IS THE LARGE CRATER ON THE ASTEROID (2867) STEINS REALLY AN IMPACT CRATER?

The Rosetta spacecraft was launched in 2004 and its primary goal is a cometary mission, arriving at comet 67/P Churyumov-Gerasimenko in 2014. However, during the cruise phase of the mission, the spacecraft has undertaken various tasks, including the observation of the asteroid (2867) Steins during a close flyby in 2008 (Schulz 2009; Keller et al. 2010). Remote observations of the asteroid took place during this close encounter (e.g., Jorda et al. 2012). Of particular interest here is the overall shape of the body combined with the observation of a large impact crater-like feature. A shape model existed preencounter based on the interpretation of light curve data (e.g., Fornasier et al. 2006, 2008; Lamy et al. 2008a), which suggested a tri-axial shape. This was confirmed by images taken during the encounter (see Figure 1), which also provided a more accurate shape model and physical dimensions, giving a set of radii of 3.41 × 2.85 × 2.21 km³ and a spherical equivalent radius of 2.63 km (Keller et al. 2010; Besse et al. 2012; Jorda et al. 2012). Based on analysis of the images and crater counting, the asteroid was considered to be a rubble-pile-like body, with an age (predicted as 0.15 Gyr or 0.49–1.6 Gyr depending on the method used; see Marchi et al. 2010) less than the collisional lifetime of a main-belt asteroid of its size (predicted to be 2.2 Gyr from Marchi et al. 2006).

Zoom In Zoom Out Reset image size

The appearance of an apparently relatively large impact crater-like feature on Steins (the crater was later named "Diamond"; see Besse et al. 2012) has attracted considerable attention. It was initially thought to be abnormally large. However, comparison with other bodies in a similar size range showed that the crater was well within the relative size range observed for other large impact craters on small bodies (e.g., Keller et al. 2010; Burchell & Leliwa-Kopystynski 2010). This removed much of the interest in the crater, and subsequent papers tend to acknowledge it as just a large impact crater (e.g., Besse et al. 2012). This crater does have some odd features, however, in that its rim is irregular in shape and indented. In Besse et al. (2012) it is suggested that this oddity concerning the rim may arise from some previous structure prior to the impact that formed the crater or a second impact.

That the collisional lifetime in the asteroid belt for an object of the size of Steins is less than the age of the solar system, combined with a young age estimate for Steins from crater counts suggests that Steins is a fragment from the catastrophic disruption of a larger body. It could be a single, monolithic fragment or a reassembled rubble pile. Based on its shape with an equatorial bulge, it has been suggested that it is a rubble pile with the YORP effect helping redistribution of material to the equator (Keller et al. 2010). However, based on numerical simulations of a possible impact origin for the Diamond crater, Jutzi et al. (2010) suggest that Steins was more likely a monolithic body before the Diamond-crater-forming impact. This impact may then have caused internal damage and fractured the body.

Here we consider another possible origin for the Diamond crater, namely that it is not an impact crater but was an original feature of Steins that arose during the fragmentation of the original parent body when it underwent a catastrophic disruption. Catastrophic disruption occurs when a body is struck at high speed by another body. The key parameter in a catastrophic impact is the energy density Q (J kg⁻¹), defined as the kinetic energy of the impactor divided by the total mass of the bodies involved in the collision. If the collision involves a relatively small Q value, the outcome is a crater on the target body. If Q is very large, the target body is broken into many small pieces. These fragments may reaccumulate into a rubble pile body or disperse and form an asteroid family—see Cellino et al. (2009) for a recent review of asteroid families. At an intermediate value of Q, referred to as Q*, the target body breaks up such that the largest surviving fragment has 50% of the mass of the original target body. There are many observations of asteroid families believed to have arisen from catastrophic disruption events. Most reports concern the size distribution of the members and these data can be used to estimate the size of the parent and place constraints on the nature of the catastrophic impact (e.g., Durda et al. 2007; Leliwa-Kopystynski et al. 2009). However, here we base our discussion not upon observations of asteroid families or simulations, but upon the results of laboratory experiments where we disrupt target bodies and examine the shape and mass distribution of the fragments.

The laboratory experiment used a horizontal two-stage light gas gun to fire a projectile at a target (Burchell et al. 1999). The projectile (here a 2.5 mm diameter, stainless steel sphere) is released from a sabot and then crosses two light curtains which are focused onto photodiodes. Passage of the projectile through the curtains interrupts the light and changes the output voltage of the photodiodes. This provides timing information at two positions with a known separation. This in turn allows a speed to be obtained which is accurate to within 1%. The range of the gun is evacuated to approximately 0.5 mbar during a shot, so the projectile does not decelerate in flight. The impact speed in the shot here was 4.54 km s⁻¹, similar to the mean collisional speed in the asteroid belt of 5 km s⁻¹ (Bottke et al. 1994). The target chamber was a cube with an internal volume of 1.7 m³ volume.

