MAGNESIUM ISOTOPE EVIDENCE FOR SINGLE STAGE FORMATION OF CB CHONDRULES BY COLLIDING PLANETESIMALS

Chondrules are millimeter- to centimeter-sized spherical objects formed as molten droplets during transient heating events in the first few million years of the solar system (Taylor et al. 1983; Grossman 1988; Hewins 1997; Zanda 2004; Krot et al. 2009). Judging by their sheer abundance in chondrite meteorites, they must reflect one of the most energetic process that operated in the early solar system. Chondrules are mainly composed of olivine, low-Ca pyroxene, and glassy or microcrystalline mesostasis, with textures varying between porphyritic, barred/skeletal, radial, and cryptocrystalline (Gooding & Keil 1981; Hewins 1997; Connolly & Love 1998; Krot et al. 2009). Although it has long been accepted that chondrule formation began ∼2 Myr after condensation of the solar system's first solids, calcium–aluminum–rich inclusions (CAIs; MacPherson et al. 1995; Swindle et al. 1996; Kurahashi et al. 2008; Villeneuve et al. 2009; Kita & Ushikubo 2012), a recent absolute chronology of chondrules based on U-corrected Pb-Pb dating, indicate ages ranging from 4567.32 ± 0.42 to 4564.71 ± 0.30 Myr (Connelly et al. 2012). These data establish that chondrule formation began contemporaneously with CAIs and lasted at least ∼3 Myr. Currently, proposed heat sources for the thermal processing of chondrule precursors include shock waves (Boss & Graham 1993; Connolly & Love 1998; Hood 1998), current sheets (Joung et al. 2004), colliding molten planetesimals (Sanders & Taylor 2005; Asphaug et al. 2011), x-winds (Shu et al. 1997), and magnetized disk winds (Salmeron & Ireland 2012).

Of interest are chondrules from the CB metal-rich carbonaceous chondrites, characterized by unusual chemical and petrologic features. In contrast to chondrules preserved in other classes of chondrites, chondrules from CB chondrites have exclusively non-porphyritic (skeletal olivine and cryptocrystalline) textures and magnesium-rich compositions (Weisberg et al. 2001). The CB chondrules lack relict grains and coarse-grained igneous rims, indicative of multiple melting events recorded by chondrules (Rubin 2000). In addition, the CB chondrules lack fine-grained matrix-like rims, suggesting formation in a dust-poor environment (Krot et al. 2001). The CB chondrites are subdivided into two subgroups, CB_a and CB_b, comprising nine meteorites (Bencubbin, Weatherford, Fountain Hills, Gujba, North West Africa (NWA) 1814, HH237, Queen Alexandra Range (QUE) 94411, and NWA 4025). Members of the CB_b subgroup (HH237 and QUE 94411) differ from the CB_a representatives by being finer-grained and having chemically zoned Fe,Ni-metal grains. Recent age-dating of individual chondrules from Gujba suggest single stage formation of these objects ≲4562 Myr (Krot et al. 2005; Bollard et al. 2013), making them the youngest dated chondrules. Moreover, cryptocrystalline chondrules from CB_b chondrites record a uniform Δ¹⁶O composition, indicating that they formed from an isotopically homogeneous reservoir (Weisberg et al. 2001; Krot et al. 2010). This is interpreted as reflecting the formation of CB chondrules in a single stage energetic event, possibly by an impact between planetary embryos after dust in the protoplanetary disk dissipated (Krot et al. 2005). However, it has been suggested that the young age of CB chondrules does not necessarily rule out their formation in an X-wind-type environment during the late evolutionary stage of the protoplanetary disk (Gounelle et al. 2007).

