AN INVESTIGATION INTO THE ORIGIN OF Fe-RICH PRESOLAR SILICATES IN ACFER 094

Grain size separates (0.1–0.5 μm and 0.5–1.0 μm) originally prepared by A. N. Nguyen were used for the measurements in this study. The procedure is discussed in detail in Nguyen (2005) and is briefly outlined here. A small fragment of Acfer 094 was placed in distilled water and disaggregated by a rapid "freeze-thaw" process, followed by ultrasonification. The sample was then repeatedly washed in distilled water and soluble organic compounds, such as toluene, to remove precipitates that might have formed during the disaggregation process and to achieve optimal dispersion of the grains. Size separates of the grains were produced by repeated centrifugation, and the grains were then dispensed onto gold foils pressed into stainless steel stubs. Areas on the gold foils that were densely covered with grains were chosen for measurements in the NanoSIMS.

Oxygen isotopic measurements were done with a Cs⁺ primary beam of a few pA in the Washington University NanoSIMS 50. The beam was rastered over 10 × 10 μm² or, in some cases, 20 × 20 μm² areas in an imaging mode (256² or 512² pixels). Secondary ions of ¹²C⁻, ¹³C⁻, ¹⁶O⁻, ¹⁷O⁻, and ¹⁸O⁻ or ¹⁶O⁻, ¹⁷O⁻, ¹⁸O⁻, ²⁸Si⁻, and ²⁴Mg¹⁶O⁻ were measured in the multi-collection mode together with secondary electrons (SEs). The analyses consisted of multiple (5–25) scans, with analysis times of up to 20 minutes for each scan. The individual layers were added together to make a single image for each analyzed species. Isotopic compositions of the presolar grains were determined following the data reduction procedures discussed in Stadermann et al. (2005), assuming normal (solar) compositions for the bulk of the matrix grains analyzed (¹⁷O/¹⁶O = 3.8 × 10⁻⁴ and ¹⁸O/¹⁶O = 2.0 × 10⁻³). A grain was considered presolar if at least one of its O isotopic ratios differed by more than 5σ from those of the normal grains and the anomaly was present in at least three consecutive layers.

Following the O isotopic analyses, four ¹⁸O-enriched presolar grains were also measured for Si isotopes. For these measurements the Cs⁺ beam was rastered over a 3 × 3 μm² area (256² pixels) centered on the grain, while detecting ¹⁶O⁻, ¹⁸O⁻, ²⁸Si⁻, ²⁹Si⁻, and ³⁰Si⁻ secondary ions for ten imaging scans. The Si isotopic ratios of the presolar grains were normalized by assuming that the surrounding material has solar ratios of ²⁹Si/²⁸Si (5.08 × 10⁻²) and ³⁰Si/²⁸Si (3.36 × 10⁻²). The normalized Si isotopic ratios were then computed as delta values, indicating deviations from solar ratios in parts per thousand.

Iron isotopic measurements were performed for two relatively large (∼300 × 200 nm²) grains using an O⁻ primary ion beam and combined analysis mode, in which multi-collection is combined with magnetic peak jumping; the details of the measurement procedure are described by Floss et al. (2008). A correction was made for the contribution of ⁵⁴Cr⁺ on ⁵⁴Fe⁺ and the Fe isotopic ratios were normalized to the average ⁵⁴Fe/⁵⁶Fe and ⁵⁷Fe/⁵⁶Fe ratios of surrounding isotopically normal silicate grains present on the grain mounts.

The PHI 700 Auger Nanoprobe, which was recently installed at Washington University, was used to obtain the elemental compositions of the O-anomalous grains. Prior to acquiring spectra or elemental maps, controlled sputter cleaning with a diffuse 2 keV, 1 μA Ar⁺ ion beam was done in order to remove surface contaminants. Iterative Auger electron energy spectra of the presolar grains were obtained at a primary beam accelerating voltage of 10 keV and a current of 0.25 nA in the energy range 30–1730 eV. The low current used during the measurements, combined with rastering the electron beam over the area of the grain, reduces the potential for beam damage to the grains (Stadermann et al. 2009). Multiple iterations of the spectra of a given grain were added together to obtain a single spectrum. The average spectrum was differentiated, using a seven-point Savitsky–Golay smoothing and differentiation routine, followed by identification of the elements on the basis of their characteristic peaks (Childs et al. 1995). Compositions of the grains are calculated from peak-to-peak heights in the differentiated spectra and are then corrected for elemental sensitivities. The sensitivity factors computed from repeated Auger measurements on powdered olivine and pyroxene standards, under the same analytical conditions as used in this study, are as follows: Ca, 0.626; O, 0.194; Fe, 0.150; Mg, 0.234; Al, 0.160; and Si, 0.121 (Stadermann et al. 2009). The Auger peaks used for quantification are the same as those specified in Stadermann et al. (2009), e.g., KLL transition peak of O and LMM peak of Fe. The sensitivity factors have the following 1σ uncertainties: Ca, 10.8%; O, 3.6%; Fe, 11.2%; Mg, 9.4%; Al, 24.9%; Si, 11.0% (Stadermann et al. 2009). The high uncertainty for Al reflects the fact that the standards measured to date contain relatively low concentrations of this element. Oxide grain compositions may have greater uncertainties associated with them because the sensitivity factors listed above were computed from measuring silicate mineral standards. Additional errors, such as those associated with the presence of surface contaminants, sample charging, and background noise, are not included in the reported errors, which should therefore be considered lower limits.

