Abstract– Samples returned from comet 81P/Wild 2 by the Stardust mission provided an unequaled opportunity to compare previously available extraterrestrial samples against those from a known comet. Iron sulfides are a major constituent of cometary grains commonly identified within cometary interplanetary dust particles (IDPs) and Wild 2 samples. Chemical analyses indicate Wild 2 sulfides are fundamentally different from those in IDPs. However, as Wild 2 dust was collected via impact into capture media at approximately 6.1 km s−1, it is unclear whether this is due to variation in preaccretional/parent body processes experienced by these materials or due to heating and alteration during collection. We investigated alteration in pyrrhotite and pentlandite impacted into Stardust flight spare Al foils under encounter conditions by comparing scanning and transmission electron microscope (SEM, TEM) analyses of preimpact and postimpact samples and calculating estimates of various impact parameters. SEM is the primary method of analysis during initial in situ examination of Stardust foils, and therefore, we also sought to evaluate the data obtained by SEM using insights provided by TEM. We find iron sulfides experience heating, melting, separation, and loss of S, and mixing with molten Al. These results are consistent with estimated peak pressures and temperatures experienced (approximately 85 GPa, approximately 2600 K) and relative melting temperatures. Unambiguous identification of preserved iron sulfides may be possible by TEM through the location of Al-free regions. In most cases, the Ni:Fe ratio is preserved in both SEM and TEM analyses and may therefore also be used to predict original chemistry and estimate mineralogy.
Iron sulfides are one of the major mineral types found in cometary dust. They are common in chondritic porous (CP) interplanetary dust particles (IDPs) (Bradley 2003) and have been identified as a major constituent of the comet 81P/Wild 2 samples collected by NASA’s Stardust mission (Zolensky et al. 2006). Iron sulfides in CP IDPs are predominantly pyrrhotite with up to 20 atom% Ni although occasionally other iron sulfides such as pentlandite, troilite, and sphalerite are found (e.g., Fraundorf 1981; Christoffersen and Buseck 1986; Zolensky and Thomas 1995; Dai and Bradley 2001; Bradley 2003). In particle analyses from Stardust aerogel impact tracks, the preliminary examination petrology team reported the occurrence of pentlandite (rare) and Ni-free pyrrhotite (Zolensky et al. 2006). However, unlike the CP IDPs there were no intermediate phases between the iron sulfides and iron-nickel sulfides, and there was a range of nonstoichiometric Ni-free iron sulfide compositions with varying S content (Zolensky et al. 2006). Although the range of S content could be attributed to an original diverse nonstoichiometric composition (and therefore to real differences between CP IDPs and comet 81P/Wild 2), it was also suggested that it could reflect a loss of S as a result of capture heating, thereby indicating that the initial impactors were pyrrhotite and pentlandite (Brownlee et al. 2006; Zolensky et al. 2006). Previous experiments by Barrett et al. (1992) identified iron blebs in aerogel tracks produced by pyrrhotite and deemed them as being the result of S-volatilization and loss during impact. Indeed sulfur is considerably more volatile than iron and nickel (based on its 50% condensation temperature; Lodders 2003) and the mobilization and loss of S from S-bearing minerals upon moderate heating has been previously observed and documented (e.g., Lauretta et al. 1997). To ensure correct interpretations of comet 81P/Wild 2 mineralogy, and therefore preaccretional or parent body processes, an investigation into the effects of capture on sulfides is vital.
The Stardust encounter conditions (well-constrained, relatively low velocity of approximately 6.1 km s−1, and a perpendicular angle of impact incidence; Brownlee et al. 2003) are such that they can be replicated in the laboratory using the two-stage light gas gun (LGG). We can therefore take known minerals, impact them into flight spare Stardust collection media, and make direct comparisons between the preimpact and postimpact materials to determine whether the impactor remains the same or, if not, whether it is even recognizable. Analyses of Stardust aerogel impact analogs of pyrrhotite have previously been undertaken. Using a hard X-ray scanning fluorescence microprobe, spectra were obtained over traverses through a track perpendicular to its length, and S from the pyrrhotite was found to have mobilized and diffused through the aerogel (Ishii et al. 2008a). Of more concern, transmission electron microscopy (TEM) analyses showed that these impacts also resulted in the production of nanometer-sized inclusions of metal and sulfide embedded in a silicate glass (molten aerogel) (Ishii et al. 2008b). This material is almost identical to the CP IDP constituent known as GEMS (glass with embedded metal and sulfides) expected to be found in comet 81P/Wild 2. Therefore, the alteration experienced by sulfides captured in aerogel is not simply the loss of S but is considerably more complex, with the resulting material potentially being indistinguishable from GEMS. Therefore, although larger terminal particles may preserve some sulfide, for the vast majority of aerogel captured grains, deciphering the original mineralogy of sulfides is virtually impossible.
