Trace element depth profiles in presolar silicon carbide grains

Authors


Corresponding authors. E-mail: ashley.king@manchester.ac.uk; ian.lyon@manchester.ac.uk

Abstract

Abstract– We have analyzed eleven presolar SiC grains from the Murchison meteorite using time-of-flight secondary ion mass spectrometry. The Si isotopic compositions of the grains indicate that they are probably of an AGB star origin. The average abundances of Mg, Fe, Ca, Al, Ti, and V are strongly influenced by their condensation behavior into SiC in circumstellar environments. Depth profiles of Li, B, Mg, Al, K, Ca, Ti, V, Cr, and Fe in the SiC grains show that trace elements are not always homogenously distributed. In approximately half of the SiC grains studied here, the trace element distributions can be explained by condensation processes around the grains’ parent stars. These grains appear to have experienced only minimal processing before their arrival in the presolar molecular cloud, possibly due to short residence times in the interstellar medium. The remaining SiC grains contained elevated abundances of several elements within their outer 200 nm, which is attributed to the implantation of energetic ions accelerated by shockwaves in the interstellar medium. These grains may have spent a longer period of time in this region, hence increasing the probability of them passing through a shockfront. Distinct groups of presolar SiC grains whose residence times in the interstellar medium differ are consistent with previous findings based on noble gas studies, although some grains may also have been shielded from secondary alteration by protective outer mantles.

Introduction

Presolar grains are dust that condensed in the environments of dying stars. The grains traveled through the interstellar medium (ISM) survived the formation of the solar system and are present within primitive meteorites. Nanodiamonds were the first type of presolar dust isolated from meteorites (Lewis et al. 1987). This was soon followed by the extraction of presolar SiC (Bernatowicz et al. 1987) and graphite (Amari et al. 1990) grains. Other presolar phases discovered since include oxides (Al2O3, spinel, and hibonite; Huss et al. 1993; Nittler et al. 1994, 1997; Choi et al. 1999), silicon nitride (Nittler et al. 1995), and silicates (Messenger et al. 2003; Nagashima et al. 2004; Nguyen and Zinner 2004).

Presolar SiC is the most extensively studied presolar phase, with an abundance in the Murchison (CM2) meteorite of approximately 6 ppm. Over 99% of SiC within the Murchison meteorite is presolar in origin (Amari et al. 1994), with isotopic anomalies in C, N, and Si defining several distinct populations. Most grains (>90%) are “mainstream” and condensed in stellar envelopes around asymptotic giant branch (AGB) stars of 1–3M (Zinner et al. 1989; Alexander 1993; Hoppe et al. 1994; Lugaro et al. 2003). The A and B grains (approximately 4%) are thought to have come from J-type carbon stars or born-again AGB stars (Amari et al. 2001a), while X grains (approximately 1%) are from supernovae (Amari et al. 1992; Hoppe et al. 1996, 2000; Nittler et al. 1996) and Y (Amari et al. 2001b) and Z grains (Hoppe et al. 1997) (approximately 1% each) are from low metallicity AGB stars.

Presolar SiC grains tend to incorporate and retain higher trace element abundances than other presolar phases, making many of them suitable for studies of trace elements combined with major isotope compositions (e.g., Amari et al. 1995). Despite this, there have been few quantitative studies of trace element abundances and distributions within presolar SiC grains (Amari et al. 1995; Huss et al. 1997; Hoppe et al. 2000, 2010; Kashiv et al. 2001, 2002, 2010; Henkel et al. 2007a; Lyon et al. 2007; Knight et al. 2008; King et al. 2011). As stellar condensates, the grains must have experienced conditions in circumstellar envelopes, the ISM, and the early solar nebula before arriving in the laboratory. Each environment may have influenced the abundance and distribution of the trace elements contained within the grains.

Lodders and Fegley (1995) modeled the condensation of SiC and other trace element compounds at C/O ratios of 1.05 and pressures in the range of 10−4–10−5 bar, conditions typical of the gaseous circumstellar envelopes of C-rich stars. Silicon carbide first begins to condense in this region at temperatures of approximately 1600 K. Initially, little SiC forms, but as the temperature decreases, eventually approximately 55% of the Si in the source gas can condense as SiC. The abundances of trace elements within the SiC are then directly influenced by both the composition of the gas and condensation processes in this environment. The elements Al, Mg, Ca, and Fe form compound species (AlN, MgS, CaS, and Fe3C) that are compatible within the lattice structure of SiC (Lodders and Fegley 1995). As these species are more volatile than SiC, with condensation temperatures in the range 1150–1250 K, the abundances of Al, Mg, Ca, and Fe are depleted within presolar SiC grains relative to chondritic (CI) values. This is supported by the data of Amari et al. (1995), who measured significant depletions of Al, Mg, Ca, and Fe (relative to CI) in presolar SiC grains. Additionally, the fact that these species condense in solid solution with SiC probably leads to homogenous distributions of Al, Mg, Ca, and Fe within presolar SiC grains. However, SiC can condense over a wide range of temperatures, and the composition of the source gas alters as its temperature drops from 1600 K to approximately 1200 K due to the condensation behavior of the elements within it. This could produce not only variations in trace element abundances between grains from the same parent star but also elemental zoning within individual grains.

Trace element compositions and distributions in presolar SiC can also be affected by compound species formed as separate phases. Titanium carbide is expected to condense prior to SiC at temperatures of approximately 1700 K. This could result in the presence of TiC within presolar SiC, and several such subgrains have been reported (Bernatowicz et al. 1992; Stroud and Bernatowicz 2005; Hynes et al. 2010a). On the basis of crystallographic alignment of a TiC within a presolar SiC grain, Bernatowicz et al. (1992) suggested that the TiC formed either through exsolution from the SiC or co-condensation of the two phases. Several subgrains of CaS have been reported within one SiC grain of type AB (Hynes et al. 2010b), while MgS and CaS have also been calculated to condense as coatings on the surfaces of presolar SiC grains (Zhukovska and Gail 2008). The presence of MgS coatings on circumstellar grains has been used to explain the 30 μm emission feature observed around some stars (Forrest et al. 1981; Goebel and Moseley 1985).

In the ISM, the grains are subjected to grain–grain or grain–gas collisions, with the effects of this sputtering expected to be displayed upon their surfaces (Jones et al. 1996). Bernatowicz et al. (2003) studied 81 pristine presolar SiC grains and found no evidence of any cratering on the grain surfaces. They did report that approximately 60% of the grains had amorphous, potentially organic coatings, indicating that grain surfaces may have been shielded from sputtering in the ISM. It has been suggested that grains could have been protected by ices accreting onto their surfaces, which were subsequently broken down into organic mantles by exposure to UV and cosmic-ray exposure (Sandford and Allamandola 1993; Gibb and Whittet 2002). Alternatively, if the residence time of presolar SiC grains in the ISM was very short, they may not have experienced significant sputtering. On the basis of Xe isotope measurements of presolar SiC separates, Ott et al. (2005) calculated cosmic ray exposure ages for SiC grains of <100 million years (Ma). Measurements of He and Ne in large (>5 μm) individual presolar SiC grains found most to have cosmic ray exposure ages of 3–400 Ma, although a few grains gave ages of approximately 1000 Ma (Heck et al. 2009). Similarly, Li isotopic compositions of large presolar SiC grains indicate that the grains were in the ISM for 40–1000 Ma (Gyngard et al. 2009). In general, cosmic ray exposure ages for most presolar SiC grains are shorter than the estimated lifetimes of dust in the ISM of >500 Ma (Jones et al. 1996).

There is evidence that some presolar SiC grains did experience ion implantation in the ISM. Lyon et al. (2007) presented Li and B isotopic compositions, and Li/Si and B/Si abundance ratios in eleven acid-extracted and ten pristine (isolated using the gentle separation procedure of Tizard et al. [2005]) presolar SiC grains. The 7Li/6Li and 11B/10B ratios were close to the solar values of approximately 12 and 4, respectively, while the Li/Si and B/Si ratios in some grains were found to be elevated at, or near, the surface, to depths of approximately 200–300 nm. This was interpreted as evidence for the implantation into the grains of Li and B ions accelerated by supernova shockwaves in the ISM. It has also been suggested that some noble gases may have been implanted into SiC grains, although this has been attributed to stellar winds associated with AGB stars (Lewis et al. 1994; Verchovsky et al. 2004).