With a half-life of ∼730,000 yr, the ²⁶Al-to-²⁶Mg system is a powerful tool to date solid formation in the early solar system. Moreover, Mg-isotopes can be fractionated in a predictable manner during condensation and evaporation processes such that the ²⁵Mg/²⁴Mg and ²⁶Mg/²⁴Mg ratios track genetic relationships between solar system reservoirs. A recent high-precision Mg-isotope study of inner solar system solids, asteroids, and planets has identified the existence of widespread heterogeneity in the mass-independent ²⁶Mg composition (²⁶Mg*) of bulk solar system reservoirs with near-solar Al/Mg ratios (Larsen et al. 2011). This variability may represent the heterogeneity in the initial abundance of ²⁶Al across the protoplanetary disk. Regardless of how this mass-independent Mg-isotope variability is interpreted, it can be used to explore the formation history of CB chondrules. For example, single stage formation of CB chondrules during a highly energetic event such as a giant impact would result in isotopic homogenization. Given the low inferred initial ²⁶Al abundance at the time of the CB forming event (∼5 × 10⁻⁷), these objects are predicted to record identical μ²⁶Mg* values regardless of their ²⁷Al/²⁴Mg and/or ²⁵Mg/²⁴Mg values. Thus, Mg-isotope measurements of CB chondrules can be used to explore their formation history, but the precision required for this approach to be exploited has been hitherto unattainable. Here, we take advantage of improved methods for Mg-isotope measurements to investigate the formation of CB_b chondrules from the HH237 carbonaceous chondrite.

A piece of the CB_b chondrite HH237 (Figure 1(a)) was imaged in a backscattered electron mode and mapped in Mg, Ca, and Al Kα X-rays at the University of Hawaii at Manoa. Chondrules were selected based on size to yield sufficient Mg for precise isotopic analysis, which resulted in a selective sampling of skeletal chondrules. Therefore, the smaller cryptocrystalline chondrules (Krot et al. 2010) were not sampled. Based on elemental maps, 10 chondrules were selected to ensure an acceptable range of ²⁷Al/²⁴Mg compositions and sampled using a computer-assisted micro-drill. Similarly, we sampled three chondrules from the CB_a chondrite Gujba for comparison. Analytical procedures for the purification of Mg and isotope analysis by high-resolution, multi-collector, inductively coupled, plasma mass spectrometer are described in Bizzarro et al. (2011). The Al/Mg ratios were measured using a ThermoFisher X-series II ICPMS and are accurate to 2%. Uncertainties shown in graphical displays reflect the external reproducibility or internal precision, whichever is larger.

Zoom In Zoom Out Reset image size

Mass-dependent fractionation of Mg-isotopes can be modeled by different fractionation laws, including kinetic and equilibrium mass fractionation (Young & Galy 2004). The fractionation laws can be described by a relationship between the fractionation factors for the isotope ratios. For magnesium, the three isotopes ²⁴Mg, ²⁵Mg, and ²⁶Mg define two fractionation factors:

$$\alpha _{25/24} = {[^{25}\mathrm{Mg}/^{24}\mathrm{Mg}]_a \over [^{25}\mathrm{Mg}/^{24}\mathrm{Mg}]_b},$$

$$\alpha _{26/24} = {[^{26}\mathrm{Mg}/^{24}\mathrm{Mg}]_a \over [^{26}\mathrm{Mg}/^{24}\mathrm{Mg}]_b},$$

where a and b refer to different magnesium reservoirs. The relationship between the two fractionation factors can be expressed as:

$$\alpha _{25/24} = (\alpha _{26/24})^{\beta }.$$

The exponent β is a function of the mass of the isotope species and determines the mass-dependence of the laws and the slope of the mass fractionation line in three-isotope plots (ln[²⁵Mg/²⁴Mg] versus ln[²⁶Mg/²⁴Mg]) (Wombacher & Rehkämper 2003). In a generalized form (Maréchal et al. 1999), the exponent β is related to the isotope masses (m_i) by the following relationship:

