Abstract– High-precision isotope imaging analyses of reversely zoned melilite crystals in the gehlenitic mantle of Type A CAI ON01 of the Allende carbonaceous chondrite reveal that there are four types of oxygen isotopic distributions within melilite single crystals: (1) uniform depletion of 16O (δ18O ≈ −10‰), (2) uniform enrichment of 16O (δ18O ≈ −40‰), (3) variations in isotopic composition from 16O-poor core to 16O-rich rim (δ18O ≈ −10‰ to −30‰, −20‰ to −45‰, and −10‰ to −35‰) with decreasing åkermanite content, and (4) 16O-poor composition (δ18O ≥ −10‰) along the crystal rim. Hibonite, spinel, and perovskite grains are 16O-rich (δ18O ≈ −45‰), and adjoin 16O-poor melilites. Gas-solid or gas-melt isotope exchange in the nebula is inconsistent with both the distinct oxygen isotopic compositions among the minerals and the reverse zoning of melilite. Fluid-rock interaction on the parent body resulted in 16O-poor compositions of limited areas near holes, cracks, or secondary phases, such as anorthite or grossular. We conclude that reversely zoned melilites mostly preserve the primary oxygen isotopic composition of either 16O-enriched or 16O-depleted gas from which they were condensed. The correlation between oxygen isotopic composition and åkermanite content may indicate that oxygen isotopes of the solar nebula gas changed from 16O-poor to 16O-rich during melilite crystal growth. We suggest that the radial excursions of the inner edge of the protoplanetary disk gas simultaneously resulted in both the reverse zoning and oxygen isotopic variation of melilite, due to mixing of 16O-poor disk gas and 16O-rich coronal gas. Gas condensates aggregated to form the gehlenite mantle of the Type A CAI ON01.
Calcium-aluminum-rich inclusions (CAIs) in primitive meteorites are thought to have formed by gas-solid condensation, or originated from condensate precursors (e.g., Grossman 1972; Grossman et al. 2002). Melilite (åkermanite, Ca2MgSi2O7– gehlenite, Ca2Al2SiO7 solid solution) is a common primary mineral in CAIs. Chemical zoning pattern of melilite crystal is diagnostic of its formation mechanism. Formation of normally zoned melilite grains showing increase in åkermanite content from core to rim can be explained either by igneous or condensation processes (Osborn and Schairer 1941; Grossman 1972). In contrast, reversely zoned melilite, commonly observed in fluffy Type A CAIs (FTAs) is likely to be formed by condensation from gas (MacPherson and Grossman 1984). Thus, reversely zoned melilite may have recorded oxygen isotope composition of the solar nebula gas, if no subsequent oxygen-isotope exchange occurred. After the formation, however, some CAIs may have experienced various thermal processes in the nebula and/or on the meteorite parent body. The oxygen isotopes in CAI melilites may have exchanged with the surrounding gas in the nebula and aqueous fluid on the parent body.
Reversely zoned melilites were found in a coarse-grained Type A CAI, named ON01, from the Allende carbonaceous chondrite. Oxygen isotopic measurements of a reversely zoned melilite by line traverse revealed a change in the oxygen isotopic composition toward 16O-rich as the åkermanite content decreased from the core to the rim of the crystal (Park et al. 2012). This observation suggests that the oxygen isotopic composition of the solar nebula gas changed during the crystal growth of the melilite by condensation, assuming that secondary processes have not altered the oxygen isotopes of the melilite. To better understand nebular settings and parent body process for oxygen isotopic variations of CAI melilites, we report the in situ spot analyses and 2-D isotopic distributions of individual reversely zoned melilite crystals in the ON01 analyzed by isotopography with submicron scale.
A polished thin section of the coarse-grained Type A CAI, ON01, from the Allende CV3 chondrite was coated with approximately 20 nm thick carbon film to conduct charge on the sample surface. A field emission scanning electron microscope (FE-SEM; JEOL JSM-7000F) equipped with an energy-dispersive X-ray spectrometer (EDS; Oxford INCA Energy) system was utilized to obtain backscattered electron (BSE) images and digital X-ray maps, and to conduct quantitative chemical analyses.
