EARLY SOLAR NEBULA CONDENSATES WITH CANONICAL, NOT SUPRACANONICAL, INITIAL ²⁶Al/²⁷Al RATIOS

The presence of the short-lived (t_1/2 = 0.73 Myr) radionuclide ²⁶Al in the early solar system is well established (Lee et al. 1976; MacPherson et al. 1995), but the abundance of ²⁶Al relative to stable ²⁷Al at the time of solar system birth ("initial" ²⁶Al/²⁷Al, or (²⁶Al/²⁷Al)₀) is the subject of renewed debate. The question is important for two reasons. First, ²⁶Al may have been a major heat source for early planetary melting and differentiation. Second, (²⁶Al/²⁷Al)₀ represents "time zero" for a high-resolution relative chronometer of closely spaced events that occurred at ∼4.57 Ga, the birth of the solar system.

Evidence for extinct ²⁶Al, as excesses of its decay product ²⁶Mg, is found in primordial solar system materials that are preserved in chondritic meteorites and interplanetary dust particles. The highest inferred ²⁶Al/²⁷Al ratios occur in calcium–aluminum-rich inclusions (CAIs), which are the earliest solar system solids and which define (²⁶Al/²⁷Al)₀. Lower inferred ratios are found in objects such as chondrules, which thus probably formed later than CAIs (Krot et al. 2009 and references therein).

Internal mineral isochrons measured in CAIs by secondary ion mass spectrometry (SIMS) over the past ∼30 years have yielded a relatively consistent ("canonical") range of (²⁶Al/²⁷Al)₀ ∼ (4–5) × 10⁻⁵ (MacPherson et al. 1995). Since the development of multicollection techniques for both SIMS and laser-ablation inductively coupled plasma mass spectrometry (LA-ICPMS), there have been several reports of "supracanonical" isochrons yielding (²⁶Al/²⁷Al)₀ up to 7 × 10⁻⁵ (Taylor et al. 2005; Young et al. 2005; Cosarinsky et al. 2007), but none is unambiguous (see the discussion). Two recent studies (Bizzarro et al. 2005; Thrane et al. 2006) reported (²⁶Al/²⁷Al)₀ = (5.85 ± 0.05) × 10⁻⁵, based on a whole-rock isochron from high-precision magnesium isotope and Al/Mg analyses of bulk CAIs. Because this whole-CAI supracanonical value differs from those generally determined from internal mineral isochrons, it has been asserted (e.g., Thrane et al. 2006) that the supracanonical value records the time of primary aluminum–magnesium chemical fractionation during nebular condensation of the CAI precursors. According to this model, the canonical value (²⁶Al/²⁷Al)₀ ∼ (4–5) × 10⁻⁵ determined from internal isochrons records only the time that each CAI was last melted or otherwise reprocessed (Thrane et al. 2006; Young et al. 2005); thus, the true "time zero" should be based on (²⁶Al/²⁷Al)₀ = (5.85 ± 0.05) × 10⁻⁵. One obvious test of this hypothesis is to locate and analyze samples of the precursors to CAIs, i.e., primary nebula condensates. Nebular condensates exist, in the form of fine-grained, spinel-rich CAIs from reduced CV3 chondrites such as Leoville and Efremovka. These distinctive CAIs have complex and delicately layered structures that are not compatible with an igneous origin (Krot et al. 2004). Their defining characteristic is a volatility-fractionated trace-element pattern ("Group II") that can only be explained by high-temperature fractional condensation of the nebular gas (Boynton 1975; Davis & Grossman 1979). Because they preserve intact the unmistakable chemical and textural imprints of actual condensation, these fine-grained CAIs should preserve the initial ²⁶Al/²⁷Al ratios at the time of condensation. Thus they are ideal candidates on which to test the supracanonical model. The small grain sizes of such objects (typically ⩽∼10 μm) have made isotopic studies difficult in the past, but the greater sensitivity of the newest generation of SIMS instruments has enabled smaller beam spots to be used to acquire high-precision data.