Trace element abundances in presolar SiC grains may also be affected by contamination in the laboratory. This could arise from the meteorite matrix or through the acid extraction procedure. Henkel et al. (2007a) reported the apparent deposition of meteoritic matrix material in crystal defects etched into grain surfaces by the harsh acids used to isolate the grains (Amari et al. 1994). Knight et al. (2008) measured abundances of trace elements, including Ti, V, Cr, Mn, Fe, Pb, and W, in SiC grains isolated using both standard procedures (Amari et al. 1994), and with additional clean acid treatments. They found that using the extra clean acid steps led to a decrease in contamination of W and Pb on grain surfaces.

Most previous studies of trace elements in presolar SiC have used direct current (DC) beam ion probes (Zinner et al. 1989; Alexander 1993; Hoppe et al. 1994; Amari et al. 2000). Sample consumption during measurements with early ion probes (e.g., Amari et al. 1995) was often high, and data represented an average over a large volume of a grain, although for more recent instruments, such as the NanoSIMS, this has become less of an issue. Ion probes use magnetic and electric fields to separate secondary ions and often only 5–7 isotopic species can be measured in a single analysis. Generally, C, N, and Si isotopic compositions are determined first with measurements that require some grain sputtering before trace element analyses. Attempts have therefore been made to determine trace element abundances in individual SiC grains using synchrotron x-ray fluorescence (SXRF) (Kashiv et al. 2001, 2002, 2010; Knight et al. 2008; King et al. 2011). This method has the advantage of being nondestructive, so other analytical techniques may be later applied to the same grains. It also does not suffer from the matrix effects associated with secondary ion mass spectrometry (SIMS), and several s-process elements, such as Rb, Sr, Y, and Zr, have been successfully measured using SXRF. The disadvantages of this technique include measured concentrations representing the bulk grain and beam diameters of approximately 2.5 μm.

In this study, we have used time-of-flight secondary ion mass spectrometry (TOFSIMS). TOFSIMS allows for comprehensive studies of samples, with a complete mass spectrum for either positive or negative secondary ions acquired in a single analysis. Isotopic ratios and trace element abundances can be obtained at the same time providing a more efficient sample usage. High spatial resolutions of approximately 300 nm can be achieved, although with a loss of signal, and detection limits can be within the ppb-range (Stephan 2001). TOFSIMS uses a pulsed primary ion beam with a low duty cycle and a high-transmission TOF mass spectrometer, so sample consumption during a single measurement is low. This makes it suitable for high-resolution depth profiling of individual presolar grains. The effects of heterogeneities, for example, subgrains, on trace element abundances may potentially be established.

We have undertaken a series of systematic TOFSIMS analyses to measure trace element abundances and distributions in presolar SiC grains. The resulting depth profiles can help constrain the effects of any processing experienced by the grains during their transit from parent stars to the laboratory. Here, we present data for eleven acid-extracted presolar SiC grains. For each grain, near-complete depth profiles for a range of trace elements have been obtained.

Experimental Procedure

Samples

Silicon carbide grains from two residues of the Murchison (CM2) meteorite were analyzed. Four of the grains came from the KJG acid residue prepared using the “Chicago procedure” (Amari et al. 1994). These grains were received as a deposit spread upon an Au-foil. Grains from this sample were transferred onto a new, cleaned, high-purity (>99.999%) Au-foil, upon which a copper finder-grid (Agar H15, 3.05 mm, 125 mesh) had been imprinted, by pressing the foils together. These grains are referred to as AK-KJG. The other seven analyzed grains were extracted by John Arden using the procedure of Amari et al. (1994). Murchison matrix material was treated with HF/HCl to produce a sample named “MM.” Some of the “MM” sample was then oxidized using Cr2O7 before treatment with HClO4 to give a new sample, “MM1.” Kerogen and sulfur were removed to leave a SiC and spinel-rich residue named “MM2.” A small aliquot of the MM2 residue, suspended in a 1:1 isopropanol/water mixture, was distributed upon a second high-purity gold foil with imprinted finder-grid. These grains are referred to as AK-MM2.

Silicon carbide grains on both foils were located using a Phillips XL30 environmental scanning electron microscope (ESEM). Energy dispersive X-ray (EDX) analysis was used to confirm the composition of grains as SiC. Electron beam energies were kept low (10–15 kV) to minimize potential damage to the grain surfaces. The grains were imaged using the ESEM and their locations relative to the finder-grid were recorded. The average major-axis diameter of the AK-KJG grains was 1.60 μm. The AK-MM2 grains were slightly larger at 2.01 μm. Grains were named according to their sample name, grid location (letters A–Z, numbers 0–9) and numbered sequence within each grid area.

Analytical Procedure

The SiC grains were analyzed with high mass resolution and a spatial resolution of approximately 500 nm in two separate rounds of measurements using our “IDLE” (Interstellar Dust Laser Explorer) TOFSIMS instruments. These instruments are based upon the designs of Braun et al. (1998) and previous versions have been described elsewhere by Henkel et al. (2006, 2007b). The AK-KJG grains were analyzed using “IDLE2,” which at the time was equipped with a 25 kV Ga+ liquid metal ion gun (LMIG) (IOG25 from Ionoptika Ltd, Southampton, UK). We then built a new instrument, “IDLE3,” equipped with a 25 kV Aun+ LMIG (IOG 25Au from Ionoptika Ltd, Southampton, UK; Davies et al. 2003; Hill and Blenkinsopp 2004). The AK-MM2 grains were analyzed with Au+ primary ions using the IDLE3 instrument. The analytical procedure used was the same for both sets of grains.

Grains were analyzed by rastering a pulsed primary ion beam over their surfaces. Positive secondary ions were detected and a single analysis often lasted several hours to achieve good counting statistics for low-abundance elements within the grains. The primary ion beam was applied over areas larger than the grain sizes (typically 7 × 7 μm, although this was adjusted according to grain size), as during long measurements, we occasionally observed shifts of up to a few micrometers in the position of the beam relative to the grain, probably due to temperature changes or electronic instabilities. Data were acquired as a series of scans covering the rastered area, usually 64 × 64 pixels with 150 primary ion shots per pixel, and each pixel containing a complete mass spectrum (1–300 amu) for that scan.

Raw data were collected for offline analysis. The region of interest was defined using the 28Si+ secondary ion image to mark the grain area and produce a total spectrum for the grain. As only spectra from the region of interest were selected for further analysis, this ensured that the background signal from the Au-foil contributing to the spectrum was minimal. As long as the grain remained within the field of view over the entire analysis period, any lateral movements in the grain position within that field of view could be compensated during postacquisition analysis.

Mass Resolution

Accurate determination of isotopic ratios and quantification of trace element abundances within presolar SiC grains required that high mass resolution (e.g., m/δm approximately 3500 [FWHM] at 28Si) was achieved, to enable hydride and oxide interferences to be resolved throughout the spectrum. In TOFSIMS, high mass resolutions can be obtained by using very short primary ion pulses (approximately 2–3 ns) (Stephan 2001). However, using such short pulses results in low secondary ion signals and may degrade the spatial resolution of analyses due to the rapid switching (i.e., blanking and unblanking) of the primary ion beam. Instead, we used a delayed secondary ion extraction technique to achieve high mass resolutions.

During delayed extraction, primary ion pulses approximately 40 ns long hit the sample while it was at ground potential. Secondary ions formed earlier in the pulse traveled further from the sample than those formed toward the end. The sample potential was then switched to 1.4 kV to accelerate the secondary ions toward the extractor, with those secondary ions formed earliest accelerated the least due to their greater distance from the sample surface. The secondary ions were then time-focused at the detector using a two-stage reflectron. The major advantage of delayed extraction was that high mass resolutions (m/δm of 3000–4000) could be achieved despite using long primary ion pulses. The long primary ion pulse lengths boosted secondary ion signals and, as the pulses were long compared with the switching time, also helped to maintain submicron spatial resolution of the focused ion beam.

Even at high mass resolutions, the Si-isotope analysis was affected by a significant interference on the 29Si-peak from 28SiH+. We therefore used the peak deconvolution technique described by Stephan (2001) to determine the Si isotope ratios. The same technique was also used to derive δ26Mg values. This method involves fitting the peak shape of the major isotope (i.e., 28Si) to that of the minor isotope. The isotope ratio is given by the scaling factor between the two peaks. The presence of the 28SiH+ interference has previously been shown to add 20–30‰ uncertainty to the δ29Si value (Lyon et al. 2007). Isotopic ratios were corrected for instrumental mass fractionation using an average of multiple analyses of silicate standards made by Henkel et al. (2007b).

