Determination of Osmium Concentration and Isotope Composition at

Apr 8, 2018 - We analyze snow collected from Summit, Greenland during 2009, 2014, and 2017. We find that the average Os concentration of the snow is 0...
3 downloads 4 Views 784KB Size
Download PDF
Article Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/ac

Determination of Osmium Concentration and Isotope Composition at Ultra-low Level in Polar Ice and Snow Ji-Hye Seo, Mukul Sharma,* Erich C. Osterberg, and Brian P. Jackson Department of Earth Sciences, Dartmouth College, Hanover, New Hampshire 03755, United States S Supporting Information *

ABSTRACT: Here we use two chemical separation procedures to determine exceptionally low Os concentrations (∼10−15 g g−1) and Os isotopic composition in polar snow/ ice. Approximately 50 g of meltwater is spiked with 190Os tracer solution and frozen at −20 °C in quartz-glass ampules. A mixture of H2O2 and HNO3 is then added, and the sample is heated to 300 °C at 100 bar. This allows tracer Os to be equilibrated with the sample as all Os species are oxidized to OsO4. The resulting OsO4 is separated using either distillation (Method-I) or solvent-extraction (Method-II), purified, and measured using negative thermal ionization mass spectrometry (N-TIMS). A new technique is presented that minimizes Re and Os blanks of the Pt filaments used in N-TIMS. We analyze snow collected from Summit, Greenland during 2009, 2014, and 2017. We find that the average Os concentration of the snow is 0.459 ± 0.018 (95% C.I.) fg g−1 corresponding to an Os flux of 0.0579 ± 0.0023 (95% C.I.) fmol cm−2 yr−1. The average R(187Os/188Os) ratio of the Summit snow is 0.264 ± 0.026 (95% C.I.). Assuming that the volcanic source is negligible, the average ratio indicates that about 0.0518 ± 0.0040 (95% C.I.) fmol cm−2 yr−1 of Os is of cosmic derivation, corresponding to an accretion rate of extra-terrestrial Os to the Earth of 264 ± 21 mol yr−1.

T

the methods1,15,16 developed to determine PGEs in ancient ice cannot be used to separate Os. Here we present two different techniques that allow the determination of Os concentration and isotope ratios in polar ice melt weighing ∼50 g. The methods use a mixture of HNO3 and H2O2 at high temperature to convert sample and tracer Os to volatile OsO4. In the first method, OsO4 is distilled from water.13 In the second method, OsO4 is separated from water by solvent extraction using Br2(l).12 Using these procedures, we present results for recent snow from Summit, Greenland and assess extra-terrestrial Os contributions.

he intent of this study is to develop a highly sensitive procedure that allows accurate and precise determination of Os concentration and isotope composition from ∼50 g of melted ice. Accumulations of Pt and Ir belonging to the platinum group elements (PGEs: Ru, Rh, Pd, Os, Ir, and Pt) in ancient polar archives have been argued to trace terrestrial (continental/volcanic dust) and extra-terrestrial sources.1−4 The latter include cosmic dust/micrometeorites (mean size of ∼200 μm)5 and nanometer-sized “meteoric smoke” from vaporization of cosmic dust or larger meteoroids.6−8 The PGE concentration data, however, lack specificity. For example, the extent to which terrestrial dust compared with cosmic dust has contributed to the PGE inventory of polar ice cannot be readily evaluated from the PGE concentration data alone. The Os isotope ratios(R(187Os/188Os) ratio) of terrestrial (= 1.40 ± 0.30)9 and extraterrestrial/volcanic sources (= 0.13)10,11 are distinctly different from each other. Depth variations in the Os isotope ratio in polar ice can therefore quantitativelyelucidate changes in the accretion rate of extraterrestrial matter over last several hundred thousand years. However, the determination of Os isotopes in polar ice core archives is challenging due to extremely low concentrations and due to the availability of small sample sizes (tens of grams); the latter issue also makes it desirable to determine PGE concentrations from the sample solution from which Os has been separated. However, the extant procedures12−14 that can be potentially used to determine Os isotopes in polar ice are incompatible with those developed for the determination of other PGEs. Similarly, © XXXX American Chemical Society

■

EXPERIMENTAL SECTION

An outline contrasting the two experimental procedures is given in Figure 1. In this section we give details of (1) preparation of reagents and materials, (2) samples, (3) oxidation of Os in ice melt and its equilibration with tracer Os, and (4) mass spectrometry. In the subsequent section (Results and Discussion), we give the first results of selected samples from Greenland, assess the reproducibility of these results using repeated independent measurements, and compare two methods. Received: January 11, 2018 Accepted: April 8, 2018 Published: April 8, 2018 A

