Accepted Manuscript The formation and chronology of the PAT 91501 impact-melt L-chondrite with vesicle-metal-sulfide assemblages G.K. Benedix, R.A. Ketcham, L. Wilson, T.J. McCoy, D.D. Bogard, D.H. Garrison, G.F. Herzog, S. Xue, J. Klein, R. Middleton PII: S0016-7037(08)00082-3 DOI: 10.1016/j.gca.2008.02.010 Reference: GCA 5552 To appear in: Geochimica et Cosmochimica Acta Received Date: 31 May 2007 Accepted Date: 11 February 2008 Please cite this article as: Benedix, G.K., Ketcham, R.A., Wilson, L., McCoy, T.J., Bogard, D.D., Garrison, D.H., Herzog, G.F., Xue, S., Klein, J., Middleton, R., The formation and chronology of the PAT 91501 impact-melt L- chondrite with vesicle-metal-sulfide assemblages, Geochimica et Cosmochimica Acta (2008), doi: 10.1016/j.gca. 2008.02.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 1 - 1 2 3 4 5 6 7 8 9 The formation and chronology of the PAT 91501 impact-melt L-chondrite with 10 vesicle-metal-sulfide assemblages 11 12 13 14 15 G.K. Benedix1, R.A. Ketcham2, L. Wilson3, T.J. McCoy4, D.D. Bogard 5 , D.H. Garrison 6 , 16 G.F. Herzog 7 , S. Xue 7 , J. Klein 8 , and R. Middleton 8,* 17 18 19 20 21 22 1Impact and Astromaterials Research Centre (IARC)/Department of Mineralogy, The 23 Natural History Museum, Cromwell Road, London, SW7 5BD UK 24 (g.benedix@nhm.ac.uk) 25 2Dept. of Geological Sciences, Jackson School of Geosciences, Univ. of Texas at Austin, 26 Austin, TX 78712 USA 27 3Environmental Sci. Dept., Lancaster Univ., Lancaster LA1 4YQ UK. 28 4Dept. of Mineral Sciences, National Museum of Natural History, Smithsonian 29 Institution, Washington, DC 20560-0119 USA 30 5 ARES, NASA-JSC, Houston TX 77058 31 6 ESCG/Barrios Technology, Houston, TX 77058 32 7 Rutgers Univ., Piscataway, NJ 08854-8087 33 8 Univ. Pennsylvania, Philadelphia, PA 19104. 34 35 36 37 38 Submitted to Geochimica et Cosmochimica Acta May 2007 39 Revision submitted in February 2008 40 *Deceased 41 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 2 - Abstract ? The L chondrite Patuxent Range (PAT) 91501 is an 8.5-kg unshocked, 42 homogeneous, igneous-textured impact melt that cooled slowly compared to other 43 meteoritic impact melts in a crater floor melt sheet or sub-crater dike (Mittlefehldt and 44 Lindstrom, 2001). We conducted mineralogical and tomographic studies of previously 45 unstudied mm- to cm-sized metal-sulfide-vesicle assemblages and chronologic studies of 46 the silicate host. Metal-sulfide clasts constitute about 1 vol.%, comprise zoned taenite, 47 troilite, and pentlandite, and exhibit a consistent orientation between metal and sulfide 48 and of metal-sulfide contacts. Vesicles make up ~2 vol.% and exhibit a similar 49 orientation of long axes. 39Ar-40Ar measurements probably date the time of impact at 50 4.461 ?0.008 Gyr B.P. Cosmogenic noble gases and 10Be and 26Al activities suggest a 51 pre-atmospheric radius of 40-60 cm and a cosmic ray exposure age of 25-29 Myr, similar 52 to ages of a cluster of L chondrites. PAT 91501 dates the oldest known impact on the L 53 chondrite parent body. The dominant vesicle-forming gas was S2 (~15-20 ppm), which 54 formed in equilibrium with impact-melted sulfides. The meteorite formed in an impact 55 melt dike beneath a crater, as did other impact melted L chondrites, such as Chico. 56 Cooling and solidification occurred over ~2 hours. During this time, ~90% of metal and 57 sulfide segregated from the local melt. Remaining metal and sulfide grains oriented 58 themselves in the local gravitational field, a feature nearly unique among meteorites. 59 Many of these metal-sulfide grains adhered to vesicles to form aggregates that may have 60 been close to neutrally buoyant. These aggregates would have been carried upward with 61 the residual melt, inhibiting further buoyancy-driven segregation. Although similar 62 processes operated individually in other chondritic impact melts, their interaction 63 produced the unique assemblage observed in PAT 91501. 64 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 3 - 1. INTRODUCTION 65 Impact is one of three fundamental processes that, along with accretion and 66 differentiation, formed and modified asteroid bodies. From nebular accretion through to 67 modern times, impact has left its traces in the ubiquitous cratered surfaces, distorted 68 shapes, and telltale signs of shocked minerals and melts recorded in meteorites. 