New Constraints on the Abundance of ⁶⁰Fe in the Early Solar System

Iron-60 is a short-lived radionuclide (T_1/2 = 2.62 ± 0.04 Ma; Rugel et al. 2009), and its abundance in the early solar system is highly debated. The initial value is generally defined as the ⁶⁰Fe/⁵⁶Fe ratio at the time of condensation of the first solar system solids, hereafter (⁶⁰Fe/⁵⁶Fe)₀, and is determined by finding excesses of ⁶⁰Ni due to in situ ⁶⁰Fe decay in minerals with a known Fe/Ni elemental ratio. Measurements of bulk chondrules, mineral separates, and bulk measurements of achondrites by inductively coupled plasma mass spectrometry give an initial (⁶⁰Fe/⁵⁶Fe)₀ ratio of (1.01 ± 0.27) × 10⁻⁸ (Tang & Dauphas 2012, 2015). In situ analyses by secondary ion mass spectrometry (SIMS) of early solar system phases, however, give (⁶⁰Fe/⁵⁶Fe)₀ ratios between 5 × 10⁻⁸ and 3 × 10⁻⁷ (Telus et al. 2012, 2018), and values up to 1 × 10⁻⁶ have been reported (Mishra & Goswami 2014; Mishra & Chaussidon 2014; Mishra et al. 2016). Because ⁵⁶Fe is a primary isotope with respect to nucleosynthesis, while ⁶⁰Fe is a secondary isotope made from ⁵⁶Fe, first-order galactic chemical evolution models indicate that the ⁶⁰Fe/⁵⁶Fe ratio should be constant over time (e.g., Huss et al. 2009). From γ-ray observations (Wang et al. 2007) we expect an average ⁶⁰Fe/⁵⁶Fe ratio in the interstellar medium (ISM) of ∼3 × 10⁻⁷. However, the initial solar system (⁶⁰Fe/⁵⁶Fe)₀ ratio is likely lower, as ⁶⁰Fe ejected from a supernova (SN) goes primarily into a hot phase and thus cannot mix with a cold protostellar cloud without first cooling down (see, e.g., Dwarkadas et al. 2017). The measurements by Tang & Dauphas (2012), Tang & Dauphas (2015), as well as some SIMS measurements (e.g., by Telus et al. 2012, 2018; Mishra & Goswami 2014; Mishra & Chaussidon 2014; Mishra et al. 2016) are in agreement with the expectation of an (⁶⁰Fe/⁵⁶Fe)₀ ratio from galactic background (hereafter called "low"). Some SIMS measurements by the same authors, however, show a "high" ⁶⁰Fe/⁵⁶Fe level, i.e., larger than galactic background. The most likely stellar source to contribute such high amounts of ⁶⁰Fe is a core-collapse SN (Wasserburg et al. 1998; Hester & Desch 2005; Huss et al. 2009; Ouellette et al. 2009; Mishra & Goswami 2014). Such an injection event could have also triggered the collapse of the protosolar nebula to form the solar system (e.g., Boss & Keiser 2010). A high ⁶⁰Fe abundance in the early solar system would also have provided a significant heat source for the first ∼10 Ma and would thus have contributed significantly to planetesimal differentiation and core formation (Moskovitz & Gaidos 2011).

Here we present new resonance ionization mass spectrometry (RIMS) measurements of the nickel isotopic composition of chondrule DAP1 from the unequilibrated ordinary chondrite Semarkona (LL3.00) and a re-evaluation of previous SIMS measurements of the same chondrule. While SIMS cannot measure ⁵⁸Ni and ⁶⁴Ni due to interferences with ⁵⁸Fe and ⁶⁴Zn, respectively, the selective ionization method of RIMS allows accurate measurements of these isotopes with minimal isobaric interferences.

Tang & Dauphas (2015) reported an ⁶⁰Fe/⁵⁶Fe ratio of (5.4 ± 3.3) × 10⁻⁹ at the time of chondrule formation, based on an isochron consisting of several individual Semarkona chondrules. Chen et al. (2013) measured DAP1 by thermal ionization mass spectrometry (TIMS) and reported a solar ⁶⁰Ni/⁵⁸Ni ratio, which translates to an upper limit on the initial ⁶⁰Fe/⁵⁶Fe ratio of 9.1 × 10⁻⁸ (2σ).