Information from the spectra was supplemented with Auger elemental distribution maps of the presolar grains and surrounding areas. Maps were acquired by rastering the areas of interest (3 × 3 μm² or 5 × 5 μm²) at 10 keV and 10 nA for 5–30 scans. The elemental maps provide detailed qualitative information about the elements present in the grains, their spatial distribution, and possible contributions to the Auger spectra from surrounding grains.

Forty O-anomalous grains (i.e., 83%) exhibit ¹⁷O enrichments of up to ∼7 times solar, with sub-solar to solar ¹⁸O/¹⁶O ratios (Table 1). These ¹⁷O-rich grains (Figure 1) are classified as group 1 grains (Nittler et al. 1997). While there are no group 2 grains in our presolar grain inventory, which would have greater ¹⁸O depletions than group 1 grains, there is a possibility that a few grains currently identified as group 1 actually belong to the group 2 category (discussed later in Section 4.1). Grains with ¹⁶O-rich compositions do not exist in our inventory.

Zoom In Zoom Out Reset image size

Four grains with ¹⁸O excesses, classified as group 4 (Figure 1), were identified. The Si isotopic ratios of these grains are plotted in Figure 2. The grains have normal Si isotopic compositions within 1σ errors, except for grain 34A-4, which has a slight enrichment (64‰ ± 26‰) in ³⁰Si (Table 1).

Zoom In Zoom Out Reset image size

Two group 1 grains that were measured for Fe isotopes have normal ⁵⁴Fe/⁵⁶Fe and ⁵⁷Fe/⁵⁶Fe isotopic ratios. Grain 42Fe14 has δ⁵⁴Fe/⁵⁶Fe = −107‰ ± 53‰ and δ⁵⁷Fe/⁵⁶Fe = −192‰ ± 87‰, and grain 42Ff11 has δ⁵⁴Fe/⁵⁶Fe = −113‰ ± 46‰ and δ⁵⁷Fe/⁵⁶Fe = −75‰ ± 87‰. Nickel isotopic ratios could not be calculated because count rates were too low.

All 48 presolar grains were characterized in the Auger Nanoprobe via elemental spectra and high-resolution maps. Figure 3 shows an example of an undifferentiated and differentiated Auger spectrum of the presolar ferromagnesian silicate grain 42Fb2 (size ∼160 × 120 nm²). Cesium was implanted in the sample during NanoSIMS measurements and appears as a complex peak, at around 572 eV in the spectrum. The elemental compositions of the 44 silicate and 4 oxide grains are listed in Table 2, along with the Mg+Fe(+Ca)/Si and (Mg/Mg+Fe) × 100 (mg#) ratios. Grains containing O, Fe, and Mg (with or without Al, Ca, and Si) with mg#s less than 45 are considered Fe-rich and those with mg#s greater than 55 are considered Mg-rich. Grains are classified as either olivine-like or pyroxene-like if their Mg+Fe(+Ca)/Si ratios fall within 1σ of the theoretical ratios of 2 for olivine (Mg,Fe)₂SiO₄ and 1 for pyroxene (Mg, Fe)SiO₃.

Zoom In Zoom Out Reset image size

Twelve grains have Mg+Fe(+Ca)/Si ratios with pyroxene-like compositions and six grains have olivine-like compositions (Figure 4). Since O is not included in the Mg+Fe(+Ca)/Si ratio, but must also be present in stoichiometric proportions, two additional ratios, namely cation/O and O/Si, were evaluated. These ratios can provide additional verification of whether the classification of a silicate grain is consistent with olivine or pyroxene and can identify grains that do not contain O in stoichiometric proportions. The classification of three olivine-like and six pyroxene-like silicate grains is less certain based on these criteria; these grains are marked with asterisks in Table 2 and Figure 4. A large number of grains (16 out of 40) have Mg+Fe(+Ca)/Si ratios intermediate between 1 and 2 (1.3–1.7; Figure 4). Finally, ten grains have Mg+Fe(+Ca)/Si ratios that are either significantly lower than 1 or higher than 2; these are listed under the category of "Other Silicates" in Table 2.

