP 49−52 regression-based model is employed. This method allows OSIS-1 Os standard solution from (NH4)2OsCl6 SCP calibrating isotope ratios of an element using a known isotope Os-2 Os standard solution from (NH4)2OsCl6 Merck · ratio of another element without assuming that the two Os-3 OsCl3 3H2O Alfa Aesar elements must necessarily display identical mass bias. This Os-4 (NH4)2OsCl6 Sigma-Aldrich approach has been successfully applied to many elements such DROsS Os standard solution from (NH4)2OsCl6 IAGEO Ltd. as ,49 ,52 ,53−55 ,56 ,57 57 57 58 51,59 60 , , , molybdenum, , standard solutions (1000 mg kg−1) were sourced from SCP ,61,62 and .63 In this work, we have employed the ’ ́ 51,52 Science (Baie D Urfe QC, Canada) and one from Merck improved regression mass bias correction model using (Darmstadt, Germany)all originally prepared from IRIS-1 iridium isotopic standard as calibrator.52 (NH4)2OsCl6 salt. High-purity osmium chloride, OsCl3· 3H2O (99.99% purity, Alfa Aesar, Tewksbury, MA, USA) 2. EXPERIMENTAL SECTION was dissolved in 10% HCl to obtain the osmium standard 2.1. Instrumentation. A Thermo Fisher Scientific solution (Os-3) with mass fraction of osmium w(Os) = 350 Neptune Plus (Bremen, Germany) MC-ICPMS equipped mg kg−1.Afifth osmium standard solution (Os-4), w(Os) = with nine Faraday cups and a combined Scott-type on the top 350 mg kg−1, was prepared by dissolution of high-purity of a cyclonic spray chamber with a PFA self-aspirating ammonium chloroosmate, (NH4)2OsCl6 (99.99% purity, nebulizer (Elemental Scientific, Omaha NE, USA), operating Sigma-Aldrich, Oakville ON, Canada) in 10% HCl. The at 50 μL min−1, was used for all osmium isotope ratio Durham Romil Osmium reference material (DROsS) was measurements. A guard electrode was fitted to the sourced from IAGEO Ltd. (Keyworth, Nottingham UK). plug-in quartz torch with quartz injector. The instrument was The primary isotope ratio calibrator employed in this work operated under low mass resolution mode to perform isotope was Certified Reference Material of iridium (IRIS-1, 1000 mg ratio measurements. Briefly, the instrument was tuned for kg−1) from National Research Council Canada (NRC, Ottawa, fi 193 maximum sensitivity and optimal peak shape as well as stable Canada) with certi ed isotope ratio value R193/191 = N( Ir)/ −1 191 52 signals using a 1.0 mg kg osmium solution. The gain N( Ir) = 1.6866(5)k=1. calibration of the Faraday cups was then performed to ensure 2.3. Sample Preparation and Analysis. Stock solutions normalization of their efficiencies. Typical operating conditions of osmium and iridium were diluted with 1% HCl to prepare are summarized in Table S1. Under these experimental test solutions with mass fractions w(Os) = 1.3−1.8 mg kg−1 conditions, acquisition of each measurement point takes and w(Ir) = 1.0−1.8 mg kg−1. A self-aspiration mode was used

9282 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288 Analytical Chemistry Article

Figure 1. Typical linear regression plots of osmium isotopes in OSIS-1 against iridium (IRIS-1) as the calibrator. for sample introduction. Two Faraday cup configurations were ng kg−1, while platinum was present at levels below 40 ng kg−1. employed to collect all isotopes of osmium and iridium. Although the concentration of platinum was somewhat higher, Isotopes 184Os, 186Os, 187Os, 188Os, 189Os, 190Os, 191Ir, and 193Ir interferences contributed from platinum and other elements were measured using the main cup configuration, and isotopes were not significant since the mass fractions of osmium and 187Os, 188Os, 189Os, 190Os, 191Os, 192Os, and 193Ir were iridium in the analyzed samples were several orders of measured using the subcup configuration (see Table S1). magnitude higher than the mass fraction of platinum. A 1.8 −1 190 Similar to the settings for the determination of iridium isotopic mg kg Os standard produced 29.5 and 47.2 V for Os and 192 composition in our previous study,52 the instrument plasma Os, respectively, whereas the 1% HCl blank produced 2.0 radio frequency (RF) power was increased from the optimum mV and 3.1 mV for 190Os and 192Os, respectively. This was fi −1 value P0 (which corresponds to the highest sensitivity and further con rmed by the fact that a 5000 μgkg platinum stable signal, typically at 1240 W) to Pmax wherein the osmium solution produced blank level intensities of 2.0 mV and 3.1 mV isotope signal decreased by approximately 25%, compared to for 190Os and 192Os, respectively, the same as in the 1% HCl its value at P0. The isotopic composition of all samples was blank solution. determined five times at incrementally increasing the RF power 3. RESULTS AND DISCUSSION with values of (Pmax − P0)·N/4 where N = 0, 1, 2, 3, and 4. This results in five sets of osmium (and iridium) isotope ratios 3.1. Correction for Instrumental Fractionation (Mass and takes approximately 35 min. The