sign in sign up
<<
Home , Merrillite, Mesosiderite

& Planetary Science 36, 869-88 1 (200 1 ) Available online at http://www.uark.edu/meteor

Formation of by fragmentation and reaccretion of a large differentiated

EDWARDR. D. SCOTT'*, HENNINGHAACKlt AND STANLEYG. LOVE2

IHawai'i Institute of Geophysics and Planetology, School of Ocean and Earth Science and Technology, University of Hawai'i at Manoa, Honolulu, Hawai'i 96822, USA *Mail Code CB, NASA Lyndon B. Johnson Space Center, Houston, Texas 77058, USA ?Present address: Geological Museum, University of Copenhagen, Oster Voldgade 5-7, DK-I 350 Ksbenhavn K., Denmark *Correspondence author's e-mail address: escottahigp.hawaii.edu

(Received 2000 October 26; accepted in revisedform 2001 March 5)

Abstract-We propose that mesosiderites formed when a 200-400 km diameter asteroid with a molten core was disrupted by a 50-1 50 km diameter projectile. To test whether impacts can excavate core iron and mix it with crustal material, we used a low-resolution, smoothed-particle hydrodynamics computer simulation. For 50-300 km diameter differentiated targets, we found that significant proportions of scrambled core material (and hence potential metal material) could be generated. For near-catastrophic impacts that reduce the target to 80% of its original diameter and about half of its original mass, the proportion of scrambled core material would be about 5 vol%, equivalent to -1 0 vol% of mesosiderite-like material. The paucity of in mesosiderites and the lack of metal-poor or -rich from the mesosiderite body probably reflect biased sampling. Mesosiderites may be olivine-poor because mantle material was preferentially excluded from the metal-rich regions of the reaccreted body. Molten metal globules probably crystallized around small, cool fragments of crust hindering migration of metal to the core. If mantle fragments were much hotter and larger than crustal fragments, little metal would have crystallized around the mantle fragments allowing olivine and molten metal to separate gravitationally. The rapid cooling rates of mesosiderites above 850 "C can be attributed to local thermal equilibration between hot and cold ejecta. Very slow cooling below 400 "C probably reflects the large size of the body and the excellent thermal insulation provided by the reaccreted debris. We infer that our model is more plausible than an earlier model that invoked an impact at -1 km/s to mix projectile metal with target silicates. If large impacts cannot effectively strip mantles from asteroidal cores, as we infer, we should expect few large eroded to have surfaces composed purely of mantle or core material. This may help to explain why relatively few olivine-rich (A-type) and metal-rich asteroids (M-type) are known. Some S-type asteroids may be scrambled differentiated bodies.

INTRODUCTION dunites and rarer anorthosites. The largest clasts are fragments of coarse-grained orthopyroxene, olivine, gabbro and basalt Mesosiderites, which are composed of roughly and are -10 cm in length (e.g., McCall, 1966; Ikeda et al., equal proportions of silicate and metallic Fe,Ni plus troilite, 1990; Mittlefehldt, 1980). Chemical, mineralogical and are among the most perplexing differentiated meteorites known. isotopic studies suggest that all these clasts were derived from About 45 mesosiderites are known including samples from the diverse levels of an igneously stratified asteroid that was similar 1-5 km diameter Eltanin asteroid, which was one of the larger but not identical to the of the , objects to hit the Earth in the last two million years (Gersonde and (the HED body), probably Vesta (Mittlefehldt et al., 1997). Eltanin samples were identified in impact ejecta ef al., 1998). Metallic Fe,Ni in mesosiderites is mostly in the found in sediment cores from the southeast Pacific Ocean (Kyte form ofmillimeter or submillimeter grains, which are intimately and Brownlee, 1985). mixed with similar-sized silicate grains (Fig. 1). Metal-troilite- The silicate fraction of mesosiderites consists of mineral silicate textures indicate that except for rare metallic nodules, and lithic clasts in a fine-grained fragmental or igneous matrix the metal crystallized after being mixed with silicate (Floran (see Mittlefehldt et al., 1998). The lithic clasts are largely 1978; Hewins, 1983). Metal in mesosiderites is remarkably basalts, gabbros, and pyroxenites with minor amounts of homogeneous and contains roughly chondritic proportions of

869 PYelLtde preprint MS#4453 0 , 2001. Printed in USA. 870 Scott et al.

FIG. 1. Photograph of a 16 x 10 cm slice of the Emery mesosiderite showing an intimately mixed matrix of grains of metallic Fe,Ni (white) and silicates (gray), which are millimeter or submillimeter in size. Embedded in this matrix are a few volume percent of subcentimeter-sized silicate clasts, one rounded, centimeter-sized, metal nodule and a very large pyroxenite clast. (American Museum of Natural History sample number 444 1 ; see Floran, 1978.)

siderophiles, unlike iron meteorites, which show large chemical range of 39Ar-40Ar ages, 3.74.1 Ga (Bogard et al., 1990; variations due to fractional crystallization (Hassanzadeh et al., Bogard and Garrison, 1998). (These Ar-Ar ages and those in 1990; Shen et al., 1998). Since impacts do not efficiently melt Table 2 have been corrected for the -300 Ma error discovered asteroidal material and impact melt is largely lost (e.g.,Keil et by Bogard and Garrison (1998).) The 39Ar-40Ar ages of al., 1997), it is most likely that the metal was molten prior to mesosiderites have been interpreted in two different ways. A impact and that it crystallized when mixed with silicate. catastrophic impact may have disrupted the parent body at Various techniques have been used to constrain the cooling 4.2 Ga, heated the mesosiderites and left them deeply buried rates of mesosiderites over different temperature ranges (Bogard et al., 1990; Mittlefehldt et al., 1998). Alternatively, (Table 1) and these are now broadly consistent with one another. the 39Ar-40Ar ages may simply result from very slow cooling In the temperature range, 850-1 150 "C, mesosiderites cooled at depth in a large asteroid (Haack et al., 1996a). relatively rapidly at 104 to 105 "C/Ma, but below 500 OC they There are many puzzling features about mesosiderites. cooled at <0.5 "C/Ma, more slowly than any other meteorites. Why are crustal and core materials so abundant in mesosiderites Chronological constraints on mesosiderite history were yet fragments of olivine mantle are rare? Why are metal and reviewed by Mittlefehldt et al. (1998) and are summarized in silicate grains so intimately mixed? Why were mesosiderites Table 2. The mesosiderite parent body was globally cooled rapidly at high temperature yet very slowly at low differentiated at 4.56 Ga (Ireland and Wlotzka, 1992), -2 Ma temperatures? Although a wide variety of origins have been after the HED body (Wadhwa et al., 1999). From the Sm-Nd proposed (Hewins, 1983), most recent authors infer that molten ages of 4.42-4.52 Ga of four lithic clasts, Stewart et al. (1994) metal and silicate were first mixed by impact in the regolith of infer that clasts were mixed with molten metal -100-150 Ma a Vesta-like asteroid to form mesosiderite breccias that cooled after the solar system formed. Mesosiderites have a narrow rapidly. After many localized impact melting events, the Formation of mesosiderites by fragmentation and reaccretion of a large differentiated asteroid 87 1

TABLE1. Thermal history of mesosiderites. ____ Feature Temperature Cooling rate Reference ("C) (OC/Ma)

Pyroxene exsolution > 1 OOO? 5.104 Miyamoto and Takeda (1 977) Pyroxene zoning 850-1 I50 2104 Ganguly et al. (1 994) I 05 Schwandt et al. (I 998) Plagioclase overgrowths 1100 2105 Ruzicka et al. (1 994)

Taenite zoning 400 -0.03 Haack et al. (1 996a) zoning 400 0.2 Hopfe and Goldstein (2001) zoning 400-300 -0.01 Haack et al. (1 996a) Fe-Mg ordering in pyroxene 250 51 Ganguly et al. (1994) 39Ar-4oAr ages <300 -0.2 Bogard and Garrison (1 998)

TABLE2. Chronology of mesosiderites and comparison of two models for the evolution of the parent asteroid.