An electron backscatter diffraction (EBSD; HKL Channel 5, Oxford Instruments) system equipped on the FE-SEM was used to determine grain boundaries. EBSD maps were obtained with a 1 μm step size, and compared with BSE images to accurately determine the grain boundaries of small grains that were vague when viewed using petrographic microscope, due to tiny cracks or holes. The chemical zoning pattern of a single melilite crystal could then be identified based on the well-defined grain boundaries. Using this technique, candidate melilite grains were selected from the gehlenite mantle of the CAI for isotopography (Fig. 1). Most EDS and EBSD examination was conducted prior to isotope analysis of the grains.
Spot Analysis for Oxygen Isotopes
Oxygen isotope analyses were conducted using a secondary ion mass spectrometer (SIMS; Cameca ims-1270) at Hokkaido University (Hokudai) before or after isotopography. A mass-filtered Cs+ primary ion beam with energy of 20 keV was focused onto the sample surface. The primary beam current was set to 20 pA, and the beam was focused to a diameter of approximately 3 μm. Negative secondary ions were accelerated to 10 keV and transmitted through entrance and exit slits adjusted to 60 and 180 μm, respectively, to acquire a flat-top shaped peak with a mass resolving power (MRP) of >5000, which is sufficient to separate 16OH− from 17O−. A normal incidence electron gun was applied to compensate electrical charge build-up of the analysis area.
16O- was measured using a Faraday cup (FC), and 17O− and 18O− were measured using an electron multiplier (EM) by magnetic peak jumping from one mass to another. 16O− signals were corrected by subtracting background signals for the FC measured at mass 15.8. The EM was operated in pulse counting mode with a dead time of 30 ns. Each measurement consisted of 60 cycles with the counting sequence of mass 15.8 for 1 s, 16O− for 1 s, 17O− for 2 s, and 18O− for 1 s with a waiting time of 2 s for every mass to stabilize the sector magnetic field.
The oxygen isotopic compositions are re ported as per mil deviations relative to standard mean ocean water (SMOW);
The deviation from the terrestrial fractionation line is expressed as Δ17O;
The instrumental mass fractionation was corrected using synthetic åkermanite as a standard. More detailed analytical procedures for the spot analyses are described in Park et al. (2012).
Quantitative isotope images (isotopographs) were obtained using the Hokudai isotope microscope system consisting of the Cameca ims-1270 SIMS instrument and a high-efficiency stacked CMOS-type active pixel sensor (SCAPS) ion imager (Takayanagi et al. 1999; Yurimoto et al. 2003). A 20 keV Cs+ primary ion beam with a current of approximately 1.0 nA was uniformly irradiated onto the sample surface as an oval shape with a traverse diameter of approximately 70 μm. Negative secondary ions were transmitted through the 50 μm diameter contrast aperture to obtain high spatial resolution. The exit slit was set to 750 μm and the magnetic field was adjusted to eliminate 18O− peak interference.
A normal incidence electron gun was utilized for charge compensation of the analysis area. Incomplete charge compensation by electron flooding on the sample surface can cause a small difference of secondary ion emission (Yurimoto et al. 2003); therefore, fine tuning of the electron gun is necessary to obtain high spatial resolution and minimize the mass fractionation effect on the oxygen isotopographs. Secondary ion images of 16O− and 28Si− were useful for fine-tuning within a short time, due to their high intensities generated from silicate minerals. The electron gun parameters were finely tuned on an isotopically homogeneous region of the CAI (core melilite; Park et al. 2010), and then moved to a candidate melilite in the gehlenite mantle (area #1–#8 in Fig. 1).
The intersite occupancy of Al + Al ↔ Mg + Si determines the chemical composition of melilite; therefore, the in situ analysis of Si and Al was designed together with the oxygen isotope analysis to investigate the relationship between the chemical and oxygen isotopic compositions of melilite. Secondary ion images were acquired using the following cycle in the order of 28Si−, 27Al−, 28Si−, 16O−, 18O−, and 16O− with respective accumulation times of 25, 500, 25, 5, 1000, and 5 s. In the case of melilite, the average numbers of ions per pixel (0.17 × 0.21 μm2) integrated in given accumulation periods were approximately 700, 1000, 6100, and 1000 for 28Si−, 27Al−, 16O−, and 18O−, respectively. The sputtering depth was approximately 200 nm for one cycle. The Si/Al ion ratio, obtained from 28Si− ions divided by 27Al− ions, was converted to a Si/Al atomic ratio by comparison with the quantitative data obtained from FE-SEM-EDS using a linear calibration curve.