Previous SIMS measurements of chondrule DAP1 by Huss et al. (2010) and Telus et al. (2011) found high ⁶⁰Fe/⁵⁶Fe ratios. A re-evaluation of these measurements (Telus et al. 2012), accounting for a previously unappreciated statistical bias (Ogliore et al. 2011; Coath et al. 2013), gave a corrected ⁶⁰Fe/⁵⁶Fe ratio of (2.1 ± 1.2) × 10⁻⁷. Using synchrotron X-ray fluorescence to map the iron and nickel distribution in DAP1, Telus et al. (2016) concluded that only minor nickel mobilization took place in this sample and that it therefore could be used to study the initial (⁶⁰Fe/⁵⁶Fe)₀ ratio. Recently, Telus et al. (2018) re-analyzed the previously measured SIMS isochrons and their significance. Based on the large scatter of the SIMS isotope measurements, Telus et al. (2018) concluded that DAP1 is disturbed due to nickel mobilization, and therefore its ⁶⁰Fe/⁵⁶Fe ratio is not significantly different from zero. Despite these disturbances, Telus et al. (2018) argued that the significant excesses in ⁶⁰Ni measured by SIMS meant that the initial ⁶⁰Fe/⁵⁶Fe ratio at chondrule formation was higher than 5 × 10⁻⁸, which, by correcting for the age of Semarkona chondrules (Kita & Ushikubo 2012) relative to calcium–aluminum rich refractory inclusions (CAIs), translated into an initial (⁶⁰Fe/⁵⁶Fe)₀ ratio of (0.8–5.0) × 10⁻⁷.

2.1. Sample and Standards

2.2. RIMS Measurements

2.3. Notation and Regression

The following notation is used throughout this Letter to discuss the isotopic results. For measurements that were standard-corrected, but uncorrected for mass-dependent fractionation in the sample, we use δ-notation (Equation (1)), which represents the deviation of the sample measurement (Smp) from a measured standard (Std) in permil:

$$\begin{eqnarray}&&{\delta }^{x}{\mathrm{Ni}}_{y}(\%o)=\left[\displaystyle \frac{{({}^{x}\mathrm{Ni}{/}^{y}\mathrm{Ni})}_{\mathrm{Smp}}}{{({}^{x}\mathrm{Ni}{/}^{y}\mathrm{Ni})}_{\mathrm{Std}}}-1\right]\times 1000.\end{eqnarray} \tag{ 1 }$$

Thus, instrumentally induced fractionations, e.g., isotope fractionation due to the odd–even effect in photoionization, are removed. We assess the precision of δ-values by examining the reproducibility of standard measurements.

The isotopic variations that we are primarily interested in are ⁶⁰Ni excesses that would show up as departures from the law of mass-dependent isotopic fractionation. Correction for natural and laboratory-induced mass fractionation is accomplished by internal normalization and ⁶⁰Ni excess is reported as Δ⁶⁰Ni_x/y, where x and y indicate the normalizing ratio. As is commonly done (e.g., Dauphas & Schauble 2016) mass-dependent fractionation is corrected for using the exponential law, which takes the form

$$\begin{eqnarray}&&{\beta }_{x/y}=\displaystyle \frac{\mathrm{log}[{({}^{x}\mathrm{Ni}{/}^{y}\mathrm{Ni})}_{\mathrm{Smp}}/{({}^{x}\mathrm{Ni}{/}^{y}\mathrm{Ni})}_{\mathrm{Std}}]}{\mathrm{log}({m}_{x}/{m}_{y})}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{\left(\displaystyle \frac{{}^{60}\mathrm{Ni}}{{}^{y}\mathrm{Ni}}\right)}_{\mathrm{Smp}\ \mathrm{or}\ \mathrm{Std}}^{* }=\displaystyle \frac{{({}^{60}\mathrm{Ni}{/}^{y}\mathrm{Ni})}_{\mathrm{Smp}\mathrm{or}\mathrm{Std}}}{{({m}_{60}/{m}_{y})}^{\beta }}\end{eqnarray} \tag{ 3 }$$

$$\begin{eqnarray}&&{{\rm{\Delta }}}^{60}{\mathrm{Ni}}_{x/y}(\%o)=\left[\displaystyle \frac{{({}^{60}\mathrm{Ni}{/}^{y}\mathrm{Ni})}_{\mathrm{Smp}}^{* }}{{({}^{60}\mathrm{Ni}{/}^{y}\mathrm{Ni})}_{\mathrm{Std}}^{* }}-1\right]\times 1000.\end{eqnarray} \tag{ 4 }$$