The Al 1100 foils on the Stardust spacecraft had the primary function of securing aerogel blocks in place and allowing their safe removal upon return (Brownlee et al. 2003). However, they also provided an additional capture surface totaling approximately 153 cm2 (Tsou et al. 2003) upon which cometary materials may be examined. Unlike the silica aerogel, the material captured by the foils is highly localized in the form of impact residues that line impact craters (Fig. 1). In addition, there is no external source of silicate, and impacts occur into a heat sink rather than an abrasive insulator. This last point could have significant implications for sulfides. Many sulfides can occur in multiple structural states having formed under different pressures and temperatures (e.g., orthorhombic cubanite forms and is stable at temperatures below 210 °C, above these temperature it is transformed to cubic cubanite which is unstable when temperatures are removed and exsolves to chalcopyrite and pyrrhotite; Putnis 1977). Those formed under extreme conditions that can be quenched to lower temperatures and pressures often only persist metastably, transforming to more stable phases with the application of heat. Therefore, metastable phases and any low temperature phases together with the formation record they preserve can be lost when heated without the need for melting to occur. Consequently, having a target like the foil, which acts to remove heat quickly, is far more desirable than one which retains heat like the aerogel. The foils therefore have a great potential for providing us with information on impacting cometary grains. In fact, previous investigations into laboratory analogs have already shown that the foils surpass the aerogel in their ability to provide information such as original impactor size, density, and structure (e.g., Kearsley et al. 2007, 2008; Burchell et al. 2008a). Analyses of the Stardust cometary impact craters show that they typically contain residues composed of a mixture of crystalline and amorphous materials (Leroux et al. 2008) which are tentatively assumed to be relict cometary dust (crystals) and shock-induced melt that subsequently quenched (amorphous material). Determining the value of these materials to our understanding of comets requires that we understand the impact process and identify the products of capture through analyses of impact analogs. Laboratory analog analyses suggest precursor chemistry (e.g., Wozniakiewicz et al. 2009) and even crystallinity (Raman signals detected by Burchell et al. 2008b) may be preserved for anhydrous silicate minerals. An investigation into the preservation and/or recognizability of sulfides is the next logical step. A variety of foils and other surfaces have been exposed in low Earth orbit over the past few decades, successfully sampling materials passing through the local Earth environment. However, the level of preservation varied due to the diverse and unknown impact velocities and angles of incidence. Hence, these findings cannot be used to gain insight into Stardust foils (see Wozniakiewicz et al. 2009, Graham et al. 2001, and Kearsley et al. 2009 for more complete discussion of previous impact crater work). It is the aim of this article to analyze preimpact and postimpact sulfides to determine the state of preservation and extent of alteration, and therefore evaluate the recognizability, of these minerals as collected by Stardust.
The largest Stardust crater identified has a diameter of 680 μm which, according to the particle-crater diameter correlations by Kearsley et al. (2007), would require an impacting particle with a diameter >100 μm. This crater is indeed large compared to the majority of craters on the Stardust foils; however, 63 Stardust craters with diameter >20 μm were identified during the preliminary examination alone, and it is craters of this size that contain the majority of the cometary dust mass collected on Stardust foils (Kearsley et al. 2008).1 The analyses in this article therefore focus on craters >50 μm in diameter formed by impacting particles of >10 μm in diameter (Kearsley et al. 2007; Price et al. 2010).
A large array of analytical tools are currently available for mineral analyses (e.g., Zolensky et al. 2000), each presenting different advantages and disadvantages, based on factors that include resolution, ease of sample preparation, speed of analysis, instrument availability, cost, and the extent of sample destruction during analysis. To fully exploit the unique sample set returned by Stardust in its Al foils, the data obtained must be maximized by performing nondestructive, quick, widely available methods to identify samples that might merit further investigation by higher resolution, more laborious and destructive methods. Scanning electron microscopy (SEM) with energy dispersive X-ray microanalysis (EDX) is a widely available technique used extensively to image and to perform initial compositional analyses of residues in craters on the Stardust Al foils (e.g., Zolensky et al. 2006; Kearsley et al. 2008). This was the chosen method for initial analysis, and later, for location of residue in our impact analog foils. TEM is a technique capable of high resolution compositional and structural analyses required for mineralogical classification over the small scales applicable to crater residues. It was therefore used for the more indepth analyses of the preimpact and postimpact samples to identify those features generated by the impact and those inherent to the impactor. These TEM results also allowed us to make a direct practical assessment of the quality of data previously obtained by the SEM. Estimates of the impact parameters (peak pressure, peak postshock temperature, and cooling time) were calculated to complement and clarify the findings of these analyses.
Samples and methodology
Impact Analog Production
The sulfide impact analog foils were generated using the two-stage LGG in the Centre for Astrophysics and Planetary Sciences at the University of Kent, Canterbury. The powder shot and velocity measurement techniques employed herein are described in Burchell et al. (1999). In each shot the target chamber was evacuated to a few ×10−1 mbar, or better, to minimize velocity loss in flight. The projectiles are loaded and launched within a sabot which protects them from heating via frictional interactions with the (limited) atmosphere and via direct contact with ignition and pressurized gasses, ensuring heating during acceleration is minimal. Indeed, Burchell et al. (2009) observe the survival of 20 nm organic polymer coatings on particles captured in aerogel shot at 1 km s−1 suggesting that acceleration does not heat the sample appreciably.
The target used in each case was a sample of flight-spare Stardust Al foil (provided by P. Tsou of the Jet Propulsion Laboratory). This foil is an Al 1100 series alloy (meeting the specifications of Al 1145) with temper grade 0. The foil is approximately 101 μm thick (Kearsley et al. 2007) and, for experimental shots, was wrapped around a 1 mm thick square Al alloy plate (Al 6082) measuring approximately 1.5 × 1.5 cm, to simulate the mounting on the Stardust collector. This was then held with conductive adhesive putty onto an Al base-plate (10 × 10 cm), drilled with a hole in each corner for support within the target chamber and to enable handling throughout the shot and subsequent analysis without damage or contamination to the craters and their residues. Throughout this preparation procedure, gloves and pliers were used to avoid contamination of the foil.
The iron sulfides chosen as projectile material were pentlandite and Ni-free pyrrhotite. The pyrrhotite was taken from collections at the Natural History Museum, London (specimen BM.2005,M317 from Drag, Tysfjord, Norway). Available museum samples of pentlandite were generally very small crystals incorporated in larger masses of pyrrhotite, chalcopyrite, and pyrite making them impossible to extract without including a high concentration of impurities. These were therefore deemed unsuitable for this investigation and a sample of pentlandite from Sudbury (Ontario, Canada) containing a crystal large enough to extract was purchased from a commercial source. These samples were prepared as projectiles with diameters of 53 μm or less by crushing with a pestle and mortar to produce powders and by passing through a 53 μm sieve. This range in impactor sizes (<53 μm) ensures that craters covering the range of those found on Stardust are produced. Between each projectile, the pestle and mortar and the sieve were thoroughly cleaned to avoid cross-contamination of different projectiles.