The δ18O isotopograph was obtained by calculating the secondary ion ratios of 18O/16O, and then converted to the SMOW scale by normalization using spot analyses data in the imaging area for the corresponding pixel. Image smoothing with a moving-average of 5 × 5 pixels was applied to reduce random noise caused by the statistical fluctuation of secondary ions.
The CAI ON01 is a round-shaped inclusion with a size of approximately 6.5 × 5.5 mm2 that is dominated by melilite with small amounts of spinel, perovskite, and fassaite (Fig. 1). The mineral assemblage, shape, and texture lead to classification of this inclusion as a coarse-grained compact Type A (CTA) CAI. A combined X-ray elemental map shows that this CTA CAI consists of an åkermanitic melilite core and a gehlenitic melilite mantle (Fig. 1b). A Wark-Lovering (W-L) rim completely encloses the gehlenitic melilite mantle. Melilite grains in the core are typically hundreds of micrometers in size with a mean åkermanite content of about 28 mol%.
The gehlenite mantle is distinguished from the core by occurrences of aluminous melilite and hibonite grains (Fig. 1b). Melilites in the mantle are more gehlenitic (Åk<∼10) and smaller (<∼100 μm) than those in the core. Hibonite often coexists with spinel and perovskite. Some hibonite grains are replaced by spinel. Hibonite has TiO2 contents up to about 8 wt%, which is comparable to the hibonite typical in CAIs (see Brearley and Jones 1998). Spinel and perovskite are nearly pure MgAl2O4 and CaTiO3, respectively. These features of the gehlenite mantle are similar to those of the “aluminous rind” observed in some CTAs (Beckett and Stolper 1994; Simon et al. 1999; Wark and Boynton 2001). Many of the mantle melilites are reversely zoned; they have a continuous decrease in the åkermanite content from the core to rim within a single crystal. The outermost part of the inclusion is irregularly shaped, although the overall shape of the inclusion is rounded. The petrography and chemical composition of the entire CAI will be described in detail elsewhere (Park et al., in preparation).
Quality of Isotopography
The quality of isotopography was evaluated by analyzing an area composed of multimineral phases with different chemical compositions (area #1, Fig. 2). Hibonite, spinel, and perovskite are enclosed by melilite in this area. Spinel partially replaces hibonite and perovskite grains are attached at the edge of spinel grains. The replacement of melilite by grossular is observed near cracks (Fig. 2a).
A 28Si− isotopograph of the analysis area (white circle of Fig. 2a) shows a clear boundary between melilite and hibonite (Fig. 2b). The sharpness of the 28Si− isotopograph is comparable to that of the BSE image. The right side of the analysis area is somewhat blurred probably due to holes on the sample surface. The 27Al− isotopograph distinguishes each mineral according to the Al content (Fig. 2c). Tiny perovskite grains are well identified as dark spots in the 27Al− isotopograph and as bright spots in the BSE image. The homogeneous distribution of Si and Al intensities in each mineral phase confirms that the charge compensation by electron flooding on the analysis area was successful.
Grossular is emphasized in the Si/Al isotopograph compared with the 28Si− and 27Al− isotopographs (Fig. 2d). A moving average operation of 5 × 5 pixels was applied to the isotopographs for image smoothing; however, the sharpness of the Si/Al isotopograph was not significantly changed with respect to the original isotopograph, because each pixel size is smaller than the spatial resolution under the magnification of the isotope microscope. To quantify the spatial resolution of the Si/Al isotopograph, a Si/Al ion ratio profile was obtained from the line A–B across hibonite and melilite (Fig. 2d). The spatial resolution defined by the width between the 16% and 84% level was calculated to be 0.7 μm (Fig. 2e).
In the δ18O isotopograph, hibonite and spinel are 16O-rich (δ18O ≈ −55‰) and are clearly distinguished from 16O-poor compositions of melilite (δ18O ≈ −5‰) (Fig. 2f). The spatial resolution of the δ18O isotopograph was also calculated to be 0.7 μm using the same method as that for the Si/Al isotopograph (Fig. 2g). A small mottled pattern with a length of about 5 pixels is evident in the δ18O isotopograph. This pattern is an artifact of the smoothing operation. After image smoothing by the moving-average of 5 × 5 pixels, the precision of δ18O for a pixel in the isotopograph was calculated to be ±7‰ (1 SD) from counting statistics, assuming isotopic homogeneity within the phase, and the precision is consistent with the peak-to-peak amplitude of the line profile within phases (Fig. 2g).
