Characterization of Exogenic Boulders on Near-Earth Asteroid (101955) Bennu from OSIRIS-REx Color Images

A small number of anomalously bright boulders on the near-Earth, rubble-pile asteroid (101955) Bennu were recently identified as eucritic material originating from asteroid (4) Vesta. Building on this discovery, we explored the global presence of exogenic boulders on Bennu. Our analysis focused on boulders larger than 1 m that show the characteristic 1-micron pyroxene absorption band in the four-color MapCam data from the OSIRIS-REx mission. We confirm the presence of exogenic boulders similar to eucrites and find that mixtures of eucrites with carbonaceous material is also a possible composition for some boulders. Some of the exogenic boulders have spectral properties similar to those of ordinary chondrite (OC) meteorites, although the laboratory spectra of these meteorites have a higher albedo than those measured on Bennu, which could be explained by either a grain size effect, the presence of impact melt, or optical mixing with carbonaceous material owing to dust coating. Our Monte Carlo simulations predict that the median amount of OC mass added to the parent body of Bennu is 0.055% and 0.037% of the volume of a 100- and 200-km-diameter parent body, respectively. If Bennu was a uniformly mixed byproduct of parent body and S-type projectiles, the equivalent mass of OC material would be a sphere with a diameter of 36 to 40 m (or a volume of 24,200 to 33,600 m3). The total amount of OC material in the interior of Bennu estimated from the MapCam data is slightly higher (91,000-150,000 m3).


INTRODUCTION
Catastrophic collisions between planetary bodies are among the most fundamental geological processes in our solar system. Collisions are thought to have created planetary and asteroidal moons and caused mass extinctions on Earth. The visible evidence for collisions is the large and small impact craters that dot planetary surfaces. Impactors have delivered exogenic materials that have enriched native regolith with minerals, volatiles, and organic material.
Our understanding of this medley of solar system materials comes from five main sources: (1) laboratory studies of brecciated meteorites and Apollo lunar samples that contain exogenic xenoliths (Baedecker et al. 1973;Brilliant et al. 1992); (2)  The surface of asteroid (4) Vesta, a 525-km differentiated basaltic asteroid, is covered with material from primitive carbonaceous impactors on a hemispherical scale, with a global average of 8 wt % of impactor materials. The howardite, eucrite and diogenite (HED) meteorites from Vesta show evidence of carbonaceous xenoliths that in some cases (e.g., howardite PRA04401) make up 60% of the host meteorite (Herrin et al. 2011). Shepard et al. (2015) proposed the presence of exogenic impactor material on large metallic (M-type) asteroids that have a prominent 3-µm absorption band indicative of hydration. These "wet" M-type asteroids are thought to contain exogenic carbonaceous chondrites rich in volatiles (OH/H2O) on a hemispherical scale. Recent observations of 252-km M-type asteroid (16) Psyche confirmed this prediction and showed that threequarters of Psyche's surface is covered by low-albedo carbonaceous material (Shepard et al. 2017;Sanchez et al. 2017;Takir et al. 2017).
Smaller asteroids also appear susceptible to exogenic contamination. The Hayabusa mission to near-Earth asteroid (25143) Itokawa observed a possibly exogenic dark boulder on the surface of this stony (S-type) asteroid (Hirata & Isiguro 2011). In addition, the impact of 2008 TC3 in 2008 delivered the Almahata Sitta meteorites, which showed that this small 4.1 m asteroid was made of 70-80% ureilite and 20-30% enstatite and ordinary (H and L) chondrites (Bischoff et al. 2010). And finally, the telescopic optical navigation camera onboard the Hayabusa2 spacecraft made observations of distinct bright boulders on the surface of carbonaceous asteroid (162173) Ryugu, and some of them have nearinfrared color spectra consistent with ordinary chondrites meteorites (Tatsumi et al., 2020).