The target was a cement sphere made in the laboratory with a radius (r) of 37.5 ± 0.5 mm and a mass of 338.6 ± 0.1 grams. The cement was cured for four days before use. It was positioned in a target holder which held it with between two clamps, one above and one below the target; these define a vertical axis about which the target was spun. The spin rate at the moment of impact was 3.44 Hz (angular speed ω = 21.6 rad s⁻¹). The purpose of this rotation was to simulate an impact on a spinning asteroid. The spin rate was chosen so that if we conserve angular momentum rω, and scale the target radius up to that of an asteroid of Stein's mean size (2.63 km), then a constant value of rω would imply a 5.66 hr rotation period. This is similar to the rotation period of Steins which is 6 hrs (Lamy et al. 2008b) and indeed that of many other asteroids (see Pravec et al. 2003), although we note that most large asteroids are assumed to be rubble piles with little tensile strength. After a shot, the target fragments were collected and individually weighed and imaged.

In the shot here, given the impact speed and target mass, the Q value was 1061 J kg⁻¹. The target was significantly disrupted, with the largest fragment having a mass of 108.4 ± 0.1 g, i.e., 32% of the original target mass; this places the event in the catastrophic disruption regime. The cumulative mass distribution of the collected fragments is shown in Figure 2. A single power law does not fit the whole set of data, and at least two power laws are needed, with different slopes at low and high mass (see Figure 2). This produces a slightly "convex" shape to the distribution, consistent with the observation of Durda et al. (2007) that this is characteristic of catastrophic disruption when Q is larger than Q*.

Zoom In Zoom Out Reset image size

An image of the 12 largest fragments is shown in Figure 3. It is the fragment shown circled in Figure 3 that is of particular interest here and this is shown in more detail in Figure 4. It had a mass of (11.7 ± 0.1) g, i.e., 3.5% of the original target mass. In Figures 4(a)–(c) it can be seen that the fragment has a distinct shape with an equatorial bulge and a central crater-like feature on the upper surface which gives the upper region a flat appearance. A sketch of this shape is shown in Figures 4(d) and (e). In general it resembles the diamond-shaped appearance of asteroid Steins (Figure 1). It has a shape, which when seen from above (Figures 3 and 4), that had approximately five sides in the equatorial plane, i.e., more pentagonal than hexagonal (which Steins is). Using orthogonal axes, the tri-axial shape which describes the fragment is 19.5 ± 1.1 mm (height) by 29.4 ± 2.3 mm by 32.1 ± 0.6 mm (equatorial plane). Thus, the 3-axes are in the ratio of 1:1.51: 1.65. This compares to asteroid Steins whose axes are in the ratio 1:1.3:1.5 (Keller et al. 2010).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The mean diameter of the fragment in its equatorial plane was 29.7 ± 1.7 mm. The crater diameter was 11.9 ± 0.9 mm, with a depth of 7.5 ± 1.6 mm. The uncertainly in the depth arises because the rim is slightly irregular, giving different crater depths dependent on which point on the rim it is measured from. The key ratios which emerge are that the crater has a depth (d) divided by diameter (D) ratio of d/D = 0.63 ± 0.14, while the ratio of the crater diameter divided by body equatorial radius (R) is D/R = 0.80 ± 0.08. These values can be compared to those of the crater Diamond on the asteroid Steins, where d/D = 0.14 and D/R value = 0.68 (using the mean equatorial radius).

The fragment obtained here was entirely from the interior of the original target—it had no surface which was from the original surface of the target. Given the number of fragments obtained, it is hard to fully reconstruct the parent and thus to accurately determine the original location of the fragment with regard to the impact point. If the impact event began with the formation of a crater, and then fractures opened up across the whole body, causing it to split into multiple pieces, then we can imagine a scenario where the pit in the fragment studied here was the bottom of the initial impact crater. This is our favored interpretation of the original location of the fragment, as no other piece(s) fit into the depression on the "Steins" fragment. Such a process is shown diagrammatically in Figure 4(f).

To test this hypothesis, the fragment was examined with a scanning electron microscope. Images were obtained with a Hitachi S3400N scanning electron microscope in backscatter mode using an accelerating voltage of 20 kV. EDX spectra were collected with an Oxford Instruments "Xmax-80" silicon drift detector and "Inca" software, calibrated using a cobalt standard. The floor of the pit contained several high density fragments (see Figure 5(a)) which had a twisted, torn look typical of projectile fragments recovered after impact. The EDX spectrum of this fragment in shown in Figure 5(b), and has peaks characteristic of iron and chromium. These peaks are typical of the stainless steel projectile but not the cement target. This strongly supports the scenario shown in Figure 4(f), where the pit is the floor of the original impact crater in the parent body.