Oxygen Isotopic Compositions of Mantle Melilites
Oxygen isotope data of melilites in the gehlenite mantle were obtained by spot analyses (#2–#8 in Fig. 1a) and are given in Table 1. The oxygen isotopic compositions of the mantle melilites vary from −45 to −10‰ for δ18O. These values are clearly different from the homogeneously 16O-depleted composition of the core melilites (δ18O ≈ 0‰; Park et al. 2010). This wide variation reflects the heterogeneous oxygen isotopic distribution among the melilite grains (area #2–#4) and also within a single melilite grain (area #5–#7). In the three-oxygen isotope diagram, all melilites are plotted along the carbonaceous chondrite anhydrous mineral (CCAM) line (Fig. 3a). The δ18O values obtained from isotopographs are also listed in Table 1 and have a 1:1 correspondence with those obtained by spot analyses over a wide range of δ18O values. This correspondence demonstrates the high accuracy of oxygen isotopography.
Table 1. Oxygen isotopic compositions of melilites from each analysis area.
Area (Fig. #)
δ17O ± 2σ
δ18O ± 2σ
Δ17O ± 2σ
aThe number-1 grains correspond to the melilite crystals showing reverse zoning in Figs. 4–6, and the number-2 grains are adjacent to the grains of number-1.
bδ18O values were obtained from an area of 300 (20 × 15) pixels corresponding to the beam position and a typical beam spot size (approximately 3.5 × 3 μm2) for the spot analysis.
c2 standard errors calculated from the 16O and 18O ions accumulated in the 300 pixels by counting statistics.
Oxygen isotopographs were obtained for seven areas in the gehlenite mantle (area #2–#8 of Fig. 1a; Figs. 4–6). Four types of oxygen isotope distributions are observed in the reversely zoned melilite crystals: (1) uniform depletion of 16O (area #2, Fig. 4a); (2) uniform enrichment of 16O (area #3–#4, Figs. 4b and 4c); (3) oxygen isotopic variation from 16O-poor core to 16O-rich rim, correlated with reverse chemical zoning (area #5–#7, Fig. 5); and (4) 16O-poor composition in the crystal rim (area #8, Fig. 6). Detailed descriptions of each analysis area are given below.
In area #2, the melilite grain in the center exhibits reverse zoning, which is clearly represented in the Si/Al isotopograph (Fig. 4a). The åkermanite content continuously decreases from the core (Åk ∼ 10) toward the rim (Åk ∼ 1). Secondary anorthite and grossular partly replace melilite near small cracks. The oxygen isotopic composition of the reversely zoned melilite crystal is uniformly depleted in 16O (δ18O ≈ −10‰). In contrast, melilite in the upper-right corner of the area is enriched in 16O (δ18O ≈ −35‰). The δ18O values between these melilite grains change sharply at the grain boundary.
In area #3, the åkermanite content of the reversely zoned melilite crystal changes from Åk ∼ 10 to Åk ∼ 1 (Fig. 4b). Regardless of the chemical zonation, the melilite crystal is uniformly enriched in 16O (δ18O ≈ −45‰). The degree of 16O enrichment in this melilite is similar to that of spinel. The 16O-poor compositions are observed along the cracks of grain boundaries and appear to diffuse slightly into melilite with an 16O-rich composition. All melilite grains in area #4 are 16O-rich (−35 to −40‰ in δ18O; Fig. 4c).
In areas #5–#7, oxygen isotopic variations are observed within the individual melilite single crystals (Figs. 5a–c). The variation was first discovered in a relatively large grain of approximately 100 μm in size (area #5) by SIMS spot analysis, and was investigated by traversing from core to rim with an approximately 3 μm beam (Park et al. 2012). The line traverse revealed that the δ18O of −10‰ in the core decreased to −30‰ at the rim. The oxygen isotopograph of the same grain shows a more 16O-rich composition in the rim than in the core, which is in good agreement with the line traverse data (Table 1, Fig. 5a). Small fassaite grains (approximately 1μm) are evident in the core of the melilite crystal; however, the oxygen isotopic compositions of the fassaite grains could not be distinguished from those of melilite, probably because they are 16O-poor or their composition in the isotopograph is a result of mixing with surrounding 16O-poor compositions due to their small grain sizes, even though they actually have 16O-rich compositions. A small hibonite grain (about 3 μm) within melilite is enriched in 16O. A perovskite, spinel, and hibonite assemblage adjoining the relatively 16O-poor portion of the melilite grain is enriched in 16O (δ18O ≈ −45‰; Fig. 5a-box).