Element Quantification

In SIMS, measured elemental abundances can vary when analyzed in samples of different mineralogy due to matrix effects (e.g., Benninghoven et al. 1987). To quantify elemental abundances in presolar SiC grains, it was therefore necessary to calculate relative sensitivity factors (RSFs) from standards of similar composition. We have previously calculated RSFs for Ga+ (Henkel et al. 2007b) and Au+ (King et al. 2010) primary ions using a series of silicate glasses (MPI-DING standards [Jochum et al. 2000, 2006]) and NIST reference material SRM 610, which have a wide range of known trace element concentrations. For this study, the SiC standard SRM 112b, for which only abundances of Al, Ca, and Fe are quantified, was also measured for comparison. Figure 1 shows that RSFs measured in SiC with Au+ primary ions were elevated (on average, by a factor of approximately 7) relative to their respective values in the silicate glasses. This indicates that RSFs from the silicate glasses need correcting for quantification of elemental abundances in presolar SiC.

Figure 1.

 Comparison of RSFs (relative to Si) obtained from analysis of several silicate glasses (data from King et al. 2010) and SiC standard SRM 112b using Au+ primary ions. The RSFs for Al (5.99eV), Ca (6.11eV), and Fe (7.90eV) in SiC are on average a factor of 7 higher than those in silicate glass.

It has been shown that sputtering of samples of different mineralogy can affect the relative abundance of cluster secondary ion species in the mass spectrum (e.g., Lyon et al. 2010). During sputtering of silicate glass with Au+ primary ions, we observed O cluster ion species (e.g., SiO+, SiO2+, Si2O+, Si2O2+, etc.), while for SiC, we instead saw C cluster species (e.g., SiC+, SiC2+, Si2C+, Si2C2+, etc.). In both cases, Si ions tended to partition into cluster species, resulting in a decrease in the practical secondary ion yield (number of secondary ions per primary ion) for Si+. The yield for Si+ in SiC was determined to be 3.1 (±0.7) × 10−5, a factor of 8.9 (±1.9) lower than the yield for Si+ in silicate glasses of 27.6 (±1.5) × 10−5 measured by King et al. (2010) (see Table 1). As the bond between Si and C is stronger than that for Si and O (Krantzman et al. 2007, 2008), it is anticipated that the partitioning of Si ions into Si-containing clusters is less severe during analyses of silicate materials than SiC. We therefore attribute lower yields for Si+ to the increased formation of SimCn+ clusters when analyzing SiC.

Table 1.   Practical secondary ion yields (number of secondary ions/number of primary ions) for Si, Al, Ca, and Fe obtained by analysis of silicate glasses (data from King et al. 2010) and SiC standard SRM 112b with Au+ primary ions. Practical yields have been normalized by the element’s atomic abundance in the standards. Errors are 1σ.
SampleSilicon (×10−5)Aluminum (×10−5)Calcium (×10−5)Iron (×10−5)
Silicate27.6 ± 1.5129 ± 7154 ± 845.4 ± 2.6
SiC3.1 ± 0.7151 ± 26101 ± 1838.1 ± 5.9
Silicate/SiC8.9 ± 1.90.86 ± 0.151.53 ± 0.291.19 ± 0.20

The RSFs in silicate glasses given by Henkel et al. (2007b) (for Ga+ primary ions) and King et al. (2010) (for Au+ primary ions) are calculated from the positive ion yield ratios of the trace elements relative to Si+. As a factor of 8.9 more Si ions are detected during analysis of silicates than SiC, the resulting RSFs for all other elements (which are relative to Si) must be lower by an equal amount if there is no change in the other elements’ ion yields. This assumption seems reasonable as we measured no significant difference in the yields for Al, Ca, and Fe between SiC and silicate glasses (Table 1). Elemental abundances relative to Si measured within our SiC grains were therefore quantified according to Equation 1 using the well-characterized RSFs for the analyzed elements (Li, B, Mg, Al, K, Ca, Ti, V, Cr, and Fe) in silicate glass, and then correcting the calculated abundances for the change in Si yield by a factor of 8.9. The errors on the RSFs used for quantification were typically approximately 30%.

image(1)

where E is the element of interest; EREF, the reference element (Si); I, the measured secondary ion intensity; and RSF, the known sensitivity factor for the element of interest in silicate glass.

Depth-Profiling Procedure

Sputtering during SIMS measurements is a destructive process that removes atoms from the surface of a sample. The longer a grain is sputtered, the more material is removed. Subsequent measurements therefore sample material from a greater depth within that grain and a depth profile can be produced. Each grain was initially sputtered using a DC-beam rastered for approximately 1–2 min over a 50–100 μm2 area containing the grain. This was sufficient to both remove hydrocarbons, which may have become deposited on the sample during handling or in the ESEM/TOFSIMS instruments, and to start sputtering into the grain to ensure that sputter equilibrium was reached. The initial “cleaning” sputtered an estimated 10–30 nm from the surface of the grains and was then followed by TOFSIMS measurements using a pulsed primary ion beam. The low duty cycle of the pulsed beam produced a lower sputter rate than that of the DC-beam. However, measurements with the pulsed beam often lasted several hours and a significant amount of material was still sputtered from a grain.

After each measurement, a grain was again sputtered with a DC-beam before the next measurement was started. By carefully recording the primary ion beam current, length of time the beam was applied, field-of-view for each DC-clean and measurement, and using known sputter rates (King et al. 2010), it was possible to estimate the amount of material sputtered from a grain during a combined cleaning and measurement step. Elemental abundance data were then plotted on the depth profile at the mid-point of the total amount of material sputtered in each step. For example, if during the first step, the DC-beam removed 10 nm of material and the measurement a further 90 nm, the abundance data were plotted at a depth of 50 nm. If the next step was identical to the first, the abundance data were plotted at a depth of 150 nm (with a total of 200 nm having been sputtered from the grain). Depths given for our trace element profiles are therefore only estimates, with the error on the known sputter rate probably around 20% (King et al. 2010) and geometrical factors expected to increase this error further (see the following section).

The cleaning and measurement steps were repeated until approximately half of a grain had been sputtered away. At this point, some grains were removed from the TOFSIMS, re-imaged using the ESEM, and then returned to the TOFSIMS for analysis until they had been completely sputtered away. This enabled us to monitor the accuracy of the depths, by comparing the amount of a grain remaining after several cleaning and measurement steps with the images of the grains collected before any analyses were performed.

Ideally, each step would have sputtered a comparable amount of material to produce uniform depth profiles through each grain. Unfortunately, during several measurements, instabilities in the electronics of the LMIGs caused the primary ion beam to “drop-out,” i.e., no primary ion current was hitting the sample. This resulted in measurements of varying length and therefore uneven amounts of sputtered material. Occasionally, there were also difficulties in relocating grains after re-imaging in the ESEM, which led to extra sputtering with the DC-beam prior to beginning some measurements.

We tested the depth-profiling procedure (repeated steps of DC-beam sputtering and measurements with a pulsed beam) by first applying it to two silicate grains (named A and B) separated from Murchison matrix material. Figure 2 shows that the elemental abundances measured in the grains remained relatively constant across the profiles. The relative standard deviations of the measured abundances with depth ranged between 2% and 45%, with an average of approximately 25%. We also calculated the difference between the maximum and minimum abundance measured for each element in the grains. The largest difference was a factor of 3.2 for K in grain A, while the average for all elements in both grains was only a factor of 2.2. At a depth of approximately 300 nm, silicate grain A was removed from the TOFSIMS and re-imaged using the ESEM. The abundances for all of the elements in the measurements immediately after this were similar to those before removal. This indicated that, in this instance, re-imaging using the ESEM and extra sample handling apparently resulted in no significant contamination or alteration of the grain surface.

Figure 2.

 Trace element depth profiles (not corrected for RSFs) for two silicate grains (A and B) from Murchison matrix material. The grains were depth-profiled using the procedure outlined in the text. As the grains were sputtered away, trace element abundances displayed little variation with depth. Errors for the element/Si ratios were on the order of a few percent and for depths were tens of percent. The dashed vertical line in the upper left panel indicates where grain A was removed from the TOFSIMS and re-imaged using the ESEM.

Limitations of Depth Profiling

Correlating the measured ion abundances as a function of time with the real elemental abundance depth profile in a sample is only a simple relationship for the case where there is a flat surface sputtered by ions impacting normally to the surface. Rough samples lead to the detection of secondary ions from different depths in the sample during the same measurement, while three-dimensional shapes cause the primary ion beam to hit samples at varying angles of incidence, resulting in differential sputtering of the surface (Rost et al. 1999). Preferential sputtering of any subgrains, known to be present in some SiC grains (e.g., Stroud and Bernatowicz 2005; Hynes et al. 2010a), would further degrade the depth resolution. The depths given for our trace element profiles are therefore only estimates checked, where possible, by repeated imaging of the sputtered grains in the ESEM.