measurement duration at Bias). Calibrated isotope ratio measurements are typically each RF power was kept identical. Signals of all monitored performed with the use of gravimetric mixtures of near-pure isotopes in every point were subtracted by their corresponding isotopes.43 This primary measurement method (not to be intensity in 1% HCl blank solution at optimum RF power, P0. confused with double spike calibration) requires all separated All data sets reported in this study were collected between near-pure stable isotopes of an element in weighable quantities August of 2017 and January of 2018. which is not feasible for osmium. In this work we have Since the abundance of osmium-184 is 3 orders of employed the cost-effective regression mass bias correction magnitude lower than the most abundant isotopes of osmium, model which presents itself as an alternative secondary method the signal of 184Os+ in a 1.8 mg kg−1 Os standard solution was for isotope ratio measurements and relies on the availability of lower than 20 mV. Thus, a 1013 Ω resistor64−66,67 was used for primary isotope reference material of another element. The the Faraday cup L4 which measures the osmium-184. The regression mass bias correction model is based on the observed quality of each set of data was monitored via the determination correlated temporal drift between several isotope ratios of the 31,50,52 coefficient (R2) of the regressions, which was R2 > 0.9995 for same or distinct elements, as shown in Figure 1 for all isotopes except osmium-184. Due to the low abundance of osmium and iridium isotope ratios: 184 2 Os, the R values for the regressions ln(r184/188) against rabrOs =+ Ir 2 lnli/188 iin193/191 (1) ln(r193/191) were R > 0.99. fi ffi 2.4. Spectral Interferences. Impurities of speci c where coe cients ai and bi are the intercept and slope of the elements in the samples could create spectral interferences corresponding linear regression which are obtained using the (as shown in Table S2) for the measurements of osmium and least-squares fitting of data. Here the true isotope ratio is iridium isotopes. Quantitative analysis of six different source directly linked to its measured ratio by a correction factor K, −1 Os Os Os lr lr lr osmium solutions (1.8 mg kg ) spiked with iridium (1.3 mg Ri/188 = Ki/188·ri/188 and R193/191 = K193/191·r193/191 (note that kg−1) revealed that the mass fractions of , , KOs does not need to be equal to KIr). The regression model Os , , , and rhenium were all less than 2 provides values for all Ki/188 separately without the need to

9283 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288 Analytical Chemistry Article impose any mass-dependent relationship between these values We note that osmium isotope ratio uncertainties reported in (such as the Russell law). Equation 1 can be used to obtain the the literature are significantly smaller than we report here corrected isotope ratios without assuming identical mass bias because they are invariably based on the assumed fixed values for different isotope pairs:50,52 (e.g., Nier’s value (1937) of 3.083 for 192Os/188Os) without uncertainty for the corresponding normalizing ratios. In this ba R Os =·ReIr ii i/188 ()193/191 (2) work, we have not constrained the measurement results to Nier’s values. Instead, our results are ultimately traceable to Although MC-ICPMS is equipped with nine Farday cups NIST isotopic reference materials of rhenium and and which would, in principle, allow for simultaneous measurement our results fully incorporate the corresponding uncertainties of of all seven osmium isotopes and both iridium isotopes, spatial these standards. restrictions of the Faraday cups made it necessary to split the 3.3. Isotopic Abundances and Atomic Weight of analysis in two lines. Osmium isotope ratios 184/188, 186/ Osmium. The isotopic abundances and atomic weight of 69 188, 187/188, 189/188, and 190/188 were established from osmium were calculated from the isotope ratios. The the main line. The intensity of iridium-193 was imputed from uncertainty of isotopic abundances and the atomic weight of the main line and linearly scaled to the iridium-191 intensity osmium were propagated from the corresponding isotope ratio from the subline measurements (see cup configuration in uncertainties, while taking into account the covariances. Table S1). Such imputation was made posible due to large Uncertainty propagation of the isotope ratios into the corresponding isotopic abundances and atomic weight was (half percent) gradual changes in the measured isotope ratios 70 at different ICPMS RF power. Osmium isotope ratios 187/ done using the R package CIAAWconsensus. The atomic masses of osmium isotopes used for calculations in this report 188, 189/188, 190/188, and 192/188 were then established 71 and the results from the two measurements averaged. are from the 2016 Evaluation. It