Zoom In Zoom Out Reset image size

Silicate grains that are olivine-like include one grain, 42Fw3, with a forsterite (i.e., Mg-rich end-member) composition (Figure 5). The other olivine-like grains have elevated Fe concentrations (mg#s ranging from 17 to 31). Among the pyroxene-like silicates, one grain, 42Fs13, has an enstatite-like composition. However, the majority of pyroxene-like grains are Fe-rich with mg#s from 27 to 40, including two grains that have compositions that are consistent with Fe-rich end-member pyroxene and fall in the "forbidden zone" of the pyroxene quadrilateral (Deer et al. 1992, p.145). None of the pyroxene-like grains contain Al and only one grain contains Ca.

Zoom In Zoom Out Reset image size

The majority of the silicate grains that have intermediate Mg+Fe(+Ca)/Si ratios are also Fe-rich. Two of these intermediate silicates have non-stoichiometric Fe-rich end-member compositions. A few grains in this category contain small detectable amounts of Ca (1.8–5.7 at.%) in addition to O, Fe, Mg, and Si. The exact classification of individual grains as olivine- or pyroxene-like on the basis of elemental ratios is accompanied by significant uncertainties, but the study of the statistical distributions of various grain types is much more robust. Olivine and pyroxene grain standards cluster around their expected Mg+Fe(+Ca)/Si ratios of 2 and 1, respectively (Stadermann et al. 2009), yet a number of standard grains have Mg+Fe(+Ca)/Si ratios between 1 and 2. Thus, it is possible that some of the silicate grains with intermediate compositions belong to either the olivine- or pyroxene-like categories. However, the number of such grains in our study is significantly larger than what is expected statistically if the inventory was only comprised of olivine and pyroxene grains, indicating that most of the intermediate grains represent a distinct group.

Six of the ten presolar grains labeled "Other Silicates" in Tables 1 and 2 contain Ca and/or Al along with Fe and Mg. Three of these grains, 34A-6, 42F-13, and 42Fv113, are Al-rich with only the first two exhibiting very high Mg+Fe(+Ca)/Si ratios. One grain, 42Fe14, is rich in Si with an O/Si ratio of 1.8 ± 0.2 and contains only minor amounts of Fe and Mg. Two other grains (42Fs4_I & 42Fs4_II) also have Si-rich compositions with O/Si ratios of ∼2. Eight grains classified as "Other Silicates" contain more Fe than Mg (mg# ≤ 40) and two grains show no detectable Mg at all. Grain 42C-5 has the highest Fe content (47.1 ± 5.3 at.%) among all the presolar silicate grains characterized in this work. Finally, grain 42Fu7 has Mg+Fe(+Ca)/Si, cation/O, and O/Si ratios that are only slightly outside the 1σ range for pyroxene.

Among the four oxides, two grains, 34C-6 and 42Fs6, contain Al, Fe, and Mg, one grain, 42Fb3, is spinel-like and the final grain, 42Ff26, is probably a TiO₂ grain (O/Ti = 1.9; Figure 6). Quantification of the Ti-oxide spectrum was done using a sensitivity factor for Ti determined from a TiO standard, which suggests a Ti concentration of 34 at.% in the grain. The Ti error was calculated by assuming a relative uncertainty of 10% (Table 2).

Zoom In Zoom Out Reset image size

A majority of the silicate and oxide grains (44 out of 48) identified in the grain size separates of Acfer 094 are group 1 grains, rich in ¹⁷O and with close-to-solar or sub-solar ¹⁸O/¹⁶O ratios (e.g., Nittler et al. 1997). When a star runs out of H fuel in its core, it expands, cools, and undergoes an episode of deep convection called the first dredge-up, which mixes material from the deep layers of the star into the envelope, thus changing its composition (Boothroyd & Sackmann 1999). This mixing brings CNO products of partial H-burning to the surface. The CNO cycle produces ¹⁷O but destroys ¹⁸O, and as a result, in the first dredge-up the ¹⁷O/¹⁶O ratio of the surface increases drastically and the ¹⁸O/¹⁶O ratio decreases slightly, relative to the initial composition of the star. For stars with masses between 1 and 2.5 M_☉, the ¹⁷O/¹⁶O ratio increases with increasing mass but it decreases in stars that are more massive (Boothroyd et al. 1994; Boothroyd & Sackmann 1999). The ¹⁸O/¹⁶O ratio, however, is not a function of mass but depends primarily on the initial metallicity of the parent star. These predictions are in reasonable agreement with O isotopic ratios that have been measured spectroscopically in evolved stars (e.g., Harris et al. 1988). All the group 1 silicate grains found in this study have compositions consistent with an origin in low-mass RG and AGB stars with near solar metallicities. Copious amounts of silicate dust can form in stars with initial masses > 4 M_☉ that undergo third dredge-up (Gail et al. 2009). However the O-isotopic composition of silicate grains from such stars should exhibit ¹⁷O/¹⁶O > 0.002–0.004 (Nittler 2009), which is rarely observed for silicate grains, including ones identified in this study. Two ferromagnesian silicate grains, 42F-13 and 42Fe8O17, are characterized by higher ¹⁷O/¹⁶O isotopic ratios than the other group 1 grains (Figure 1), but are still well below the maximum ¹⁷O/¹⁶O ratios predicted for first dredge-up in stars of near solar metallicity (Nittler et al. 2008).