We measured magnesium isotopes in one fine-grained, spinel-rich CAI from the Leoville CV3 chondrite, labeled Leoville 3536-1 (Figure 1; analyzed and described by Mao et al. 1990 and Krot et al. 2004), whose somewhat larger average grain size than most similar objects makes it particularly amenable to high-precision microbeam isotopic analysis. It consists entirely of the primary high-temperature phases melilite, spinel, anorthite, and aluminous diopside; it is devoid of alkali- and halogen-bearing secondary phases that are abundant in Allende CAIs.

Zoom In Zoom Out Reset image size

We used both laser-ablation multicollector inductively coupled plasma mass spectrometry (LA-MC-ICPMS; see Appendix A) and IMS-1280 large radius SIMS (see Appendix B). The LA-MC-ICPMS analyses were collected using a ∼90 μm laser spot size which, for such a fine-grained object, resulted in multiple mineral grains within each spot. The modal mineralogy of each laser spot was determined using image analysis of backscattered electron (BSE) images and element X-ray area maps taken with a scanning electron microscope (SEM). The SIMS analyses utilized 7 μm beam spots for melilite and 12 μm beam spots for spinel and pyroxene, thus permitting analysis of a single phase within most spots.

The aluminum–magnesium isotopic data for Leoville 3536-1 are given in Table 1 and plotted on an isochron diagram in Figure 2. The LA-ICPMS data for multi-mineral spots define a correlation line between ²⁶Mg/²⁴Mg versus ²⁷Al/²⁴Mg corresponding to (²⁶Al/²⁷Al)₀ = (5.25 ± 0.86) × 10⁻⁵, with an intercept (initial δ²⁶Mg) of 0.06‰ ± 0.21‰, MSWD = 1.8, and ²⁷Al/²⁴Mg values up to ∼7.6. Because of the low ²⁷Al/²⁴Mg values, the slope of the LA-ICPMS isochron is not resolved from those of either the canonical (∼5 × 10⁻⁵) or supracanonical (5.85 ± 0.05) × 10⁻⁵ ones. However, the SIMS data for individual mineral phases extend to much higher ²⁷Al/²⁴Mg values (up to 45) and define a precise correlation line yielding (²⁶Al/²⁷Al)₀ = (5.29 ± 0.18) × 10⁻⁵, initial δ²⁶Mg = −0.05‰ ± 0.13‰, and MSWD = 0.9. The smaller error on the slope of the SIMS-based isochron is due to the higher ²⁷Al/²⁴Mg values relative to the LA-ICPMS data. The combined data sets are within error of one another, and define a correlation line (MSWD = 1.3) yielding (²⁶Al/²⁷Al)₀ = (5.27 ± 0.17) × 10⁻⁵, initial δ²⁶Mg = 0.02‰ ± 0.06‰, and MSWD = 1.3. The initial ²⁶Al/²⁷Al ratio of Leoville 3536-1 is clearly resolved from the supracanonical value of (5.85 ± 0.05) × 10⁻⁵ (Thrane et al. 2006).