DOI: 10.1021/acs.analchem.8b00150 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Water. For this study we used 18.2 MΩ·cm water (Milli-Q) obtained using a Millipore water purification system that was further purified by distillation. Batches of pure waters called DMQ and DDMQ were obtained by, respectively, one- and two-times sub-boiling Milli-Q water distillation in the CUPOLA STILL. The DMQ water was used for cleaning of the materials used in the procedure, whereas the DDMQ water was used for all Os chemistry and for rinsing Pt filaments (see below). For the determination of the procedural blanks, the DDMQ water was further purified using a two-bottle distillation apparatus (= TDMQ).17 The Os blank for TDMQ in the collecting bottle (distillate) is higher than that of the water remaining in the feed bottle (residue).13 For the determination of procedure blank we used the TDMQ residue (≡ Os-free water) prepared by distilling ∼50% of DDMQ into the collecting bottle. Pt Ribbon. For mass-spectrometry, purified Os, hexabromoosmate is loaded on a high-purity Pt ribbon (H. Cross or ESPI Metals, 99.996%), 0.0254 mm thick and 0.478 mm wide. The ribbon is cut into strips of ∼2.0 cm length and are welded directly onto side filament posts. The filaments are outgassed for 20 min at ∼1470 °C (∼3.2A) under a slight vacuum using an aspirator. The outgassed Pt filaments, including filament posts, are first rinsed with the DDMQ and dried. The cleaned filaments are used for both sample loading and for providing an external source of heat in a double filament assembly. Ba(OH)2 Emitter Solution. Deposition of electropositive elements such as Ba reduces the electron work function of Pt filaments and promotes the production of negative thermal ions.18 While initial experimentation with Ba(NO3)219 and Ba(OH)220 emitter solutions produced intense OsO3− ion beams, a stable and robust ion emission is obtained when Ba(OH)2 is mixed with NaOH (see e.g., Chen and Sharma, 200913). For this study the emitter solution is freshly made prior to loading by dissolving 0.56 g of Ba(OH)2·8H2O (ACS grade, J.T. Baker) and 0.12 g NaOH (ACS grade, Fisher) in 30 mL of DDMQ. Cleaning of Lab Ware and Mass Spectrometer Parts. All labware is cleaned sequentially in concentrated HNO3 (ACS plus grade, Fisher Scientific), concentrated HCl (ACS plus grade, Fisher Scientific), and Milli-Q water overnight on a hot plate. Glassware is then rinsed with Milli-Q water, followed by DMQ water, and air-dried under a laminar flow hood. All fluorinated ethylene propylene (FEP; Savillex) and perfluoroalkoxy alkane (PFA; Savillex) labware used for Os chemical separation is assembled and cleaned further using HBr (2× lab distilled ACS grade, Alfa Aesar) at 80 °C for 12 h. Polypropylene pipet tips are cleaned by soaking at room temperature in 20% v/v HNO3 (ACS plus grade, Fisher Scientific) for >2 weeks, followed by 10% v/v HCl (1× distilled trace metal grade, Fisher Scientific) for >1 week, followed by Milli-Q water for 24 h. All filament posts and coverslips are cleaned in RBS pF detergent concentrate (Thermo Scientific), sonicated for 1h, rinsed with Milli-Q water, and dried under the laminar flow hood. All extraction slits are first cleaned with aluminum oxide/alumina powder (Micropolish II, Beuhler) and rinsed with Milli-Q water. They are then cleaned with the same procedure as filament posts and coverslips. Samples. The c(Os) and isotope compositions were measured in fresh snow collected in 2009, 2014, and 2017 from the clean air sector in the Summit, Greenland (72°34′N, 38°28′W) and in late 1700s firn collected during the North Greenland Eemian Ice Drilling Project (NEEM; 72°27′N,

Figure 1. Flow diagram of two Os separation methods.

Preparation of Reagents and Materials. Establishing ultra-low procedural blanks is imperative due to the extremely low PGE concentrations of polar ice. Polar ice is expected to have an Os concentration (hereafter = c(Os)) of ∼1 fg g−1 (= 10−15 g g−1, i.e., 3 × 106 atoms g−1). For ∼50 g of ice-melt, the Os blank needs to be 1900 fg g−1). It is cleaned by dissolving in concentrated HBr and double-distilling the mixture in the CUPOLA STILL by PicoTrace, Germany (Supporting Information). Extreme care must be taken when handling Br2(l), as it is highly oxidizing. It reacts violently with metals and dissolves a number of polymers. Inhalation exposure to Br2 (bp = 59 °C) is fatal even at low concentrations (IDLH = 3 ppm, https://www.cdc. gov/niosh/idlh/7726956.html). B

DOI: 10.1021/acs.analchem.8b00150 Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

■

RESULTS AND DISCUSSION Mass Spectrometric Measurements. Platinum filaments coated with Ba salts have been found to provide the most intense OsO3− ion beams and least interferences.19,20 However, even the highest purity zone refined Pt ribbons contain hundreds of ng g−1 of Os, which contributes to filament blank during thermal ionization of the sample. A traditional method to reduce Os blank is by repeatedly heating the Pt filament at low temperature in air.13 This procedure oxidizes Os present near the filament surface to volatile OsO4 and has yielded Os loading blanks of ∼1 fg.13 Following extensive experimentation (Supporting Information), we have developed a procedure that reduces the filament loading blank to ∼0.2 fg (Table 1). This

