Tungsten isotopes in bulk meteorites and their inclusions—Implications for processing of presolar components in the solar protoplanetary disk

Abstract

We present high precision, low- and high-resolution tungsten isotope measurements of iron meteorites Cape York (IIIAB), Rhine Villa (IIIE), Bendego (IC), and the IVB iron meteorites Tlacotepec, Skookum, and Weaver Mountains, as well as CI chondrite Ivuna, a CV3 chondrite refractory inclusion (CAI BE), and terrestrial standards. Our high precision tungsten isotope data show that the distribution of the rare p-process nuclide ¹⁸⁰W is homogeneous among chondrites, iron meteorites, and the refractory inclusion. One exception to this pattern is the IVB iron meteorite group, which displays variable excesses relative to the terrestrial standard, possibly related to decay of rare ¹⁸⁴Os. Such anomalies are not the result of analytical artifacts and cannot be caused by sampling of a protoplanetary disk characterized by p-process isotope heterogeneity. In contrast, we find that ¹⁸³W is variable due to a nucleosynthetic s-process deficit/r-process excess among chondrites and iron meteorites. This variability supports the widespread nucleosynthetic s/r-process heterogeneity in the protoplanetary disk inferred from other isotope systems and we show that W and Ni isotope variability is correlated. Correlated isotope heterogeneity for elements of distinct nucleosynthetic origin (¹⁸³W and ⁵⁸Ni) is best explained by thermal processing in the protoplanetary disk during which thermally labile carrier phases are unmixed by vaporization thereby imparting isotope anomalies on the residual processed reservoir.

Introduction

The process of planetesimal growth and planet formation in the nascent solar system was governed by accretion of dust into km-sized bodies and the subsequent differentiation of large bodies into a metallic core and silicate mantle (Kleine et al. 2002; Yin et al. 2002; Scherstén et al. 2006; Kruijer et al. 2013; Wittig et al. 2013). Constraints on the timescales of planetesimal melting and differentiation are provided by the short-lived ¹⁸²Hf-¹⁸²W chronometer (T_1/2 = 8.90 ± 0.09 Myr; Vockenhuber et al. 2004). This decay scheme, comprising a lithophile parent and a siderophile daughter, effectively constrains metal-silicate differentiation events and has been applied to a range of meteoritic materials, with a focus on iron meteorites (Kleine et al. 2005; Markowski et al. 2006a; Scherstén et al. 2006; Qin et al. 2008a; Kruijer et al. 2013; Wittig et al. 2013). Its use relies on the critical assumption of isotopic homogeneity in the protoplanetary disk and among solids formed within the first few million years after initial collapse of the protosolar molecular cloud core.

Knowledge of the distribution of tungsten isotopes requires the assessment of their origin and introduction into the protosolar system. Tungsten isotopes are produced in the terminal phases of stellar evolution during the asymptotic giant branch phase of low and intermediate mass stars and during supernova nucleosynthesis. The two main processes responsible for W isotope synthesis are the slow and rapid neutron capture processes, the s-, and r-process, respectively (Burbidge et al. 1957; Cameron 1957; Arlandini et al. 1999; Ávila et al. 2011). Aside from ¹⁸⁰W, which is a p-process isotope produced in highly energetic stellar environments by photodisintegration and/or proton capture, all W isotopes have contributions from both the s-, and r-process (Arlandini et al. 1999; Vockenhuber et al. 2007). Altogether, W isotopes provide a potential means for studying the distribution of p-, s-, and r-process carriers in the early solar system (Kleine et al. 2008; Qin et al. 2008a; Burkhardt et al. 2012b; Wittig et al. 2013).

This is important, as variability in the abundances of key isotopes is ubiquitous in meteoritic material (e.g., Trinquier et al. 2009; Burkhardt et al. 2012a) suggesting a heterogeneous distribution of the carriers of nucleosynthetic isotope anomalies within the protoplanetary disk. The initial abundance of short-lived radionuclides ²⁶Al, ⁵³Mn, and ¹⁰Be has been shown to vary in the solid forming regions of the disk (Nyquist et al. 2009; Larsen et al. 2011; Wielandt et al. 2012; Schiller et al. 2015). For example, a recent high-resolution comparison of U-corrected Pb-Pb and ²⁶Al-²⁶Mg ages for three angrite meterorites supports a model of initial disk ²⁶Al heterogeneity. Indeed, Schiller et al. (2015) showed that ²⁶Al-²⁶Mg ages for three rapidly cooled angrites are systematically younger by ~1.5 Myr relative to their assumption free absolute ages, requiring that the angrite parent body formed from material with an initial ²⁶Al/²⁷Al of ${1.33}_{-0.18}^{+0.21}\times {10}^{-5}$. Furthermore, there is clear evidence that stable nonradiogenic isotopes, particularly the neutron-rich iron group nuclei, are not homogeneous at bulk planetesimal and individual grain scales (Andreasen and Sharma 2006; Trinquier et al. 2007, 2009; Larsen et al. 2011; Burkhardt et al. 2012a). Such heterogeneity could reflect incomplete mixing of freshly synthesized stellar ejecta that were injected (e.g., Ouellette et al. 2009, 2010) into the forming solar system with its associated protoplanetary disk. Alternatively, it could result from disk processing that generated isotopically distinct reservoirs from an initially well-mixed protoplanetary disk (Trinquier et al. 2009).

Variability in the isotopes of tungsten has been shown for both calcium-aluminum-rich inclusions (CAIs; Burkhardt et al. 2008, 2012a; Kruijer et al. 2014) and several groups of iron meteorites (Qin et al. 2008a; Schulz et al. 2013; Wittig et al. 2013; Cook et al. 2014; Peters et al. 2014). The protoplanetary disk may thus have been characterized by W isotope heterogeneity for p-, s-, and r-process nuclides at the scale of parent bodies and down to individual refractory inclusions.

We address this topic by the stable ¹⁸⁰W and ¹⁸³W isotopes in a twofold study employing novel analytical techniques, including both low- and high-resolution multiple-collector inductively coupled plasma-mass spectrometry (MC-ICPMS). It has been shown that low-resolution measurements could be affected by molecular interferences, imparting spurious anomalies on W isotope data (Holst et al. 2011) and thereby possibly affecting ε¹⁸⁰W and ε¹⁸³W. The application of high-resolution mass spectrometry may thus be required to obtain accurate ¹⁸⁰W and ¹⁸³W results.

We report robust high-precision, high-resolution data for ¹⁸⁰W/¹⁸⁴W of a suite of well-studied iron meteorites, chondrites, and a refractory inclusion to ascertain the degree of homogeneity of this rare heavy p-process nuclide in the young protoplanetary disk. Secondly, we report high-resolution measurements of ¹⁸³W/¹⁸⁴W, to assess the s- and r-process variability in tungsten in the formation region of the investigated material. Importantly, if tungsten isotope ratios vary due to nucleosynthetic s- and r-process anomalies, internal normalization to ¹⁸⁶W/¹⁸⁴W or ¹⁸⁶W/¹⁸³W will impact chronology based on normalized ¹⁸²W data. Thus, to construct a robust and accurate chronology of metal-silicate differentiation in the early solar system, it is important to assess the presence of nucleosynthetic variability in W isotopes.

Samples and Analytical Methods

Iron Meteorites

Six samples were chosen to represent a variety of iron meteorite types, some of which have previously been characterized for their W isotope composition including ¹⁸⁰W (Schulz et al. 2013). These included Cape York (IIIAB), Rhine Villa (IIIE), and Bendego (IC) as well as the IVB iron meteorites Tlacotepec, Skookum, and Weaver Mountains, all of which were chosen for direct comparison to the data set of Schulz et al. (2013). Furthermore, we provide ¹⁸⁰W measurements of a bulk CI chondrite Ivuna and a CAI from CV3 chondrite NWA 8722. Sample preparation methods were modified from previously established procedures (Kraus et al. 1955; Faix et al. 1981; Salters and Hart 1991; Horan et al. 1998; Münker et al. 2001; Kleine et al. 2004, 2008).