69 Radiometric chronology reveals impact disturbance or resetting of isotopic systems on 70 several meteorite parent bodies over a wide range in time, but mostly in the periods of 71 3.5-4.0 Gyr and <1 Gyr ago (Bogard, 1995). Cosmic ray exposure ages of stony 72 meteorites often date the time of impact-induced ejection from their parent objects and 73 are essentially all less than 0.1 Gyr (Herzog, 2005). Despite abundant evidence, 74 however, our knowledge of the physical processes and the timing of impact on asteroidal 75 bodies remains incomplete. While we have systematic sampling from many well-76 documented impact craters on Earth to serve as our guide to interpreting impact 77 phenomena on other planets, we have only sparse and random sampling of asteroidal 78 impact craters. Further, our knowledge of chronology is limited because the samples we 79 do have may not accurately reflect the impact flux throughout the history of the Solar 80 System. Examples within meteorites of relatively large impacts during the first ~0.2 Gyr 81 of planetary history, when some parent objects were still experiencing metamorphism, are 82 particularly rare. Continuing study of impact-derived meteorites can help fill these gaps 83 in our knowledge. 84 Melt veined meteorites, impact melt breccias, and impact melts are not 85 uncommon among the known ordinary chondrite population and have been studied 86 extensively (Rubin, 1985; St?ffler et al., 1991). In particular, considerable effort has 87 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 4 - focused on relating these meteorites to the physical setting of their formation, starting 88 with the location of their melt. Melt can form on the floor and walls during formation of a 89 crater or in subsurface dikes that may extend below the crater into the country rock. Each 90 of these settings provides a unique physical and thermal environment for the 91 incorporation of clastic material and cooling history. In the case of the floor melts, they 92 will be exposed to space prior to being covered by falling debris or wall collapse. The 93 dike melts will be shielded from space exposure and allow the melt to remain clast-free. 94 Impact rates were much higher in the early history of the solar system (Hartmann 95 et al., 2000), but those impacts are probably recorded in the asteroid belt by the 96 population of small bodies produced by the break-up of larger precursor asteroids. 97 Impact craters observed on asteroids today are more recent events, consistent with the 98 fact that most strong impact events in chondrites occurred within the past 1 Gyr, as 99 determined by Ar-Ar radiometric dating (Bogard, 1995, and references therein). For 100 example, many L-chondrites show Ar-Ar impact heating ages clustering near 0.5 Gyr, 101 perhaps dating the time of disruption of the parent body (Haack et al., 1996; 102 Korochantseva et al., 2007). Interestingly, chronological evidence for collisional events 103 very early in asteroid history is sparse. 104 This paper presents a multidisciplinary study of PAT 91501, a vesicular, impact 105 melted L chondrite (Score and Lindstrom, 1992). Vesicles have been reported in only 106 two other ordinary chondrite impact melts: Shaw (Taylor et al., 1979) and Cat Mountain 107 (Kring et al., 1996). Although these meteorites are chemically and petrologically well-108 characterized (Harvey and Roeder, 1994, Mittlefehldt and Lindstrom, 2001), no study has 109 addressed the implications of the presence of vesicles in impact melt rocks. Our 110 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 5 - objectives were to document the meteorite?s impact and cosmic ray exposure history and 111 to understand the genesis of the unusual vesicular nature of this meteorite. 112 2. PREVIOUS WORK 113 The petrology of the silicate portion of PAT 91501 is reported by Harvey and 114 Roedder (1994) and Mittlefehldt and Lindstrom (2001). We briefly review these studies 115 here. PAT 91501 is an unshocked, homogeneous, igneous-textured rock of broadly L 116 chondrite mineralogy and chemistry. Major element mineral chemistries were shown to 117 be consistent with those of L-type chondrites; the minor element chemistry of olivine and 118 low-Ca pyroxene, on the other hand, is consistent with melting (Mittlefehldt and 119 Lindstrom, 2001); depletions of Zn and Br and sequestration of siderophile and 120 chalcophile elements into the large, heterogeneously-distributed metal-sulfide aggregates 121 were observed (Mittlefehldt and Lindstrom, 2001). Relic material includes rare 122 chondrules, as well as opaque-inclusion-rich olivine and some low-Ca pyroxene grains 123 that constitute ~10 vol.% of the meteorite, but distinct unmelted clasts, commonly found 124 in impact melt breccias, are absent. Mittlefehldt and Lindstrom (2001) concluded that 125 PAT 91501 is an impact melt of an L chondrite that crystallized at a cooling rate slower 126 than that typical for impacts melts and likely formed in a melt sheet on the crater floor or 127 in a sub-crater melt dike. 128 3. SAMPLES AND ANALYTICAL TECHNIQUES 129 Patuxent Range (PAT) 91501 was recovered in Antarctica during the 1991-1992 130 collecting season. Numerous large and small pieces, totaling more than 8.5 kg, were 131 collected. Their relative positions in the meteoroid are unknown. In the same locale, two 132 small metal-sulfide nodules (PAT 91516 and 91528; Clarke, 1994) were recovered. As 133 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 6 - discussed below, these meteorites are petrologically identical to metal-sulfide nodules 134 from PAT 95101. 