All these lines of evidence point to a size-independent contamination of asteroid surfaces ranging from 500 km down to 4 m in size. In addition, analyses of Apollo samples show that on average, the lunar regolith contains 3.5 wt.% exogenic material (Baedecker et al. 1973) and 1-2 wt.% carbonaceous chondrite material (Brilliant et al. 1992).
The carbonaceous near-Earth asteroid (101955) Bennu, a rubble pile made from the reaccumulated fragments of a larger parent body, is the target of NASA's OSIRIS-REx (Origins, Spectral Interpretation, Resource Identification, and Security-Regolith Explorer) mission, which is bringing back a sample of regolith to Earth (Lauretta et al. 2017, 2021. Using multispectral data collected by the MapCam imager of the OSIRIS-REx Camera Suite (OCAMS) (Rizk et al. 2018),  discovered six exogenic basaltic boulders on the surface of Bennu thanks to their distinctive spectral shape and higher albedo relative to Bennu's overall dark regolith (e.g., . These exogenic boulders have the characteristic absorption band at ~1 µm due to the mineral pyroxene, which is detected in three of the four MapCam color filters. This absorption feature has been independently confirmed by observations made using the OSIRIS-REx Visible and InfraRed Spectrometer (OVIRS) (Reuter et al. 2018) that showed the complete 1-µm band, as well as the associated 2-µm absorption band expected for pyroxenes. Based on the analysis of OCAMS and OVIRS data,  concluded that the exogenic basalts, in the form of HED meteorites (in particular eucrites), were incorporated into the parent body of Bennu following their liberation from Vesta, the largest source of basalts in the asteroid belt.
In this paper, we focus on the analysis of the four-color MapCam data of a broader set of potential exogenic boulders to constrain their color properties and identify their meteorite affinities. Although our color dataset lacks the spectral resolution of the OVIRS instrument, the increased spatial resolution of our data helps avoid spectral 'contamination' from Bennu's background terrain. We analyzed color ratios and spectrally matched them with resampled laboratory spectra of meteorites to draw our conclusions.

DATA PROCESSING
We used a multi-band mosaic built with images at 25 cm/pixel from the OCAMS MapCam camera ( Figure 1A&B) to analyze the color properties of boulders . We also used a mosaic constructed from OCAMS PolyCam panchromatic images at ~5 cm/pixel (Bennett et al., 2020) to study the morphology of the boulders with higher-resolution images (Figure 2A, 2B and 2C). The absolute uncertainty for the OCAMS calibrated images is 5% and the relative uncertainty is 1% (Golish et al. 2020a). For the MapCam image mosaic, camera pointing was adjusted so that the image-to-image registration was improved to facilitate extraction of color spectra. All images were also registered to a shape model of Bennu (Barnouin et al. 2020) for the creation of Both mosaics were assembled with an equirectangular projection using a Bennu radius of 250 m and based on a high-resolution shape model of Bennu (version v28 of the shape model from Barnouin et al. (2020)) with a ground sample distance (GSD) of 80 cm. Global mosaics were created using applications from the ISIS3 software as described in DellaGiustina et al. ( , 2020b for MapCam data and Bennett et al. (2020) for PolyCam data. The color mosaic consists of four bands in the visible and near-infrared with filters b′ (473 nm), v (550 nm), w (698 nm) and x (847 nm). Boulders with a distinct shape were mapped as polygons using the MapCam image mosaic in a Geographic Information System, as described in the Supplementary Materials from DellaGiustina et al. (2020b). The median value of I/F was extracted in each of the color filters for all the 1593 mapped boulders studied in this work.