Both mineral projectiles were shot individually, generating separate impact analog foils to ensure that any identifications made were associated with the correct mineral. Pyrrhotite was fired in shot G080507#3 at a velocity of 5.82 km s−1. Pentlandite was fired in shots G2205072 and G291106#2 at velocities of 6.21 km s−1 and 5.85 km s−1, respectively. The average impact speed was 5.96 ± 0.12 km s−1 (see Table 1), measured individually in each shot with an accuracy to within 2%.
Table 1. Details of projectiles and parameters for two-stage LGG shots performed and analyzed as part of the work in this article.
Projectiles (NHM catalog no.)
Projectile size and shape
Impact speed (km s−1) ± 2%
Pentlandite (specimen awaiting cataloging)
Samples of the preimpact projectiles were mounted in resin blocks that were polished down to expose areas for analysis in the SEM (Fig. 2A).
The SEM analyses in this article sought not only to investigate sulfide impacts but also to determine the value of such initial investigations performed on Stardust craters. Therefore, residues had to be analyzed in situ (matching real Stardust SEM analysis conditions). Although the SEM is capable of obtaining EDX data from crater residues with no need for sample preparation, the complicated geometry involved and the location of the EDX detector on this instrument (see the Analytical Instrumentation and Sample Analysis section) means that analyses are limited to a crescent-shaped region opposite the detector. These regions are located on the steep crater walls resulting in oblique beam-to-sample incidence. Such beam-sample incidence angles result in X-ray excitation from a very shallow volume and consequently relative enhancement of low-energy X-rays, preventing the use of available quantitative correction routines (Goldstein et al. 1992; Kearsley et al. 2007; Wozniakiewicz et al. 2009). However, Kearsley et al. (2007) noted that, if the foil is tilted directly toward the EDX detector, it is possible to collect X-rays emitted from the crater wall-to-floor arc, with beam-normal incidence on the residue enabling the successful application of correction routines. Fortuitously, these (deeper) locations tend to contain a higher abundance of residue (Kearsley et al. 2008). The impact analog foils were therefore dismounted from the base plates and fixed to an SEM stub which could then be mounted at an angle of 40° (the tilt required for this SEM) during EDX analyses (Fig. 2B).
To analyze samples by TEM, they must be prepared as electron transparent thin sections that are approximately 100 nm thick or less. Currently the preferred method available for creating these sections from large samples utilizes the focused ion beam (FIB) to extract sections typically 10–20 μm long by 10 μm deep. This preparation method utilizes ions to remove material such that two trenches are milled progressively closer to one another until a section of approximately 1 μm thickness remains. The section can be removed and attached to a TEM grid by micromanipulators and then thinned down to 100 nm. As the section is thinned, progressively lower currents are used to avoid heating and therefore damage of the sample.2 This preparation method is sensitive to the topography of the sample: The flat, embedded projectiles are ideal; however, complex geometries like that of a crater cause complications to beam-sample interactions (with secondary ions and atoms being redeposited elsewhere in the crater, covering and modifying existing residue) and make section extraction by micromanipulators impossible. For very small craters (with diameters of less than a few micrometers), infilling with carbon to create a flat geometry has enabled sections across whole craters to be made and analyzed (e.g., Leroux et al. 2006, 2008, 2010; Stroud et al. 2010). However, larger craters like those investigated herein cannot be infilled in this way and, to date, only sections taken from their well-exposed lips or the tops of their crater walls have been attempted (e.g., Graham et al. 2006, 2008). Given that the majority of residue is often found at the bottom of these craters (Kearsley et al. 2008), and that our SEM analyses were focused on these lower regions, it was vital that a sample preparation method be developed to enable access to these deeper regions by FIB. Hence, a section of each foil was mounted vertically within a resin block. These blocks were then polished down until a crater cross section was revealed and residue was exposed at the surface (Fig. 2C). Due to the nonconductive nature of these blocks, they required C-coating prior to SEM analyses. SEM EDX was then used to determine the location of outcropping residue for FIB.
Analytical Instrumentation and Sample Analysis
All SEM imaging and EDX analyses for this investigation were conducted using the JEOL 5900 LV SEM in the Electron Microscopy and Mineral Analysis Division (EMMA) of the Department of Mineralogy at the Natural History Museum, London (NHM). An accelerating voltage of 20 kV, beam current of 2 nA, and a working distance of 10 mm were used. BSE images, SE images, and EDX microanalyses were obtained through an Oxford Instruments system, running version 16 INCA software. Before all compositional analyses, the X-ray microanalyzer was calibrated with the acquisition of a 50 s spectrum from a cobalt standard. Point spectra were then acquired from the preimpact projectiles and from five craters over 50 μm in diameter from each of the impact analogs. The atom% data for the major element Kα lines (S, Fe, and Ni) were then used for comparing precursor projectiles and resulting residues.
The FIB sections were produced using the FEI Nova600 NanoLab dual-beam FIB microscope at Lawrence Livermore National Laboratory (LLNL). They were then analyzed using the 200 keV FEI Tecnai T20 G2 field emission gun (FEG) monochromated scanning transmission electron microscope (STEM) at LLNL. Sections were imaged in bright field (BF) TEM and high angle annular dark field (HAADF) STEM. STEM-EDX atom% data for the major element Kα lines were obtained for projectile and residue bulk chemistry comparisons, using k-factors that were evaluated against thin film mineral standards. STEM-EDX maps provided detailed chemical data for the highly complex residues. Electron diffraction patterns were used to determine crystal structures where possible.
Results and discussion
Initial SEM Analyses: Preservation of Chemistry
The SEM-EDX data obtained from preimpact and postimpact pyrrhotite and pentlandite are plotted on separate ternary diagrams of S, Fe, and Ni in atom% (Fig. 3). The theoretical compositions of pyrrhotite and pentlandite in literature are given as Fe(1−x)S with 0 < x < 0.2 and (Fe,Ni)9S8, respectively (Deer et al. 1992). Our quantitative SEM-EDX analyses of the preimpact projectiles found that these particular samples had average compositions of Fe(1−x)S where x = 0.15 (pyrrhotite) and (Fe,Ni)1.04S where the Fe to Ni ratio is 0.92 (pentlandite). Lines marking constant Fe to S, Ni to Fe, and Ni to S ratios based on these average projectile compositions have been added to these ternaries.