38°04′W). The snow samples collected in 2009 and 2014 were obtained from the surface to 10 cm deep using an acid-washed high-density polyethylene (HDPE) scoop. These samples were then stored unacidified in 1 L FEP bottles. In 2017, the snow was collected from a 1 m deep snowpit. The snow was then stored in a 1 L PFA collection jar. This was then acidified and vortexed. The firn was melted in Korea Polar Research Institute (KOPRI) using a melting system.21 The outer layer was removed for possible contamination from drilling, and the middle inner layer of the core was then used for the development and testing of our methods. All samples arrived melted in acid-cleaned sampling containers. Oxidation of Os in Ice Melt and Its Equilibration with Tracer Os. About 50 g of the water is spiked with tracer solutions containing 190Os in a quartz-glass reaction vessel. The sample is then frozen at −20 °C to minimize oxidation of tracer/sample Os in air during subsequent addition of reagents. Chilled 2.5 mL of the purified HNO3 and 1.5 mL of H2O2 are added to the frozen sample, and the vessel is closed immediately by securing a glass lid using polytetrafluoroethylene (PTFE) tape. The vessel is then placed in a highpressure asher (HPA-S; Anton Paar) and effectively sealed by pressurizing the HPA-S to 100 bar using nitrogen gas. Once the pressure is reached, the sample is allowed to thaw for about an hour and then heated to 300 °C for 16 h, allowing all Os species to oxidize to OsO4. Following this step, the sample is cooled for ∼1.5 h in the HPA-S, which is then vented slowly (10 bar min−1) to prevent sample leakage from rapid decompression. The bulk of the resulting OsO4 remains dissolved in aqueous solution and is separated by distillation13 (Method-I) or by solvent extraction using liquid bromine12 (Method-II). The resulting hexabromoosmate, H2(OsBr6), is dried and further purified using microdistillation.22 Mass Spectrometry. Purified hexabromoosmate in HBr is dried to a volume of ∼0.4 μL and then placed on a precleaned Pt filament using polyethylene tubing (PE-20, 0.04 cm I.D. × 0.1 cm O.D., Scientific Commodities) fitted to a micropipette (Hamilton). Samples are loaded under a microscope in a small area of about 1.5 mm × 1.5 mm and dried electrothermally using a DC power supply at 0.6 A. Approximately 0.2 μL of emitter solution is then placed on top of the evaporated sample and dried. To ensure complete dryness of the sample, the filament current is slowly raised to 1.2 A for 6 s as a final step.13 On the surface of the heated Pt filament and at pressures below ∼10−6 mbar Os ionizes preferentially as OsO3−.19,20 We use double filament geometry13 to obtain an OsO3− ion beam at a relatively low temperature (1900 Lab 2nd distilled Br2 (l) 0.051 ± 0.013 Loading Blank (Pt Filament + Ba(OH)2) H.Cross Pt filamentc 0.10 ± 0.13 fg ESPI metals Pt filamentd 0.22 ± 0.22 fg Reagent Blank + Loading Blank NP + Method-I 2.28 fg (1.15−4.74)e NP + Method-II 2.68 fg (1.02−4.36)e Procedural Blank 25.77 g TDMQ residuef 11.65 ± 0.26 fg 28.83 g TDMQ residuef 12.44 ± 0.62 fg 28.35 g TDMQ residueg 12.67 ± 0.45 fg 26.86 g TDMQ residueg 11.62 ± 0.26 fg TDMQ residue 0.35 ± 0.31 fg g−1 h NP + Method-I 2.3 ± 2.0 fg NPh + Method-II 2.8 ± 2.4 fg JRi + Method-I 2.7 ± 2.4 fg JRi + Method-II 3.1 ± 2.8 fg

R(187Os/188Os) 0.246 ± 0.031 0.218 ± 0.044 0.58 ± 0.51 0.31 ± 0.19 0.26 ± 0.27 0.313 ± 0.091 1.392 ± 0.079 0.134 ± 0.030 ∼0.17 ∼0.17 0.31 ± 0.17 0.31 ± 0.17 0.351 ± 0.036 0.722 ± 0.077 0.388 ± 0.053 0.303 ± 0.030 0.17 (0.09−0.19)e 0.42 ± 0.20 0.40 ± 0.19 0.95 ± 0.49 0.85 ± 0.44

a

Precisions in Os concentrations and the R(187Os/188Os) ratios are 2 SD. bAssessed using the measured reagent blanks and procedural blanks. cUsed in Method-I. dUsed in Method-II. eRange of values indicates max and min of assessed blanks. fDetermined by Method-I using 0.5 mL of Jones reagent. gDetermined by Method-II using 2.5 mL of HNO3 and 1.5 mL of H2O2. hNP refers to 2.5 mL of HNO3 and 1.5 mL of H2O2. iJR refers to 0.5 mL of Jones reagent.

procedure has minimized Re isobaric interference as well as hydrocarbon interference (Figure 2). Using this procedure we obtain R(187Os/188Os) (= 0.1070 ± 0.0017, 2 S.D.) and R(190Os/188Os) (= 1.983 ± 0.010, 2 S.D.) ratios for 19 independent loads of 25 fg of our house Os standard (= MPI1). Ultra-high precision measurements of six 5 ng loads of MPI1 standard using static multicollection by Faraday collectors give weighted averages of R(187Os/188Os) and R(190Os/188Os) of 0.1068873 ± 0.0000065 (2 S.D.) and 1.98393 ± 0.00025 (2 S.D.), respectively. Thus, there is no systematic offset between high-precision isotope ratios and those measured by a single C

DOI: 10.1021/acs.analchem.8b00150 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 2. Measurements of R(187Os/188Os) ratios in MPI-1 Os standards with 2 S.D. Dashed line at 0.1069 is the reference value of the MPI-1 standard. Solid line is the average of measurement results of 50 and 25 fg of the standard. Error bars show 2 S.D. standard deviation. The total standards measured, the mean standard ratios, and 2 S.D. external reproducibilities (shaded area) are indicated at the bottom of each panel. No mass interference corrections from 185ReO3− on 187OsO3− were applied in all panels.

