²⁶Al AND THE FORMATION OF THE SOLAR SYSTEM FROM A MOLECULAR CLOUD CONTAMINATED BY WOLF–RAYET WINDS

Primitive meteoritic materials contain compelling evidence for the short-lived radionuclides (SLRs) ¹⁰Be, ²⁶Al, ³⁶Cl, ⁴¹Ca, ⁵³Mn, ⁶⁰Fe, ¹⁰⁷Pd, and ¹⁸²Hf in the early solar system (Goswami et al. 2005). These radionuclides have a half-life τ_1/2 < 10 Myr and are potential high-resolution chronometers of events during the epoch of planet formation (Kita et al. 2005; Krot et al. 2008b), but only if they were introduced at discrete times and were uniformly distributed in the solar system. The potential sources of these radionuclides inform us about the young Sun's stellar neighborhood (Hester et al. 2004), or its magnetic interaction with an accretion disk (Shu et al. 1997). The decay of one SLR, ²⁶Al, might have been the principal heat source in planetesimals and responsible for the differentiation of the parent bodies of magmatic iron meteorites in the first 1–2 Myr of the solar system (Greenwood et al. 2005; Scherstén et al. 2006; Markowski et al. 2006).

The origin of SLRs is controversial. The half-life of each is much shorter than the ∼100 Myr mixing time of the interstellar medium (de Avillez & Mac Low 2002) and the excess abundances of at least five radionuclides (¹⁰Be, ³⁶Cl, ²⁶Al, ⁴¹Ca, and ⁶⁰Fe) require one or more "local" sources in addition to the average Galactic background (Jacobsen 2005). Two principal scenarios emerged soon after the first reports of fossil SLRs in meteorites: (1) an origin in one or more neighboring massive stars, either Type II supernova (SN) progenitors (Cameron & Truran 1977) or Wolf–Rayet (W-R) stars (Arnould & Prantzos 1986) and (2) production by irradiation of gas or dust with energetic particles from the active young Sun (Heymann & Dziczkaniec 1976). An alternative scenario invoking an origin in a nearby intermediate-mass asymptotic giant branch (AGB) star (Wasserburg et al. 1994) is generally discounted because of the very low probability of such an encounter (Kastner & Myers 1994). Each of the two schools of thought has developed elaborate models (Lee et al. 1998; Gounelle et al. 2006; Ouellette et al. 2007) but neither has produced a comprehensive explanation of the origin, abundance, and distribution of all the SLRs (Goswami et al. 2005; Gounelle et al. 2006; Duprat & Tatischeff 2007).

The origins of three SLRs seem unambiguous: (1) ¹⁰Be (τ_1/2 ∼ 1.5 Myr) inferred from excess ¹⁰B (McKeegan et al. 2000) could have been produced by energetic particle irradiation but not by stellar nucleosynthesis. Although magnetic trapping of Galactic cosmic rays in the protosolar molecular cloud has been advanced as an alternative explanation (Desch et al. 2004), it is inconsistent with variations in the inferred initial ¹⁰Be/⁹Be ratio (McKeegan et al. 2000; Sugiura et al. 2001; Marhas et al. 2002; MacPherson et al. 2003). (2) Excess ³⁶S correlated with the ratio ³⁵Cl/³⁴S and attributable to the decay of ³⁶Cl (τ_1/2 ∼ 0.3 Myr) was reported in sodalite, an alteration phase that replaced anorthite in calcium–aluminum-rich inclusions (CAIs) and chondrules from CV chondrites (Lin et al. 2005; Hsu et al. 2006). The inferred ³⁶Cl/³⁵Cl ratio at the time of sodalite formation is 5 × 10⁻⁶. If the sodalite formed late (greater than 1.5 Myr after CAI crystallization, based on absence of ²⁶Al), the initial ³⁶Cl/³⁵Cl ratio was >1.6 × 10⁻⁴. This is inconsistent with a massive stellar source and requires a late episode of irradiation. (3) ⁶⁰Ni excess correlated with the ⁵⁶Fe/⁵⁸Ni ratio is evidence for live ⁶⁰Fe, a radionuclide that cannot be produced by irradiation and must have originated in one or more massive stars (Tachibana & Huss 2003). High-precision nickel isotope measurements in several groups of magmatic iron meteorites indicate that ⁶⁰Fe was uniformly distributed in the solar nebula (Dauphas et al. 2008) but its initial abundance, (⁶⁰Fe/⁵⁶Fe)₀ ∼ (0.5–1) × 10⁻⁶, is uncertain (Tachibana et al. 2006; Bizzarro et al. 2007; Quitté et al. 2007; Guan et al. 2007). The lower end of the range of estimates is consistent with the expected abundance in star-forming regions if star formation rates were approximately twice as high at the epoch of solar system formation (Williams 2008, Gounelle & Meibom 2008a) and/or the half-life of ⁶⁰Fe is actually longer than the published value (τ_1/2 ≈ 1.5 Myr).