Iron meteorite chips of 0.63–2.9 g were cut using diamond coated saw-blades and polished with tungsten-free abrasive paper and diamond-coated dental drills. Prior to dissolution, sample chips were cleaned successively in ethanol and 0.05 M HNO₃ in an ultrasonic bath. The W purification procedure for iron meteorites was modified from Kleine et al. (2004, 2008). All hydrofluoric acid used in the purification protocol was pre-cleaned on AG1 anion resin at a concentration of 4 M, which was found to reduce the W reagent blank by a factor of ~25. Initial sample digestion was achieved in 10–25 mL 6 M HCl + 0.06 M HF at 130 °C. Samples were then dissolved in 4:1 HNO₃:H₂O₂ to remove organics and Os as a volatile oxide. Following this step, full dissolution was achieved in 6 M HCl + 0.06 M HF and each sample was taken up in 1 M HF + 0.1 M HNO₃. This volume was then loaded on pre-cleaned and conditioned 20 mL BioRad polypropylene columns containing AG1-X8, 200–400 mesh. Sample matrix was subsequently eluted in 100 mL 1 M HF + 0.1 M HNO₃, followed by elution of W and other high field strength elements (HFSE) in 100 mL 6 M HNO₃ + 0.2 M HF. The HFSE fraction was dried and re-dissolved in 2 mL 1 M HF + 0.1 M HNO₃. Here, it was critical to flux the sample at 100–130 °C to re-dissolve any sample tungsten adsorbed on the vial surface. Once fluxed, this cut was loaded onto a 1 mL anion column. Matrix elements were eluted with 5 mL 1 M HF + 0.1 M HNO₃ and HFSE such as Zr and Hf, the latter a direct isobaric interference, were eluted in 2 mL 6 M HCl + 0.01 M HF. Finally, the purified W was recovered in 8 mL 6 M HCl + 1 M HF.

To attain a sufficiently low interference level of ¹⁸⁰Hf on ¹⁸⁰W, a third column step was necessary for all samples. Prior to this step, samples were converted to NO₃⁻ form, and then dissolved in 1 M HF + 0.1 M HNO₃. Again, hot plate fluxing at 130 °C was necessary before loading onto 1 mL anion columns. The elution protocol was identical to the second column step. For some samples, the third column step was repeated to reduce the ¹⁸⁰Hf/¹⁸⁰W to ≤10⁻³.

Silicate Samples

For chondrite and CAI samples of 1–3 g, as well as NIST 3163 W and NIST SRM 361 standards and a BCR-2 rock standard, a new W purification protocol designed for silicate matrices was applied. This was modified from Fritz et al. (1961) and Strelow et al. (1972) and utilized a cation matrix separation step followed by an anion purification step. The advantage of this approach is that W is eluted directly from the cation column, thereby avoiding the potential loss of W due to column saturation. This enables larger samples to be processed on fewer parallel columns, as the cation column efficiently retains most major elements. We used AG50W-X8, 200–400# and samples were loaded in 4–6 column volumes (c.v.) 0.25 M HNO₃ + 0.1 M HF + 0.1% H₂O₂ depending on sample size. Prior to loading, they were fluxed at 100 °C for 1–2 days to ensure proper oxidation of Cr to Cr³⁺. At this oxidation state, Cr is strongly adsorbed on the cation resin and hence effectively separated from W. Residual adsorbed tungsten and other HFSE were eluted in 2.5 c.v. 0.1 M HF. After collection of W, adsorbed Fe was eluted in 2.5 c.v. 1 M HF followed by removal of matrix elements with 10 c.v. 6 M HCl.

The second column consisted of pre-cleaned and conditioned AG1-X4, 200–400 mesh. The converted and fluxed samples were loaded in 1–4 mL 1 M HF. Aluminum and other matrix elements were eluted in 1 M HF, followed by elution of Ti, Hf, and Zr in 2 M HCl + 0.1% H₂O₂. Residual Hf was removed with 2 c.v. 6 M HCl + 0.01 M HF followed by W elution in 4 c.v. 6 M HCl + 1 M HF. As with the above protocol, the second column step was repeated 1–3 times to ensure that ¹⁸⁰Hf/¹⁸⁰W ≤10⁻³. The final two column steps used 0.2 mL columns to minimize resin-derived organics and minimize procedural blanks. Following W recovery, all samples were re-dissolved six times with 4:1 concentrated HNO₃: 30% H₂O₂ to remove residual Os and organic molecules. We note that residual organics can cause (1) a significant decrease in the ionization efficiency of W in the plasma source mass spectrometer resulting in variable instrumental mass fractionation and poor standard-sample intensity matching and (2) differences in uptake rate between sample and standard, also resulting in poor sample-standard signal intensity matching. Moreover, an organic molecular interference has been known to cause significant effects on ¹⁸³W (Kleine et al. 2002, 2004), necessitating reduction in resin-derived and sample-related organics. Total procedural blanks were ~4.5 ng for the most elaborate six-column chemistry, and result from the use of large columns and large quantities of reagents as well as the use of 350 mL Teflon beakers during sample handling. In addition, a substantial part of the procedural blank is caused by the use of steel-jacketed Parr bombs during silicate sample digestion. However, owing to sample sizes containing 450–3000 ng W, coupled with the relatively small magnitude of the potential isotopic anomalies, this blank level has a negligible impact on the results, given the uncertainties of our measurements.

The Elemental Hf/W Ratio of CAI BE

The Hf/W ratio of the CV chondrite refractory inclusion dubbed CAI BE was determined by doping with a mixed ¹⁸⁰Hf-¹⁸⁶W tracer as described in Holst et al. (2013). A doped 0.3% aliquot of the bulk sample was passed over a cation column (1 mL AG50W-X8) to remove most matrix elements. It was then converted to NO₃⁻ form and fluxed on a hotplate before analysis. The measurements were performed on a ThermoFisher X-series II quadrupole ICPMS. In the same run, we conducted a tracer calibration against Alfa Aesar solution standards of known concentration and isotopic composition to obtain the elemental Hf/W ratio in the mixed tracer. The tracer calibration is necessary as the tracer composition changes with time due to the variable behavior of Hf and W when stored in solution.

Mass Spectrometry

Samples were fluxed on a hotplate at 130 °C in 0.5 M HNO₃ + 0.1 M HF and run on the ThermoFisher Neptune Plus at the Centre for Star and Planet Formation in Copenhagen, using a Cetac Aridus II desolvating nebulizer sample introduction system and combining a sampler Jet Cone with the skimmer X-cone. The typical sample aspiration rate for this introduction system was ~0.05 mL min⁻¹. The large sample sizes permitted the measurement of each sample in both low- and high-resolution mode.

Tungsten isotopes were measured in static mode using seven Faraday collectors with the following configuration: ¹⁸³W in the axial collector; ¹⁷⁸Hf, ¹⁸⁰W, and ¹⁸²W in the L3, L2, and L1 collectors on the low-mass side of the axial Faraday; and ¹⁸⁴W, ¹⁸⁶W, and ¹⁸⁸Os in the H1, H2, and H3 collectors on the high mass side of the axial Faraday. All masses were measured using Faraday detectors connected to amplifiers with 10¹¹ Ω feedback resistors, with the exception of ¹⁷⁸Hf and ¹⁸⁰W, which were measured using 10¹² Ω feedback resistors. ¹⁷⁸Hf was monitored to correct for ¹⁸⁰Hf interference on ¹⁸⁰W. Sample-standard bracketing using the NIST 3163 W solution standard was applied to correct for instrumental mass fractionation using the exponential law, and data were acquired in both low- and high-resolution modes for each sample. Total sensitivity of the instrument for W in low- and high-resolution mode was 1200 and 120 V/ppm, respectively. Low-resolution measurements were acquired based on techniques described in Holst et al. (2013). Measurements conducted in high-resolution mode were performed on the low-mass side of the peak, at a position allowing for an effective mass resolving power (M/ΔM) of ~4500. Samples and standards were analyzed with signal intensities matched to better than 5%. The typical intensity on mass 180 (¹⁸⁰W+¹⁸⁰Hf) was 25–60 mV. Isobaric interferences from ^184,186Os were monitored on ¹⁸⁸Os, and the ¹⁸⁴Os/¹⁸⁴W ratio was ~1 × 10⁻⁶ for the high-Os IVB iron meteorite samples. The Os interference was substantially less for silicate samples and was thus negligible in all cases. Therefore, only the interference from ¹⁸⁰Hf on ¹⁸⁰W was significant and required correction. Although our ion exchange protocol reduced the ¹⁸⁰Hf/¹⁸⁰W to ≤10⁻³, we are experienced minute levels of a few picograms of Hf-blank addition during sample preparation immediately prior to mass spectrometry. To account for this, in addition to matching the concentration of W in the bracketing standard with the sample, the former was also doped with Hf standard solution to a level matching that of the sample, so that it best reflected the measurement conditions of unknowns.

