PHOBOS AS A D-TYPE CAPTURED ASTEROID, SPECTRAL MODELING FROM 0.25 TO 4.0 μm

During the observations, a set of 10 WAC and 11 NAC images, respectively, were obtained showing a resolved target (see Figure 4 of Pajola et al. 2012a). Due to the limited scale, 11.8 km pixel⁻¹ for WAC and 2.2 km pixel⁻¹ for NAC, it is not possible to discern any feature of the surface of Phobos. Consequently, we decided to integrate all the pixels and create a single spectrum for the entire area observed.

Using the same calibration pipeline as in Magrin et al. (2012), the bias image was subtracted from the raw data, then the resulting image was divided by the flat field image and the background sky was subtracted. We then used the aperture photometry technique to measure the flux, and its error, coming from the target. After checking that no cosmic ray hit the CCD detector, we decided to split the WAC and NAC observations into two parts. With the WAC images, we used a inner radius of 9 pixels centered on the Phobos photocenter in order to get all of the light coming from the target. We then computed the flux of the sky background in an annulus between 19 and 21 pixels from the photocenter of the target. The selection of these values was driven by the fact that we had to be close enough to Phobos to compute the sky background value, but not too close to get light coming from the satellite inside. For the 11 NAC images we used the same technique, but we selected an inner radius of 10 pixels centered on the Phobos photocenter and then we computed the flux of the sky background from an annulus at a distance between 19 and 21 pixels from the center of the observed target. After all these steps were followed, we measured Phobos I/F for both cameras, from 0.25 to 0.61 μm for the WAC and from 0.26 to 0.99 μm for the NAC. The resulting values of Phobos I/F and their respective percentage errors⁶ are provided in Table 2.

Different sets of Phobos data from the literature show different albedo (or I/F) levels. Fraeman et al. (2012) report I/F consistent with an albedo value of 0.07 reported by Simonelli et al. (1998). OSIRIS, OMEGA, and CRISM data show lower I/F levels that differ from each other. Different I/F levels are usually attributed to variations in physical or observational conditions, e.g., composition, grain size, and/or phase angle, therefore differences among the datasets could partly be due to sampling of compositionally distinct geographical regions and/or partly to different observing geometries.

Phase angle has been shown to play an important role in the albedo level of data. As shown by Hiroi et al. (2003) using laboratory spectra of the Tagish Lake meteorite (Brown et al. 2000), changing the phase angle from 0° to 30° decreases the albedo (or reflectance) by more than a factor of two. Since the Phobos observations were all taken during flybys, the phase angles range from a few degrees to several, yielding an average I/F of ∼0.028 at 0.55 μm. The geographic coverage offered by the OSIRIS data overlaps with some of the Fraeman et al. (2012) data on the anti-Mars hemisphere. Simonelli et al. (1998) do not report a significant albedo variability on Phobos. As a result, we scaled the OSIRIS I/F data to match the albedo of Simonelli et al. (1998) at 0.55 μm.

4.1.1. Phobos UV Spectrum

Various spectral analyses of Phobos have included comparisons to D-type asteroids (Pajola et al. 2012a; Fraeman et al. 2012; Lynch et al. 2007; Rivkin et al. 2002; Murchie 1999). A comparison to 624 Hektor, a D-type Trojan asteroid, is shown in Figure 2 where the albedo of Phobos from OSIRIS data is scaled to match Hektor's (Cruikshank et al. 2001).⁸ It is apparent that the overall slope and curvature of the two bodies over the common wavelength range are very similar. Also, both Phobos and Hektor lack a 3.0 μm feature that would be indicative of water in some form (Cruikshank et al. 2001; Rivkin et al. 2002).

Zoom In Zoom Out Reset image size

The similarity in spectral slope at wavelengths up to 1.0 μm, along with a lack of a 3.0 μm absorption for both Phobos and Hektor, is consistent with Phobos being a captured D-type asteroid as previously suggested by Pajola et al. (2012a) and Rivkin et al. (2002).

