4 Spectroscopic results and discussion

The obtained spectra, normalized at 5500 Å, are given in the Appendix. In these figures only one spectrum per object is given and the crosses indicate observations from the ECAS survey (Zellner et al. 1985). Most of the asteroids show similar featureless spectra, characteristic of C-, B-, F- and G-type asteroids. Two of them, 461 Saskia and 1171 Rusthawelia, display P-type spectra. In effect, Tholen (1989) also classifies the latter as P-type.

In order to compare the overall characteristic of the Themis family, a polynomial has been fitted to each spectrum and the result is presented in Fig. 2a along with ECAS spectra represented by dash lines. In this figure, spectra of a same asteroid obtained in different nights are also individually included. The figure shows a limited spread of the superficial composition among the members of the Themis family. The compositional distribution is completely contained in the distribution of the C-type class and its sub-classes B, F and G as can be seen in Figs. 2b to 2e. In these figures, the distribution of broadband spectra obtained from ECAS's survey for each of the above classes is plotted. This result, therefore, is consistent with a probable common origin of the observed asteroids except 461 Saskia and 1171 Rusthawelia which might be background objects. It must be recalled that in the region around 3.2 AU the P class objects should be abundant (Bell et al. 1989). The compositional analysis of a significative sample of Themis members, therefore, supports the dynamical/statistical indication of a common origin of the family, probably resulting from the breakup of a C-type parent body. In this scenario, the asteroids 461 and 1171 would be background objects. It is noteworthy that only 1171 Rusthawelia lies at the border of the family while 461 Saskia is contained well inside the nuclear region.

  

Figure 2: a) Spectra of the Themis family presented in this work normalized around 5500 Å and fitted by a polynomial for clarity along with ECAS data (Zellner et al. 1985) represented by the dashed lines. b) Distribution of broadband spectra of C-type objects obtained from ECAS's survey. c) Same as b) for B-type objects. d) Same as b) for G-type objects. e) Same as b) for F-type objects

Considering that a collisional origin of the Themis family can be inferred, a question remains on whether the original parent body, or its fragments, experienced some degree of aqueous alteration. This information can set some constraints on the primordial nebula and the temperature gradient, or events, during the first stages of the Solar System formation. We recall that aqueous alteration is the low temperature (< 320 K) chemical alteration of materials by liquid water (Vilas & Sykes 1996, and references therein). In order to investigate this aspect of the Themis family, aqueous alteration was searched in the spectra following the procedure described by Vilas et al. (1993, 1994). Indication of aqueous alteration, identifiable by the presence of absorption bands with minimum at 0.60-0.65, 0.7 or $0.8-09\, \mu$m, is clearly present in the spectra of 15 asteroids among the observed ones. Since these bands are sometimes very swallow, their identification is highly dependent on the noise of the spectra so that we adopted the criterion defined by Vilas et al. (1993) in which a band is identified if its depth is greater than the peak-to-peak scatter. If a more relaxed criterion is adopted, 11 more objects could also present aqueous alteration in their surface. As already pointed out by Vilas et al. (1993), in some cases the 0.65 and $0.90\,\mu$m absorption bands seem to overlap the $0.70\,\mu$m band making their identification difficult. In Figs. 3a to 3d we show the polynomial fit to the spectra of asteroids that show clear indication of aqueous alteration through the 0.7, 0.65 or $0.90\,\mu$m absorption band. The asteroids 316, 637, 846, 848, 954, 1487, 1539, 1691, 2519, 2524, and 3128 may also present indication of aqueous alteration but with a lesser degree of confidence and, therefore, are not included in these plots.