is important to note that this refined regression calibration 3.4. Measurement Results for OSIS-1 Standard. model is not derived from either the exponential or the Osmium isotope ratio measurements in OSIS-1 standard ’ were performed with replicate solutions containing 1.2−1.8 mg Russell s isotope ratio fractionation law as it is commonly −1 −1 perceived and orginally presented.49 Rather, it relies on the kg of osmium and 1.0−1.8 mg kg of iridium during a three- ffi month from August of 2017 to January of 2018. During invariability of the ratio of the fractionation coe cients and is ff capable of correcting both mass-dependent and mass- this period, six di erent sets of ICP cones were used and a total − independent isotope ratio fractionation occurring in MC- of 156 sets of osmium iridium regressions were obtained. The ICPMS.40,50,52 Given that each osmium isotope ratio is results of osmium isotope ratios (R) and isotopic abundances calibrated separately as shown in Figure 1, this refined (x) are shown in Table 2 and the corresponding atomic weight of osmium in the OSIS-1 standard is A (Os) = 190.2407(7) . regression model does not invoke the assumption that the r k=1 mass bias between the various isotope ratios of an element Table 2. Osmium Isotope Ratios and Abundances in the varies smoothly as a function of nuclide mass differencean a assumption that is at the core of double spike calibration which OSIS-1 Standard is another frequently employed secondary isotope ratio Atomic mass Isotope ratio, Isotope ratio, Isotopic abundance, i R R x i calibration method. The regression model allows for different number, i/187 i/188 ( Os) mass bias between osmium and iridium. In addition, both the 184 0.011 20(6) 0.001 30(1) 0.000 172(1) analyte and the calibrator are measured simultaneously in the 186 1.032(2) 0.119 73(5) 0.015 902(14) same solution, thus eliminating any effect of the sample matrix. 187 1 (exact) 0.1159(2) 0.015 397(26) Matching the absolute signal levels of the analyte and the 188 8.626(13) 1 (exact) 0.132 817(70) calibrator has an insignificant effect on the results, as 189 10.53(2) 1.2201(3) 0.162 053(53) demonstrated in our previous study.52 190 17.12(3) 1.9850(8) 0.263 642(50) 3.2. Uncertainty Evaluation. Uncertainty estimations for 192 26.63(5) 3.087(3) 0.410 017(185) the calibrated osmium isotope ratios were done in accordance aValues are presented in a concise notation whereby the combined with the JCGM 100:2008 “Guide to the Expression of standard uncertainty is given in parentheses next to the least 68 fi Uncertainty in Measurement” and its Supplement 1. In signi cant digits to which it applies; for example, R192/188 = 3.087(3) short, ordinary least-squares fit was applied to each regression represents a mean value of R192/188 = 3.087 with combined standard uncertainty u (R ) = 0.003. set, and the corresponding intercept, slope, and their c 192/188 uncertainties were obtained. From these values, one calculates the osmium isotope ratios as per eq 2. The values for the three 3.5. Effect of the Sample Provenance. Acore variables in eq 2 (R193/191, a, and b) were modeled as random assumption in isotope systematics of many elements, including numbers drawn from the probability distributions representing osmium, is that the ratios of nonradiogenic isotopes are the available knowledge about them. In particular, R193/191 was expected not to vary in nature whereas ratios involving modeled as normal distribution with the mean R193/191 and its osmium-187 will vary significantly. We have surveyed the reported standard uncertainty u(R193/191) whereas a and b for isotopic composition of five commercial osmium standards each measurement set are modeled jointly as a bivariate normal relative to OSIS-1 showing significant spread in the distribution with the mean estimates and covariance matrix 187Os/188Os ratio (Table 3). deriving from the ordinary least-squares fitting of the data No significant differences in the osmium isotope ratios were (parametric bootstrap resampling). This procedure was found at 1 part per thousand level measurement uncertainty repeated 105 times, and the best estimate of the calibrated with the exception of the ratios involving osmium-187; relative osmium isotope ratios and their uncertainty is obtained from variations in the 187Os/188Os ratio, however, reached 50%. For the resulting histograms. example, R187/188(Os) = 0.1159(2) in OSIS-1 standard whereas

9284 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288 Analytical Chemistry Article