The apparent distribution of trace elements shown in our depth profiles may also have been affected by primary ion beam mixing of layers with different elemental compositions. The Au+ primary ions used in this study have energies of 25 kV when impacting into a grain surface and they are implanted to depths of tens of nanometers. This creates collisions between the primary ions and atoms in the grain, which subsequently become transported to greater depths. Layers in a grain containing different elemental abundances could therefore have become mixed to some extent.

The finite extent of the primary ion beam focus will also cause some mixing between layers during sputtering in the lateral direction, if such layers of different elemental abundances in a grain exist. For example, if an element is enriched in the outer part of a grain relative to the core and this outer layer is thinner than the spatial resolution of the ion beam, then the core and rim cannot be fully resolved from each other when part of the core is exposed by sputtering. In an attempt to account for this scenario, we have produced a simple geometrical model (described in Appendix S1) that assumes depth profiling through a spherical grain of 1 μm diameter with an outer region, the thickness of which can be varied, enriched in any chosen element. The abundance of the element is taken as 1 in the enriched outer region and 0 in the grain core. In an ideal situation, when sputtering through this grain, the depth profile should show a clear transition between the two layers of differing composition.

In reality, mixing of atoms due to sputtering effects and an inability to laterally resolve the rim from the core prevent this, and the measured abundance in the core is elevated due to a contribution from the enriched outer region. The model indicates that, assuming the spatial resolution of the primary ion beam is greater than the thickness of any enriched outer region, the variation measured in the depth profile will be determined by the thickness of the enriched outer region relative to the grain core. The thicker the outer region, the higher the abundances in the grain core will appear. It is important to note that this is a simple model that assumes a spherical grain, constant sputtering rate, and an extreme abundance gradient. It therefore does not take into account the other depth-profiling issues such as irregular grain shapes, preferential sputtering, crystal orientation, or heterogeneous inclusions, and provides only an indication as to the depth profiles we might observe.

Results

Isotopes

Figure 3 shows the measured δ29Si and δ30Si values were within the range −32‰ to +296‰. The δ26Mg values were in the range −81‰ to +92‰, with 1σ errors on average of approximately 50‰. Within the analytical uncertainty, none of the grains contained a significant 26Mg-excess that could be attributed to the decay of 26Al. Lithium and B isotopic compositions measured in the SiC grains are also provided in Table 2. No significant 7Li/6Li or 11B/10B isotopic anomalies were detected, with all ratios within 3σ of the solar values of approximately 12 and 4, respectively.

Figure 3.

 Three-isotope plot of the Si isotopic compositions (expressed as δ values, inline image) measured in presolar SiC grains as part of this study. Dashed gray lines indicate the solar composition. The Si isotopic compositions are compared with previous studies using both DC-SIMS (Hoppe et al. 1994) and TOFSIMS (Henkel et al. 2007a; Lyon et al. 2007). Mainstream SiC grains, as all of those in this study appear to be, are often characterized by enrichments in 29Si and 30Si relative to solar, and fall along the mainstream correlation line of slope 1.3 (dashed diagonal black line). Error bars are 1σ.

Table 2.   Elemental abundances and approximate depths sputtered to during TOFSIMS measurements of presolar SiC grains. Elemental abundances are given as atomic-% relative to Si. Errors on the RSFs used for quantification are approximately 30%. Errors for the isotopic compositions are 1σ and include counting statistics and corrections for mass fractionation. A gap indicates that no data were obtained, either due to unresolved interferences or lack of counts in the mass spectra. Values given in bold are the geometric mean abundances for each element within a grain. Values given in italics have been excluded from the interpretation of depth profiles (see the Discussion), although these data points are still included on Figures S3–S13 in Appendix S1.
Grain NameGrain Size (μm) 7Li/6Li 11B/10Bδ29Siδ30Siδ26MgApprox Depth (nm)Li/Si
(×10−2)
B/Si
(× 10−2)
Mg/Si
(× 10−2)
Al/Si
(× 10−2)
K/Si
(× 10−2)
Ca/Si
(× 10−2)
Ti/Si
(× 10−2)
V/Si
(× 10−2)
Cr/Si
(× 10−2)
Fe/Si
(× 10−2)
MM2-G-12.9 × 0.415.1 ± 1.64.5 ± 0.8296 ± 1970 ± 15−1 ± 21420.0240.00280.0850.410.440.340.0360.0110.0370.32
       1530.51 0.8649.463.262.933.690.671.131.23
       3980.410.0132.233.568.472.431.010.760.155.79
       6210.490.00880.738.4110.383.543.350.180.122.82
         0.22 0.0069 0.59 4.96 3.34 1.71 0.82 0.18 0.17 1.59
MM2-G-21.1 × 0.514.0 ± 1.24.6 ± 2.6228 ± 3362 ± 3213 ± 19610.0560.00401.813.251.521.250.400.0160.532.58
       1670.180.00540.160.940.460.330.160.00400.170.84
       2560.170.00210.331.540.500.230.120.00420.0780.60
       3430.310.0112.693.562.000.440.0930.0180.100.16
       4300.370.0132.763.880.611.270.110.0140.230.14
       5230.350.00993.982.691.321.350.360.0160.233.39
         0.20 0.0063 1.19 2.36 0.91 0.64 0.17 0.010 0.18 0.68
MM2-M-11.4 × 0.917.2 ± 1.9 185 ± 68−32 ± 534 ± 20310.390.0102.483.742.870.930.0670.0220.61 
       840.41 2.713.512.910.0940.0430.0250.39 
       1250.056 4.043.781.840.500.0740.0360.36 
       1790.00950.00463.091.851.070.660.048 0.19 
       2490.26 5.111.880.241.410.0400.0120.38 
       3500.29 1.882.701.632.330.140.0200.25 
         0.14 0.0069 3.05 2.78 1.36 0.68 0.062 0.022 0.34  
MM2-N-11.8 × 1.013.9 ± 0.82.4 ± 1.5207 ± 55146 ± 3524 ± 15620.410.0131.623.440.710.220.020 0.26 
       1790.420.00941.001.371.160.510.0330.00360.27 
       4110.210.0104.322.230.851.340.0540.00600.11 
       6590.280.00793.543.481.091.500.0860.0110.16 
       7850.150.00362.312.251.110.690.0990.00660.019 
       9200.27 4.143.200.331.120.0490.00280.027 
         0.27 0.0081 2.48 2.53 0.81 0.75 0.049 0.0053 0.093  
MM2-delta-12.0 × 1.412.7 ± 1.42.9 ± 1.9197 ± 6788 ± 53−81 ± 61140.00230.00372.650.0300.00980.00940.010 0.0230.58
       1900.45 0.473.123.460.880.160.0250.322.43
       4660.330.00180.0960.340.120.0670.033 0.0780.49
       7050.0370.00120.120.630.201.250.10 0.140.33
       9100.0500.00470.100.470.0820.840.041 0.0360.45
         0.057 0.0025 0.27 0.39 0.15 0.23 0.047 0.025 0.079 0.64
MM2-delta-22.4 × 1.613.5 ± 0.84.4 ± 1.675 ± 2639 ± 2615 ± 17300.150.00333.299.310.263.410.350.0110.223.49
       980.0970.00392.4317.130.655.460.880.0140.253.13
       1570.130.00384.1216.920.725.141.240.0150.381.53
       2060.028 3.1017.511.792.451.60 0.112.66
       2940.170.000613.9618.741.391.326.160.0180.0741.00
       4890.110.00210.141.000.980.570.0340.00460.241.58
       6770.160.00260.270.380.310.610.036 0.0440.51
       8010.230.000890.170.360.270.680.00900.00120.0240.26
       10790.340.00300.0471.080.840.400.350.00880.120.81
         0.13 0.0021 0.79 3.70 0.65 1.45 0.29 0.0079 0.12 1.25
MM2-5-13.9 × 0.613.5 ± 2.2 62 ± 3672 ± 2835 ± 201060.0560.00594.520.270.650.120.0240.0110.0850.57
       3710.11 0.488.831.530.250.0180.0440.232.02
       6680.087 0.1511.231.450.15 0.0710.0290.67
        826 0.12   0.31 3.77 1.07 9.14 0.10 0.29 0.16 1.15
        867 0.058 0.0022 0.15 0.85 0.20 0.97 0.090 0.030 0.050 0.14
         0.080 0.0059 0.70 3.00 1.13 0.17 0.021 0.033 0.083 0.92
KJG-D-11.3 × 1.110.2 ± 3.1 95 ± 3155 ± 29 210.0740.0131.561.5214.5659.340.650.0483.5916.85
       700.076 0.260.840.440.521.10 1.856.17
       1600.17 0.650.420.34   1.63 
       3000.060 0.370.350.150.150.88 0.342.52
       4600.056 0.410.370.23 1.83  2.84
       5870.0065 0.0142.110.300.420.35 0.341.68
       6700.00910.0042 3.950.210.260.540.0320.481.91
       7870.0081 0.00842.190.150.140.55 0.211.55
       870   1.680.0950.100.57 0.311.56
       10770.0030 0.0172.270.160.200.430.0380.361.70
         0.025 0.0074 0.13 1.17 0.32 0.44 0.68 0.039 0.62 2.78
KJG-D-22.0 × 1.513.7 ± 4.2 237 ± 164  510.240.221.120.440.832.10  5.10 
       1290.24 0.890.771.123.75  3.79 
       1450.041 0.120.966.622.080.028 0.63 
       2320.061 0.190.742.271.32    
       2940.070  0.642.301.20    
       3660.34 2.171.671.895.81  6.54 
         0.12 0.22 0.55 0.80 1.98 2.31 0.028   2.98  
KJG-M-11.6 × 1.517.2 ± 4.0 239 ± 22111 ± 1848 ± 161340.00420.00200.0101.090.0500.0490.330.00840.0220.60
       790.0039 0.0212.620.0340.0700.560.010 0.98
       1240.0020 0.0212.530.0330.0680.530.0130.190.95
       1520.0031 0.0192.480.0240.0550.640.0190.210.83
       2370.0016 0.00952.050.0240.0550.460.0200.210.85
       3450.0019 0.00821.710.0180.0580.550.0300.220.82
       4650.0011 0.0141.430.0160.0560.540.0330.220.86
       5280.0012 0.0131.260.0160.0740.530.0210.210.85
       5600.00094 0.0141.100.0190.0760.530.0260.210.86
       6100.00130.000460.0111.270.0200.0950.460.0230.190.85
       7550.0020 0.0121.370.0160.0570.340.0130.230.87
       8260.0017 0.0121.680.0190.0730.520.0400.230.92
       9120.0045 0.0171.780.0160.0750.540.0200.260.91
       9470.0027  1.490.0200.0830.490.0490.180.62
       10010.00300.00160.0172.260.0240.0790.370.0270.220.79
         0.0021 0.0011 0.014 1.67 0.022 0.067 0.49 0.021 0.18 0.83
KJG-6-11.5 × 1.113.5 ± 1.3 134 ± 54159 ± 5192 ± 48220.0056 0.237.018.431.060.620.0170.51 
       920.0420.00140.925.851.254.350.450.00711.20 
       2240.067 1.059.011.485.620.54 1.58 
       3970.084 0.979.301.995.181.000.0241.46 
       5700.16 0.9710.111.695.060.610.0241.17 
        750 0.0086   0.044 2.98 0.19 0.15 0.17   0.068  
       8950.18 1.4510.203.027.010.88 1.80 
       11830.100.00340.249.170.832.210.560.0310.32 
         0.063 0.0022 0.68 8.52 2.01 3.75 0.65 0.018 0.99  

