An amoeboid olivine inclusion (AOI) in CK3 NWA 1559, comparison to AOIs in CV3 Allende, and the origin of AOIs in CK and CV chondrites




An amoeboid olivine inclusion in CK3 NWA 1559 (0.54 × 1.3 mm) consists of a diopside-rich interior (approximately 35 vol%) and an olivine-rich rim (approximately 65 vol%). It is the first AOI to be described in CK chondrites; the apparent paucity of these inclusions is due to extensive parent-body recrystallization. The AOI interior contains irregular 3–15 μm-sized Al-bearing diopside grains (approximately 70 vol%), 2–20 μm-sized pores (approximately 30 vol%), and traces of approximately 2 μm plagioclase grains. The 75–160 μm-thick rim contains 20–130 μm-sized ferroan olivine grains, some with 120º triple junctions. A few coarse (25–50 μm-sized) patches of plagioclase with 2–18 μm-thick diopside rinds occur in several places just beneath the rim. The occurrence of olivine rims around AOI-1 and around many AOIs in CV3 Allende suggests that CK and CV AOIs formed by the acquisition of porous forsteritic rims around fine-grained, rimless CAIs that consisted of diopside, anorthite, melilite, and spinel. Individual AOIs in carbonaceous chondrites may have formed after transient heating events melted their olivine rims as well as portions of the underlying interiors. In AOI-1, coarse plagioclase grains with diopside rinds crystallized immediately below the olivine rim. Secondary parent-body alteration transformed forsterite in the rims of CV and CK AOIs into more-ferroan olivine. Some of the abundant pores in the interior of AOI-1 may have formed during aqueous alteration after fine-grained melilite and anorthite were leached out. Chondrite groups with large chondrules tend to have large AOIs. AOIs that formed in dust-rich nebular regions (where CV and CK chondrites later accreted) tend to be larger than AOIs from less-dusty regions.


Amoeboid olivine inclusions (AOIs), also known as amoeboid olivine aggregates (AOAs), are the most abundant variety of refractory objects in carbonaceous chondrites (e.g., Kornacki and Wood 1984; Weber and Bischoff 1997; Krot et al. 2002, 2004a). These irregularly shaped inclusions are composed of major olivine; major to minor Ca-pyroxene; minor anorthite; accessory metallic Fe-Ni and perovskite; and in some minimally altered carbonaceous chondrites, accessory spinel ± low-Ca pyroxene ± melilite (Krot et al. 2004a). AOIs in altered carbonaceous chondrites contain secondary phases such as nepheline, sodalite, andradite, magnetite, carbonate, Ca,Fe-rich pyroxene, ferrous olivine, Fe-Ni sulfide, and phyllosilicate (e.g., Grossman and Steele 1976; Hashimoto and Grossman 1987; Krot et al. 2004a). Some AOIs enclose small CAIs (or they partly mantle large CAIs) with Wark-Lovering rims (Komatsu 2003; Krot et al. 2004a; Weisberg et al. 2004). Rare, extensively melted AOIs containing low-Ca pyroxene texturally resemble ferromagnesian type-I porphyritic chondrules, suggesting that AOIs might form a link between CAIs and chondrules (Krot et al. 2004b; Weisberg et al. 2004). This notion is consistent with the report of a Ca,Al-rich chondrule enclosed within an AOI in CO3.5 Lancé (McSween 1977a).

AOIs have been described in most major groups of carbonaceous chondrites (CM, CO, CV, CR, CH, CB), in ungrouped carbonaceous chondrites (e.g., Acfer 094 and Adelaide), and in the Semarkona LL3.0 ordinary chondrite (e.g., Grossman and Steele 1976; Kornacki and Wood 1984; MacPherson et al. 1988; Rubin 1998; Komatsu et al. 2001; Aléon et al. 2002; Chizmadia et al. 2002; Krot et al. 2004; Weisberg et al. 2004; Itoh et al. 2007).

Although AOIs have been noted in CK3 NWA 1559 (Brandstätter et al. 2003), CK4 Karoonda (McSween 1977b), and CK4-an Maralinga (Keller et al. 1992), the present manuscript is the first detailed description of a CK AOI. As documented below, AOIs vary in size and in their modal abundances of constituent phases among different chondrite groups. Thus, a detailed description of the first AOI in a chondrite group has the potential of broadening the known range of AOI properties, potentially leading to new insights into the origin of these inclusions. This particular inclusion is important because it is in a type-3 chondrite; presumably, rare AOIs in CK4 chondrites have been more severely recrystallized.