BOULDER CLASSIFICATION
The first criterion for classification of the boulders in our dataset was the presence of an absorption feature centered around 1 µm attributed to pyroxene and detected as a reflectance peak in the w filter (698 nm), Wpeak. We call the group of boulders that bear this feature PYR. If instead there is an absorption band at this wavelength (drop in reflectance in the w filter), as observed in the spectra of carbonaceous chondrites, it could be due to the presence of iron oxides in clay minerals that were formed by aqueous alteration of the parent asteroidal surface. We call the corresponding group of boulders PHY due to the possible presence of phyllosilicates. Analysis of OVIRS data by  confirmed the presence of the 1-µm band due to pyroxene for six of the boulders or clasts included in our PYR group. A formula based on the work presented in Bell and Crisp (1993) and Bell et al. (2000) was employed to compute the band depth of spectral feature at 698 nm used for distinguishing between the PYR (Wpeak > 0) and PHY (Wpeak < 0) groups. The formula is as follows: Wpeak =(Rw/(0.50168*Rv+0.49831*Rx))-1 (1) with Rw, Rv, and Rx corresponding to the reflectance values for the filter w, v, and x, respectively.
Within these two groups, boulders were further divided using the b′/v ratio values to separate boulders with b′ reflectance lower than v reflectance (PYR1, PHY1), which is typical of pyroxene-rich material, and those with b′ reflectance higher than v reflectance (PHY2, PYR2). These two last groups exhibit color spectra with an overall blue (negative) spectral slope, more consistent with the moderately blue average spectrum of the global surface of Bennu (DellaGiustina et al. 2020b).
These groups were additionally subdivided using the v reflectance level with a threshold value of 0.049. This value corresponds to the median value of the bimodal distribution observed for all mapped boulders (DellaGiustina et al., 2020b). For boulders with reflectance in the v filter lower than 0.049, we  Table   1. identified these boulders as exogenic candidates based on the detection of an absorption in the x filter and not based on Wpeak, therefore, some of them were not classified as exogenic (i.e., PHY1b) in our work.

SPECTRAL LIBRARY OF METEORITES
We compared the boulders from the subgroups PYR1a and PYR1b that have a color spectrum indicative of pyroxene to spectra of ordinary chondrites (OCs; originating from S-type bodies) and HED meteorites obtained in the laboratory and resampled to MapCam filters. Most of the meteorite data that we included in our spectral library for this study were extracted from the RELAB database, but some are laboratory spectra of more recent falls or finds ( represented as PYR1b (sites 1, 2, 3, 4, 6 with Wpeak > 10% and site 5 with Wpeak < 4%). We also evaluated the boulders' affinity with OCs or HEDs based on the proximity of data points in scatterplots of color parameters such as ratios, relative slope, albedo and Wpeak.

BOULDER DISTRIBUTION AND MORPHOLOGY
We plotted the spatial distribution of the different classes of PYR1 boulders, which are the most likely candidates for an exogenic origin, on the RGB color map of Bennu ( Figure 1A). Most of the PYR1 boulders appear to be distributed at There is also a group of boulders scattered radially around what appears to be an ancient impact crater centered at 60°W, 45°N. This pattern most likely corresponds to redistribution of boulders after an impact rather than the remaining material from the crater-forming impactor itself. Multi-meter exogenic boulders could not have been delivered directly to Bennu via impact without disrupting it, though it is plausible that Bennu was assembled from larger precursors that did go through disruption. It is more likely these boulders were delivered to Bennu's parent body by an impactor , and are therefore distributed throughout Bennu's interior. The floor of these craters might have been covered by a layer of dark regolith particles from Bennu (via mass wasting or redeposition after the impact) hiding any potential exogenic boulders inside the crater while exposing many others along the crater rim. It is interesting to note that some boulders from the PYR2 classes also show similar distribution patterns ( Figure 1B). It could be due to a bias in mapping boulders larger than a few meters that are more readily visible in crater rims than crater interiors where accumulation of regolith can occur due to mass movements. The remaining PYR1 boulders appear to be randomly scattered at the mid-latitudes. A.

B.