Ironically, the origin of ²⁶Al (τ_1/2 ∼ 0.73 Myr), the first SLR to be discovered (Lee et al. 1976) and the one best studied, remains an enigma. A significant contribution by irradiation is disputed on the grounds that (1) production models adjusted to achieve the required ²⁶Al/²⁷Al ratio overpredict the observed initial abundance of ⁴¹Ca (τ_1/2 ∼ 0.1 Myr) whose co-occurrence indicates a common origin (Sahijpal & Goswami 1998; Goswami et al. 2005), (2) the flux of energetic particle inferred from X-ray observations of young solar-type stars would have been insufficient to produce the ²⁶Al (Duprat & Tatischeff 2007), and (3) a lack of correlation between the presence of ²⁶Al and ¹⁰Be, which was formed by irradiation (MacPherson et al. 2003; Marhas & Goswami 2004). Finally, the canonical value [(4.5–5) × 10⁻⁵] of the initial ²⁶Al/²⁷Al ratio in the majority of CAIs from primitive chondrites (MacPherson et al. 1995; Jacobsen et al. 2008; Makide et al. 2008), the consistent chronology of CAI and chondrule formation between ²⁰⁷Pb–²⁰⁶Pb and ²⁶Al–²⁶Mg systematics (Amelin et al. 2002; Halliday & Kleine 2006; Connelly et al. 2008), and the apparently uniform Mg–isotope compositions of bulk chondrites, Mars, Moon, and the Earth (Thrane et al. 2006) are evidence for a uniform distribution thought inconsistent with a central (solar) irradiation source.

Although these observations favor an origin of ²⁶Al in massive stars, the mechanism and timing of its delivery to the early solar system remains unclear. Two models have been proposed: instability-induced mixing of gaseous SN ejecta during the molecular cloud core phase (Cameron & Truran 1977) and injection of SLR-bearing dust grains into the later protoplanetary disk (Ouellette et al. 2005). The relatively small cross section of the solar nebula dictated that the progenitor was within ∼1 pc of the solar system (Looney et al. 2006). Such a circumstance is possible only in the dense environment of a large stellar cluster (Hester et al. 2004). Incorporation of hot, low-density gas from SN ejecta into the denser, cooler disk gas is inefficient (Vanhala & Boss 2002; Ouellette et al. 2007; Boss et al. 2008). Instead, Ouellette et al. (2005) proposed that several SLRs, including ²⁶Al, were delivered to the early solar system as grains that condensed from SN ejecta and vaporized upon entering the relatively high-density gas in the disk (Ouellette et al. 2007). Disks around low-mass cluster stars persist for up to 6 Myr (Haisch et al. 2005; Jayawardhana et al. 2006), longer than the main-sequence lifetime of the most massive SN progenitors (Schaller et al. 1992).