Zoom In Zoom Out Reset image size

Lateral fractures developing from partway up the sides of an impact crater are a well observed feature of impacts in brittle materials at small sizes. In targets that are large compared to the crater size (e.g., effectively semi-infinite targets) the resulting craters have a shape like an upside down fried egg, i.e., there is a deep central pit surrounded by a shallow surface spallation zone. This morphology is well known in laboratory impact studies on brittle materials: for example, impacts on ice show the effect well (see Fendyke et al. 2013 for a recent example) and the phenomenon is discussed for impacts on rocky bodies in Gault (1973) who pointed out that on Earth this spall effect is only significant for craters up to sizes of the order of 1 m. Impact experiments on (effectively semi-infinite) gabbro targets by Polanskey & Ahrens (1990) show fracture patterns beneath the craters that are suggestive of a general shape like that of Steins. Polanskey & Ahrens (1990) also reported that material from the spall region around an impact crater can have a maximum velocity ≈0.5% of the impact speed. Thus, for a 5 km s⁻¹ impact in the asteroid belt, spall material would be removed from around a crater on a body up to 47 km diameter. A more recent analysis by Holsapple & Housen (2013) suggested that, on asteroids between 5 and 20 km diameter, all craters up to 1 km in diameter will be spall dominated during their formation. If we divide the "Steins" fragment equatorial diameter found here by the diameter of the parent body we get the ratio 0.396 ± 0.023. Naively assuming this is similar for the real Steins, then using the spherical equivalent size of asteroid Steins implies a predisruption parent body of diameter of 13.3 km. In a catastrophic disruption event on such a body, it is therefore not implausible that lateral fracturing around the initial pit-like impact crater played a role in the break-up of the body.

In our approach we have studied a catastrophic impact on a spinning body and found a fragment which has similarities with asteroid Steins. It is not clear if the spin of the target had a significant role on the outcome of the impact. To explore this further, more data are being obtained at a range of impact energy densities and will be published separately. However, what we have demonstrated is that, based on morphology, a subsequent impact origin for the large crater on Steins cannot be simply assumed. In addition, whereas equatorial bulges on rubble pile bodies are usually held to arise from redistribution of material due to spin effects (e.g., see Walsh et al. 2012 for a discussion of this mechanism), the equatorial bulge seen on the real asteroid Steins could potentially also arise directly from the break-up of the parent as seen here.

Not all of the observed aspects of the asteroid (2867) Stein's are fully reproduced. For example, the real Stein's Diamond crater is not as relatively deep as the pit observed in the laboratory experiment. This could partly be due to random chance during formation in the lab experiment, and of course the crater on Stein's will have filled in somewhat after its formation. Or it could result here from a denser projectile (stainless steel) impacting a less dense target (cement), which normally results in a relatively deep impact crater: if the projectile and target had been similar densities (i.e., the impactor on asteroid Steins was another rocky body rather than a denser M type asteroid), a shallower pit may have resulted. One feature not explained in our experiment is the chain of small pit-like features observed on Stein's (see Figure 1) which Keller et al. (2010) attributes to sunken pits along fracture lines caused by the impact that made the large crater. However, Besse et al. (2012) dispute this somewhat, as not all the pit/crater alignments seen on Stein's can be explained in this fashion.

That all circular features on the surface of planetary bodies are not impact craters has long been recognized, but is often neglected. On Earth, the desire to find new impact craters has perhaps led some to prematurely label such features as craters; this is discussed in Reimold (2007). On Earth, there is a particular problem and a solution. The problem is that Earth's surface is very active, with strong resurfacing and erosive processes on geological timescales. There is also a thick atmosphere shielding the surface against smaller impactors (below ≈50 m) which decelerate or break up in the atmosphere. The combined result is a relative paucity of impact craters observed today. In parallel, many other processes can lead to shallow, circular depressions, particularly on an active planetary surface with many erosive processes (both surface and sub-surface). The risk of confusing the two possible causes for creating circular surface features thus arises. The solution is to visit the site and inspect it (via collecting surface samples and cores as well as close visual inspection of outcrops) for evidence of shock effects characteristic of a high speed impact.

For non-terrestrial bodies it is different. Impact craters are much more prevalent due to fewer erosive effects and often an absence of an atmosphere to shield the body. Therefore, once the features on the Moon had been identified (from returned lunar samples) as impact craters, it has become commonplace to accept most circular features as impact craters. There are some exceptions, with sink holes (or pit holes, aligned along a possible underlying fault line) identified as likely features on some bodies (including Steins as discussed above). However, in general a common assumption is that observed features are craters.

In the particular case of asteroid Steins, we present laboratory evidence that the assumption of a subsequent impact origin for the large crater-like pit need not be automatic—it could arise from the fragmentation of the parent body. Since it is not possible to sample the crater to look for shock effects it is difficult to rule out this other potential origin. Indeed, given that in the approach here the fragment itself originated in a catastrophic disruption event and the large pit is actually the base of the impact crater that caused the disruption, there may well be shock features present throughout the body. The question also arises if all catastrophic disruption events can produce such an outcome. We are continuing experiments with a range of Q values and it seems that the outcome here requires Q > Q*, but not excessively so, i.e., not super-catastrophic disruption, as there is a need for the base of the original impact crater to survive the disruption event. This suggests that the fragment shape found here arises from a limited range of Q values just above Q*.

We thank STFC (UK) for funding and K. Holsapple for useful comments.