We PYR1b boulders are generally much brighter than the background regolith in the PolyCam and MapCam images. These boulders are also redder than their surroundings due to their higher reflectance in the w filter relative to the other filters (which is also true for the PYR1a boulders in Figure 2C). Most of these boulders appear as a single rock standing on Bennu's surface, but some seem partially buried in the regolith, whereas a few others look like clasts embedded within a typical dark boulder found on Bennu (as previously shown in DellaGiustina et al., 2020a,b). An example of this is shown in the largest image of Figure 2B; some clasts are casting a shadow on the rock underneath and could represent debris from a pyroxene-rich rock, whereas others are embedded within the matrix of the dark boulder itself. Alternatively, these clasts could represent material freshly exposed on the surface of an exogenic boulder coated with carbonaceous dust from Bennu after an impact. Some of the PYR1b boulders in Figure 2A and 2B have both bright and greyish surfaces, possibly indicating a partial dust coating. The darker pyroxene-bearing boulders from the PYR1a subgroup are shown in Figure 2C. These boulders appear slightly reddish compared to their surroundings in the RGB images but have also a much lower albedo than the PYR1b boulders; thus, they stand out less from the average dark regolith of Bennu. B. C.

COMPARISON OF BOULDER COLOR SPECTRA WITH METEORITES
With b′ < v, PYR1 boulders show the typical spectral shape of pyroxene-rich material but with a reflectance in the v filter (550 nm) ranging from 0.039 to 0.13, which is significantly lower reflectance than for HEDs (v ~ 0.3) or OCs (v ~ 0.23). This implies the existence of a mechanism that lowers the reflectance without erasing the spectral shape characteristic of pyroxene-rich material. As discussed in section 3, the PYR1 group includes the subgroups PYR1a (with v < 0.049) and PYR1b (with v > 0.049). Figure 3A shows the average four-band color spectrum for PYR1a boulders, PYR1b boulders with Wpeak > 10%, and PYR1b boulders with Wpeak < 4%, with error bars indicating the extreme values found for each filter to give a measure of the boulder diversity within each subgroup. Figure   3B has the same spectra normalized at 550 nm. All these groups show the characteristic drop in reflectance in the x filter (centered at 847 nm) due to the mineral pyroxene. They also have a positive b′-v slope, in contrast with the negative slope observed in the global average spectrum of Bennu. A. B.

C.
To constrain the meteorite analogs for these candidate exogenic boulders, we surveyed the distribution of meteorite types in the terrestrial collection available via the Meteoritical Bulletin (https://www.lpi.usra.edu/meteor/). Of the 63,900 meteorites in that collection, the two most dominant classes are the OC (85.6%) and HED meteorites (3.6%). Together, these make up 89.2% of all known meteorites, and they both have the characteristic 1-µm pyroxene absorption band observed in the color spectra of candidate exogenic boulders on Bennu. All other meteorite types that have pyroxene-dominated spectra make up ~1% of the meteorites that fall on Earth. We chose to investigate these two meteorite types 10% have a v-albedo range of 0.08-0.13, which is well within the range measured for eucrites. However, we cannot exclude the possibility that exogenic boulders are covered with a thin layer of fine dust from Bennu, which would bring down the albedo (and lower the 1-µm band depth) enough for some howardites and diogenites to be good matches as well. The only potential meteorite spectral match for the PYR1b with Wpeak < 4% is a sample of eucrite PCA82501 with a particle size range from submicrons to 1000 µm, but its 1-µm band depth is higher that of than these boulders, as is its v albedo (0.23 vs. 0.049-0.092).