$$\beta = {\left(m_1^n-m_2^n\right) \over \left(m_1^n-m_3^n\right)},$$

where the exponent n is varied to quantify a range of different mass-dependencies. The equilibrium law (β = 0.521) and the power law (β = 0.501) are particular cases of this generalized form with n values of −1 and 1, respectively, whereas the kinetic law (β = 0.511) is approximated with n→0. The standard-sample bracketing technique provides a means to correct for the instrumental mass fractionation induced during analysis by MC-ICPMS, thereby allowing us to characterize the potential natural mass fractionation of a sample relative to the standard. The reference material typically used in Mg-isotope studies is the DSM-3 standard (Young & Galy 2004). However, DSM-3 is isotopically heavier than the Bulk Silicate Earth (BSE) by ∼150 ppm, which suggest that this standard has been fractionated during the purification process (Bizzarro et al. 2011). Thus, to minimize potential inaccuracies in the mass bias corrected ²⁶Mg/²⁵Mg ratios resulting from inappropriate correction of the natural mass-dependent fractionation of the reference standard, we elected to use a reference material more representative of BSE, namely the DTS-2 rock standard. We dissolved a large amount of DTS-2 and purified the Mg from four aliquots of this dissolution using our chemical purification scheme. The DTS-2 aliquots were first analyzed against DSM-3 to ensure that their compositions were identical to earlier studies (Table 1), and then combined and used as reference material for the analysis of unknowns.

The 10 skeletal chondrules selected for study show variability in both ²⁷Al/²⁴Mg and μ²⁵Mg values, ranging from 0.09 to 0.45 and from −33 ± 17 ppm to 292 ± 7.1 ppm, respectively (Figure 2(b)). The observed variability in ²⁷Al/²⁴Mg and μ²⁵Mg values are correlated such that chondrules with high ²⁷Al/²⁴Mg ratios record heavy magnesium stable isotope composition relative to BSE. Based on a similar relationship for chondrules from Allende, Galy et al. (2000) suggested that the ²⁷Al/²⁴Mg–μ²⁵Mg variability reflects the admixing of isotopically heavy and aluminum-rich CAI material to a normal chondritic composition. However, as noted by Gounelle et al. (2007), there is no unique CAI composition. Moreover, some CAIs are characterized by an isotopically light magnesium composition (Larsen et al. 2011) such that admixing variable amounts of diverse CAIs to a chondritic magnesium composition would not result in a binary relationship. We conclude that the correlated ²⁷Al/²⁴Mg–μ²⁵Mg variability must have been imparted during chondrule formation and, thus, provides insights into the CB_b chondrule-forming process.

Zoom In Zoom Out Reset image size

To understand the significance of the correlated ²⁷Al/²⁴Mg–μ²⁵Mg variability, we explore potential variations in the mass-independent bulk ²⁶Mg composition (μ²⁶Mg*). Variations in ²⁶Mg* can reflect (1) the in situ decay of ²⁶Al in chondrules with variable ²⁷Al/²⁴Mg ratios (2) ²⁶Al and/or Mg-isotope heterogeneity of chondrule precursors, or (3) inappropriate correction for the natural mass-dependent fractionation. Accepting the age of 4562.52 ± 0.44 Myr for CB chondrules (Bollard et al. 2013), we calculate a radiogenic ingrowth of ²⁶Mg (μ²⁶Mg*) from the decay of ²⁶Al of only ∼1.3 ppm in the chondrule with the highest Al/Mg value if its precursor formed with the canonical ²⁶Al/²⁷Al ratio of (5.252 ± 0.019) × 10⁻⁵ (Larsen et al. 2011) and the solar ²⁷Al/²⁴Mg of 0.09781 ± 0.00029 (Paton et al. 2012). Thus, no variability related to the in situ decay of ²⁶Al is expected to be detected in the selected chondrules within the uncertainty of our measurements. Although heterogeneity in the initial abundance of ²⁶Al amongst chondrule precursors will be solely reflected in the abundance of ²⁶Mg, Mg-isotope isotope heterogeneity can, in principle, affect all magnesium isotopes masses. However, because the ²⁶Mg/²⁴Mg ratios are internally normalized to a fixed ²⁵Mg/²⁴Mg value during analysis, both ²⁶Al and Mg-isotope heterogeneity will be reflected by variations in the μ²⁶Mg* values. Alternatively, μ²⁶Mg* variations can be imparted by inappropriate correction for the natural mass-dependent fractionation affecting individual chondrules. In this case, the apparent excesses or deficits will be correlated with the magnitude of the mass-dependent fractionation.

