MULTIPLE AND FAST: THE ACCRETION OF ORDINARY CHONDRITE PARENT BODIES

To explore the mineralogical composition of our asteroid sample, we utilized the SpeX instrument (Rayner et al. 2003) over the near-infrared 0.8–2.5 μm range on the NASA Infrared Telescope Facility (IRTF) on Mauna Kea, HI. This wavelength range covers the diagnostic absorption bands (at 1 and 2 μm) due to olivine and pyroxene, which are characteristic for both OCs and S-type asteroids (see Appendix A.1 for a brief description of the observation protocol and the data reduction procedure). Combining our new near-infrared measurements with available visible wavelength spectra (Bus 1999) allows for the first time a nearly complete spectral database of main-belt S-types with D > 60 km (95% or 54/56; accounting for >85% of all S-type mass). Previous near-IR measurements were acquired for 32 main-belt S-types (Gaffey et al. 1993) while De León et al. (2010) acquired near-IR spectra for 22 main-belt S-types and Gietzen et al. (2012) for 5 main-belt S-types.

While spectral data for more than 60 EOCs have existed within the RELAB database for a long time (http://www.planetary.brown.edu/relab/), similar data have been quasi-inexistent for type 3 UOCs (there were spectra for only 11 UOCs: Suwahib-Buwah (H3.7), Dhajala (H3.8), ALHA77214 (L3.4), Hallingeberg (L3.4), Khohar (L3.6), Mezo-Madaras (L3.7), Hedjaz (L3.7), Bishunpur (LL3.1), Krymka (LL3.1), Chainpur (LL3.4), Parnallee (LL3.6)). Considering that the range of properties within petrologic type 3 (subtypes 3.0–3.9) is as great as that from types 4 to 6 (Sears et al. 1980), it is clear that there is a major lack of data within the RELAB spectral database.

In order to reduce the current gap in data between UOCs and EOCs, we obtained a large number of Antarctic samples (http://curator.jsc.nasa.gov/antmet/) with the specific goal of measuring the spectral properties of UOCs. As UOCs are compositionally diverse, we asked for a representative sample (53 meteorites in total) for H, L, and LL UOCs (i.e., the 3.0–3.9 continuum needing to be well-covered for H, L, and LL—see Table 2). The meteorites were sieved to grain sizes in the 0–45 μm range and spectra were subsequently collected at the RELAB facility (http://www.planetary.brown.edu/relab/) over the 0.3–2.6 μm range. The spectra are available in the RELAB database.

Our broadened spectral survey indicates the presence of two compositional groups (shown by the bimodality) among S-type asteroids, namely (1) several objects have the same spectral signature (and quantitatively the same surface composition) as asteroid (6) Hebe (Figure 4) and equilibrated H chondrites and (2) several others have Flora-like (8 Flora) and LL chondrite-type spectral signatures. Collisions cannot be responsible for the existence of these two groups as these compositionally similar asteroids are found at large sizes (D > 100 km) and are unlikely to have experienced disruptive collisions since the formation of the asteroid belt (Bottke et al. 2005a, 2005b). Our observations thus indicate that "clones," namely identical compositions among multiple large asteroids (D ≈ 100–200 km), are a natural outcome of planetesimal formation.

Zoom In Zoom Out Reset image size

Lyra & Kuchner (2013) recently showed that due to interactions between gas and dust, a disk might, under the right conditions, produce narrow rings on its own without planets being needed. Ring-like structures, which are thus proposed as a natural step in the evolution of a protoplanetary disk, would naturally lead to the formation of several compositionally similar asteroids. The observation of two compositional groups only among S-type asteroids—implying that prior to accretion the disk must be locally quite homogeneous—appears very coherent with those findings. Our results are also consistent with current models of planetesimal formation via secular gravitational instabilities with turbulent stirring, which predict the formation of clans of chemically homogeneous planetesimals (e.g., Youdin 2011).