Analytical Procedures

Thin section LMT 34 of NWA 1559, borrowed from the Naturhistorisches Museum in Vienna, was examined in transmitted and reflected light with an Olympus BX60 petrographic microscope. Also examined were nine large, rimmed AOIs selected from Allende in UCLA thin sections 459, 473, 474, and 502. Mineral compositions were determined with the JEOL electron microprobe at UCLA using natural and synthetic standards, a sample current of 15 nA, an accelerating voltage of 15 keV, 20 s counting times per element, ZAF corrections, and a focused beam. Backscattered electron (BSE) images were also made with the JEOL probe. The EDS function of the probe was used in an attempt to identify the small high-Z grains within plagioclase in the NWA 1559 AOI. The sizes of AOIs in CM chondrites were measured microscopically with a calibrated reticle.


Petrography and Mineralogy

NWA 1559 AOI-1

The amoeboid olivine inclusion in NWA 1559, dubbed AOI-1, is 0.54 × 1.3 mm in plan view (Fig. 1). The AOI interior, which constitutes approximately 35 vol% of the entire inclusion (in two dimensions), consists of irregular 3–15 μm-sized Al-bearing diopside grains (Fs1.6Wo49) (approximately 70 vol%), 2–20 μm-sized irregularly shaped pores (approximately 30 vol%), and traces of approximately 2 μm-sized plagioclase grains (Table 1; Fig. 2). Many of the diopside grains in the interior appear to have eroded margins; some of the grains are elongated and are similar in shape to the diopside rinds that surround coarse plagioclase grains near the AOI rim (see below). In several cases, diopside grains partially or completely surround rounded pores. Within a network of locally connected pores in the AOI interior, there are small Ca-pyroxene grains with a host of irregular shapes, some resembling runic letters (Fig. 3).

Table 1. Mean compositions (wt%) of phases in AOIs in NWA 1559 and Allende
SampleCK3 NWA 1559 AOI-1CV3 Allende AOI 473-3
No. of points121291131
SiO236.7 ± 0.358.6 ± 1.954.1 ± 0.537.80.1652.948.5
TiO2<0.04<0.040.10 ± 0.03<<0.04
Al2O3<0.0426.1 ± 1.11.9 ± 1.1<0.0464.82.30.07
Cr2O3<0.04<0.040.07 ± 0.06<0.040.900.09<0.04
FeO30.6 ± 0.40.30 ± 0.241.0 ± 0.327.421.41.127.7
MnO0.24 ± 0.04<0.04<
MgO32.9 ± 0.2<0.0418.1 ± 0.436.113.718.50.48
CaO0.15 ± 0.117.8 ± 1.524.6 ±
Na2O<0.047.2 ± 0.80.10 ± 0.09<
K2O<0.040.19 ± 0.060.06 ± 0.02<0.04<0.04<0.04<0.04
EndmemberFa 34.3 ± 0.4Ab 61.8 ± 6.8Fs 1.6 ± 0.4Fa 30 Fs 1.7Fs 48
  Or 1.1 ± 0.3Wo 48.6 ± 0.3  Wo 48Wo 51
Figure 1.

AOI-1 in CK3 NWA 1559 showing its coarse ferroan olivine rim enclosing a fine-grained diopside-rich porous interior. The bottom of the rim extends below the edge of the image. The clastic matrix material of the CK host that surrounds the inclusion is recrystallized and appears to have experienced impact crushing. The outer boundary of the rim of the inclusion is difficult to discern, but can be taken to be at those areas where the olivine grain size decreases appreciably or where large pyroxene and plagioclase grains occur (dark gray). Backscattered electron (BSE) image.

Figure 2.

Interior of AOI-1 showing its fine-grained porous interior and portions of the coarse rim. The rim contains coarse olivine grains, some with 120° triple junctions; also present just beneath the rim are coarse plagioclase grains with elongated diopside rinds. BSE image.

Figure 3.

Numerous pores and irregularly shaped diopside grains in the interior of AOI-1 in NWA 1559. Coarse plagioclase grains with diopside rims near the inclusion rim occur at right and bottom. Coarse olivine in the rim occurs at bottom right. BSE image.

Diopside in the inclusion has lower Al2O3 (1.9 wt%; Table 1) than diopside in AOIs from reduced CV chondrites (3.3–13.1 wt%, n = 8; table 1 of Komatsu et al. 2001), but it is within the range of diopside in Allende AOIs (1.6–8.4 wt%; table 3 of Grossman and Steele 1976) and in CR AOIs (1.3–13.2 wt%; table 4 of Aléon et al. 2002).

The 75–160 μm-thick rim of the AOI in NWA 1559 constitutes approximately 65 vol% of the inclusion (in two dimensions) (Figs. 1 and 2). The rim contains ferroan, 20–130 μm-sized olivine grains (Fa34 ± 0.4; Table 1) with faceted and fragmental shapes, some forming 120º triple junctions. Coarse (25–50 μm-sized) patches of plagioclase (Ab62 Or1.1) with 2–18 μm-thick diopside rinds occur in several places just beneath the olivine rim (Table 1; Fig. 4). EDS analysis indicates that the bright-shaded submicrometer-sized grains within the plagioclase are probably spinel-hercynite, but they are too small to analyze quantitatively with the electron microprobe. Because the matrix olivine grains have grown coarser during parent-body thermal metamorphism (cf. Kallemeyn et al. 1991), it is difficult to discern the boundary between the olivine rim of the inclusion and the surrounding matrix. Nevertheless, it is likely to be at those locations where the olivine grain size decreases appreciably or where large pyroxene and plagioclase grains from the matrix occur (Fig. 1).