135 PAT 91501 was originally classified as an L7 chondrite (Score and Lindstrom, 136 1992) based on textural features, mineral chemistry and oxygen isotopic composition, 137 although it was noted that it was similar in many respects to the Shaw L chondrite impact 138 melt. On further investigation (Harvey and Roedder, 1994; Mittlefehldt and Lindstrom, 139 2001), it was determined to be a near-total impact melt of an L chondrite. 140 One of the most striking features of PAT 91501 is the mm- to cm-sized vesicles 141 seen on cut surfaces of the sample (Fig. 1), as originally noted by Marlow et al. (1992). 142 We focused on PAT 91501 because it contains large vesicles that are visible in hand 143 sample (Fig. 1), there is abundant material, and it has been described as a total impact 144 melt (Mittlefehldt and Lindstrom, 2001). Visual inspection of PAT 91501 ,50 (2814.3 g) 145 and ,78 (127.6 g) show clastless, light colored surfaces with cm-sized vesicles and 146 metal/troilite aggregates. 147 3.1 Petrology 148 All thin sections of PAT 91501 (,26; ,27; ,28; ,95; and ,111) available at the 149 Smithsonian National Museum Natural History, as well as sections of PAT 91516 and 150 PAT 91528, were examined in both reflected and transmitted light with an optical 151 microscope. Metal and troilite compositions were analyzed using a JEOL JXA 8900R 152 electron microprobe at the Smithsonian. Analytical conditions were 20kV and 20nA. 153 Well-known standards were used and analyses were corrected using a manufacturer 154 supplied ZAF correction routine. Sulfur isotopes were analyzed using the 6f ion 155 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 7 - microprobe at the Carnegie Institution of Washington utilizing Canyon Diablo troilite as 156 the standard. 157 The two hand samples described above (,50 and ,78) were imaged at the High 158 Resolution X-ray Computed Tomography facility at the University of Texas at Austin 159 (UTCT), which is described in detail by Ketcham and Carlson (2001). The focus of our 160 work was to determine the distribution of vesicles, metal, and sulfide, which are easily 161 distinguished based on their large density contrast from the silicate matrix. Sample PAT 162 91501 ,50 was scanned using the high-energy subsystem, with the X-ray energy set at 163 420 kV and 4.7 mA, with a focal spot size of 1.8 mm. The samples were scanned in air 164 and, to reduce scan artifacts, the beam was pre-filtered with 1.58 mm of brass. Each slice 165 was reconstructed from 1800 views, with an acquisition time of 128 ms per view. A total 166 of 141 (1024x1024) slices were acquired with a thickness and spacing of 0.5 mm, 167 imaging a 196 mm field of view. The final scan images were post-processed for ring 168 artifact removal. Sample PAT 91501,78 was imaged using the microfocal subsystem, 169 with X-rays at 180 kV and 0.25 mA, and a focal spot size of approximately 0.05 mm. 170 The sample was sealed and placed in a cylinder and surrounded by water, which was used 171 as a wedge calibration to reduce scan artifacts. Data for 31 slice images were acquired 172 during each rotation of the sample; over each rotation, 1000 views were acquired with an 173 acquisition time of 267 ms per view. A total of 927 (1024x1024) slices at 0.0726 mm 174 intervals, each showing a 67 mm field of view, were acquired. Scans were reconstructed 175 using a software correction to further reduce beam-hardening artifacts. Animations, 176 including flipbooks for the 2D computed tomography scans and 3D rotational renderings 177 are available in the electronic annex. 178 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 8 - Measurements of vesicles and metal/troilite particles from the CT data volume 179 were made using Blob3D software (Ketcham, 2005), and visualizations were made using 180 Amira? version 3.1. 