Zoom In Zoom Out Reset image size

CAIs are the oldest objects formed in the solar system (Amelin et al. 2002), and preserve the highest measured value for initial ²⁶Al/²⁷Al (excepting presolar grains). This canonical ratio for initial ²⁶Al/²⁷Al is chronologically important because it marks the approximate time of solar system birth. However, the exact value of the ratio is less precisely defined than is commonly assumed. MacPherson et al. (1995) compiled all data then available for (²⁶Al/²⁷Al)₀ in early solar system materials and showed a histogram of these values, in which most CAI data define a peak centered on 4.5 × 10⁻⁵. Accordingly, some authors have defined canonical (²⁶Al/²⁷Al)₀ to be 4.5 × 10⁻⁵. But the histogram peak has substantial width, mostly in the range (3.2–5.5) × 10⁻⁵, due at least in part to analytical uncertainties in the data at that time (especially for data having low Al/Mg ratios). Therefore, MacPherson et al. (1995) specifically did not define the canonical value based on the histogram peak center. Instead, they recommended the value of 5 × 10⁻⁵ as a reasonable approximate upper limit, because many analyzed phases with ²⁷Al/²⁴Mg > 1200 have initial ²⁶Al/²⁷Al ∼ 5 × 10⁻⁵ but none are significantly above it. Recent high-precision bulk CAI data (Jacobsen et al. 2008) give a slightly higher value of (5.23 ± 0.13) × 10⁻⁵ (±2σ). We accept the latter value as canonical for the purposes of the following discussion, and the importance here has to do with whether some recently reported supracanonical values really are higher than canonical.

The putative existence of (²⁶Al/²⁷Al)₀ = ∼6 × 10⁻⁵ preserved in some CAI material, relative to the lower "canonical" ratios generally inferred from internal CAI isochrons, implies the interesting possibility that a measurable time gap existed between condensation of the CAI precursors and the final crystallization of the CAIs themselves. The evidence for the higher ratio comes from two types of data: in situ analyses from a number of CAIs, using high-precision SIMS and LA-ICPMS techniques, and analyses of bulk CAIs.

The in situ data (Young et al. 2005; Taylor et al. 2005; Cosarinsky et al. 2007) were interpreted to be the partially preserved remnants of internal supracanonical isochrons. Cosarinsky et al. (2007; who incorporated the data of Taylor et al. 2005) analyzed (via SIMS) only phases with very low ²⁷Al/²⁴Mg ratios (spinel and aluminous diopside), on the assumption that high Al/Mg phases such as melilite and anorthite may have been partially reset. They also forced their regressions to pass through the origin, which may not be a safe assumption (e.g., note that Thrane et al. 2006, who advocate supracanonical (²⁶Al/²⁷Al)₀, do not force their regressions through the origin). However, other recent high-precision in situ SIMS studies that specifically include melilite and anorthite (Kita et al. 2007, 2008; Makide et al. 2009) have yielded well-defined isochrons with only canonical ratios. Connolly et al. (2009) used high-precision SIMS and LA-ICPMS both on the same CAI and showed that both techniques yield similar supracanonical results for some melilite grains at low ²⁷Al/²⁴Mg, but anorthite in the same CAI is disturbed and thus the origin of the "supracanonical" melilite values remains open to question. Young et al. (2005) used LA-ICPMS to obtain data for several CAIs, and nearly 79% of their analyses yielded apparent supracanonical (²⁶Al/²⁷Al)₀ relative to the histogram peak value of 4.5 × 10⁻⁵. However, relative to (²⁶Al/²⁷Al)₀ = 5 × 10⁻⁵, that percentage drops to just over 50%. One highlighted inclusion (Leoville 144A) yielded (²⁶Al/²⁷Al)₀ = (5.9 ± 0.3) × 10⁻⁵ (±2σ). However, the data for this object show significant scatter on the isochron diagram (MSWD = 3.3) that likely indicates isotopic disturbance. Moreover, the relatively large laser spot size (50–100 μm) used to acquire their LA-ICPMS data prevented obtaining any analyses with high ²⁷Al/²⁴Mg. Together these factors preclude a precise estimate of the true initial ²⁶Al/²⁷Al ratio at the time Leoville 144A formed. Another contributing factor to the uncertainty over whether their data are resolved from canonical rests on their use of the value 0.521 for the mass fractionation correction factor, β, for calculating the radiogenic component of excess ²⁶Mg in their analyses. Davis et al. (2005) argued that β = 0.521 is ill suited for low-Al/Mg phases in CAIs that show (and many do) significant intrinsic mass-dependent isotopic fractionation of magnesium, and recommended a value for β = 0.514 based on isotopic measurements of laboratory-produced evaporation residues. By using β = 0.514 instead of 0.521 to correct for mass fractionation, the inferred initial ²⁶Al/²⁷Al ratio of CAI 144A decreases from (5.9 ± 0.3) × 10⁻⁵ to (5.6 ± 0.42) × 10⁻⁵, which is not resolved from (5.23 ± 0.13) × 10⁻⁵. Lastly, Leoville 144A is a compact Type A CAI, which by virtue of being probably once molten is one of the very objects least expected by the model of Young et al. (2005) to preserve any evidence for supracanonical (²⁶Al/²⁷Al)₀.