Table 3. Relative Isotope Ratios (Isotope Deltas) of ratios is proportional to the mass difference of these isotopes in Osmium in Several Commercial Reagents against OSIS-1 a logarithmic scale: δ a Standard, OSIS‑1(Os) lnKK (AB Os/ Os)/ln ( CD Os/ Os) Material 187Os/188Os 189Os/188Os =−(lnmm (AB Os) ln ( Os)) Os-1 −78.7(1) ‰ +0.05(9) ‰ CD Os-2 +405.9(2) ‰ −0.32(14) ‰ /(lnmm ( Os)− ln ( Os)) (3) Os-3 −78.1(1) ‰ +0.24(10) ‰ i j Os-4 −24.0(1) ‰ −0.29(12) ‰ Here K( Os/ Os) is the bias of the corresponding isotope ratio, fi DROsS +390.8(2) ‰ −0.16(10) ‰ de ned as the ratio of the true (calibrated) and the measured a1s is given in parentheses next to the least significant digits to which (raw) isotope ratios. it applies. In order to evaluate the relationships between the observed mass bias values, we have evaluated the measurements of the OSIS-1 standard. For each measurement, the value 190 188 in DROsS it is R187/188(Os, DROsS) = R187/188(Os, OSIS-1) × K( Os/ Os) was then used to calculate all other values as [1 + δ (Os, DDrOsS)] and results in a value of R (Os, ’ A 188 Os‑1 187/188 per Russell s law (eq 3). The obtained values, KR( Os/ Os), DROsS) = 0.1159(2) × [1 + 0.3908(2)] = 0.1612(3). were then compared to the actual values of K(AOs/188Os). The “ fi ” The certi ed values for radiogenic isotope ratios of R187/188 ratio of the calculated and observed mass bias values for all and R186/188 in DROsS are listed as 0.160 924(4) and 0.119 osmium isotope ratios is shown in Figure 2. 9293(58),33 shown in Table 4, which are in agreement with our measurements. This agreement, however, is largely due to the fact that our estimate of 192Os/188Os, 3.087(3), aligns well with the frequently used fixed value of 3.083 (Nier’s value in 1937). It is important to appreciate that the values frequently used in geosciences, including the DROsS value, do not carry full uncertainty statements as they are based on assumed fixed values of a nonradiogenic isotope ratio of osmium (by a fixed value we mean one that carries no associated uncertainty). 3.6. Isotopic Abundance of Osmium-184. Despite osmium-184 being the least abundant of all isotopes, it plays a key role in the study of nucleosynthetic processes, in particular the P-process, occurring in stars.33,72,73 Luguet et al.33 used the ratio 184Os/188Os as a second normalizing ratio for mass bias correction despite the fact that its true value is Figure 2. Deviations from Russell’slawintheMC-ICPMS 184 192 29 measurements of OSIS-1. All mass bias values were calculated from not well-known. Nier reported Os/ Os = 0.000 43, 190 188 whereas Luck and Allegre reported ca. 30% higher value in the Os/ Os isotope ratio which were then compared to the values Merck standard, 184Os/192Os = 0.000 58 ± 0.000 07. Our determined independently using the regression-based method (from 184 192 the results shown in Table 2). The black lines connect the observed results for OSIS-1 provide Os/ Os = 0.000 421 ± 0.000 ’ ’ deviations from the Russell s law of osmium isotope ratios from each 005 (k = 2) which supports Nier s value and other measurement sequence. compilations made by Masuda et al., who reported 184Os/192Os ratios of 0.000 42, 0.000 50, and 0.000 45, from three commercial osmium reagents,74 and is also close to Our results show that the 190Os/188Os isotope ratio can be 184Os/192Os = 0.000 48 ± 0.000 11 reported by the value of used reasonably well in conjunction with the Russell’s model to Völkening et al.3 correct the instrumental discrimination in MC-ICPMS of 3.7. Mass-Independent Fractionation. Typical analyses isotope ratio 189Os/188Os. However, the mass bias of of osmium isotopes rely on the invariance of nonradiogenic 187Os/188Os does not follow the mass-dependent model. We ratios and on the applicability of a mass bias correction model. observed relative deviations from the mass-dependent This approach adopts a fixed true value, as an example, for the fractionation model in the order of a few parts per thousand 192Os/188Os ratio and then uses the measured isotope ratio for the 187Os/188Os ratio. Deviations from Russell’s law can value to correct all other osmium isotope ratios for the also be evidenced from the slopes of the measured isotope instrumental fractionation using a function that extends the ratios. The slope of the lnr187/188 vs lnr190/188 plot is the observed mass bias from one isotope ratio to all others. (logarithmic) ratio of the corresponding average mass biases, ’ Frequently such function is the Russell s mass bias model which, according to eq 3 implies that the slope of the lnr187/188 which asserts that the relative bias of the measured isotope vs lnr189/188 plot is (lnm187 − lnm188)/(lnm189 − lnm188)=

Table 4. Summary of Osmium Isotope Ratios Obtained for the DROsS Standard R R Method Lead author 187/188 186/188 Traceability ref a DS N-TIMS Luguet, 2008 0.160924 ± 4 0.1199293 ± 58 R192/188 = 3.083 33 a DS MC-ICPMS Nanne, 2017 0.160916 ± 20 0.119909 ± 17 R192/188 = 3.083 24 a DS N-TIMS Nanne, 2017 0.160916 ± 43 0.119909 ± 43 R192/188 = 3.083 24 MC-ICPMSb this work 0.16119 ± 60 0.11986 ± 10 NRC IRIS-1 aReproducibility standard deviations are quoted, ±2s. bCombined expanded uncertainty is quoted in the parentheses, ±2u.