However, any explanation for solar system ²⁶Al invoking a late introduction of SN ejecta has four significant shortcomings : (1) not all stars form in large clusters and most clusters are dynamically unbound and disperse in ∼10 Myr (Lada & Lada 2003). It is statistically unlikely that the Sun would have been sufficiently close to a massive star at the end of the latter's main-sequence life (Williams & Gaidos 2007; Gounelle & Meibom 2008b). (2) Even the most massive stars have a main-sequence lifetime of at least 3 Myr and, if they and the Sun formed simultaneously, the former would have ended in SNe late in the evolution of the solar nebula and probably long after CAIs containing the canonical ²⁶Al/²⁷Al ratio had formed. These CAIs, the oldest dated solids from the solar system (Amelin et al. 2002), have a narrow range of inferred initial ²⁶Al/²⁷Al ratios suggesting that they formed in ≪1 Myr (Thrane et al. 2006; Jacobsen et al. 2008), consistent with the duration of Class 0–I stages of protostars (Smith 2004; Ward-Thompson et al. 2007). A short interval of CAI formation is also consistent with the narrow range of their oxygen isotope compositions (Δ¹⁷O = −24 ±  2% MacPherson et al. 2008; Makide et al. 2008), which are similar to the inferred oxygen isotopic composition of the Sun (Hashizume & Chaussidon 2005). (Later CAI formation would presumably have reflected the rapid oxygen isotopic evolution of the solar nebula along the slope-one carbonaceous chondrite anhydrous mineral (CCAM) line toward the terrestrial value (Δ¹⁷O = 0%), a trend attributed to CO photochemical self-shielding and radial mixing in the disk (Yurimoto et al. 2007; Aléon et al. 2007).) (3) Late injection of ²⁶Al into the protoplanetary disk would likely have disturbed the oxygen isotopic composition of the disk from the CCAM line, and this is not observed (Gounelle & Meibom 2007). (4) SN models that produce the canonical ²⁶Al/²⁷Al ratio invariably overpredict the abundance of ⁵³Mn and must impose fallback of the innermost layers, e.g., Meyer (2005). (These models also overpredict the abundance of ⁶⁰Fe, see Section 5.3.)

A scenario in which SLRs from coeval SN progenitors were injected into the protoplanetary disk can be evaluated for its statistical plausibility. Gounelle & Meibom (2008b) estimated that the probability that any given disk is contaminated by a SN with enough ²⁶Al to reach the canonical ²⁶Al/²⁷Al ratio is ⩽0.3%. However, there are three limitations to their model: (1) it underestimated the probability by assuming a maximum disk radius of 50 AU based on observations and theoretical arguments that disks within 0.1 pc of massive (O) stars suffer from photoevaporation (Johnstone et al. 1998; Chevalier 2000). But disks further from O stars are invariably larger (Vicente & Alves 2005; Andrews & Williams 2007; Balog et al. 2007) and this effect compensates the greater distance from the source of radionuclides. (2) Their model overestimated the probability by neglecting the expansion of the host star cluster and using the ⩽1 Myr old Orion Nebular Cluster as a template, rather than a cluster at the minimum age (3 Myr) when SNe occur. In older clusters, stars are more widely separated. (3) It only included the contribution of SNe to ²⁶Al. ²⁶Al is also produced and ejected from stars with initial masses ⩾40 M_☉ in their luminous blue variable (LBV) and W-R phases (Arnould & Prantzos 1986; Palacios et al. 2005; Sahijpal & Soni 2006).

