Eileen V. Ryan, Donald R. Davis, and Ian Giblin

Planetary Science Institute

620 North Sixth Avenue

Tucson, Arizona 85705

Submitted to: ICARUS

22 October 1998

18 Pages

6 Figures

Eileen V. Ryan

Visiting Assistant Professor of Physics

Department of Physical Sciences

New Mexico Highlands University

Las Vegas, New Mexico 87701

Phone: (505) 454-3148

FAX: (505) 454-3103

Email: eryan@cs.nmhu.edu

Running Title:

This paper reports on a series of laboratory impact experiments designed to provide basic data on how simulated Edgeworth-Kuiper Belt Objects (EKOs) fragment in an impact event. Fundamental questions to be addressed are: Do porous, icy targets fragment to produce a power law size distribution, as do silicate targets? What is the impact strength of such bodies? Do collisional outcomes depend significantly upon the porosity of the targets? In September and October 1997 we completed 20 low velocity airgun shots at the Ames Vertical Gun Range (AVGR) into porous (30 - 45%) and homogeneous ice spheres using aluminum, fractured ice, and solid ice projectiles. We found that the very porous ice targets behaved just as strongly as solid ice in collision. Energy is apparently well dissipated by the void spaces within the target, such that these fragile, porous ice structures are strong under impact conditions. Therefore, it would appear that if EKOs are porous, they are not necessarily collisionally weak.

Also, our data have revealed a caveat regarding impact strengths derived using strong projectiles (e.g., aluminum, steel, etc.). The shots we performed in this study provide fundamental information on how porous ice targets fragment when hit by projectiles ranging in nature from very strong to very weak (e.g., fractured ice). We observed that if the target and projectile have the same material strength, we can expect a particular fragmentation outcome for the target body. However, in cases where the projectile is stronger than the target, more of the impacting energy is coupled into target breakup, causing increased fragmentation of that body at the same impact energy. In effect, the target responds as if it had a lower-than-normal impact strength. In cases where the projectile is much weaker than the target, more energy is instead coupled into it during the collision, and the same target material sustains less damage and thus appears stronger at a particular collisional energy. One possible explanation for this behavior is the variable

depth of penetration of the projectile for the different cases. For example, a weak object will deposit its collisional energy at a shallower target depth than a stronger projectile, resulting in less efficient energy coupling, and less overall fragmentation. Based on this, if we assume that there has not been significant heating or differentiation in the Edgeworth-Kuiper (E-K) Belt, the most applicable impact strength for the low-velocity E-K belt collisions is likely to be that derived from similar target/projectile materials impacting each other. The laboratory data from this analysis indicate that a value for impact strength >5x10⁵ erg/cm³ is appropriate for porous ice targets impacted with solid/porous ice projectiles.

Since the end of the accretion phase of the Solar System approximately 4.5 Byr ago, three distinct heliocentric regions are believed to have undergone significant collisional evolution. These are the main asteroid belt (Davis et al. 1989), the Trojan swarms (Marzari et al. 1997) and the recently discovered population of small bodies orbiting at Pluto's distance and beyond, a region collectively referred to as the Edgeworth-Kuiper (E-K) belt. In the asteroid belt, collisions occur frequently on the astronomical timescale, and the same is true of the E-K belt (Stern 1995, 1996; Farinella and Davis 1996). The E-K belt is strongly depleted in mass relative to the Uranus-Neptune zone, a condition which almost certainly did not exist during the accretion phase of the formation of EKOs. Otherwise, the timescale for growing a body the size of 1992 QB1--the first identified E-K belt object (Jewitt and Liu 1992) -- exceeds the age of the solar system. Recently, Stern and Colwell (1997) suggested that collisions may have ground down >90% of the mass of the primordial E-K belt, and this mass was subsequently removed by radiation forces, similar to one mechanism proposed for the asteroid zone. Also, Farinella and Davis (1996) argued that most E-K objects smaller than about 50 km in diameter are collisionally derived fragments rather than primordial objects from the formative era of the solar system.

Work to date on the formation and evolution of Edgeworth-Kuiper Belt Objects (EKOs) is predicated on the assumption that EKOs have a similar collisional response to that of asteroids. Yet, there are major differences between these two populations. For one thing, the mean collision speed in the E-K belt is less than 1 km/s (Stern 1996; Davis and Farinella 1997), substantially less than the 5.8 km/s mean impact speed among the main belt asteroids. Collisions then occur not as hypervelocity impacts (as is the case with asteroids), but in the subsonic regime. Many of the assumptions for hypervelocity impacts, such as the point source approximation (Housen and Holsapple 19xx; Melosh 198x) do not apply for E-K belt collisions. Furthermore, EKOs are thought to consist primarily of water ice and silicates, the latter dominantly in the form of dust, and may have a high degree of porosity.

