Space weathered rims found on the surfaces of the Itokawa dust particles

Authors


Abstract

On the basis of observations using Cs-corrected STEM, we identified three types of surface modification probably formed by space weathering on the surfaces of Itokawa particles. They are (1) redeposition rims (2–3 nm), (2) composite rims (30–60 nm), and (3) composite vesicular rims (60–80 nm). These rims are characterized by a combination of three zones. Zone I occupies the outermost part of the surface modification, which contains elements that are not included in the unchanged substrate minerals, suggesting that this zone is composed of sputter deposits and/or impact vapor deposits originating from the surrounding minerals. Redeposition rims are composed only of Zone I and directly attaches to the unchanged minerals (Zone III). Zone I of composite and composite vesicular rims often contains nanophase (Fe,Mg)S. The composite rims and the composite vesicular rims have a two-layered structure: a combination of Zone I and Zone II, below which Zone III exists. Zone II is the partially amorphized zone. Zone II of ferromagnesian silicates contains abundant nanophase Fe. Radiation-induced segregation and in situ reduction are the most plausible mechanisms to form nanophase Fe in Zone II. Their lattice fringes indicate that they contain metallic iron, which probably causes the reddening of the reflectance spectra of Itokawa. Zone II of the composite vesicular rims contains vesicles. The vesicles in Zone II were probably formed by segregation of solar wind He implanted in this zone. The textures strongly suggest that solar wind irradiation damage and implantation are the major causes of surface modification and space weathering on Itokawa.

Introduction

Surfaces of airless bodies exposed to interplanetary space have their structures, optical properties, chemical compositions, and mineralogy gradually changed by solar wind irradiation, implantation, and sputtering, irradiation by galactic and solar cosmic rays, and micrometeorite bombardment. These alteration processes and the resultant optical changes are collectively known as space weathering (e.g., Hapke 2001; Clark et al. 2002; Chapman 2004), and one of the major problems in comparing meteorite studies with asteroid observations. Space weathering effects are also relevant to resource utilization on the Moon, for example, because of the possibility that they are a source of H2 and -OH (McCord et al. 2011).

Our knowledge of space weathering has depended almost entirely upon studies of the surface materials returned from the Moon and regolith breccia meteorites (Hapke et al. 1975; Keller and McKay 1997; Pieters et al. 2000; Hapke 2001; Taylor et al. 2001; Noble et al. 2005) until the surface material of the asteroid Itokawa was returned to the Earth by the Hayabusa spacecraft (Noguchi et al. 2011). Lunar studies show that space weathering darkens the albedo of lunar regolith, reddens the slopes of their reflectance spectra, and attenuates the characteristic absorption bands of their reflectance spectra (Hapke 2001; Clark et al. 2002; Chapman 2004). These changes are caused by vapor deposition of small (<40 nm, typically 2–3 nm in diameter) nanophase metallic Fe (npFe0) within the grain rims of lunar regolith and agglutinates (Keller and McKay 1997; Pieters et al. 2000; Noble et al. 2005).

The initial analysis of the Itokawa dust particles revealed that the particles are composed of minerals that are practically the same as minerals comprising LL5-6 chondrites (Nakamura et al. 2011a, 2012) and that the Δ17O values of olivine, low-Ca pyroxene, and plagioclase are very similar to those of LL chondrites, although δ18O values are highly variable (Yurimoto et al. 2011; Nakamura et al. 2012; Nakashima et al. 2013). As the interiors of the Itokawa particles are very similar to those of LL5-6, the interior material should produce reflectance spectra similar to those of LL chondrites. Because it is expected that thin grain coatings on semitransparent regolith materials greatly change the reflectance spectra of S-type asteroids (e.g., Chapman 2004), Itokawa particles must have thin surface layers that modify the reflectance spectra. Because no impactors were successfully fired by the Hayabusa spacecraft sampling mechanism (Yano et al. 2006), the collected samples are unlikely to have been significantly fragmented during sampling and are thus likely to retain the outermost layer of surfaces. Therefore, the harvested Itokawa particles are ideal for investigation of asteroidal space weathering. Noguchi et al. (2011) showed that 5 of 10 investigated Itokawa particles have nanoparticle-bearing rims, which are probably the major cause of modification in the reflectance spectra of Itokawa. The structure of the nanoparticle-bearing rims suggests that radiation damage, sputtering, and ion implantation by solar wind irradiation are major agents of space weathering on Itokawa. Noguchi et al. (2012) found two additional types of surface modification on the surface of the particles. Here, we report the detailed results of the initial analysis of space weathering of the Itokawa dust particles performed in 2011.

Samples and Methods

We investigated 12 Itokawa particles from Room A of the sample catcher of the Hayabusa sample container, the samples of which were collected on 25 November 2005. Their average diameter varies from 26 to 88 μm (mean value 53 μm; Table 1). We investigated Itokawa particles both preserved in N2 purge environment (four particles) and exposed to the Earth's atmosphere (nine particles) (Table 1). In the first procedure, we tried to prevent contact with atmospheric oxygen and water vapor from acting on the Itokawa particles during the entire process of embedding, ultramicrotomy, and transportation to Hitachi High-Technologies Corporation where our STEM observation was performed.

Table 1. Itokawa dust particles investigated in this study
Sample IDSample preparationaAverage diameter (μm)Type of space weathering rimbUltramicrotomy under a nitrogen purgec
  1. a

    FIB = FIB lift-out; UM = ultramicrotomy.

  2. b

    RR = redeposition rim; CR = composite rim; CVR = composite vesicular rim.

  3. c

    Dehydrated ethylene glycol was used as trough liquid instead of pure water.

RA-QD02-0009FIB70CVRNo
RA-QD02-0022UM46RRYes
RA-QD02-0032UM and FIB38CRNo
RA-QD02-0033UM50CVRNo
RA-QD02-0034UM41RRNo
RA-QD02-0035UM60CRYes
RA-QD02-0041UM75CVRNo
RA-QD02-0042UM88CRNo
RA-QD02-0050UM52RRNo
RA-QD02-0051UM49RRYes
RA-QD02-0054UM26CRNo
RA-QD02-0056UM46CRYes

The other Itokawa samples were exposed to the Earth's atmosphere. Total exposure time to the atmosphere was less than 6 h for each of six particles, and less than 10 h for RA-QD02-0009. RA-QD02-0033 was kept in a regular desiccator for about 1 month. These eight particles were investigated at the synchrotron radiation facility of High Energy Accelerator Research Organization (KEK-PF) (Nakamura et al. 2011a) and the Spring-8 synchrotron facility (Tsuchiyama et al. 2011) before our investigation. In addition, the surface morphology of RA-QD02-0033 was observed by field emission scanning electron microscopy (FE-SEM) at Osaka University (Tsuchiyama et al. Forthcoming). For the samples except for RA-QD02-0009, ultramicrotomy was applied to make ultra-thin sections. Ultra-thin sections prepared in the Earth's atmosphere were promptly introduced into a vacuum desiccator after ultramicrotomy. The ultra-thin sections were approximately 100 nm thick. Ultramicrotomy was performed using a Reichert Ultracut-N ultramicrotome at Ibaraki University. Two ultra-thin (approximately 100 nm) samples were prepared by focused ion beam (FIB) system Hitachi FB-2100 at Hitachi High-Technologies Corporation. Because the FIB system is equipped with a microsampling manipulator, a target sample with few tens of micrometers in size can be extracted with a high positional accuracy (FIB lift-out technique). After the target sample was attached to a TEM grid, it was thinned to approximately 100 nm by using a 30 kV Ga+ ion beam.

Spherical aberration-corrected scanning transmission electron microscopy (Cs-corrected STEM) is suitable to locate approximately 1-nm-sized nanophase Fe because this type of STEM can obtain high-resolution (<0.1 nm resolution) high-angle annular dark-field (HAADF)-STEM images, in which the nanoparticles appear as bright spots (Z-contrast imaging). We utilized a Cs-corrected STEM equipped with a cold FEG, Hitachi HD-2700 at Hitachi High-Technologies Corporation. EDAX Genesis EDS equipped on the Hitachi HD-2700 STEM was used for analysis of chemical compositions of constituent materials in the ultra-thin sections. To analyze compositional variations near the surfaces of the Itokawa particles, we utilized long and thin raster beams (20 nm × 200–500 nm) that were parallel to the edges of the sections to decrease compositional changes during analysis. Semiquantitative analysis was based on the Cliff-Lorimer thin film approximation. Experimental k-factors were obtained using unweathered minerals in the sections, although k-factors of minor elements such as Cl and K were calculated by linear interpolation between the two measured k-factors of major elements. To search for solar flare tracks, a conventional FE-TEM Hitachi HF-2000 was used at Hitachi High-Tech Manufacturing and Service Corporation.