Age Source Comparison of models: (G4 (A) This work; (B) Rubin and Mittlefehldt (1993) and Mittlefehldt et al. (1998)

4.56 Pb/Pb age of clast (I) (A) and (B): Global differentiation and formation of original crust. 53Mn-53Cr systematics (2)

4.42-4.52 Sm-Nd ages of four clasts (3) (A) After crustal remelting, an impact at -5 km/s mixed molten core with cooler crust and mantle. Metal-silicate mixtures were buried deeply in the reaccreted body, cooling to -800 "C within lo4 years. Slow cooling below 400 "C started after local thermal equilibration. (B) Projectile with molten core impacted at 1 kmls causing crustal remelting. Metal-silicate mixtures resided near surface of body.

4.42- 4.2 (A) Slow cooling at depth. (B) Many impacts caused localized melting and heating of near-surface metal- silicate mixtures followed by rapid cooling.

3.7-4.1 39Ar-4oAr ages of mesosiderites (4, 5) (A) Ar closure during slow cooling at depth. (B) An impact at 4.2 Ga caused disruption and reaccretion of body, Ar degassing, deep burial of mesosiderites, and the start of slow cooling.

~ ~ References: (1) Ireland and Wlotzka (1992); (2) Wadhwa et al. (1999); (3) Stewart et al. (1 994); (4) Bogard et al. (1990); (5) Bogard and Garrison (1 998). mesosiderites were deeply buried by an impact that disrupted Here we explore whether mesosiderites could have formed and reassembled the parent body, causing the Ar-Ar ages to be by impact-induced break-up and reassembly of a Vesta-like reset and the mesosiderites to cool slowly at low temperatures asteroid with a molten core (Haack et af., 1996a; Scott et af., (Bogard et al., 1990; Rubin and Mittlefehldt, 1993; Mittlefehldt 1996). Petrologic studies of several types of and et al., 1998). In this scenario, the silicates in mesosiderites differentiated meteorites suggest that many parent bodies were derived from the regolith of the target asteroid and the experienced catastrophic impacts that mixed together materials metal from a neighboring differentiated projectile that collided from diverse depths (Keil et af., 1994). In the case of the IVA at a relatively low velocity (- 1 km/s) (Wasson and Rubin, 1985; iron and stony-iron meteorites (Haack et al., 1996b) and the Hassanzadeh et al., 1990). However, Haack et al. (1996a) IAB iron meteorites and (Benedix et al., 1996, argued that mean impact velocities at the time of metal-silicate 2000), catastrophic impacts appear to have mixed molten metal mixing would probably have been close to 5 km/s and suggested and silicate materials. Theoretical and experimental studies that metal and silicate were mixed together in a fragmentation suggest that impacts converted asteroids larger than about a and reaccretion event. kilometer in size into rubble piles (e.g.,Love and Ahrens, 1996; 872 Scott et al.

Ryan, 2000), and the low densities inferred for many asteroids experiments for targets with metallic cores to check whether support this concept (Wilson et al., 1999). Here we investigate impacts can excavate core iron and mix it with crust material. whether such catastrophic impacts can scramble differentiated asteroids and whether such a process might account for the Numerical Experiments strange properties of mesosiderites. We used a smoothed-particle hydrodynamics (SPH) FRAGMENTATION AND REACCRETION MODEL computer simulation. The SPH method (Lucy, 1977; Gingold and Monaghan, 1977) models a continuous medium with a We have investigated the possibility that metal-silicate grid of discrete particles whose physical properties are mixing in mesosiderites occurred during disruption and mathematically "smoothed" or blurred together. It is a robust, reassembly of an asteroid with a molten core (Fig. 2). Our proven method for modeling high-speed impacts (e.g., Benz thermal models confirm that an asteroid several hundred and Asphaug, 1994). Our three-dimensional SPH code (Love kilometers in diameter that accreted and melted at 4.55 Ga could and Ahrens, 1996) runs on a Cray YMP supercomputer and have retained a molten core until 4.454.4 Ga, as required in treats collisions between two spherical asteroids. In each this model, without the necessity for invoking extra insulation simulation, the smaller of the two collision partners-the from a regolith. Such a large body is also consistent with the projectile-is a stone body modeled with the Tillotson equation conclusion of Wilson and Keil(l991) that only bodies 100 km or of state parameters for granite (Tillotson, 1962; Melosh, 1989). more in diameter could have retained basalts on their surfaces. (The granite equation of state is commonly used to represent Love and Ahrens (1996) showed that catastrophic impacts generic stony material in impact calculations.) The larger, could exhume a significant amount of material from the core differentiated target body is modeled with the Tillotson of homogeneous stony targets. But for differentiated targets, equations of state for iron in the core and granite in the mantle. the shock impedance mismatch at the core boundary and the In each case, the core occupies 8.75 vol% of the target (44% greater density (hence stronger gravitational cohesion) of the of the target radius), the impact speed is 5 km/s (typical for core makes applying the conclusions based on homogeneous main belt asteroids) and the impact angle is 45", the most likely targets questionable. We report results from comparable value.

Basalt Dunite 0 Pyroxenite Molten Iron-Nickel Metal-silicate

FIG.2. Schematic diagrams to illustrate our model for mesosiderite formation. (a) Layered differentiated asteroid with a molten core; (b) the same body after a hypervelocity collision with a projectile that mixed metal and silicate to generate globs of mesosiderite material, which were deeply buried in a breccia of rock fragments. Formation of mesosiderites by fragmentation and reaccretion of a large differentiated asteroid 873