  

Figure 3: Asteroids presenting absorption bands with minimum at 0.6-0.65 (1), 0.7 (2), and $0.80-0.90\,\mu$m (3), indicative of aqueous alteration in their surfaces. Each spectrum is fitted by a polynomial to increase the identification of the bands. For each spectrum is given the identified band (1, 2 or 3) at its left and the asteroid number at its right

In Figs. 4a and 4b we present the distributions of the observed asteroids as a function of the presence or not of aqueous alteration. In these figures the triangles and squares represent objects with and without indication of aqueous alteration, respectively. The stars represent objects that may present aqueous alteration and the size of all the symbols is proportional to the diameter of the asteroids. Considering the taxonomy of the observed objects, all those classified by Tholen (1989) as F do not present indication of thermal metamorphism. This is clear in three of them: 268 Adorea, 468 Lina, and 621 Werdandi, classified as FC, CPF and FCX, respectively. Another asteroid classified as FCX, 954 Li, might or not present aqueous alteration by our analysis. These results confirm earlier suggestions (Sawyer 1991; Vilas et al. 1993; Vilas et al. 1994) that F-class asteroids may be unaltered material or that they are less or differently altered. In a recent paper, Hiroi et al. (1996) plot the UV absorption strength (the most prominent absorption band indicative of aqueous alteration) vs. IRAS diameter for the C, G, B and F asteroids. In their plot the F-type asteroids all present the smallest absorption band or the total absence of it. This fact is interesting since it can indicate a different evolution of this type of objects. In particular for the Themis family, it would imply a different degree of thermal alteration in different parts of the original parent body or a distinct origin of these objects. This last assumption seems not to be supported by the limited spread out in composition shown by our spectra.

  

Figure 4: Distribution, in the proper elements space, of the observed asteroids as a function of the presence (triangles) or not (squares) of aqueous alteration. The stars represent objects which may present aqueous alteration but with a lesser degree of confidence. The size of all the symbols is proportional to the diameter of the asteroids

Analyzing the percentage of asteroids presenting indication of aqueous alteration as a function of the diameter, a trend is apparent: it decreases with the diameter of the objects. The percentage changes from nearly 60% to 35% for asteroids with diameter greater and smaller than 50 km, respectively. This trend is expected by Vilas & Sykes (1996) if we suppose the break-up of a thermally altered parent body. In their model, the bigger fragments should retain more information on the interior of the parent body where the heating event should have been more effective. If in the 11 objects which may have aqueous alteration this fact is confirmed, then the above percentages would increase to nearly 80% and 70%, respectively, and the trend is less evident.

  

Figure 5: Spectra normalized at $0.7\,\mu$m of bakground objects from ECAS (Zellner et al. 1985), of all the asteroids observed in the present work and of the observed asteroids with diameter greater and smaller than 50 km

The above results tend to indicate that the parent body of the Themis family was thermally altered and the distribution of compositions can be attributed to fragments coming from different parts of the original body. Following Vilas & Sykes (1996), if the original parent body was altered due to a thermal event in the later phases of the Solar System formation, then all the resulting fragments from a breakup should also present signs of aqueous alteration, in particular the greatest ones. This is true for the Themis family and the percentage of asteroids presenting absorption bands due to aqueous alteration decreases with the object size, as expected by Vilas & Sykes (1996).

The dependence of the distribution of surface compositions with the diameter was further investigated. In Fig. 5 it can be seen the spectra, normalized at $0.7\,\mu$m, of the observed asteroids of the Themis family along with background objects from ECAS (Zellner et al. 1985). The figure also presents our spectra divided in asteroids with diameter greater and lesser than 50 km, respectively. As can be seen, the distribution is slightly larger for the small asteroids. It is noteworthy to point out that this result is also expected by Vilas & Sykes (1996). In their model the small asteroids would be fragments resulting from the breakup of a thermally altered large asteroid. Since these fragments would come from diverse compositional units of the original body, they will present a great degree of differences between them. On the other hand, the larger asteroids should be reaccreted bodies representing the original core displaying, therefore, a smaller degree of diversity.

In Fig. 5 we notice that the distribution of spectra of the Themis family is similar or slightly smaller than that of background objects. It must be pointed out, however, that a direct comparison is quite imprecise since the two types of spectra, ECAS and ours, were obtained with different techniques. A better comparison would be obtained with a background defined from SMASS (Xu et al. 1995) or from an analogous survey. Since there are too few spectra of this region in the SMASS, the real representation of the background is not guaranteed.