Table 2. Os Concentration and Isotope Composition of Antarctic Sea Snow, Sea Water, and NEEM Firna oxidizing reagents

extraction method

mass (g)

0.5 mL Jones reagentc 0.5 mL Jones reagentc 0.5 mL HNO3 + 0.5 mL H2O2d 0.5 mL HNO3 + 1 mL H2O2d 1 mL HNO3+ 0.5 mL H2O2d 2.5 mL HNO3 + 1.5 mL H2O2d

Distillation Distillation Method-I Method-I Method-I Method-I

∼50 ∼50 51.73 50.6 50.06 50.15

Distillation Distillation Method-II Method-II Method-I Method-I Method-I Method-II

sample Antarctica Ross Sea snow CORSACS2 IS-3 (acidified)

Deep Seawater Atlantic Ocean 2007 at depth 2000 m GEOTRACES (Acidified) Greenland NEEM firn late 1700s (unacidified)

0.5 0.5 0.5 2.5

mL mL mL mL

Jones reagentc Jones reagentc Jones reagente HNO3 + 1.5 mL H2O2

2.5 2.5 2.5 2.5

mL mL mL mL

HNO3 HNO3 HNO3 HNO3

+ + + +

1.5 1.5 1.5 1.5

mL mL mL mL

H2 O2 H2 O2 H2 O2 H2 O2

c(Os) (fg g−1) 0.83 0.81 0.2697 0.697 0.895 0.945

± ± ± ± ± ±

0.01 0.01 0.0070 0.018 0.023 0.025

60 mL 60 mL 52.2 52.22

8.67 8.91 8.62 8.55

± ± ± ±

49.84 46.16 51.31 50.26

1.437 1.339 0.734 0.910

± ± ± ±

[c(Os)]C (fg g‑1)b

R(187Os/188Os)

[R(187Os/188Os)]Cb

0.77 0.76 0.215 0.644 0.814 0.76

± ± ± ± ± ±

0.05 0.05 0.045 0.051 0.071 0.10

0.46 0.46 0.392 0.451 0.423 0.402

± ± ± ± ± ±

0.01 0.01 0.023 0.026 0.025 0.023

0.42 0.42 0.423 0.465 0.440 0.433

± ± ± ± ± ±

0.07 0.07 0.028 0.028 0.027 0.028

0.05 0.05 0.22 0.22

8.6 8.7 8.55 8.49

± ± ± ±

0.4 0.4 0.22 0.23

1.04 1.04 1.072 1.026

± ± ± ±

0.01 0.01 0.063 0.060

1.04 1.05 1.073 1.030

± ± ± ±

0.02 0.01 0.063 0.060

0.037 0.035 0.019 0.024

1.396 1.286 0.691 0.853

± ± ± ±

0.054 0.057 0.038 0.052

0.483 0.430 0.465 0.426

± ± ± ±

0.028 0.025 0.027 0.026

0.484 0.428 0.466 0.425

± ± ± ±

0.029 0.027 0.029 0.028

a

Uncertainties in Os concentrations and the R(187Os/188Os) ratios are 2 S.D. bCorrected for procedural blanks. cData from Chen et al. (2009). Blanks and their isotope ratios for each of the solutions were based on the amounts of reagents used: (2.47 fg: 0.23), (2.58 fg, 0.26), (3.81 fg, 0.23), and (8.03 fg, 0.24). eCorrected for 3.1 ± 2.4 fg blank with the R(187Os/188Os) ratio of 0.9 ± 0.4

d

SEM collector when only ∼75 million atoms of Os are loaded on the filament. Osmium Analytical Blanks and Yields. Blanks of the reagents and total procedure are given in Table 1. Because the Os concentration of polar ice/snow is extremely low, we expended considerable effort to determine and individual reagent blanks, which in large part, govern the total procedural blank (Supporting Information). Total blanks for different procedures were determined as follows. (Note that the data are given below as x ± y, where x = assessed blank and y = 2 S.D.) Four different aliquots of TDMQ residue weighing between 26 to 29 g were processed using 0.5 mL of JR and Method-I13 and using a mixture of 2.5 mL of HNO3 and 1.5 mL of H2O2 (nitric-peroxide; NP) and Method-I. These measurements yielded TDMQ with c(Os) of 0.35 ± 0.31 fg g−1 and a procedural blank for NP and Method-I of 2.3 ± 2.0 fg. Procedural blank for NP and Method-II (= 2.8 ± 2.4 fg) was therefore determined by simply adding the Br2 (l) blank to the NP and Method-I blank. Interestingly, >75% of the Os in these measurements is from our Os-free water. A regression between