Average Grain Data

Elemental abundance data for Li, B, Mg, Al, K, Ca, Ti, V, Cr, and Fe relative to Si for each grain are presented in Table 2. An element’s “average” (calculated as a geometric mean) abundance within a grain is also provided along with the size of each grain and the estimated sputter depth during each individual measurement.

Figure 4 shows the average abundances, relative to Si and normalized to CI abundances (Anders and Grevesse 1989), of Mg, Fe, Ca, Al, Ti, and V compared with data measured using DC-SIMS by Amari et al. (1995) (in 34 mainstream SiC grains from the KJH series [3.4–5.9 μm]), and TOFSIMS by Henkel et al. (2007a) (three mainstream SiC grains from the KJG series [1.5–3 μm], and four extracted using acids from Tieschitz). In all grains, Mg, Fe, Ca, and Al are depleted relative to Si and CI abundances.

Figure 4.

 Comparison of the average abundances of Mg, Fe, Ca, Al, Ti, and V in individual presolar SiC grains analyzed by TOFSIMS in this study and Henkel et al. (2007a), and DC-SIMS by Amari et al. (1995). The data shown for both Amari et al. (1995) and Henkel et al. (2007a) represent only that measured in mainstream SiC grains in those studies. The maximum and minimum abundances measured in this study are also shown. Abundances are relative to Si and normalized to CI. Elements more volatile than SiC, such as Mg and Fe, are depleted in the grains.

Our average abundances for Mg, Fe, Ca, and Al were all approximately an order of magnitude higher than those measured by Amari et al. (1995) and Henkel et al. (2007a). However, Amari et al. (1995) did measure similarly high abundances of Mg, Fe, Ca, and Al in some individual SiC grains, while the lower ranges of our measured abundances were comparable to the average abundances reported by both Amari et al. (1995) and Henkel et al. (2007a). The average abundances of Ti and V were similar to those reported by Amari et al. (1995), although the lower limit was more consistent with the average Ti and V abundances measured by Henkel et al. (2007a).

The average Li/Si and B/Si ratios for the grains were approximately 10−3 and 10−4, respectively. Hoppe et al. (2001) measured B/Si ratios of 10−3–10−5 in mainstream SiC grains from the Murchison meteorite. Huss et al. (1997) measured Li/Si ratios of 10−4–10−6 in SiC grains from the Orgueil meteorite, at least an order of magnitude lower than those reported here, while in large (>5 μm) presolar SiC grains, Gyngard et al. (2009) measured Li abundances as low as approximately 10−8. In contrast, Lyon et al. (2007) measured Li/Si up to approximately 10−2 in the outer regions of both acid and gently separated SiC grains.

High abundances of Li may be the result of contamination from the meteorite matrix or laboratory. The Li/Si ratio in bulk meteorite matrix is approximately 10−5 (Curtis et al. 1980). During depth profiling of the silicate grains from the Murchison matrix, no Li (or B) was detected suggesting that the high abundances measured in the SiC grains are unlikely to be due to contamination with the matrix material. Furthermore, secondary ion images from the analyses of the SiC grains showed measured Li to be strongly localized to the grains, with no significant contribution arising from the surrounding Au-foil. We therefore argue that the measured Li abundances are not from contamination and must have been produced prior to the grains’ incorporation into the meteorite parent body.

Depth Profiles

Due to the variety of factors that can influence the elemental composition of presolar SiC grains, trace element depth profiles for individual grains are complex. Full descriptions and color figures (S3–S13) of individual grains’ depth profiles are provided in Appendix S2. Seven of the grains produced symmetrical trace element depth profiles, the other four asymmetrical in element abundance relative to the center of the grain. This is either a true reflection of the distribution of trace elements within some grains, or, we suspect, indicates that in some instances, cleaning steps with the DC-beam, necessary to combat the build-up of hydrocarbons upon grain surfaces both in vacuum and during handling, were too large and information was lost from the grains.

The grains have been grouped according to whether their highest trace element abundances occurred either near the grain surface (at depths of <100 nm) or below it (depths >100 nm) (Table 3). In four grains, elemental abundances showed little variation with depth resulting in “flat” depth profiles. It should be noted that in none of the grains did every trace element display the same depth-profile trend. The grains have therefore been assigned to a group if at least 50% of the measured trace elements had the same depth profile. Each group is described in detail below and representative depth profiles for each group are shown in Fig. 5.

Table 3.   Summary of the main characteristics of each presolar SiC grains trace element depth profile. The grains either have symmetrical or asymmetrical depth profiles. They are grouped according to whether any significant changes in trace element abundances occur near (<100 nm), or below (>100 nm), the grain surfaces.
Variation?SymmetricalAsymmetrical
Near surface (<100nm)AK-KJG-M-1
AK-MM2-G-2
AK-KJG-D-1
Below surface (>100nm)AK-KJG-D-2AK-MM2-G-1
AK-MM2-delta-1
AK-MM2-delta-2
FlatAK-KJG-6-1
AK-MM2-M-1
AK-MM2-N-1
AK-MM2-5-1
 
Figure 5.