We also analyzed the distribution of IRAS albedo in the Themis family, where a slight correlation with the diameter seems apparent. In Fig. 6 it can be seen a larger dispersion of albedos as the objects' sizes decrease. In particular, higher albedos are displayed by smaller asteroids. However, it must be pointed out that the error bars associated to the individual IRAS albedos are greater as the diameter decreases. Taking into account this fact the trend is less apparent even if it still remains an indication, which is compatible with our spectra (Fig. 5). If this specific albedo distribution among family members is real it can imply in, at least, three distinct scenarios:

• A space weathering process acting distinctly depending on the size of the objects.
• The original parent body presented a slightly differentiated crust. In this case most of the smaller asteroids would be part of the crust presenting an albedo distinct from the bigger ones, which would be mostly reaccreted bodies from the nucleus. If this is the case, the nucleus would have an albedo around 0.02 to 0.07, and crust values greater than 0.08. This hypothesis would imply that lighter materials present in the crust would be more reflective than the denser materials of the nucleus.
• The collision that gave rise to the Themis family occurred between a parent-body and a projectile of slightly different composition. We recall that Marzari et al. (1995) in their model of Themis' family collisional evolution, obtained the best result with a parent-body and projectile sizes of 380 km and 190 km, respectively. Since the two bodies probably were not formed in the same region (otherwise the collision velocity would not lead to the catastrophic break-up of a so big object) different compositions can be expected. Moreover, since the projectile body was relatively large, its fragments would be a considerable part of the remnants. This hypothesis is, obviously, the most difficult to be accepted in view of the quite homogeneous results from the spectroscopic observations.

  

Figure 6: Distribution of the albedos of Themis family members as a function of their diameters. A larger dispersion is apparent as the object size decreases

Among the observed objects with a known albedo, the presence of aqueous alteration was found in both low and (relatively) high albedo objects of the Themis family. Unfortunately the number of objects with available albedos and spectra is not statistically significative to permit us to find a relation between aqueous alteration and the above scenarios.

Finally the results for the Themis family were also compared to those obtained by Di Martino et al. (1997) for the Veritas family. It must be recalled that these two families occupy nearly the same heliocentric region in the outer belt. From a dynamical point of view, the greatest difference between these families is that while Themis family presents relatively high eccentricities (from .12 to .19, proper elements) and small inclinations, Veritas family members have small eccentricities (around 0.06) but relatively high inclinations (around 10 degrees). Since the two families lie at the same heliocentric distance, it seems obvious to expect that a temperature gradient, or event, should act similarly on the two original bodies. However, a remarkable greater inhomogeneity is apparent in the Veritas family, as can be seen comparing our Fig. 5 with Fig. 3 of Di Martino et al. (1997) paper (note that the scales in the two plots are identical) Moreover, only three asteroids of this family seem to present indications of aqueous alteration. These results set the basis for several scenarios for the origin of the two families:

1)

The heating event that thermally altered Themis original body occurred after the breakup of the Veritas family. This assumption, however, is not supported by the dynamical analysis of this family. In effect, Milani & Farinella (1994) proved that Veritas should be one of the youngest families in the asteroid belt due to the chaoticity of 490 Veritas, as well as other members;

2)

Since the original parent body of the Themis family was larger than that of Veritas (around 100 km) the thermal event was more efficient in the former. This result would favor the hypothesis that the thermal event was due to the ²⁶Al decay (Grimm & McSween 1993), since larger bodies would retain more radionuclides which would generate a greater and throughout heating of the asteroid;

3)

The Veritas family was not originated from the breakup of a large body, but it would just represent a clustering of background objects. Again this assumption is not supported by the statistical clustering analysis of this family, which confirms a highly reliable Veritas family (Zappalà et al. 1995).