In area #6, more continuous variations in both the chemical and oxygen isotopic compositions are observed within a single melilite crystal (Fig. 5b). The δ18O values of this melilite grain gradually change from −20 to −45‰ with decreasing åkermanite content (Åk ∼ 9 to Åk ∼ 2) from the core to the rim. Melilites near cracks have 16O-poor compositions.
The melilite crystal in area #7 exhibits reverse zoning (Åk ∼ 10 to Åk ∼ 0) that appears to correlate with the oxygen isotopic variation (−10 to −35‰ in Fig. 5c). Melilites near holes and secondary phases within this melilite crystal have also 16O-poor compositions, although they are gehlenitic.
Melilite with reverse zoning (Åk ∼ 6 to Åk ∼ 0) in area #8 has a δ18O variation of 25‰ (Fig. 6). Melilite in the center of the crystal is 16O-rich (δ18O ≈ −35‰). Some areas distributed along cracks or near secondary phases are 16O-poor (δ18O ≈ −10‰). Melilites at a triple junction are 16O-rich, the same as those in the core area of the crystal. Hibonite is uniformly 16O-enriched, even though the grain is surrounded by 16O-poor melilite.
Oxygen Isotopic Variation in Reversely Zoned Melilite
Reversely zoned melilites found in the gehlenite mantle of the CAI ON01 have various oxygen isotope distributions within a single crystal. Some melilites show a correlation between chemical zoning and oxygen isotopic compositions, while others have homogeneous oxygen isotopic compositions either depleted or enriched in 16O. However, reversely zoned melilites that exhibit such variations in oxygen isotopic compositions appear to be randomly distributed over the gehlenite mantle. The oxygen isotopic compositions of the melilites have no systematic relation with the distance from the W-L rim (Fig. 1).
The formation of reverse zoning throughout melilite crystals is best explained by gas-solid condensation (MacPherson and Grossman 1984). It is reasonable to expect that the oxygen isotopic composition of reversely zoned melilite should record the isotopic composition of the gas in which the melilite condensed. However, the melilites may have experienced various thermal processes in the nebula and/or on the parent body after condensation. To understand the mechanism of oxygen isotopic variations in the reversely zoned melilites, we discuss several possible thermal processes that could have modified primary oxygen isotopic signatures of the melilites. These include (1) isotope exchange between gas and partially molten CAI (Yurimoto et al. 1998; Aléon et al. 2007), (2) gas-solid diffusive exchange in the nebula over a long accumulated time (Fagan et al. 2004a; Simon et al. 2011), and (3) aqueous alteration and thermal metamorphism on the parent body (Wasson et al. 2001; Krot et al. 2008). If these processes have not disturbed the oxygen isotopes of melilites, then the oxygen isotopic variation in the reversely zoned melilite has been directly inherited from the solar nebula gas.
Before discussing oxygen isotope exchange between the melt and ambient nebula gas, it should be assessed whether the reverse zoning of melilite as shown in this study could be formed by igneous process. Reversely zoned melilites shown in Figs. 4–6 have åkermanite contents that continuously decrease from the crystal core to the rim. The reverse zoning throughout the crystal is very unlikely to be formed by simple melt-crystallization, because equilibrium crystallization of melilite predicts that the åkermanite content increases as the temperature decreases (Osborn and Schairer 1941).
In Type B inclusions, some melilites exhibit a restricted interval of reverse zoning that has been reproduced by dynamic crystallization experiments (MacPherson et al. 1984). If anorthite crystallization is suppressed in the melt with the Type B CAI composition, then pyroxene could be cocrystallized with melilite. In such a case, the Al/Mg ratio of the melt would be elevated as pyroxene begins to crystallize, which would result in a decrease in the åkermanite content in the melilite cocrystallizing with the pyroxene (MacPherson et al. 1984). However, the zoning profile of melilite crystal in Type B inclusions generally exhibit normal zoning in the core of crystal, and reverse zoning is restricted to some portion of the crystal. Moreover, melilite compositions with the reversely zoned intervals are more åkermanitic than those in this study. Thus, the reverse zoning profile of melilite in Type B inclusions is significantly different from the observations in this study. In addition, pyroxene is almost absent in the gehlenite mantle, which supports that the reverse zoning of melilites in ON01 cannot be explained by melt crystallization.