 Representative trace element depth profiles. In some presolar SiC grains, variations in trace element abundances occurred from the grain surface. For grain AK-KJG-D-1 (top left), this can be attributed to surface contamination, while in other grains, such as AK-KJG-M-1 (top right), it may be the result of implantation. In some grains, such as AK-MM2-G-1 (bottom left) where abundances rose to a peak at approximately 150 nm, variations in trace element abundance took place below the surface. The trace elements in several grains showed no abundance variations with depth, e.g., AK-MM2-M-1 (excluding Li, bottom right). Dashed vertical lines show where a grain was removed and re-imaged using the ESEM. Trace element depth profiles for all eleven analyzed grains are provided in Appendix S2. Error bars are 1σ from counting statistics.

Highest Elemental Abundances Near the Surface

The grains AK-MM2-G-2, AK-KJG-D-1, and AK-KJG-M-1 had depth profiles where the highest trace element abundances were observed near the grain surfaces. In each grain, these abundances then decreased as the grain cores were exposed by primary ion beam sputtering. For example, in AK-KJG-D-1, at a depth of <100 nm, the Ca/Si ratio was 0.59, K/Si was 0.15, and Fe/Si was 0.17. Figure 5 shows that as the grain was sputtered away, these abundances rapidly decreased by orders of magnitude to lower values within the grain core. The abundances then remained low and did not rise again as the far side of the grain (i.e., closest to the Au-foil) was reached. Most elements measured in AK-KJG-D-1 followed the same trend, although the decreases between the outer surface and grain core were less striking.

In the grains AK-MM2-G-2 and AK-KJG-M-1, the highest trace element abundances were also measured near the grain surfaces. However, the abundances of elements such as K, Ca, and Fe were at least an order of magnitude lower than those near the surface of AK-KJG-D-1. Trace element abundances in AK-MM2-G-2 and AK-KJG-M-1 were lower in the grain cores, although the decreases between the surface and core were typically only a factor of approximately 4. In contrast to AK-KJG-D-1, the abundances in AK-MM2-G-2 and AK-KJG-M-1 then increased by a comparable amount at the far side of the grains resulting in symmetrical profiles.

Highest Elemental Abundances Below the Surface

In grains AK-MM2-G-1 and AK-MM2-delta-1, the highest trace element abundances occurred as peaks below the grain surface (i.e., at depths >100 nm). Figure 5 shows that for AK-MM2-G-1, the elements Li, Al, Ti, V, and Cr increased, on average, by a factor of approximately 80, from near the grain surface to a peak at a depth of approximately 150 nm. The abundance of these elements then fell (although less significantly for Li) as the grain was sputtered away. In AK-MM2-delta-1, we observed a similar depth profile for Li, Al, K, Ca, Ti, Cr, and Fe, whose abundances rose to a peak at approximately 200 nm before decreasing in the grain core. In neither case did we observe a similar peak at the far side of the grains.

In the grains AK-MM2-delta-2 and AK-KJG-D-2, the abundances of most trace elements were highest within the outer 150–300 nm. In contrast to grains AK-MM2-G-1 and AK-MM2-delta-1, however, the abundances did not rise to a peak, instead remaining relatively constant to these depths. As greater depths in grains AK-MM2-delta-2 and AK-KJG-D-2 were reached, the elemental abundances did decrease by factors of 3−25. The abundances remained low in AK-MM2-delta-2 producing an asymmetrical profile, but in AK-MM2-D-2, we measured an abundance rise at the far side of the grain.

Flat Depth Profiles

Grains AK-MM2-M-1, AK-MM2-N-1, AK-MM2-5-1, and AK-KJG-6-1 all had “flat” depth profiles. In each of these profiles, the majority of elemental abundances showed very little variation with depth. Often, the elemental abundances in the grains changed by less than a factor of 3 across the entire depth profile, comparable to the differences measured when using the same procedure to analyze the meteoritic silicate grains (see the Depth-Profiling Procedure section). We therefore consider that the variations measured in these SiC grains cannot be said to be significant and group them separately from those which, as described above, displayed very clear changes in their trace element abundances with depth.

Discussion

Isotopes

Figure 3 shows that all of the SiC grains were enriched in 29Si, and all except one also in 30Si, ruling out the possibility of them being X grains (depleted in 29Si and 30Si) from supernovae. However, there is a tendency that compared with the well-defined Si-isotopic systematics of mainstream presolar SiC grains (e.g., data taken from Hoppe et al. [1994] in Fig. 3), our data are biased toward 29Si, with δ29Si values at the upper end of those measured previously. This is likely due to an insufficient deconvolution of the 28SiH+ interference from the 29Si-peak. Lyon et al. (2007) showed that the presence of the 28SiH+ interference can add a 20–30‰ uncertainty to the δ29Si value when using the peak deconvolution technique. Within the analytical uncertainties, we cannot determine between the mainstream, A, B, Y, or Z grains, but since mainstream grains make up >90% of all presolar SiC, our grains probably belong to this group. They therefore probably originated around 1–3M AGB stars (Zinner et al. 1989; Hoppe et al. 1993, 1994; Lugaro et al. 2003).

Average Abundance Patterns

Amari et al. (1995) and Lodders and Fegley (1995) suggested that average elemental abundance patterns reflect the volatility of an element and its compound species relative to SiC. Our average elemental abundance patterns for Mg, Fe, Ca, Al, Ti, and V (Fig. 4) match qualitatively with those measured by Amari et al. (1995) and Henkel et al. (2007a). The patterns show that the more volatile elements (Mg, Fe, Ca, and Al) are depleted in the grains relative to CI abundances.

Our average elemental abundances of Ti and V (relative to Si and CI) are close to 1, similar to the abundances measured by Amari et al. (1995). Figure 4 shows that the Ti and V abundances of Henkel et al. (2007a) were approximately an order of magnitude lower, but those measurements only sampled the grain surfaces (typically <100 nm). Around stars, Ti is predicted to condense as TiC prior to the condensation of SiC (Lodders and Fegley 1995), while strong correlations between the abundances of Ti and V in presolar SiC suggest that V condenses into the TiC grains (Amari et al. 1995). This should result in a depletion of Ti and V in the gas and subsequently in SiC grains. However, the very low abundances of Ti in the source gas restrict the condensation of TiC due to nucleation effects. Therefore, Ti and V are not entirely removed from the gas and are still available for condensation into SiC.

The average elemental abundance patterns of Amari et al. (1995), Henkel et al. (2007a), and those presented in this work, indicate that the condensation behavior of each element in stellar atmospheres must play a major role in controlling their abundances in presolar SiC grains. The range of our average elemental abundances for Mg, Fe, Ca, Al, Ti, and V are comparable to those of Amari et al. (1995) and other similar studies (e.g., Huss et al. 1997; Hoppe et al. 2000). To measure trace element abundances, Amari et al. (1995) used DC-SIMS, which sputters a much larger volume of each grain during a single analysis than the TOFSIMS technique used here (and also in the study of Henkel et al. 2007a), which has a depth resolution of tens of nanometers. In the Amari et al. (1995) study, trace element measurements were also made after some previous sputtering of the grains had taken place to determine isotopic compositions (Hoppe et al. 1994). The data of Amari et al. (1995) therefore represented an average elemental abundance over a large volume of each analyzed grain. However, trace element depth profiles of individual grains show that the elements are clearly not always homogeneously distributed within all SiC grains.

Contamination

During depth profiling of a silicate grain and many of the SiC grains, we observed no significant variation in trace element abundances in the measurements acquired immediately after the grain had been re-imaged in the ESEM. However, trace element abundances in the SiC grains AK-MM2-5-1 and AK-KJG-6-1 did decrease noticeably in the first measurements after re-imaging, before returning to values comparable to those previously recorded in the grains. Possible explanations are the deposition of contaminants upon the surface of these grains while being transferred between the instruments, or that during analysis within the ESEM, interaction with the electron beam altered the chemical environment on the grain surfaces. Electron beam cracking of hydrocarbons in the residual gas of the ESEM sample chamber is known to cause a deposition of carbon films upon sample surfaces (e.g., Kondoleon et al. 2000). Influences on this process, including the cleanliness of the sample chamber and irradiation time, can vary both between analysis sessions and also from grain to grain. Any variation may have led to the deposition of carbon films of differing thicknesses on the grain surfaces. We suggest therefore that the surfaces of grains AK-MM2-5-1 and AK-KJG-6-1 were possibly coated by carbon films during re-imaging by the ESEM, and these were not efficiently removed by the DC beam. The films probably yielded lower trace element abundances until sputtered away during the TOFSIMS measurements and this data has been excluded.