Figure 4.

Higher magnification view of a portion of the rim of AOI-1 in NWA 1559. There is low porosity in the rim and much higher porosity in the interior. The small (submicrometer-sized) light-shaded grains within the plagioclase are spinel-hercynite. BSE image.

The ferroan olivine in the rim (Fa34) is homogeneous and similar to the mean composition of olivine in some equilibrated CK chondrites (e.g., Fa33 in Maralinga and EET 83311; table 1 of Kallemeyn et al. 1991). In contrast, olivine in Allende AOIs is highly variable (Fa1–36; Grossman and Steele 1976) and has an average composition near Fa17.

Allende AOIs

The nine studied AOIs from Allende (459-1, 459-2, 459-3, 459-4, 473-1, 473-3, 474-1, 474-2, 502-1) are irregularly shaped, partly lobate objects ranging in size from 340 × 350 μm (473-1) to 2090 × 3470 μm (459-3). They each consist of a 10–80 μm-thick olivine rim (most rims are 20–40 μm thick) enclosing isolated 30–50 μm-sized porous patches of Al,Ti-diopside and anorthite ± spinel (Table 1; Fig. 5). As in AOI-1 from NWA 1559, diopside tends to form rims around the plagioclase (especially evident at the top of Fig. 5). In some cases, olivine peninsulas from the rims extend into the interior patches. The olivine grains in the rims consist of low-FeO cores surrounded by ferroan rinds (Fig. 6). The ferroan olivine occurs along fractures within the low-FeO olivine cores and at their grain boundaries; in a few cases, the olivine grain boundaries form 120º triple junctions. Minor phases in the AOIs include hedenbergite (e.g., Table 1), sodalite, and nepheline (common phases produced by secondary alteration in Allende AOIs; e.g., Grossman and Steele 1976; Krot et al. 1998, 2004a). Similar phases occur in the fine-grained matrix of oxidized CV chondrites (e.g., Krot et al. 1998), in the igneous rims enclosing some Allende chondrules (Rubin 1984a), and within many Allende CAIs (e.g., MacPherson and Grossman 1984). The compositions of the phases in Allende AOI 473-3 (Table 1) are within the ranges of those previously reported for Allende AOIs (e.g., Grossman and Steele 1976).

Figure 5.

Allende AOI 473-1 showing a coarse olivine rim enclosing an interior consisting of smooth diopside rinds, plagioclase grains, and large pores. Olivine in the rim consists of low-FeO cores surrounded by ferroan olivine. BSE image.

Figure 6.

Allende AOI 502-1 with a very thick olivine rim enclosing small isolated interior patches consisting of plagioclase, diopside, and large pores. Olivine in the rim consists of low-FeO cores surrounded by ferroan olivine; the lobate periphery of the rim consists exclusively of ferroan olivine. BSE image.


Is AOI-1 Really an AOI?

AOI-1 differs from normal AOIs in other carbonaceous chondrites in four principal respects: (1) there are no olivine grains in the AOI-1 core, (2) olivine is restricted to the mantle of this object, (3) olivine is coarser and more ferroan in AOI-1 than in AOIs from other carbonaceous chondrites, and (4) plagioclase is more sodic than in AOIs from more-pristine unoxidized carbonaceous chondrites. As an inclusion consisting principally of diopside, olivine, and plagioclase, AOI-1 is clearly related to AOIs, but is it a genuine AOI, a new type of AOI, or a closely related type of inclusion?

I address these points here:

  • 1.Although olivine is absent from the core and restricted to the mantle of AOI-1, this texture is not very different from that of the majority of Allende AOIs. Figure 5 shows that Allende AOI 473-1 contains separated patches (i.e., cores) that consist mainly of plagioclase, diopside, and pores with very little olivine. Although smaller and more compressed, these patches are completely analogous to the core of AOI-1 from NWA 1559. Allende AOI 502-1 (Fig. 6) has even smaller plagioclase-diopside core patches; they differ principally in containing some olivine, but the olivine in these cores appears to have crystallized from an olivine-normative melt that invaded the patches during AOI melting.
  • 2.Olivine is essentially restricted to the mantle of Allende AOI 473-1, but the mantle constitutes the majority of the inclusion. This is the main difference between its texture and that of NWA 1559 AOI-1. The possibility exists, however, that the high proportion of olivine mantle in Allende 473-1 is an artifact caused by slicing the inclusion near its mantle-rich end.
  • 3.Because the olivine grains in the mantle of the Allende AOIs impinge on one another, it is difficult to determine their original sizes; the low-FeO cores of the olivine grains range up to approximately 10 μm. The approximately 40–50 μm-average size of the olivine grains in the mantle of NWA 1559 AOI-1 (Fig. 4) may originally have been of comparable size, but grew coarser during parent-body metamorphism. Equilibration during metamorphism led to the high FeO contents of the olivine grains in the rim of AOI-1; they are compositionally equivalent (Fa34) to the rims of compositionally zoned olivine grains in the whole rock (Brandstätter et al. 2003).