Thus, the principal evidence for a supracanonical (²⁶Al/²⁷Al)₀ ratio of ∼ 6 × 10⁻⁵ rests with the bulk CAI data (whole rock isochrons) of Bizzarro et al. (2005)⁷ and Thrane et al. (2006) and which, according to the model of those authors, are independent of whether any of their CAIs have canonical internal isochrons, because the latter presumably reflect only the time of final solidification.

Their results, however, have proven to be controversial. Subsequent studies of bulk CAIs, using techniques similar to those of Bizzarro et al. (2005) and Thrane et al. (2006), have not reproduced supracanonical initial ratios. Jacobsen et al. (2008) in particular carefully validated both the magnesium isotopic analyses and the measurements of Al/Mg in several ways, and obtained canonical (²⁶Al/²⁷Al)₀ = (5.23 ± 0.13) × 10⁻⁵. Baker (2008) obtained results in close agreement with those of Jacobsen et al. (2008).

The reasons for the discrepancy between the results of Thrane et al. (2006) and more recent data (Jacobsen et al. 2008; Baker 2008) are not yet understood. Nevertheless, the principal rationale for our study was to find the best example of an unprocessed CAI, one that formed by primary condensation, because such an object would represent the best possibility of finding a well-defined supracanonical internal isochron. Leoville 3536-1 has both the petrologic and chemical signatures of a condensate, and yet our isotopic data define an undisturbed isochron that yields canonical (²⁶Al/²⁷Al)₀, (5.27 ± 0.17) × 10⁻⁵.

The difference between our results and those of Bizzarro et al. (2005) and Thrane et al. (2006) cannot be explained as due to reprocessing or recrystallization of Leoville 3536-1 or other fine-grained CAIs (cf. Young et al. 2005). But such CAIs certainly never have been melted: had they been, their anorthite-normative bulk compositions would result in distinctive igneous (ophitic) textures (characteristic of, e.g., Type C CAIs; Tronche et al. 2007). Krot et al. (2004) noted that the melilite-rich mantles of fine-grained CAIs such as Leoville 3536-1 are difficult to explain by single-stage condensation, and might have formed as a result of later reheating accompanied by evaporation. But the isotopically light magnesium compositions of the melilite-rich mantle and anorthite-rich core of the Leoville 3536-1 (Table 1) indicate that the mantle could not have resulted from melt evaporation (Richter 2004). Moreover, even solid-state evaporation cannot explain the melilite mantles, as those mantles have higher Si/Al and Mg/Al relative to the cores—opposite what would be expected. Also, our δ²⁵Mg data (both SIMS and ICPMS) are isotopically light (negative; Table 1) in the melilite mantles just as in the cores, indicating that the mantles also formed by condensation (e.g., Niederer and Papanastassiou, 1984), possibly from a gas of non-solar composition.