The extraction process used to isolate presolar SiC grains has been identified as a source of potential trace element contamination. Knight et al. (2008) compared trace element abundances measured in KJG SiC grains extracted by Amari et al. (1994) with those in a set of grains that underwent additional HF and HCl cleaning steps. They found that although Ti, V, Cr, and Mn abundances between the two sets of grains were largely similar, W and Pb were noticeably more abundant in the KJG grains. This was attributed to contamination from the sodium polytungstate used for a density separation of the KJG SiC grains during the extraction process.

In this study, we analyzed KJG SiC grains, plus grains obtained using the Amari et al. (1994) protocol, so cannot fully exclude trace element contamination from the extraction procedure. However, if it did occur, we would expect it to affect all SiC grains from the same separation equally, although potentially the effects could vary between the KJG and MM2 samples. All grains from the same separation would probably have similar surface abundances of any contaminating elements, yet we measured orders-of-magnitude differences in trace element abundances near the surfaces of SiC grains from each separation. For example, the Cr/Si ratio near the surface of AK-MM2-M-1 was 0.0061, but, for AK-MM2-delta-1, it was 2.3 × 10−4. Similarly, for AK-KJG-D-2, the Cr/Si ratio was 0.051 near the surface, compared to 2.2 × 10−4 in AK-KJG-M-1.

Recent evidence has indicated that the use of harsh acid treatments to extract presolar SiC grains from meteorites can also etch grain surfaces, leading to a high density of surface pits (e.g., Bernatowicz et al. 2003). The acids may also at least partially dissolve any nonacid resistant phases, such as AlN, present at the surface of a presolar SiC grain (Stephan et al. 1997). Henkel et al. (2007a) suggested that depletions of Al at the surfaces of grains, plus correlations between the abundances of Al, Mg, K, and Ca, which are major constituents of meteorite matrix material, were caused by the dissolution of AlN and subsequent deposition of meteoritic matrix material in crystal defects. This could lead to high abundances of Mg, K, Ca, and Fe at SiC grain surfaces. In addition, during analysis of pristine SiC, Stroud and Bernatowicz (2005) noted an apparent in-filling of surface pits by meteorite matrix minerals.

Quantifying the effects of any potential grain surface contamination from the meteorite matrix is difficult, as the likelihood of any particular SiC grain acquiring contaminants will be strongly influenced by its individual structural characteristics. Presolar SiC grains occur as two polytypes, either cubic 3C (80%) and hexagonal 2H (3%), or as intergrowths between the two (17%) (Daulton et al. 2002, 2003). Each grain may have a varying density of surface pits, and each may also contain different abundances of nonacid resistant subgrains such as AlN. This means that the deposition of meteorite matrix material is unlikely to be a uniform process on SiC grain surfaces. The grains in this study were all of similar sizes and did not appear significantly different in morphology from one another, but we cannot rule out that some grains may have taken up more meteoritic contamination than others. Nevertheless, if this contamination did occur, it should be reflected by very high surface abundances of elements such Mg, K, Ca, and Fe.

Contamination was only suspected for AK-KJG-D-1, where the abundances of many elements near the grain surface were significantly higher than those measured in the other SiC grains. In particular, the initial Ca/Si ratio was 0.59, and the K and Fe initial abundances were similarly high. During subsequent measurements, these abundances decreased rapidly, until at a depth of <100 nm they reached values typical of the other presolar SiC grains. The detection of exceptionally high trace element abundances at a shallow depth seems to indicate that there was contamination upon the grain surface, probably caused by the deposition of meteoritic matrix material. This maybe suggests that grain AK-KJG-D-1 had many crystal defects, possibly formed through the dissolution of abundant AlN subgrains during extraction.

Grain Profiles

The abundances of most elements analyzed in grains AK-MM2-M-1, AK-MM2-N-1, AK-MM2-5-1, and AK-KJG-6-1 did not show any significant variation with depth. Elemental abundances within the grains were largely consistent with those measured in other SiC grains (e.g., Amari et al. 1995; Henkel et al. 2007a). The homogeneous distribution of elements is most likely due to their condensation in solid solution with the SiC in the atmospheres of their parent stars. That the elemental homogeneity is preserved suggests that these grains suffered only minimal alteration prior to their arrival in the laboratory, although homogenization by secondary alteration processes cannot be ruled out.

In grain AK-MM2-G-1, the elements Li, Al, Ti, V, and Cr, and in grain AK-MM2-delta-1, the elements Li, Al, K, Ca, Ti, Cr, and Fe, showed depth profiles that rose to peaks in abundance at depths within the grains. The depths, approximately 150–200 nm, at which abundance peaks occurred in grains AK-MM2-G-1 and AK-MM2-delta-1, were below the grain surfaces and comparable to those reported by Lyon et al. (2007), although care must be taken when referring to exact depths in the profiles. As outlined previously, differential sputtering due to uneven grain surfaces combined with the errors on the known sputter rates, lead to errors on our depths of >20%. Also, due to the amount of material sputtered within the first two measurements of both grains, we can only confidently state that the peak in AK-MM2-G-1 occurred between 60 and 150 nm, and for AK-MM2-delta-1, between 10 and 200 nm. Nonetheless, we argue that the peaks in trace element abundances in these grains are real variations, with elemental abundances in grains AK-MM2-G-1 and AK-MM2-delta-1 increasing by up to a factor of 190 compared to only a factor of 2.2 when analyzing meteoritic silicate grains (Fig. 2), and are consistent with the trends observed by Lyon et al. (2007).

The peaks in trace element abundances may have been caused by the sputtering of any TiC or AlN subgrains or inclusions present within the SiC grains, although there was no evidence of any such grains in the secondary ion images. A further argument against the TiC or AlN origin for the abundance peaks is that they are unlikely to contain significant abundances of elements such as Li or K, which also peaked at depth within the grain AK-MM2-delta-1. Additionally, the possibility of AlN and TiC being present at the same depth within two different grains is unlikely, although as such subgrains can be widespread in some presolar SiC we cannot entirely exclude this explanation.

The peaked depth profiles appear to qualitatively support the scenario whereby elements may be implanted into presolar SiC grains by supernova shockwaves. Calculations using the SRIM code (Ziegler 2004) indicate that the implantation of Li, Al, Ca, Ti, Cr ions, etc. to depths of 150–200 nm in SiC requires implantation velocities on the order of 1000 km s−1 (Fig. 6). This is toward the upper end of the velocity range expected for supernova shockwaves (Jones et al. 1996; Truelove and McKee 1999). Lyon et al. (2007) argued that the relative uniformity of 7Li/6Li approximately 12 recorded within a range of extra-terrestrial samples (e.g., Sephton et al. 2004; Chaussidon et al. 2006; Seitz et al. 2007), and observed spectroscopically in local star-forming regions (Knauth et al. 2003), could only be achieved by mixing of several sources on a molecular cloud scale. This led them to conclude that the implantation of Li ions into presolar SiC grains must have taken place as a shockwave passed through an interstellar molecular cloud. The presolar SiC reported here also contain solar Li isotopic compositions, supporting the model outlined by Lyon et al. (2007).

Figure 6.

 SRIM calculations simulating the implantation of Li, B, Al, Ca, and Ti ions into a 1 μm SiC grain at a velocity of 1000 kms−1. The ions impact the SiC at normal angle of incidence. At this velocity, most ions are implanted into the SiC to a depth of approximately 200 nm, consistent with peaks in elemental abundances observed in some presolar SiC grains. The same calculation for Mg, K, Cr, V, and Fe ions produces similar results.

The depth profiles for grains AK-MM2-G-2, AK-KJG-D-2, and AK-KJG-M-1 were symmetrical with respect to the center of the grains, with the highest elemental abundances observed at or just below the grain surfaces. The highest elemental abundances in grain AK-MM2-delta-2 were also recorded just below the grain surface, although in this case the profile was not symmetrical. As previously discussed, interpreting such depth profiles, where elemental abundances are elevated in the outer rims relative to the grain core, can be complicated by an insufficient spatial resolution that hampers distinguishing between the two regions. Abundances recorded in the cores must therefore be upper limits. Using our geometrical model (see Appendix S1), we can infer that decreases in trace element abundances of a factor of approximately 4, which were typically measured between the outer regions and cores of grains AK-MM2-G-2, AK-KJG-D-2, and AK-KJG-M-1, would be consistent with enriched outer rim thicknesses of approximately 200 nm. In grain AK-MM2-delta-2, the elemental abundances fell by up to a factor of 25, which cannot be explained by our model. However, it is important to note that the model does not take into account irregular grain morphologies or preferential sputtering by the ion beam.