Each analysis comprised a total of 1678 s of baseline measurements (obtained on peak) and 839 s of data acquisition (100 scans integrated over 8.39 s) interspaced by 120 s washouts using the Cetac QuickWash accessory module. All data reduction was conducted off-line using the freely distributed Iolite data reduction package, which runs within Igor Pro (Paton et al. 2011). Baseline intensities were interpolated using a smoothed cubic spline, as was instrumental drift with time. Typical baseline levels were 0.5–2 mV of total tungsten. Iolite’s smooth spline auto choice was used in all cases, which determines a theoretically optimal degree of smoothing based on variability in the reference standard throughout an analytical session, which corresponds typically to 12–24 h of continuous measurement without adjusting the instrument’s tuning parameters. For each analysis, the mean and standard error of the measured ratios were calculated, using a 2 SD threshold outlier rejection. Individual sample analyses were combined to produce an average, weighted by the propagated uncertainties of individual analyses. The reported analytical uncertainties include the propagated 2 SD error on the bracketing standard for each analytical session. W isotope data are reported in the ε notation as deviations from the NIST 3163 W standard in parts per 10⁴:

$$\left\{{\text{ε}}^{\text{18*}}\text{W}={\left({}^{\text{18*}}\text{W}/{}^{184}\text{W}{}_{\text{std}}\right)}_{\text{sample}}/\left({}^{\text{18*}}\text{W}/{}^{184}\text{W}{}_{\text{std}}\right)-1\right\}*{10}^4$$ (1)

$$\left\{{\text{ε}}^{\text{18*}}\text{W}={\left({}^{\text{18*}}\text{W}/{}^{183}\text{W}\right)}_{\text{sample}}/\left({}^{\text{18*}}\text{W}/{}^{183}\text{W}{}_{\text{std}}\right)-1\right\}*{10}^4$$ (2)

and were internally normalized using the exponential law. As shown in equations (1) and (2) this was performed separately using both ¹⁸⁶W/¹⁸⁴W = 0.92767 and ¹⁸⁶W/¹⁸³W = 1.98594 (Völkening et al. 1991) to test for isotopic anomalies on the normalizing isotopes. The two normalization schemes are denoted (6/4) and (6/3), respectively. For (6/3) normalized data, refer to Table 1. We prefer (6/4) normalization due to the potential presence of analytical artifacts (see Discussion section) affecting the measured ¹⁸³W, causing shifts in (6/3) normalized data.

Table 1.

Tungsten isotope data for iron meteorites, chondrites, inclusion CAI BE, and terrestrial standard materials

Sample		ε180W(6/4)	ε182W(6/4)	ε183W(6/4)	ε180W(6/3)	ε182W(6/3)	ε184W(6/3)	n
Cape York (IIIAB)	LRa	1.27 ± 0.4	−3.37 ± 0.08	−0.08 ± 0.05	1.54 ± 0.4	−3.25 ± 0.05	0.05 ± 0.03	4
	HRb	0.40 ± 1.0	−3.40 ± 0.12	−0.12 ± 0.08	0.43 ± 1.3	−3.20 ± 0.13	0.08 ± 0.05	5
Rhine Villa (IIIE)	LR	1.08 ± 0.7	−3.67 ± 0.10	−0.09 ± 0.07	−0.72 ± 0.5	−3.53 ± 0.09	0.06 ± 0.04	5
	HR	−0.03 ± 1.4	−3.63 ± 0.10	−0.11 ± 0.07	0.29 ± 1.2	−3.50 ± 0.09	0.07 ± 0.04	6
Bendego (IC)	LR	−0.81 ± 1.3	−4.16 ± 0.09	−0.16 ± 0.11	−0.20 ± 1.2	−3.91 ± 0.11	0.11 ± 0.06	10
	HR	0.14 ± 1.6	−4.11 ± 0.06	−0.15 ± 0.07	0.48 ± 1.1	−3.90 ± 0.11	0.10 ± 0.05	8
Tlacotepec (IVB)	LR	3.83 ± 0.8	−3.85 ± 0.10	0.02 ± 0.06	3.82 ± 0.8	−3.87 ± 0.07	−0.01 ± 0.04	5
	HR	5.80 ± 2.0	−3.88 ± 0.11	0.03 ± 0.09	5.50 ± 1.8	−3.91 ± 0.05	−0.02 ± 0.05	3
Weaver Mts (IVB)	LR	1.90 ± 1.3	−3.13 ± 0.11	0.05 ± 0.09	1.90 ± 1.2	−3.19 ± 0.06	−0.05 ± 0.08	9
	HR	2.20 ± 1.2	−3.27 ± 0.19	0.00 ± 0.08	2.40 ± 1.2	−3.19 ± 0.15	0.00 ± 0.05	6
Skookum (IVB)	LR	2.66 ± 0.8	−3.29 ± 0.08	0.25 ± 0.06	2.31 ± 1.2	−3.64 ± 0.07	−0.16 ± 0.04	10
	HR	2.94 ± 1.1	−3.27 ± 0.10	0.04 ± 0.08	2.92 ± 0.8	−3.32 ± 0.08	−0.03 ± 0.06	6
Ivuna (CI)	LR	−0.12 ± 2.3	−1.77 ± 0.13	0.39 ± 0.12	−0.69 ± 2.2	−2.16 ± 0.15	−0.26 ± 0.08	1
	HR	0.38 ± 2.2	−1.98 ± 0.23	0.36 ± 0.12	−0.37 ± 2.3	−2.50 ± 0.31	−0.24 ± 0.08	1
CAI BE (CV incl.)	LR	0.35 ± 1.7	−1.50 ± 0.13	1.00 ± 0.11	−1.31 ± 1.8	−2.79 ± 0.10	−0.67 ± 0.07	1
	HR	1.34 ± 1.4	−2.21 ± 0.30	0.51 ± 0.17	0.30 ± 1.5	−2.92 ± 0.25	−0.34 ± 0.11	1
Allende (CV3)c	LR	–	−1.84 ± 0.34	0.20 ± 0.12	–	−2.06 ± 0.20	−0.13 ± 0.18	5
NIST 3163	LR	−0.08 ± 1.0	−0.04 ± 0.05	−0.02 ± 0.03	−0.16 ± 0.9	0.01 ± 0.05	0.02 ± 0.02	9
	HR	0.18 ± 1.3	0.05 ± 0.08	0.03 ± 0.07	0.08 ± 1.3	0.01 ± 0.09	−0.03 ± 0.04	10
BCR-2	LR	−0.03 ± 0.8	0.00 ± 0.09	−0.09 ± 0.04	0.23 ± 0.7	0.13 ± 0.10	0.06 ± 0.05	10
	HR	0.22 ± 1.5	0.06 ± 0.11	−0.06 ± 0.08	0.42 ± 1.8	0.10 ± 0.08	0.04 ± 0.06	10
NIST SRM 361	LR	−0.80 ± 1.3	0.01 ± 0.16	−0.04 ± 0.06	−0.54 ± 1.2	0.09 ± 0.08	0.03 ± 0.04	17
	HR	−0.50 ± 1.3	0.05 ± 0.14	−0.05 ± 0.08	−0.30 ± 1.2	0.08 ± 0.08	0.04 ± 0.05	10

^aLow-resolution mode.

^bHigh-resolution mode MC-ICPMS data. The ε notation is the deviation from the terrestrial standard in parts per 10,000. (6/4) designates data that are normalized to ¹⁸⁶W/¹⁸⁴W, whereas (6/3) are for normalization to ¹⁸⁶W/¹⁸³W.

^cData from Holst et al. (2013). Quoted values are the weighted means of multiple sample analyses during one analytical session and errors for samples with n > 1 are 2 SE including the propagated 2 SD of the bracketing standard for each session. Errors on samples with n = 1 are 2 SE.

Results

Tungsten isotope data for iron meteorites, chondrites, and CAI BE are summarized below and presented in Table 1 and Fig. 1.

Fig. 1.

Tungsten isotope data for iron meteorites, Ivuna, Allende (from Holst et al. 2013), and CAI BE as well as rock, steel, and processed solution standards. Top and bottom panels show low- and high-resolution data, respectively. Errors are 2 SE for samples with n > 1 and propagated with the 2 SD uncertainty on the bracketing standard for each analysis.