the reciprocal of total Os contributed by different volumes of TDMQ residue processed using a single method and the corresponding measured R(187Os/188Os) ratios then allows us to determine the R(187Os/188Os) ratio of TDMQ. To this end, we combine previously published TDMQ data for JR and Method-I13 with that obtained in this study (Table 1) and find that the R(187Os/188Os) ratio of TDMQ residue is 0.17, which yields a R(187Os/188Os) ratio of the JR and Method-I procedural blank of 0.95+0.80 −0.18, corresponding to 2.7 ∓ 2.4 fg. Procedural blank for JR and Method-II (= 3.1 ± 2.8 fg) was again determined in the same manner as the blank for NP and Method-II. The calculated R(187Os/188Os) ratios of procedural blanks for NP and Methods-I and -II are 0.42 ± 0.20 and 0.40 ± 0.19, respectively. For JR and Methods-I and -II, the R(187Os/188Os) ratios are 0.95 ± 0.49 and 0.85 ± 0.44, respectively. The Os yields for Methods-I and -II are 9013 and 46%, respectively. Optimization of Os Oxidation. Closed-system hightemperature (300 °C) oxidation of Os dissolved in natural waters leads to its isotopic equilibration with a calibrated Os D

DOI: 10.1021/acs.analchem.8b00150 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Table 3. Os Concentration and Isotope Composition of Modern Snow from Summit, Greenlanda sample

extraction method

mass (g)

c(Os) (fg g−1)

[c(Os)]C (fg g‑1)c

R(187Os/188Os)

[R(187Os/188Os)]Cc

snow 2009 (unacidified)

JR + Method-I NPe + Method-I

50.21 52.46

0.522 ± 0.014 0.489 ± 0.013

0.471 ± 0.050 0.445 ± 0.040

0.359 ± 0.021 0.269 ± 0.016

0.30 ± 0.10 0.250 ± 0.039

snow 2014 (unacidified)

NPe + Method-1 NPe + Method-I NPe + Method-II

50.35 52.19 49.34

1.019 ± 0.027 0.492 ± 0.013 0.2953 ± 0.0077

0.966 ± 0.049 0.453 ± 0.043 0.233 ± 0.053

0.370 ± 0.022 0.210 ± 0.012 0.332 ± 0.019

0.363 ± 0.025 0.188 ± 0.041 0.294 ± 0.078

snow 2017 (acidified)

NPe + Method-II JRd + Method-II

50.02 51.27

0.664 ± 0.017 0.443 ± 0.012

0.614 ± 0.049 0.391 ± 0.053

0.258 ± 0.015 0.1550 ± 0.0091

0.245 ± 0.031 0.118 ± 0.066

d

a

Precisions in Os concentrations and the R(187Os/188Os) ratios are 2 S.D. bArea-time product of all snow samples is 2.1 cm2 yr. cCorrected for procedural blanks. dJR refers to 0.5 mL of Jones reagent. eNP refers to 2.5 mL of HNO3 and 1.5 mL of H2O2.

comes from a middle layer of the firn core and therefore may have been contaminated during the drilling and handling procedure. Interestingly, while the measured R(187Os/188Os) ratios of the NEEM aliquots vary from 0.43 to 0.48 (∼12%; Table 2), the c(Os) varies from 0.69 to 1.40 fg (∼64%; Table 2). The variance of the mean of the measurement results is much higher than indicated by our assessed analytical variability. This observation suggests that while the bulk of Os resides in nanoparticles, there may still be some larger refractory particles, which likely require acidification to release their Os load. Os Isotopes in Recent Greenland Snow. We further investigated the reproducibility of Os data using JR and the NP mixture in fresh snow from Summit, Greenland collected in 2009, 2014, and 2017 (Table 3). We find that the 2009 snow sample (unacidified) gives c(Os) and R(187Os/188Os) ratios that are identical within the stated precision for oxidation using JR and the NP mixture, followed by distillation (Method-I). A comparison of NP and Methods-I and -II was made using the snow collected in 2014 (unacidified). Intriguingly, when duplicate analyses were performed using Method-I, this sample gives c(Os) and R(187Os/188Os) ratios that are not within the precision. Processing of this sample through Method-II yields c(Os) and isotope ratio that overlap with the above two results. Method-II, using JR and NP, was used to process the 2017 snow sample, which was acidified using HCl (final concentration = 0.2% v/v), homogenized by shaking for 12 h, and then immediately analyzed. Two aliquots of this sample give widely different c(Os) and isotope compositions (Table 3). These data indicate that the Greenland snow hosts Os-bearing phases that are relatively insoluble in dilute HCl at low temperature. This suggests that long-term storage following acidification or some other procedure that homogenizes the sample without compromising their PGE contents is needed to provide a more homogeneous sample. Implications for the Accretion Rate of Extra-Terrestrial Os. The c(Os) and R(187Os/188Os) ratio of the recent Greenland snow should reflect contributions from cosmic dust/ meteoric smoke, volcanic activity, anthropogenic Os, and continental mineral dust. The average c(Os) of recent snow = 0.459 ± 0.018 fg g−1 and R(187Os/188Os) = 0.264 ± 0.026 (C.I. = 95%; N = 3). In comparison, the c(Os) and isotope composition of the NEEM firn is 0.98 ± 0.55 fg g−1 and 0.449 ± 0.046 (C.I. = 95%; N = 4), respectively. Clearly, the Os isotope ratio of recent snow lies in between cosmic/volcanic (R(187Os/188Os) ≈ 0.13)10,11 or anthropogenic (R(187Os/188Os) ≈ 0.15)24,25 endmembers on one hand and the continental mineral dust (R(187Os/188Os) = 1.40 ± 0.30)9