9285 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288 Analytical Chemistry Article

−1.003. Our measurements of OSIS-1 provide regression against conventional values which are not fully standardized. slopes ranging from −1.23 to −0.81 (see Figure 3). The Nier himself noted that the abundance of 188Os relative to 192Os “is correct to one percent” which means that the ratio 192Os/188Os provided by Nier is to be interpreted as 3.08 ± 0.03 and not “3.083”. Conferring any additional digits to Nier’s value is unjustified. Chatterjee and Lassiter, for example, have noted that the canonical values used in most studies are mutually incon- sistent90 referring to a parts per million level discrepancy in the 189Os/188Os ratios used by others (1.219 78 vs 1.219 73). In 1983, Luck and Allegre admitted that these reference values are “arbitrary”85 and the situation has not changed since then. In this vein, although the aim of this work was to provide determination of osmium isotope ratios independent of any conventions, the values obtained for the OSIS-1 (Figure 4) might be of relevance in debating the choice of conventional Figure 3. Deviations from the Russell’s law in the MC-ICPMS osmium isotope ratios which ideally would coincide with our measurements of OSIS-1 as evidenced from the random variation of best estimates of the corresponding true values. the three-isotope plot slope. random character of the three-isotope plot slopes demonstrates that the instrumental fractionation of osmium isotope ratios in MC-ICPMS is not entirely driven by mass-dependent fractionation and that a significant component remains unexplained (and is commonly known as the mass- independent fractionation). The 20% variation of the three- isotope plot slope shown in Figure 3 has the effect of imparting an uncertainty to the mass-bias corrected 187Os/188Os isotope ratio from the assumed 189Os/188Os value in the order of a few parts per thousand. Figure 4. Histograms of the osmium isotope ratios in the OSIS-1 We suggest that deviations from the mass-dependent standard along with typical canonical reference values. behavior are a reason in the observed 3 parts per thousand spread in the MC-ICPMS interlaboratory comparison data.75 This study was conducted under the auspices of the International Association of Geoanalysts (IAG) in 2011 and 4. CONCLUSIONS was concluded with the following comment by the organizers: We have determined the isotopic composition of osmium “discrepancies between the results [of 187Os/188Os] of the standard solution which will become certified reference participating laboratories are all too obvious, challenging their material OSIS-1 available from the NRC. Our MC-ICPMS 75 presumed “expertise”. measurement is the first independent determination of osmium The exact causes of MIF in MC-ICPMS are still isotopic composition since Nier (1937). In the absence of 35,36,38,40,41,76−83 unclear and two physical effects are frequently separated osmium isotope standards, we rely on the optimized thought to cause MIF: the nuclear volume effect and the regression mass bias correction model with the values 77,78 magnetic isotope effect. Detailed explanations of both ultimately traceable to the NIST thallium standard 997 and 76,78,79 effects can be found elsewhere. However, these effects NIST rhenium standard 989 through NRC iridium standard