Low-Resolution Data

For ε¹⁸⁰W, there is no resolved effect of using (6/3) normalization instead of (6/4) and all measurements overlap for the two different normalizations. For analyses performed in low-resolution mode, there is an apparent spread of ε¹⁸⁰W with a negative anomaly for Rhine Villa and positive anomalies for Cape York and all three measured IVB iron meteorites. The positive ε¹⁸⁰W for Cape York is intermediate between the terrestrial standard value and the IVB iron group that span a range in ε¹⁸⁰W of ~2–6. The CI chondrite Ivuna and the CV chondrite refractory inclusion CAI BE both have ε¹⁸⁰W overlapping with the terrestrial value. Also, all three processed standards, NIST 3163, BCR-2, and SRM 361, overlap within uncertainty with unprocessed NIST 3163.

As with ε¹⁸⁰W, the ε¹⁸²W of all processed standards are in good agreement with the unprocessed terrestrial standard. The ε¹⁸²W of the iron meteorites range from −4.1 to −3.1 with Bendego and Tlacotepec having the most negative anomalies. Ivuna and CAI BE have ε¹⁸²W between −1.5 and −1.8 for (6/4) normalization and between −2.1 and −2.8 for (6/3) normalization. We observe this clear discrepancy only for the two silicate samples and the cause is explained in the Discussion section.

The ε¹⁸³W of processed standards also overlap with the terrestrial value. In contrast, there is a clear negative ε¹⁸³W (6/4) for Cape York, Rhine Villa, and Bendego, corresponding to a positive ε¹⁸⁴W (6/3). We explore this minor deficit relative to the standard in The Distribution of ¹⁸³W section. Note that for Skookum ε¹⁸³W = 0.25 ± 0.06, which is most likely caused by an interference on mass 183 (see Organic Interferences section and cf. Kleine et al. 2002, 2004; Holst et al. 2011). Ivuna has ε¹⁸³W of 0.36 ± 0.12, whereas CAI BE ε¹⁸³W = 1.00 ± 0.11.

We include W isotope data measured in low-resolution mode for Allende from Holst et al. (2013), as these data were obtained under identical analytical conditions and substantiate our conclusions regarding ¹⁸³W variability (Table 1).

High-Resolution Data

As for low-resolution data, all ε¹⁸⁰W in high-resolution mode is consistent when using (6/3) and (6/4) normalization. In contrast to the low-resolution data, the high-resolution ε¹⁸⁰W of Cape York and Rhine Villa is not resolved from the terrestrial standard. The ε¹⁸⁰W values for Cape York and Rhine Villa are 0.40 ± 1.0 and 0.03 ± 1.4, which is in excellent agreement with terrestrial standard. The weighted mean ε¹⁸⁰W of Bendego is also in closer agreement with the terrestrial value when measuring in high-resolution mode. For CAI BE and Ivuna, there is no indication of anomalous ε¹⁸⁰W. Again, we note that all processed standards agree with the terrestrial ε¹⁸⁰W. Interestingly, the general lack of ε¹⁸⁰W anomalies is contrasted by the IVB iron group that in high resolution still displays clearly resolved ε¹⁸⁰W excesses ranging from 2 to 6 ε with Tlacotepec having the highest measured ε¹⁸⁰W of 5.80 ± 2.0.

The measured ε¹⁸²W is identical to the pure W standard for all processed standards. High-resolution data for iron meteorites show the same range of ε¹⁸²W from −4 to −3 ε and each sample measurement is identical in both low- and high-resolution mode. Ivuna and CAI BE both overlap within uncertainty with the average value for chondritic meteorites of −1.9 ± 0.1 (Kleine et al. 2004). However, due to effects on ε¹⁸³W (or ε¹⁸⁴W), the ε¹⁸²W for Cape York, Rhine Villa, Bendego, Ivuna, and CAI BE vary according to the applied normalization scheme. This will be discussed in the Discussion section.

As for ε¹⁸²W, the ε¹⁸³W of processed NIST 3163, SRM 361, and BCR-2 overlap with the unprocessed bracketing standard. Moreover, the IVB irons have terrestrial ε¹⁸³W with no indication of variability. The deficits in ε¹⁸³W for Cape York, Rhine Villa, and Bendego in low resolution are confirmed in high-resolution mode and range between −0.15 and −0.11, whereas Ivuna and CAI BE are positive with ε¹⁸³W of 0.36 ± 0.12 and 0.51 ± 0.17, respectively. Note that CAI BE is less positive in high resolution compared to low resolution (see the Organic Interferences section). The validity and implications of the observed spread in ε¹⁸³W among different types of bulk meteorites are explored in The Distribution of ¹⁸³W section.

Lastly, we determined the Hf/W ratio for CAI BE following the analytical protocol for elemental ratio determination as described by Holst et al. (2013) and found a Hf/W of 0.27 ± 0.016. This corresponds to ~0.3 ε of ingrowth on ¹⁸²W if a canonical initial ¹⁸²Hf/¹⁸⁰Hf of 9.85 × 10⁻⁵ is assumed (Burkhardt et al. 2012b).

Discussion

Reproducibility of W Isotope Data

The reproducibility of our analytical protocols is different for metal and silicate samples. This difference is due to the application of different chromatographic procedures, matrix effects, and lower W concentrations in silicate samples.

The reproducibility of the isotopic measurements for iron meteorites was evaluated based on the 2 SD uncertainty of low- and high-resolution measurements of the NIST SRM 361 Fe-Ni steel standard with tungsten concentrations comparable to that of iron meteorite samples. For Fe-Ni-rich samples in low-resolution mode, the 2 SD is ε¹⁸⁰W (6/4) = 1.3, ε¹⁸²W (6/4) = 0.16, and ε¹⁸³W (6/4) = 0.06. In high-resolution mode it is ε¹⁸⁰W (6/4) = 1.3, ε¹⁸²W (6/4) = 0.14, and ε¹⁸³W (6/4) = 0.08.

For silicate matrices, the long-term 2 SD was assessed from measurements of column processed NIST 3163 and USGS rock standard BCR-2. In low-resolution mode the external reproducibility for ε¹⁸⁰W (6/4) = 1.2, ε¹⁸²W (6/4) = 0.10, and ε¹⁸³W (6/4) = 0.05. In high-resolution mode ε¹⁸⁰W (6/4) = 2.0, ε¹⁸²W (6/4) = 0.14, and ε¹⁸³W (6/4) = 0.11. We note that our 2 SD on ε¹⁸⁰W is comparable to those of recent studies (Schulz et al. 2013; Cook et al. 2014), despite our study including measurement of ¹⁸⁰W in high-resolution of small silicate samples with low W concentrations such as CAI BE (~300 ng W). In addition, the ε¹⁸²W of all iron meteorites and chondritic samples yield values that are consistent with literature data (e.g., Kleine et al. 2005; Scherstén et al. 2006; Qin et al. 2008a; Schulz et al. 2013).

Tungsten isotopic anomalies outside of the estimated reproducibility can have several causes including (1) cosmic ray induced production/burnout during a prolonged meteoroid phase, (2) nucleosynthetic anomalies, or (3) alpha-decay of rare, long-lived ¹⁸⁴Os. As small anomalies in the abundances of nonradiogenic W isotopes help constrain the extent and nature of nucleosynthetic variability in the solar protoplanetary disk it is important to assess their cause.

However, before drawing conclusions from the data set, there are a number of potential analytical artifacts that must be identified and accounted for. Repeat measurements of several standards, including processed synthetic and rock standards as well as a NIST SRM 361 steel standard yield W isotope data that are indistinguishable from the terrestrial value as determined from an unprocessed NIST 3163 solution standard. Nevertheless, deficits and excesses in ε¹⁸³W are observed in bulk meteorites and also, conditions may differ for meteoritic samples relative to standards due to various effects including isobaric interferences, nonkinetic mass fractionation, nuclear charge related odd-even effects (Shirai and Humayun 2011; Willbold et al. 2011; Kruijer et al. 2012), and organic interferences. These potential effects are evaluated before the data are interpreted in terms of early solar system processes.

¹⁸⁰Hf Interference Correction

Apart for IVB irons, all samples show ε¹⁸⁰W values identical to the terrestrial standard, indicating that interference correction for ¹⁸⁰Hf on ¹⁸⁰W does not result in inaccuracies beyond the 2σ reproducibility of our method. The accuracy of the Hf interference correction was empirically tested by doping pure W solutions with variable amounts of Hf. These experiments show that the ¹⁸⁰Hf interference correction is accurate for interference levels of up to ~8000 ppm (Fig. 2), which is substantially beyond the levels encountered in meteorite samples (<3000 ppm). In high-resolution mode all samples, excluding the IVB irons, overlap with the terrestrial ε¹⁸⁰W, thereby validating the interference correction. The inferred accuracy of the interference correction substantiates the observed positive ε¹⁸⁰W anomalies in the IVB iron meteorites.