Modeling by Nguyen et al. (2007) has shown that ion imaging measurements of tightly packed grains in the NanoSIMS are diluted by the surrounding isotopically normal silicates, resulting in measured isotopic compositions that are closer to solar values than the intrinsic compositions of the grains. While this dilution affects all measured compositions, the effect is most pronounced for grains with compositions that are depleted in the low-abundance isotopes relative to solar, such as the ¹⁸O-depleted group 2 grains (Nittler et al. 1997). Because of this dilution effect, some of our group 1 grains with ¹⁸O/¹⁶O ratios close to 10⁻³ (e.g., 42B-1) may actually belong to group 2. Large ¹⁸O depletions in group 2 grains indicate that they probably formed in low-mass AGB stars (1.2–1.8 M_☉) undergoing cool bottom processing (Wasserburg et al. 1995; Nollett et al. 2003). Cool bottom processing, also referred to as "deep mixing," occurs when material from the envelope circulates down into the radiative zone above the H shell, and rapidly brings up nuclear processed material. The processed material is devoid of ¹⁸O that has been rapidly destroyed, via the reaction ¹⁸O (p, α) ¹⁵N.

Our oxide grain population contains a presolar TiO₂ grain (42Ff26) with a solar ¹⁸O/¹⁶O ratio and a ¹⁷O/¹⁶O ratio that is 1.9 times the solar value. Individual presolar Ti oxide grains have previously been identified in residues of several meteorites (Nittler et al. 2008). Three group 1 Ti oxide grains found by Nittler et al. (2008) have close-to-solar Ti isotopic ratios. Titanium isotopes are affected by neutron capture reactions and small shifts in the Ti abundances are predicted to occur after third dredge-up in low-mass AGB stars (Zinner et al. 2007). However, we are not able to acquire Ti isotopic data for the TiO₂ grain found in this study due to the small size (270 × 120 nm²) of this grain (Figure 6) and the fact that Ti isotopic measurements in the NanoSIMS need to be made with an O⁻ primary beam, which has a spatial resolution of no better than 300 nm.

Another group 1 oxide grain 42Fb3 has a spinel-like (MgAl₂O₄) composition with an Al/Mg ratio of 1.6 ± 0.4. Abundant presolar spinel grains have been identified in acid residues from meteorites including Tieschitz, Bishunpur, and Semarkona (Nittler et al. 1994, 1997, 2008; Choi et al. 1998). Most of these spinels have O isotopic compositions that are consistent with theoretical values obtained from models of O-rich RG or AGB stars (e.g., Lugaro et al. 2007). Several authors (e.g., Choi et al. 1998; Zinner et al. 2005) have reported variable Al/Mg ratios of presolar spinel grains determined from NanoSIMS isotopic measurements, suggesting that these grains may be non-stoichiometric. However, the Auger measurement of grain 42Fb3 shows that it is, within errors, consistent with stoichiometric spinel. Transmission electron microscopy (TEM) analysis of another presolar spinel, UOC-S2, also indicates a stoichiometric composition (Zega et al. 2009). It remains unclear whether most presolar spinels tend to be non-stoichiometric or whether the variable ratios are due to SIMS matrix effects. Spinel grains form in the envelopes of AGB and RG stars as a result of a solid–gas interaction of Al₂O₃ with gaseous Mg (Lodders & Fegley 1999) and condensation of our spinel grain may have occurred via such a process. Finally, the O isotopic compositions of the remaining two oxide grains, 34C-6 and 42Fs6 are also compatible with an AGB source.

In this study, four silicate grains (Table 1) show ¹⁸O enrichments and normal Si isotopic compositions (within 2σ; Figure 2). Elemental analyses of the four grains indicate that two have Fe-rich end-member compositions, including one with a pyroxene-like stoichiometry. The remaining grains have Fe-rich ferromagnesian compositions, one with an olivine-like stoichiometry, while the other is non-stoichiometric and contains Al and Ca.