The morphologies of the Itokawa particles were observed using a Hitachi S-4300SE/N FE-SEM at the curation facility of JAXA. Observation of the surface of RA-QD02-0033 and its potted butt was performed using field emission-scanning electron microscopy (FE-SEM). The surface was observed by a JEOL JEM-7001 FE-SEM at Osaka University operated under 2 kV accelerating voltage and the potted butt was observed using a Hitachi S-4300 FE-SEM at Ibaraki University operated under 3 kV accelerating voltage. All the potted butts remaining after ultramicrotomy were observed using a JEOL JSM-5400LV SEM at Ibaraki University. A manually polished section of RA-QD02-0009 was observed using a JSM-7000F FE-SEM at Kyushu University. More information about sample preparation and analysis of STEM images is available in Appendix A.

Results

Surface and Interior Textures of the Itokawa Particles Investigated

Figure 1 shows backscattered electron (BSE) images of all the Itokawa particles investigated in this study. The Itokawa particles in Fig. 1 are arranged according to three different textural types of their rims: those in the left, middle, and right columns have redeposition rims, composite rims, and composite vesicular rims, respectively. Descriptions of these rims are presented in the following sections. It is difficult to predict which type of rim an Itokawa particle has based only on its surface morphology. Among these particles, only RA-QD02-0033 (Fig. 1L) has a unique lumpy surface. Its unique morphology is described and discussed elsewhere (Tsuchiyama et al. Forthcoming). Among the Itokawa particles not investigated here, RA-QD02-0013 has a similar feature to this particle (Fig. 1C in Nakamura et al. 2011a).

Figure 1.

Backscattered electron (BSE) images of the 12 Itokawa dust particles investigated in this study. Scale bars are 20 μm long. The numeral in the upper right corner of each BSE image represents the last four letters of each sample number. For example, 0022 means RA-QD02-0022 in full. The numerals in the following figures express the same thing.

Figure 2 shows BSE images of the cross sections of the Itokawa particles shown in Fig. 1. These images were taken from the surfaces of potted butts after ultramicrotomy, except for Fig. 2J, which was taken from the surface of a manually polished section. Scratches (linear features) shown on the images, except for Fig. 2J, and dimples that appear as dark spots in Figs 2A, 2B, 2F, and 2L are artifacts during ultramicrotomy. These features are common when hard materials are ultramicrotomed. Because no rims were visible in these low-magnification images, rim thickness could obviously not be measured. FIB sample preparation was conducted to RA-QD02-0009 and 0032. The positions where FIB sections were lifted out are indicated by dotted lines. The FIB lift-out sections of the external surfaces of RA-QD02-0032 and -0009 existed at the bottom of these samples (Figs. 2F and 2J).

Figure 2.

Backscattered electron images of the cross sections of the 12 Itokawa dust particles investigated in this study. Scale bars are 20 μm long. Dotted lines in Figs. 2F and 2J indicate the positions where FIB lift-out sections were prepared. Lineations on the BSE images except for 2J are scratches resulting from ultramicrotomy. Ol, olivine; LPx, low-Ca pyroxene; Tr, troilite; Pl, plagioclase; HPx, high-Ca pyroxene; Cl-ap, chlorapatite.

We identified three types of particle surface modifications probably formed by space weathering: redeposition rims, composite rims, and composite vesicular rims (Fig. 3). The basic structures of the external surface of the Itokawa particles can be described by a combination of three zones: Zone I (redeposition zone), Zone II (partially amorphized zone), and Zone III (unchanged mineral) (Table 2). The redeposition rims are composed only of Zone I, which is directly adjacent to Zone III. The composite rims and the composite vesicular rims consist of Zone I and Zone II, below which Zone III exists. The lunar samples show no such simple systematics, having more varied and complicated rims including amorphous rims (amorphized by solar wind irradiation), inclusion-rich rims (vapor redeposit including abundant nanophase Fe), combinations of these two rim types, and vesicular rims (Keller and McKay 1997; Noble et al. 2005). In this article, we follow the usage of the lunar science, and use the term nanophase (Fe,Mg)S to describe nanoparticles rich in Fe, S, and Mg, and the term nanophase Fe to describe nanoparticles rich in Fe, most of which are metallic iron. We use the abbreviation np(Fe,Mg)S for nanophase (Fe,Mg)S and the abbreviation npFe for nanophase Fe. We use the term nanophase Fe0 (npFe0) when npFe was identified as metallic Fe. In the following sections, the characteristics of each Itokawa particle rim are described.

Table 2. Observed combinations of zones in the space weathered rims on Itokawa dust particles investigated in this study
Types of rimsRedeposition rimsComposite rimsComposite vesicular rims
ZonesNanophases and vesiclesOlivine, low-Ca pyroxenePlagioclaseOlivine, low-Ca pyroxenePlagioclaseOlivine, low-Ca pyroxene
  1. a

    Zone I often lacks nanophase (Fe,Mg).

Zone I (redeposition zone)NoYesYesYesaYesaYesa
Nanophase (Fe,Mg)SYesYesYes
Zone II (partially amorphous zone)NoYes
Nanophase FeYes
Nanophase Fe and vesiclesYes
Zone III (crystalline substrate zone)YesYesYesYesYes
Figure 3.

Three typical surface features observed on ferromagnesian silicates in the Itokawa particles investigated in this study and corresponding schematic diagrams with interpretation of their textures. np(Fe,Mg)S, nanophase (Fe,Mg)S; npFe, nanophase Fe.

Redeposition Rim

Noguchi et al. (2011) reported that 5 of 10 Itokawa particles do not show any evidence of space weathering; in other words, they do not have rims containing np(Fe,Mg)S and/or npFe. We re-examined the ultra-thin sections of these particles that initially did not have rims containing these nanophases by STEM. Although remarkable textural changes could not be detected on the surface of the particles at relatively low magnification (Figs. 4A and 4B) except for RA-QD02-0056, high-magnification BF and HAADF-STEM images revealed that there was a quite thin (2–3 nm) amorphous rim (indicated by parallel short lines) when viewed edge-on (Figs. 4C and 4D). Lattice fringes of the olivine substrate can be observed just below the rims (Fig. 4C). EDX spectrum of the thin rim indicates that the rims contain elements that are not included in the olivine substrate (Figs. 4E and 4F); they are enriched in Si and contain Na, Al, Cl, K, and Ca, as well as Mg and Fe that are present in the olivine substrate. It is obvious that the thin rims are a mixture of elements derived from both the substrate mineral and the surrounding minerals. Therefore, we call the rims redeposition rims. Although the redeposition rims do not include npFe0, which modifies the shapes of the reflectance spectra of fresh materials, the rims are obviously products formed during exposure to interplanetary space. Therefore, in this article we treat them as a type of space weathering product. RA-QD02-0056 has the composite rim, which is described in the next section.

Figure 4.

Redeposition rims on Itokawa dust particles. A) Bright field- (BF-) and B) High-angle annular dark field- (HAADF-) STEM images of an edge-on cross section at the surface of olivine in the Itokawa particle RA-QD02-0050. C) and D) Enlarged BF- and HAADF-STEM images of a redeposition rim. There is an approximately 3-nm-thick redeposition rim (indicated by two parallel short lines) on an unchanged crystal. Two numbered box areas indicate the positions of EDX analysis. E) and F) EDX spectra of the redeposition rim (boxed area No. 1 in 3D) and the interior (olivine: boxed area No. 2 in 3D), respectively.

Composite Rims

Texture of the Composite Rims

Space weathering rims containing nanoparticles were reported in Noguchi et al. (2011). Here, we call them composite rims. The structure of the composite rims is different between ferromagnesian silicates (olivine and low-Ca pyroxene) and plagioclase (Fig. 5). Both ferromagnesian silicates and plagioclase have a thin (<15 nm) layer composed of amorphous silicate often containing a densely arranged np(Fe,Mg)S, which often appears as a continuum, at the top of the rims (Noguchi et al. 2011). In addition to S, the thin layer contains elements that are not included in the substrate minerals, as described later in this section. Irrespective of the presence or absence of np(Fe,Mg)S, this feature is similar to the redeposition rims. We call this surface layer Zone I (redeposition zone) (Fig. 3; Table 2) (Noguchi et al. 2011). In this respect, the redeposition rim is composed only of the redeposition zone (Zone I).

Figure 5.

BF- and HAADF-STEM images of composite rims on olivine in RA-QD02-0041, low-Ca pyroxene in RA-QD02-0042, and plagioclase in RA-QD02-0042, respectively. The surfaces are edge-on. Roman numerals in the images represent three zones near the surface. Zone I is characterized by np(Fe,Mg)S embedded in amorphous material. Zone II is characterized by partially amorphized crystals. There are abundant npFe particles in Zone II of ferromagnesian silicates. Zone III is unchanged mineral.