Fig. 2.

Hf doping experiment of terrestrial NIST 3163 W standard showing the impact of the direct isobaric ¹⁸⁰Hf interference on the final, interference corrected ε¹⁸⁰W ratio. The interference correction is effective to less than 1ε for ¹⁸⁰Hf levels beyond 8000 ppm. Given that no samples have ¹⁸⁰Hf interference higher than 3000 ppm, the interference correction is considered valid for all meteoritic samples.

Instrumental Mass Fractionation Correction

Nonradiogenic isotope anomalies as observed for ε¹⁸³W can potentially arise from improper mass fractionation correction. To evaluate this, the effects of mass fractionation behavior were investigated by the application of distinct fractionation laws. In this study, the exponential law is applied to correct for mass fractionation, but because of minor loss of tungsten on chromatographic columns (~90% W recovery), any induced mass fractionation not following this law can influence the results by yielding residual anomalies relative to the bracketing standard. To determine the role of isotopic mass fractionation during sample processing, mass fractionation corrections for the measured raw isotope ratios were calculated using both the exponential and equilibrium fractionation factors (Maréchal et al. 1999; Wombacher and Rehkämper 2003). The maximum effect, corresponding to using the exponential law to correct for purely equilibrium governed fractionation for ε¹⁸³W (6/4) was ~4 ppm, which is within the uncertainty of the measurements. This is a consequence of the low degree of stable isotope mass fractionation of samples relative to the terrestrial standard (typically ≤0.5 ‰/a.m.u.) and the small relative mass difference for the isotopes of tungsten. Thus, mass fractionation correction using the exponential law is valid and does not cause residual variability outside stated 2σ uncertainties.

Nuclear Charge-Related ¹⁸³W Deficit

Small analytically derived deficits in ¹⁸³W have been reported in previous high-precision studies (Shirai and Humayun 2011; Willbold et al. 2011; Kruijer et al. 2012) and may cause anomalies in (6/3) corrected data as well as ε¹⁸³W (6/4). Correcting for such effects can be performed using different normalization schemes (¹⁸⁶W/¹⁸⁴W and ¹⁸⁶W/¹⁸³W) combined with data for terrestrial standards of known W isotope composition. However, all high-resolution data for processed standards have mean ε¹⁸³W within error of the terrestrial value, suggesting that our analytical protocol does not result in a deficit in ¹⁸³W due to a mass independent isotope fractionation effect within the reproducibility of our approach. As such, we infer that the observed deficits in ε¹⁸³W for Cape York, Rhine Villa, and Bendego are not a result of nuclear charge effects and thus do not make any correction to our meteorite data. It is not possible to completely rule out the presence of minute, analytically derived ¹⁸³W deficits in Ivuna, Allende, and CAI BE but for these samples the magnitude of this effect is expected to be smaller than the 2σ uncertainty given that it is not observed in our rock or pure W column-processed standards. It may be speculated that the terrestrial ε¹⁸⁴W (6/3) of the IVB iron Tlacotepec measured in this study is the result of a deficit in ¹⁸³W, shifting it to a terrestrial value, i.e., more positive than in earlier reports (~−0.1ε, Qin et al. 2008a; Kruijer et al. 2012; Wittig et al. 2013) and such an effect cannot be ruled out. We note however that the two other analyzed IVB irons, Skookum and Weaver Mountains, have ε¹⁸³W ~0 suggesting no unique nucleosynthetic ^183,184W anomalies for the IVB parent body and there is thus no a priori reason to expect anomalies in Tlacotepec.

Organic Interferences

It is known that tungsten isotope anomalies can be affected by an organic interference on mass 183, impacting not only ε¹⁸³W but all (6/3) normalized data (Kleine et al. 2002, 2004). Terrestrial basalt standards are low in organics compared to meteoritic material so these may not yield anomalous results rendering it difficult to test the magnitude of such matrix effects. Because many chondritic samples are rich in organics, it is important either to remove chemically all such material through strong oxidation with HNO₃:H₂O₂ and/or to fully resolve molecular interferences during acquisition of isotope data. The applied purification protocol uses HNO₃:H₂O₂ both before and after the column chemistry to ensure thorough oxidation and removal of organic molecules. In particular, each sample is oxidized in this way a total of six times subsequent to the column chemistry. Moreover, the elution of W for chondritic samples is carried out in 6 M HCl − 1 M HF, which results in less damage to the anion resin than if elution was carried out in HNO₃ and thus less organics in the W eluate. Nevertheless, the presented data for Ivuna and CAI BE appear to indicate a slight excess on ¹⁸³W with ε¹⁸³W of 0.36 ± 0.12 and 0.50 ± 0.17, respectively. A similar but more pronounced signature was observed for CI chondrite Orgueil (Kleine et al. 2004) indicating that it may be particularly difficult to fully remove organics from this class of meteorites. However, we rule out an interference on ε¹⁸³W for CAI BE by considering the ¹⁸²Hf-¹⁸²W systematics of this inclusion. From the decay-corrected ε¹⁸²W, it is possible to distinguish an interference-related anomaly on ¹⁸³W from a nucleosynthetic s-deficit/r-excess as illustrated in Fig. 3. Thus, the ε¹⁸³W in high resolution for CAI BE is caused by nucleosynthetic variability and not an organic interference. Figure 3 also indicates that Ivuna and Allende are not affected by an interference on ¹⁸³W as will be discussed in The Distribution of ¹⁸³W section. This argument is strengthened by the application of high-resolution mass spectrometry where interfering organic molecules are resolved from the ¹⁸³W beam. Note that the low-resolution measurements of CAI BE and Skookum are affected by such an interference that is resolved in high-resolution mode. Indeed, the excess ε¹⁸³W observed for Skookum in low-resolution mode corresponds to a deficit in ε¹⁸⁴W of the same magnitude observed by Qin et al. (2008a, 2008b) and Wittig et al. (2013) indicating that this signature results from an unresolved molecular interference when measuring in low-resolution mode.

Fig. 3.

Plot showing ε¹⁸³W/¹⁸⁴W versus ε¹⁸²W/¹⁸⁴W for all investigated iron meteorites, chondrites Ivuna (CI) and Allende (CV), and refractory inclusion CAI BE. Also shown are data for USGS rock standard BCR-2 and the NIST 3163 W solution standard. Filled symbols represent magmatic iron meteorites and all symbols are as in Fig. 1. All plotted data are normalized using the exponential mass fractionation law and ¹⁸⁶W/¹⁸⁴W, see text. The arrow indicates the direction of s-process deficit/r-process excess forming a mixing line with solar system material at an initial ε¹⁸²W/¹⁸⁴W of −3.51 (Burkhardt et al. 2012b). Note that the ordinate intercept for bulk irons may be slightly higher than solar system initial due to decay of ¹⁸²Hf on the parent body prior to core formation. Gray data symbols represent measured data, not corrected for ingrowth on ¹⁸²W from the decay of ¹⁸²Hf. For Ivuna, the decay correction is based on the most recent Hf/W ratio for this meteorite (Barrat et al. 2012) whereas for CAI BE, we have directly determined the Hf/W ratio on a sample aliquot. For Allende, the decay correction is calculated using a chondritic Hf/W ratio of 1.2. The large uncertainty associated with decay correction of Ivuna means that it is not possible to distinguish an analytical artifact from a nucleosynthetic excess on ¹⁸³W (see text). However, when also considering that Allende and CAI BE plot on the expected mixing line between solar system material and a component characterized by an s-process deficit/r-process excess (in agreement with earlier reports; Burkhardt et al. 2012b), we conclude that Ivuna’s composition is most likely a nucleosynthetic anomaly.

In addition, positive ε¹⁸⁰W may result from a nonresolved molecular interference during mass spectrometry indicated by the low-resolution data for Cape York (Table 1). Thus, it appears that acquiring W isotope data in high-resolution mode is necessary to ensure the accuracy of high-precision W isotope measurements by MC-ICPMS, at least for the sample introduction system and cone configuration utilized here.