We conducted a series of impact experiments between September 28 and October 8 1997 using the NASA Ames Vertical Gun Range (AVGR). Data were obtained for 20 low-velocity (ranging from 73 m/s to 308 m/s) airgun shots, using porous (30 - 45%) and homogeneous ice spheres as targets. In this low velocity impact regime, the nature of the impactor may influence the collisional outcome, so we used a variety of projectiles: aluminum, fractured-ice, and solid ice projectiles. (see Table 1).

Targets and projectiles for this experiment run were prepared in a cold room (kept at 5° F) near the impact chamber. Prior to impact, targets were suspended in a large (~6 ft x 6 ft x 6 ft) vacuum chamber, which is evacuated to a pressure of about 10 mbars for the impact event. Targets were shot at ambient temperature within 10 minutes of being removed from the cold room. The impact chamber floors and walls were padded to protect against secondary fragmentation of the ejecta. Three high-speed (400 frames per second) Milliken cameras are positioned at viewing windows at the side and top of the chamber, perpendicular to the direction of the incoming projectile . A NAC video camera running at 1000 frames per second gave us a real time evaluation of our shots, and provided a quick measurement of the projectile impact speed. All fragments with a mass greater than 0.5 g were collected and weighed to determine a fragment mass (size) distribution for each of the shots. Smaller fragments melted almost immediately, and could not be collected. The percentage of the target mass recovered and measured for each shot was generally from 80 - 98%. In addition to the size distribution, a mass-velocity distribution for the ejecta will also be determined by measuring fragment speeds from the filmed records of the experiments. This will be the subject of a future paper.

We used both solid ice and porous ice targets in our program. Porous ice targets have not been used in previous experimental program, however, there have been a number of impact experiments have been carried out using solid ice targets (e.g., Lange and Ahrens 1981, Cintala et al. 1985, Kawakami et al. 1983) with primarily non-ice projectiles. We included solid, ice targets in our study primarily to compare our results with those of other workers.

For the porous targets, three different structures were fabricated and labeled as follows:

· Pellet Ice targets (see Fig. 1a), formed of cylindrical ice pellets each 0.25in in both diameter and length, having a fairly uniform particle-size distribution.

· Chipped ice targets were created by molding frozen commercial water-ice chips into a roughly spherical target body. The constituent particles for this target type were less regularly formed than the pellet particles, and although still approximately similar in size and shape, had a somewhat wider spread in initial particle masses. There were almost no small particles in this distribution.

· Crushed Ice targets were designed to give the constituent particles a form that mimicked the experimental mass distribution's character: a power-law dependence. This was accomplished by randomly crushing the chipped-ice particles so that small masses were far more numerous than large ones. We called this target type "crushed" ice (Figure 1b), it will be most closely related to a future end-member target design, the previously mentioned "snowball" target constructed from highly comminuted ice chips.

In all cases the constituent ice particles were formed into targets by allowing them to melt slightly in room temperature air before re-freezing them in the target mould. This was necessary in order to produce targets with sufficient mechanical strength that they could be suspended in the target chamber. Porosity for all three target types described above ranged from 30 to 45%.

For comparison with earlier impact experiments using ice targets, we used aluminum projectiles. However, a more realistic collisional set-up should involve icy bodies impacting icy bodies, so we endeavored to shoot ice projectiles with the AVGR airgun. A major experimental problem was keeping the ice projectiles intact through launch. In our first shots, as shown on our film and video records, the projectile broke up in the gun barrel and delivered a spray of fragments to the target. By reducing the air pressure in the gun and ensuring that the projectile was entirely seated within the sabot to reduce shear stresses, we were able to launch intact projectiles. Fortunately, the projectile spray shots could be used to test another variation in collisional condition: the response of our porous targets to a fractured (and presumably much weaker) ice projectile.