It seems that the textural differences between AOI-1 and the Allende AOIs are due mainly to the different thermal histories of these objects. I conclude that NWA AOI-1 has a texture that is within the range attributable to that in normal CV-chondrite AOIs. Its ferroan olivine composition was caused by parent-body metamorphism.

  • 4.Because plagioclase in AOIs from reduced CV3 chondrites is close to endmember anorthite in composition (Komatsu et al. 2001), the more sodic plagioclase in AOI-1 (Ab62 Or1.1; Table 1) is anomalous. Although Na is mobile during parent-body alteration (Tomeoka and Itoh 2004), it does not seem likely that the sodic plagioclase in AOI-1 developed from more-anorthitic plagioclase during metamorphism because (1) diffusion rates of the coupled NaSi-CaAl substitution in plagioclase are sluggish (e.g., Hart 1981; Grove et al. 1983; Morse 1984; Cherniak 2003); (2) olivine grains in the NWA 1559 whole-rock are compositionally zoned (i.e., incompletely equilibrated) (Brandstätter et al. 2003), and diffusion in plagioclase is much slower than in olivine (Hart 1981); (3) the homogeneity of plagioclase compositions in type-4 OC may not be due to parent-body metamorphism, but instead to the equilibration of chondrule mesostases prior to feldspar crystallization (Kovach and Jones 2010); and (4) even in the paired Y-75097 and Y-793241 L6 breccias, plagioclase is only partly equilibrated in H-like chondrite clasts that were metamorphosed along with their hosts (Prinz et al. 1984). (Nevertheless, the small spinel-hercynite grains within the plagioclase in AOI-1 were probably modified by parent-body metamorphism and initially were very low in FeO; diffusion in spinel is many orders of magnitude faster than in plagioclase [Sheng et al. 1992; Cherniak 2003].)

There is another parent-body process that can modify plagioclase compositions—shock metamorphism. CK chondrites have highly variable plagioclase compositions, e.g., An14–69 in CK4 Karoonda and An22–82 in CK4 ALH 84038 (table 1 of Kallemeyn et al. 1991). Plagioclase has a low impedance to shock compression (Schaal et al. 1979) and the compositional variability of plagioclase in CK chondrites is probably due to shock-induced melting, fractionation, and recrystallization (Rubin 1992). NWA 1559 appears crushed (Fig. 1) and has obviously been affected by impact processes. It is plausible that the plagioclase in AOI-1 was melted and recrystallized. It also could have been compositionally modified by aqueous alteration (cf. Kovach and Jones 2010). It thus seems likely that the plagioclase in AOI-1 was initially much more calcic. Because the differences between AOI-1 and AOIs in other carbonaceous chondrites can be accounted for by differences in thermal history, shock history, and aqueous alteration, I conclude that AOI-1 is a genuine AOI.

Secondary Alteration

CK and CV carbonaceous chondrites are closely related groups containing similarly sized chondrules, similar proportions of different chondrule textural types, similar bulk O-isotopic compositions, and similar opaque mineralogy (metallic Fe-Ni, sulfide, magnetite); they also have overlapping distributions of major and trace element abundances (Kallemeyn et al. 1991; Greenwood et al. 2010; Rubin 2010). Both groups contain some centimeter-sized CAIs (Grossman 1975; MacPherson and Grossman 1984; Meteoritical Bulletin Database entry for CK4 Hammadah al Hamra 280). It is plausible, even likely, that CK chondrites are impact-heated, crushed, and oxidized/altered samples of CV3 chondrites (e.g., Greenwood et al. 2010; Isa et al. 2012).

CK chondrites are highly oxidized; they have abundant magnetite and very little metallic Fe-Ni (Kallemeyn et al. 1991; Rubin 1993; Geiger and Bischoff 1995; Neff and Righter 2006). Oxidized CV chondrites comprise two subgroups: an Allende-like oxidized subgroup (CV3OxA) characterized mainly by chondrule primary minerals having been replaced by magnetite, Ni-rich sulfide, ferroan olivine (Fa30–60), and feldspathoids; and a Bali-like oxidized subgroup (CV3OxB) characterized by abundant phyllosilicates and chondrules with primary minerals having been replaced by phyllosilicate, magnetite, Ni-rich sulfide, very ferroan olivine (Fa95–100), and hedenbergite (McSween 1977c; Weisberg et al. 1997; Krot et al. 1998, 1997). It seems likely that both CK chondrites and the two subgroups of oxidized CV chondrites formed from more reduced materials by aqueous alteration on their parent bodies (Krot et al. 1998).