Our observations of several compositionally similar asteroids do open the possibility that a given meteorite group (e.g., H chondrites) can have more than one parent body, but do not imply that this is necessarily the case. Indeed, we want to stress here that it is not incompatible with our results that meteorite measurements should imply that a large fraction of meteorites in each OC group come from the same parent body. About 50% of the H chondrites with measured cosmic ray exposure ages have ages between 7 and 8 Myr suggesting a similar origin (Eugster et al. 2006). Similarly, about two-thirds of L chondrite meteorites were heavily shocked and degassed with ³⁹Ar–⁴⁰Ar ages near 470 Myr (Korochantseva et al., 2007) implying a common parent body for those meteorites (Haack et al. 1996).

It is worth noting that none of the properties that define the OC groups actually requires a single parent-body, not even oxygen isotopic signatures, which have been shown to be directly related to petrography (Zanda et al. 2006), more specifically to the relative proportions of reduced and oxidized chondrules. Chondrules constitute 80% of OC materials and therefore control their chemistry and mineralogy, as well as their oxygen isotopic signatures (Zanda et al. 2006), hence both the OC group to which they belong and their spectral properties. If groups of asteroids with similar spectra are a product of accretion, namely if they accreted from the same reservoir of chondrules and matrix, it is thus expected that they will also share their oxygen isotopic signatures.

Finally, we want to stress that asteroid (6) Hebe has many twins—including families—that lie as close or even closer (Figure 5) to a resonance (3:1 resonance in particular). This implies that there are several alternatives to asteroid Hebe as plausible source(s) for H chondrites.

Zoom In Zoom Out Reset image size

Our deduction of equilibrated surface compositions on current day left-peak S-type asteroids (ol/(ol+low-Ca px) <65; the left peak (Figures 2 and 3) is analogous to thermally metamorphosed (type 3.6–6; minimum T > ∼400°C) H and possibly L chondrites) is unexpected since onion skin models for ordinary chondrite parent bodies (e.g., Ghosh et al. 2003; Henke et al. 2012a, 2012b, 2013) predict original surfaces of UOC material (type 3.0–3.4). EOC material is formed in volumetrically large abundance, but in the context of the onion skin models resides entirely in the metamorphosed interior of the original asteroid. Thus, we propose that the left peak is composed of asteroids whose surfaces are not primordial; rather their ol/(ol+low-Ca px) ratios reveal that today we are seeing the exposed interiors.

While many D <∼ 100 km sized bodies in the left peak are likely collisional fragments (Bottke et al. 2005a; Consolmagno et al. 2008; Morbidelli et al. 2009; Carry et al. 2012, and references therein)—thus naturally explaining that their surface composition is best matched by previously interior, thermally metamorphosed chondrites (type 3.6–6), large bodies (D > 100 km) are more likely to be intact survivors capable of preserving at least some of their original surface compositions. Indeed, both current collisional models (Bottke et al. 2005a; Morbidelli et al. 2009) as well as the recent measurements performed by Rosetta on the asteroid Lutetia (Pätzold et al. 2011; Sierks et al. 2011) and by Dawn on the asteroid Vesta (Marchi et al. 2012) minimize the possibility that large bodies (D > 100 km) are collisional fragments themselves; instead these objects are predominantly primordial (undisrupted) planetesimals, which means they are unlikely to have undergone a catastrophic disruption since their formation. Specifically, an impactor population capable of producing widespread collisional disruption among D = 100–200 km asteroids would