The uniform distribution of ²⁶Al, the existence of the canonical ²⁶Al/²⁷Al ratio in CAIs, the adherence of primitive oxygen isotope compositions to the CCAM line, and the low likelihood of an SN injection event all point to the introduction of ²⁶Al before the collapse of the protosolar cloud. Indeed, the oxygen isotopic composition of the entire solar system appears to be displaced from the locus of mean Galactic evolution (Young et al. 2008), suggesting primordial contamination. The source of ²⁶Al must also have introduced ⁴¹Ca, which co-occurs with an inferred initial abundance of ⁴¹Ca/⁴⁰Ca ∼1.5 × 10⁻⁸ (Sahijpal & Goswami 1998). The half-life of this radionuclide is only 0.1 Myr, limiting any time delay between production at the source and its incorporation into CAIs. The source of ²⁶Al cannot have been accompanied by substantial ⁵³Mn, as is the case of SN ejecta without fallback onto the remnant (Meyer 2005). (The predicted ⁵³Mn abundance in the interstellar medium is consistent with its inferred initial abundance in the solar system and a "local" source is not required.) Any relationship between the source of ²⁶Al and that of ⁶⁰Fe remains to be determined. There is as yet no evidence that ⁶⁰Fe and ²⁶Al are correlated and had the same origin. Indeed, there may be a deficit of ⁶⁰Fe relative to ²⁶Al compared with SN ejecta that cannot be explained by free decay of the two radionuclides: the initial ratio of ⁶⁰Fe to ²⁶Al in the solar system was 0.1–0.2 (Tachibana et al. 2006), lower than the 0.3 deduced from the Galactic average γ-ray emission (Wang et al. 2007) and theoretical predictions (Limongi & Chieffi 2006; Woosley & Heger 2007).

Very massive (⩾40 M_☉) stars eject ²⁶Al (and other SLRs) during the W-R phase of mass loss near the end of hydrogen core burning, as well as in SN (Arnould & Prantzos 1986). W-R winds might account for a large fraction of the total fluence and Galactic distribution of γ-rays from the decay of ²⁶Al (Palacios et al. 2005; Diehl 2006; Voss et al. 2008; Martin et al. 2008). (The nondetection of γ-ray emission in the decay line of ²⁶Al from the nearest W-R star γ²-Velorum can be explained by the dispersal of most of the radionuclide to large angular separation Mowlavi & Meynet 2006). We propose that most or all of the ²⁶Al in the early solar system originated in W-R winds from one or more massive stars that contaminated the molecular cloud from which the Sun formed. (Sahijpal & Soni 2006 also considered the contribution of W-R winds to solar system inventories of SLRs.) These stars could have been members of the same embedded cluster as the Sun, or, more likely, members of another cluster that formed in the same giant molecular cloud (GMC; Figure 1). An analogous "self-contamination" scenario has been invoked to explain anomalous abundance patterns in some globular clusters (Smith 2006).

Zoom In Zoom Out Reset image size

Our proposal is based upon the following: (1) the amount of ²⁶Al ejected in the winds of a single 60 M_☉ star (Limongi & Chieffi 2006) is sufficient to contaminate 2 × 10⁴ M_☉ of solar-metallicity gas to the canonical ²⁶Al/²⁷Al ratio of the solar system. (2) The W-R phase can occur as soon as 1–2 Myr after star formation and more than 1 Myr before the SN (Palacios et al. 2005), making it more likely to contaminate residual star-forming molecular gas. (3) W-R winds have speeds of up to 2000 km s⁻¹ (Niedzielski & Skórzynński 2002) and can traverse star-forming regions (10–100 pc) in 10⁴–10⁵ yr. (4) In contrast to single clusters where star-formation may be coeval, star formation in a molecular cloud can occur over an interval of at least a few Myr (Hartmann et al. 2001), and possibly longer (Williams et al. 2000) as clumps of gas with a mass spectrum M^−1.7±0.1 (Pudritz 2002) form multiple embedded stellar clusters (Williams et al. 2000). For example, the Orion star-forming complex contains several subgroups that are several Myr older than the Orion Nebula Cluster (Bally 2008). (5) W-R winds contain multiple SLRs, including ²⁶Al, ⁴¹Ca, and ³⁶Cl, but little or no ⁵³Mn or ⁶⁰Fe (Arnould et al. 2006). We propose that the collapse of the protosolar cloud homogenized the distribution of these isotopes (but see Section 5.2). Our scenario does not preclude an SN-triggered collapse (Cameron & Truran 1977; Boss et al. 2008), which would have occurred after the W-R phase.