Because there are no known samples of reduced CK chondrites, it is reasonable to examine AOIs from reduced CV chondrites (such as Vigarano, Leoville, and Efremovka) to approximate the primary properties that AOIs in NWA 1559 and Allende may have possessed in the solar nebula. As described by Krot et al. (2004a), AOIs in reduced CV chondrites consist of forsterite, metallic Fe-Ni, and variable amounts (<1 to >90 vol%) of fine-grained CAIs containing Al-diopside, anorthite, spinel, and melilite; the melilite was partially replaced in the solar nebula by anorthite or a fine-grained mixture of spinel and Al-diopside (Krot et al. 2004a).

The minor phases in the Allende AOIs (hedenbergite, sodalite, and nepheline) are secondary minerals widely interpreted as having formed by alteration on the CV parent body (Krot et al. 1998). The olivine grains in the Allende AOI rims presumably were originally all forsterite; diffusion of FeO during parent-body metamorphism/alteration (not via condensation as proposed by Peck and Wood 1987 and Hua et al. 1988) resulted in the grains being transformed into ones consisting of low-FeO olivine cores surrounded by ferroan olivine rinds (Fig. 6). This interpretation is favored by Krot et al. (2004a) and is consistent with the O-isotopic evidence: forsterite, spinel, and diopside are 16O rich (Δ17O ≅ −20 to −25‰) in Allende AOIs, whereas ferroan olivine and other secondary phases are 16O poor (Δ17O ≅ −5‰) (Imai and Yurimoto 2003) as is the Allende whole rock (Δ17O = −3.52‰; Clayton and Mayeda 1999).

Similarly, the ferroan olivine grains in the rim of AOI-1 in NWA 1559 (Fa34.3 ± 0.4) (Fig. 4) were probably also initially forsterite. The beginnings of the process of transforming forsterite into more ferroan olivine are evident in AOIs from CO3.2 Kainsaz wherein thin veins of ferroan olivine transect small grains of forsterite (Fig. 7c of Chizmadia et al. 2002).

Figure 7.

Type-1 porphyritic olivine chondrule F3o in CR2 chondrite LAP 02342. It consists of a primary chondrule enclosed by a secondary shell and an outlying igneous rim. Discontinuous metal rims (white) occur between the primary chondrule and the secondary shell and between the secondary shell and the igneous rim. Low-FeO olivine is dark grey; mesostasis is light grey. BSE image.

If the interior of AOI-1 in NWA 1559 was similar to those in reduced CV chondrites, then it initially consisted of Al-diopside, anorthite, spinel, and melilite. Hydrothermal experiments (e.g., Nomura and Miyamoto 1998) show gehlenite (the main component of melilite in most refractory inclusions; Brearley and Jones 1998), to be very susceptible to parent-body alteration (forming anorthite, grossular, and feldspathoids; Allen et al. 1978; MacPherson et al. 1981; Wark 1981). This is consistent with its paucity in CM CAIs (Armstrong et al. 1982; Rubin 2007) and its evident corrosion in Allende CAIs (e.g., MacPherson et al. 1983; MacPherson and Grossman 1984). Under conditions wherein gehlenite breaks down, diopside and spinel are unaffected (Nomura and Miyamoto 1998). In Allende AOIs, Al,Ti-diopside was partially replaced by phyllosilicates (Krot et al. 2004a). Although anorthite is more resistant to alteration than melilite (Nomura and Miyamoto 1998), in the Bali-like oxidized CV chondrites, anorthite was either replaced by phyllosilicates or leached out to form pores (Krot et al. 2004a). It is thus plausible that many of the pores in the interior of AOI-1 in NWA 1559 were formed on the CK parent body by the corrosion of melilite and/or the leaching out of anorthite. It seems possible that some pores represent bubbles formed during melting of the inclusion. A few pores could also have formed by plucking during sample preparation.

Nature of AOI Interiors and Olivine Rim Acquisition

In their studies of CR2 EET 92105, Weisberg et al. (2004) reported three discrete AOIs attached to the Wark-Lovering rim surrounding a Compact type-A CAI. Forsterite occurs in these three AOIs as well as in portions of the Wark-Lovering rim around the CAI. This association of AOIs and CAIs is consistent with the interpretation of Krot et al. (2004a) that the Al-diopside, anorthite, spinel, and melilite in the interiors of AOIs in reduced CV chondrites are components of (rimless) fine-grained CAIs. If this interpretation is correct, it implies that the nine Allende AOIs in the present study as well as AOI-1 from CK3 NWA 1559 formed from fine-grained CAIs that acquired thick forsterite rims. In fact, this appears to have been a common process: the majority of Allende AOIs have rims (Figs. 5 and 6); these are the “rimmed olivine aggregates” of Kornacki and Wood (1984) and Kornacki et al. (1983). In these cases, parent-body alteration caused the original forsterite grains in the rims to be replaced by ferroan olivine. The rims are composed of 10–30 μm-sized olivine grains, some of which form 120º triple junctions (Kornacki and Wood 1984).