Several possible stellar sources have been suggested for group 4 grains: early dredge-up events in evolved stars (Nittler et al. 1997), high-metallicity stars (Nittler et al. 1997), and SN explosions (e.g., Choi et al. 1998; Nittler et al. 2008). In the first scenario, the production of ¹⁸O occurs via the ¹⁴N + α reaction during He-burning and the ¹⁸O-rich material is brought up to the stellar surface during third dredge-up events, which occurs before ¹⁸O is destroyed in later pulses. However, AGB models do not predict the occurrence of third dredge-up during the earliest thermal pulses, as is required to produce large surface enhancements in ¹⁸O (Nittler et al. 2008) and, therefore, our group 4 grains probably did not form by this process. The high ¹⁸O/¹⁶O ratios of our group 4 grains could also reflect the initial ratios of high metallicity parent stars. Galactic chemical evolution (GCE; Figure 1) predicts that the abundances of the heavier isotopes (e.g., ^17,18O relative to ¹⁶O) increase over time and, therefore, lead to higher metallicities (Timmes et al. 1995). Similarly, Si isotopes of grains that condense in higher metallicity stars should exhibit ^29,30Si enrichments relative to ²⁸Si and evolve along the SiC mainstream line (Figure 2). Therefore, grains condensing in high-metallicity stars should exhibit high ^17,18O/¹⁶O and ^29,30Si/²⁸Si isotopic ratios. If the very high ¹⁸O/¹⁶O ratios of our grains are due to GCE, this would imply a parent star with metallicity several times higher than solar. Aside from the fact that such stars are exceedingly rare and probably were even more so at the time of solar system formation, such stars are also expected to have high ¹⁷O/¹⁶O ratios, both because of GCE and the first dredge-up. Moreover, an enrichment in ^29,30Si would also be expected. However, none of the ¹⁸O-rich group 4 grains identified in this work show correlated excesses in ¹⁷O. In addition, all the grains exhibit normal Si isotopic ratios (within 2σ). Thus, the group 4 grains from this study probably did not originate in high-metallicity stars.

Finally, an origin in an SN environment has been used to explain the compositions of several group 4 grains (e.g., Choi et al. 1998; Messenger et al. 2005). Grains that condense from the debris of a core-collapse SN may contain material from different zones of a pre-SN star. Recently, Nittler (2007) showed that it was possible to reproduce the compositions of some group 4 grains by mixing material from the H-envelope with contributions from the He/C zone, which is ¹⁸O-rich, and the inner O-rich zones (O/Si, O/Ne, and O/C). We attempted to reproduce the O isotopic compositions of the group 4 grains found in this study by using this mixing model to see how this would affect the Si isotopes of the same grains. A comparison with the observed compositions could provide additional constraints about these grains' provenance. The O and Si isotopic abundances in the zones of a 15 M_☉ pre-SN star (Rauscher et al. 2002), specifically model s15a28c, were used to model the compositions of the four group 4 grains identified in this work (Table 3). The O and Si isotopic abundances of the different zones are shown in Figure 7. These mixing calculations were also carried out for ten group 4 grains from Mostefaoui & Hoppe (2004), Bland et al. (2007), Yada et al. (2008), and Floss & Stadermann (2009). The average O isotopic ratios from each of the five zones (H-envelope, He/C, O/Si, O/Ne, and O/C) were mixed in different proportions to achieve the measured O isotopic compositions of the grains. An additional constraint for the condensation of O-rich dust grains, that the carbon content in the gas should be less than the oxygen content, i.e., C/O < 1, was also satisfied. All O isotopic compositions could be reproduced by mixing 92%–99% material from the H envelope and considerably less material from the O-rich zones and the He/C zone (Table 3). A small amount (1%–1.9%) of material from the ¹⁸O enriched He/C zone is sufficient to account for the observed ¹⁸O anomalies.

Zoom In Zoom Out Reset image size

We then assumed that the Si isotopic ratios were mixed in the same proportions as the O isotopes and calculated the predicted Si compositions of the grains. However, using the same mixing ratios results in predicted δ²⁹Si and δ³⁰Si values that range from +87‰ to +207‰ and +425‰ to +1035‰, respectively, significantly higher than those measured in the grains. This is a reflection of the fact that the O-rich zones are also Si-rich and are enriched in the heavy isotopes of Si (Figure 7). Thus, including even small amounts of material from the O-rich zones alters the ²⁹Si/²⁸Si and ³⁰Si/²⁸Si ratios of the resulting mixture dramatically. Clearly, this mixing model is unable to simultaneously produce the ¹⁸O enrichments and the solar Si isotopic ratios of the analyzed grains. Next we tested if mixing of material from the H envelope, the He/C zone, and only the O/C zone (which has the lowest Si abundance of the three O-rich zones) could explain the O and Si isotopic compositions of the 14 group 4 grains. In this case, the predicted Si isotopic ratios fall on the broad gray line in Figure 2, in qualitative agreement with almost all of the grain data, indicating that it may indeed be possible to obtain close-to-normal Si isotopic ratios with this SN mixing model.