cannot fully explain the observed MIF in Os isotopes. For IRIS-1. We also report significant departures in MC-ICPMS example, both isotopes of 187Os and 189Os have nonzero from the Russell’s mass-dependent fractionation model (mass nuclear values, yet significant MIF was observed only for independent fractionation) for isotopes 187Os and 192Os. 187Os. More research in this area is needed to facilitate our understanding of MIF occurring in MC-ICPMS, but it is ■ ASSOCIATED CONTENT beyond the scope of current study. *S Supporting Information 3.8. Canonical Reference Values. To the best of our The Supporting Information is available free of charge on the knowledge, all previous osmium isotope ratio measurements ACS Publications website at DOI: 10.1021/acs.anal- rely, directly or indirectly, on osmium isotope ratios measured chem.8b01859. by Nier in 1937.29 Although the 1937 study reported isotope 192 188 Tables S1. MC-ICP-MS operating conditions. Table S2. ratio N( Os)/N( Os) = 100/32.4 = 3.0864, in a later Possible interferences on measured isotopes. (PDF) publication,84 Nier summarized these values as N(192Os)/ N(188Os) = 41.0/13.3 = 3.0827, which is the value often used 85,86 ■ AUTHOR INFORMATION today. Further to this, others have renormalized Nier’s values, which alters the value for 188Os from 32.4 to 32.44, thus Corresponding Author giving rise to yet another normalizing ratio for 192Os/188Os = *E-mail: [email protected]. 3.082 614 (=100/32.44).2,25,87−89 Despite these discrepancies, ORCID most researchers rely on normalizing osmium isotope ratios Lu Yang: 0000-0002-6896-8603

9286 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288 Analytical Chemistry Article

Notes (34) Vance, D.; Thirlwall, M. Chem. Geol. 2002, 185, 227−240. The authors declare no competing financial interest. (35) Newman, K.; Freedman, P. A.; Williams, J.; Belshaw, N. S.; Halliday, A. N. J. Anal. At. Spectrom. 2009, 24, 742−751. ■ ACKNOWLEDGMENTS (36) Newman. J. Anal. At. Spectrom. 2012, 27,63−70. (37) Xu, L.; Hu, Z.; Zhang, W.; Yang, L.; Liu, Y.; Gao, S.; Luo, T.; China Scholarship Council is gratefully acknowledged for Hu, S. J. Anal. At. Spectrom. 2015, 30, 232−244. financial support of Z. Zhu and S. Tong during the study. (38) Shirai, N.; Humayun, M. J. Anal. At. Spectrom. 2011, 26, 1414− 1420. ■ REFERENCES (39) Irrgeher, J.; Prohaska, T.; Sturgeon, R. E.; Mester, Z.; Yang, L. Anal. Methods 2013, 5, 1687−1694. (1) Riley, G. H.; Delong, S. E. Int. J. Mass Spectrom. Ion Phys. 1970, (40) Yang, L.; Mester, Z.; Zhou, L.; Gao, S.; Sturgeon, R. E.; Meija, 4, 297−304. J. Anal. Chem. 2011, 83, 8999−9004. (2) Creaser, R. A.; Papanastassiou, D. A.; Wasserburg, G. J. Geochim. (41) Yang, L.; Zhou, L.; Hu, Z.; Gao, S. Anal. Chem. 2014, 86, Cosmochim. Acta 1991, 55, 397−401. 9301−9308. (3) Völkening, J.; Walczyk, T.; Heumann, K. G. Int. J. Mass Spectrom. (42) Miyazaki, T.; Kimura, J. I.; Chang, Q. J. Anal. At. Spectrom. Ion Processes 1991, 105, 147−159. 2014, 29, 483−490. (4) Shirey, S. B.; Walker, R. J. Annu. Rev. Earth Planet. Sci. 1998, 26 (43) Meija, J. Anal. Bioanal. Chem. 2012, 403, 2071. (1), 423−500. (44) Qi, H. P.; Berglund, M.; Taylor, P. D. P.; Hendrickx, F.; (5) Carlson, R. W. Lithos 2005, 82, 249−272. Verbruggen, A.; De Bievre,̀ P. Fresenius' J. Anal. Chem. 1998, 361, (6) Walker, R. J.; Horan, M. F.; Morgan, J. W.; Becker, H.; − Grossman, J. N.; Rubin, A. E. Geochim. Cosmochim. Acta 2002, 66, 767 773. 