The ternary diagrams (Fig. 3) show that pyrrhotite and pentlandite are easily distinguished as projectiles, and they have remained so as residues. The majority of the pentlandite residue data have maintained the original Fe to Ni ratio. However, these ternary diagrams echo that produced during the preliminary examination for the sulfide terminal grains in aerogel (Fig. 2 of Zolensky et al. 2006), with the majority of data for both minerals exhibiting varied degrees of loss of S. There are also a few data points that do not fit with these trends, being slightly enriched in Fe relative to Ni and vice versa (pentlandite), or slightly enriched in S relative to Fe and Ni (both pyrrhotite and pentlandite). These plots could highlight real changes in chemistry; however, there are several potential complications to the analyses that need to be taken into consideration. In a previous article, the ability of the SEM to investigate in situ the compositions of crater residues of a variety of silicate minerals were studied (Wozniakiewicz et al. 2009). Those analyses were performed without the foils being tilted toward the EDX detector, and were therefore subject to the effects of beam-sample incidence (as mentioned in the Sample Preparation section); however, several further important factors were highlighted as having potential to complicate the relative counts (and therefore atom%) detected. Those relevant to the present sulfide analyses are: secondary fluorescence (SF), target impurities, projectile impurities, and gun-derived debris (GDD).
Secondary fluorescence (SF), the phenomenon whereby the X-rays generated by one element in the sample can stimulate the production of lower energy X-rays from the same sample (see description in Goldstein et al. 1992), is expected to occur between the elements that compose these sulfides, with Fe Kα X-rays generating S Kα X-rays and, in the pentlandite, Ni Kα X-rays generating both Fe and S Kα X-rays. Quantitative correction routines assume the sample is homogeneous to the extent of beam penetration and attempt to account for SF. However, residues are thin layers, and if their thickness is less than that of the beam penetration depth, then SF between their elements is reduced compared to that of a large mass. As a result, the quantitative correction routines will be overcompensating so that, when compared against projectile analyses, both residues may appear to be enhanced in higher energy X-rays. Those data points where Fe is increased relative to S for pyrrhotite, and where Ni is increased relative to S and Fe for pentlandite, may represent locations of thin residue (where SF is reduced and thus overcompensated).
The target Al 1100 foil contains discrete, randomly distributed Fe- and Si-rich inclusions typically a few microns in size (Kearsley et al. 2006). If an impact occurred over Fe-rich inclusions, they could lie beneath a residue (within the beam penetration depth) or even have mixed with the residue, adding to the Fe signal of an analysis. The Fe-rich inclusions could have contributed to both mineral residues, producing some of the pyrrhotite data with apparent decreases in S and the few data points for pentlandite that are high in Fe relative to S and Ni.
The impurities observed in the preimpact samples were chalcopyrite (in pyrrhotite) and pyrrhotite (in pentlandite) (see Table 1). The lack of Cu signals in analyzed residue data suggests pyrrhotite impactors were pure. For the pentlandite, the low abundance of these impurities meant there was a low probability of their occurrence in impactors, but their influence cannot be ruled out for those outliers high in Fe relative to S and Ni.
GDD is an impurity introduced to the samples by the LGG after the impact has occurred and is composed of cartridge powder, sabot, previous projectiles, and targets, and material broken away from the inside of the gun itself (Wozniakiewicz et al. 2009). It results in a fine coating of C and a littering of larger particles which exhibit compositions that are C-rich or Fe- and Si-rich oxides, often with accompanying minor concentrations of Ca, Mn, Cu, and Cl. The fine C-coating is unavoidable; however, great care was taken to steer clear of the larger particles during analysis. Although no signs of Si, Ca, Mn, Cu, or Cl peaks were detected, we cannot be sure that the Fe oxide GDD were avoided in our analyses, and therefore these could be contributing to those data enhanced in Fe relative to S (and Ni).
The differences observed between these iron sulfide projectile and residue analyses could therefore be the result of multiple factors: real changes in chemistry, SF, target impurities, projectile impurities, and GDD. (We note, however, that only the first three of this list are applicable to real Stardust impacts). To further investigate and qualify these, we look to the TEM analyses of the next section.
TEM Analyses: Preservation of Mineralogy
The BF TEM images, diffraction patterns, and EDX spectra were obtained from the projectiles to confirm homogeneity, initial crystallinity, and composition. HAADF STEM images were obtained from residue sections, with differences in Z contrast between the resin, residue, and foil enabling identification of areas of residue. EDX spectra were obtained to compare chemistry against that of the original projectile and also used together with EDX maps to highlight compositional homogeneity/heterogeneity. Electron diffraction patterns were obtained where possible to compare crystallinity against that of the initial projectile. Together these data enabled a direct comparison of projectile and residue.
The Preimpact Projectiles
The BF TEM images and diffraction patterns from the projectile sections are displayed in Figs. 4 and 5. The diffraction patterns were indexed as pyrrhotite (2H hexagonal pyrrhotite) and pentlandite. STEM-EDX spectra obtained from the projectile sections gave an average composition of Fe1.02S for pyrrhotite and (Fe,Ni)1.35S (where the Fe to Ni ratio is 0.88) for pentlandite. These data are plotted on all ternary diagrams in Figs. 6B–D and 8B–F along with lines marking constant Fe to S, Ni to Fe, and Ni to S ratios based on the average projectile compositions to enable comparisons with residue data.