Other mass-fractionation laws might be more appropriate, depending on the processes responsible for nickel mass fractionation (Dauphas & Schauble 2016), but the differences in calculated Δ⁶⁰Ni_62/58 values between those laws and the exponential law are negligible at the precision of our nickel isotopic analyses. The uncertainties on Δ-values reflect counting errors propagated in the internal normalization scheme through a Monte Carlo approach (Bevington & Robinson 2003).

All weighted, linear regressions are done using the method from Mahon (1996). This method is an updated and corrected calculation following York (1969), which had underestimated the uncertainties in slope and intercept.⁸

Table 1 shows the isotope ratios for all RIMS measurements as well as the ⁵⁶Fe/⁵⁸Ni ratios that we determined from the SIMS measurements.

3.1. Mass-dependent Fractionation

Figure 1 shows δ⁶⁰Ni₅₈ versus δ⁶²Ni₅₈ for all RIMS measurements of Semarkona chondrule DAP1, along with the mass-dependent fractionation line (MFL). The total observed range in mass fractionation in the two isotope ratios is 23‰ and 40‰, respectively, which corresponds to a mass-dependent fractionation of up to ∼10‰/amu. A weighted regression through all measurements results in a mass-fractionation relationship exponent of 0.5 ± 0.1, which encompasses the range of 0.506–0.516 for commonly used mass-dependent fractionation laws for nickel atoms (exponential, equilibrium, Rayleigh; Davis et al. 2015; Dauphas & Schauble 2016).

Zoom In Zoom Out Reset image size

3.2. Internally Normalized Measurements

Figure 2 shows the internally normalized Δ⁶⁰Ni_62/58 versus the standard-corrected δ⁶²Ni₅₈, along with the SIMS measurements of Δ⁶⁰Ni_61/62 versus δ⁶¹Ni₆₂ by Telus et al. (2012, 2018). For the SIMS measurements, we distinguish between the mono- and multicollector data sets as described in Telus et al. (2018). Correlated uncertainties for SIMS and RIMS data were calculated using Monte Carlo error propagation, starting with the uncorrelated uncertainties that are given in the supplementary material of Telus et al. (2018). The mass-dependent fractionation discussed in Section 3.1 shows variations up to 10‰/amu, which are effectively corrected for using internal normalization. Only one out of 16 RIMS measurements shows an excess in Δ⁶⁰Ni_62/58 at the 2σ level, consistent with statistical noise (Bevington & Robinson 2003). The SIMS measurements (Telus et al. 2012, 2018) show a similar spread in δ⁶¹Ni₆₂ (∼10‰/amu), however, they have excess Δ⁶⁰Ni_61/62 of up to 23‰. Uncertainties in the SIMS measurements are larger and more correlated than those of the RIMS measurements, primarily due to the low abundance of the normalizing isotopes in SIMS. Nickel-61 and ⁶²Ni used in SIMS are 1.34% and 3.63% of the total nickel, while ⁶²Ni and ⁵⁸Ni, as used in RIMS, have abundances of 3.63% and 68.08%, respectively. Normalizing to ⁶²Ni/⁵⁸Ni, as was done here for the RIMS measurements, is not an option in SIMS because of a major interference of ⁵⁸Fe on ⁵⁸Ni that cannot be resolved or eliminated.

Zoom In Zoom Out Reset image size

Because Figure 2 does not contain any information on the elemental Fe/Ni ratio, an excess in ⁶⁰Ni due to in situ decay of ⁶⁰Fe would show in this representation in the form that the measurements would be randomly distributed at Δ⁶⁰Ni_62/58 or Δ⁶⁰Ni_61/62 values ≥0. The SIMS measurements, however, show a strong correlation. Within the newly calculated correlated uncertainties, however, no individual SIMS spot has a statistically significant excess or deficit in ⁶⁰Ni.