In Section 2, we revisit the scenario of ²⁶Al introduction by SN into the protoplanetary disk with a Monte Carlo approach that used more realistic disk sizes, accounted for the expansion of clusters, and included the contribution from both W-R winds and SN. While our results differ quantitatively from those of Gounelle & Meibom (2008b), the calculated probability of a disk having the canonical solar system ²⁶Al/²⁷Al ratio is nevertheless small (less than 2%). We also consider Bondi–Hoyle accretion of contaminated intracluster gas, as proposed by Throop & Bally (2008) and find its inclusion does not significantly alter this result (Section 3). We then use a similar Monte Carlo model to investigate a scenario where ²⁶Al is introduced by W-R winds into the parent molecular cloud of the Sun (Section 4). This model readily reproduces the canonical ²⁶Al/²⁷Al ratio of the solar system. We discuss the delivery of ²⁶Al from W-R winds into the molecular cloud, interpret CAIs that lack the canonical ²⁶Al/²⁷Al ratio, and, in the context of the cloud contamination scenario, present additional calculations for three other SLRs (⁴¹Ca, ⁶⁰Fe, and ³⁶Cl) in Section 5.

Our scenario requires a plausible mechanism for the efficient introduction of the ²⁶Al carrier in W-R winds into the surrounding molecular cloud. In the calculations above, we assumed a 100% delivery efficiency but this will clearly not be the case. In general, mixing between the hot, tenuous gas from massive stars and cooler, denser molecular gas is thought to be very inefficient (de Avillez & Mac Low 2002). SN-enriched gas from H ii regions is thought to find its way back into the interstellar medium (if at all) through a circuitous route taking ∼100 Myr (Tenorio-Tagle 2000). Impact of SN ejecta onto a protostellar cloud core may induce mixing via Rayleigh–Taylor and Kelvin–Helmholtz instabilities (Boss & Foster 1998; Boss et al. 2008) but the injection efficiency is too low to explain the solar system's canonical ²⁶Al/²⁷Al ratio.⁶ Gas in W-R winds will be less dense than SN ejecta by several orders of magnitude and efficient mixing in the gas phase is even less likely. Instead, the high-velocity (500–2000 km s⁻¹) winds will develop a reverse shock upon encountering the much denser molecular cloud (Weaver et al. 1977).

We propose that refractory dust grains are the principal carrier of ²⁶Al and ⁴¹Ca in W-R winds and that these can dynamically decouple from the shocked wind and imbed themselves into the surrounding molecular cloud, analogous to the "aerogel" model described by Ouellette et al. (2005). Presolar grains of Al₂O₃ (corundum or other forms) are found in primitive meteorites but their oxygen isotopes indicate a source in red giant or AGB stars, not in very massive stars (Hutcheon et al. 1994; Ott 2007; Nittler et al. 2008). Although many of these grains have large ²⁶Mg excesses produced by the decay of ²⁶Al, the abundance of these grains is insufficient to account for the canonical ²⁶Al/²⁷Al ratio (Hutcheon et al. 1994). In fact, the ²⁶Al in CAIs must have been processed through the gas phase and subsequently recondensed into the refractory inclusions. The presolar grains are also much larger (∼1 μm) than the silicate grains predicted to form from winds and ejecta. Their size may be why they survived the incorporation process and the latter did not.

Simple models of grain nucleation and growth predict oxide grain growth to sizes of 0.01–0.1 μm in SN ejecta (Nozawa et al. 2007). SN are predicted to be copious sources of dust, but observations have so far produced evidence only for a few times 10⁻⁵ M_☉ of dust in individual events (Meikle et al. 2007). Dust production in W-R stars is poorly investigated, although such stars appear to be minor contributors to the overall interstellar dust budget (Tielens et al. 2005). Copious amorphous carbon dust is observed around carbon-rich W-R CO stars and is thought to be the result of colliding stellar winds in binary systems (Crowther 2003). WCO stars do not produce ²⁶Al, but dust production by predecessor LBV and WN phases predicted to contain ²⁶Al in their winds has recently been established (Rajagopal et al. 2007; Barniske et al. 2008). The η Carinae LBV star ejected ∼10 M_☉ of dust-rich material during its 1843 eruption (Smith et al. 2003), including aluminum oxide (de Koter et al. 2005), although its ²⁶Al content has yet to be definitely established by γ-ray observations (Knödlseder et al. 1996).