Several group 4 silicate grains studied elsewhere (Messenger et al. 2005; Bland et al. 2007; Vollmer et al. 2008) were found to be ²⁹Si-depleted (Figure 2) and their compositions cannot be explained by either of the mixing models discussed here. One of these group 4 silicates shows an extreme ¹⁸O isotopic enrichment, accompanied by a large ¹⁷O depletion. Its formation was suggested to have occurred as a result of the mixing of O-rich material from the He/C and O/C zones with the Si-rich material from the Si/S and Ni zones (Messenger et al. 2005). Vollmer et al. (2008) also identified a few group 4 grains that are ²⁹Si-depleted. These authors argued for an SN origin for these grains by performing SN mixing calculations that required mixing material from six zones, including the He/N and inner ²⁸Si-rich, Si/S zones. While contributions from the inner ²⁸Si-rich Si/S zone are required to explain the Si isotopic compositions of the ²⁹Si-depleted group 4 grains, the compositions of the four grains from our study can be reproduced with contributions from only the several SN zones described above.

Three group 1 grains (42Fe14, 42Fs4_I, & 42Fs4_II) have Si-rich compositions. Two grains are ferromagnesian silicates while one grain contains no Mg at all (Table 2). The formation of Si-rich phases is possible under non-equilibrium conditions in low-metallicity stars with low Mg/Si ratios (Ferrarotti & Gail 2001). However, stars with low metallicities exhibit low ¹⁸O/¹⁶O ratios. The ¹⁸O/¹⁶O ratios of our grains are too high i.e., greater than (1.75 ± 0.03) × 10⁻³ to have originated in low-metallicity stars, even if we consider substantial dilution of the isotopic ratios (e.g., Nguyen et al. 2007). Moreover, it is questionable whether dust grains can form under such low metallicity conditions at all (Gail et al. 2009). Nagahara & Ozawa (2009) concluded that Si-rich phases may condense in kinetically fractionating systems where the condensation of Fe metal onto forsterite prevents the equilibration of Si-rich gas with the forsterite to produce enstatite, thereby allowing the formation of Si-rich grains.

Four ferromagnesian grains (34A-4, 34A-6, 42F-13, and 42Ff16) that exhibit Al and Ca Auger peaks in their spectra have Si-poor compositions given by Mg+Fe(+Ca)/Si ratios of ∼2.9–4.6. None of these grains resemble any stoichiometric Al- and/or Ca-bearing mineral phases. All the grains exhibit Fe-rich compositions with mg#s from ∼10 to 35. Another grain (42Fv113) is similar with respect to the high Al contents (∼23 at.%), but has a lower Mg+Fe(+Ca)/Si ratio. Abundant presolar Al₂O₃ grains have been identified in acid residues of meteorites (e.g., Nittler et al. 1994, 1997). Aluminum oxide is predicted to be the first condensate to form in an O-rich circumstellar environment (Lodders & Fegley 1999). As temperatures decrease, Al₂O₃ can act as a nucleation site for the subsequent condensation of ferromagnesian silicates (Gail & Sedlmayr 1999) or can react with the gas to form Al-bearing silicates. The "complex" silicate grains containing Al-rich subgrains (>200 nm) identified by Vollmer et al. (2009b) provide evidence for the former process. The Al-bearing silicate grains from our study may be products of the latter scenario, in which the influence of kinetics is significant, resulting in non-stoichiometric compositions. Grain 34A-4 is a group 4 grain and may have condensed under non-equilibrium conditions in SN ejecta, where a sharp drop in temperature resulting in the incorporation of Fe into the condensing grains may result in the formation of ferromagnesian silicate minerals (Gail & Sedlmayr 1999; Ferrarotti & Gail 2001; Gail 2003).