• 1.  
  strongly affect asteroids like Vesta by creating ∼9–10 basin-forming events equivalent to the one that made the D ∼ 500 km Rheasilvia basin. Right now, we know of two such basins on Vesta (Marchi et al. 2012). Each basin is also associated with a band of deep equatorial troughs as measured from the central axis passing through the center of the basin. For now, it seems highly unlikely that one could (1) hide seven additional such basins on Vesta's surface and, at the same time (2) avoid bringing lots of the diogenites to the surface. Note that the relaxation of Vesta's surface may have occurred during the first ∼100 Myr of the solar system implying that primary basins may be no longer observable (Fu et al. 2013). As such, Vesta's actual surface may thus not be fully incompatible with an early period of more intense bombardment.
• 2.  
  create enumerable asteroid families filled with large fragments. (A family is produced when a large asteroid undergoes a catastrophic collision, leaving behind numerous fragments with similar proper orbital elements). However, the telltale evidence that should have been produced by such disruption events—numerous asteroid families with large fragments—is not obvious in the main belt today, as we only see a few large S-type asteroid families, the oldest being ∼2 Gyr old (e.g., Koronis family). Thus, some mechanism would be needed to eliminate this evidence.

A natural explanation for the surfaces of D > 100 km left-peak S-type asteroids consisting mostly of material formed in the interior, is that impacts by small asteroids over the course of solar system history (D < 10 km; subcatastrophic collisions), and in particular those that occurred during the first ∼20 Myr of solar system formation (Ciesla et al. 2013), brought deep material up to the surface. Indeed, Ciesla et al. (2013) recently showed that early impacts on "warm" bodies (up to 20 Myr after the bodies were formed) would be far more efficient than later ones on "cold" bodies (500 Myr after solar system formation or even later) in bringing deep material to the surface. Following their work, early impacts during the first ∼20 Myr of solar system formation—where the interior of OC parent bodies would still be at high temperature from the decay of short-lived nuclides such ²⁶Al—would result in hot material from the deep interior (material mainly from 10 to 30 km depth in a D = 200 km body) being brought to, and flowing out over, the surface of the planetesimal, covering most of its surface. This contrasts with the effect of later impacts with cold planetesimals by material from no deeper than ∼10 km below the surface (in a D = 200 km body) being slightly heated by the impact and then exposed around the point of impact.

For asteroids to the right of the left peak (ol/(ol+low-Ca px) > 65), the ol/(ol+low-Ca px) ratios do not distinguish between a surface and interior origin. However, since the sizes of these bodies do not differ significantly from those in the left peak (though they do comprise larger (D > 75 km) bodies; see Figures 11 and 12), a different collisional history seems unlikely. We therefore suggest that the surfaces of these bodies are also exposed interiors, which would imply that bodies in the left peak are H-like bodies while bodies in the right peak are LL-like bodies only (the presence of H-like bodies with unequilibrated surfaces in the right peak thus seems unlikely). Compositionally equilibrated L chondrites are intermediate between H and LL. Thus, the compositional gap (Figure 2) between the two peaks (60 < ol/(ol+low-Ca px) < 72) implies that asteroids with L chondrite surfaces must be rare.

In summary, our observations suggest that current day S-type asteroid surfaces are exposed interiors. Independent support for this finding is provided by the metallographic cooling rates of H chondrites (discussed in the third section of the Introduction). Finally, we would like to stress that S-type asteroids are not the only objects whose surfaces are exposed interiors. Indeed, this is also the case for the asteroid Vesta whose surface composition is best matched by howardites (Hiroi et al. 1994; De Sanctis et al. 2012), which are, by definition, a mixture of the primordial crust (eucrites) and of the underlying layer (diogenites).

It is well known that atmospheric entry biases our meteorite collections by preferentially selecting the denser and more compact objects, which explains for example why ordinary chondrites are largely overrepresented in our collections with respect to carbonaceous chondrites, such as CI and CM chondrites. However, it is still unclear whether, at similar densities as it is the case for H, L, and LL chondrites, these collections are fairly representative of the compositional diversity of the asteroid belt. We have accumulated data for the majority of the most massive S-types (i.e., we sample more than 85% of the main-belt mass for S-types) and we can therefore use our survey to make progress on this question.