Below Zone I, there is a partially amorphized zone (Zone II), where original crystal structure is preserved locally. The texture of Zone II is different between ferromagnesian silicates and plagioclase (Noguchi et al. 2011). Zone II of ferromagnesian silicates contains npFe (Figs. 5A–D). Figure 6 clearly shows the structure of Zone II of olivine in RA-QD02-0032. NpFe that appears as bright spots intimately coexists with dark patches in HAADF-STEM images (Fig. 6B). The dark appearance of the patches suggests that they are depleted in heavy elements (Fe in this case) from the host olivine. Because the npFe and the patches are embedded in crystalline olivine showing clear lattice fringes (Fig. 6A), we suggest that they were formed within the crystalline olivine. The total thickness of the composite rims on ferromagnesian silicates is less than approximately 60 nm. Below Zone I of plagioclase, there is a partially amorphized area (Zone II) without npFe (Figs. 5E and 5F). Identification of the boundary between Zone II and unchanged substrate plagioclase (Zone III: unchanged mineral) is difficult because plagioclase is susceptible to electron beam damage from the Cs-corrected STEM.

Figure 6.

Edge-on BF- and HAADF-STEM images of composite rims on olivine in RA-QD02-0032. These images clearly show that Zone II is partially amorphized as shown in 5A and that it contains abundant npFe (bright spots in 5B).

Noguchi et al. (2011) did not observe thin sections prepared by FIB lift-out sections. Thus, we prepared nine FIB lift-out sections from six Itokawa particles and investigated the mineralogy of these samples (Nakamura et al. 2011a). In this study, we re-examined the FIB lift-outs and found that lift-out sections of RA-QD02-0032 and RA-QD02-0009 do, indeed, exhibit the composite rims. Because their exterior surfaces were located at the bottom of the sample embedded in epoxy resin and because the surfaces faced to the opposite direction of the incident Ga+ ion beam of the FIB, it is unlikely that their surface texture was modified considerably during FIB processing. Because both ultra-thin sections and FIB lift-out sections were prepared for RA-QD02-0032, we compared the texture of the composite rims prepared by both methods (Fig. 7). Composite rims shown in Fig. 7 are composed of a thin (approximately 5 nm thick) Zone I that does not contain np(Fe,Mg)S and npFe in Zone II (approximately 55 nm thick) directly beneath Zone I. In Fig. 7A, because the section was viewed edge-on, the exterior surface of the sample is not observed. On the other hand, the exterior surface of the sample appears as a slope in Fig. 7B because the section is not viewed edge-on. By considering the difference in configuration of the exterior surface to the incident electron beam between these two figures, no remarkable difference is apparent. In both images, Zone II contains dark irregular patches elongated parallel to the surface.

Figure 7.

Comparison of HAADF-STEM images of composite rims on olivine in RA-QD02-0032 prepared by A) ultramicrotomy and B) focused ion beam (FIB), respectively. The ultramicrotomed sample is viewed edge-on. The boundary between Zones I and II is indicated by a dotted line sandwiched by Roman numbers I and II. On the other hand, because the FIB sample is not edge-on, the inclined surface can be seen to the left of a dotted line.

Elemental Distribution and Compositional Variation in the Composite Rims

Figure 8 shows HAADF-STEM images and element maps of the corresponding areas. Noguchi et al. (2011) reported that Fe, S, and Mg are enriched in Zone I of RA-QD02-0041 and -0042. We re-examined these elemental distribution data as well as newly obtained data on RA-QD02-0033. In Fig. 8, the boundary between Zone I and Zone II, and that between Zone II and Zone III are indicated by white dotted lines in the HAADF-STEM images. Element maps show the distribution of C, O, Na, Mg, Al, Si, Si, Cl, K, Ca, and Fe Kα lines. By comparing the HAADF-STEM images and corresponding element maps, Zone I has different chemical composition from Zones II and III. Elevation of Na, Al, K, and Ca Kα intensities is observed in Zone I of RA-QD02-0033. Elevation of Mg and Fe Kα intensities is shown in Zone I of RA-QD02-0042 and -0041. Because Zone I of RA-QD02-0033 does not contain np(Fe,Mg)S, Mg, Fe, and S Kα do not show elevated abundances in Zone I. Interestingly, Cl Kα intensity increases in Zone I of all the samples. Because the rims on RA-QD02-0041 and -0033 are composite vesicular rims, a variant of the composite rims, we describe their texture in detail later.

Figure 8.

Elemental distribution maps of composite rim and its vesicular variants. A) composite rim of low-Ca pyroxene (LPx) in RA-QD02-0042. B) and C) composite vesicular rims (described later) of olivine (Ol) in RA-QD02-0041 and 0033.

Plagioclase and troilite also have rims, which contain elements that are not included in these minerals. Figure 9 shows HAADF-STEM images and corresponding EDX spectra of plagioclase and troilite. The EDX spectrum of Zone I of plagioclase in RA-QD02-0042 shows the peaks of Mg, S, Cl, and Fe (Figs. 9A and 9C). In Fig. 9B, the exterior surface of troilite appears as a bumpy slope because this is the first ultra-thin section for the troilite crystal in RA-QD02-0054. Its EDX spectrum shows that the surface of troilite contains Na, Mg, Si, Cl, K, and Ca, which are, of course, not characteristic of troilite. Titanium and Cu in Figs. 9C and 9D are derived from the STEM sample holder and TEM grids, respectively.

Figure 9.

HAADF-STEM images and corresponding EDX spectra. A) and C) Fe-rich layer in Zone I of plagioclase in RA-QD02-0042 (boxed area no. 1 in 8A) contains Fe, Mg, S, and Cl, which are not included in plagioclase as major components. B) and D) Inclined surface of troilite in RA-QD02-0054 (boxed area no. 2 in 8B) contains Na, Mg, Si, Cl, and K, which are not included in troilite. Tr, troilite; Ol, olivine. Ti and Cu peaks are attributed to TEM sample holder and TEM grid, respectively.

Compositional variations that traverse from Zone I to Zone III in four Itokawa particles are illustrated in Fig. 10. Considering the analytical errors (indicated by bars in Fig. 10), there is no remarkable change in the composition of Zone II of olivine and low-Ca pyroxene. On the other hand, we recognize a decrease in Na in Zone II of plagioclase in RA-QD02-0054. On the other hand, it is clear that Zone I has different chemical compositions from those of Zone II and substrate minerals, which is consistent with results shown in Figs. 8 and 9.

Figure 10.

Compositional variations of minerals near the surface of composite rims in Itokawa particles. Roman numerals in these graphs represent Zone I and Zone II. Each analyzed area is a boxed area, the dimensions of which are 20 nm perpendicular to the surface and 200–400 nm along the surface depending on the shape of the surface. All the data were obtained from sections viewed edge-on.

High-Resolution Images of Nanophase (Fe,Mg)S and Nanophase Fe

Although it was very difficult to image the lattice fringes of the np(Fe,Mg)S and npFe, we successfully obtained images of lattice fringes of some nanophases from all of the samples containing them. Figure 11 displays high-resolution BF- and HAADF-STEM images. NpFe in Zone II of olivine in RA-QD02-0033 appears as bright spots in Fig. 11B. By comparing Figs. 11A and 11B, the npFe shows 0.20–0.21 nm lattice fringes. The npFe in Zone II of olivine in the other samples also has 0.20–0.21 nm lattice fringes. Lattice fringes of np(Fe,Mg)S in both Zones I and II of low-Ca pyroxene in RA-QD02-0042 are observed at the same time (Figs. 11C and 11D). Although the np(Fe,Mg)S in Zone I has 0.23 nm lattice fringes (Fig. 11C), the spacing varies from 0.21 to 0.23 nm at places. Similarly, although lattice fringes of the npFe in Zone II are 0.21 nm in Fig. 11C, they vary from 0.20 to 0.22 nm at places. Because the lattice fringes of npFe0 in lunar regolith 15004, 194 show 0.20 nm spacing at almost the same magnifications, these npFe particles probably contain both metallic iron and the other phases. This point is discussed later.

Figure 11.

High-resolution BF- and HAADF-STEM images of nanophases. A) and B) NpFe in Zone II of olivine in RA-QD02-0033 show 0.20 to 0.21 nm lattice fringes. C) and D) Np(Fe,Mg)S in Zone I and npFe in Zone II of low-Ca pyroxene in RA-QD02-0042 display 0.23 and 0.21 nm lattice fringes, respectively. In both samples, the areas displaying lattice fringes in A and C correspond to brighter areas in B and D. Dotted lines in C and D indicate approximately the position of the boundary between Zone I and II.