We show in Figure 2(a) that calculating μ²⁶Mg* with the kinetic mass fractionation law returns values ranging from −8.1 ± 1.2 to −1.6 ± 3.4. Given our external reproducibility of 2.5 ppm for the μ²⁶Mg* value (Bizzarro et al. 2011), this suggests the presence of μ²⁶Mg* heterogeneity amongst the chondrules investigated. However, the variability in μ²⁶Mg* is negatively correlated to the magnitude of the mass-dependent fractionation, defining a slope and intercept of −0.0138 ± 0.0047 and −3.82 ± 0.94, respectively. In contrast, correcting the data instead using the equilibrium fractionation law results in a positive correlation with a slope and intercept of 0.0231 ± 0.0042 and −3.89 ± 0.92, respectively (Figure 2(b)). These observations establish that the μ²⁶Mg* variability obtained using different fractionation laws predominantly reflects inappropriate correction for the natural mass-dependent fractionation experienced by the CB_b chondrules and, therefore, our new high-precision μ²⁶Mg* supports a single stage formation of the CB_b chondrules from a homogenous magnesium isotope reservoir. Moreover, these data demonstrate that the mass-dependent fractionation experienced by the CB_b chondrule population can neither be modeled by pure kinetic nor equilibrium fractionation laws.

The correlated variability between the ²⁷Al/²⁴Mg and μ²⁵Mg values reported in Figure 1(b) is consistent with an origin via volatility-controlled processes. Based on a similar relationship for chondrules from HH237 and QUE94411, Gounelle et al. (2007) proposed that the observed correlated variability reflects evaporation under conditions that prevented Mg-isotope fractionation. The isotopic fractionation associated with the ideal evaporation of magnesium atoms in an open system is expected to follow a Rayleigh-type fractionation characterized by a β value of 0.5160 (Richter 2004). Using a β of 0.5160 to calculate the μ²⁶Mg* values still results in correlated μ²⁶Mg*–μ²⁵Mg variability, indicating that the natural mass-dependent fractionation observed in the chondrule population cannot be modeled by an ideal Rayleigh-type fractionation (Figure 2(c)). A best fit through our data to minimize scatter in μ²⁶Mg* returns a β of 0.5142 and 2sd of 2.6 ppm for the μ²⁶Mg* values of the chondrule population (Figure 2(d)). The scatter of 2.6 ppm is comparable to the external reproducibility of 2.5 ppm inferred for our approach, confirming that the CB_b chondrule precursors formed from an homogenous Mg-isotope reservoir. The β of 0.5142 deduced from our chondrule data is identical to the experimentally determined value of 0.51400 ± 0.00024 from melt evaporation experiments under vacuum conditions (Davis et al. 2005). In Figure 3(a), we show that using a β value of 0.5142 to correct the μ²⁶Mg* data returns a ²⁶Al/²⁷Al upper limit that is consistent with the absolute age of 4562.43 ± 0.40 Myr inferred for the chondrule-forming event. In contrast, using a kinetic (Figure 3(b)) or equilibrium (Figure 3(c)) fractionation law results in, respectively, a negative slope in the ²⁶Al–²⁶Mg isochron diagram, or an age of <3 Myr after CAI formation (assuming ²⁶Al homogeneity), which is inconsistent with the absolute age.