To reach the molecular cloud, grains must survive sputtering after they pass the reverse shock and move at high speed (∼10³ km s⁻¹) with respect to the shocked gas in the H ii region. They also must not be completely decelerated within the reverse shock zone. These conditions place a lower limit on grain size. Grains too small are sputtered to destruction or eventually vaporized in the hot, shocked gas (Nozawa et al. 2007). The density of the wind 1 pc from a W-R star-losing mass at a rate of 10⁻⁵ M_☉ yr⁻¹ is ∼10⁻² cm⁻³ and the stopping distance of grains even as small as 0.01 μm grains is 10 pc, comparable or larger to the size of H ii regions. The deceleration across the scale of the shocked region (∼1 pc; Weaver et al. 1977) will be low and the fraction of material sputtered from the grains, which is related to the deceleration (Equation (1) in Nozawa et al. 2007) will be likewise small.⁷

Grains that escape the W-R wind will not penetrate far into a GMC. Typical hydrogen number densities in clouds are 10²–10³ cm⁻³ and the stopping distance of 0.01–0.1 μm grains will be only of order ∼10³ AU. The gas densities in portions of the molecular cloud that are shocked and swept up by the expanding wind or SN ejecta will be higher and the stopping distances proportionally shorter. Thus, only the surfaces of clouds will be initially contaminated by ²⁶Al. Further transport of dust grains into cloud depends on their kinematics, which are poorly understood. The smallest scale on which turbulence in clouds can affect the mass distribution and cause mixing is the sonic transition where the turbulent velocity is equal to the sound speed; this is roughly 1 pc in solar-metallicity clouds (Padoan 1995). Thus, mixing might be inefficient at the surfaces of clouds. The degree to which this controls the incorporation of ²⁶Al into new low-mass stars depends on the extent of large-scale mixing and whether star formation is triggered or at least assisted, by the interaction of winds or SN ejecta with cloud gas (Zavagno et al. 2007). In that case, star formation is spatially correlated with ²⁶Al abundance.

W-R progenitors may themselves migrate into and contaminate regions of a molecular cloud where low-mass star formation has yet to occur. 10%–30% of O stars have large peculiar velocities (up to 200 km s⁻¹) relative to most early-type stars as a result either of dissolution of binary systems by SN explosions, or encounters between two binary systems (Hoogerwerf et al. 2001). One of these so-called runaway O stars moving at a typical speed of 30 km s⁻¹ would traverse a molecular cloud in ∼1 Myr. Mass loss from this star, if occurring, could contaminate a larger region of the molecular cloud with SLRs.