Size Distribution of Nanophase Fe

Diameters (equivalent circle diameters) of npFe in Zone II obtained from HAADF-STEM images for three Itokawa particles and npFe0 in a lunar regolith grain collected by Apollo 14 are shown in Fig. 12. Three samples show the mode around 1.5–2 nm. However, in RA-QD02-0041, the npFe tends to be smaller than the others. NpFe in RA-QD02-0033 tends to be larger than those in the other samples. The mode of the npFe0 in the lunar regolith sample is within the range of npFe in the three Itokawa samples. However, there are outliers in the lunar regolith, the diameters of which are larger than 4 nm. There is no npFe of a comparable size with the outliers in the Itokawa samples.

Figure 12.

Size distribution of npFe in Zone II of three Itokawa particles RA-QD02-0041, -0042, -0033 and comparison with npFe0 in lunar regolith 15004, 194. The majority of npFe and npFe0 are concentrated around 2 nm in diameter. However, lunar regolith contains npFe0 larger than 5 nm in diameter.

Composite Vesicular Rims

A variant of composite rims was identified from three particles: RA-QD02-0009, -0033, and -0041. The composite rims in these particles contain vesicles (Fig. 13). Their typical size is approximately 20 nm normal to the exterior surface by approximately 50 nm parallel to the exterior surface. We call these “composite vesicular rims.” Because these rims were found from samples prepared by both ultramicrotomy and the FIB lift-out method, it is not likely that they are artifacts of sample preparation.

Figure 13.

Composite vesicular rims on three Itokawa particles: A) to C) RA-QD02-0009; D) and E) RA-QD02-0041; F) RA-QD02-0033. A) HAADF-STEM image of composite vesicular rim observed in an oblique direction to the surface. B) and C) HAADF- and BF-STEM images of the same rim when viewed edge-on. Many vesicles are observed on the surface. D) and E) edge-on HAADF- and BF-STEM images of a composite vesicular rim. F) BF image of a composite vesicular rim. Cracks in D–F are artifacts from ultramicrotomy.

When observed from an oblique direction against the exterior surface, many dark patches were observed on the surface of low-Ca pyroxene in RA-QD02-0009 (Fig. 13A). When viewed edge-on, they are identified as vesicles (Figs. 13B and 13C). Figures 13D–F show other examples of the composite vesicular rims on olivine in RA-QD02-0041 and -0033. The number density of vesicles in RA-QD02-0009 is obviously higher than in the other samples.

Vesicles exist within Zone II and near the boundary between Zone I and Zone II (Fig. 14). Because the vesicles were formed within Zone II, npFe showing 0.204 nm lattice fringes could be identified in the inner wall of the vesicles (Figs. 14C and 14D). Except for the presence of vesicles, the other structural features of the composite vesicular rims are the same as those of the composite rims described in the previous section.

Figure 14.

BF- and HAADF-STEM images of a composite vesicular rim on olivine in RA-QD02-0041. A) and B) These images show that a vesicle exists within Zone II and near the boundary between Zone I. C) and D) Enlarged BF- and HAADF-STEM images of the vesicle shown in A and B. Lattice fringes of npFe0 are observed.

RA-QD02-0033: The Most Atmosphere-Exposed Particle Investigated in This Study

Figure 15A is an optical photomicrograph of RA-QD02-0033 embedded in epoxy resin before ultramicrotomy. The surface of this particle is not transparent, but translucent (indicated by an arrow), which is distinct from transparent and colorless surfaces of all the other Itokawa particles investigated (indicated by arrows in Figs. 15B and 15C). Their transparent and colorless feature is common to the particles with the composite rim (Fig. 15B) and those with the redeposition rim (Fig. 15C). However, because this particle was kept in a desiccator for about a month, it is possible that its cloudy surface resulted from the relatively long-term exposure to the Earth's atmosphere (>100 times longer than the other atmosphere-exposed samples).

Figure 15.

The most atmosphere-exposed Itokawa particle RA-QD02-0033 investigated in this study. A) An optical photomicrograph taken by using a transmitted light of this particle shows that it has a translucent surface (indicated by an arrow). Its translucent surface is different from the colorless and transparent surfaces of the other particles, B) RA-QD02-0035 and C) RA-QD02-0022 (indicated by arrows). A dotted area in A indicates an approximate position where ultra-thin sections were cut. D) A secondary electron image of this particle, which was attached to a glass fiber by glue. A dotted area in D indicates the approximate position where ultra-thin sections were cut.

Secondary electron image of this particle shows that its surface has a marked unevenness (Fig. 15D). BSE image of the cross section of the particle (Fig. 2L) revealed that it is composed of an aggregate of 5–10 μm-sized minerals with interstitial pores. The markedly uneven surface probably results from its aggregate structure.

Because it was difficult to identify solar flare tracks by using HAADF-STEM images, we searched the tracks by using conventional dark-field (DF) TEM images. Definite solar flare tracks could be observed only in RA-QD02-0033 (Fig. 16A). Figure 16A is a DF TEM image of olivine that is a shard of olivine, which was formed during ultramicrotomy. The composite vesicular rim appears as a dull and scalloped rim on the exterior surface. Nanophases and vesicles cannot be resolved in this DF TEM image. Beneath the rim, solar flare tracks appear as faint linear features in olivine (indicated by arrowheads in Fig. 16A). These textures are similar to a combination of solar-wind sputtered rim on the exterior surface and solar flare tracks found in interplanetary dust particles (IDPs) (Bradley and Brownlee 1986) and in a plagioclase grain in lunar regolith (Keller and McKay 1997). Because we could not find any other particles showing solar flare tracks as densely as this one, this particle probably has the longest exposure to interplanetary space of the particles investigated in this study.

Figure 16.

Dark-field (DF) and bright field (BF) TEM images of RA-QD02-0033. A) A composite vesicular rim (Rim) can be observed as a dull rim in the DF image. The linear features that appear in the core of olivine (Ol), indicated by arrowheads, are solar flare tracks. B) A porous high-Ca pyroxene (HPx) contacts with olivine (Ol). Selected area electron diffraction of HPx indicates that it is a single crystal.

Figure 16B is a DF TEM image of porous high-Ca pyroxene that contacts olivine. In spite of its porous structure, a selected area electron diffraction pattern (shown in an inset of Fig. 16B) indicates that the high-Ca pyroxene is a single crystal. Its texture is similar to that of vesicular plagioclase grains in RA-QD02-0060 (Nakamura et al. 2011b). Because RA-QD02-0060 was estimated to be a sintered regolith particle, RA-QD02-0033 may be similar in origin.

Discussion

Formation of Zone I (Redeposition Zone)

Recoil mixing of elements by implanted solar wind ions, deposition of sputtered material from other grains (Bernatowicz et al. 1994; Toppani et al. 2006), and impact vapor deposition (Keller and McKay 1997) may contribute to the chemical compositions of Zone I. The abundances of elements derived from minor minerals vary remarkably among the Itokawa particles. For example, reflecting the small abundance of troilite (2 vol%; Tsuchiyama et al. 2011), np(Fe,Mg)S was not observed in all Zone I rims. On the other hand, a small amount of Al is common among Zone I because plagioclase is the third most abundant mineral among the samples (11 vol%; Tsuchiyama et al. 2011). It is worthwhile to point out that Zone I often contains a minor amount of Cl (Figs. 4E and 8). Because Cl appears only in Zone I in elemental distribution maps (Fig. 8), Cl is also one of the exogenous components. Because xenolithic halides have been already reported from H chondrite falls, Zag, and Monahans (Zolensky et al. 1999, 2000; Whitby et al. 2000; Fries et al. 2011), halides may be also the source of Cl of Zone I of Itokawa particles.

Zone II as a Product of Irradiation Damage

Because both npFe and intimately coexisting Fe-poor amorphous patches are embedded in crystalline ferromagnesian silicates in Zone II (Fig. 6), and because the bulk composition of Zone II is almost the same as that of the substrate minerals (Fig. 10), we suggest that npFe was segregated in situ and that Fe-poor amorphous patches are residues after the segregation. The partial amorphization texture observed in olivine and low-Ca pyroxene in Zone II is similar to that of radiation-damaged zircon except for the segregation of nanophases (e.g., Ewing et al. 2003).

Many irradiation experiments have been performed to investigate the origin of space weathering (e.g., Dukes et al. 1999; Carrez et al. 2002; Lemelle et al. 2003; Brunetto and Strazzulla 2005; Strazzulla et al. 2005; Cantando et al. 2008; Davoisne et al. 2008; Loeffler et al. 2009). In these experiments, many ions, such as H+, He+, C+, Ne+, Ar+, Xe+, as well as electrons, were irradiated onto minerals with different acceleration voltages (i.e., different kinetic energy) and fluences. Spectral changes such as reddening and darkening were observed after ion irradiation when both irradiation and spectral measurements were performed under vacuum conditions (e.g., Strazzulla et al. 2005; Loeffler et al. 2009). In the following discussion, we would like to discuss the formation of Zone II using the results of the irradiation of low-energy (approximately 1 keV amu−1) H+ and He+ because the mean value of the solar wind is 468 km s−1 (approximately 1 keV amu−1) (Loeffler et al. 2009 and references therein) and because proton and 4He+ occupy 95.41% and 4.57% of the solar wind (Reisenfeld et al. 2007).