Grossman et al. (2002) demonstrated that reversely zoned melilite without a normal zoning core could be reproduced by evaporation of a melt. However, they pointed out that this process requires significantly large Mg isotope fractionation beyond the range observed in CAIs; therefore, this process is unrealistic.
The strikingly different oxygen isotopic compositions among the melilite grains (Fig. 4a) are unlikely to be formed from a single melt composition, even if the melting is accompanied by evaporation. Moreover, the wide and gradual change of oxygen isotopes in a single crystal (Fig. 5) is inconsistent with the origin from a melt. Oxygen isotopes in the melt would be quickly homogenized, because oxygen self-diffusion in a silicate melt is very fast (Zhang and Ni 2010). Partial melting may result in oxygen isotope variation within a single crystal. However, the oxygen isotopic distribution changes abruptly within a single crystal (Yurimoto et al. 1998; Ito et al. 2004).
Therefore, the best explanation for the reverse zoning observed in the mantle melilites of ON01 is direct condensation from a gas (see discussion in MacPherson and Grossman 1984). A small degree of pressure decrease could result in the reverse zoning of melilite. For example, reverse zoning with variation of Åk ∼ 10 to Åk ∼ 1 could be formed with a total pressure decrease from 1 × 10−3 atm to 4 × 10−4 atm at about 1468K (fig. 11 in MacPherson and Grossman 1984).
The inferred condensation origin of reverse zoning in the mantle melilite grains of ON01 excludes gas-melt interaction as a mechanism to explain the observed oxygen isotopic variations in these grains.
Gas-Solid Diffusive Exchange
Recently, Simon et al. (2011) reported oxygen isotopic variations in a Type A CAI from Allende. In this CAI, 16O-poor compositions of melilites in the core gradually change to 16O-rich compositions around the W-L rim. It was suggested that gas-solid interaction followed by solid-state diffusion could occur by repetitive heating events at high temperature. The upper limit of temperature is approximately 1600 K to prevent oxygen isotope homogenization of small spinel grains with 16O-enriched compositions. However, our observation showing distinctive δ18O values across the grain boundary of melilites (Fig. 4a) is inconsistent with this proposed model.
If the CAI has experienced repetitive heating events over a long accumulated time, then solid-state diffusion could occur among melilite grains, which would give rise to the equilibration of oxygen isotopes at the grain boundary. Moreover, if such heating occurred, then perovskite adjoining the 16O-poor melilite shown in Fig. 5a-box should become 16O-poor, because the oxygen self-diffusion in perovskite (Gautason and Muehlenbachs 1993; Sakaguchi and Haneda 1996) is tens of times faster than that in melilite (Yurimoto et al. 1989; Ryerson and McKeegan 1994) at approximately 1600K.
The relationship between diffusion time and diffusion length for melilite and perovskite at 1600 K is illustrated in Fig. 7a. The time required for approximately 5 μm diffusion in melilite is about 10 hours. For the same time, diffusion over approximately 35 μm can occur in perovskite. This is inconsistent with the present observation. Therefore, the gas-solid diffusion in the nebula is also excluded as a model to explain the oxygen isotopic variations of the ON01 inclusion.
Fluid-Assisted Thermal Metamorphism
Allende is thought to be one of the most metamorphosed CV3 chondrites (Bonal et al. 2006). Based on the Fe/Mg exchange between olivine and spinel, the peak metamorphic temperature experienced by Allende is estimated to be approximately 800 K (Weinbruch et al. 1994). Thermal metamorphism on the Allende parent body is generally associated with aqueous alteration (fluid-assisted thermal metamorphism; Krot et al. 1998). The oxygen isotopic composition of this fluid was 16O-poor (Δ17O ≈ −3–0‰) as inferred from composition of the Allende magnetite formed by oxidation of metal by asteroidal water (Choi et al. 1997). This oxygen isotopic composition is similar to those of the secondary anorthite and grossular in ON01 (Figs. 4a, 5c, and 6), which implies that their oxygen isotope compositions reflect that of the fluid on the CV parent asteroid (Yurimoto et al. 2008).
Oxygen self-diffusion in melilite by isotope exchange with 16O-depleted fluid probably leads to 16O-poor compositions of the outermost portions of melilite grains. However, the oxygen self-diffusion in melilite is extremely slow even at 800 K. For example, 10 μm diffusion needs about 10 Myr, which is unrealistic for the thermal history of the Allende parent body (Weinbruch et al. 1994). It could be possible that lattice diffusion becomes faster in the presence of fluid (Krot et al. 2008), but there are no experimental data for oxygen self-diffusion in melilite under “wet” conditions. In addition, 16O-rich melilites at the triple junction (Fig. 6) and the distinctive δ18O values across the grain boundary of the melilites (Fig. 4a) do not correspond well with the concept of fast oxygen self-diffusion.