Figure 3 shows the isochron diagram for DAP1 as determined in this study, the solid lines show the linear relation

$$\begin{eqnarray}&&{\left(\displaystyle \frac{{}^{60}\mathrm{Ni}}{{}^{58}\mathrm{Ni}}\right)}^{* }={\left(\displaystyle \frac{{}^{60}\mathrm{Fe}}{{}^{56}\mathrm{Fe}}\right)}_{0}\times \displaystyle \frac{{}^{56}\mathrm{Fe}}{{}^{58}\mathrm{Ni}}+{\left(\displaystyle \frac{{}^{60}\mathrm{Ni}}{{}^{58}\mathrm{Ni}}\right)}_{0}.\end{eqnarray} \tag{ 5 }$$

Here, (⁶⁰Fe/⁵⁶Fe)₀ is the slope and (⁶⁰Ni/⁵⁸Ni)₀ the intercept of the linear relation if all measurements corresponded to a single Fe/Ni fractionation event from a homogeneous reservoir; a valid assumption for a chondrule formed by rapid crystallization of a molten droplet. To compare the RIMS and SIMS data, we start with the ⁶²Ni-normalized mono- and multicollector SIMS measurements of Telus et al. (2018). We measured the same pits by RIMS that were previously measured by multicollector SIMS (Telus et al. 2012, 2018). These SIMS measurements are also shown in Figure 3 using red triangles. As the elemental ⁵⁶Fe/⁵⁸Ni ratio could not be determined by RIMS, the corresponding SIMS values were used in the isochron diagram. However, the Fe/Ni ratios changed with depth in the SIMS pit, and these measurements show up to a factor of 30 variation in the Fe/Ni ratio, likely due to sample heterogeneity, i.e., small metal inclusions that cannot be avoided by SIMS due to the large spot size. Because the RIMS measurements were performed at the bottom of the SIMS pits, we determined the Fe/Ni ratio for the RIMS measurements by only considering the last 20 cycles of the SIMS measurements, which represent the bottom ∼10% of the depth of the SIMS pits. The RIMS pits were also much smaller and shallower than the SIMS pits, further justifying this approach. This results in different Fe/Ni ratios for the RIMS (see Table 1) compared to the SIMS data. For the uncertainty of the Fe/Ni ratio, we use the standard deviation of these 20 cycles. We note that the slope and intercept calculated from the isochron diagram do not change if we use the same Fe/Ni ratios that were used for interpreting the SIMS results.

Zoom In Zoom Out Reset image size

The linear regression of the RIMS measurements is shown in Figure 3. We find that the initial ⁶⁰Fe/⁵⁶Fe ratio of DAP1 is (3.8 ± 6.9) × 10⁻⁸, with a 2σ upper limit of ⁶⁰Fe/⁵⁶Fe = 1.1 × 10⁻⁷, which agrees with the mono- and multicollector SIMS measurements by Telus et al. (2018) and is consistent with the low (⁶⁰Fe/⁵⁶Fe)₀ ratio as measured by bulk techniques (Tang & Dauphas 2012, 2015) for other samples. However, it is inconsistent with high (⁶⁰Fe/⁵⁶Fe)₀ values as found in some in situ studies (Telus et al. 2012; Mishra & Goswami 2014; Mishra & Chaussidon 2014; Mishra et al. 2016). In addition, we find a solar initial (⁶⁰Ni/⁵⁸Ni)₀, which translates to Δ⁶⁰Ni_62/58 value of +0.1 ± 2.4‰.

The SIMS measurements show a larger spread and larger uncertainties than the RIMS measurements (see Figure 3). All uncertainties were calculated using the described Monte Carlo error propagation. A linear regression through the SIMS measurements results in a mean square weighted deviation (MSWD) of 0.99 for the monocollector measurements and 0.86 for the multicollector measurements, which are acceptable values for the given degrees of freedom. The regressions of the mono- and multicollector SIMS measurements are given in Figure 3. The initial (⁶⁰Ni/⁵⁸Ni)₀ translates to Δ⁶⁰Ni_61/62 ratios of −1 ± 4‰ and +11 ± 9‰ for the mono- and multicollector SIMS measurements, respectively.