The irradiation of low-energy He+ and/or H+ causes amorphization of San Carlos olivine (approximately Fo90) (Demyk et al. 2001; Carrez et al. 2002) based on TEM observation, formation of metallic iron (Dukes et al. 1999; Carrez et al. 2002; Loeffler et al. 2009), based on X-ray photo electron spectroscopy (XPS) of redeposit of sputtered material (Loeffler et al. 2009). Carrez et al. (2002) also reported that a large amount of bubbles were observed in the samples with fluences in the range >5 × 1017 He+ cm−2. These data suggest that solar wind irradiation and implantation sputter the surface of Fe-bearing silicates and form sputter deposits, amorphize the Fe-bearing silicates, and cause in situ reduction of Fe2+ to Fe0 to form minute metallic iron crystals in Zone II.

In addition, the thickness of the composite rims and the composite vesicular rims also provides evidence for their origins. The thickness of the composite rims ranges from 30 to 60 nm (Figs. 5 and 7; Noguchi et al. 2011) and that of the composite vesicular rims ranges from 60 to 80 nm (Fig. 13). Because the latter values include the thickness of the swelled vesicles (5–20 nm thick), the thickness after subtraction of the vesicles overlaps with the range of thickness of nonvesicular rims. The similarity of the thickness of these rims suggests that their thickness was controlled by the same mechanism.

Wieler (1998) and Wieler et al. (2002) estimated that the solar wind is implanted approximately 30 nm deep and that solar energetic particles are implanted approximately 3 μm deep into solid matter based on the measurement of the trapped noble gases in the lunar regolith and asteroidal regolith breccias. In particular, solar wind protons that compose approximately 95% of solar wind (Reisenfeld et al. 2007) are implanted to 5–15 nm deep (McCord et al. 2011 and references therein). Demyk et al. (2001) stated that a surface layer of approximately 45 ± 15 nm of olivine was amorphized by 4 keV He+ irradiation. Carrez et al. (2002) reported that the thickness of the amorphized layer corresponds to the penetration depth of the irradiated ions. We estimated penetration depth of 1 keV H+ and 4 keV He+ into Fo70 olivine (density 3.3 g cm−3), which is within the range of olivine in the equilibrated Itokawa samples (Fo72.9–69.3: Nakamura et al. 2011), by using the SRIM code (stopping and range of ions in matter: Ziegler et al. 2008). The penetration depth by 1 keV H+ and 4 keV He+ was estimated by the sum of the projected range and the longitudinal straggle. The penetration depth by 1 keV H+ and 4 keV He+ was approximately 27 nm and approximately 51 nm, respectively. Because the range of the thickness of the composite rims (30–60 nm) is consistent with the estimated penetration depth formed by 4 keV He+ (approximately 51 nm), it is likely that the thickness of the composite rims was controlled by the penetration depth of solar wind He ions that occupy approximately 4% of solar wind (Reisenfeld et al. 2007). Dukes et al. (1999) reported that He+ ions change the chemical state of Fe even more efficiently than protons do, which is consistent with the STEM observation that npFe exists throughout Zone II.

Dran et al. (1984) and Cantando et al. (2008) reported on the extremely reactive nature of the surfaces of the irradiated samples. In particular, Cantando et al. (2008) have pointed out that this reactive nature should be considered during the analysis of irradiated extraterrestrial specimens acquired via sample return missions. Because the composite rims of the Itokawa samples are indeed very reactive (Appendix B), irradiation of solar wind is a plausible formation mechanism of the rims.

A combination of a radiation-damaged rim and solar wind track-bearing interior is common on the surface of plagioclase in lunar regolith (e.g., Dran et al. 1970; Bibring et al. 1972; Keller and McKay 1997), on the surface of isolated minerals in IDPs (Bradley and Brownlee 1986), and on the surface of olivine and pyroxene grains in a dark inclusion in the Ningqiang carbonaceous chondrite (Zolensky et al. 2003). Irrespective of the size of mineral grains and occurrences, the ranges of the thickness of the amorphous radiation-damaged rims on crystals in these samples (approximately 40–80 nm thick) are similar to the range of the thickness of the composite rims. Therefore, approximately 4 keV He+ released from the past sun probably played an important role in establishing the structures of these rims.

Nanophase Fe in Zone II and Nanophase (Fe,Mg)S in Zone I

Because irradiation experiments involving both low-energy H+ and He+ form metallic iron in Fe-bearing olivine, it is plausible that the irradiation of solar wind segregates npFe in Zone II of ferromagnesian silicates. The segregation of nanophase at concentrations well below the solubility limit is pervasive among irradiated alloys, which is known as radiation-induced segregation (RIS) (e.g., Zinkle 2012; Nastar and Soisson 2012). RIS proceeds more effectively by irradiation of light ions such as H+ and He+ than by irradiation of heavy ions for a given displacement per atom condition (Nastar and Soisson 2012). Because solar wind is composed mainly of light ions (approximately 95.4% proton, approximately 4.6% 4He ions, approximately 0.02% heavier ions; Reisenfeld et al. 2007), npFe segregation in Zone II may have been promoted by process(es) similar to the RIS.

As already pointed out by Noguchi et al. (2011), it is difficult to image lattice fringes of the np(Fe,Mg)S in Zone I and npFe in Zone II. Lattice fringes of the host minerals in Zone II that contain the npFe probably obscure the lattice fringes of the nanophase. In addition, many npFe particles may be poorly crystalline. It is obvious that most of the np(Fe,Mg)S in Zone I is poorly crystalline because it does not show lattice fringes, although embedded in amorphous silicate.

Based on the interpretation of the reflectance spectra of Itokawa, it was expected that the surface material would contain a small amount of metallic iron (Binzel et al. 2001; Abe et al. 2006; Hiroi et al. 2006). Noguchi et al. (2011) showed that some of the np(Fe,Mg)S in Zone I and some of the npFe in Zone II show approximately 0.22–0.23 nm and approximately 0.20 nm lattice fringes, respectively. This observation supports the proposal that there are npFe0 in the rims. Additional detailed observation of these nanophases revealed that crystalline nanophases show a variation in lattice fringes; those in Zone I range from 0.21 to 0.23 nm and those in Zone II range from 0.20 to 0.22 nm (Figs. 11 and 14).

Interplanar spacing of the strongest diffraction of minerals that may constitute these nanoblobs is as follows: 0.20 nm for (110)α-iron, 0.21 nm for (114)troilite, 0.21 nm for pyrrhotite (ascribed index depends on its structure), 0.22 nm for (200)wüstite, 0.26 nm for (200)niningerite, and 0.26 nm for (200)keilite. If we assume that the lattice fringes of nanophases are ascribed to the strongest diffraction of constituting minerals and if we postulate a typical error of lattice spacing measurements of 5%, it is likely that the np(Fe,Mg)S in Zone I contains α-iron, troilite, and/or pyrrhotite, and that npFe in Zone II contains α-iron and wüstite. Although the nanophases in Zone I contain Mg, their lattice fringes do not coincide with those of niningerite and keilite. Because the solubility of Mg in troilite is quite low (Skinner and Luce 1971), Mg may not be contained in the crystalline portion of the np(Fe,Mg)S in Zone I, but rather in the poorly crystalline portion that comprises the majority of the np(Fe,Mg)S in Zone I.

Poor crystallinity and variable spacing of the lattice fringes of crystalline nanophases in both zones may have occurred after the samples arrived on Earth. Itokawa particles were apparently exposed to air for about a week before the sample container was opened in the curation facility of ISAS/JAXA because 5000 Pa unfractionated air was detected during the capsule opening operation at the curation facility of ISAS/JAXA (Okazaki et al. 2011). The immediate leaking of air into the sample container must have occurred, possibly due to the detachment of the parachute from the re-entry capsule on the ground.

Rapid oxidation of the metallic iron formed during low-energy H+ or He+ irradiation onto Fe-bearing olivine has been reported based on Fe-2p XPS measurements (Dukes et al. 1999; Loeffler et al. 2009). They reported that in their experiments, about half of the metallic iron was oxidized after exposure to air for 10 min. Therefore, it is likely that a considerable amount of npFe0 in the Itokawa samples had been already oxidized before the initial analysis, although it is very difficult to trace the oxidation process of npFe0 by using TEM. This idea is consistent with the fact that we could not find remarkable textural, mineralogical, or chemical differences between ultra-thin sections prepared in a N2 purge environment (approximately 0.1% O2 and ≲100 ppm H2O; Noguchi et al. 2011) and those contacted with air for about 6 h. Loeffler et al. (2009) also reported that 15% metallic iron was left after 1 month exposure. This result is consistent with the STEM observation of RA-QD02-0033, which was kept in a desiccator for about a month. In RA-QD02-0033, although we could find npFe showing 0.204 nm lattice fringes, which can be interpreted as (110) of α-Fe (Fig. 11), the npFe0 particles were fewer than the other samples. Therefore, we believe that most npFe in Zone II was metallic iron when the Itokawa particles were on Itokawa. We think that Zone II was formed by irradiation damage of solar wind ions and that formation of npFe was promoted by implanted solar wind ions.