The TEM-EDX derived compositions appear depleted in S relative to those determined by SEM-EDX and to a lesser extent, also depleted in Fe relative to Ni (pentlandite). This depletion may be partially the result of trying to compare the results from two different analysis techniques which are used for different types of samples (volumes versus thin sections) and which consequently employ different corrections. However, there is also the possibility that there could be some real preferential loss of S during the FIB preparation process. The FIB is known to produce an amorphous layer on each side of the section (e.g., Kato et al. 1999; McCaffrey et al. 2001; Rubanov and Munroe 2004). We can envisage that a volatile element like S may be easily lost from these regions without the confining crystalline bonds. Even with the use of progressively lower currents during thinning and a final low keV polish at higher angles of incidence to remove some of the damaged material (as done in this work), some authors have found this amorphous layer to be approximately 10 nm in thickness for silicon samples (e.g., Delille et al. 2001; Gao et al. 2004; Alexandre et al. 2008). Such amorphous layers on our projectile FIB sections could represent on the order of 20% of the sample and therefore could certainly have contributed to the observed difference in S content for the TEM sections if loss occurs from these regions. The variation in S content observed in the projectile sections was used as an indication of the range of measured compositions attributable to unaltered samples. Therefore, the discussion below recognizes changes in composition when data lie beyond the range of projectile compositions.
The Postimpact Pyrrhotite Residue
The HAADF STEM image of the pyrrhotite residue section is shown in Fig. 6A. The atomic number (Z) contrast between the residue, resin and Al foil were sufficiently distinct to allow these components to be easily distinguished (see inset drawing of Fig. 6A). The pyrrhotite residue appears to be made up of four distinct components: fractured blocks, high Z spheres, low Z spheres, and fluffy material. To determine the nature of these components (composition, structure, genesis) electron diffraction patterns, STEM-EDX spectra, and STEM-EDX elemental abundance maps were obtained. The results of analyses of the EDX spectra are displayed in the ternary diagrams of Al, S, and Fe atom% in Figs. 6B–D, and the individual EDX elemental abundance maps are displayed in Fig. 7.
Fractured blocks: The bulk of the residue is in the form of blocky material that is bright in the HAADF STEM image against the resin and foil and has large fractures running throughout. Compositionally, the majority of this material matches the preimpact projectile, although some data points, taken from the edges of these blocks, plot toward enrichment in Al (Fig. 6B). The interfaces between these blocks and the foil exhibit textures reminiscent of melting, suggesting the presence of a rim of material that began to melt and to mix with the Al foil, thereby resulting in the varied Al content observed in such locations. The EDX element abundance maps (Fig. 7) also highlights this. Electron diffraction patterns taken from the main body of these blocks are well-indexed as pyrrhotite (2H hexagonal pyrrhotite) although there is some elongation of diffraction spots indicating shock.
High Z spheres: Some residue takes the form of bright spheres lying above the bulk residue, either completely separate from these blocks or apparently frozen in the process of leaving them. These spheres exhibit varying amounts of Fe, Al, and S (see EDX maps of Fig. 7), with S severely depleted relative to the original projectile (see Fig. 6C). These spheres are crystalline, however, none of the patterns generated could be indexed to the pyrrhotite cell or known equilibrium Fe-Al alloys (probably due to the minor S content). These features are not GDD: the only GDD that would contain S would be remnants of previously shot iron sulfide projectiles which would exhibit such chemistries and diffraction patterns and, due to being prepared by crushing, would be angular and not spherical. Instead we appear to have generated impact melts which have crystallized into Fe-Al (+S) alloy spheres.3
Low Z spheres: Within some of these bright spheres were dark spheres. From the EDX maps (Fig. 7), these spheres appear to be dominated in composition by S. No STEM-EDX spectra were obtained solely from these spheres because they are too small (smaller than the section thickness). However, a single spectrum obtained from an area of a high Z (bright) sphere containing a low Z (dark) sphere exhibited slightly higher S content when compared to those analyses of the high Z spheres alone (the cross in Fig. 6C), suggesting that the dark sphere composition is S-rich. As they are small and encased in the Fe-Al alloy spheres, any diffraction patterns they might generate were masked.
Fluffy material: Surrounding the bright spheres is an apparent “fluffy” material. This fluffy material appears to have compositions dominated by O and C with minor Al, S, and Fe (Figs. 6D and 7). The presence of C is not clear in the EDX abundance maps because the high C content of the resin has forced the contrast up, hiding its true extent in the fluffy material. Al, Fe, and S were found to be present in at least a few atom%, with Al exceeding 10 atom% in most cases. In general, this material maintained the Fe to S ratio of pyrrhotite (Fig. 6D). No clear diffraction patterns could be obtained. This material is probably a mixture of very fine (nanometer-scale, therefore unable to generate clear diffraction patterns) GDD, mixed with fragments of late impact ejecta with low energy rendering it unable to leave the crater.
In summary, the pyrrhotite residue appears to be made up of several components. The majority are in the form of fractured blocks which are pyrrhotite in composition and structure. The crystalline, Fe-Al alloy (high Z) spheres and S-rich (low Z) spheres are likely the result of melting, segregation, and loss of more volatile S from Fe followed by mixing with molten Al foil. The areas of blocks at the foil-residue boundary that exhibit a molten texture represent the material at the front of the projectile that impacted first. This material experienced melting and mixing with Al, but may have been unable to lose S due to confinement by incoming material from the rear of the projectile. The fractured blocks are pyrrhotite projectile material that experienced fragmentation but remained structurally and chemically intact after impact. The pyrrhotite projectile therefore appears to have been largely preserved after impact, but some areas have experienced melting which has resulted in loss of S and the production of new phases, some of which may have increased their Fe content by incorporating target Fe-rich inclusions. Finally, the surface of the pyrrhotite residue is also littered with GDD (fluffy material).
The Postimpact Pentlandite Residue
The HAADF STEM image of the pentlandite residue section is shown in Fig. 8A. The Z contrast between the residue, resin, and Al foil are sufficiently distinct to allow these components to be easily distinguished (see inset drawing of Fig. 8A). Like the pyrrhotite, the pentlandite residue is also complex, taking the form of five different components: high Z spheres and blocks, low Z spheres, fluffy material, solid boundary material, and diffuse boundary material. The nature of these components (composition, structure, genesis) was investigated by obtaining electron diffraction patterns, STEM-EDX spectra, and STEM-EDX elemental abundance maps. The results of the EDX spectra are displayed in the ternary diagrams of S, Fe, and Ni atom% in Figs. 8B–F and the individual EDX elemental abundance maps are displayed in Fig. 9.