tracer and permits accurate determination of c(Os).12,13,23 Two different combinations of oxidants have been successfully used in the past to fully oxidize and equilibrate Os dissolved in ∼50 g of natural waters with Os tracer: (1) 0.5 mL of JR13,14,22 and (2) 1:1 mixture of H2SO4 and H2O2 (total volume = 1 mL).12 These reagents are, however, incompatible with subsequent concentration determination of other PGEs in the same sample aliquot using Method-I or Method-II. Sulfuric acid cannot be readily dried and interferes with PGE separation. Similarly, CrVI interferes with PGE separation and would need to be substantially diluted for the Inductively couple plasma mass spectrometry (ICPMS) preventing PGE determination. We used a mixture of HNO3 and H2O2 as an oxidant and optimized their amounts by processing water samples, whose c(Os) values were determined by oxidation with JR. To compare the extent of tracer-sample equilibration by oxidation, we kept the temperature and duration of oxidation the same as in Chen and Sharma.13 For determining the optimal quantities of reagents, a sample of snow collected from sea ice in the Ross Sea (CORSACS II cruise, Ice Station-03, 70°19′S, 177°07′W) previously reported in Chen et al.13 was repeatedly analyzed. Table 2 compares c(Os) and R(187Os/188Os) ratios obtained using 0.5 mL of JR with those using different volumes of HNO3 and H2O2. For these experiments, OsO4 was separated by distillation (MethodI). Mixtures of 1 mL of HNO3 with 0.5 mL of H2O2 and 2.5 mL of HNO3 with 1.5 mL of H2O2 reached maximum c(Os) of 0.81 ± 0.07 and 0.80 ± 0.10 fg g−1, matching those obtained using JR. In contrast, lower reagent volumes provided incomplete oxidization of Os species in the Ross Sea snow samples. Although a mixture of 1 mL of HNO3 with 0.5 mL of H2O2 can provide full Os oxidation, we used a mixture of 2.5 mL of HNO3 with 1.5 mL of H2O2 for this study. Seawater is a complex high-salinity solution in which Os is present in multiple oxidation states that frustrated early attempts to accurately determine its concentration. The fact that the NP mixture could be used to accurately measure seawater c(Os) and isotope composition was tested next using GEOTRACES large volume seawater sample collected in 2007 from the Atlantic Ocean (31°40′N, 64°10′W) from a depth of 2000 m (see Chen et al.24). For this experiment, the bromine extraction method12 (Method-II) was used with both the 0.5 mL of JR and 2.5:1.5 NP mixture. We find that seawater c(Os) and isotope composition obtained using the NP mixture are identical to those obtained using the JR (Table 2) and are consistent with our previous results.24 Finally, aliquots of an unacidified meltwater sample of NEEM firn are analyzed using both methods. This sample E

DOI: 10.1021/acs.analchem.8b00150 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry endmember on the other. While the R(187Os/188Os) ratio of the NEEM sample seems to suggest more continental Os contribution, we cannot uniquely interpret this result because the sample does not represent the decontaminated firn core. In the discussion below, we use data from recent snow to investigate the present-day extraterrestrial cosmic dust flux. The relative contributions of cosmic/volcanic/anthropogenic and terrestrial end-members can be investigated by comparing the Os flux at the Summit site to previously estimated global fluxes. With regards to the cosmic dust flux, we note that our sample size corresponds to area-time product of ∼2 × 10−4 m2 yr, which could introduce sampling artifacts if accreting particles follow the size distribution derived from satellite measurements.26 However, the bulk of Os accreted to the earth’s surface is in the form of nanoparticles created due to partial melting of cosmic dust during its passage through the atmosphere.27 This would suggest that sampling artifacts would be substantially reduced for ice-core samples with small areatime product. We estimate the Os flux at Summit to be 0.0579 ± 0.0023 fmol cm−2 yr−1. In comparison, the global anthropogenic Os flux as obtained from analyses of rainwater24 is much higher (= 2.47 fmol cm−2 yr−1). It is harder to estimate Os flux from volcanic activity, which is spatially dispersed and temporally variable in magnitude. An indirect estimate can be made using Ir data from Summit snow collected between 1991 and 1995 when the snow chemistry was impacted from the eruptions of Hekla (Iceland) and Mount Pinatubo (Philippines).28 The estimated Ir flux is 0.13 ± 0.02 fmol cm−2 yr−1. The Os/Ir of the earth’s mantle is ∼1, and at face value the Ir flux seems to be a good proxy for Os flux. However, Os and Ir can fractionate from each other during volcanic eruptions and atmospheric transport as there is some evidence that Ir may form volatile IrF6 (bp = 53.6 °C) (e.g., Zoller et al.29) and could be more readily transported over Os.30,31 Moreover, our snow collection seasons did not follow significant global volcanic activity. The Os flux at Summit from volcanic input should therefore be ≪0.13 fmol cm−2 yr−1 during times when our samples were collected. In contrast, the global Os flux from cosmic dust is ∼0.007 fmol cm−2 yr−1 and could potentially be as high as 0.06 fmol cm−2 yr−1 when Os from meteoric smoke is added.32 The Os flux at Summit therefore appears to be within the range of estimated cosmic dust flux. It is over 100 times less than the estimated anthropogenic Os flux. The most straightforward interpretation of these data would be that little anthropogenic Os makes it to central Greenland. This inference is consistent with the observation that anthropogenic Pt transport to Summit is orders of magnitude less than that inferred from smelting and production activities.28 Assuming that Summit received insignificant volcanic Os during collection times, we can use measured Os isotopes of the snow to evaluate cosmic and continental dust flux. The Os isotope ratio in snow (snow) is given by