Rock targets exhibit several different modes of fragmentation depending on the impact speed of the projectile (Fujiwara et al. 1977, Matsui et al. 1982, Takagi et al. 1984). Core shattering occurs in the high velocity regime, while in low-velocity impacts, cone-type shattering or longitudinal splitting is commonly observed (see Fujiwara et al 1989 for a more complete discussion of fracturing modes), Neither of these fragmentation modes had been previously observed for homogeneous, ice targets. Instead, these bodies were reported to break into a few large pieces (Fujiwara et al. 1989).

Several of our shots (No.'s 971011, 971012, and 971025) with homogeneous ice spheres displayed what appears to be a combination of cone-type/longitudinal splitting fragmentation mode (see Figure 2) with fractures originating from the impact point and extending through the target. It is not clear that this fracture pattern has the same origin as cone-type shattering in silicate targets.

The largest fragment mass (normalized to the initial target mass) is shown as a function of the total collisional specific energy Q in Figure 3 for the ice targets of various structures using different projectile types. In general, our results show that the fragmentation outcomes are strongly dependent on the type of projectile used.

A convenient way to characterize collisional outcomes is through the impact strength of the target material. Impact strength is defined as the total kinetic energy per unit volume of the target needed to produce a largest fragment that contains 1/2 of the target mass (see Fujiwara et al. 1977, Davis and Ryan 1990, and Ryan et. al. 1991). We determined the impact strength of the solid ice targets hit by aluminum projectiles as about 1x10⁵ erg/cm³, similar to that found by previous workers (Lange and Ahrens 1981, Cintala et al. 1985, Kawakami et al. 1983). When the same solid ice target was hit with an ice projectile, either solid or broken, there was much less damage, implying a higher impact strength for this case. Our data also show that the porous ice target behave collisionally as if they were as strong, or nearly so, as silicates. Energy is apparently well dissipated by the void spaces within the target, such that these fragile, porous ice structures are strong in collision. Ryan et al. (1991) also found this to be true for porous, preshattered rock targets. Thus, one of the more interesting results of this study is that even though the very porous ice targets should have a static material strength well below that for solid ice, they actually behave just as strongly as solid ice. So it would appear that if EKOs are porous (at least, on a macroscopic scale as in our experiments), they are not necessarily collisionally weak. A significant result from our data is that there is systematically more collisional damage to the icy targets at the same specific energy as the projectile type varies from aluminum to solid ice to fractured ice. This suggests a major difference between the low impact velocity shots that characterize the E-K belt and the high velocities that are found in the asteroid belt. In the latter case, the so-called point source approximation holds; where the nature of the projectile is relatively unimportant to the collisional outcome. Only the energy delivered by the projectile is important. When a solid or porous ice target is hit with a solid ice projectile, its impact strength increases to about 5x10⁵ erg/cm³, a factor of 5 larger than that found when using Al projectiles. When a porous target is hit with fractured ice its apparent impact strength approaches that found for rock, ~2x10⁶ erg/cm³.

These results confirm work done previously on collisional energy partitioning between target and projectile at low velocities. Ryan and Davis (1991) noted that if the target and projectile have the same material strength, the incoming energy is partitioned equally between them. However, as the projectile strength increases relative to that of the target, an increasing fraction of the collisional energy goes into target breakup, approaching 90% for extreme strength differences. In effect, the target responds as if it had a lower than "normal" impact strength. One possible explanation for this behavior is the variable depth of penetration of the projectile for the different cases. A weak object will deposit its collisional energy at a shallower target depth than a stronger projectile, resulting in less efficient energy coupling, and less overall fragmentation. Based on this, if we assume that there has not been significant heating or differentiation in the E-K Belt, the most applicable impact strength for the low-velocity E-K belt collisions is likely to be that derived from similar target/projectile materials impacting each other. The laboratory data from this analysis indicate that a value for impact strength >5x10⁵ erg/cm³ is appropriate for porous ice targets impacted with porous ice projectiles.

A basic question in our program is: What is the relationship between the pre- and post-impact size distributions? One possibility is that energetic collisions reduce the target structure back to the original particles from which the body was originally constructed. Figure 4 compares the mass distribution of the initial particles with the fragmental size distribution for shots 971019 and 971020. It is clear from this figure that while the size distribution for the constituent particles influences the ejecta size distribution, it also contains many particles both larger and smaller than the original building blocks.

Bearing in mind our fabrication technique of "gluing" the particles together with ice, it is not surprising that some members of the size distribution should be moved upward (that is, toward larger fragments) as members of the initial population become bound together by freezing. So we cannot necessarily attribute the increased number of large fragments to any impact process. However, this would not explain the addition of small fragments as seen here (indeed, it works against any small fragments) -- so the comminution of initial particles into small fragments must be a result of the impact.