The forsterite rims around CV and CK AOIs may be related to the “accretionary rims” containing 20–40 μm-sized anhedral forsterite grains that surround many CV CAIs (MacPherson et al. 1985). The forsterite rims in CV chondrites were probably originally rich in 16O (e.g., Cosarinsky et al. 2008) and were acquired by the AOIs early in the history of the solar nebula (e.g., Krot et al. 2004a), prior to chondrule formation. The rims became coarser during episodes of flash heating (see below). Their present ferroan compositions resulted from parent-body alteration processes (Imai and Yurimoto 2003; Krot et al. 2004a). Because CK chondrites may be metamorphosed and oxidized CV chondrites (Greenwood et al. 2010; Isa et al. 2012), it is plausible that CK AOIs started out as CV AOIs and, by analogy, also probably had 16O-rich forsterite rims.

One alternative is that the ferroan olivines in the AOI rims in Allende and NWA 1559 resulted from late acquisition of ferromagnesian chondrule material followed by remelting during episodes of chondrule formation. I reject this idea because (1) most Allende AOIs have ferroan olivine rims (Kornacki and Wood 1984), rendering such collisions statistically unrealistic; (2) many of the ferroan olivine grains in the rims have low-FeO cores (indicative of parent-body diffusion); and (3) none of the AOIs in the reduced (i.e., the most pristine) CV chondrites have these features (Krot et al. 2004a).

CAI populations in different chondrite groups tend to plot on or close to the CCAM (carbonaceous-chondrite-anhydrous minerals) line on the standard three-O-isotope graph (e.g., figs. 24–27 of MacPherson 2005); the most 16O-rich CAIs in each group have similar bulk O-isotopic compositions (Δ17O approximately –25‰). This suggests that CAIs from different chondrite groups formed within the same O-isotopic reservoir (although it is not clear how physically large this reservoir was). The evidence in some CAIs for once-live 10Be and, perhaps, 7Be (McKeegan et al. 2000; Sugiura et al. 2001; MacPherson et al. 2003; Chaussidon et al. 2006) and the fact that 10Be is not produced by stellar nucleosynthesis (but rather from 16O in CAI precursors by nuclear spallation reactions probably caused by intense radiation in the solar nebula; McKeegan et al. 2000; Gounelle et al. 2001, 2006; Marhas et al. 2002; Chaussidon et al. 2006) imply that the majority of CAIs were produced relatively close to the Sun.

Rubin (2012) proposed that, after formation, some Compact type-A CAIs were transported to the CV region of the nebula where abundant 16O-rich, forsterite-rich dust had accumulated. Possible transportation mechanisms included jets, turbulent diffusion, gravitational instabilities, and advection due to viscous expansion of the disk (e.g., Gounelle et al. 2001; Shu et al. 2001; Cuzzi et al. 2003; Boss 2004; Wood 2004; Ciesla 2010; Jacquet et al. 2011). At the CV location, many of the Compact type-A CAIs stuck together; the aggregates were mantled by forsteritic dust and flash-heated, leading to the formation of large Fluffy type-A CAIs and type-B CAIs.

AOIs in Allende and NWA 1559 may have formed in an analogous way: fine-grained CAIs consisting largely of Al-diopside, anorthite, spinel, and melilite were transported by aerodynamic forces to the CV-CK region of the nebula. There they were mantled by porous clumps of 16O-rich forsterite (e.g., Imai and Yurimoto 2003) and subjected to flash heating (as discussed below). Large portions of the nebula may have contained 16O-rich dust at this time (long before chondrule formation) due to the radial transport of multimillimeter-to-centimeter-sized, highly porous, 16O-rich, forsterite-rich dustballs (Rubin 2011).

Heating of Rims

Many chondrules exhibit features that appear to have formed by the flash melting of dust mantles surrounding the chondrules. These features include (1) igneous rims around chondrules (Rubin 1984; Rubin and Wasson 1987; Rubin et al. 1990; Krot and Wasson 1995) (e.g., Fig. 7), (2) secondary shells of enveloping (i.e., nested) compound chondrules (Wasson et al. 1995) (e.g., Fig. 7), and (3) microchondrule-bearing fine-grained rims around chondrules (Krot and Rubin 1996; Krot et al. 1997). In many cases, the heating mechanism must have had a steep thermal gradient and was capable of melting the porous dusty mantles around the chondrules without melting the primary chondrules themselves. Such fine-grained porous dusty mantles have very large surface areas and are more readily melted than intact chondrules, particularly if there was little superheating (Wasson 1993).