One possibility is that the grains AK-MM2-G-2, AK-MM2-delta-2, AK-KJG-D-2, and AK-KJG-M-1 had small (approximately 100s nm) non-SiC grains attached to their surfaces that, until completely sputtered away, resulted in elevated elemental abundances. We do not believe that this occurred, as even the highest abundances in these grains were comparable to those in the other analyzed SiC grains. Alternatively, depth profiles for these grains may represent elemental zoning within them due to their formation in a cooling source gas. Presolar SiC grains formed at high temperatures (approximately 1600 K) would have initially acquired low abundances of more volatile trace elements in their cores. As the source gas cooled (down to temperatures of approximately 1200 K), some SiC grains may have continued to grow, or new grains could have condensed The lowering temperature of the gas would have permitted the condensation of increasingly volatile elements that could have been incorporated into grains’ outer regions. However, at these temperatures, diffusion coefficients for several trace elements in SiC (e.g., Van Opdorp 1971) suggest that their homogenization within a 3 μm diameter grain would occur on a timescale of only a few minutes, which is much shorter than the expected formation time of the grains (Lodders and Fegley 1995).

We suggest that depth profiles for grains AK-MM2-G-2, AK-MM2-delta-2, AK-KJG-D-2, and AK-KJG-M-1 may also reflect the implantation of interstellar matter. As most supernova shockwaves occur at velocities of <1000 km s−1, we would expect to see evidence in some presolar grains for implantation at shallower depths (i.e., <200 nm) than observed in grains AK-MM2-G-1 and AK-MM2-delta-1 and other studies (Lyon et al. 2007). SRIM calculations show that implantation into SiC at 500 km s−1 would produce peaks in elemental abundances at a depth of only approximately 50 nm. Analytical constraints in our measurements would make any such peaks difficult to resolve, and it is likely that we would only observe a decrease in abundance from the grain surface. As this trend is seen in the depth profiles for grains AK-MM2-G-2, AK-MM2-delta-2, AK-KJG-D-2, and AK-KJG-M-1, they may be explained by implantation via supernova shockwaves traveling at velocities of <1000 km s−1.

An important consideration is that ions accelerated by supernovae shockwaves in the ISM not only become implanted into dust but also lead to sputtering and erosion of grain surfaces (Jones et al. 1996; Slavin et al. 2004). A study of pristine SiC grains found no features of erosion on grain surfaces, indicating either that the grains’ residence time in the ISM was either very short or they were protected by coatings (Bernatowicz et al. 2003). There is evidence to support both theories, with 60% of the pristine grains reported to have amorphous coatings, and the majority of SiC grains found to have interstellar residence times of <200 Ma (Ott et al. 2005; Gyngard et al. 2009; Heck et al. 2009). As many of the grains do not appear to be coated, while a few have interstellar residence times of approximately 1000 Ma, it suggests that at least a fraction of the SiC population probably suffered significant erosion in the ISM. If true, then during analyses of such grains, we would not have sampled their original surfaces.

The grains AK-MM2-G-2, AK-MM2-delta-2, AK-KJG-D-2, and AK-KJG-M-1 may have passed through a high-velocity supernova shockwave that initially caused the implantation of ions, and produced peaks in elemental abundance at depths of approximately 200 nm. As the number of ions impacting the grain surfaces increased over time, sputtering and grain erosion would become the dominant process, and the number of atoms removed would overtake the number implanted into the grains (Jones et al. 1996; Slavin et al. 2004). Assuming that these grains were uncoated, this erosion would be sufficient to remove enough material so that only a portion of the elemental abundance peak produced by the initial implantation was retained within the grains. We argue therefore that in grains AK-MM2-G-2, AK-MM2-delta-2, AK-KJG-D-2, and AK-KJG-M-1, we measured the tail of an elemental abundance peak caused by ion implantation and subsequently removed by sputtering in the ISM. As the peak no longer existed within the grains, elemental abundances in AK-MM2-G-2, AK-KJG-D-2, and AK-KJG-M-1 only fell by around a factor of 4 between the surface and core, rather than the orders-of-magnitude decreases observed in other grains with implantation profiles. The decrease in abundances of up to a factor of 25 in grain AK-MM2-delta-2 was consistent with other implantation profiles, perhaps suggesting that less material was sputtered from the grain surface, or that the implantation peak was present at a greater depth within the grain.

Implications

Based upon our interpretations of trace element depth profiles, presolar SiC grains can be separated into two distinct groups. The grains AK-MM2-M-1, AK-MM2-N-1, AK-MM2-5-1, and AK-KJG-6-1 had trace element distributions consistent with those expected for condensation processes around parent stars. The grain AK-KJG-D-1 showed evidence for grain surface contamination followed by homogeneously distributed trace elements, suggesting that it also belongs to this group. In contrast, the SiC grains AK-MM2-G-1, AK-MM2-G-2, AK-MM2-delta-1, AK-MM2-delta-2, AK-KJG-D-2, and AK-KJG-M-1 had trace element depth profiles that could possibly be attributed to ion implantation caused by supernova shockwaves in the ISM. However, the elemental abundances and distributions measured in the cores of these grains are almost certainly due to the grains’ original condensation environment.

The two groups may reflect the large variations in ISM residence times previously determined for other presolar SiC grains (Ott et al. 2005; Gyngard et al. 2009; Heck et al. 2009). Cosmic ray exposure ages indicate that most grains spent <200 Ma in the ISM compared with lifetimes for a few grains of approximately 1000 Ma. Unfortunately, within the analytical uncertainties, we detected no Li isotopic anomalies that could be used to calculate cosmic ray exposure ages for the grains in this study. It seems reasonable to assume that the grains present in the ISM for the longest periods of time were more likely to have experienced at least one, if not more, shockfronts fast enough to implant ions below their surfaces. Approximately 55% of the SiC grains in this study had trace element profiles that were interpreted as due to implantation, although we would expect these grains to be rarer, as interactions with faster, or even possibly multiple, shockwaves would probably destroy them. The remaining grains (approximately 45%) analyzed did not contain evidence for ion implantation, which, we suggest, is because they were only in the ISM for a short time, consistent with the cosmic ray exposure ages measured in other presolar SiC grains.

Another factor is whether the grains were protected from ion implantation in the ISM by mantles of ices or organic materials. The percentage of grains in this study seemingly unaltered by implantation is slightly lower than the fraction of pristine SiC grains reported by Bernatowicz et al. (2003) to have surface coatings. Icy and/or organic coatings on grains may play a significant role in protecting them from the effects of sputtering, regardless of their residence time in the ISM. A critical next step in trying to quantify the influence coatings have on dust in the ISM will be to further investigate the isotopic and elemental compositions of pristine presolar grains.

Summary

We have obtained complete depth profiles of Li, B, Mg, Al, K, Ca, Ti, V, Cr, and Fe in eleven acid-extracted presolar SiC grains. The average elemental abundance patterns of Mg, Fe, Ca, Al, Ti, and V indicate that condensation processes heavily dictate the abundances of these elements in presolar SiC.

We find evidence for two distinct groups of presolar SiC grains based upon the distribution of trace elements within them. Approximately 45% of the SiC grains contained homogenous trace element distributions that can be explained by condensation processes around the grain’s parent stars. Trace element depth profiles for six of the presolar SiC grains analyzed can be interpreted as due to the implantation of ions at velocities up to approximately 1000 km s−1. Supernova shockwaves accelerating material in the ISM are the suggested source of this implantation, although they are also likely to cause significant erosion of grain surfaces.

The two groups may suggest that some SiC grains experienced no, or only minimal, processing prior to arrival in the presolar molecular cloud, whereas others must have passed through high-velocity shockwaves. This is consistent with cosmic ray exposure ages that show that many presolar SiC grains had short (<200 Ma) residence times in the ISM, while others had residence times of up to approximately 1000 Ma. Grains with longer residence times are more likely to have passed through a high-velocity supernovae shockwave, although this also increases their chances of being destroyed and should make them rarer in our presolar grain collection. In the ISM, some grains may also have been protected from shockwaves by the presence of surface coatings.

Acknowledgments—  The authors thank S. Amari and R. Lewis for providing the KJG SiC grains and permission to publish data obtained from them; J. Arden for separating the MM2 grains; and D. Blagburn and B. Clementson for instrument maintenance. We thank P. Heck, T. Croat, and an anonymous reviewer for their constructive and helpful comments, and are extremely grateful to P. Hoppe, M. Schonbachler, and I. Franchi for suggestions that significantly improved the manuscript. This work was supported by the Science and Technology Facilities Council (STFC) through a studentship for AK, research position for TH, and the STFC UKCAN program for DR.

Editorial Handling—  Dr. Ian Franchi

Ancillary