4187−4201. (45) Malinovsky, D.; Dunn, P. J. H.; Goenaga-Infante, H. J. Anal. At. − (7) Meisel, T.; Walker, R. J.; Morgan, J. W. Nature 1996, 383, 517− Spectrom. 2013, 28, 1760 1771. 520. (46) Rienitz, O.; Pramann, A.; Schiel, D. Int. J. Mass Spectrom. 2010, − (8) Brandon, A. D.; Walker, R. J.; Morgan, J. W.; Norman, M. D.; 289,47 53. Prichard, H. M. Science 1998, 280, 1570−1573. (47) Yang, L.; Mester, Z.; Sturgeon, R. E.; Meija, J. Anal. Chem. − (9) Walker, R. J. Chem. Erde 2009, 69, 101−125. 2012, 84, 2321 2327. (10) Snow, J. E.; Reisberg, L. Earth Planet. Sci. Lett. 1995, 136, 723− (48) Song, P.; Wang, J.; Ren, T.; Zhou, T.; Zhou, Y.; Wang, S. Anal. Chem. 2017, 89, 9031−9038. 733. ́ ́ ̀ (11) Meibom, A.; Sleep, N. H.; Chamberlain, C. P.; Coleman, R. G. (49) Marechal, C. N.; Telouk, P.; Albarede, F. Chem. Geol. 1999, Nature 2002, 419, 705−708. 156, 251−273. (12) Brandon, A. D.; Creaser, R. A.; Shirey, S. B.; Carlson, R. W. (50) Yang, L.; Meija, J. Anal. Chem. 2010, 82, 4188−4193. Science 1996, 272, 861−863. (51) Malinovsky, D.; Dunn, P. J. H.; Goenaga-Infante, H. J. Anal. At. (13) Pearson, D. G.; Parman, S. W.; Nowell, G. M. Nature 2007, Spectrom. 2016, 31, 1978−1988. 449, 202−205. (52) Zhu, Z.; Meija, J.; Zheng, A.; Mester, Z.; Yang, L. Anal. Chem. (14) Dale, C. W.; Gannoun, A.; Burton, K. W.; Argles, T. W.; 2017, 89, 9375−9382. Parkinson, I. J. Earth Planet. Sci. Lett. 2007, 253, 211−225. (53) Yang, L.; Sturgeon, R. E. Anal. Bioanal. Chem. 2009, 393, 377− (15) Stein, H. J.; Sundblad, K.; Markey, R. J.; Morgan, J. W.; 385. Motuza, G. Miner. Deposita 1998, 33, 329−345. (54) Meija, J.; Yang, L.; Sturgeon, R. E.; Mester, Z. J. Anal. At. (16) Burton, K. W. J. J. Geochem. Explor. 2006, 88, 262−265. Spectrom. 2010, 25, 384−389. (17) Pegram, W. J.; Krishnaswami, S.; Ravizza, G. E.; Turekian, K. K. (55) Xie, Q.; Lie, S.; Evans, D.; Dillon, P.; Hintelmann, H. J. J. Anal. Earth Planet. Sci. Lett. 1992, 113, 569−576. At. Spectrom. 2005, 20, 515−522. (18) Peucker-Ehrenbrink, B.; Ravizza, G. Terra Nova 2000, 12, 205− (56) Yang, L.; Meija, J. Anal. Chem. 2010, 82, 4188−4193. 219. (57) Yang, L.; Sturgeon, R. E.; Mester, Z.; Meija, J. Anal. Chem. (19) Ravizza, G. E.; Bothner, M. H. Geochim. Cosmochim. Acta 1996, 2010, 82, 8978−8982. 60, 2753−2763. (58) Hou, Q.; Zhou, L.; Gao, S.; Zhang, T.; Feng, L.; Yang, L. J. (20) Klemm, V.; Levasseur, S.; Frank, M.; Hein, J. R.; Halliday, A. N. Anal. At. Spectrom. 2016, 31, 280−287. Earth Planet. Sci. Lett. 2005, 238,42−48. (59) Anbar, A. D.; Knab, K. A.; Barling. Anal. Chem. 2001, 73, (21) Gannoun, A.; Burton, K. W. J. Anal. At. Spectrom. 2014, 29, 1425−1431. 2330−2342. (60) Arnold, G. L.; Weyer, S.; Anbar, A. D. Anal. Chem. 2004, 76, (22) Mathur, R.; Ruiz, J.; Tornos, F. Miner. Deposita 1999, 34, 790− 322−327. ̀ ́ 793. (61) White, W. M.; Albarede, F.; Telouk, P. Chem. Geol. 2000, 167, (23) Walker, R. J.; Morgan, J. W. Science 1989, 243, 519−522. 257−270. (24) Nanne, J. A.; Millet, M. A.; Burton, K. W.; Dale, C. W.; Nowell, (62) Collerson, K. D.; Kamber, B. S.; Schoenberg, R. Chem. Geol. G. M.; Williams, H. M. J. Anal. At. Spectrom. 2017, 32, 749−765. 