Like chondrule formation itself (e.g., Gooding et al. 1980; Grossman and Wasson 1985; Wasson 1993, 1996; Alexander 1994; Jones 1996; Connolly et al. 1998; Rubin 2000; Hewins et al. 2005; Jones et al. 2005; Rubin and Wasson 2005; Russell et al. 2005; Alexander et al. 2008; Weisberg et al. 2011), flash-heating processes affecting chondrule rims are widely assumed to have taken place in the solar nebula (e.g., Rubin 1984a, 2010; Rubin and Wasson 1987; Krot and Wasson 1995; Wasson et al. 1995).

It seems plausible that the AOIs with coarse olivine-rich rims were heated in the nebula by a flash-melting process analogous to the one responsible for producing igneous rims and secondary shells around chondrules. However, AOI melting took place long before chondrule formation. In the case of AOI-1 in NWA 1559, the porous forsteritic rim was melted along with those pyroxene and plagioclase grains that were present in some places just below the rim. Relatively coarse grains of olivine (20–130 μm) and plagioclase (25–50 μm) with 2–18 μm-thick diopside rinds crystallized from the melt, forming a hypidiomorphic-granular-like texture. These olivine grains from the AOI rim are approximately the same size as many of the forsterite phenocrysts in low-FeO chondrules in ordinary and carbonaceous chondrites (e.g., fig. 3 of Scott and Taylor 1983).

Flash-melting of the porous olivine-rich rim around AOI-1 might be expected to have produced olivine grains that crystallized with chain structures or lattice morphologies due to rapid cooling instead of the polyhedral shape that they possess (Donaldson 1976). This difference is probably due to the relatively low viscosity of the AOI rim melt. Silica and alumina are network formers and tend to increase the viscosity of silicate melts (e.g., Blatt et al. 2006). The bulk composition of a melt consisting of 95 wt% forsterite, 4 wt% anorthite, and 1 wt% diopside (approximating the composition of the rim of AOI-1) will contain 42.8 wt% SiO2 and 1.5 wt% Al2O3. The basalts and peridotites studied by Donaldson (1976) have slightly higher mean SiO2 (45.4 wt%) and much higher mean Al2O3 (11.7 wt%). This indicates that the rim would produce a melt of an appreciably lower viscosity than the rocks studied by Donaldson. Donaldson (1976) found that in melts of lower viscosity, olivine grains that are rapidly cooled assume morphologies normally characteristic of those grains that crystallize at lower cooling rates from more-viscous melts. This is consistent with the hypothesis that the polyhedral morphologies of the olivines in the AOI-1 rim formed at fairly high cooling rates during flash melting. The olivine grains may also have grown somewhat coarser and moderately changed their morphologies during parent-body metamorphism.

The variable CaO contents in olivine in AOIs from primitive chondrites (Sugiura et al. 2009) may result from variations in the number of reheating episodes and the intensities of reheating experienced by different inclusions. (Reheating can increase the CaO content of forsterite in AOIs [Sugiura et al. 2009].) The relatively Mn-rich magnesian olivines in some AOIs (Grossman and Steele 1976; Aléon et al. 2002; Weisberg et al. 2004; Sugiura et al. 2009) may have been inherited from the nebula (Ebel et al. 2012).

During the flash-heating event that melted the porous olivine rim of AOI-1, the phases in the interior of the inclusion were not melted. (There are many analogous cases wherein the outer surfaces but not the interiors of millimeter-sized objects were melted in the nebula. These include normal-sized chondrules with microchondrule-bearing rims [Krot and Rubin 1996; Krot et al. 1997], enveloping compound chondrules [Wasson et al. 1995], and chondrules with igneous rims [Rubin 1984; Rubin and Wasson 1987; Krot and Wasson 1995]. In the latter two cases, the rims were porous structures prior to melting.) After the porous rim of AOI-1 was melted, the grains in the interior of AOI-1 remained small (with high surface/volume ratios) (Fig. 2), thereby facilitating their alteration and leaching during subsequent parent-body processing.

The majority of Allende AOIs experienced analogous flash-heating events. They developed relatively coarse forsteritic rims, while much of the interior remained as small grains of Al-diopside, anorthite, spinel, and melilite (Figs. 5 and 6). During melting, the cores of the Allende AOIs were compressed and, in the case of AOI 502-1 (Fig. 6), the olivine-normative melt invaded the core patches, causing olivine crystallization adjacent to relict grains of plagioclase and diopside. The difference between AOI-1 in NWA 1559 and the Allende AOIs in the sizes of olivine and plagioclase reflect the different thermal histories of these objects.

Subsequent alteration on the parent body made the olivine more ferroan and produced such secondary phases as hedenbergite, nepheline, and sodalite (e.g., Krot et al. 2004a). Many of the pores (Fig. 2) could have been created by the leaching of primary melilite during alteration.