Galactic Cosmic Ray Exposure

Galactic cosmic ray exposure during meteoroid transit to Earth is known to create substantial isotopic anomalies in iron meteorites by (epi)thermal neutron capture (Masarik 1997; Leya et al. 2000, 2003; Kruijer et al. 2013; Wittig et al. 2013). These anomalies affect chronologies based on the Hf-W short-lived isotope decay system by causing shifts in the daughter isotope abundances. Tungsten-182 in iron meteorites is consumed by cosmic ray irradiation through capture of (epi)thermal neutrons during the meteoroid phase, such that ε¹⁸²W is lowered with time. This effect was noted by Kleine et al. (2005), Markowski et al. (2006b), and Scherstén et al. (2006) and expected depletions for a probable range in shielding depths are ~0.1−0.2 ε per 100 Myr (Qin et al. 2008b; Kleine et al. 2009). Methods have recently been developed to empirically correct for burnout of ¹⁸²W (Kruijer et al. 2013; Wittig et al. 2013) using cosmic ray dosimetry. This is based on induced shifts in Os and Pt isotopes that are particularly sensitive to (epi)thermal neutrons, in analogy to ¹⁸²W. Although analytically challenging and time-consuming, these methods provide a direct means of quantifying burnout effects and apply accurate cosmic ray exposure correction of ¹⁸²W to individual iron meteorite samples. As such, cosmic ray exposure correction based on Os and Pt dosimetry is superior to previously developed models for determining exposure effects on ¹⁸²W. Unfortunately, our iron meteorite samples were consumed by analysis prior to the reports of Pt and Os dosimetry rendering it impossible to quantify accurate ¹⁸²W burnout. We emphasize that accurate ε¹⁸²W and Hf-W chronology is not the focus of the present work and we do not further discuss the cosmic ray exposure effects for ¹⁸²W. We do note, however, that our uncorrected ε¹⁸²W are in good agreement with uncorrected ε¹⁸²W from literature (e.g., Markowski et al. 2006a, 2006b; Scherstén et al. 2006).

It is debated whether ¹⁸⁰W is subject to burnout or production during prolonged exposure to cosmic rays (Schulz et al. 2013; Cook et al. 2014). Indeed, spurious burnout signatures were observed in Bendego (IC) and Ainsworth (IIAB) (Schulz et al. 2013) indicating that the net effect of prolonged cosmic ray exposure is to lower ε¹⁸⁰W so that burnout dominates putative cosmogenic production of ¹⁸⁰W. In contrast, Cook et al. (2014) showed evidence of cosmogenic production of ¹⁸⁰W, particularly for the IVB iron meteorites. In our study, Tlacotepec has excess ε¹⁸⁰W coupled with a prolonged cosmic ray exposure, whereas no resolved deficits are observed for iron meteorites. This is in agreement with data from Cook et al. (2014) suggesting that cosmogenic production dominates putative burnout of ¹⁸⁰W. In summary, we observe no resolvable effects of cosmic ray burnout on the ¹⁸⁰W abundance, which suggests that the samples were shielded, or that potential effects are small and may be masked by the comparatively large uncertainties on ¹⁸⁰W measurements. If the IVB iron meteorites were affected by ¹⁸⁰W burnout (e.g., Tlacotepec has an exposure age of 945 Myr [Voshage and Feldmann 1979]) then their excess ¹⁸⁰W would represent lower bounds on their original composition.

Cosmic ray irradiation causes changes in both ε¹⁸²W and ε¹⁸⁴W, but for a shift of −0.5 in ε¹⁸²W, a corresponding shift in ε¹⁸⁴W of −0.03 is estimated (Qin et al. 2008a). Assuming that the difference between our most negative measured ε¹⁸²W (Bendego, −4.11 ± 0.06) and the average pre-exposure ε¹⁸²W for iron meteorites (Kruijer et al. 2013) of −3.23 represents the maximum CRE effects (i.e., 0.88ε), the corresponding shift in ε¹⁸⁴W is 0.05, i.e., within the stated 2σ uncertainties. Moreover, Bendego and Tlacotepec, which have the longest exposure ages in our data set, do not record deficits in ε¹⁸⁴W supporting the view that they were not significantly affected by burnout for any other isotope than ¹⁸²W. Thus, the ε¹⁸⁴W of the analyzed iron meteorites do not record cosmic ray burnout, and can be considered primary signatures.

Comparison with Earlier Studies

During the course of our analyses, we have not observed the strong fractionation effects causing deficits in ε¹⁸⁰W that were noted by Schulz et al. (2013) when using the Jet sampler cone in high-resolution mode on the Neptune Plus. In contrast, our terrestrial rock and steel standard data, as well as that of most meteoritic samples agree with the terrestrial ε¹⁸⁰W, supporting the reliability of our analytical setup. Overall, our high-resolution ε¹⁸⁰W data agree with Cook et al. (2014) wherein only the IVB iron meteorite group clearly displays positive anomalies. The IVB irons show a large spread in ε¹⁸⁰W, indicating that the anomalies do not represent a reservoir uniformly enriched in p-nuclei and sampled by the IVB parent body prior to its differentiation as proposed by Schulz et al. (2013). Notably, our low-resolution ε¹⁸⁰W data resemble that of Schulz et al. (2013) where IVB irons are significantly positive and at least one other sample is negative, which we do not observe in high-resolution mode. As a consequence, we again emphasize the need for high-resolution mass spectrometry when assessing ¹⁸⁰W variability.

The ε¹⁸²W data are in agreement with previous reports (Kleine et al. 2005; Scherstén et al. 2006; Qin et al. 2008a; Kruijer et al. 2013; Schulz et al. 2013; Wittig et al. 2013) where magamatic iron meteorites range from ε¹⁸²W ~ −4 to −3. For reasons noted above, we consider this a good quality check and do not attempt to correct for cosmic ray exposure effects as we have not combined the W isotope measurements with Pt or Os isotopes (Kruijer et al. 2013; Wittig et al. 2013).

Finally, our ε¹⁸³W data are anomalous relative to the terrestrial standard for some magmatic iron meteorites (Cape York, Rhine Villa, and Bendego) as well as Ivuna, Allende, and CAI BE. In contrast to previous studies (Shirai and Humayun 2011; Kruijer et al. 2012) these anomalies are not present in processed standards (see The Distribution of ¹⁸³W section). In addition, we do not identify a negative ε¹⁸⁴W (corresponding to a positive ε¹⁸³W) for the IVB iron Tlacotepec as was shown by Markowski et al. (2006a), Qin et al. (2008a), and Wittig et al. (2013). The reason for this discrepancy is unclear but all other iron meteorites either overlap with the terrestrial ε¹⁸⁴W or are slightly positive. In particular, the two IVB iron meteorites Skookum and Weaver Mountains overlap with the terrestrial ε¹⁸⁴W. Therefore, the IVB iron Tlacotepec may also be characterized by a terrestrial ε¹⁸⁴W, as the core of the IVB parent body is expected to be homogeneous (Walker et al. 2008).

The Distribution of ¹⁸⁰W

High-resolution measurements show that widespread positive ¹⁸⁰W anomalies are not present in meteoritic samples. Positive ε¹⁸⁰W anomalies exist only for the IVB iron meteorites. Based on both low- and high-resolution data, there is no indication of variability in ε¹⁸⁰W coupled with inferred parent body groupings (Fig. 4). This is in agreement with Cook et al. (2014) but contrasts the low-resolution work of Schulz et al. (2013), in which distinct and large ¹⁸⁰W anomalies were observed for different suites of iron meteorites. Our data suggest that ¹⁸⁰W does not constrain the provenance of individual iron meteorite parent bodies. Instead, a viable explanation proposed for excess ¹⁸⁰W is that it reflects α-decay of ¹⁸⁴Os (half-life ~10¹³ yr) that could generate the observed signatures in iron meteorites with high Os/W. The IVB irons have high Os/W (Walker et al. 2008) and as shown recently (Peters et al. 2014), anomalies on ¹⁸⁰W correlate with the elemental Os/W ratio possibly resulting from decay of ¹⁸⁴Os. We have not measured Os/W ratios but point out that IVB irons, the only type for which we observe excess ε¹⁸⁰W, are generally characterized by high Os/W compared to other iron meteorite groups (Campbell and Humayun 2005; Walker et al. 2008). Considering the ε¹⁸⁰W (6/4) of Tlacotepec (5.80 ± 2.0) and the evolution of ¹⁸⁴Os decay (T_1/2 = 2.2 × 10¹³ yr [Cook et al. 2014]) for 4.567 Gyr, this iron meteorite sample would have an ¹⁸⁴Os/¹⁸⁴W of approximately 0.0158 ± 0.0054. Correspondingly for Weaver Mountains (ε¹⁸⁰W [6/4] = 2.20 ± 1.2), an ¹⁸⁴Os/¹⁸⁴W = 0.0060 ± 0.0031 would be required. These ¹⁸⁴Os/¹⁸⁴W ratios are somewhat higher than IVB iron data from Peters et al. (2014), but we note that variability is likely, as it has been shown that Os/W can vary substantially within individual iron meteorites and/or groups due to elemental fractionation during core crystallization on their parent bodies (Righter 2003; Chabot et al. 2003). The long half-life of ¹⁸⁴Os produces only minor radiogenic ingrowth on ε¹⁸⁰W even on timescales of several Gyr (Cook et al. 2014), so that the only clear indication of decay comes from high Os/W samples. Although providing an explanation for elevated ε¹⁸⁰W, this potential chronometer does not yield high temporal resolution, and its potential use in early solar system chronology is therefore limited.