For the 50-300 km target diameters treated here, gravity- Final diameter (km) not material strength-controls how the target reacts to the 300 275 250 225 collision (e.g.,Love and Ahrens, 1996; Benz and Asphaug, 1999), so we neglect strength effects and treat the entire medium Scrambled as a cohesionless gravitating fluid. We numerically model the hydrodynamic phase: contact, compression, and shock propagation to the antipode, which takes seconds to minutes. -2 60 - We treat the later ballistic flight of particles analytically. After m the hydrodynamic phase, we determine the fate of each particle by comparing its center-of-mass reference frame kinetic energy (K)with the gravitational binding energy (W)between it and the rest of the mass in the simulation. Particles with K > I WI escape to interplanetary space; the rest remain to form the Core remnant target "rubble pile". Note that we followed the t trajectories of the particles with the numerical model for no 0 100 200 300 400 500 more than a couple of minutes, which is much less than total Specific Impact Energy (J/g) time for reaccretion. Most of the body reaccretes within hours to days, but some particles take a year to reaccrete (Love and Ahrens, 1996). Our simulated targets are represented using - 1700 hydrocode particles, giving a resolution of 15 particles across the target diameter. The projectiles contain 7 to 57 particles of the same size as the target's. In all but one case, all of the projectile particles escape and are not retained on the final target. In the case of the largest projectile, one particle is retained which occupies 0.1 vol% of the final target. Love m Mantle and Ahrens (1 996) note that the possibility that more projectile s - material might be retained in a higher resolution model cannot % 40 be excluded. Nevertheless, it is clear that accretion of projectile z Scrambled material is not efficient under the conditions considered here, 20 - Core \ consistent with the inference that all of the mesosiderite silicate - - (7- - \ material could have been derived from a single differentiated ,Core . I I\ body. The limited number of particles or, equivalently, the model's low resolution, is the greatest uncertainty in our treatment. Resolution testing suggests, however, that the errors are not FIG. 3. Results of low-resolution numerical modeling experiments significant, and recent independent work (Benz and Asphaug, showing the proportions of core, scrambled core, mantle and 1999) at much higher resolution agrees very well with earlier scrambled mantle in the final body as a function of the specific impact energy, which is the projectile kinetic energy per unit target mass, results from our code. Accordingly, we believe our model is for targets with diameters of (a) 300 km and (b) 100 km. The reliable enough to support a semi-quantitative check on our projectiles impact at 5 km/s at a 45" impact angle. Scrambled particles concept for the formation of mesosiderites. are those that are launched with sufficient kinetic energy to reaccrete To investigate impact excavation of core material, we note anywhere on the final rubble pile. For both targets, impacts that that particles with IWI > K > 0.5 IWI do not escape, but are reduce the diameter of the target by 20% and halve its mass generate launched on suborbital trajectories and can reaccrete anywhere about 5 vol% of scrambled core material in the final body. We infer that such impacts could form about 10 vol% of mesosiderite-like on the final target remnant. Especially, they may land alongside material. cold crustal rock. We call these particles "scrambled," and use scrambled core particles as a proxy for mesosiderite metal. entirely. For impacts that reduce the target to 90% of its original Our results for the 50-300 km diameter differentiated size, or about 73% of its initial mass, the proportion of targets (Table 3, Fig. 3) show that significant proportions of scrambled core material in the final target remnant is only about scrambled core material (and hence potential mesosiderite metal 1 ~01%.However, for near-catastrophic impacts that reduce material) can be generated. These results demonstrate that the target to 80% of its original size and about halfof its original nearly catastrophic impacts on differentiated asteroids can mass, the proportion of scrambled core material is about 5 ~01%. dredge up significant amounts of core material and mix it If this scrambled metal mixes with an equal volume of the throughout the final rubble pile without destroying the target appropriate rocky materials, about 10 vol% of mesosiderite- 874 Scott et al.

TABLE3. Results of modeling impacts on a differentiated target asteroid.*

Target Projectile Final target Final target Final target diameter diameter diameter core scrambled core (km) (km) (km) (vol%) (vole/,)

300 79 268 12 0.83 300 95 242 16 4.9 100 16 92 10 1 .o 100 22 80 11 5.8 50 8 43 9.2 I .7

~ ~~

*All runs at 5 km/s and 45" impact angle. Target initially contained a core occupying 8.75 ~01%.

like material may form after an impact that halves the asteroid's Ahrens, 1996). For differentiated targets, the temperature mass. Higher proportions are formed in more catastrophic increases are higher but still insignificant. events (Fig. 3). Some mesosiderite matrices have igneous silicate textures Our results also show that catastrophic impacts are and resemble matrices of clast-laden impact melts (Floran, inefficient at stripping mantles from cores of asteroids or 1978; Hewins, 1984; Hewins and Harriott, 1986). These enriching them in core metal (Love and Ahrens, 1996). For authors inferred that the mesosiderites had formed by impact the 100 km target (Fig. 3b), there is only a small ircrease in melting but we suggest that molten metal-not shock heating- the proportion of core material in the final target, even for supplied the heat to locally melt silicate. If the silicates had collisions that remove half the target's mass. For larger targets, been shock melted, shocked clasts should be present, but there core material is retained with slightly greater efficiency, as Love are only minute amounts of shocked material in mesosiderites, and Ahrens ( 1996) inferred. Nevertheless, the proportion of even in those with igneous-textured matrices (-0.1% shocked core material increases from 9 to only 16 vol% for an impact above 15 GPa; Haack et al., 1996a). that removes half the mass of a 300 km asteroid (Fig. 3a). Mesosiderites cooled through 1 150 to 850 "C at -0.1 "C/year, Unfortunately, model limitations currently prevent us from much more slowly than clasts in shock-formed impact melts. doing much more than proving the concept. The detailed In the temperature range 400-250 "C, they cooled exceptionally histories of the scrambled core particles could not be followed. slowly at <0.5 "CMa (Table 1). Although the early thermal In the future, it would be instructive to improve the model's history of individual mesosiderites may have been very resolution, to model three-layer (liquid iron core, plastic olivine complex because of the diversity of sizes and temperatures of mantle, and rigid basalt crust) targets, to track the degree of the components in metal-rich and metal-poor regions, we infer damage and fragment size distribution in the silicate parts of that the cooling rate changed drastically between 850 and the target, to explore the size distribution of liquid iron globs 400 "C. In our model, we attribute rapid cooling at -0.1 "C/year in the ejecta, and to estimate the ballistic flight time of each in the range 1150-850 "C to thermal equilibration between particle and thereby model the degree of mixing and burial hot and cold regions of the reaccreted asteroid. This cooling depth of the reaccreted materials. These improvements would rate suggests that the hot regions were about a few hundred yield predictions that could be checked against the meters in size. The exceptionally slow cooling below 400 "C mesosiderites to hrther test the model's validity. is attributed to the large size of the reacccreted body and the superior thermal insulation provided by break up and Thermal History of Mesosiderites reaccretion. We equate the transition temperature of 450-800 "C with the equilibration temperature of the reaccreted body. To test whether our model is consistent with the thermal The equilibration temperature of the reaccreted body is not history of mesosiderites (Table l), we need to understand the well constrained because the thermal profile of the target processes that control the thermal history of diverse types of asteroid prior to impact and the temperatures of the dispersed ejecta in a catastrophic impact. We have previously shown fragments are uncertain. Nevertheless, a lower limit to the that radiative heat losses during catastrophic impacts for bodies equilibration temperature can be estimated if we assume that <400 km in diameter are insignificant because of the short the core had just begun to crystallize prior to the impact and lifetime of the ejecta cloud (most ejecta accrete within hours ignore the preferential loss of surface material. Then the mean or days) and the likelihood that the debris cloud is dusty and temperature of the reaccreted body would have been close to opaque (Haack et a/., 1996b). For near-catastrophic impacts the average of the core and surface temperatures prior to impact into isothermal, undifferentiated asteroids up to 300 km in (Haack et al., 1996b). If the core and surface temperatures diameter, the mean temperature of the asteroid is raised only a were 1400 and -100 "C, the mean temperature of the accreted few degrees by the kinetic energy of the projectile (Love and body would have been -650 "C. This lower limit lies within Formation of mesosiderites by fragmentation and reaccretion of a large differentiated asteroid 875 the transition temperature range of 450-800 "C suggesting that are also observed in nearly all other metal-rich differentiated our model is consistent with the cooling history of the meteorites. The rarity of troilite-rich , iron meteorites mesosiderites. (excluding those in groups IAB and IIICD), and mesosiderites can also be attributed to the differential resistance of phases to Mineral Proportions in the Mesosiderite Parent Body impact disruption and erosion in space and during entry into the Earth's atmosphere (e.g.,Ulff-Msller et al., 1998). To provide additional constraints on our impact model for Mineral abundances in the silicate matrix of mesosiderites mesosiderites, we need estimates of the bulk composition of based on modal analyses of thin sections (Prinz et al., 1980) the mesosiderite body prior to metal-silicate mixing, assuming show that olivine is depleted by a factor of -20 relative to it was chondritic, and the mean composition of mesosiderites. ordinary chondrites, while pyroxene and plagioclase show two- Olivine and metal abundances in chondrites vary enormously; fold enrichments (Table 4). Preferential loss of silicate during some are very deficient in olivine or enriched in metal like the catastrophic impact might enrich the metal twofold in the mesosiderites (CH and E chondrites) but their silicates have asteroid (Fig. 3) but pyroxene and plagioclase would have been very low Fe/(Fe + Mg) molar ratios of<0.05 (Brearley and Jones, preferentially lost in crustal ejecta. Olivine may be much more 1998). Since the silicates in mesosiderites have Fe/(Fe + Mg) abundant in some multi-kilogram samples as centimeter-sized molar ratios that are mostly in the range 0.2-0.4 (Mittlefehldt olivine clasts appear to be relatively common in several et al., 1998), we infer that ordinary chondrites, which have a mesosiderites (Mittlefehldt, 1980) and Mount Padbury contains comparable range of values (0.15 to 0.35), provide the best olivine crystals and dunite clasts that are 10 cm in length guide to the chondritic material that differentiated to form (McCall, 1966). The Lamont mesosiderite, which is the only mesosiderites. It seems unlikely that mesosiderites were one for which quantitative estimates of clast abundances are derived from olivine-poor asteroids, as Hewins (1983) available, contains 6 vol% olivine: 4 vol% of olivine clasts suggested. 0.5-2 cm in size and 2 vol% of finer matrix grains (Boesenberg Comparisons of bulk chemical analyses of mesosiderites et al., 1997). and ordinary chondrites (Table 4) show that mesosiderites are Additional evidence that mesosiderites may be very greatly enriched in metal (52 f 15 wt%) relative to the inferred unrepresentative samples of their parent body is provided by chondritic precursors (4-1 8 wt%). This absence of the kilometer-sized Eltanin asteroid. Analyses of unmelted complementary metal-poor meteorites from the mesosiderite fragments and vesicular impact melt particles from body enrichment of metal can plausibly be attributed to biased Eltanin suggest that it contained less metal than regular sampling of the mesosiderite body due to the failure of metal- mesosiderites (Gersonde et al., 1997; Kyte, 200 1). The estimated poor material to survive the journey to the Earth's surface. metal content of 4 f 1 wt% is consistent with arguments that Similar arguments are needed to explain the absence of metal-rich from the mesosiderite body survive meteorites from the mantles of the parent bodies of most iron longer in space (Kyte, 2001). The impact-melt glasses, which meteorites (e.g.,groups IIAB, IIIAB and IVA) and pallasites. are derived from the projectile and not from target rocks, are The mean troilite concentration (5 k 5 wt%) in mesosiderites less silicic than silicates in regular mesosiderites, but are still is comparable to concentrations in ordinary chondrites, but far from chondritic in composition. The inferred normative relative to metal, troilite shows threefold or higher depletions olivine abundance in these glasses is 4 wt%, which is higher in mesosiderites. Strong depletions of troilite relative to metal than the olivine abundance in the silicate matrix of normal mesosiderites (2.5 wt%; Table 4), providing weak support for the suggestion that the olivine missing from normal TABLE4. Mineral abundances in mesosiderites and ordinary mesosiderites may be concealed in larger asteroidal fragments. chondrites (wt%).