As for the np(Fe,Mg)S in Zone I, a RIS-like formation mechanism may have also played an important role. Chemical compositions of nanophases were probably controlled by the bulk chemical compositions of zones, in which nanophases were formed.

There was considerable controversy about the relative contributions of vapor deposition versus irradiation on lunar regolith grains in the 1990s (e.g., Bernatowicz et al. 1994; Hapke et al. 1994; Keller and McKay 1993). Valence electron energy loss spectroscopy reflecting bond breaking or disruption in solids indicates that some rims on lunar regolith sample 10084 show irradiated features, suggestive of multiple formation mechanisms such as vapor deposition and solar wind irradiation (Bradley et al. 2012). The different size distribution between npFe in the Itokawa particles and npFe0 in lunar samples may also provide another hint on relative contribution of these two agents (Fig. 12). Size distributions of the npFe in Zone II are unimodal, and their modes are at approximately 2 nm (Figs. 12A–C). Although the size distribution of npFe0 in lunar sample 15004 is also unimodal and although the mode is at approximately 2 nm, there are outliers with larger grain sizes (Fig. 12D). If npFe0 the diameter of which overlapped with that of npFe (1–2 nm in diameter) had been formed by the RIS and in situ reduction processes, large outliers might have been formed by mechanisms different from these processes.

Formation of Vesicles in the Rims

Carrez et al. (2002) observed abundant bubbles or voids at the bottom of the amorphous layer formed by irradiation of low-energy He+ with fluences on the orders of 1017–1018 ions cm−2. The bubble or void-rich zone exists just above the unaltered part of the minerals, i.e., at the boundary. They suggested that the bubble layer is located at the depths at which the He+ ions stop in the specimen and that the bubbles are filled with He gas. Their results are consistent with the observation of the composite vesicular rims in the Itokawa samples, which probably means that the vesicles were formed by segregation of the implanted solar wind gases.

In Figs. 5B, 7, and 13D, lenticular dark patches are apparent along with vesicles. These patches may be areas where He ions started to segregate but had not formed vesicles. If this is true, because the vesicles and their probable precursors exist throughout Zone II, there was a range in the depths at which the He+ ions stop in the Itokawa particles, which is consistent with the wide distribution of solar wind velocities (250–850 km s−1; Reisenfeld et al. 2007).

During comparison of a HAADF-STEM image of a vesicular rim observed from an oblique direction and HAADF-STEM images of the vesicular rims viewed from edge-on (Fig. 13), it is evident that there are many hillocks about 50 nm across on the surfaces of the Itokawa particles having the vesicular rims. Similarly abundant hillocks typically 50 nm across were also reported on the surface of olivine after irradiation of 4 keV He+ with fluences on the order of 1019 ions cm−2 (Davoisne et al. 2008). Although TEM observation was not a feature of this study, the morphological similarity between these hillocks suggests that they have similar internal structures and were formed by similar mechanisms. Davoisne et al. (2008) also reported that 1 keV He+ with similar fluences made smaller (20 nm) hillocks. Formation of lenticular bubbles parallel to the external surfaces (blistering) has been observed on the surfaces of materials exposed to irradiation of gas ions in nuclear environment (e.g., Igarashi et al. 2002). For example, abundant He+-filled blisters and minor H+-filled blisters on the surface of Si and SiC were observed after irradiation of H+ and He+, in which the fluences were the same: 2.0 × 1018 ions cm−2 (Muto et al. 2001; Igarashi et al. 2002). Helium-filled bubbles in alloys are also commonly observed in irradiated alloys (e.g., Zinkle 2012). The morphology of blisters is very similar to that of the vesicles in the Itokawa samples. These experiments suggest that the majority of the vesicles in the Itokawa samples were formed by solar wind—approximately 4 keV He+ ions implanted into the surface to depths of approximately 40–60 nm deep—and that the He+ ions segregated to form vesicles in Zone II. Because the calculation of the SRIM program suggests that the sputtering from He+ may be an order of magnitude higher than H+, He+ still is a significant component of the sputtering to form Zone I, even though H+ is approximately 20 times more abundant than He+ in solar wind.

Vesicular rims have been reported from lunar regolith (Phakey et al. 1972; Keller and McKay 1997; Assonov et al. 1998) and lunar regolith breccias (Noble et al. 2005). These studies concluded that the vesicular rims had been formed during implantation of solar wind gas. The sizes of the vesicles in the vesicular rims in the lunar regolith breccias are considerably larger (100–200 nm in diameter) than the vesicles in the Itokawa samples, and the rims in the regolith breccias were probably formed in situ by heating during lithification processes (Noble et al. 2005). In the unconsolidated lunar regolith, vesicular rims are amorphous without nanoparticles, approximately 100 nm thick, and do not show any evidence for sputtering or recondensation (Keller and McKay 1997). The sizes of vesicles (typically 20–80 nm across; Assonov et al. 1998) are comparable to those of the vesicles observed in the Itokawa samples (approximately 50 nm across). Keller and McKay (1997) suggested that the solar wind gases were released by heating due to nearby impact of micrometeoroids because irradiation experiments of 4 keV He+ ions onto pyrrhotite show that heating is needed to form bubbles (Brownlee et al. 1998). On the other hand, it is also possible that small bubbles can be formed by irradiation of 4 keV He+ ions onto olivine without heating (Carrez et al. 2002). Nearby impact of micrometeoroids could easily dissipate surrounding particles on Itokawa because of the low gravity (approximately 10−4 m s−2; Hirata et al. 2009) on Itokawa. Therefore, further studies are needed to clarify the formation mechanisms of vesicles in the Itokawa samples.

Time Scale of the Rim Formation on the Itokawa Particles

Because the redeposition rims directly attach Zone III (unchanged minerals) (Fig. 4), the formation rate of Zone I is probably higher than the formation rate of Zone II (amorphization rate of the substrate minerals). As discussed earlier, the composite rims and their vesicular equivalents in the Itokawa samples were probably formed by the implanted solar wind He+ ions and in situ reduction of Fe2+ to Fe0. Loeffler et al. (2009) obtained a characteristic time for space weathering of S-type asteroids at 1 AU, which is approximately 5000 yr by considering He+ ion irradiation experiments onto olivine. Because the space weathering of Itokawa is still in progress (Hiroi et al. 2006), it is likely that ≪5000 yr have passed to form the composite rims on the Itokawa particles. Dukes et al. (1999) reported that surface Fe3+ ions in the irradiated samples had been reduced to Fe2+ and Fe0 states after a fluence of 5 × 1016 He+ ions cm−2, which corresponds to <300 yr of solar wind irradiation at 1.3 AU. Therefore, it may have taken less than a few hundred years to form the observed npFe in Zone II.

After the structure of the composite rims was established, the total thickness of the composite rims would not have changed remarkably, but the amount of implanted solar wind gases and the number density of solar flare tracks in Zone III (unchanged mineral) would have accumulated according to the duration of exposure to solar wind and flare. Nagao et al. (2011) estimated the time necessary to accumulate the observed concentration of 20Ne ion in three Itokawa samples to be from 150 to 550 yr by assuming single-stage irradiation, no removal of grain surface, and no backscatter effects of Ne ions. These particles might have had the composite rims and even some might have contained vesicles in Zone II if their surfaces had been saturated in He.

The general lack of solar flare tracks in Zone III below the composite rims and the composite vesicular rims suggests that the exposure duration to solar wind was short. We measured the number density of solar flare tracks in RA-QD02-0033. Track formation rate by solar flares at 1 AU was estimated to be 3 × 105−3 × 107 tracks cm−2 yr−1 (Fraundorf et al. 1980). Because track density of 1010–11 tracks cm−2 corresponds to an exposure age duration of approximately 1 × 104 yr (Bradley 2004), approximately 2 × 109 tracks cm−2 in RA-QD02-0033 is on the order of 103 yr. Because the composite rim of this particle contains vesicles and because the unchanged minerals (Zone III) below the composite vesicular rims in the other samples contain rare tracks, the surfaces of Itokawa particles must have been saturated by solar wind He within a period of time shorter than on the order of 103 years. Nagao et al. (2011) found that Ne isotopic ratios of all the three Itokawa particles that they investigated coincide with the ratio of solar wind Ne. Because energetic solar Ne, or solar flare Ne, has different isotopic ratios from that of solar wind Ne due to isotopic fractionation during implantation (Grimberg et al. 2008), their result is consistent with our interpretation that Itokawa particles were exposed to the Sun for a relatively short period. The measured directions of solar flare tracks seem to be randomly distributed in RA-QD02-0033, which might reflect efficient stirring of fine-grained particles on Itokawa due to its low gravitational field. Maurette and Price (1975) have proposed the same thing on the basis of their observation that the degree of anisotropy of solar flare tracks in a meteorite regolith breccia is lower than that in the lunar regolith samples. Levitation of the Itokawa particles due to electrostatic repulsion between the Itokawa particles and the surface of Itokawa (Lee 1996; Hartzell and Scheeres 2013) might also have played a role in the origin of the random distribution of the tracks.