Alternatively, Wasson et al. (2001) suggested that the oxygen isotopic composition of melilite could be disturbed by the action of water, which would allow dissolution and immediate reprecipitation of the primary melilite being in a high energy state due to small grain size or crystal defects. If the fluid has low Mg, Al, and Si concentrations, then the oxygen isotope compositions of melilite could become 16O-poor while keeping the chemical composition. 16O-poor compositions are limited to a narrow range of about 10 μm near holes, cracks, or secondary phases (Figs. 4b, 5b, 5c, and 6). Thus, such 16O-poor areas within a mostly 16O-rich melilite likely reflect alteration of the oxygen isotopes by dissolution and reprecipitation that occurred during fluid-assisted thermal metamorphism (Nakamura et al. 2005).
However, a reset of oxygen isotopes by dissolution and reprecipitation cannot occur in the core of a large melilite grain that is energetically stable. This indicates that 16O-poor compositions throughout the large crystal shown in Fig. 4a cannot be simply explained by aqueous alteration on the parent body. In addition, the parent body process could not lead to a change of 16O-poor composition to 16O-rich composition, because the fluid is depleted in 16O. Only melilite grains with 16O-rich cores with 16O-poor rims, as shown in Fig. 6, may have the result of fluid-assisted thermal metamorphism.
Direct Records of Gas Composition
Both gas-melt and gas-solid interactions have been determined to be implausible explanations for the oxygen isotopic variations observed in the mantle melilites of the CAI ON01. Fluid-assisted thermal metamorphism on the parent body could partly disturb the oxygen isotopes of the melilites, but the effect is restricted to a narrow area accompanied by alteration phases, holes, or cracks. Thus, the mantle melilites in ON01 mostly preserve their original oxygen isotopic compositions. The reverse zoning of melilite observed in this study is successfully explained only by condensation from gas; therefore, we conclude that the three types of oxygen isotopic compositions of melilites, uniform depletion of 16O, uniform enrichment of 16O, and variation from 16O-poor to 16O-rich compositions with crystal growth, directly reflect the oxygen isotopic compositions of the solar nebula gas in which the melilites formed.
Formation of the Gehlenite Mantle
Reversely zoned melilites observed in the gehlenite mantle imply that the mantle consists of solids condensed from gas. If this hypothesis is correct, then other minerals observed in the mantle should also be formed by condensation. Hibonite, spinel, and perovskite assemblages are observed in the gehlenite mantle.
The texture of the assemblage is that spinel replaces hibonite, and small perovskite grains are attached to the spinel (Fig. 2a). Similar texture is also observed in area #5 (Fig. 5a). Spinel could be formed by the reaction of pre-existing hibonite with gas, due to the similarities of the spinel and hibonite crystal structures (Simon et al. 2006), although this deviates from the predictions of thermodynamic calculations of the solar gas (e.g., Grossman 1972; Petaev and Wood 1998; Ebel 2006). Ca and Ti ions contained in pre-existing hibonite cannot enter into the spinel therefore these ions are expected to form perovskite. Thus, the texture of this mineral assemblage can be explained by condensation.
If spinel is formed by the reaction of hibonite with gas, then melilite condensation could be kinetically hindered.
During this reaction, the oxygen isotopic compositions of spinel and perovskite products are inherited from that of the reactant hibonite, and are therefore less affected by the oxygen isotopic composition of the gaseous reactants. Assuming that hibonite is uniformly 16O-enriched (−55‰ in δ18O; Figs. 2f and 2g), then the oxygen isotopic compositions of spinel and perovskite are expected to be 16O-rich (approximately −42‰ in δ18O at x = 0.5), even when hibonite reacts with 16O-poor gas (0‰ in δ18O). Thus, spinel and perovskite could have 16O-rich composition, regardless of the oxygen isotopic composition of the gas.