~ ~ ~ Mineral Mesosiderites H L LL Olivine Deficiency in Mesosiderites

Metallic Fe,Ni 52 k 1st 18 8.3 3.6 There does not appear to be any plausible way to mix Troilite 525 5.4 5.8 5.8 samples of metallic core and crustal material by impact without Silicate 462 11 77 86 91 retaining much of the associated olivine mantle. Either mesosiderites are very unrepresentative samples of the metal- Olivine* 2.5 46 52 57 rich breccias or else olivine was preferentially excluded from Plagioclase* 23 12 12 11 metal-rich breccias. In the latter case, the deficiency of olivine Pyroxene* 68 40 34 30 can be attributed to the very low survival rate of unarmored *Abundances in silicate portion. fragments. ?la of analyses. If core-metal was mixed with crustal silicates as we suggest, References: Mittlefehldt et al. (1998); Prinz et al. (1980); Jarosewich we should consider whether olivine could have been (1 990); McSween er al. (1 99 1). preferentially excluded from metal-rich breccias. For a hot 876 Scott et al.

differentiated target with a solid core that is turned inside out Mesosiderites contain a few metal globules with tiny by a large impact we would expect that the final body consists dispersed silicates (Hassanzadeh et al., 1990). These globules of a mixture of submillimeter to multi-kilometer fragments of may have formed early by mixing of molten metal and rock basalt, gabbro, pyroxenite, dunite and metal. Mean grain sizes fragments in the ejecta cloud. Metal-silicate mixing would of the fragments formed in the initial stages of the impact would have continued during reaccretionary impacts and ended in a increase in this sequence from metal and dunite to basalt slurry of molten metal, molten silicate and metal-coated silicate because prior impacts and the catastrophic impact itself would fragments that percolated through the asteroidal debris. The preferentially fracture surface materials. Dunite fragments typical fine-grained mesosiderite probably formed from a would also tend to be larger than basalt because of pressure viscous slurry in which the proportion of molten metal and strengthening at depth and because hot olivine is somewhat silicate relative to solid metal, silicate crystals and clasts was malleable and able to dissipate impact energy by deforming low enough to prevent further flow or gravitational separation rather than shattering. Subsequent impacts in the ejecta cloud of metal and silicate. may have been too gentle to cause much fragmentation, but Although mantle was more abundant than crust, the impacts during reaccretion onto the growing body would break multitude of tiny, crustal fragments would have comprised most all materials. For the 300 km target, impact velocities during of the total surface area of the debris, if (as is reasonable to reaccretion could exceed 100 m/s, more than sufficient to expect) the fragment size distribution followed that measured shatter igneous rocks (e.g., Hartmann, 1978). During the final in laboratory impact experiments. Thus, molten metal was stages, we would also expect cooler rocks to have been more likely to hit cold, crustal fragments than any other surfaces shattered more readily than hot rocks. Thus in the final body, during debris rainout, assuming good mixing. Contact between mantle and core fragments would be coarser than crustal molten metal and large, hot mantle fragments would have fragments, and fragments from all depths would be intimately caused relatively little metal to crystallize. Molten metal would mixed. have crystallized preferentially around the small, cool crustal In mesosiderites, the distributions of grain sizes of metal fragments as these were most effective at cooling and trapping and rock fragments do not resemble the size distribution of molten metal. We infer that basalt, gabbro and pyroxenite may fragments found from solid targets. Mesosiderites consist be enriched in mesosiderites because the cool rock fragments mostly of fine mixtures ofmetal and silicate grains with a rather that trapped molten metal were largely composed of these uniform grain size (Fig. 1). Pure silicate or metallic regions materials. Olivine may be deficient in mesosiderites because that are centimeter-sized or larger are relatively rare. Silicate- most of the mantle broke into fragments that were too large to fit free metal regions in mesosiderites do not exceed a few inside meteorites and too hot to crystallize much molten metal. centimeters in size (e.g., Clarke et al., 1997; Mittlefehldt et al., 1998) and the few iron meteorites that are derived from FURTHER DISCUSSION the mesosiderite body are <40 g in size (Wasson et al., 1989). Except for an unsampled residual core (Fig. 2b), it appears To understand the nature of the impact that mixed metal that all the metal was remarkably well mixed with silicate in and silicate in mesosiderites, we have focussed on their metal- the reaccreted asteroid. Metal-troilite-silicate textures and silicate textures, formation ages, and thermal histories. metal compositions strongly suggest that almost all the metal However, other properties of mesosiderites must be explained solidified after mixing with silicate. A few rare centimeter- to construct a plausible geological history for the mesosiderite sized globules (one is visible in Fig. 1) probably solidified in body. Here we discuss three other properties that have been flight. The remarkable grain size distribution of metal and used to construct a very different geological history from the associated silicates can reasonably be attributed to the molten one we propose. Table 2 compares the geological history state of the metal prior to impact and to a very dynamic devised by Rubin and Mittlefehldt (1993) and Mittlefehldt et environment when the metal crystallized. We suggest that the al. (1 998) with our model. lack of olivine in mesosiderites may also reflect the molten An important difference between these models concerns state of the metal during metal-silicate mixing. the length of time for which metal and silicate were mixed Initially, the molten core of a scrambled, differentiated body together. We infer that mesosiderites formed in a single, large might break into large fragments like a solid core, because and complex impact that mixed metal and silicate in less than gravity is more important than material strength for large a year. Rubin and Mittlefehldt (1993) infer that impact mixing asteroids. However, metallic globules would be readily of metal and silicate continued for several hundred million years fragmented by gentle collisions in the ejecta cloud, and further (4.5 to 4.2 Ga). During this period, heating and melting events, impacts during reaccretion would splash molten metal over which they attribute to impacts, caused metamorphism and the surface of the growing body. Mineral and rock fragments extensive melting in some mesosiderites. In addition, basin- would be cooler than molten metal so that molten metal would forming impacts on the mesosiderite asteroid are thought to start to crystallize around the silicate fragments hindering their have melted the crust creating large melt sheets (Mittlefehldt gravitational separation in the body. et al., 1992). Formation of mesosiderites by fragmentation and reaccretion of a large differentiated asteroid 877