Telus et al. (2018) found a large MSWD when fitting an isochron and concluded that DAP1 represents a disturbed system. However, this argument does not hold true when correlated uncertainties are considered, as the MSWD values are near one. Furthermore, the initial ⁶⁰Fe/⁵⁶Fe SIMS ratios agree well with the RIMS measurement. The initial ⁶⁰Ni/⁵⁸Ni ratio of the multicollector SIMS measurements, however, agrees neither with RIMS nor the monocollector SIMS measurements. Telus et al. (2018) pointed out that this discrepancy could be due to different experimental conditions, which seems highly likely when compared with the new measurements of the same pits. The measured ⁶⁰Ni/⁵⁸Ni intercept thus agrees with Telus et al. (2016), which concluded the nickel isotopes in DAP1 are undisturbed.

Using RIMS, we re-measured the Semarkona chondrule DAP1, which had been previously analyzed by SIMS and used as evidence that the initial solar system abundance of ⁶⁰Fe was high (Telus et al. 2012). The RIMS data, as well as the re-evaluation of the SIMS measurements, reveal no statistically significant excesses in ⁶⁰Ni outside of 2σ. Using the Fe/Ni elemental ratios determined by SIMS and the isotope ratios determined by RIMS, we derive an initial ⁶⁰Fe/⁵⁶Fe of (3.8 ± 6.9) × 10⁻⁸ (2σ). Using a formation age of 2.0 ± 0.8 Ma for Semarkona chondrules (based on Al-Mg systematics for other Semarkona chondrules; Kita & Ushikubo 2012), we calculated an (⁶⁰Fe/⁵⁶Fe)₀ ratio of (6.4 ± 11.9) × 10⁻⁸ for the time of CAI condensation. The initial Δ⁶⁰Ni_62/58 of DAP1 measured by RIMS is +0.1 ± 2.4‰ (2σ) and is therefore consistent with terrestrial Ni isotope ratios (Gramlich et al. 1989; Elliott & Steele 2017).

The improvement in precision for RIMS relative to SIMS can be explained by the fact that the SIMS measurements were internally normalized for mass-dependent fractionation using ⁶¹Ni, a task that is especially difficult as ⁶¹Ni is a minor isotope with an abundance of only 1.14%. RIMS avoids this issue as the more abundant ⁵⁸Ni (68.08%) can be measured, because ⁵⁸Fe ionization is suppressed in the resonance ionization scheme employed here.

The RIMS measurements of DAP1 do not provide any evidence for live ⁶⁰Fe in the early solar system. From these results, an injection of material into the early solar system by an SN is unlikely. As shown above, the solar system initial (⁶⁰Fe/⁵⁶Fe)₀ inherited from galactic background is expected to be lower than ∼3 × 10⁻⁷. Our new measurements agrees with such a low value for DAP1. Other short-lived radionuclides that were present in the early solar system, e.g., ²⁶Al, which was present at a canonical level of ²⁶Al/²⁷Al = 5.2 × 10⁻⁵ (Jacobsen et al. 2008), thus require a different origin than a SN injection event. These radionuclides could be, e.g., the result of stellar winds from a Wolf–Rayet star (Gaidos et al. 2009; Gounelle & Meynet 2012; Young 2014; Dwarkadas et al. 2017). Recently, Dwarkadas et al. (2017) showed that such a scenario can explain the amounts of ²⁶Al as well as the low amounts of ⁶⁰Fe simultaneously, thus agreeing with the measurements Tang & Dauphas (2012, 2015), the low SIMS measurements, as well as our new RIMS measurements.

We would like to acknowledge helpful comments from an anonymous reviewer as well as from the editor Richard de Grijs. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344 and was supported by the Laboratory Directed Research and Development Program at LLNL under project 17-ERD-001, by the University of Chicago through the Chamberlin Postdoctoral Fellowship (to P.B.), by NASA through grants NNX15AF78G, 80NSSC17K0250, and 80NSSC17K0251 (to A.M.D.), grants NNX17AE86G, NNX17AE87G, NNX15AJ25G (to N.D.), grants NNX11AG78G, NNX14AI19G (to G.R.H.), and by the NSF through grants EAR1502591 and EAR1444951 (to N.D.). LLNL-JRNL-740364.