We did not find a good meteorite match for the PYR1a group as all HEDs have much higher 1-µm band depths than this group. Further in-depth analysis of the HED mineralogy for a subset of exogenic boulders using the OVIRS data is  In addition to simple spectral matching, we also investigated color ratio trends to verify the affinity of these candidate exogenic boulders to HEDs. 27% of PYR1a and 44% of the PYR1b boulders are distributed within the range of eucrites in the b′/v vs. v/x plot, and for 5% PYR1b boulders, within the range of howardites as well ( Figure 5). The rest of the PYR1a and PYR1b boulders are within the range of values for v/x observed for HEDs but their b′/v ratio is slightly higher than the HEDs' range due to their flatter b′-v slope compared to HED spectra. The higher b′/v ratio could be explained by coating of the boulders with dust from Bennu for example, which would subdue the typical spectral shape for HEDs, or by a different mineralogical composition implying a different spectral shape. The affinity with HEDs is more obvious in Figure 6 (v/x vs. b′/x) where PYR1a and PYR1b boulders plot within the HED trendline and more specifically the eucrites and howardite regions.  We investigated in more detail the possibility that some boulders could be HEDs covered with a layer of carbonaceous dust from Bennu by comparing them with laboratory spectra of mixtures of eucrite (Millbillillie) and CM2 carbonaceous chondrite (Murchison). Figure 7 shows the color spectra of PYR1b (with Wpeak > 10%) boulders with the best matches from mixtures of HED and CM2 with different proportions. The v-band albedo of these mixtures (0.08-0.16) and PYR1b boulders (with Wpeak > 10%; 0.08-0.13) match. The apparent depth of the 1µm absorption feature is similar to that of some boulders for areal mixtures but too low for the intimate mixtures presented. For lower percentages of CM2 in the intimate mixtures (10-20%), the pyroxene feature better matches the boulders, but the v albedo is too high. None of the mixtures were a good match for boulders from the PYR1a group. The closet match for PYR1b (Wpeak < 4%) is the intimate mixture with 60% CM2, but higher percentages of CM2 (which are not available) would likely provide a better match with lower v-band albedo and weaker Wpeak. However, the b′/v ratio for the mixture at 60% does not match well. In summary, spectra from brighter boulders show a similar spectral shape than some of the spectra from mixtures, but most low-albedo boulders with the pyroxene absorption band (PYR1a and PYR1b with Wpeak < 4%) do not match. Intimate mixtures with other type of eucrites might help obtain a better spectral match.
This becomes more evident in Figure 8, where we see that the color ratios of HED and carbonaceous chondrite mixtures are offset from most PYR1a and -b boulders, with a slight overlap for some PYR1b boulders. This would imply that most Bennu boulders are pristine eucritic material, and that the subdued absorption bands and lower albedo relative to laboratory spectra might be a  PYR1a, a selection of three impact melt OCs (from H, L, and LL) is the closest match to this group; the Wpeak is slightly higher but is better than any HED fit that we found. PYR1a boulders have a v-band albedo averaging 0.045, which is lower than the selected OC melts (0.09-0.18). Computing areal mixtures of the selected OC impact melts with the global average spectrum of Bennu ( Figure 9B) provided a better fit to some PYR1a boulders (for example, with 10% L6 Mreira melt). Similarly, the best meteorite matches for the PYR1b boulders with Wpeak < 4% are also H, L, and LL impact melts, but the best fits to some of the boulders With the boulders selected in Figure 9, plus four additional boulders from the PYR1b sub-group that were not plotted in this figure for clarity, we   We further investigated the similarity of candidate exogenic boulders to ordinary chondrites by plotting color ratio scatterplots. Figures 10 and 11 show the ratios of b′/v and v/x color filters for the PYX1a and PYX1b boulders along with the three OC subtypes. We found that several outlying boulders in these color spaces match those of ordinary chondrites. 55% of PYR1a and 19% of the PYR1b boulders are distributed within the range of OCs in the b′/v vs. v/x plot ( Figure 10). The rest of the PYR1a and PYR1b boulders are within the range of values for v/x observed for OCs, but their b′/v ratio is slightly higher than the OCs' range due to their flatter b′-v slope compared to OC spectra. The resemblance with OCs is also observed in Figure 11 (v/x vs. b′/x) where some PYR1a boulders plot close to the OCs, and ~10% of PYR1b boulders overlap with the OC clusters of points. PYR1a and PYR1b boulders also plot close to the OC trendline. Although four-band color data from MapCam is not as diagnostic of composition as hyperspectral data from the spectrometers onboard OSIRIS-REx (OVIRS and the OSIRIS-Rex Thermal Emission Spectrometer, OTES), it provides spectral information at a higher spatial resolution than either instrument. A follow-up study using the full spectral resolution and coverage of the OVIRS instrument should be undertaken to confirm our findings.  2. Bennu's parent body may have been disrupted by an OC projectile, and some of that exogenic material may have mixed into the reaccumulated Bennu.