181 3.2 Chronology (39Ar-40Ar ages and cosmic-ray exposure ages) 182 A 48-mg whole rock sample of PAT91501 ,109 was irradiated with fast neutrons, 183 along with multiple samples of the NL-25 hornblende age standard. This irradiation 184 converted a portion of the 39K into 39Ar, and the 40Ar/39Ar ratio yields the K-Ar age. The 185 irradiation constant (J-value) was 0.025210 ?0.000125. Ar was released from PAT 91501 186 in 34 stepwise temperature extractions and its isotopic composition was measured on a 187 mass spectrometer. Experimental details are given in Bogard et al. (1995). Two 188 unirradiated whole rock samples of PAT 91501, taken from different locations in the 189 meteorite (see below), were degassed in either two or four stepwise temperature 190 extractions and the He, Ne, and Ar released were analyzed on a mass spectrometer. All 191 noble gas analyses were made at NASA-JSC. 192 We analyzed chips from four different specimens (subsamples 34, 38, 40, and 42) 193 of PAT 91501 for cosmogenic radionuclides. Sample ,34 was located adjacent to sample 194 ,106, which was analyzed for noble gases. Using facilities at Rutgers, the four specimens 195 were ground and weighed. After addition of Al and Be carriers, the powders were 196 dissolved in strong mineral acids. Beryllium and aluminum were separated by ion 197 exchange, precipitated as the hydroxides, and ignited to the oxides as described by Vogt 198 and Herpers (1988). The activities of 10Be and 26Al were measured by accelerator mass 199 spectrometry at the University of Pennsylvania as described by Middleton and Klein 200 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 9 - (1986) and Middleton et al. (1983) (Table 1). As the 10Be activity of sample ,40 was 201 unaccountably low and inconsistent with the 26Al activity, we do not report it. 202 203 4. RESULTS 204 We report our analyses of the metal-sulfide-vesicle assemblages, based on both 205 microscopic examination and computed tomography, and the results of the chronological 206 analyses for both 39Ar-40Ar and cosmogenic noble gases and radionuclides. 207 4.1 Petrography of metal-sulfide assemblages 208 PAT 91501 contains both vesicles and rounded metal-sulfide nodules that reach 1 209 cm in diameter. Previous studies have focused primarily on the silicate portion (Harvey 210 and Roedder, 1994; Mittlefehldt and Lindstrom, 2001), with neither study reporting 211 detailed examination of a metal-sulfide nodule in thin section. As we discuss later, 212 metal-sulfide nodules are rare, with less than 1 per cm3. Apart from a single chemical 213 analysis of troilite reported by Mittlefheldt and Lindstrom (2001), they have not been 214 studied . 215 We examined a 5-mm-diameter metal-sulfide nodule adjacent to a 5-mm-diameter 216 vesicle in subsample ,111. This nodule consists of a core of Fe,Ni metal (2 by 3.5 mm) 217 rimmed by sulfide, with the two phases exhibiting numerous mutual protrusions into each 218 other. The sulfide is dominantly troilite, although minor (<1 vol.%) pentlandite 219 ((Fe6.15Ni2.62Co0.10) =8.88S8) is observed at troilite-metal, troilite-silicate and troilite-vesicle 220 boundaries. Schreibersite rims are often found at the border between metal and 221 troilite/pentlandite. The S isotopic composition of pentlandite (3 analyses yield ?34S of 222 0.5-1.9?) and troilite (7 analyses yield ?34S of 0.4-1.2?) are essentially identical. No 223 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 10 - polycrystallinity or twinning is observed in the troilite, confirming the observation of 224 Mittlefehldt and Lindstrom (2001) that PAT 91501 experienced minimal secondary shock 225 after its crystallization. The metal is composed of two domains (Fig. 2a). Rimming each 226 domain is a 50 ?m thick region of high-Ni (up to 45 wt.%; Fig. 2b) taenite that is 227 relatively inclusion free. Adjacent to this, Ni decreases systematically from ~40 wt.% to 228 ~20 wt.% and this zoned metal often contains 10-30 ?m troilite and 1-5 ?m schreibersite 229 inclusions. The center of the largest domain, which appears to have been bisected, is 230 martensitic, with irregular Ni concentrations of 20-25 wt.%, and contains troilite and 231 schreibersite inclusions that can reach tens of microns. 232 While metal-sulfide nodules from PAT 91501 have not been previously described, 233 the published descriptions (Clarke, 1994) for the small iron meteorites PAT 91516 (1.58 234 g) and PAT 91528 (3.34 g) are essentially identical to that given here for PAT 91501. 235 The only substantive difference is that Clarke observed a larger number of metal 236 domains, particularly in PAT 91516, and these were often separated by sinuous troilite. 237 4.2 Computed Tomography 238 We used computed tomography (CT) to survey the distribution of vesicles, metal 239 and sulfide in two samples of PAT 91501 (see electronic annex EA-2 and EA-4). Figure 240 3 is a single frame of a 3 dimensional, rotational visualization (see electronic annex EA-241 1) made from the CT scan of PAT 91501 ,50, in which vesicles and metal-sulfide 242 intergrowths are highlighted. In this sample, 5085 vesicles were measured, which 243 comprise ~2 volume percent of the sample. The sizes of the vesicles range in diameter 244 from 0.6 to 14 mm. In contrast, analysis of PAT 91501 ,78 yielded 36685 vesicles 245 ranging in size from 0.2 to ~6mm in diameter. The difference in the numbers and the size 246 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 11 - range of vesicles is due to the fact that the smaller sample was scanned at a much higher 247 resolution. In both samples, tiny vesicles (<1mm diameter) dominate the population. 