Any scenario that explains the canonical solar system ²⁶Al/²⁷Al ratio must also accommodate the exceptions. Several classes of CAIs show either no excess of ²⁶Mg produced by the decay of ²⁶Al or have an inferred ²⁶Al/²⁷Al ratio ≪1 × 10⁻⁵, much lower than the canonical value of (4.5–5)× 10⁻⁵. These include (1) igneous CAIs associated with chondrule-like materials (relict CAIs inside chondrules and CAIs surrounded by chondrule-like, ferromagnesian silicate rims; Krot et al. 2005a, 2005b); (2) some igneous CAIs in metal-rich (CB and CH) carbonaceous chondrites (Gounelle et al. 2007, Krot et al. 2008a); (3) fractionation and unidentified nuclear (FUN) effects CAIs (Lee 1988); (4) isolated platy hibonite crystals (PLACs; Ireland & Fegley 2000); (5) pyroxene–hibonite spherules (Ireland et al. 1991; Russell et al. 1998); (6) some corundum-bearing CAIs (Simon et al. 2002); and (7) most of the grossite- and hibonite-rich inclusions in CH chondrites (Kimura et al. 1993; Weber et al. 1995, Krot et al. 2008a). (1 and 2) CAIs associated with chondrule materials and some igneous CAIs in CB and CH carbonaceous chondrites (Krot et al. 2001, 2005a, 2005b, 2008a) are ¹⁶O-depleted to varying degrees (Δ¹⁷O ranges from −25 to −5) relative to typical CAIs in primitive chondrites which uniformly have ¹⁶O-rich compositions (Δ¹⁷O ∼ −25; Itoh et al. 2004; Makide et al. 2008). We infer that the ²⁶Al-poor and ¹⁶O-depleted CAIs experienced late-stage melting and oxygen isotope exchange, probably during chondrule formation, which could have reset their ²⁶Al–²⁶Mg systematics. (3–7) The lower than the canonical ²⁶Al/²⁷Al ratio in FUN CAIs, PLACs, pyroxene–hibonite spherules, some corundum-bearing CAIs, and most of the grossite- and hibonite-rich inclusions in CH chondrites can be explained by (1) their late formation, after decay of ²⁶Al; (2) their early formation, prior to introduction of ²⁶Al; or (3) the lack of the canonical budget of ²⁶Al in their precursors. Most of these CAIs have ¹⁶O-rich compositions Goswami et al. 2001; Simon et al. 2002; Krot et al. 2008a, 2008c), indistinguishable from those of typical CAIs with the canonical ²⁶Al/²⁷Al ratio. The rapid evolution of the oxygen isotopic composition of the inner solar system (Krot et al. 2005c; Aléon et al. 2007) and the short (less than 10⁵ yr) duration of CAI formation (Thrane et al. 2006; Jacobsen et al. 2008) thus make (1) unlikely. Similar arguments can be used against (2). Although an early formation is possible if complete melting and exchange of oxygen isotopes occurred, this seems unlikely considering evidence for incomplete melting of CAIs (MacPherson et al. 2005).

We infer that the CAIs of categories 3–7 formed contemporaneously with ²⁶Al-rich inclusions. The absence of canonical ²⁶Al in their precursors suggests either they formed prior to homogenization of ²⁶Al in the solar system Sahijpal & Goswami 1998; Krot et al. 2008a, 2008c), or preferential loss of the (uniformly distributed) ²⁶Al carrier during thermal processing of the CAI precursors (Wood 1996). Both explanations can be reconciled with the presence of nucleosynthetic anomalies in some of these CAIs (Lee et al. 1998) if the ²⁶Al carrier contributed a distinct component to the solar system's stable isotope composition (Lee et al. 1998). The second explanation is more speculative, however. It hypothesizes that (1) the precursors of these CAIs was isotopically heterogeneous and, contrary to typical refractory inclusions, escaped a cycle of complete evaporation-condensation; (2) the carrier of ²⁶Al was relatively volatile; and (3) it was lost to varying degrees by sublimation of these CAI precursors prior to their melting. Although these inclusions can be used as an evidence for heterogeneous distribution of ²⁶Al among CAI precursors, the scale of any heterogeneity was probably limited because such inclusions are rare relative to ²⁶Al-rich CAIs.

⁴¹Ca: A further test of the wind model is whether it can also reproduce the inferred initial ⁴¹Ca/⁴⁰Ca ratio of 1.5 × 10⁻⁸ (Sahijpal & Goswami 1998). Published calculations of ⁴¹Ca yields in winds from massive stars are limited. We considered a 60 M_☉ progenitor for which ⁴¹Ca and ²⁶Al yields from the W-R winds were separately published (Arnould et al. 2006; Limongi & Chieffi 2006). We accounted for the additional free decay of ²⁶Al, which is ejected during the hydrogen-burning WN W-R phase, while ⁴¹Ca is produced in the later core He-burning WCO phase 1.9 × 10⁵ yr later. The corrected ratio of ⁴¹Ca to ²⁶Al in the ejecta is ∼25 times higher than in the solar system, but would be consistent if an additional time Δ ≈ 0.5 Myr elapsed before CAI formation. It is interesting that this is approximately the same duration as the W-R phase itself before the final SN Ib/c.