Both amorphous and crystalline silicates have been discovered around young main sequence stars (Waelkens et al. 1996; Malfait et al. 1998) as well as O-rich AGB stars (Waters et al. 1996; Demyk et al. 2000). Demyk et al. (2000) suggest that the broad stellar spectral features of amorphous silicates primarily indicate olivine-like compositions, with pyroxene-like compositions being relatively rare (<10%). On the other hand, enstatite appears to be much more abundant than forsterite among crystalline silicates (Molster et al. 2002). Grains with pyroxene-like compositions are twice as abundant as olivine-like grains in our inventory. The (Mg+Fe)/Si ratios of the majority of silicate grains identified by Vollmer et al. (2009b) in Acfer 094 scatter around a value of 1, which also indicates an over-abundance of pyroxene-like grains compared to olivine-like compositions. Although Auger spectroscopy does not provide structural information about the phases and, therefore, we do not know whether the grains are crystalline or amorphous, the higher proportion of pyroxene-like grains may suggest that some grains are crystalline (e.g., Molster et al. 2002). However, the Fe high contents of many of these grains (see Section 4.5) may indicate that they are amorphous, as astronomical observations suggest that crystalline silicates are Mg-rich (Demyk et al. 2000). Stoichiometric presolar silicates in the CR3 chondrites QUE 99177 and MET 00426 consist of approximately equal numbers of olivine-like and pyroxene-like grains (Floss & Stadermann 2009), which suggests that the apparent over-abundance of pyroxene-like grains in Acfer 094 may be simply a statistical effect. To date, only a small fraction of presolar silicates have been analyzed using TEM. A few of these silicate grains are crystalline circumstellar grains with olivine compositions (Messenger et al. 2003, 2005); Vollmer et al. (2007) also found an unusual presolar silicate in Acfer 094 with a perovskite structure. However, the majority of presolar silicate grains studied in the TEM are heterogeneous on the scale of ∼20–50 nm, have overall non-stoichiometric compositions, and amorphous structures (Nguyen et al. 2007; Vollmer et al. 2009a; Stroud et al. 2008, 2009). Therefore, it is possible that most of the silicate grains are indeed amorphous.

Shocks, irradiation, and sputtering processes in the ISM may alter the structure and composition of crystalline dust grains (e.g., Lodders & Amari 2005). For example, Demyk et al. (2001) showed that crystalline olivine can transform into amorphous pyroxene when irradiated by low-energy He⁺ ions. Such an irradiation process is probably responsible for the low abundance of crystalline silicates in the ISM and the transition from olivine grains around evolved stars to pyroxene grains around protostars (Demyk et al. 2001). The large number of grains with pyroxene-like compositions in our study may have resulted from such a process in the ISM. Amorphous silicate grains with stoichiometries in-between olivine and pyroxene may also be the result of irradiation and sputtering events (Jones 2000; Min et al. 2007). A large fraction of the silicates found in this study have elemental compositions intermediate between olivine and pyroxene (Figure 4). However, Min et al. (2007) predict the existence of Mg-rich grains (mg#s > 90) with stoichiometry between olivine and pyroxene in the ISM, which is in sharp contrast to the Fe-rich compositions of the grains in our inventory.

Figure 8 shows the distribution of mg#s for olivine- and pyroxene-like silicates, as well as for grains with compositions intermediate between these two stoichiometries. Silicates with high Fe/Mg ratios dominate our presolar grain inventory. The addition of the oxide grains or the "other" silicate grains does not change the distribution substantially. Relatively Fe-rich presolar silicate grains have been previously reported in Acfer 094 (Nguyen & Zinner 2004; Nguyen et al. 2007; Vollmer et al. 2009b), as well as in the CR3 chondrites, QUE 99177 and MET 00426 (Floss & Stadermann 2009), and the CO3 chondrite, ALHA77307 (Nguyen et al. 2008; Bose et al. 2009). The possible origin of these Fe-rich silicates is discussed in the following sections.

Zoom In Zoom Out Reset image size

4.5.1. Primary Processes: Condensation in Stars

Equilibrium condensation models predict that silicates with Mg-rich end-member compositions, such as forsterite and enstatite, are the first silicate minerals to condense at high (∼1000 K) temperatures in stellar envelopes (Lodders & Fegley 1999; Ferrarotti & Gail 2001; Gail 2003). Although earlier studies had suggested higher Fe contents in amorphous silicates (e.g., Molster et al. 2002; Demyk et al. 2000), recent models using realistic particle shapes suggest that amorphous silicate grains are more Mg-rich (mg# > 90) than previously assumed (Min et al. 2007). In addition, the best fit to the scattered emission spectra from interstellar dust requires an Mg:Fe abundance of ∼2–3.6 (Costantini et al. 2005). A viable primary mechanism that may produce Fe-rich silicates is non-equilibrium condensation of grains occurring in the outflows of stars. The dusty gas in stars is lost rapidly by radiation pressure, resulting in a sharp drop in temperature in which equilibrium may not be maintained, allowing a considerable fraction of Fe to be incorporated into condensing silicate grains (Gail & Sedlmayr 1999; Ferrarotti & Gail 2001; Gail 2003). In such a non-equilibrium environment, a large fraction of Mg may condense into other solid phases (e.g., Ferrarotti & Gail 2001). Evidence for the incorporation of Fe into silicates in circumstellar environments has also been obtained from laboratory experiments carried out to understand non-equilibrium condensation of silicates (e.g., Rietmeijer et al. 1999; Nuth et al. 2000). These laboratory experiments show that condensates with ferrosilica and magnesiosilica compositions form directly from the vapor, but that grains with FeMgSiO compositions are not among the primary condensates (Rietmeijer et al. 1999). Thus, silicates with Fe-rich end-member compositions, such as those found in this study, may have formed under similar circumstances. Rietmeijer et al. (1999) suggest further that dust aggregates form by accretion of simple end-member silicates, which then undergo energetic reactions and result in the formation of grains with ferromagnesian silicate compositions. High-resolution SE images of the silicate grains from our study (Figure 9) show a predominance of silicate grains with single complete grain boundaries with uniform O isotopic and elemental compositions, unlike the aggregates expected from such a process. However, the physical separation procedure carried out to produce the grain size separates could have resulted in the preferential survival of larger and more compact grains. Aggregates such as those postulated by Rietmeijer et al. (1999) have been observed in TEM studies, which show that many presolar silicates exhibit composite structures with heterogeneous elemental distributions (e.g., Stroud et al. 2009; Vollmer et al. 2009a).