248 Vesicles in both samples are homogenously distributed and have a median aspect ratio of 249 1.4, indicating moderate elongation. 250 The CT scans revealed the existence and distribution of several large metal-251 sulfide intergrowths (Fig. 3). Together, metal and sulfide represent less than 1 volume 252 percent of the sample. We measured 255 and 142 metal grains in sample ,50 and ,78, 253 respectively. Metal ranges in size from 0.7 to 8.6 mm in the larger sample and comprises 254 0.27 vol%, while in the smaller sample metal ranges from 0.1 to 3.8 mm and represents 255 0.35 vol%; as with the vesicles, the higher-resolution scan of the smaller sample 256 permitted us to measure particles too small to be resolved in the scan of the larger 257 specimen. Sulfide is more abundant than metal in both samples and occupies 258 approximately twice the volume as metal. Sulfide accounts for 0.4 vol% in the larger 259 sample (,50) and for 0.56 vol% in the smaller sample. We measured 404 sulfide grains in 260 sample ,50 and 540 grains in sample ,78. Sulfide is overall larger than metal and ranges 261 from 0.6 to 12.7mm in ,50 and from 0.2 to 4.7mm in ,78. As with vesicles, tiny grains (< 262 1mm) comprise the mode of both the metal and sulfide size distributions (see flipbooks 263 and 3D renderings in supplemental data). 264 PAT 91501 (,50) contains 169 grains in which metal and sulfide are in contact. 265 These particles were noted by earlier workers (Score and Lindstrom, 1992; Mittlefehldt 266 and Lindstrom, 2001) and attributed to formation as immiscible melts prior to silicate 267 crystallization. Interestingly, these particles exhibit a consistent orientation of the metal 268 and sulfide relative to each other and to the meteorite as a whole. Figure 4a is a stereo 269 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 12 - plot of the normals to the planes defined by the contact between metal and sulfide with 270 the size of the each circle proportional to the area of the contact. Although some scatter 271 is observed in this plot, particularly for smaller metal-sulfide pairs, the majority of larger 272 particles defines a tight cluster trending 255? and plunging 45?; note that these 273 orientations are with respect to the scan data, and are not geographical. 274 The CT scans also document the relationship between vesicles, metal, and sulfide. 275 Larger vesicles appear to have metal-sulfide intergrowths associated with them. In the 276 CT scan of the larger sample of PAT 91501, we found 18 instances where vesicles are in 277 contact with metal only, sulfide only or metal-sulfide intergrowths. In the higher 278 resolution CT scan of PAT 91501, 78, there are nearly 200 vesicles in contact with metal, 279 sulfide or metal-sulfide. The vast majority of the largest vesicles are in contact with 280 metal and/or sulfide. 281 The elongation of vesicles allows us to examine their orientation as well. Fig. 4b 282 is a stereo plot of the orientations of the vesicle long axes from the main mass of PAT 283 91501 (,50) with the circle areas proportional to vesicle volume. Again, considerable 284 scatter is observed, particularly among the smaller vesicles. However, the larger vesicles 285 define a distinct cluster trending 300? and plunging 40?, with the main outlier attributable 286 to contact with an irregular metal-sulfide mass. This cluster is offset ~33? from the 287 orientation defined by the normals to the metal-sulfide contacts. 288 4.3 Ar-Ar Age 289 The PAT 91501 Ar-Ar age spectrum (Fig. 5) appears complex but can be 290 interpreted to yield a reliable age. The rate of release of 39Ar and changes in the K/Ca 291 ratio and the Ar-Ar age as a function of extraction temperature all suggest that 39Ar is 292 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 13 - contained in three distinct diffusion domains? 0-17%, 17-80%, and 80-100% 39Ar release 293 (Fig 5). The 39Ar release data can be modeled by standard diffusion theory in terms of 294 the parameter D/a2, where D is the diffusion coefficient and a is the average diffusion 295 length for Ar in the degassing grains. On an Arrhenius plot (argon released vs. 1/T; not 296 shown), data for these three domains give separate linear trends, each one characterized 297 by a different value of D/a2. Above 80% 39Ar release, the observed decreases in age and 298 K/Ca are interpreted to represent release of excess 39Ar recoiled during irradiation into 299 the surfaces of pyroxene grains. Below ~17% 39Ar release, the higher ages are 300 interpreted to represent loss of recoiled 39Ar from surfaces of grains possessing a 301 relatively larger