One plausible way to reconcile our data with those of Bizzarro et al. (2005) and Thrane et al. (2006) is the possibility that fine-grained, spinel-rich CAIs with Group II trace element patterns are a different, younger breed of CAI arising from a later episode of condensation subsequent to the hypothetical primordial condensates. This requires the special coincidence that Group II CAIs were condensing at precisely the same time as the Types A and B CAIs were being processed. There admittedly are no absolute age data to show that fine-grained spinel-rich CAIs are older (or younger) than any other CAIs, but the nature of the volatility-controlled Group II trace element pattern argues against their being younger. The Group II signature, which is understood primarily in terms of volatility-controlled fractionation of the rare earth elements (REE), reflects fractional condensation from gas to solid (Boynton, 1975; Davis & Grossman, 1979). Condensation of a very high temperature phase (likely hibonite, CaAl₁₂O₁₉; MacPherson & Davis 1994) selectively incorporates the most refractory REE into the condensing solids and leaves the more volatile REE in the gas. Subsequent condensation from the remaining reservoir will produce phases depleted in those most refractory REE and relatively enriched in the more volatile REE, i.e., the Group II pattern. However, even Group II patterns are depleted in the two most volatile REE, europium and ytterbium. Most CAIs, especially in CV3 chondrites, have relatively unfractionated REE patterns: they are uniformly enriched in both refractory and volatile REE (including europium and ytterbium), meaning that they represent condensation of the REE to lower temperatures than do the Group II patterns. In the sense that lower temperatures correspond to a later time, the unfractionated "normal" CAIs are relatively younger than the CAIs with Group II trace element patterns, not the other way around. This is further supported by the observation that some coarse-grained CAIs have Group II trace element patterns (e.g., Wark 1987), suggesting that those inclusions formed by melting of fine-grained precursors with the Group II trace element patterns. Finally, fine-grained spinel-rich inclusions like Leoville 3536-1 have subchondritic bulk Ca/Al ratios, so they cannot have recondensed from vapor that was fractionally evaporated from an earlier generation of CAIs or CAI precursors. Such vapor would be depleted, not enriched, in aluminum.

We find no compelling arguments that Leoville 3536-1 and its kin are anything other than primordial condensates. Thus our data do not support the existence of supracanonical (²⁶Al/²⁷Al)₀: we analyzed a type of CAI that should best preserve it yet we find only canonical (²⁶Al/²⁷Al)₀. The existence of supracanonical initial ²⁶Al/²⁷Al in the solar system will remain in doubt until convincing internal mineral isochrons unaffected by local re-equilibration are found that clearly indicate such ratios.

Leoville 3536-1 records the time of the condensation event that formed this particular inclusion. The data of Jacobsen et al. (2008) record the time of high-temperature Mg/Al fractionation of a suite of whole CAIs. These two events are contemporaneous within analytical uncertainty. Moreover, both are approximately contemporaneous with the melting event recorded by a coarse-grained igneous CAI measured by our group (e.g., Kita et al. 2008). Collectively, these observations indicate that nebular Mg/Al fractionation, condensate CAI formation, and melting of some CAIs occurred within a very short time interval, perhaps only a few tens of thousands of years.

This work was supported by NASA grants NNX07AJ05G (G.J.M.), NNX07AI46G (N.K.), NNG06GE52G (M.W.), NNG06-GF19G and NNX09-AG39G (A.M.D.), and NAG5-10610 (A.N.K.). Wisc-SIMS is supported partly by NSF (EAR03-19230 and EAR07-44079, PI: J. W. Valley). Detailed comments by three anonymous reviewers led to improvements in the manuscript.

LA-MC-ICPMS measurements were made at the Field Museum in Chicago using a GV Instruments IsoProbe with a New Wave UP193HE excimer laser ablation system. The mass spectrometer was run at a mass-resolving power of approximately 400, and with a slight positive voltage applied to the extraction lens to minimize magnesium background intensities to <5 × 10^‑15 A. Individual measurements consisted of 20 integration cycles (7 s each). The laser spot size was 90 μm, with a repeat rate of 4 Hz and a laser energy density of ∼2 J cm⁻². Signal intensities for ²⁴Mg ranged from 0.6 to 3.0 × 10⁻¹¹ A. All measurements employed sample-standard bracketing. Standards were isotopically well-characterized synthetic glasses from evaporation experiments having CAI-like compositions (Richter et al. 2007), chosen to approach the composition of the unknown analyzed phase assemblage. Based on the analyses of these glasses, external reproducibilities (2σ) for our measurements are ±0.20‰ for δ²⁶Mg. Internal precisions of individual measurements of unknowns typically are larger, ±0.3‰–1.3‰ for δ²⁶Mg, because it was not possible to make repeat measurements of the same spot. Both natural and instrumental mass fractionation were corrected using an exponential law with β = 0.514 (Davis et al. 2005).