3. Bennu, or an immediate precursor, may have been hit by OC material during its transit from the main asteroid belt to its current orbit in near-Earth space (semimajor axis, eccentricity, and inclination of Bennu (a, e, i) are 1.126 au, 0.204, and 6.035º, respectively).
We address each scenario in sequence.
7.1. Adding S-type material to Bennu's parent body prior to its disruption Bennu's parent body was probably hit by many different types of asteroids prior to its disruption, including S-types. If we assume that many of these impactors accreted to the parent body over billions of years, some portion of this material must have mixed into the fragments produced by the parent body's disruption.
Moreover, if we assume that mixing during the parent body's disruption was 100% efficient, we can estimate the fraction of exogenic material that Bennu would have had when it or its immediate precursor was formed by this collision event.
This calculation requires knowledge of: A. The orbit and size distribution of S-type material in the asteroid belt.
B. The collision probability of these objects with Bennu's parent body.
C. The size of Bennu's parent body and the interval that it was hit by the S-type population prior to disruption.

Component A
An important yet underappreciated issue about main belt collisions is that most asteroids can strike one another, even if the target of interest is located at the inner regions of the main belt (e.g., Farinella and Davis 1992;Bottke et al. 1994). Accordingly, when we consider S-type impacts on the parent body of Bennu, we need to evaluate the entire population of S-types in the main belt, since most of them have some chance of hitting the parent body. This is accounted for in the collisional probability calculation that we will discuss in Component B.
Our initial task is to identify S-type material in the main belt and determine its approximate orbit and size distribution. We estimated the former using the Wide-field Infrared Survey Explorer (WISE) diameter-limited catalog of main belt objects (e.g., Masiero et al. 2011). The WISE catalog provides albedos for numerous asteroids with diameter D > 1 km in the main belt, but it is incomplete. Partially this is because some small portions of the main belt went unobserved over WISE's mission lifetime, but also because WISE and other asteroid surveys cannot detect main belt bodies below a threshold size that approaches a few kilometers on average. For this reason, we limited our study to asteroids with D ≥ 5 km, a population that is arguably complete within existing asteroid catalogs for the inner, central, and outer main belt regions.
Using this catalog, we assumed that S-type asteroids have albedos that broadly fit within a range between 0.1 and 0.3 (Masiero et al., 2011;Morbidelli et al., 2020). This range is slightly larger than that assumed by Mainzer et al. (2012)  by an S-type asteroid (i.e., one projectile has to hit Eulalia that has a diameter larger than the disruption threshold). Given that S-types only make up 11-16% of the main belt, this can be considered a modestly low probability event but one that cannot be ignored.
Rerunning our Monte Carlo code, and assuming Bennu is a uniform mixture of projectile and target, we find that the median contamination is between 3 and 5%, which is higher than our estimated concentration of 0.1-0.2% based on the MapCam data. We hesitate to rule out this scenario, however, because an unknown fraction of the projectile may have been lost in the disruption event.
If enough disruptor mass was lost, this scenario could readily explain observations.
Note that this fractional value of exogenic material above is high enough to suggest that ample projectile material could be returned by sample collection via the OSIRIS-REx spacecraft. The odds favor that such exogenic samples would be fragments from some C-complex bodies, which dominate the main belt. We consider this the most likely scenario, with the existence of such material difficult to detect among the predominantly B-type rocks seen on Bennu's surface.