Arguments in favor of impact heating of asteroids are based 39Ar-40Ar ages and metallographic cooling rates for unshocked on studies of many types of meteorites (Mittlefehldt, 1979; ordinary chondrites. Extrapolation of this trend suggests that Rubin, 1995). Counter arguments that are based on studies of Bogard et al. (1990) might have predicted that meteorites that terrestrial impact craters, laboratory shock experiments, and cooled as slowly as mesosiderites would have 39Ar-40Ar ages theoretical arguments indicate that impacts were not an effective 14.2 Ga. Given the considerable uncertainty in the absolute heat source for asteroids (Keil et al., 1997). Only two types of metallographic cooling rates, Haack et al. proposed that meso- features in mesosiderites have been offered as evidence for siderites began slow cooling-not at 4.2 Ga-but at 4.4-4.5 Ga, impact heating. The first, which is discussed below, concerns when they were well buried in an asteroid 200-400 km in chemical evidence that the crust of the mesosiderite parent body radius. Bogard and Garrison (1998) agreed that the 39Ar-40Ar was remelted. Mittlefehldt (1979) infers that 26A1 could not ages of mesosiderites were probably not caused by direct impact have caused remelting of asteroids and argues that impact heating but were simply independent evidence for very slow heating must have been responsible. The second is that some cooling at depth. mesosiderites contain clasts and areas of matrix with silicate crystals with textures like those found in rapidly cooled impact Rare Earth Element Concentrations in Clasts melts (e.g.,Floran, 1978). Although a few rare clasts may conceivably have formed by shock heating, we suggest that Abundances of rare earth elements (REE) in achondritic most quenched silicate melt probably formed as a direct result clasts in mesosiderites have been affected by many processes. ofmixing molten metal at >l300-1400 "C with silicate clasts. The REE concentrations of some clasts were modified after In his review of this paper, A. E. Rubin suggested that high the clasts formed when they were mixed with molten metal levels of porosity would greatly enhance the efficiency of (Stewart et al., 1994; Kennedy et al., 1992; Wadhwa et al., impact heating and abundance of impact melt on the 1999). Some clasts have REE fractionations that may reflect mesosiderite body. Porosity may indeed increase the abundance terrestrial weathering (Ikeda el al., 1990). In Vaca Muerta, a of impact melt by a factor of 3-5, but the overall heating effects few clasts, which mostly have gabbroic textures, show extreme are still minimal and large impact-melt sheets should not be depletions of REEs with very high EdSm ratios, including the anticipated on asteroids. In any case, large differentiated highest known EdSm ratios in solar system rocks. Mittlefehldt asteroids like the mesosiderite parent body are unlikely to have et al. (1992) and Rubin and Mittlefehldt ( 1992) argue that these been very porous when they were hot. REE patterns were not established by reactions between clasts and matrix or during weathering. They infer instead that the Argon-39lArgon-40 Ages of Mesosiderites REE patterns result from complex igneous processing prior to the formation of mesosiderites. Since comparable REE The 39Ar-4OAr ages of mesosiderites were originally found fractionations are not observed in the basaltic , to be 3.4-3.8 Ga (Bogard et al., 1990) but were later revised Mittlefehldt et al. (1992) inferred that this processing occurred upwards to 3.74.1 Ga by Bogard and Garrison (1998) because when the crust of the mesosiderite parent body was melted by of a rounding error in the decay constant. These ages are much a large impact to form melt sheets, which fractionally younger than the 39Ar-40Ar ages of normal unshocked crystallized. Rubin and Mittlefehldt (1993) later suggested chondrites, which are 4.4-4.5 Ga (Turner et al., 1978), and that crustal remelting was caused by the projectile that mixed coincide with a period of heavy bombardment in the inner solar molten metal with crustal rocks on the mesosiderite body. This system. Bogard et al. (1990) therefore proposed that the interpretation is not consistent with our model. 39Ar-40Ar ages of mesosiderites resulted from a catastrophic There are numerous problems in attributing crustal impact around 3.9 Ga, which heated the mesosiderites causing remelting on the mesosiderite parent asteroid to an impact (Keil Ar loss and buried them deeply so that they cooled slowly. et al., 1997) and additional problems in invoking an impact Mittlefehldt et al. (1998) revised the time of this impact to that delivered projectile metal to the surface of an asteroid. 4.2 Ga to accommodate the error discovered by Bogard and Impacts on asteroids at the current mean velocity of 5 km/s Garrison (1 998). generate and retain only minor amounts of impact melt (Keil Haack et al. (1 996a) studied the low-temperature thermal et al., 1997). For an impact velocity of 1 kds, which is the history of mesosiderites and concluded from four separate velocity specified by Hassanzadeh et al. (1990) to mix projectile metallographic constraints and one based on cation ordering metal with target silicates, the amount of shock-formed impact in orthopyroxene (Ganguly er af., 1994) that mesosiderites melt would have been negligible. If the clasts with high EdSm cooled more slowly than any other metal-bearing meteorites. ratios did form after crustal remelting, it seems unlikely that Since Haack er al. (1996a) found little evidence for shock an impact was responsible. Below, we discuss an alternative heating, they proposed that the relatively young 39Ar-40Ar ages explanation. of mesosiderites resulted not from impact heating but from Around the time that the crust on the mesosiderite body slow cooling in a large asteroid. In support of this conclusion, was remelted, the crust of the HED body was strongly Haack et al. (1996a) found an inverse correlation between metamorphosed. Eucrites and diogenites were metamorphosed 878 Scott et al.