Previous investigations of the composition of the surface of Phobos made use of spectral matching. In this study, we undertake spectral modeling of Phobos' surface albedo using a radiative transfer code based on the Shkuratov et al. (1999) formulation of the slab approximation. The code calculates the albedo of a powdered surface from the optical constants of candidate materials. The best-fitting models, shown in Figures 3 and 4, were achieved with a simple intimate mixture (salt-and-pepper like) varying the composition, the relative amounts of the components, and the grain sizes to minimize the weighted residual sum of squares (WRSS), where

$$\begin{equation*} {\rm WRSS} = \sum \limits _{i=1}^{n}\frac{{(o_{i}-m_{i})^{2}}}{{\epsilon }_{i}}, \end{equation*}$$

in which o and m are the observed and model albedos, respectively, is the percentage error provided in Table 2 multiplied by the I/F in Table 2, and i is the wavelength. Table 3 lists the parameters characterizing the best fitting models and the corresponding WRSS, which was calculated in the wavelength interval between ∼0.25 and 1.0 μm covered by the OSIRIS data.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Model 1 best fits the OSIRIS data (see Table 3 for details) and yields a composition that includes organic carbonaceous material, serpentine, olivine, and basalt glass and is shown in Figure 3. However, the presence of organic material makes this model questionable as it produces a strong absorption at 3.0 μm (see Figure 4) not seen in previously obtained Phobos data (Murchie & Erard 1996; Rivkin et al. 2002). Since the OSIRIS data do not cover longer wavelengths, it is impossible for us to know whether the geographical region covered by this dataset might indeed include organic carbonaceous material, however, we consider this possibility unlikely. It should also be noted that the basalt glass used in the model is only a convenient endmember with a flat spectrum whose composition is consistent with the kinds of mineral found in carbonaceous chondrites and does not reflect a detection of this material on the surface of Phobos.

The OSIRIS Phobos spectrum exhibits a minimum near 0.6 μm which is not matched by our model. This absorption was first noted by Murchie et al. (2008) in the Phobos red unit and was attributed to ferric iron-bearing phyllosilicates on the surface. Fraeman et al. (2013) show that the 0.65 μm feature is present in independent telescopic observations that were centered on the Phobos red unit. Among the various candidates that might give rise to this feature, Fraeman et al. (2013) include carbonaceous materials. Our best attempt at modeling the 0.6 μm feature, at a slightly different wavelength than that mentioned in Fraeman et al. (2013), is illustrated in Model 1, where serpentine yields an absorption at 0.65 μm. Unfortunately, the wavelength of the observed band falls at a slightly shorter wavelength than the one produced in the model, making the fit still imperfect and leaving doubts as to the nature of the material that might be causing the absorption. For this reason, serpentine was not included in the second model.

Hiroi et al. (2001, 2003) in their study of Tagish Lake compare spectra of the meteorite to different kinds of asteroids and conclude on the basis of spectral matching that the best candidates are D-type asteroids. The Tagish Lake meteorite is a very friable carbonaceous chondrite with a complex structure that includes a fine-grained opaque matrix consisting of phyllosilicates, magnetite, siderite, and Fe-Ni sulfides peppered with chondrules, isolated grains of olivine, and nodules of Fe–Mn–Mg–Ca carbonates (Rauf et al. 2010; Izawa et al. 2010).

As a further and more rigorous test of the nature of Phobos, we modeled the OSIRIS data using the Tagish Lake meteorite optical constants (Roush 2003). Our best fitting model in this case (Model 2 in Table 3) consists of Tagish Lake and magnesium-rich pyroxene glass (Mg80) and is shown in Figures 3 and 4. Model 2 has a higher reduced WRSS than Model 1, probably due to a mismatch with the OSIRIS data at shorter wavelengths and lack of a band at ∼0.6 μm. However, the overall shape of the Model 2 spectrum over a large wavelength range shows a curvature and albedo level that match both the OSIRIS and the Murchie & Erard (1996) and Rivkin et al. (2002) data much better than Model 1. Both models contain complex organic molecular material, however, Tagish Lake seems to have a composition and/or molecular structure that yields a better fit, particularly in view of the fact that it is a naturally occurring material rather than a synthetic one like the tholin. This gives weight to this model, making it our most promising fit for Phobos. In Figure 5, we plot the individual components included in the two models considered in this work and presented in Table 3.

Zoom In Zoom Out Reset image size