K/Ca ratio. Between ~19% and 80% of the 39Ar release, the K/Ca ratio 302 is relatively constant and the Ar-Ar ages describe a plateau. Ten extractions releasing 19-303 78% of the 39Ar define an age of 4.463 ?0.009 Gyr, where the age uncertainty is 304 approximately one-sigma and includes the uncertainty in the irradiation constant, J. 305 Seven extractions releasing 30-78% of the 39Ar give an age of 4.461 ?0.008 Gyr. To 306 examine these data in an isochron plot, we adopted the cosmogenic 38Ar concentration 307 given below and used the measured 37Ar/36Ar ratios for each extraction to apportion the 308 measured 36Ar into trapped and cosmogenic components. An isochron plot of 40Ar/36Ar 309 versus 39Ar/36Ar, using trapped 36Ar, is highly linear (R2=0.9995) and its slope yields an 310 age of 4.466 ?0.0012 Myr, in agreement with the plateau age. The isochron intercept 311 value of 40Ar/36Ar =-79?156 suggests all 40Ar released in these extractions is radiogenic. 312 The total age summed across all extractions is 4.442 Gyr and suggests that little to no 313 40Ar was lost from the sample by diffusion over geologic time. 314 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 14 - Although, we conclude that impact resetting of the K-Ar age in PAT91501 315 occurred 4.46 ?0.01 Gyr ago, several unshocked L and H chondrites show Ar-Ar ages 316 across a wide range of ~4.52-4.44 Gyr, presumably as a result of extended parent body 317 metamorphism (Turner et al., 1978; Pellas & Fi?ni, 1988). Therefore, we cannot exclude 318 the possibility that the 4.46 Gyr Ar-Ar age of PAT91501 dates this L-chondrite 319 metamorphism and that the impact event occurred earlier. 320 4.4 Cosmogenic Noble Gases and Radionuclides. 321 PAT 91501 ,109 (33.4 mg) was heated in two temperature steps and sample ,106 322 (50.0 mg) was heated in four steps (Table 1). In both samples approximately half of the 323 3He was released at 500?C. In sample ,106 the peak of the Ne release occurred at 900-324 1200 ?C, and the peak of the 38Ar release occurred at 1200 ?C. Measured 3He is entirely 325 cosmogenic. The summed 20Ne/22Ne ratios of 0.845-0.847 indicate that measured Ne is 326 also entirely cosmogenic. Consequently we summed concentrations for each Ne isotope 327 across all extractions to obtain total cosmogenic abundances. Measured 36Ar/38Ar ratios 328 varied over 0.72-1.75 and indicate the release of trapped Ar, which is mostly adsorbed 329 atmospheric Ar, particularly at lower extraction temperatures. We assumed 36Ar/38Ar 330 ratios of 5.32 for trapped Ar and 0.67 for cosmogenic Ar and calculated the abundances 331 of cosmogenic 38Ar for each extraction. The 38Arcos abundances were then summed 332 across each extraction to obtain the total abundance of 38Arcos. From analyses of He, Ne, 333 and Ar delivered from a standard gas pipette, we estimate the uncertainty in these 334 abundances as ~?10%. 335 Cosmogenic abundances and 22Ne/21Ne ratios for the two PAT samples are given 336 in Table 2. Cosmogenic abundances of 3He, 21Ne, and 38Ar in the two samples agree with 337 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 15 - each other within their individual uncertainties of ?10%. The measured 20Ne/21Ne ratios 338 for ,109 and ,106 are identical at 0.847 ?0.005 and 0.845 ?0.015. The measured 22Ne/21Ne 339 ratios of 1.084 ?0.003 and 1.097 ?0.003 differ slightly, which probably reflects a shielding 340 difference. A plot of 3He/21Ne versus 22Ne/21Ne defines a shielding trend for many 341 chondrites (Eberhardt et al., 1966). Sample ,109 plots on this shielding trend, but sample 342 ,106 plots slightly above the trend, as a consequence of its lower 21Ne concentration. 343 This observation may imply that in our ,106 sample the concentration of Mg, the main 344 target for 21Necos production, was slightly less than the chondritic value. There is no 345 suggestion of diffusive loss of 3He in either sample, in spite of the observation that He 346 degassed at relative low temperature in the laboratory (Table 1). 347 The measured abundances of 10Be for three PAT samples agree within their 348 uncertainties (Table 2). The measured activities of 26Al in four PAT samples (Table 2) 349 span a range of ~17%. 