The oxygen isotopic compositions of these minerals do not correspond well with melt-crystallization as the origin of the assemblage. Hibonite, spinel, and perovskite are 16O-rich and are in contact with 16O-poor melilite (Figs. 2f and 5a-box). Crystallization sequence from a melt of Type A CAI composition is spinel, gehlenitic melilite, followed by perovskite (Beckett 1986). If melilite and perovskite have crystallized from the melt from which spinel crystallized, then oxygen isotopic composition of melilite should be the same as those of spinel and perovskite. Therefore, it is very unlikely that these minerals are formed by melt crystallization.
The observations indicate that the gehlenite mantle in the CAI ON01 formed by aggregation of gas condensates to the pre-existing inclusion (now the core region of the inclusion). After aggregation of the condensates, a W-L rim surrounded the entire CAI. The irregular shape of the W-L rim shown in Fig. 1 may reflect the original shape of the gehlenite mantle, which indicates an aggregate of gas condensates.
Mechanism for the Correlation Between the Oxygen Isotope and Chemical Compositions
Reverse zoning of melilite could have formed by the condensation of solid from the solar nebula gas with a decrease in pressure (MacPherson and Grossman 1984). Therefore, it is reasonable that the oxygen isotopic composition of the solar nebula gas changed from 16O-poor to 16O-rich during the crystal growth of reversely zoned melilite. The astrophysical setting proposed by Itoh and Yurimoto (2003) and developed by Yurimoto et al. (2008) is a likely scenario to explain the coexistence of the two oxygen isotopic reservoirs, 16O-poor and 16O-rich, and also the formation of reversely zoned melilite.
According to their model, distinct 16O-enriched and 16O-depleted gaseous reservoirs coexisted around the inner edge of the protoplanetary disk. The space inside the inner edge of the protoplanetary disk was filled with plasma enriched in 16O, due to a high flux of coronal flow from the active proto-Sun. This is consistent with the oxygen isotopic composition of the Sun, which is highly enriched in 16O (δ17,18O ≈ −60‰; McKeegan et al. 2011). On the contrary, the disk gas was depleted in 16O during the classical T-Tauri stage. The inner edge of the disk was radially fluctuated due to magnetic flux variations (Shu et al. 1997).
In this setting, the oxygen isotopic composition of the gas in the fluctuation zone can be changed from 16O-poor to 16O-rich composition, as the inner edge of the disk moves outward. At the same time, the gas pressure of the fluctuation zone could be decreased as the low-pressure coronal gas gradually fills in the space previously occupied by the disk gas. As a result, reversely zoned melilite with compositional variation from 16O-poor to 16O-rich could be formed by condensation of the gas in the fluctuation zone. Reversely zoned melilites with uniformly 16O-enriched or 16O-depleted compositions could have formed in the 16O-rich or 16O-poor gasses in this setting when the gas pressure decreased.
Intracrystalline oxygen isotope distributions of reversely zoned melilites in the gehlenite mantle of the ON01 inclusion from Allende were obtained using high-precision isotopography with SCAPS. Four types of oxygen isotopic distributions within melilite crystals were observed: uniform depletion of 16O, uniform enrichment of 16O, variation from 16O-poor in the core to 16O-rich in the rim, and 16O-poor composition along the crystal rim.
The reverse zoning of melilite and distinct oxygen isotopic compositions among minerals contradict secondary oxygen isotopic exchange by both gas-melt interaction and solid-state diffusion accompanied with gas-solid interaction in the solar nebula for this CAI. Parent body process could partially alter the primary oxygen isotopic composition of a melilite grain to provide 16O-rich core with 16O-poor rim; however, this effect is limited to a narrow range near holes, cracks, or secondary phases.
Thus, the reversely zoned melilites mostly preserve the primary oxygen isotopic compositions, indicating that these melilites condensed separately in 16O-depleted or 16O-enriched nebula gasses. The correlation between the oxygen isotopes and chemical zonation indicates that the gas composition changed from 16O-poor to 16O-rich during crystal growth of the melilite by condensation. Thus, fluctuation of the inner edge of the protoplanetary disk is a suitable setting for the formation of reversely zoned melilite.
Hibonite, spinel, and perovskite are 16O-enriched, in contrast to the adjacent 16O-poor melilite. The replacement of hibonite by spinel could occur by the reaction of pre-existing hibonite and gas, where perovskite could be a by-product of the reaction. The gehlenite mantle of the inclusion is an aggregate of solids condensed from gas and the irregular shape of the W-L rim supports this idea.
Acknowledgments— We are grateful for detailed reviews by Kazuhide Nagashima and Alexander N. Krot. We thank Shoichi Itoh for helpful discussions. This work was supported by the Monka-sho.