Considering the fall statistics of the H (∼34%), L (∼37%), and LL (∼8%) classes, both the paucity of L-like bodies among the asteroids and the dominance of LL-like bodies in the main-belt population are very surprising. This implies that even at similar densities and tensile strength (which is typically the case for the H, L, and LL chondrites) meteorite falls are not representative of the compositional diversity of the asteroid belt. A direct implication is that atmospheric entry is not the only factor that governs (i.e., biases) the inventory of our collections. Other factors, such as the location of the source region versus the location of the main resonances, the nature of the source region (e.g., asteroid collisional family versus cratering event on a large asteroid, size frequency distribution of the collisional family), or the age of the source region (e.g., age of the collisional family), may play a fundamental role in the current statistics of meteorite falls.

It is clear that large bodies do not necessarily dominate the delivery of material to the Earth. For example, the largest main-belt asteroid, Ceres (D ∼ 950 km, ∼25% of the mass of the main belt), may be unsampled by our meteorite collections as suggested by spectroscopy. Similarly, there are few lunar and Martian meteorites. Instead, asteroid families that are the result of catastrophic collisions producing an abundance of small fragments may be the main source of meteorites. This is possibly the case for the Gefion family that has been proposed (Nesvorny et al. 2009), based on dynamical arguments, to be the source of L chondrites. Here, we confirm that Gefion is the only known family whose composition is compatible with those of L chondrites, consequently reinforcing families as the most plausible sources of meteorites. The age of the families may also be a factor, causing older families to possibly contribute less to the meteorite flux than younger ones as their smaller members may have been removed a long time before via the Yarkovsky effect (Rubincam 1995; Bottke et al. 2006). Future dynamical work will shed light on this question and, as a byproduct, may also explain the compositional difference between meteorites and NEAs (Vernazza et al. 2008; De León et al. 2010; Dunn et al. 2013).

The accretion duration for asteroids, from the disk phase to their final size, is a fundamental and still open question. Both the size distribution of main-belt asteroids and the most recent planetesimal formation models predict that the asteroids formed big, namely that the size of solids in the proto-planetary disk "jumped" from a submeter scale to a multi-kilometer scale, without passing through intermediate values (Johansen et al. 2007; Cuzzi et al. 2008; Morbidelli et al. 2009). Here, we show that the surface composition of the members of H-like families is consistent with those predictions, thus providing another piece of evidence for the fast accretion scenario of asteroids. However, it should be noted that the present result does not allow us to distinguish between instantaneous accretion and a growth over ∼10⁵ yr and that it is highly dependent on current thermal models (e.g., Henke et al. 2012a, 2012b; Henke et al. 2013). Specifically, it relies on the assumption that the main source of asteroid heating (metamorphism) was the decay of short-lived radioactive nuclides leading to an onion-shell structure for early asteroid parent bodies. If future findings demonstrate that this assumption is wrong, the present result will need to be reinterpreted.

As shown by Ghosh et al. (2003), the internal structure and therefore the thickness of the primordial unheated crust of type 3.0–3.4 material is directly related to the duration of accretion (Ghosh et al. 2003; Henke et al. 2012a, 2012b, 2013):

• 1.  
  in the case of rapid accretion (instantaneous accretion or growth over ∼10⁵ yr), asteroids have been quasi-metamorphosed throughout (onion shell structure) and formed with a thin crust of type 3.0–3.4 material (<∼5 km assuming a D = 200 km sized parent body; see Ciesla et al. 2013, and references therein; Figure 6);
• 2.  
  in the case of a long duration of accretion (incremental accretion over at least 1 Myr), only the cores have been metamorphosed leaving primordial bodies with a thick outer shell of type 3.0–3.4 material (>∼20 km assuming a D = 200 km sized parent body, more than ∼40% in volume; Figure 6).

Zoom In Zoom Out Reset image size

This difference of the internal structure as a function of duration of accretion is particularly interesting in the case of H-like bodies since one can use spectroscopy to distinguish the outer shell from the metamorphosed interior and thus to estimate the primordial percentage (in volume percent) of type 3.0.3–4 material in H-like bodies. To estimate this percentage, asteroid families are an ideal laboratory, because they allow us to access the interior composition of primordial parent bodies.