⁶⁰Fe: W-R winds contain negligible amounts of ⁶⁰Fe (Arnould et al. 2006). Live ⁶⁰Fe in the early solar system could have originated in SN from the same early generation of massive stars that produced the ²⁶Al, or in an even earlier generation of stars (Gounelle & Meibom 2008a). We repeated the calculations described in Section 4 but included SN contributions and calculated ⁶⁰Fe/⁵⁶Fe ratios in the same manner as ²⁶Al/²⁷Al, using the yields of Limongi & Chieffi (2006) and the solar iron abundance of Lodders (2003). We assume Δ = 0.5 Myr based on the ⁴¹Ca abundance. In Figure 4, we plot Monte Carlo realizations for different epochs (3–8 Myr before the Sun) for the earlier generation of massive stars. If all SN ejecta is incorporated, the ⁶⁰Fe/⁵⁶Fe is overpredicted by an order of magnitude relative to ²⁶Al/²⁷Al. A comparison with the ratio of the two SLRs (black line) inferred from γ-ray measurements (Wang et al. 2007) suggests that this discrepancy may be in part the result of an overprediction of ⁶⁰Fe yield—or underprediction of ²⁶Al yield—by the nucleosynthesis models (see Section 6). There are two other explanations suggested by Figure 4: (1) the Sun formed 3 Myr after the massive stars, when many massive stars were in the W-R phase but few SN had occurred (a scenario represented by the red dots extending below the primary locus); or (2) the SN contribution was attenuated by an effect such as described in Section 5.1.

Zoom In Zoom Out Reset image size

Explanation (1) is statistically unlikely if star formation is uncorrelated and demands precise timing between the formation of massive stars and the Sun (∼3 Myr later), but could be demanded in a scenario where the Sun's formation was triggered by an SN (Cameron & Truran 1977; Boss et al. 2008). Our simulations indicate that the initial solar ²⁶Al/²⁷Al and ⁶⁰Fe/⁵⁶Fe ratios can be reproduced in this manner only in star clusters with N_*>10⁵ whose most massive members have ∼100 M_☉. As such large clusters are relatively rare, this scenario is a priori less likely. Explanation (2) requires a reduction in the SN contribution to ⁶⁰Fe (and ²⁶Al) by a factor of 20. This could be due to a combination of effects; retention or fallback of the central region of the progenitor (Meyer 2005), inefficient delivery of SN ejecta into the cloud (Section 5.1) or the collapse of the protosolar cloud and a decrease in its cross section by the time the SN ejecta arrived. An alternative scenario (3) is that ⁶⁰Fe in the solar system is the relict of an even earlier episodes of massive star-formation and contamination (Gounelle & Meibom 2008a) of which the ⁴¹Ca and most ²⁶Al has decayed. ⁶⁰Fe will decay to 5% of its initial abundance in 6.5 Myr, during which ²⁶Al decays to 0.2%, and ⁴¹Ca essentially vanishes. This last explanation is viable only if this earliest generation of stars ceased to contribute SLRs to the molecular cloud after ∼6 Myr.

³⁶Cl: We estimated the abundance of ³⁶Cl, which is also ejected by W-R stars during the WCO phase (Arnould et al. 2006), and calculated the ³⁶Cl:³⁷Cl ratio in the same manner as ⁴¹Ca/⁴⁰Ca. We find that our model underpredicts the ratio by at least 3 orders of magnitude. It has already been recognized that stellar nucleosynthetic models cannot account for this isotope, especially if it was introduced at the epoch of CAI formation 1–2 Myr before the host sodalite alteration phases were formed. At the present time, the only viable explanation appears to be a late episode of irradiation by energetic particles (Hsu et al. 2006)