The fragment size distributions shown in Figure 4 exhibit a feature found in other impact experiments, namely a two segment power law representation. However, the physical basis for the presence of a 'knee" is probably quite different in the experiments discussed here. Here, the change is slope indicates a memory of the constituent particle size distribution. The influence of this initial distribution is seen where the inflection point for the post-impact fragment masses occurs at the same mass as in the initial particle distribution.

In Figure 5, target/projectile material combinations are fixed, with a similar projectile material impacting a similar target material. We see that the fragmentation trends are directly related to the collisional energy: fi gets smaller when energy is increased, and the steepness of the mass distribution slope increases with increasing fi . Figure 6 shows the wide range of collisional outcomes that result when having different projectile/target combinations at the same collisional specific energy Q. Results show no similarities to one another, confirming that fragmentation outcome is not simply governed by the energy of impact, but is more complexly related to the material properties of the bodies involved.

When ice projectiles are used to shatter ice targets at low impact speeds (hundreds of m/s), the impact strength is found to be variable, depending on the nature of the target and the type of projectile. For porous bodies impacted by ice projectiles, the impact strength is about 5x105 erg/cm³ for laboratory sized bodies. This value is higher by a factor of about 5 from the impact strength inferred using aluminum projectiles into solid ice targets.

While some "memory" of the initial (constituent particle) size distribution was retained in a few cases of ice target disruption, this size distribution is modified in the disruption process, as more small particles are formed by fragmentation and larger fragments are possibly formed by agglomeration.

Capaccioni, F., P. Cerroni, M. Coradini, P. Farinella, E. Flamini, G. Martelli, P. Paolicchi, P.N. Smith, and V. Zappala. Shapes of Asteroids Compared with Fragments from Hypervelocity Impact Experiments. Nature 308, 832-834, 1984.

Capaccioni, F., P. Cerroni, M. Coradini, M. Di Martino, P. Farinella, E. Flamini, G. Martelli, P. Paolicchi, P.N. Smith, A. Woodward and V. Zappala. Asteroidal Catastrophic Collisions Simulated by Hypervelocity Impact Experiments, Icarus 66, 487-514, 1986.

Stern, S.A. 1995, Astron. J. 110, 856-868.

Stern, S.A. 1996, Astron. Astrophys. 310, 999-1009.

Farinella, P., and D.R. Davis 1996, Science 273, 938-941.

Davis, D.R., and P. Farinella 1997, Icarus 125, 50-60.

Stern, S.A., and J. Colwell 1997, Astron. J. 114, 841-849.

Lange, A., and T. Ahrens 1981, LPSC XII 1667-1668.

Cintala et al. 1985, LPSC XVI 129-130.

Kawakami et al. 1983, JGR 88, 5806-5814.

Ryan et al. 1991, Icarus 94, 283-298.

Ryan, E., and D.R. Davis 1994, LPSC XV, 1175-1176.

Fujiwara et al. 1989, In Asteroids II, 240-265.

Figure 1. a) Finished pellet target and constituent particles. b) Finished chipped ice target prior to impact. c) Finished crushed ice target prior to impact.

Figure 2. Cone type breakup in ice targets. a) Fragmentation outcome for the solid ice target (shot 971025), showing distinct cone-type fragmentation. b) Series of impact frames for shot 971025 (solid ice target) confirming cone-shaped fragment formation. Each frame is 0.02 seconds apart.

Figure 3. Results on the mass of the largest fragment (fi ) vs. collisional specific energy (Q) for our new results on porous ice targets compared with fits to data from a database of collisional outcomes for ice, silicates), and cooled iron meteorite targets.

Figure 4. A comparison of the initial and fragmental size distribution for two types of porous targets. a) Cumulative mass distributions for the constituent particles used in shots 971019 (chipped ice target aluminum projectile at 308 m/s), and b) 971020 (pellet target, aluminum projectile at 304 m/s).

Figure 5. Cumulative mass distributions for four shots all of which had a crushed ice target and a broken ice projectile, but having different Q's.

Figure 6. Cumulative mass distributions for like target/projectile materials and different target/projectile materials, but all at about the same collisional specific energy.

Tables

Figure 1a

Figure 1b

Figure 1c

Figure 2a

Figure 2b

Figure 3

Figure 4a

Figure 4b

Figure 5

Figure 6