In the case of H-like asteroid families, we would expect to observe a clear compositional variation among the fragments in the case of a long duration of accretion as the volume of type 3.0–3.4 material (∼40%) would be roughly equivalent to the volume of type 3.6–6 material (∼60%). Specifically, we would expect to observe an extended compositional range, namely some members with a type 3.0–3.4 composition (LL-like), some other members with a type 3.6–6 composition (equilibrated H-like) and possibly some members being mixtures of type 3.0–3.4 and 3.6–6 materials (L-like, see Figure 6). This is, however, not the case since families in the H peak (Agnia, Merxia, and Koronis) show little compositional variation among their members. In addition, all the members of each family have a composition that is only compatible with equilibrated H chondrites. We estimate that type 3.0–3.4 materials can only represent a small fraction of the current surface material (<∼15%), as for more than ∼15%, there would be an obvious shift between the asteroid compositions and the compositions of equilibrated chondrites. Our observations are, therefore, consistent with an initially thin crust of type 3.0–3.4 materials (<∼5 km assuming a D = 200 km sized parent body) and thus with a short duration of accretion (<0.3 Myr; see Ghosh et al. 2003).

The small fraction of type 3.0–3.4 material implied by our observations is consistent with meteorite fall statistics (type 3 OCs represent ∼15% of the falls and type 3.0–3.4 OCs less than 5%; see Hutchison 2004), while the short duration of accretion implied by our observations is consistent with current models of planet formation through streaming and gravitational instabilities (Youdin & Goodman 2005; Johansen et al. 2007; Chiang & Youdin 2010) that show that bodies of several hundred kilometers in size form on the timescale of a few orbits (Johansen et al. 2011; Youdin 2011; Johansen et al. 2012).

The distribution of asteroids across the main belt has been studied for decades to understand the current compositional distribution and what that tells us about the formation and evolution of our solar system (Gradie & Tedesco 1982; DeMeo & Carry 2013, 2014). The result of those studies has shown the existence of a global compositional heliocentric gradient in the asteroid belt (S-types in the inner part; C-types in the outer part; Gradie & Tedesco 1982; DeMeo & Carry 2013, 2014). However, the compositional gradient among S-types, that is the respective formation locations of the H, L, and LL parent bodies, remain unknown.

We computed the average semi-major axis of asteroids in the two peaks in order to trace any initial difference in terms of formation location between the H and LL parent bodies. Since small objects (D < 20 km) are sensitive to the Yarkovsky effect, which affects their semi-major axis, the current semi-major axis of D < 20 km planetesimals is quite different from their initial one; on the contrary, larger objects have not been moved from their current location since the last episode of major migration (e.g., Nice and Grand Tack models; Gomes et al. 2005; Tsiganis et al. 2005; Morbidelli et al. 2005; Walsh et al. 2011). We therefore restricted our sample to asteroids with D > 30 km (to be far enough from the D = 20 km limit) in the two peaks.

We found that LL-like bodies are located, on average, closer to the Sun (2.56 ± 0.04 AU) than H-like bodies (2.66 ± 0.03 AU) (Figure 7). Note that, such as in the case of other compositional classes (e.g., S- and C-types), we observe a strong overlap in heliocentric distance between these two groups, which suggests that the radial mixing of planetesimals after their formation must have been significant, a likely consequence of giant planet migrations (e.g., Walsh et al. 2011). Since migration models (e.g., Walsh et al. 2011) for small bodies and planets that try to reproduce the architecture of the solar system conserve the initial relative positions of the bodies with respect to the Sun, LL-like bodies are likely to have formed closer to the Sun than H-like ones.

Zoom In Zoom Out Reset image size

Since chondrules in OCs are size-sorted, with LL chondrites containing on average larger chondrules than H chondrites (Kuebler et al. 1999; Zanda et al. 2006), we explore here whether size sorting of chondrules in the young protoplanetary disk might be an efficient physical mechanism that would naturally explain this observed difference in formation locations.