High Z spheres and blocks: The majority of the residue is in the form of bright spheres and blocks. The STEM-EDX spectra (Fig. 8B) and maps (Fig. 9) show that these components are dominated by Fe, Al, and Ni with minor S, with compositions typically containing around 30 atom% of each of Fe and Ni, 40 atom% Al, and up to a few atom% S. Variation in the Al content is evident from feature to feature, although most maintain a close-to-original projectile Fe to Ni ratio after impact despite losing S (Fig. 8B). One sphere does, however, exhibit a very high Fe content, suggesting either varied partitioning of Fe and Ni, or the addition of Fe from an Fe-rich foil inclusion. These materials are all crystalline; however, none of the diffraction patterns obtained could be indexed to the pentlandite cell or any documented equilibrium Ni-Fe-Al alloy. This may be due to the minor S content. These diffraction patterns and Fe-, Al-, Ni-rich compositions suggest these are crystalline alloy spheres produced from impact-generated melts.3
Low Z spheres: The low Z spheres occur within high Z spheres and blocks. They are amorphous and typically enriched in S relative to Fe and Ni (see Figs. 8C and 9), with those closer to the original pentlandite Fe to S to Ni ratios being enriched in Al and O. The S-rich spheres are likely the result of capture of S released during impact, while the Al- and O-rich spheres share more affinity with the material of the fluffy masses (next section), possibly having been encased by the molten residue after impact, or simply appearing encased by the alloy in this particular two-dimensional section.
Fluffy material: This fluffy material appears to have compositions dominated by O with moderate C and Al and minor S, Fe, and Ni (see Figs. 8D and 9). The level of C- and Al- content is not represented well in the EDX abundance maps because the high C-content of the resin and high Al-content of the foil increased the contrast, hiding their true abundance in the fluffy material. Ni, Fe, and S are present at levels of at least a few atom% and, although the fluffy material maintains a close-to-original pentlandite Ni to S ratio, it is enriched in Fe (Fig. 8D). This increased Fe content could indicate a GDD source for this material (which may not have caused a significant increase in Fe for the pyrrhotite fluffy masses as the level of contaminant within GDD changes with each shot). No clear diffraction patterns could be obtained. Like the fluffy material in the pyrrhotite residue section, this is probably a mixture of very fine GDD and fragments of late impact ejecta with low energy rendering it unable to leave the crater.
Solid boundary material and diffuse boundary material: In the pentlandite residue section, the exact boundary between the residue and foil is not as clear as it is in the pyrrhotite residue. Instead, the boundary appears to be characterized by the remaining two components: a solid band of bright material and a diffuse band of bright material. The solid boundary material appears to be dominated by Al, Fe, Ni, and, in one region, by S (Fig. 9). STEM-EDX spectra show that the areas low in S are similar in composition to the high Z spheres and blocks, with Fe to Ni ratios close to the original pentlandite projectile (Fig. 8F). The region with more abundant S appears to have a bubbly texture, with dark areas possibly being minute, low-Z, S-bearing spheres. This bubbly material plots close to the original pentlandite S to Fe to Ni ratio (Fig. 8F). The diffuse boundary material appears to be dominated by Al with lesser amounts of Fe, Ni, and S. This diffuse material has largely maintained the Fe to Ni to S ratio of the original pentlandite projectile (Fig. 8E). It appears that the pentlandite may have melted, but rapidly combined with Al by the force of the impact, forming a rapidly quenched “solid solution,” thereby not allowing the segregation and loss of S to occur. Although diffraction spots indicated the presence of crystalline material in both solid and diffuse components, no clear single crystal patterns were obtained.
In summary, the pentlandite residue, like the pyrrhotite residue, appears to be made up of several components. Unlike the pyrrhotite residue, however, none of these components matched the pentlandite residue in structure and chemistry––all components appeared to be the result of melting, segregation, and loss of more volatile S, followed by mixing with molten Al foil. The majority of the residue is in the form of high Z spheres and blocks which are crystalline and composed of various Fe-Ni-Al alloys. Within these are amorphous S-rich spheres, captured during solidification of the alloy spheres and blocks. Both alloy spheres and S-rich spheres were also identified in the pyrrhotite residue. The solid boundary material appears to have similar compositions to the alloy spheres and blocks unless bubbles are present in which case major element ratios closer to the original pentlandite are observed. Like the material of the fractured block boundaries in the pyrrhotite residue, the diffuse boundary material represents the projectile material that impacted first and which, after melting, was forced into the Al foil by the rear of the projectile. This has resulted in the S remaining in the melt, giving an Fe to S to Ni ratio similar to that of pentlandite. Finally, the surface of the pentlandite residue is littered with GDD (fluffy material). The pentlandite projectile was therefore completely altered in the impact, melting, segregating, and mixing with molten Al foil, and resulting in the production of new phases. The majority of residue components do, however, maintain close to original Ni to Fe ratio and those that do not are likely to be the result of mixing with Fe-rich target inclusions (ignoring the fluffy materials which are likely GDD and therefore likely deposited after impact has occurred). Previous work on Stardust Al foils has shown the Fe-rich target inclusions to be abundant, variable in size, and random in distribution. Therefore, the occurrence of this impact over an inclusion comes as no surprise as craters of this size have a high probability of interacting with several inclusions (cf. the number of inclusions visible within a 50 μm diameter circle on fig. 4 in Kearsley et al. 2006 and Fig. 3 in Wozniakiewicz et al. 2009). It is interesting, however, that these results suggest that the degree of mixing with Fe-rich target inclusions is not uniform throughout a single crater residue and therefore likely depends on the size and number of inclusions impacted. It is clear that these inclusions are a potential obstacle to deciphering all original iron sulfide compositions.
Calculating Impact Parameters
To complement and clarify the results of the TEM analyses rough estimates of the impact parameters, peak pressure (Ppeak), peak postshock temperature (Tpps), and cooling time were calculated. These calculations require the input of experimental Hugoniot data for the projectile and target and various material properties of the projectile (e.g., heat capacity, thermal conductivity). The available data are limited for minerals, with the required experiments being biased toward materials of industrial and military importance. As a result, these calculations were only performed for the pyrrhotite shots into Al 1100 foil since adequate data were not available for pentlandite.