Zoom In Zoom Out Reset image size

If the ²⁷Al/²⁴Mg variability observed in CB_b chondrules predominantly reflect evaporative loss of magnesium, then the chondrule with the highest ²⁷Al/²⁴Mg suggest ∼78% of evaporative loss assuming a chondritic starting composition with the solar ²⁷Al/²⁴Mg of 0.09781 ± 0.00029 (Paton et al. 2012). The magnitude of the predicted mass-dependent magnesium isotope fractionation associated with such evaporative loss from a molten silicate sphere into a low pressure environment is ∼2.5% per atomic mass unit, which is substantially greater than the observed value of ∼300 ppm. Thus, similar to previous work (Alexander et al. 2000; Galy et al. 2000; Gounelle et al. 2007), our results support the view that significant suppression of isotope fractionation occurred during chondrule formation. The lack of isotopic fractionation in chondrules has been attributed to diffusion-limited evaporation (Young 2000), a high surrounding gas pressure (Galy et al. 2000) or, alternatively, evaporation at high magnesium partial pressure (Young & Galy 2004; Cuzzi & Alexander 2006). However, diffusion-limited evaporation or a high surrounding H₂ gas pressure will only limit the magnitude of isotopic fractionation but not prevent it (Richter et al. 2011). Our HH237 chondrules data suggest that about 99% of the predicted isotope fractionation was suppressed. Taken at face value, this is consistent with the suppression of magnesium isotopic fractionation due to evaporation at high magnesium partial pressure, which may have been reached in a region of sufficiently high chondrule density such that partial evaporation was sufficient to maintain a magnesium-saturated surrounding gas (Young & Galy 2004; Cuzzi & Alexander 2006).

Single stage formation of CB chondrules is best understood in the framework of the giant impact model, whereby chondrules and metal grains in CB chondrites formed from a plume produced by a large-scale asteroidal collision (Krot et al. 2005). Such an energetic impact event is expected to form a two-phase silicate disk consisting of a melt layer surrounded by a vapor atmosphere (Visscher & Fegley 2013). Krot et al. (2005) suggested that skeletal chondrules in Gujba and HH237 represent crystalline products of the melt fraction produced by this event, whereas the cryptocrystalline chondrules and chemically zoned metal grains in HH237 reflect condensates from the vapor phase. Our data suggest that significant evaporative loss of Mg occurred during melting of the CB_b skeletal chondrules, which was accompanied by limited mass-dependent isotopic fractionation resulting in a modest enrichment in the heavy isotopes of magnesium in the evaporative residue. This would have resulted in light Mg-isotope enrichment of the coexisting vapor phase such that condensates from this phase are predicted to record a light stable Mg-isotope composition compared to skeletal chondrules. Although we have not analyzed cryptocrystalline chondrules in our study, Gounelle et al. (2007) report that cryptocrystalline chondrules in HH237 are systematically enriched in the light isotope of magnesium. Thus, currently available Mg-isotope data for skeletal and cryptocrystalline chondrules from CB_b chondrites are consistent with their origin as melt products and condensates, respectively, from an energetic impact event.

The existence of μ²⁶Mg* heterogeneity amongst bulk solar system reservoirs with near-solar Al/Mg ratios has been attributed to the heterogenous distribution of ²⁶Al (Larsen et al. 2011). The inferred variations in the initial abundance of ²⁶Al correlate with the ⁵⁴Cr/⁵²Cr ratios for the same reservoirs (Larsen et al. 2011). The μ²⁶Mg* composition of the bulk CB parent body deduced from our chondrule dataset is −3.85 ± 0.96 ppm, which is significantly lower than the bulk solar system value of 4.50 ± 1.10 ppm (Larsen et al. 2011). Mg-isotope data for bulk Gujba chondrules display similar negative μ²⁶Mg* values as HH237 (Table 1). Thus, accepting that the ⁵⁴Cr composition of the CB parent body can be approximated by that of Gujba (Trinquier et al. 2007), we note that the CB parent body lies off the μ²⁶Mg*–⁵⁴Cr correlation line (Figure 4). As such, its Mg and Cr isotope composition cannot be explained by selective thermal processing of presolar carriers, which is predicted to result in a binary mixing relationship (Trinquier et al. 2009; Larsen et al. 2011; Paton et al. 2013). Rather, these data suggest admixing of an additional component with either a distinct Mg-isotope composition or, alternatively, a lower initial abundance of ²⁶Al.