the fraction of continental dust calculated from our data is XCOs = 0.11+0.06 −0.04, corresponding to a ROs value of 8.5. That is, the cosmic Os flux at Summit is over eight times the continental Os flux. The estimated cosmic flux = 0.0579 × 0.89 = 0.0518 ± 0.0040 fmol cm−2 yr−1. Integrated over the surface of the Earth the extra-terrestrial Os accretion rate is 264 ± 21 mol yr−1. This result is within the estimated range of extra-terrestrial Os accretion rate close to the present-day (38 and 320 mol yr−1).32

■

CONCLUSIONS We describe a procedure to precisely measure Os isotopes in polar snow/ice by oxidizing it in a mixture of HNO3 and H2O2 at 300 °C. This selection of reagents is critical toward the development of chemical procedures to separate and measure other platinum metals using the same sample from which Os has been extracted. We separate Os by distillation in a quartz glass apparatus (Method-I) and by solvent extraction using liquid Br2 (Method-II). Both methods give comparable blanks, but the Os yield from distillation is ∼60% higher. However, this advantage of Method-I is offset by the time needed to clean glassware. We find that (a) high-temperature outgassing of Pt filaments removes Re from the filament and reduces its blank, (b) a Pt−Pt double filament geometry provides consistently high ionization efficiency (∼10%), and (c) cleaning of filaments with distilled water and baking of the ion source are needed to remove organic interference. Using this procedure, we analyzed recent snow collected from Summit, Greenland. The c(Os) and R(187Os/188Os) ratio of the snow indicate that the total amount of extra-terrestrial Os accreted to the Earth is 264 ± 21 mol yr−1. This study opens up a new chapter in fully studying changes in acrreation of extraterrestrial matter over last several hundred thousand years at very high resolution.

■

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.8b00150. Details of purification of nitric acid, purification of liquid bromine, cleaning of Pt filament, determination of Os concentration by isotope dilution, data reduction, blank determination, and procedural blank correction. (PDF)

■

*E-mail: [email protected]. Fax: 603-646-3922. ORCID

Mukul Sharma: 0000-0003-4328-3090 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank C. Han and S. Hong from the Korea Polar Research Institute and Inha University and the NEEM project drilling campaign for providing us with the NEEM firn. We also thank K. Graeter, G. Lewis, G. Wong, and J. Severinghaus for sampling snow at Summit. We are grateful to US GEOTRACES program (E. Boyle, K. Bruland, and G. Smith) for the Atlantic Ocean deep seawater sample and CORSACS cruise (P. Sedgwick and C. Chen) for the Ross Sea snow. The algorithm to correct for contributions from 17O and 18O from OsO3

where Γi is R(187Os/188Os) of i = Continent (C), Cosmic (D) C C C D D XOs = J188 /(J188 + J188 ) = 1/(1 + J188 /J C ) Os Os Os Os 188Os i 188 where J188 Os from reservoir i = Os represents flux of Continent (C), Cosmic (D). ΓC = 1.40 ± 0.309 and ΓD = 0.13,10,11 and thus the only variable governing the Os isotope D J188

Os

C J188

Os

=

1 C XOs

AUTHOR INFORMATION

Corresponding Author

C C Γsnow = XOs ·ΓC + (1 − XOs ) ·ΓD

composition of the snow is ROs ≡

ASSOCIATED CONTENT

S Supporting Information *

− 1. On average, F

DOI: 10.1021/acs.analchem.8b00150 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(28) Gabrielli, P.; Barbante, C.; Plane, J. M. C.; Boutron, C. F.; Jaffrezo, J. L.; Mather, T. A.; Stenni, B.; Gaspari, V.; Cozzi, G.; Ferrari, C.; Cescon, P. Chem. Geol. 2008, 255, 78−86. (29) Zoller, W. H.; Parrington, J. R.; Kotra, J. M. P. Science 1983, 222, 1118−1121. (30) Krähenbühl, U.; Geissbühler, M.; Bühler, F.; Eberhardt, P.; Finnegan, D. L. Earth Planet. Sci. Lett. 1992, 110, 95−98. (31) Yudovskaya, M. A.; Tessalina, S.; Distler, V. V.; Chaplygin, I. V.; Chugaev, A. V.; Dikov, Y. P. Chem. Geol. 2008, 248, 318−341. (32) Sharma, M.; Rosenberg, E. J.; Butterfield, D. A. Geochim. Cosmochim. Acta 2007, 71, 4655−4667.

species and from mass fractionation was developed by Rasmus Andreasen. Amy Jurewicz drew our attention to Ellingham diagrams and provided deep insights into how to use them. We are grateful to B. Peucker-Ehrebrink and two anonymous reviewers for insightful comments that helped in improving this paper. This work was supported by NSF OPP-1417395.