Fig. 4.

Measured ε¹⁸⁰W for iron meteorites Ivuna and CAI BE as well as standards in high-resolution mode. All samples except the IVB group irons overlap with the terrestrial ε¹⁸⁰W suggesting no widespread early solar system heterogeneity in the rare p-process isotope ¹⁸⁰W as previously proposed (Schulz et al. 2013). Errors are 2 SE propagated with the 2 SD of the bracketing standard in each analytical session.

The nonanomalous ε¹⁸⁰W, including CAI BE, does not conform to the conclusions of Schulz et al. (2013) but rather suggests that the carrier(s) of heavy p-process isotopes were homogeneously distributed at the scale of planetesimals and down to individual refractory inclusions that formed very early in the lifetime of the protoplanetary disk. The progressive homogenization model for p-process isotopes proposed by Schulz et al. (2013) would require the ε¹⁸⁰W of CAI BE to be highly anomalous and clearly resolved from the terrestrial standard, which is not the case. Likewise, although large variability was observed in Os isotopes in Murchison leachates (Reisberg et al. 2009) this is contrasted by Os isotope homogeneity (Walker 2012) for bulk chondrites, including the rare p-process isotope ¹⁸⁴Os, in good agreement with our bulk ¹⁸⁰W data. The same conclusion holds for apparent variability in p-process isotopes ⁹²Mo (Burkhardt et al. 2011) and ⁸⁴Sr (Paton et al. 2013), which instead was attributed to incorporation of variable amounts of nucleosynthetically distinct s-process carrier grains. The general idea of p-process homogeneity is, however, contrasted by a deficit observed in carbonaceous chondrites for p-nuclide ¹⁴⁴Sm (and possibly short-lived ¹⁴⁶Sm; Andreasen and Sharma 2006; Qin et al. 2011) suggesting a heterogeneous distribution of nucleosynthetic products from a nearby stellar p-process source. Because the p-process functions in distinct and rare environments with massive stellar progenitors (Woosley et al. 2002), the p-nuclides that were incorporated into the solar system may be present in the same grain carrier type/population and hence have coupled distributions in early solar system solids. This is indeed what is observed for W and Os as well as Zr (Schönbächler et al. 2003). Yet the disparity between Sm (Andreasen and Sharma 2006) and other p-nuclide abundances (Schönbächler et al. 2003; Reisberg et al. 2009; Burkhardt et al. 2011; Walker 2012; Paton et al. 2013) including the W isotope data presented here, points to a decoupling of p-process nuclides in the protoplanetary disk. Accepting the present body of data, all but Sm p-isotope is robustly explained by s-deficit/r-excess (Paton et al. 2013). The carrier of p-process isotopes and its detailed distribution in early solar system solids have yet to be constrained but it is known that the protoplanetary disk was characterized by s- and r-process nucleosynthetic isotope variability (e.g., Trinquier et al. 2009; Burkhardt et al. 2012a) and this would impart apparent anomalies on internally normalized p-process isotope data. A possible mechanism for generating a heterogeneous distribution of s- and r-process carriers is unmixing by thermal processing (Trinquier et al. 2009; Paton et al. 2013) that affected only the more volatile elements such as Mo, Ru, and possibly Sm (Trinquier et al. 2007, 2009; Chen et al. 2010; Burkhardt et al. 2011) residing within isotopically anomalous presolar phases. Unmixing imparted isotopic variability in these elements in the residual processed reservoir. Conversely, elements like W, Os, Zr, and Hf (Schönbächler et al. 2003; Yokoyama et al. 2007; Reisberg et al. 2009; Sprung et al. 2010; Burkhardt et al. 2012a) that remain highly refractory during thermal processing show no or only slight isotopic heterogeneity. These elements were thus not significantly affected by thermal processing and did not undergo volatilization. Therefore, they remained isotopically homogeneous throughout the protoplanetary disk. The study of p-isotope heterogeneity should thus be restricted to refractory elements with no demonstrable s-/r-process variability. Although minor nucleosynthetic variability for W isotopes of IVB irons (Markowski et al. 2006a; Qin et al. 2008a) could result from incorporation of variable amounts of s- and r-process carriers at a bulk planetesimal scale, our results indicate that the refractory element W was not significantly volatilized during the processing that caused volatile depletion in the IVB irons. Therefore, no nucleosynthetic W isotopic anomalies were imparted (cf. Burkhardt et al. 2012a) in the IVB parent body. Minor heterogeneity in the p-nuclei of Sm in carbonaceous chondrites argues against overall p-nuclide homogeneity, and more detailed studies are needed to shed light on this decoupling.

The Distribution of ¹⁸³W

The observation of anomalies on ε¹⁸³W (6/4) (ranging from −0.15 ± 0.07 [Bendego] to 0.51 ± 0.17 [CAI BE]; Table 1) suggests variable incorporation of nucleosynthetically distinct components. This signature cannot be explained by a nuclear charge related ¹⁸³W deficit as noted in Nuclear Charge Related ¹⁸³W Deficit section and is the reason why ε¹⁸²W (6/4) and ε¹⁸²W (6/3) anomalies differ for CAI BE and Ivuna. In a plot of ε¹⁸³W (6/4) versus ε⁶²Ni (Ni isotope data from Steele et al. 2011, 2012) both differentiated and chondritic specimens exhibit a correlation (Fig. 5). Our data for Ivuna represent the first interference-free, high-precision ε¹⁸³W data for a bulk CI chondrite. As such, the applied decay correction for Ivuna (Fig. 3) using Hf and W concentrations from Barrat et al. (2012) could suggest that this sample is affected by an interference on mass 183. As highlighted in the Organic Interferences section, our high-resolution data are not affected by a molecular intereference and the correlation with Ni isotopes includes CV3 chondrite Allende as well as several iron meteorites, emphasizing the robustness of the ε¹⁸³W result for Ivuna. Therefore, we suggest that the precision and/or accuracy of the Hf and W concentration measurements in Barrat et al. (2012) may not be sufficient to distinguish between an interference related excess on mass 183 and a real s-deficit/r-excess when applying the Hf/W ratio to correct for ingrowth (the theoretical mixing line between a nucleosynthetic endmember and initial solar system in Fig. 3 is calculated analogously to Burkhardt et al. [2012a]; Kruijer et al. [2012; and Kruijer, personal communication). Indeed, the concentration data for W reported by Barrat et al. (2012) vary by 20% among bulk CI chondrites and this variability becomes larger when also considering the concentration data of Lodders (2003) and Lodders et al. (2009). When this uncertainty is propagated into our decay correction, it is impossible to distinguish an interference on mass 183 from an s-deficit/r-excess (Fig. 3).

Fig. 5.

ε⁶²Ni versus ε¹⁸³W (6/4). ⁶²Ni data are from Steele et al. (2011, 2012). Symbols are as in Fig. 1 and filled symbols represent magmatic iron meteorites. The Ni isotopic composition of Ivuna is assumed to be identical to Orgueil as measured by Steele et al. (2012) and Cape York is assumed to have the same Ni isotope composition as Lenarto (IIIAB) as also measured by Steele et al. (2012). We make no corrections for an odd-even isotope effect, resulting in an apparent ε¹⁸³W deficit (Shirai and Humayun, 2011; Kruijer et al. 2012). Allende W isotope data are taken from Holst et al. (2013). Errors on W isotope data are 2 SE propagated with the 2 SD of bracketing standards for each session (except for Allende, which is 2 SE, n = 5). Errors on nickel isotope data are 2 SE (Steele et al. 2012). The correlation between Ni and W isotopes is most likely related to a coupled incorporation of nucleosynthetic carrier(s). The Ni isotope variability is ascribed to variations in the abundance of neutron-poor ⁵⁸Ni (Steele et al. 2012) and the correlation of ⁵⁸Ni and ¹⁸³W could result from a nebular sorting process such as thermal processing (see text).