It is interesting to note that Zhang and Keller (2011) found that there is a positive correlation between the thickness of the irradiation rim and the track density of the lunar regolith samples. On the other hand, the range of thickness of the composite rims on the Itokawa particles is narrow (30–60 nm), and the tracks are not abundant. To discuss this difference, we have to consider many physical conditions such as bombardment by ions from the solar wind, magnetosphere ions, energetic electrons, cosmic rays, ultraviolet photons and micrometeorites (Loeffler et al. 2009), gravity, shielding effect by Earth's magnetosphere for the lunar case, levitation of fine-grained surface particles by photoelectric effects, etc. This is an issue for future study.

Implications for Space Weathering of Other S-Type Asteroids

To summarize our observations, modifications observed on the surfaces of 12 Itokawa particles suggest that the main agent to modify the external surfaces of the particles was solar wind irradiation and implantation of solar wind ions (especially H+ and He+), which caused irradiation damage of minerals, in situ reduction of Fe2+ to Fe0 to form npFe0, and the formation of vesicles probably filled by He gas. In addition, another important agent of space weathering, micrometeorite impact, may have also played a role in the formation of Zone I. This simple picture is related to the low gravity (approximately 10−4 m s−2; Hirata et al. 2009) on Itokawa. Even if impacts onto Itokawa made fine-grained debris, its low gravity would prevent accumulation of abundant fine-grained debris as small as the Itokawa particles investigated on the asteroidal surface and would diminish the efficiency of accumulation of vapor deposits generated by micrometeoroid impacts. Therefore, the formation of space weathered rims on Itokawa should be simpler than that on the Moon, on which two competing processes, micrometeoroid impacts and solar wind irradiation, promote the formation of the rims. Therefore, more varied and complicated rims including amorphous rims (amorphized by solar wind irradiation), inclusion-rich rims (vapor redeposit including abundant npFe0), combinations of these two rim types, and vesicular rims are observed on the lunar samples (e.g., Keller and McKay 1997; Pieters et al. 2000; Noble et al. 2005).

Because small npFe0 (<10 nm) redden optical spectra and because large ones (≳50 nm) darken these spectra (Keller and McKay 1997; Keller and Clemett 2001; Noble et al. 2007; Lucey and Riner 2011), both darkening and reddening of the reflectance spectra are observed for lunar agglutinates, in which glass contains abundant submicrometer metallic iron along with small npFe0 (<10 nm). Probable melt splashes formed by impacts found on the surfaces of the Itokawa samples are rare as compared with the lunar samples (Nakamura et al. 2012). The low abundance of such impact glass among the Itokawa samples and abundant small npFe (1–2 nm) in Zone II are consistent with the fact that Itokawa has a high albedo (0.23), which is higher than most of the S-type asteroids (0.11–0.22) and that Itokawa is as red as a large Main Belt S-type asteroid (Thomas-Osip et al. 2008).

Our STEM observations suggest rapid formation of npFe0 within the radiation-damaged surfaces of ferromagnesian minerals, which is consistent with the conclusions of Vernazza et al. (2009) who indicated rapid reddening of Main Belt S-type asteroids, suggestive of solar wind irradiation as a major agent of space weathering. However, because most S-type asteroids exhibit both reddening and darkening (e.g., Clark et al. 2002; Chapman 2004), something to darken the spectra must have accumulated on the surfaces of S-type asteroids. Similarly, Loeffler et al. (2009) state that the low-energy He+ ion irradiation experiments can be best applied to interpret the reflectance spectra of the Main Belt S-type asteroids, showing variation in slope and absorption band depth, but minor albedo variations. The accumulation of vapor deposits generated by micrometeoroid impact may play an important role in darkening the reflectance spectra of large S-type asteroids.

As discussed earlier, only particle RA-QD02-0033 contains measurable solar flare tracks, the number density of which suggests an approximately 103 yr exposure duration. It contains larger npFe (2–4 nm) than the other samples (Fig. 12). These data suggest that growth of npFe continues over thousands of years. If degrees of reddening caused by solar wind irradiation increase over approximately 106 yr in the Main Belt S-type asteroids (Vernazza et al. 2009), we have to consider also other mechanisms such as the formation of npFe0 in amorphous sputter deposits. In such a long time scale, the formation of npFe0 in vapor deposit generated by micrometeoroid impact would start to occur because it is expected to act with a 108–109 yr time scale (Sasaki et al. 2001; Vernazza et al. 2009).

Conclusions

On the basis of observations using Cs-corrected STEM, we identified three types of surface modifications probably formed by space weathering on the surfaces of Itokawa particles. (1) Redeposition rims are extremely thin (2–3 nm thick) and amorphous. They contain elements that are not included in the substrate minerals. They are sputter deposits and/or impact vapor deposits originating from the surrounding minerals. Because their substrate minerals are crystalline just below the rims, the duration of solar wind irradiation is very short. (2) Composite rims (30–60 nm), the structure of which is quite different between ferromagnesian silicates and plagioclase. In both minerals, the surface thin (5–15 nm) amorphous layer (Zone I) often contains np(Fe,Mg)S. Because Zone I contains elements that are not included in the substrate minerals as for redeposition rims, the origin of Zone I is common to the redeposition rims; in other words, the redeposition rims are composed only of Zone I. Below Zone I, there is a partially amorphized zone (Zone II). Zone II of ferromagnesian silicates contains abundant npFe (typically 1–2 nm). Lattice fringes of the npFe indicate that they contain npFe0. We believe that RIS and in situ reduction is the probable mechanism to form npFe. Probably, npFe causes reddening of the reflectance spectra of Itokawa. (3) Composite vesicular rims (60–80 nm) are composite rims containing vesicles (typically approximately 20 × 50 nm) in Zone II. The vesicles were probably blisters formed by segregation of solar wind He implanted in Zone II. These textures strongly suggest that solar wind irradiation damage and implantation are the major causes of surface modifications observed on the surface of the Itokawa particles. We conclude that solar wind is the major cause of space weathering on Itokawa.

Acknowledgments

First, we especially thank the Hayabusa project team for the sample return. We are grateful to M. Kawamoto, R. Hinoki, A. Yamaguchi, Y. Suzuki, T. Sato, Y. Kuroda, K. Watanabe, and K. Muto for supporting N2 purge sample preparation at ISAS/JAXA and Ibaraki University, FE-SEM observation at Ibaraki University, and STEM and FE-SEM observation at Hitachi High-Technologies Corporation, and FE-TEM observation at Hitachi High-Tech Manufacturing and Service Corporation. We are most grateful to J. P. Bradley for reading the manuscript and giving us constructive comments. Discussion in a seminar at Nuclear Science Research Institute, Japan Atomic Energy Agency was also useful to improve the manuscript. Constructive and detailed comments by reviewers and the AE were quite useful to improve the manuscript. T. Noguchi was supported by JSPS KAKENHI grant number 2424408. M. E. Zolensky was supported by NASA's Muses-CN Program.

Editorial Handling

Dr. Scott Sandford

Appendix A

Details of sample preparation and analysis of STEM images

We used dehydrated ethylene glycol as the trough liquid instead of distilled water to prevent unnecessary contact with water for sample preparation in both N2 purge and Earth's atmosphere. For RA-QD02-0032, ultra-thin sections were prepared by both ultramicrotomy and focused ion beam (FIB) lift-out sample preparation technique to compare the texture of the composite rims prepared by both methods. For RA-QD02-0009, an FIB lift-out section was prepared from a manually polished section. The FIB lift-out sections were approximately 100 nm thick. FIB lift-out samples were prepared by Hitachi FB2200 FIB at Hitachi High-Technologies Corporation.

At Hitachi High-Technologies Corporation, we used an air-protection TEM holder to observe ultra-thin sections prepared in N2 purge environment and FIB lift-out sections. During this procedure, the dew temperature of the N2 purge grove-boxes was kept approximately −50 °C and oxygen concentration was less than 0.1%. Although we used a normal TEM holder for the other sections, exposure time to the Earth's atmosphere for these samples during mounting on the holder was less than a few minutes because ultra-thin sections were preserved in a vacuum desiccator until just before STEM observation and quickly returned to the vacuum desiccator after observation. By using these samples, we could examine the impact of the exposure time to the atmosphere on the internal texture of the rims. There is no remarkable textural difference in the rims between ultra-thin sections prepared in N2 purge environment and those prepared in the atmosphere. Therefore, we described the rims of all the particles investigated without distinction.