Peak pressure (Ppeak): To determine Ppeak, the graphical method outlined in Melosh (1989) was used, whereby a plot of pressure (P) against particle velocity (u) is constructed from available Hugoniot data for the Al target (plotted forward from u =0 at P =0) and the pyrrhotite projectile (plotted backward from u = the impact velocity, 5.85 km s−1, at P =0). The point at which the two curves intersect corresponds to the Ppeak. Hugoniot data were obtained from Marsh (1980) for the Al 1100 target and from Ahrens (1979) and Brown et al. (1984) for pyrrhotite. The Ppeak calculated for this impact is approximately 85 GPa (to the nearest GPa). This value falls within the ranges estimated in previous publications for impacts at this speed (Melosh 1989; Burchell and Kearsley 2009).
Peak postshock temperature (Tpps): The Tpps is the temperature of the decompressed shocked material that results from waste heat deposited after shock and release. We estimated this using the method described by Artemieva and Ivanov (2004) and Fritz et al. (2005). The peak postshock temperature (Tpps) is expressed as:
where T0 is the preshock temperature of the material (300 K), umax is the maximum particle velocity experienced in the projectile (in m s−1), ER is the energy lost from the projectile during release from high pressure (in J) and Cp is the specific heat capacity of the projectile (in J kg−1 K−1). umax was determined from the plot of u against P to be 2590 m s−1. The value of Cp used for pyrrhotite was 50.6 J mol−1 K−1 or 625 J kg−1 K−1 (derived from Grønvold and Stølen (1992) for Fe0.875S). ER can be approximated by assuming that release occurs along the Hugoniot curve and that the Hugoniot curve is adequately represented by a linear shock-particle velocity relationship:
where U is the shock velocity and c and s are constants. In this case,
(Artemieva and Ivanov 2004; Fritz et al. 2005), where Umax is the maximum shock velocity. Plotting values of U against u, taken from the Hugoniot data sources referenced previously, the c and s values derived for pyrrhotite are 3070 m s−1 and 1.5, respectively. Inserting these and u = umax into Equation 2 gives a Umax of 6955 m s−1 and therefore an ER of 1.90 × 106 J.
Inserting these values of umax, Cp and ER into Equation 1 gives a Tpps of 2600 K (to the nearest 100 K).
Cooling time: A crude estimate of characteristic cooling time can be made by simplifying the impact geometry to one dimension (1D), ignoring the effects of the surrounding foil and assuming the only method of heat transport is by conduction. In this case temperature decays exponentially, with a characteristic decay time τ defined as:
where L is the characteristic length scale of the cooling object and α is the thermal diffusivity (Turcotte and Schubert 2002), which can be expressed as α = k/ρCp, where k is the thermal conductivity (W m−1 K−1) and ρ is the density (kg m−3). For the pyrrhotite crater residues, L (thickness of residue) is assumed to be 5 μm (based on the images shown in Fig. 6), k is 3.53 W m−1 K−1 (Clauser and Huenges 1995), and ρ is 4710 kg m−3 (average for pyrrhotite from Deer et al. 1992). This gives a decay time τ of 10−5 s. We emphasize that this calculation is severely limited by the fact that the material properties applied are for solid pyrrhotite, whereas we expect at least some of this material to be molten pyrrhotite, and by simplifying the assumption of 1D conductive cooling. However, as other cooling mechanisms will reduce cooling time, this estimate serves as a useful upper bound on cooling time.
The calculated impact parameters confirm that the uncompressed melting temperature of pyrrhotite, approximately 1373–1473 K (Vaughan and Craig 1978), was significantly exceeded in those areas that experienced the Ppeak and accounts for the melt-produced phases identified in the pyrrhotite residue TEM section. However, the peak shock pressures in a finite projectile are not experienced throughout the whole body, and the rear of the projectile is likely to be less heavily shocked resulting in the survival of some crystalline material. In addition, it appears that the melting, segregation of S, mixing with Al, and crystallization exhibited by the pyrrhotite (and pentlandite) residues were not inhibited by the short cooling time (approximately 10−5 s). Since melting of pyrrhotite results in production of new phases, and since the short cooling time has clearly not inhibited the segregation of pyrrhotite components, the fragmented pyrrhotite blocks identified in the pyrrhotite residue must be original preserved projectile. As previously noted, the data required for these impact parameter calculations are not available for pentlandite; however, assuming that pentlandite has a similar Tpps as that determined for pyrrhotite, the more extensive melting exhibited could be a result of the fact that this material has a lower uncompressed melting temperature of approximately 1135 K (Kullerud 1963). It is, however, possible that the slightly higher impact velocity for the TEM-analyzed pentlandite foil (6.21 km s−1) may also have contributed to this difference.
This work has used a combination of SEM and TEM to investigate the degree of preservation in the residues of two iron sulfides, pyrrhotite and pentlandite. The SEM results highlight a possible change in chemistry, but used alone, SEM could not distinguish and remove the contribution of SF, GDD, and target and projectile impurities in these samples. TEM, however, was able to show that the structures and compositions exhibited by the pyrrhotite and pentlandite residues suggest a very complex reaction to the impact process. Both of these iron sulfide residues contain several impact-generated components that clearly indicate that the projectile has undergone heating, melting, segregation, and some loss of S from Fe (and Ni) and mixing with molten Al. On cooling, the new alloy components have crystallized and, in areas, have captured exsolved S. The process of melting, segregation of S, mixing with Al, and crystallization was therefore clearly not inhibited by the short cooling time (approximately 10−5 s). The pyrrhotite section reveals that some of the projectile remains compositionally and structurally intact, albeit with some evidence of shock. The pentlandite impact has resulted in the complete destruction of the original projectile in the examined section, although the majority of its residue components retain roughly the original Ni to Fe ratio. Those with higher Fe content than the original projectile (ignoring “fluffy material”) are likely the result of mixing with Fe-rich foil inclusions. The “fluffy material” identified in both residues likely corresponds to GDD, and likely arrived after impact and therefore did not contribute to the melt. (This material is also not applicable to real Stardust impacts.) The foils have therefore been unable to collect iron sulfides without at least some alteration occurring, but unequivocal identification of the original preserved iron sulfides may be possible by TEM through the location of Al-free areas of iron sulfide (if they exist).