Although there are no robust Ni isotopic data available for CAI BE, its ε¹⁸³W is more positive than Ivuna and Allende, thus representing a more extreme isotopic composition. This is similar to results for Cr and Ti isotopes where CAIs carry the most positive anomalies followed by CI and CV chondrites (Trinquier et al. 2009) and corroborates our interpretation of nucleosynthetic heterogeneity in W isotopes. Based on the correlation between ε¹⁸³W and ε⁶²Ni, we conclude that the ε¹⁸³W for Ivuna is not affected by analytical artifacts and is the result of an s-deficit/r-excess compared to Earth. Note that the spread in ⁶²Ni is ascribed to nucleosynthetic variability in the neutron poor ⁵⁸Ni (Steele et al. 2012) and that W and Ni isotopes are of different nucleosynthetic origin. The correlation between nucleosynthetic anomalies in W and Ni isotopes suggests (1) a coupled and variable incorporation of carrier grain(s) for ⁵⁸Ni and ¹⁸³W in bulk early solar system objects and/or (2) that an initial homogeneous mixture of carrier grains was affected by a nebular sorting mechanism that generated the observed correlated anomalies. The primary objective is thus to investigate the process(es) by which these nuclei may be heterogeneously distributed on a bulk scale in the early protoplanetary disk.

The Origin of W Isotope Variability

Previous studies have explained the coupled variability in non radiogenic isotope ratios in bulk objects and inclusions by selective thermal processing in the solar nebula (Trinquier et al. 2009; Paton et al. 2013). Above, we noted how apparent p-isotope heterogeneity may also arise from thermal processing that segregated differential amounts of presolar s- and r-process carriers. Thermal processing results in variable incorporation of refractory presolar SiC that is the main carrier of the s-process signature (Zinner 2003). Following this argument, we performed mass balance calculations to investigate the probability of SiC controlling the ¹⁸³W isotope budget for bulk objects. Paton et al. (2013) show that CAIs carry the most positive anomalies in non-s-process nuclei due to removal of SiC from their precursor reservoir relative to Earth. Our calculations based on W isotopes using SiC concentration data (Huss 1990; Huss and Lewis 1995) as well as W concentration and isotope data for SiC (Knight et al. 2008; Ávila et al. 2011) suggest that CI chondrites contain less than 20% of the SiC that the Earth incorporated. Assuming that SiC comprises the main presolar s-process carrier, this corresponds to a SiC enrichment factor of >5 × CI and implies that the reservoir from which the Earth formed had lost ~80% of its material prior to or during accretion. Such extensive material loss is highly unlikely indicating that the W isotope variability is not controlled by differential incorporation of SiC (i.e., the s-process signature) among bulk solar system objects.

Rather than an s-process deficit, the observed ε¹⁸³W anomalies in chondrites and CAI BE are likely related to a preferential incorporation of r-process-enriched material relative to Earth, and iron meteorites. This is consistent with previous conclusions based on Mo isotopes in Allende type B CAIs (Burkhardt et al. 2011). Due to uncertainties in the isotopic composition and the exact nature of the r-process carrier(s) in the early solar system, it is not possible to quantify the degree of r-process enrichment as a function of the observed anomalies. However, we suggest that the magnitude of the ¹⁸³W anomalies could be related to the selective incorporation of one or more anomalous r-process carriers rather than by bulk depletion/enrichment of SiC in planetesimals and planetary objects. Indeed, this is perfectly consistent with a thermal processing scenario in which thermally labile carrier grains, possibly carrying the r-process signature, are enriched in the gaseous CAI precursor reservoir and, in turn, would imply that chondrites, Earth, and iron meteorite parent bodies formed from the residual processed disk material. Furthermore, such a model predicts higher degrees of thermal processing in the inner versus outer protoplanetary disk (cf. Trinquier et al. 2009), consistent with the observation that carbonaceous chondrites are distinct in their ε¹⁸³W from that of Earth and iron meteorites, the former having formed further from the Sun. The r-process excess observed in carbonaceous chondrites results from either a lower degree of thermal processing (e.g., Ivuna) and/or a higher abundance of incorporated CAIs (Allende), carrying the r-process signature from thermally labile precursor grains (Trinquier et al. 2009; Burkhardt et al. 2012a).

In summary, the stable W isotope variability among bulk solar system objects is consistent with a thermal processing scenario, as has previously been proposed to explain bulk heterogeneity in Ti, Cr, Sr, Mo, and Ni isotopes (Trinquier et al. 2009; Burkhardt et al. 2012a; Steele et al. 2012; Paton et al. 2013). Due to the nucleosynthetic origin of W isotopes, these signatures are not readily explained by varying abundances of presolar SiC in bulk planetary objects. Instead a viable explanation is an excess of r-process material in CAIs and carbonaceous chondrite parent bodies such as those of CI and CV chondrites. Ultimately, this is explained by thermal processing in the solar nebula where thermally labile r-process carrier phases were vaporized in the inner protoplanetary disk to form part of the gaseous precursor reservoir to CAIs. These, as well as pristine outer disk material enriched in r-process nuclei, were then later incorporated into carbonaceous chondrites, making them distinct from Earth and iron meteorite parent bodies.

Conclusions

Our results for tungsten isotopes give constraints on the nucleosynthetic make-up of the protoplanetary disk and the processing which governed isotope variability. The conclusions are summarized as follows:

1. We observe no ¹⁸³W deficit in any processed standards. Such deficits have been reported previously as caused by a nuclear-charge–related fractionation during sample preparation. As our analytical method does not result in resolved deficits, there is no requirement for correction of the (6/3) normalized data or ε¹⁸³W (6/4) for meteorite samples.

2. In low-resolution mode, we observe spurious effects of organic interferences on mass 183 (0.25−0.5ε). However, high-resolution data for chondrites and CAI BE do not show this effect and are consistent with an s-process deficit/r-process excess. Thus, for high-precision measurements of ¹⁸³W/¹⁸⁴W, it is important to apply high-resolution mass spectrometry. In addition, ¹⁸⁰W data may be affected by molecular interferences as indicated by the positive ε¹⁸⁰W of Cape York. This signature is resolved in high-resolution mode, further supporting the use of high mass resolution mass spectrometry for high-precision W isotope measurements.

3. There are no resolved effects of meteoroid phase cosmic ray exposure on either ε¹⁸⁰W or ε¹⁸³W. Substantial effects are observed for ε¹⁸²W in iron meteorites with long exposure ages, in agreement with previous reports, yet the ε¹⁸⁰W and ε¹⁸³W data appear to be unaffected by cosmic rays.

4. The constant ε¹⁸⁰W for the studied irons (except the IVB group), chondritic materials, and a refractory inclusion provides evidence for p-process homogeneity in the young protoplanetary disk and is in agreement with recent results from Cook et al. (2014). In contrast to a recent study by Schulz et al. (2013), there is no indication of widespread heterogeneity of ¹⁸⁰W and progressive p-process nuclide homogenization in the disk. Rather, p-process nuclei of several elements (e.g., Sr, Mo, Ba, and W) appear to have been homogeneous when the solar system formed. The elevated ε¹⁸⁰W for IVB iron meteorites is likely due to their high Os/W ratio suggesting that it is a result of long-term accumulation from the α-decay of ¹⁸⁴Os, as was recently suggested by Peters et al. (2014). We did not determine an ¹⁸⁴Os/¹⁸⁰W isochron for the samples studied here but point out that the variable ε¹⁸⁰W within the group IVB irons is qualitatively compatible with ¹⁸⁴Os decay in samples with high and variable Os/W ratios.

5. Variability in ε¹⁸³W in bulk meteoritic objects and inclusions (−0.15 ± 0.07 to 0.51 ± 0.17) reflects variable incorporation of nucleosynthetically anomalous presolar components. As such, Ivuna, Allende, and CAI BE are s-process depleted/r-process enriched compared to Earth and iron meteorites. This is likely due to the preservation of a thermally labile r-process carrier(s) in Ivuna (which accreted in the outer protoplanetary disk) and the enrichment of r-process carrier material in Allende and the CAI (which condensed from a gas following thermal processing).

6. Thermal processing is further supported by the observed correlation of ¹⁸³W and ⁵⁸Ni, as these anomalous isotopes of the two elements are produced in different stellar environments. Moreover, the generation of isotopic anomalies through thermal processing of dust is backed by other isotope data (e.g., Cr and Ti) that suggest widespread thermal processing in the protoplanetary disk during the earliest stages of planetesimal formation.