On the other hand, we find it unlikely that such a disruption event could involve eucrites (i.e., a eucritic projectile disrupted the Eulalia parent body). The largest eucrite-like body in the main belt today, other than Vesta, is (1459) Magnya, which has a diameter of roughly 17 km (Hardersen et al. 2004).
All other V-type bodies are smaller than 10 km. Accordingly to our estimates above, a Magnya-size object could just barely disrupt the Eulalia parent body.
However, the probability that this took place is exceedingly low. It would require that some kind of singular body in the main belt larger than Magnya happened to end its life by striking the Eulalia parent body. Using the collision probabilities for S-type asteroids above, we find that the average interval for a single object to strike Eulalia is 6 × 10 13 yr, many orders of magnitude longer than the age of the solar system. 7.3. Adding S-type material to Bennu or its immediate precursor Finally, we consider scenario #3, the possibility that S-type material struck Bennu and that some of it remained on Bennu itself. This solution is difficult to quantify without a more sophisticated modeling effort because much depends on the evolutionary history of Bennu itself. Open questions are whether the precursor(s) of Bennu shed mass (and how much) as a result of rotational acceleration, whether Bennu is a Nth-generation body originating from a series of larger precursors who experienced collisional disruption events prior to leaving the main belt, and how many impacts of all sizes did these precursors receive from S-type bodies.
For simplicity, we used the asteroid disruption law from Bottke et al. (2020) to estimate the size of the projectile needed to disrupt Bennu. If projectiles strike at 5 km/s, we estimate that a 7.5-m body is capable of disrupting the 495-m-diameter Bennu. The ratio of the volumes for these two bodies is roughly 3.5 × 10 -6 . If Bennu did not disrupt, or only disrupted once, and the net mass added to Bennu by OC projectiles never exceeded the disruption mass, scenario #3 is less likely than #1 and #2 for producing S-type contamination.
A possible exception to this calculation would be that objects larger than the minimum disruptor size may have hit Bennu, and this could change the probabilities above. We will explore this possibility in future work.

SUMMARY
We investigated the color spectral properties and meteorite affinities of candidate exogenic boulders larger than 1 m on near-Earth asteroid Bennu using data from the OCAMS instrument on the OSIRIS-REx spacecraft. The key findings of our study are as follows: • We confirm the presence of exogenic boulders with meteorite affinities to eucrites from the HED group derived from asteroid Vesta, as first • Some of the exogenic boulders also have spectral properties similar to those of ordinary chondrite meteorites, although the laboratory spectra of these meteorites show a higher albedo than those measured on Bennu. We attribute this difference to particle size effects, but optical mixing with carbonaceous material due to dust coating could also produce it.
Mixtures of OCs with material having Bennu's average spectrum were also suggested for boulders smaller than 1 m by Tatsumi et al. (submitted). OC impact melts, which are relatively rare among the OCs, are a possible match for the darker exogenic boulders. Based on this study, the approximate net volume of potential OC-like material we identified on Bennu is 5353 m 3 . This is more than the amount of material (~70 m 3 ) with an HED-like affinity estimated from the six boulders found by .
• We explored several different ways for Bennu to be contaminated by OCs.
The most plausible mechanism is a collision of S-type asteroids with prospective parent bodies for Bennu. Out of 10,000 Monte Carlo trials, models indicate the median amount of OC mass added to a 100-and 200-kmdiameter Bennu parent body corresponds to 0.055% and 0.037%, respectively, of their volumes. This is slightly lower than the amount of OC-like material we have estimated on Bennu based on the MapCam data (0.1-0.2%).
We proposed several mechanisms to account for this difference. An alternative explanation for observations of OC material on Bennu would be the disruption of Bennu's parent body by an S-type asteroid and subsequent mixing.
• A more thorough investigation involving the analysis of data with higher spectral resolution and wider wavelength range from the OVIRS visible and near-infrared spectrometer, coupled with more sophisticated dynamical modeling, is necessary to confirm our findings.