Analyses of terrestrial mineral standards (diopside, spinel, and gehlenitic melilite) showed no resolvable excesses or deficits in δ²⁶Mg that would indicate that our CAI δ²⁶Mg data were affected by matrix effects or isobaric interferences. Specifically, we observed no significant contribution of ⁴⁸Ca⁺⁺ to the ²⁴Mg⁺ signal in melilite (Ca-rich); this is not surprising given the low isotopic abundance of ⁴⁸Ca (0.19%). We made one additional test to evaluate the potential effect of ⁴⁸Ca⁺⁺ to the ²⁴Mg⁺ signal, by ablating calcite and scanning over the mass range of 21 amu to 22 amu to determine if there was any ⁴³Ca⁺⁺ present at 21.5 amu. The isotopic abundance of ⁴³Ca (0.14%) is similar to that of ⁴⁸Ca (0.19%), and if any appreciable ⁴⁸Ca⁺⁺ were produced, we would have expected to see a signal at mass 21.5 amu. We observed no signal above background at this mass. It is likely that thermalization of the ion beam (from collisions with neutral gas atoms) in the hexapole collision cell of the IsoProbe MC-ICPMS effectively eliminates doubly charged ions.

The large radius SIMS CAMECA IMS 1280 at the University of Wisconsin-Madison (Wisc-SIMS) was used to obtain high-precision/high spatial resolution magnesium isotopic analyses. The primary beam was O⁻, operated under conditions similar to those described in Kita et al. (2000). Two different analytical conditions were used. Melilite (²⁷Al/²⁴Mg > 10) was analyzed using a 3 × 7 μm beam spot (0.1 nA intensity) in order to avoid grain boundaries with magnesium-rich phases; magnesium isotope signals were detected with a monocollection electron multiplier pulse-counting system in magnetic peak jumping mode, at a mass-resolving power of 3500. Spinel and pyroxene (²⁷Al/²⁴Mg < 10) were analyzed using a ∼10 × 15 μm beam (1.4 nA intensity) with multicollector Faraday Cup (MCFC) detectors to achieve high precision (≤0.2‰) at a mass-resolving power of 2200. ²⁷Al signals were collected on a MCFC detector simultaneously with Mg isotopes in both cases. The standards used to correct instrumental bias on the magnesium isotope analyses were synthetic zoned melilite crystals, glass, spinel, and pyroxene. Both natural and instrumental mass fractionation were corrected using an exponential law with β = 0.514 (Davis et al. 2005). Secondary ²⁴Mg intensities typically were 1 × 10⁵ cps for melilite (MgO < 2%), and 4 × 10⁷ cps for spinel and pyroxene (MgO ∼ 20%). The dead time of the monocollection electron multiplier detector was estimated to be 20.4 ± 1.8 (2σ) ns from the analyses of standards with various Mg intensities (5 × 10⁴–3 × 10⁵ cps). External reproducibilities (2σ) for δ²⁶Mg* were ∼1‰ for melilite analyses and ≤0.2‰ for spinel and pyroxene analyses. During some melilite analyses, the Mg⁺ intensity increased significantly during analyses because of μm-size magnesium-rich phases. We rejected all such data. Similarly, we rejected any analyses that hit mixed phases, based on SEM examination of all SIMS spots.