at temperatures of 800-1 000 "C for periods of the order of 1 Ma eroded and disrupted without ever having pure olivine or metal (e.g., Miyamoto et al., 1985). Yamaguchi et al. (1996, 1997) surfaces. This may help to explain why we have fragments of suggested that this metamorphism resulted from the eruption the cores of numerous asteroids in the iron meteorites but have and subsequent burial of lava flows to depths of 20 km or more identified only a few olivine-rich (A-type) and metal-rich and that deeply buried basalts were heated to their melting asteroids (M-type). points. G. J. Taylor (pers. comm., 2000) suggests that the same Davis et al. (1999) also suggested that impacts mixed up process may have caused crustal remelting on the mesosiderite differentiated asteroids. The asteroid, 16 , which is an body. If the mesosiderite body had contained more 26Al (e.g., M-type, has generally been considered to be a collisionally because of earlier accretion), or was slightly larger than the exposed core from a -500 km diameter asteroid. However, HED body, it is possible that deep burial under lava flows the absence of a dynamical family associated with Psyche caused remelting and extensive REE fractionations. argues against such an origin. Davis et al. (1999) suggest instead that a catastrophic impact mixed metal with silicate Reduction of Basaltic and Gabbroic Material onto the surface and that Psyche may even be the parent body of the mesosiderites. Basaltic and gabbroic clasts as well as the matrices ofmany mesosiderites contain abundant tridymite and and COMPARISON OF IMPACT MODELS low, correlated Fe/Mn and Fe/Mg ratios in pyroxene, which are not observed in HED meteorites (Delaney et al., 1981; According to the previous impact model, which has evolved Mittlefehldt, 1990). These features appear to result from redox over the last 15 years, mesosiderites formed when two reactions between P-rich metal and silicates but the timing and differentiated asteroids collided at -1 km/s (e.g., Rubin and location of certain reduction features is not well established. Mittlefehldt, 1993). This collision caused molten metal from Mittlefehldt (1990) attributed redox features in basaltic and the projectile to mix with the regolith of the larger body creating gabbroic clasts to reduction during crystallization in the crust the characteristic mixtures of metal and basaltic and gabbroic of the mesosiderite parent body from a source region that rock. After the hot metal-silicate mixture had cooled quickly contained P-rich metal. Thus Rubin and Mittlefehldt (1993) at the surface of the mesosiderite body, another large projectile infer that there was an extended period of geological activity broke up and reassembled the composite body so that metal- on the mesosiderite parent body between the start of metal- silicate mixtures were buried deeply. (We exclude aspects of silicate mixing and the formation of mesosiderites, which would this model that invoke impact heating, apart from that provided not be compatible with our model. by mixing of hot and cold materials.) Delaney et al. (198 1) studied matrices of mesosiderites and Wasson and Rubin (1985) originally proposed the low- observed pyroxene overgrowths on magnesian pyroxene clasts, velocity impact model to explain how metal from an asteroidal magnesian rims on pyroxene grains, olivine coronas, abundant core could have been mixed into a basalt-rich regolith without silica and phosphate, and resorbed pyroxene clasts, and argued the inclusion of major amounts of olivine from the mantle that that the reduction, growth and resorption features formed in foimed around the metallic core. They inferred that small metal situ in the mesosiderites. Similarities between the redox fragments from differentiated asteroids were mixed into the features in matrices and clasts suggest that the clasts may have basaltic regolith of the mesosiderite parent body at speeds of been greatly modified and reduced after they were mixed with -1 kds. After Hassanzadeh et al. (1990) deduced that the molten metal. We infer that the data and interpretations of metal was molten, not solid, when mixed with silicate, they Rubin and Mittlefehldt (1 993) do not provide persuasive suggested that a single asteroid with a molten core struck the arguments against our model for mesosiderite formation. basaltic surface of the mesosiderite parent body. The deficiency of olivine from the mantle around the projectile's core was IMPLICATIONS attributed by Hassanzadeh et al. to "stochastic processes combined with variable resistance to space erosion". Rubin Our conclusion that mesosiderites formed by scrambling and Mittlefehldt (1993) and Mittlefehldt et al. (1998) proposed of a differentiated asteroid suggests that differentiated asteroids that the low-velocity projectile was a "core fragment'' or an can be camouflaged by catastrophic impacts. Thus global asteroidal core that had been largely stripped of its mantle by spectra of a scrambled differentiated asteroid like that in Fig. 2b impacts. might be dominated by features due to olivine, pyroxene and It seems very unlikely that metal in mesosiderites could metal, like spectra of chondrites and S-type asteroids. If impacts have been derived from a stripped core that impacted the do riot efficiently strip crusts and mantles from cores of mesosiderite body, as impact stripping of cores appears to be differentiated asteroids as we infer, we should not expect to highly inefficient (Love and Ahrens, 1996; Fig. 3). In addition, find large eroded remnants of differentiated asteroids with given the inefficiency of impact melting (Keil et al., 1997), surfaces composed purely of mantle or core material. The the molten core must have been well insulated by a large olivine numerous parent bodies of iron meteorites may have been mantle and crust to remain molten for 100-1 50 Ma. It seems Formation of mesosiderites by fragmentation and reaccretion of a large differentiated asteroid 879 inevitable that any low-velocity projectile with a molten core of shocked material that was heated to several hundred degrees that mixed molten metal into a regolith would have deposited (Stoffler et al., 1991). There is one heavily shocked much mantle material as well. This presents problems for the mesosiderite, Chaunskij, but this was shocked after kamacite low-velocity model because mantle materials are rare. The formed (see Mittlefehldt et al., 1998). chemical, mineralogical and isotopic data for mantle and other In fairness to the low-velocity impact model, we should silicates are also consistent with an origin in a single body. point out that if both bodies were very large (hundreds of This argument is weakened by the close resemblance between kilometers in size), much of the target might have been the mesosiderite and HED bodies, but it could be tested with scrambled, as in our hypervelocity impact. However, such very more accurate 0 isotopic analyses of minerals in these large asteroids are rare and very few asteroids of any size collide meteorites. The evidence that metal was molten when mixed at 1 Ms. Since Jupiter probably formed from the solar nebula with silicates appears to have removed the major justification 100-150 Ma before metal and silicate were mixed, impact for invoking low speed impacts, viz., to explain the olivine velocities in the asteroid belt at the time ofmetal-silicate mixing deficiency, and created additional problems for this model. were probably close to their present values. Currently, impacts We propose that olivine is deficient in mesosiderites because at <1 km/s are very rare (51%) and the mean impact velocity olivine was not trapped in metal-rich volumes of the reaccreted for main belt asteroids is 5 Ws(Bottke et al., 1994). At 5 Ws, asteroid. The molten metal droplets that splashed all over projectile material is almost entirely lost from the target asteroid fragments of mantle and crust preferentially solidified around (Love and Ahrens, 1996). In howardites, which are surface the finest, coolest rock fragments, which were very largely breccias probably from Vesta (Mittlefehldt et al., 1998), there crustal. Mantle fragments were hotter than crustal fragments are only a few volume percent of projectile material. We and vastly larger in size. Although more studies of the conclude that differentiated asteroids failed to accrete fragmentation of molten metal during breakup and reaccretion significant projectile material at low velocities after they were are needed, this mechanism for excluding olivine appears covered with basalt. plausible. We note that the same mechanism for excluding olivine from metal-silicate aggregates could be invoked for Acknowledgements-We thank A. Ruzicka, A. E. Rubin, D. the low-velocity impact model. However, in this case, the Mittelfehldt, and E. Asphaug for helpful reviews and D. D. Bogard, K. Keil, D. W. Mittlefehldt, A. E. Rubin, G. J. Taylor, and L. Wilson mantle of the low-velocity projectile would have been more and other colleagues for valuable discussions. The Cray finely fragmented and more efficiently mixed with the impact supercomputer used in this investigation was provided by funding debris. from the NASA ofices of Mission to Planet Earth, Aeronautics, and According to the low-velocity impact model, mesosiderite Space Science. Curators at the American Museum ofNatural History, material was fragmented and mixed together on the surface the Smithsonian National Museum ofNatural History, and the NASA Johnson Space Center kindly provided photographs of slabs and long after it cooled below -800 "C. We should therefore expect sections of mesosiderites for this study. This work was partly to find some broken olivine coronas and pyroxene rims, which supported by NASA grant 5-4212 to K. Keil. This is SOEST are not observed (Ruzicka ef al., 1994). At high magnifications, contribution #5568 and HIGP publication #I 148. mesosiderites show igneous and metamorphic textures, not fragmental textures. The brecciated macrostructure of slabs Editorial handling: P. H. Warren ofMt. Padbury (e.g.,Fig. 55 ofMittlefehldt et al., 1998; Cowan REFERENCES and McCoy, 1998) and its metamorphic microtexture (Powell, 1969) show that some mesosiderite material was fragmented BENEDIX G. K., LOVE S. G., SCOTT E. R. D.,TAYLOR G. J., KEIL K. after the start of metal-silicate mixing and that pieces were AND MCCOY T. J. (1996) Catastrophic impact during embedded in later-formed hot material. But this does not differentiation of the IAB- parent asteroid (abstract). Lunar Planet. Sci. 27, 95-96. require that mesosiderite material experienced several BENEDIX G. K., MCCOY T. J., KEIL K. AND LOVE s. G. (2000) A hypervelocity impacts during an extended period in a regolith, petrologic study of the IAB iron meteorites: Constraints on the as is commonly inferred (e.g.,Mittlefehldt et al., 1998). Since formation of the IAB-winonaite parent body. Meteorit. Planet. metal-silicate mixtures can be formed, cooled, shattered and Sci. 35, 1127-1141. then buried in hot material during the reaccretion stage of the BENZw. AND ASPHAUG E. (1994) Impact simulations with fracture: 1. Method and tests. Icarus 107, 98-1 16. catastrophic impacts that we have modeled, we infer that BENZW. AND ASPHAUG E. (1999) Catastrophic disruptions revisited. complex breccias like the Mt. Padbury sample could have Icarus 142, 5-20. formed during asteroid break-up and reaccretion. BOESENBERGJ. S., DELANEYJ. s. AND PRlNZ M. (1997) Magnesian One argument that might be advanced in favor of a low megacrysts and matrix in the mesosiderite Lamont (abstract). speed collision model is that the amount of shocked material Lunar Planet. Sci. 28, 125-1 26. BOGARDD. D. AND GARRISOND. H. (1998) 39Ar-40Ar ages and in mesosiderites is very low (A. Ruzicka, pers. comm., 2000). thermal history of mesosiderites. Geochim. Cosmochim. Actu However, if the mean temperature of the asteroid was only 62, 1459-1468. raised a few degrees by the catastrophic collision (Love and BOGARDD. D., GARRISOND. H., JORDANJ. L. AND MITTLEFEHLDT Ahrens, 1996), there should be at most a few volume percent D. W. (1990) 39Ar-40Ar ages and thermal history of mesosiderites: 880 Scott et al.