AOIs are thus not unmelted aggregates—the rims of different AOIs have experienced variable amounts of melting. However, if the melting had been appreciably more extensive, spheroidal shapes, and porphyritic chondrule-like textures would have been produced (Komatsu et al. 2009). Remelting of AOIs could produce Ca-Al-rich chondrules (Rubin 2004).

Correlation of AOI and Chondrule Size

AOIs vary in size among chondrite groups. Groups (LL, CM, CO) with small average chondrule sizes (150–570 μm) tend to have small average AOI sizes (200–260 μm); groups (CR, CK, CV) with large average chondrule sizes (700–910 μm) tend to have relatively large average AOI sizes (600–3000 μm) (Table 2; Fig. 8). It seems plausible that the mechanisms proposed to have produced large chondrules in the CR, CK, and CV regions of the nebula (discussed below, e.g., Rubin 2010) were also responsible, at an earlier epoch, for making large AOIs.

Table 2. Sizes of AOIs and chondrules in different chondrite groups
GroupAOI size range (μm)AOI mean size (μm)Chondrule mean size (μm)g
  1. Sizes are apparent mean diameters measured in thin section.

  2. References: a = Itoh et al. (2007); b = this study; c = Chizmadia et al. (2002); d = Weisberg et al. (2004); e = Grossman and Steele (1976); f = Komatsu et al. (2001); g = Rubin (2010).

LL3 40–350a200570
CR2 70–1200d600700
Figure 8.

Correlation between the mean size of AOIs and the mean size of chondrules in carbonaceous- and ordinary-chondrite groups. The sizes of these objects are apparent diameters, measured in thin section. Chondrite groups with large chondrules tend to have large AOIs.

The relative formation locations of the different chondrite groups were modeled by Wasson (1988) and Rubin (2010, 2011); in order of increasing heliocentric distance, the groups are EH-EL, H-L-LL, R, CR, CV-CK, CM-CO, CI. Different amounts of nebular dust were inferred by Rubin (2010, 2011) to have been present at these locations. The groups that accreted closest to the Sun (enstatite and ordinary chondrites) formed in regions with small amounts of ambient dust; R chondrites formed in a region farther from the Sun with appreciably more dust. The dust abundance increased in the CR region of the nebula and probably peaked in the CV-CK region; it decreased with increasing distances where the CM-CO and CI chondrites formed.

During the epoch when AOIs were forming, the dust was likely to have contained abundant 16O-rich forsterite. This is the material that formed porous clumps that coated the fine-grained CAIs; the CAIs had been transported to those locations from near the Sun. The entire CAI-forsterite-mantle assemblage was later melted. If it is correct that more dust was present in the CV-CK location, one might expect that the AOIs from these groups would have higher modal abundances of olivine than AOIs that formed in regions inferred to have lower dust concentrations. The limited data available are consistent with this inference. Although the olivine content of individual AOIs is quite variable, when pore space is included, on average, olivine constitutes (in vol%) approximately 50% of OC, CO, and CR AOIs (Aléon et al. 2002; Chizmadia et al. 2002; Itoh et al. 2007) and approximately 80% of AOIs in reduced CV chondrites (Komatsu 2003) (e.g., Figs. 5 and 6). (The sole studied AOI in a CK3 chondrite (Fig. 1) contains 65 vol% olivine.)

By the time chondrules formed, the dust had become appreciably lower in 16O. This may have been caused by subsequent episodic influxes of 16O-poor material to the nebula from the parental molecular cloud (e.g., Wasson 2000).


I suggest that amoeboid olivine inclusions (AOIs) in CK and CV chondrites were produced from fine-grained CAIs rich in diopside, anorthite, and melilite that formed relatively near the Sun. They were later transported to the dust-rich CV-CK nebular location where they acquired porous rims of forsterite in a region with Δ17O similar to that near the Sun. Flash-heating events, akin to those that later formed chondrules, melted the porous forsterite mantles of the AOIs and created thick olivine rims. In some cases, some refractory mineral grains just beneath the rims were also melted. AOIs inferred to have formed in dusty regions of the nebula tend to be larger and richer in olivine than those that formed in less-dusty regions. Secondary processing on the CV and CK parent bodies altered the inclusions and created some of the pores in AOI interiors by the leaching out of fine-grained anorthite and melilite.


I thank F. Brandstätter of the Naturhistorisches Museum in Vienna for the loan of the NWA 1559 thin section. I am also grateful to J. T. Wasson for comments and M. Komatsu for providing data from her dissertation. Reviews by A. El Goresy, M. Komatsu, M. K. Weisberg, P. C. Buchanan, and L. Chizmadia were helpful in revising the manuscript. This work was supported by NASA Cosmochemistry grant NNG06GF95G.

Editorial Handling

Dr. Michael Zolensky