With this information, we see that at least some of the S depletion indicated by the SEM-EDX analyses represents real loss of this element (as opposed to artifacts of sample and analysis effects). In addition, those few data points that exhibit enhanced S relative to the original projectile can be explained as arising from an area containing trapped S (low Z spheres) while those richer in Fe could represent mixing with Fe-rich inclusions, or analyses including Fe-rich GDD. We do, however, note that none of the factors that could cause the observed variations in SEM-EDX derived residue chemistries can be quantified and corrected for. As a result, the original compositions of iron sulfide impactors cannot be determined from SEM-EDX analyses of their residues. The “first impression” that the SEM does provide, along with a knowledge of alteration experienced by different minerals nevertheless allows informed decisions to be made about how best to further investigate a crater residue. In the majority of cases the Ni to Fe ratios have remained identical to that of the original impactor. Therefore, these elements may also be used to distinguish these minerals (in both SEM and TEM) and determine whether comet 81P/Wild 2 differs from CP IDPs in its iron sulfide content (although Fe-rich foil inclusions and GDD need to be borne in mind).
It is clear that to comprehensively determine the state of preservation and define the products of impact for Stardust crater residues, this combined SEM and TEM investigation should be extended to include all likely minerals that may have impacted the Stardust aluminum foil collector surfaces as well as appropriate analogs of amorphous materials. In addition, more complicated impactors with varied internal structure (containing fractures, pores, grain boundaries) and gross morphology (aggregates) should be considered as we note that while craters arising from mono-mineralic cometary dust grains have been identified on Stardust foils (e.g., Leroux et al. 2008), most craters appear to be the result of polymineralic cometary dust grains. Mixed composition components in such an aggregate impactor may effectively be able to contaminate each other if melting occurs and therefore it is vital to determine the effects of impact specific to this more complex type of projectile. The Stardust sample set is unique and unlikely to be repeated soon. This, combined with its proven ability to capture and to preserve silicate residue, in some cases better than aerogel, means that the foil-collected material is exceptionably valuable, and it is vital that all efforts be made to enable its successful interpretation.
Finally, we have also highlighted the possibility that some real preferential loss of volatile S may have occurred during the FIB preparation process, with S being lost from the thin surface amorphous layers that are produced by the ion milling process. This should therefore be borne in mind when performing TEM-EDX analyses on such volatile-bearing materials prepared in this way. (We find that nonvolatile bearing minerals like silicates show no such effects; Wozniakiewicz et al. unpublished data.) However, with improvements in ion beam technologies, specifically the ability to perform even lower keV final polishes, we will see enhanced removal of this amorphous material and a reduction in the extent of these damaged (potentially volatile depleted) regions. This should result in any possible contribution they have toward chemical analyses becoming negligible.
This assumes that all craters in the cometary collector originate from impacting comet 81P/Wild 2 grains; however, it should be noted that this may not be the case. Analyses of aerogel track trajectories by Westphal et al. (2008) suggest that although 3/5 of the tracks they analyzed were consistent with a comet 81P/Wild 2 origin (tracks with normal incidence) the remainder exhibit oblique incidence. The oblique incidence tracks observed are consistent with being secondary ejecta (and therefore “preprocessed” comet 81P/Wild 2 grains) from impacts with various components of the spacecraft (Westphal et al. 2008); however, it is also possible that some tracks might be the result of IDP impacts.
Ishitani and Kaga (1995) calculate the local temperature rise experienced by three different materials (Si, GaAs, and SiO2) during FIB sample preparation utilizing a gallium ion source with an accelerating voltage of 30 keV (as used here). For a beam current of 15 nA, the maximum increase in temperature for Si, GaAs, and SiO2 was 2, 8, and 230 K, respectively. However, the maximum temperature increase drops off significantly with decreased current so that at the highest polishing current we use (1 nA when the section is still >1 μm in thickness) the maximum temperatures predicted by Ishitani and Kaga drop to approximately 0 K for Si, approximately 1 K for GaAs, and approximately 70 K for SiO2. As the sample is thinned we drop currents even further so that our final 30 keV polish (conducted when the section thickness is approximately 150 nm) uses a current of 100 pA which corresponds to heating of <20 K for SiO2 and approximately 0 K for GaAs. The increase in temperature experienced by a material is dependent on the thermal conductivity and diffusivity of the material (materials with lower thermal conductivity and diffusivity experience higher levels of heating). The thermal conductivity and diffusivity for the sulfides (see Tsatis and Theodossiou 1982) lies above that of SiO2 but below that of GaAs and therefore the degree of heating experienced by them will lie somewhere between those values. We can therefore be confident that heating experienced during FIB preparation is negligible for the sulfides.
Although quasicrystals have been reported for quenched Al-Fe and Al-Fe-Ni systems (e.g., Tanaka et al. 1993; Grushko and Velikanova 2004), to date we have found no evidence of their presence in the pyrrhotite or pentlandite residues. We also note that the compositions of our Al-Fe-Ni alloys do not fall in the range of those reported in quasicrystals, so quasicrystals may not be expected.
Acknowledgments–– We thank NASA for providing Al foils, STFC for support of the LGG and PPARC grant funding a Ph.D. studentship for P. J. W. (grant ref. PPA/S/S/2005/04118). We also thank G. Flynn, M. Zolensky, and D. Brownlee for their valuable comments and suggestions during review. G. S. C. was funded by NERC grant NE/E013589/1. Parts of this work were performed under the auspices of the U.S. DOE by LLNL under Contract DE-AC52-07NA27344. This work was supported by grants: NASA NNH07AG46I to H. A. I. & LDRD 09-ERI-004 to J. P. B.