Zoom In Zoom Out Reset image size

The ²⁶Al–²⁶Mg systematics of CAIs from the metal-rich CH and CB chondrites indicate the existence of inclusions with low initial abundance of ²⁶Al (Gounelle et al. 2007; Krot et al. 2008), perhaps with ²⁶Al/²⁷Al as low as ∼4 × 10⁻⁷. In particular, the Mg-isotope composition of CAIs from HH237 is not consistent with a late-stage disturbance of their ²⁶Al–²⁶Mg systematics, but instead requires that their precursors formed in the absence of ²⁶Al. Although the chronology of ²⁶Al-poor inclusions has long been debated, a recent report on the ¹⁸²Hf–¹⁸²W systematics of a ²⁶Al-poor CAI from Allende indicate that this object formed coevally with CAIs recording the canonical ²⁶Al/²⁷Al value of ∼5 × 10⁻⁵ (Holst et al. 2013). This indicates that ²⁶Al-poor inclusions formed early, namely prior to or during admixing of ²⁶Al to the protoplanetary disk (Sahijpal & Goswami 1998). The presence of high-temperature components, believed to have originated in the innermost protoplanetary disk, within the accretion regions of chondrite meteorites and comets suggest that an efficient mass transport process was operative during CAI formation (Krot et al. 2009). Efficient outward transport during the earliest stages of solar system evolution would have resulted in the delivery of a significant fraction of the ²⁶Al-poor material formed prior to admixing of ²⁶Al to the protoplanetary disk to the outer solar system and, by extension, to the accretion region of cometary bodies. We note that analysis of the Coci refractory particle returned from comet 81P/Wild 2 did not show detectable ²⁶Al at the time of its crystallization (Matzel et al. 2010). Although this has been interpreted as reflecting late formation of this object, a more likely explanation given the results of Holst et al. (2013) is that Coci formed prior to admixing of ²⁶Al to the protoplanetary disk.

The preservation of ²⁶Al-poor CAIs in CB and CB/CH chondrites suggest that these meteorites accreted from precursor material that included an early formed, ²⁶Al-free component. A possibility is that the collisional event proposed for the origin of CB chondrites involved a body that predominantly accreted from material formed prior to the admixing of ²⁶Al to the protoplanetary disk, possibly cometary in origin. Admixing of a considerable fraction of ²⁶Al-free material to CB_b chondrule precursors would result in the observed deficit in ²⁶Mg* for the CB_b parent body and, hence, decoupling of the ⁵⁴Cr and ²⁶Mg* systematics. We note that the Isheyevo metal-rich chondrite (CH/CB), a meteorite with a mineralogy and texture akin to CH and CB chondrites, contains pristine lithic clasts with extreme ¹⁵N-isotope enrichments (Bonal et al. 2010). Similar enrichments in heavy nitrogen are spectroscopically observed in cometary comas, suggesting that these clasts may have formed in the outer solar system. If correct, we predict that lithic clasts enriched in heavy nitrogen present in Isheyevo will record μ²⁶Mg* values similar to the solar system's initial μ²⁶Mg* composition of ∼−16 ppm defined by the CV CAI bulk isochron (Larsen et al. 2011). Thus, in agreement with earlier studies (Briani et al. 2009; Bonal et al. 2010), our results and interpretation suggest that metal-rich chondrites contain pristine outer solar system material, possibly providing a window into the composition of comets.

The Centre for Star and Planet Formation is financed by the Danish National Science Foundation (DNRF97). We thank the referee for constructive and thoughtful comments.