In a gaseous disk (Adachi et al. 1976; Weidenschilling 1977), solids with different sizes (and densities) have a different aerodynamic coupling with the gas, which automatically implies that they acquire nonzero relative radial velocities. In a minimum mass solar nebula model for the early Solar Nebula (Hayashi 1981), a chondrule with mean size (D = 0.6 mm) as seen in LL chondrites is expected to drift toward the Sun a few cm s⁻¹ faster than a chondrule with mean size (D = 0.3 mm) as in H chondrites. Therefore, if one assumes that the H and LL parent bodies formed at a similar epoch, the slight shift in their formation location favors the idea of an initial turbulent concentration (Cuzzi et al. 2008) in the protoplanetary disk that produced dense zones of aerodynamically size-sorted particles. In such a case, larger chondrules would have been pushed inwards with respect to smaller ones, exactly as predicted by current models and in agreement with recent observations of protoplanetary disks (Perez et al. 2012). Interestingly, metal grains in OCs are located in between chondrules (rather than within as in CCs; see Figure 8). Their aerodynamic properties are closer to those of the smaller chondrules (Kuebler et al. 1999), which may at the same time provide a simple explanation of the relative metal enrichment of H-chondrites. The chemistry/composition of a given OC class with respect to the other two OC classes may thus have more to do with the mean size of its chondrules (and possibly metal grains) than its formation location as suggested by Zanda et al. (2006). Our results may thus highlight the importance of transport processes in the disk prior to accretion on the local compositional heliocentric gradient in the asteroid belt.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

New data presented here are near-infrared spectral measurements from 0.7 to 2.5 μm obtained using SpeX, the low- to medium-resolution near-IR spectrograph and imager (Rayner et al. 2003), on the 3 m NASA IRTF located on Mauna Kea, HI. Observing runs were conducted remotely primarily from the Observatory of Paris-Meudon, France. The spectrograph SpeX, combined with a 0.8 × 15 arcsec slit was used in the low-resolution prism mode for acquisition of the spectra in the 0.7–2.5 μm wavelength range. In order to monitor the high luminosity and variability of the sky in the near-IR, the telescope was moved along the slit during the acquisition of the data so as to obtain a sequence of spectra located at two different positions (A and B) on the array. These paired observations provided near-simultaneous sky and detector bias measurements. Objects and standard stars were observed near the meridian to minimize their differences in airmass and match their parallactic angle to the fixed N/S alignment of the slit. Our primary solar analog standard stars were 16 Cyg B and Hyades 64. Additional solar analog stars with comparable spectral characteristics were utilized around the sky. Two to three sets of eight spectra per set were taken for each object, each with exposures typically being 120 s.

Finally, reduction was performed using a combination of routines within the Image Reduction and Analysis Facility (IRAF) and Interactive Data Language (IDL). See DeMeo et al. (2009) for a full description of the procedure.

We applied a radiative transfer model to analyze and compare the mineralogies for both asteroid and meteorite samples. We first applied this model to the ordinary chondrite meteorite spectra to (1) constrain the composition of ordinary chondrites relative to petrographic measurements; and (2) to perform a direct comparison between these meteorites and their possible asteroid parent bodies.

We already applied this model to EOCs in the past (Vernazza et al. 2008). At that time, we showed the strength of our method by demonstrating that our model-fit average ratios are within 3% of their corresponding value derived from direct laboratory analysis (Hutchison 2004).

Recently, the robustness of the method—namely its ability to link a specific meteorite to a given asteroid—has been validated by the samples brought back from the asteroid Itokawa by the Hayabusa mission (Nakamura et al. 2011). In 2008, our technique had successfully predicted that the asteroid Itokawa must have a composition similar to LL chondrites (Vernazza et al. 2008).

Last, it is important to note that our model is applied in a totally consistent way to both meteorites and asteroids, which allows us to be extremely confident in any measured difference.