Evidence for major parent body disruption <4 Ga ago. Geochim. JAROSEWICHE. (1990) Chemical analyses of meteorites: A Cosmochim. Acta 54,2549-2564. compilation of stony and analyses. Meteoritics BOTTKEW. F., NOLANM. C., GREENBERGR. AND KOLVOORDR. A. 25,323-337. (1 994) Velocity distributions among colliding asteroids. Icarus KEILK., HAACKH. AND SCOTT E. R. D. (1994) Catastrophic 107,255-268. fragmentation of asteroids: Evidence from meteorites. Planet. BREARLEYA. J. AND JONESR. H. (1 998) Chondritic meteorites. In Space Sci. 42, 1109-1 122. Planetary Materials (ed. J. J. Papike), pp. 3-1 to 3-398. Reviews KEILK., STOFFLER D., LOVE S. G. AND SCOTT E. R. D. (1997) in 36, Mineral. SOC.America, Washington, D.C., Constraints on the role of impact heating and melting in asteroids. USA. Meteorit. Planet. Sci. 32, 349-363. CLARKER. S., JR., GOLDSTEINJ. I. AND MCCOYT. J. (1997) KENNEDYA. K., STEWARTB. W., HUTCHEONI. D., PAPANASTASSIOU Reexamination of mesosiderite metal using low-temperature phase D. A. AND WASSERBURGG. J. (1992) Partitioning of REE between diagrams (abstract). Meteorit. Planet. Sci. 32 (Suppl.), A29. phosphates and silicates in mesosiderites: Evidence for differing COWAN K. A. AND MCCOY T. J. (1998) Mass equivalence and degrees of equilibration (abstract). Lunar Planet. Sci. 23, brecciation in the Mt. Padbury mesosiderite (abstract). Lunar 681-682. Planet. Sci. 29, #1194, Lunar and Planetary Institute, Houston, KYTEF. T, (2001) Composition of impact melt debris from the Eltanin Texas, USA (CD-ROM). impact , Bellingshausen Sea. Proc. Con$ Ocean DAVISD. R., FARINELLAP. AND MARZARIF. (1 999) The missing Impacts, Deep Sea Research 11, (in press). Psyche family: Collisionally eroded or never formed? Icarus KYTEF. T. AND BROWNLEED. E. (1985) Unmelted meteoritic debris 137, 140-151. in the Late Pliocene iridium anomaly: Evidence for the ocean DELANEYJ. S., NEHRUC. E., PRINZM. AND HARLOWG. E. (1981) impact of a nonchondritic asteroid. Geochim. Cosmochim. Acta Metamorphism in mesosiderites. Proc. Lunar Planet. Sci. 12B, 49, 1095-1108. 1315-1342. LOVES. G. AND AHRENST. J. (1 996) Catastrophic impacts on gravity FLORANR. J. (1978) Silicate petrography, classification, and origin dominated asteroids. lcarus 124, 141-1 55. of the mesosiderites: Review and new observations. Proc. Lunar LUCYL. B. (1 977) A numerical approach to the testing of the fission Planet. Sci. ConJ 9th, 1053-108 I. hypothesis. Astron. J. 82, 1013-1024. GANGULYJ., YANG H. AND GHOSES. (1994) Thermal history of MCCALLG. J. H. (1966) The petrology of the Mount Padbury mesosiderites: Quantitative constraints from compositional mesosiderite and its achondritic enclaves. Mineral. Mag. 35, zoning and Fe-Mg ordering in orthopyroxenes. Geochim. 1029-1 060. Cosmochirn. Acta 58,271 1-2723. MCSWEENH. Y., JR., BENNETTM. E. AND JAROSEWICHE. (1991) GERSONDER. ETAL. (1997) Geological record and reconstruction of The mineralogy of ordinary chondrites and implications for the late Pliocene impact of the Eltanin asteroid in the Southern asteroid spectrophotometry. Icarus 90, 107-1 16. Ocean. Nature 390, 357-363. MELOSHH. J. (1989) Impact Cratering: A GeologicProcess. Oxford GINCOLDR. A. AND MONAGHANJ. J. (1977) Smoothed particle Univ. Press, New York, New York, USA. 245 pp. hydrodynamics: Theory and application to non-spherical stars. MITTLEFEHLDTD. W. (1 979) The nature of asteroidal differentiation Mon. Not. Roy. Astron. SOC.181, 375-389. processes: Implications for primordial heat sources. Proc. Lunar HAACKH., SCOTT E. R. D. AND RASMUSSENK. L. (1 996a) Thermal Planet. Sci. Con$ loth, 1975-1993. and shock history of mesosiderites and their large parent asteroid. MITTLEFEHLDTD. W. (1980) The composition ofmesosiderite olivine Geochim. Cosmochim. Acta 60, 2609-26 19. clasts and implications for the origin of pallasites. Earth Planet. HAACKH., SCOTT E. R. D., LOVE S. G., BREARLEYA. J. AND McCoy Sci. Lett. 51, 29-40. T. J. (1996b) Thermal histories of IVA stony-iron and iron MITTLEFEHLDTD. W. (I 990) Petrogenesis of mesosiderites: 1. Origin meteorites: Evidence for asteroid fragmentation and reaccretion. of mafic lithologies and comparison with basaltic achondrites. Geochim. Cosmochim. Acta 60, 3 103-3 1 13. Geochim. Cosmochim. Acta 54, 1 165-1 173. HARTMANNW. K. (1 978) Planet formation: Mechanism of early MITTLEFEHLDTD. W., RUBINA. E. AND DAVISA. M. (1992) growth. Icarus 33, 50-61. Mesosiderite clasts with the most extreme positive europium HASSANZADEHJ., RUBINA. E. AND WASSONJ. T. (1990) anomalies among solar system rocks. Science 257, 1096-1099. Compositions of large metal nodules in mesosiderites: Links to MITTLEFEHLDTD. W., MCCOY T. J., GOODRICHC. A. AND KRACHER iron meteorite group IIIAB and the origin of mesosiderite A. (1998) Non-chondritic meteorites from asteroidal bodies. In subgroups. Geochim. Cosmochim. Acta 54,3 197-3208. Planetary Materials (ed. J. J. Papike), pp. 4-1 to 4-185. Reviews HEWINSR. H. (1983) Impact versus internal origins for mesosiderites. in Mineralogy 36, Mineral. SOC.America, Washington, D.C., Proc. Lunar Planet. Sci. ConJ14th, J. Geophys. Res. 88 (Suppl.), USA. B257-B266. MIYAMOTOM. AND TAKEDAH. (1977) Evaluation of a crust model HEWINS R. H. (1984) The case for a melt matrix in plagioclase-POIK of eucrites from the width of exsolved pyroxene. Geochem. J. mesosiderites. J. Geophys. Res. 89 (Suppl.), C289-C297. 11, 161-169. HEWINS R. H. AND HARRIOTTT. A. (1986) Melt segregation in MIYAMOTOM., DUKEM. B. AND MCKAYD. S. (1985) Chemical plagioclase-poikilitic mesosiderites. J Geophys. Res. 91 (Suppl.), zoning and homogenization of Pasamonte-type pyroxene and their D365-D372. bearing on thermal metamorphism of parent body. HOPFEW. D. AND GOLDSTEINJ. I. (2001) The metallographic cooling Proc. Lunar Planet. Sci. ConJ 15th, J. Geophys. Res. 90 (Suppl.), rate method revised: Application to iron meteorites and C629-C63 5. mesosiderites. Meteorit. Planet. Sci. 36, 135-1 54. POWELLB. N. (1969) Petrology and chemistry of mesosiderites-I. IKEDA Y., EBIHARAM. AND PRINZ M. (1990) Enclaves in the Mt. Textures and composition of nickel-iron. Geochim. Cosmochim. Padbury and Vaca Muerta mesosiderites: Magmatic and residue Acta 33, 789-8 10. (or cumulate) rock types. Proc. NIPR Symp. Antarct. Meteorites PRlNZ M., NEHRU C. E., DELANEYJ. S., HARLOWG. E. AND BEDELL 3, 99-131. R. L. (1 980) Modal studies of mesosiderites and related IRELAND T. R. AND WLOTZKA F. (1 992) The oldest zircons in the achondrites, including the new mesosiderite ALHA 77219. Proc. solar system. Earth Planet. Sci. Lett. 109, 1-10. Lunar Planet. Sci. Con$ llth, 1055-1071. Formation of mesosiderites by fragmentation and reaccretion of a large differentiated asteroid 88 1