Zoom In Zoom Out Reset image size

One consequence of non-equilibrium condensation is the formation of olivine as the dominant mineral species rather than pyroxene (Ferrarotti & Gail 2001). The apparent conflict with the observation of more pyroxene-like compositions among our grains may be due to possible transformation in the ISM, as discussed above.

4.5.2. Secondary Processes

4.5.3. Sources of Fe Enrichments

This study identified 44 presolar silicates and four presolar oxides (silicate/oxide ratio ∼11⁺¹¹₋₅ with 1σ errors; Gehrels 1986) in Acfer 094. The silicate/oxide ratios for the O-anomalous grains identified in other studies of the same meteorite range from 1⁺¹₋₅ to 7⁺²₋₁ (Nguyen et al. 2007; Vollmer et al. 2009b); the latter ratio is within 1σ of the silicate/oxide ratio obtained in our work. The rather low silicate/oxide ratio obtained by Nguyen et al. (2007) may be attributed to using the NanoSIMS to ascertain whether an O-anomalous grain is a silicate or an oxide grain. Phase identification on the basis of NanoSIMS secondary ion yields alone may lead to misidentifications and subsequent errors in the silicate/oxide ratios. In contrast to the low silicate/oxide ratios in Acfer 094, a total of 53 silicates and only one oxide grain were identified in the matrices of two CR chondrites: QUE 99177 and MET 00426 (Floss & Stadermann 2009). In addition, recent Auger measurements of O-anomalous grains in the CO3 chondrite ALHA77307 show high silicate/oxide ratios from 31⁺⁵⁸₋₂₅ (prelim. data from our lab.) to 35⁺⁶⁷₋₂₉ (A. N. Nguyen 2009, private communication). Such comparisons of the presolar silicate/oxide ratios in various meteorite classes can provide a gauge for the effects of secondary processing on presolar grain survival.

The variable silicate/oxide ratios observed in the primitive meteorites studied to date can be the result of a variety of factors including local heterogeneity in the ISM, secondary alteration processes or insufficient statistics. However, it is generally thought that silicate grains are more susceptible to destruction or alteration than oxide grains (e.g., Nguyen et al. 2007; Floss & Stadermann 2009). If this is the case, and if the high ratios observed in ALHA77307 and the CR chondrites MET 00426 and QUE 99177 are indeed representative of these meteorites, then these ratios may represent the initial proportion of presolar silicate and oxide grains in the protosolar molecular cloud from which the solar system evolved. This would further imply that the lower silicate/oxide ratios in Acfer 094 are the result of secondary processing and that this meteorite is, therefore, not as primitive as previously thought. This argument is supported by the low silicate/oxide ratio (3⁺²₋₃) that was found from measurements of the CR chondrite NWA 530 (Leitner et al. 2009). These authors suggested that the parent body alteration experienced by NWA 530 may have destroyed a significant fraction of the silicate grains. Similarly, the low abundance of presolar silicates in the CM2 chondrite, Murchison, is probably due to aqueous alteration experienced by this meteorite (Nagashima et al. 2005). As noted above, however, the Acfer 094 matrix is almost devoid of secondary phases that form as a result of aqueous alteration (e.g., Greshake 1997). Aqueous alteration may therefore not have played a significant role in destroying silicates in this meteorite. Similarly, the low degree of heating experienced by Acfer 094 suggests that thermal metamorphism probably did not significantly affect the silicate/oxide ratios either. On the other hand, Acfer 094 does seem to have experienced some weathering on Earth leading to the presence of abundant ferric Fe in the matrix (Bland et al. 2008) and the alteration of the IOM in Acfer 094 (Alexander et al. 2007). Thus, terrestrial weathering, resulting in the destruction of silicates, could account for the low silicate/oxide ratios in this meteorite. This suggests that the silicate grain abundance calculated for Acfer 094 may be a lower limit.