350 4.5 PAT Pre-Atmospheric Size 351 The 22Ne/21Ne ratio of ~1.09 indicates that the pre-atmospheric shielding 352 experienced by PAT 91501 was somewhat greater than that for typical chondrites. The 353 maximum dimension of the recovered meteorite was ~19 cm, which, for the purpose of 354 modeling calculations, sets a minimum radius in space of ~10 cm. Modeling of the 355 22Ne/21Ne ratio in L-chondrites (Leya et al., 2000) predicts that as the meteoroid radius 356 increases, 22Ne/21Ne ratios as low as ~1.09 are first reached in the center of a body with a 357 pre-atmospheric radius of ~ 30 cm. Thus a somewhat larger body presumably carried the 358 physically separate samples that we analyzed. According to the calculations of both Leya 359 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 16 - et al. (2000) and Masarik et al. (2001), 22Ne/21Ne ratios plateau at 1.09?0.01 for pre-360 atmospheric depths from 10 to ?30 cm in L chondrites with radii of 40 cm. 361 The 22Ne/21Ne ratio is not useful for setting an upper bound on the pre-362 atmospheric radius. For this purpose we use the 26Al activity. After a cosmic ray 363 exposure lasting more than 20 My (see below) activities of 26Al (and 10Be) would have 364 reached saturation and are therefore equal to average production rates in space ? assuming 365 the terrestrial age of PAT was less than 50 kyr or so as suggested by the normal 26Al/10Be 366 ratios for three samples. According to the calculations of Leya et al. (2000), only 367 meteoroids with radii between 32 cm and 85 cm have the range of 26Al activities 368 observed in PAT. The 10Be activities of PAT 91501 are comparable to those of the L5 369 chondrite St-Robert (Leya et al., 2001), which is thought to have had a pre-atmospheric 370 radius between 40 and 60 cm. We conclude that the pre-atmosphere radius of PAT 371 91501 was in this range. 372 4.6 Cosmic Ray Exposure Age. 373 Cosmic Ray Exposure (CRE) ages of stony meteorites were initiated by impacts 374 that reduced meteoroids to objects meter-size or smaller and are almost all <100 Myr for 375 stones (Herzog, 2005). To calculate cosmic ray exposure ages for PAT (Table 2), we 376 used the cosmogenic production rates for L-chondrites given by Eugster (1988), except 377 that the 38Ar production rate was lowered by 11%, as suggested by Graf and Marti 378 (1995). The production rates were corrected for shielding using the measured 22Ne/21Ne 379 ratios. The differences among ages calculated from He, Ne, and Ar for a given sample 380 are greater than the differences in the same age between the two samples. This pattern 381 suggests that most of the apparent variation in CRE age is produced by our choice of 382 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 17 - production rates. Because cosmogenic Ar is more sensitive to likely compositional 383 variations and because there is some chance that cosmogenic Ar was incompletely 384 extracted, we give greater weight to the 3He and 21Ne ages and obtain a CRE age for PAT 385 91501 of 25-29 Myr. 386 We also calculated CRE ages based on the 26Al-21Ne-22Ne/21Ne and 10Be-21Ne-387 22Ne/21Ne equations of Graf et al. (1990a) by using data for the two samples known to 388 have been adjacent to each other ,34 and ,106. The results, 29.6 Myr and 25.5 Myr, 389 respectively, are in the same range as the CRE ages calculated from the noble gases 390 alone. Finally, we calculated the 10Be-21Ne CRE age for the ,34 -,106 pair by using the 391 formula of Leya et al. (2000) after modifying it for a 10Be half life of 1.5 My. This age, 392 21.9 My, is about 15-26% lower than the others. Leya et al. (2000) observed that their 393 equation for 10Be-21Ne CRE ages gives a low result for another large L-chondrite, 394 Knyahinya (preatmospheric radius ~45 cm; Graf et al., 1990b). They attribute the 395 discrepancy to their model?s underestimation of 10Be production rates in meteoroids the 396 size of Knyahinya and larger. 397 5. DISCUSSION 398 Among ordinary chondrites, the L chondrites record a particularly severe history 399 of impact bombardment, with almost 5% of this group containing shock melts (La Croix 400 and McCoy, 2007). However, most of these shock features probably occurred during a 401 major disruption of the L-chondrite parent body ~0.47 Gyr ago (Scott, 2002; 402 Korochantseva et al., 2007). Although the old Ar-Ar age for PAT 91501 is atypical 403 among L-chondrites, it is similar to the impact-melted Shaw L chondrite, as was noted 404 during its initial description (Marlow et al., 1992). However, the vesicular nature, 405 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 18 - presence of preserved, cm-sized metal-troilite intergrowths, and orientation of both the 406 vesicles and metal-sulfide particles in PAT91501 are unusual. These features promise 407 new insights into the timing of and physical processes occurring during the formation of 408 this impact-melted L chondrite. 