2002, 188,65−83. (25) Hirata, T.; Hattori, M.; Tanaka, T. Chem. Geol. 1998, 144, (63) Yuan, W.; Chen, J. B.; Birck, J. L.; Yin, Z. Y.; Yuan, S. L.; Cai, 269−280. H. M.; Wang, Z. W.; Huang, Q.; Wang, Z. H. Anal. Chem. 2016, 88, (26) Nowell, G. M.; Luguet, A.; Pearson, D. G.; Horstwood, M. S. A. 9606−9613. Chem. Geol. 2008, 248, 363−393. (64) Bouman, C.; Trinquier, A.; LIoyd, N.; Schwieters, J.; (27) Meija, J. Nat. Chem. 2014, 6, 749−750. Koornneef, J.; Davies, G. Application Note 30282; Thermo Fisher (28) Meija, J.; Coplen, T. B.; Berglund, M.; Brand, W. A.; De Bievre, Scientific: Bremen, Germany, 2015. P.; Groning, M.; Holden, N. E.; Irrgeher, J.; Loss, R. D. Pure Appl. (65) Koornneef, J. M.; Bouman, C.; Schwieters, J. B.; Davies, G. R. Chem. 2016, 83, 265−291. Anal. Chim. Acta 2014, 819,49−55. (29) Nier, A. O. Phys. Rev. 1937, 52, 885. (66) Gustinelli Arantes de Carvalho, G.; Vitoriano Oliveira, P.; Yang, (30) Wang, G.; Sun, T.; Xu, J. Rapid Commun. Mass Spectrom. 2017, L. J. Anal. At. Spectrom. 2017, 32, 987−995. 31, 1616−1622. (67) Meija, J.; Yang, L.; Sturgeon, R. E.; Mester, Z. Anal. Chem. (31) Yang, L. Mass Spectrom. Rev. 2009, 28, 990−1011. 2009, 81, 6774−6778. (32) Walczyk, T. Anal. Bioanal. Chem. 2004, 378, 229−231. (68) JCGM. Evaluation of measurement data − Supplement 1 to the (33) Luguet, A.; Nowell, G. M.; Pearson, D. G. Chem. Geol. 2008, “Guide to the expression of uncertainty in measurement” − Propagation of 248, 342−362. distributions using a Monte Carlo method; JCGM 101; 2008.

9287 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288 Analytical Chemistry Article

(69) Meija, J.; Mester, Z. Metrologia 2008, 45,53−62. (70) Meija, J.; Possolo, A. Metrologia 2017, 54, 229−238. (71) Wang, M.; Audi, G.; Kondev, F. G.; Huang, W. J.; Naimi, S.; Xu, X. Chin. Phys. C 2017, 41, 030003. (72) Liu, Y.; Huang, M.; Masuda, A.; Inoue, M. Int. J. Mass Spectrom. Ion Processes 1998, 173, 163−175. (73) Brandon, A. D.; Walker, R. J.; Puchtel, I. S. Geochim. Cosmochim. Acta 2006, 70, 2093−2103. (74) Masuda, A.; Hirata, T.; Shimizu, H. Geochem. J. 1986, 20, 233− 239. (75) Meisel, T.; Kane, J. S. Accredit. Qual. Assur. 2011, 16, 407−414. (76) Fujii, T.; Moynier, F.; Albarede,̀ F. Chem. Geol. 2009, 267, 139−156. (77) Malinovsky, D.; Vanhaecke, F. Anal. Bioanal. Chem. 2011, 400, 1619−1624. (78) Bigeleisen, J. J. Am. Chem. Soc. 1996, 118, 3676−3680. (79) Bergquist, B. A.; Blum, J. D. Science 2007, 318, 417−420. (80) Buchachenko, A. L. J. Phys. Chem. A 2001, 105, 9995−10011. (81) Shirai, N.; Humayun, M. J. Anal. At. Spectrom. 2011, 26, 1414− 1420. (82) Angeli, I. At. Data Nucl. Data Tables 2004, 87, 185−206. (83) Tables of Nuclear Data: http://wwwndc.jaea.go.jp/NuC/index. html. (84) Bainbridge, K. T.; Nier, A. O. Relative Isotopic Ahundances of the Elements (Vol. 9), Preliminary Report; National Academies, 1950. (85) Luck, J. M.; Allegre,̀ C. J. Nature 1983, 302, 130−132. (86) Dickin, A. P.; McNutt, R. H.; McAndrew, J. I. J. Anal. At. Spectrom. 1988, 3, 337−342. (87) Schoenberg, R.; Nagler,̈ T. F.; Kramers, J. D. Int. J. Mass Spectrom. 2000, 197,85−94. (88) Yin, Q. Z.; Jacobsen, S. B.; Lee, C. T.; McDonough, W. F.; Rudnick, R. L.; Horn, I. Geochim. Cosmochim. Acta 2001, 65, 2113− 2127. (89) Meija, J.; Coplen, T. B.; Berglund, M.; Brand, W. A.; De Bievre,̀ P.; Gröning, M.; Holden, N. E.; Irrgeher, J.; Loss, R. D.; Walczyk, T.; Prohaska, T. Pure Appl. Chem. 2016, 88, 293−306. (90) Chatterjee, R.; Lassiter, J. C. Chem. Geol. 2015, 396, 112−123.

9288 DOI: 10.1021/acs.analchem.8b01859 Anal. Chem. 2018, 90, 9281−9288