Detection limits were approximately 0.3 wt%. We assumed that the precision of analysis can be calculated by a function math formula (%), the errors are 1.5–3% for major elements, which depend on count number of an element. In addition, there are the other errors that are ascribed to the determination of k-factors of elements. However, it is difficult to determine these values based on a limited number of analyses. Therefore, we used a typical analytical error that appears in many TEM textbooks: 5%. It is not suitable to use this value for minor elements (less than about 2 wt%). For such elements, we used the above function. The value reaches approximately 20% for elements of approximately 0.5 wt%. By using these assumptions, we added error bars in Fig. 10.

Grain sizes of the npFe in Zone II were measured using HAADF-STEM images of RA-QD02-0033, -0041, and -0042 as well as those of npFe0 in lunar regolith sample 15004, 194 for comparison. We adjusted contrast and brightness to image npFe by using Adobe Photoshop CS5 software. Then, the images were binarized and the area of each npFe was measured using Image J software. We estimated that the equivalent circle diameter represented the diameter of each npFe.

The spacings of lattice fringes of npFe and npFe0 in the samples were calibrated by using lattice fringes of gold. The spacings were obtained by averaging the distances of 3–10 fringes by using Gatan Digital Micrograph software.

Appendix B

Highly reactive nature of the sample preparation and analysis of STEM images

During initial analysis, we found that FIB lift-out sections of the Itokawa samples are indeed very reactive. Surprisingly, even if the FIB sections were kept in a vacuum desiccator, the surfaces of the sections of the composite rim reacted with Cu derived from the FIB grid to form Cu sulfide on the surface of the FIB sections (Fig. A1). This fact suggests that the extremely reactive nature of the composite rim was enhanced by the FIB thinning process, which made amorphous layers that are parallel to both surfaces of the FIB section. Even ultramicrotomed sections show severe contamination during STEM observation after being kept in aluminum-laminated plastic bags filled with pure (99.9995%) N2 gas. The bags were sealed in a N2 purge glove box for approximately 10 months (Table A2).

Figure A1.

HAADF-STEM images of the same FIB lift-out section of RA-QD02-0032. A) HAADF-STEM image of the section just after its preparation. There is nothing except for contaminants (metallic Al and tin oxide), both of which are typical contaminants among the Itokawa particles. B) HAADF-STEM image of the section 9 months later after its preparation. Many small crystals were observed at the boundary between the external surface and the embedding epoxy. C) SEM images of the small crystals indicate that they exist on the surface of the FIB section. EDX spectrum was obtained at a circle labeled as Spectrum 1. D) EDX spectrum of the small crystal. It is composed of Cu and S. Ti is derived from the TEM holder.

Table A1. Semiquantitative TEM-EDX analysis of amorphous rims and their substrates in Itokawa dust particles.

Sample IDMineralaDistance from edge (nm)ONaMgAlSiSClKCaMnFe
  1. a

    Ol = olivine; LPx = low-Ca pyroxene; Pl = plagioclase.

  2. b.d. = below detection; n.i. = not included.

RA-QD02-0042LPx1046.4b.d.13.1b.d.27.30.8n.i.b.d.b.d.b.d.12.4
RA-QD02-0042LPx3544.0b.d.15.7b.d.26.5b.d.b.d.b.d.1.20.412.1
RA-QD02-0042LPx6042.9b.d.16.9b.d.26.4b.d.b.d.b.d.0.9b. d12.6
RA-QD02-0042LPx8543.7b.d.16.2b.d.26.5b.d.b.d.b.d.1.0b.d.12.4
RA-QD02-0042LPx11043.0b.d.16.1b.d.26.4b.d.b.d.b.d.1.10.513.0
RA-QD02-0042LPx13543.3b.d.16.5b.d.26.3b.d.b.d.b.d.1.10.412.4
RA-QD02-0042LPx16043.3b.d.16.2b.d.26.4b.d.b.d.b.d.1.20.412.5
RA-QD02-0042LPx18543.3b.d.16.1b.d.26.3b.d.b.d.b.d.1.00.512.8
RA-QD02-0042LPx38544.3b.d.16.0b.d.26.1b.d.b.d.b.d.0.8b.d.12.4
RA-QD02-0042LPx58544.4b.d.16.3b.d.25.9b.d.b.d.b.d.0.50.412.5
RA-QD02-0041Ol1038.2b.d.20.8b.d.18.80.7n.i.b.d.b.d.0.521.1
RA-QD02-0041Ol3038.7b.d.22.4b.d.17.80.0b.d.b.d.b.d.0.520.5
RA-QD02-0041Ol5038.7b.d.22.6b.d.17.50.0b.d.b.d.b.d.0.520.7
RA-QD02-0041Ol7939.2b.d.22.3b.d.17.70.0b.d.b.d.b.d.0.520.3
RA-QD02-0041Ol9039.5b.d.22.1b.d.17.80.0b.d.b.d.b.d.0.420.2
RA-QD02-0041Ol11039.5b.d.22.3b.d.17.50.0b.d.b.d.b.d.0.520.3
RA-QD02-0041Ol13039.1b.d.22.3b.d.18.10.0b.d.b.d.b.d.0.520.1
RA-QD02-0041Ol15039.8b.d.22.3b.d.17.70.0b.d.b.d.b.d.0.419.8
RA-QD02-0041Ol17039.9b.d.22.1b.d.17.80.0b.d.b.d.b.d.0.419.8
RA-QD02-0054Pl1052.04.2b.d.10.129.7b.d.n.i.b.d.b.d.b.d.1.8
RA-QD02-0054Pl3051.82.0b.d.11.932.7b.d.b.d.b.d.b.d.b.d.b.d.
RA-QD02-0054Pl5051.53.1b.d.11.531.9b.d.b.d.0.7b.d.b.d.b.d.
RA-QD02-0054Pl7050.04.2b.d.12.531.0b.d.b.d.1.00.8b.d.b.d.
RA-QD02-0054Pl9051.44.6b.d.10.531.8b.d.b.d.0.6b.d.b.d.b.d.
RA-QD02-0054Pl11050.35.6b.d.10.731.4b.d.b.d.b.d.b.d.b.d.b.d.
RA-QD02-0054Pl13049.95.2b.d.10.433.0b.d.b.d.b.d.b.d.b.d.b.d.
RA-QD02-0054Pl15048.35.9b.d.11.632.2b.d.b.d.b.d.b.d.b.d.b.d.
RA-QD02-0054Pl17049.46.1b.d.11.431.2b.d.b.d.b.d.b.d.b.d.b.d.
RA-QD02-0054Pl19050.15.5b.d.11.331.5b.d.b.d.b.d.b.d.b.d.b.d.
RA-QD02-0033Ol1038.81.116.80.622.40.34.20.60.8b.d.14.4
RA-QD02-0033Ol3040.30.622.6b.d.18.7b.d.1.0b.d.b.d.b.d.16.9
RA-QD02-0033Ol5040.8b.d.22.8b.d.17.9b.d.b.d.b.d.b.d.0.418.0
RA-QD02-0033Ol7040.7b.d.22.5b.d.17.8b.d.b.d.b.d.b.d.0.418.5
RA-QD02-0033Ol9040.9b.d.22.1b.d.17.8b.d.b.d.b.d.b.d.0.418.8
RA-QD02-0033Ol11040.2b.d.22.6b.d.18.0b.d.b.d.b.d.b.d.0.418.8
RA-QD02-0033Ol13040.6b.d.22.4b.d.17.8b.d.b.d.b.d.b.d.b.d.18.9
RA-QD02-0033Ol15040.2b.d.22.5b.d.18.0b.d.b.d.b.d.b.d.0.418.9
RA-QD02-0033Ol17040.2b.d.22.5b.d.18.0b.d.b.d.b.d.b.d.0.418.9
RA-QD02-0033Ol19040.3b.d.22.5b.d.17.8b.d.b.d.b.d.b.d.0.419.0

Table A2. Contamination found on ultra-thin sections during STEM observation.

Date of analysis9-25/2/201117/7/201117-18/11/201127-28/12/2011
Duration of storage~5 months~9 months~10 months
Contamination during STEM observation
  1. a

    “Not remarkable” means that no severe contamination was observed after approximately 1 h long elemental mapping by STEM.

  2. b

    “Remarkable” means that severe contamination was observed immediately during STEM observation.

Stored in N2Not remarkableaNot remarkableaNot observedRemarkableb
Stored in vacuumNot remarkableaNot remarkableaNot remarkableaNot remarkablea

Ancillary