■

REFERENCES

(1) Gabrielli, P.; Barbante, C.; Plane, J. M. C.; Varga, A.; Hong, S.; Cozzi, G.; Gaspari, V.; Planchon, F. A. M.; Cairns, W.; Ferrari, C.; Crutzen, P.; Cescon, P.; Boutron, C. F. Nature 2004, 432, 1011−1014. (2) Karner, D. B.; Levine, J.; Muller, R. A.; Asaro, F.; Ram, M.; Stolz, M. R. Geochim. Cosmochim. Acta 2003, 67, 751−763. (3) Gabrielli, P.; Plane, J. M. C.; Boutron, C. F.; Hong, S. M.; Cozzi, G.; Cescon, P.; Ferrari, C.; Crutzen, P. J.; Petit, J. R.; Lipenkov, V. Y.; Barbante, C. Earth Planet. Sci. Lett. 2006, 250, 459−469. (4) Petaev, M. I.; Huang, S. C.; Jacobsen, S. B.; Zindler, A. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 12917−12920. (5) Love, S. G.; Brownlee, D. E. Science 1993, 262, 550−553. (6) Lal, D.; Jull, A. J. T. Geochim. Cosmochim. Acta 2003, 67, 4925− 4933. (7) Brooke, J. S. A.; Feng, W. H.; Carrillo-Sanchez, J. D.; Mann, G. W.; James, A. D.; Bardeen, C. G.; Plane, J. M. C. J. Geophys. Res.-Atmos 2017, 122, 11112−11130. (8) Peucker-Ehrenbrink, B.; Ravizza, G.; Winckler, G. Elements 2016, 12, 191−196. (9) Peucker-Ehrenbrink, B.; Jahn, B. M. Geochem., Geophys., Geosyst. 2001, 2, 26. (10) Sharma, M. In Handbook of Environmental Isotope Geochemistry; Baskaran, M., Ed.; Springer Berlin Heidelberg: Berlin, 2012; Vol. I, pp 205−227. (11) Horan, M. F.; Walker, R. J.; Morgan, J. W.; Grossman, J. N.; Rubin, A. E. Chem. Geol. 2003, 196, 27−42. (12) Paul, M.; Reisberg, L.; Vigier, N. Chem. Geol. 2009, 258, 136− 144. (13) Chen, C.; Sharma, M. Anal. Chem. 2009, 81, 5400−5406. (14) Levasseur, S.; Birck, J. L.; Allegre, C. J. Science 1998, 282, 272− 274. (15) Gabrielli, P.; Varga, A.; Barbante, C.; Boutron, C.; Cozzi, G.; Gaspari, V.; Planchon, F.; Cairns, W.; Hong, S. M.; Ferrari, C.; Capodaglio, G. J. Anal. At. Spectrom. 2004, 19, 831−837. (16) Soyol-Erdene, T. O.; Huh, Y.; Hong, S.; Hwang, H. J.; Hur, S. D. Bull. Korean Chem. Soc. 2011, 32, 2105−2108. (17) Mattinson, J. Anal. Chem. 1972, 44, 1715−1716. (18) Heumann, K. G. In Inorganic Mass Spectrometry; Adams, F., Gijbels, R., Van Grieken, R., Eds.; J. Wiley & Sons, 1988; pp 301−376. (19) Creaser, R. A.; Papanastassiou, D. A.; Wasserburg, G. J. Geochim. Cosmochim. Acta 1991, 55, 397−401. (20) Volkening, J.; Walczyk, T.; Heumann, K. G. Int. J. Mass Spectrom. Ion Processes 1991, 105, 147−159. (21) Jun, S. J.; Hong, S. B.; Hur, S. D.; Lee, J.; Kang, J.-H.; Hwang, H.; Chung, J. W.; Jung, H. J.; Han, C.; Hong, S. Ocean Polar Res. 2014, 36, 13−24. (22) Birck, J. L.; Barman, M. R.; Capmas, F. Geostand. Geoanal. Res. 1997, 21, 19−27. (23) Peucker-Ehrenbrink, B.; Sharma, M.; Reisberg, L. Eos, Transactions American Geophysical Union 2013, 94, 73−73. (24) Chen, C.; Sedwick, P. N.; Sharma, M. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 7724−7728. (25) Poirier, A.; Gariepy, C. Environ. Sci. Technol. 2005, 39, 4431− 4434. (26) Peucker-Ehrenbrink, B.; Ravizza, G. Geochim. Cosmochim. Acta 2000, 64, 1965−1970. (27) Chen, C.; Taylor, S.; Sharma, M. In Lunar and Planetary Science XXXVI: Papers Presented at the Thirty-Sixth Lunar and Planetary Science Conference: March 14−18, 2005; Lunar and Planetary Institute: League City, TX, 2005. G

DOI: 10.1021/acs.analchem.8b00150 Anal. Chem. XXXX, XXX, XXX−XXX