We applied the radiative transfer model to the newly acquired spectra of UOCs in order to constrain the relative abundance of the two main minerals present in these meteorites (olivine, orthopyroxene, see Table 2). We found that for the lowest petrologic types (<3.5), H, L, and LL meteorites have a similar ol/(ol+low-Ca px) ratio (>0.65). For petrologic types greater than 3.6, the compositional trend evolves gradually toward the more familiar compositional spread seen/known for EOCs (Dunn et al. 2010a, 2010b; Jarosewich 1990): on average 58.8 ± 4.5% for Hs; 64.2 ± 6.8% for Ls; and 75.1 ± 4.5% for LLs (Vernazza et al. 2008).

It is interesting to note that our results are supported by previous independent laboratory measurements that utilized Mossbauer spectroscopy and X-ray diffraction to quantify the modal mineralogy of UOCs. As in our case, these authors (Menzies et al. 2005) noticed a broad decrease in the olivine/pyroxene ratio in the early stages of equilibration, suggesting reduction in the least equilibrated ordinary chondrites.

These variations can possibly be understood using the bulk chemical data of UOCs (Jarosewich 1990) and were already noted by McSween & Labotka (1993) based on the same set of data. We believe these data to indicate (Figure 10) that C-induced reduction in the early stages of metamorphism provokes oxygen loss and metal formation, leading to the formation of pyroxene at the expense of olivine by the following reaction: FeMgSiO₄ + C → CO + Fe_met + MgSiO₃. This early reduction effect is more pronounced amongst Hs, which contained less oxygen and more C in their starting material, thus offering a possible explanation for the strong compositional divergence with increasing metamorphism with respect to the L and LL groups.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The dip test (mentioned in the manuscript) is actually not a direct proof of a bimodal distribution. The dip test is a proof against unimodality. A direct test of bimodality (Schilling et al. 2002) says that a mixture of two normal distributions with similar variability cannot be bimodal unless their means differ by more than the sum of their standard deviations. In our case, this criterion is largely fulfilled (see Figure 12).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We observe a correlation between the composition and the asteroid size within the H peak (see Figure 14). As detailed in Section 4.2, for all asteroids in the left peak (H-like bodies) we are seeing their exposed interiors. As such, their surfaces give us a first-order representation (insight into the composition of the external layers down to a depth of 30 km in a D = 200 km body; see Ciesla et al. 2013) of the composition of their interior. The observed correlation may thus indicate a variation of the internal composition of H-like bodies as a function of their initial sizes. In particular, the correlation may indicate that smaller H-like bodies formed with a more pyroxene-rich interior than larger H-like bodies (the data for families are in agreement with this hypothesis with Koronis members being on average more olivine-rich than Agnia or Merxia ones). In the context of the H4–H6 sequence (H4 < H5 < H6 in terms of ol/(ol+low-Ca px) ration; see Dunn et al. 2010a, 2010b) this may imply that smaller bodies were formed with a higher fraction of H4 material while larger bodies formed with more H5 and/or H6 material. Thermal models (e.g., Henke et al. 2012a, 2012b; Henke et al. 2013) could directly test this hypothesis.

Zoom In Zoom Out Reset image size

We computed the average size of asteroids in the two peaks in order to trace any initial difference in terms of mean size between the H and LL parent bodies. We selected only asteroids with D > 100 km, because smaller ones are likely the result of the collisional evolution over the course of the history of the solar system. Note that it is not implied that all D < 100 km asteroids are the results of collisional disruptions; many D = [70–100] km asteroids may indeed be primordial. In any case, the current size distribution for D > 100 km gives us a good indication of primordial differences in mean sizes between H and LL bodies.

We found that the average size within the H (left) peak is D = 132 ± 6 km, while the average size in the LL (right) peak is D = 160 ± 10 km. Our results imply that LL bodies were born, on average, bigger than H ones. Note that if we measure the average size over a wider range (for all objects with D > 60 km), this result remains unchanged (D = 94 km for the H peak and D = 119 km for the LL peak).

This Appendix contains plots (Figure 15) of the final reduced NIR spectra of S-type asteroids presented in this paper combined with available visible wavelenght spectra (Bus 1999).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size