RUBINA. E. (1 995) Petrologic evidence for collisional heating of TILLOTSONJ. H. (1 962) Metallic Equations of Statefor Hypervelocity chondritic asteroids. Icarus 113, 156-1 67. Impacts. General Atomic Report GA-32 16. RUBINA. E. AND MITTLEFEHLDTD. W. (1 992) Classification of mafic TURNER G., ENRIGHTM. c. AND CADOGAN P. H. (1 978) The early clasts from mesosiderites: Implications for endogeneous igneous history of parent bodies inferred from 40Ar-39Ar ages. processes. Geochim. Cosmochim. Acta 56, 827-840. Proc. Lunar Planet. Sci. Con$ 9th, 989-1025. RUBINA. E. AND MITTLEFEHLDTD. W. (1993) Evolutionary history ULFF-MBLLER F., CHOI B-G., RUBINA. E., TRAN J. AND WASSON J. T. of the mesosiderite asteroid: A chronologic and petrologic (1998) Paucity of sulfide in a large slab of : New perspec- synthesis. Icarus 101, 201-212. tives on formation. Meteorit. Planet. Sci. 33,22 1-227. RUZICKAA., BOYNTONW. v. AND GANGULY J. (1994) Olivine WADHWA M., SHUKOLYUKOV A., DAVISA. M. AND LUGMAIRG. W. coronas, metamorphism, and the thermal history of the (1999) Origin of silicate clasts in mesosiderites: Trace element Morristown and Emery mesosiderites. Geochim. Cosmochim. microdistributions and Mn-Cr systematics tell the tale (abstract). Act0 58,2725-274 1. Lunar Planet. Sci. 30, #1707, Lunar and Planetary Institute, RYAN E. V. (2000) Asteroid fragmentation and the evolution of Houston, Texas, USA (CD-ROM). asteroids. Ann. Rev. Earth Planet. Sci. 28, 367-389. WASSON J. T. AND RUBINA. E. (1 985) Formation of mesosiderites SCHWANDT C. S., CYGAN R. T. AND WESTRICH H. R. (1998) by low-velocity impacts as a natural consequence of planet Magnesium self-diffusion in orthoenstatite. Contrib. Mineral. formation. Nature 318, 168-1 70. Petrol. 130, 390-396. WASSON J. T., OUYANG X., WANG J. AND JERDEE. (1989) Chemical SCOTT E. R. D., HAACKH. AND LOVE S. G. (1996) Formation of classification of iron meteorites: XI. Multi-element studies of mesosiderites by fragmentation and reaccretion of a large 38 new irons and the high abundance of ungrouped irons from differentiated asteroid (abstract). Meteorit. Planet. Sci. 31 Antarctica. Geochim. Cosmochim. Acta 53, 735-744. (SUPPI.),Al25-Al26. WILSON L. AND KEIL K. (1 99 1) Consequences of explosive eruptions SHEN J. J., PAPANASTASSIOU D. A. AND WASSERBURG G. J. (1998) on small solar system bodies: The case of the missing basalts on Re-0s systematics in pallasite and mesosiderite metal. Geochim. the parent body. Earth Planet. Sci. Lett. 104, 505-5 12. Cosmochim. Acta 62,2715-2723. WILSON L., KEILK. AND LOVE s. G. (1999) The internal structures STEWARTB. W., PAPANASTASSIOU D. A. AND WASSERBURGG. J. and densities of asteroids. Meteorit. Planet. Sci. 34,479-483. (1 994) Sm-Nd chronology and petrogenesis of mesosiderites. YAMAGUCHI A., TAYLORG. J. AND KEIL K. (1996) Global Geochim. Cosmochim. Acta 58,3487-3509. metamorphism of the parent body. Icarus 124,97-112. STOFFLER D., KEIL K. AND SCOTT E. R. D. (1991) Shock YAMAGUCHI A., TAYLORG. J. AND KEIL K. (1997) Metamorphic metamorphism of ordinary chondrites. Geochim. Cosmochim. history of the eucritic crust of 4 Vesta. J. Geophys. Res. 102, Acts 55,3845-3867. 13 381-13 386.