409 5.1 Chronology 410 Chronological evidence for collisional events on asteroids very early in Solar 411 System history is sparse. Among achondrites, some unbrecciated eucrites may have been 412 excavated from depth on Vesta by a large impact ~4.48 Gyr ago (Yamaguchi et al., 2001; 413 Bogard and Garrison, 2003). Chronological evidence for relatively early impacts also 414 exists for parent bodies of the mesosiderites and IIE irons (Scott et al., 2001; Bogard et 415 al., 2000). Within the chondrites, McCoy et al. (1995) reported ages of enstatite 416 chondrite impact melts dating to before 4.3 Gyr and Dixon et al. (2004) suggested that 417 Ar-Ar ages of ~4.27 Gyr for a few LL-chondrites may date the time of one or more 418 impact events on the parent body. Portales Valley is an H6 chondrite, which gives an Ar-419 Ar age of 4.47 Gyr (Garrison and Bogard, 2001) and contains large metal veins showing 420 Widmanst?tten patterns indicative of slow cooling. These phases may have been mixed 421 by an impact (Kring et al., 1999; Haaack et al., 2000). Taken alone, Ar-Ar ages between 422 ~4.38 to ~4.52 Gyr can be ambiguous, as ancient ages may reflect either a late impact or 423 slow cooling after parent body metamorphism (Turner et al., 1978; Pellas & Fi?ni, 1988). 424 In contrast, impact melts provide a more direct means for dating the timing of collisional 425 events. Most impact melts give Ar-Ar ages less than 1 Gyr, suggesting that melting and 426 re-solidification took place recently, either during events confined to the surfaces of 427 modern asteroids or, perhaps, when collisions on asteroids melted partially and launched 428 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 19 - meteoroids into Earth-crossing orbits. This population of more recently formed impact 429 melt rocks includes the vesicular meteorites Cat Mountain (Kring et al., 1996) and Chico 430 (Bogard et al., 1995). 431 In contrast, PAT 91501 dates to the earliest history of the Solar System at 4.461 432 Gyr. Until this work, Shaw was the only ordinary chondrite known to be a near total 433 impact melt (Taylor et al., 1979) and have an Ar-Ar age consistent with an early (>4.0 434 Gyr ago) impact (Turner et al., 1978). Indeed, PAT 91501 shares a number of features 435 with Shaw, particularly its light-colored lithology, petrographic texture and clast-free 436 nature (Taylor et al., 1979). Based on its cosmogenic noble gas concentrations, Shaw has 437 a much younger, nominal one-stage CRE age of ~0.6 My, although in all likelihood, 438 Shaw had a complex exposure history with a first stage that probably lasted >10 Myr 439 (Herzog, 1997). In any event, cosmic-ray exposure ages greater than 1 Gyr are unheard 440 of in stony meteorites and thus Shaw?s old Ar/Ar age indicates that the meteoroid did not 441 melt (and hence lose an appreciable fraction of its radiogenic 40Ar) when it was launched 442 from the asteroid. The 4.46 Gyr impact event that formed PAT 91501 apparently took 443 place considerably earlier that those impacts that reset the Ar-Ar ages of Shaw (4.40 ?0.03 444 and 4.42 ?0.03 Gyr; Turner et al, 1978). We conclude that PAT 91501 and Shaw formed 445 in different impact events on the L-chondrite parent body, and that the two meteorites 446 were not located in close proximity. There seems little question that PAT 91501 is 447 closely related to the bulk of L chondrites and this relationship is supported by the CRE 448 age for PAT of 25-29 Myr, which lies within a diffuse ~22-30 Myr cluster in the 449 distribution of L-chondrite CRE ages. The 4.46 Gyr impact for PAT 91501 falls within 450 the range of Ar-Ar metamorphic ages of relatively unshocked chondrites (Turner et al., 451 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 20 - 1978; Pellas & Fi?ni, 1988). This observation implies that the L parent body experienced 452 a significant impact while it was still relatively warm. In all likelihood, these events 453 occurred on the original L chondrite parent body prior to any subsequent collisions and 454 breakups that would have formed modern asteroids. These early impacts left PAT 91501 455 deeply buried until it was excavated and launched toward Earth ~28 Myr ago. 456 5.2 Vesicle Formation 457 PAT 91501 is remarkable for its mm- to cm-sized vesicles. Vesicles of this size 458 have never before been observed in an impact-melt rock. The few vesicular meteorites 459 that have been investigated in detail are basaltic eucrites or angrites, where vesicles are 460 formed by gases liberated or generated during silicate partial melting (McCoy et al., 2006 461 and references therein). In terrestrial systems, H2O is the typical vesicle-forming gas, as 462 it is abundant in the Earth?s crust and exsolves from basaltic magmas at relatively 463 shallow depths (Oppenheimer, 2004). In contrast, chemical analyses (Jarosewich, 1990) 464 and the presence of abundant Fe,Ni metal suggest that ordinary chondrites likely were 465 very dry systems and, thus, H2O is unlikely as a major vesicle-forming gas. McCoy et al. 466 (2006) argued that a mixed CO-CO2 gas was responsible for vesicle formation in 467 asteroidal basalts and the contribution of such a gas cannot be unequivocally eliminated. 468 The contribution of volatiles from the impactor, such as ice in a cometary body, or 469 volatilization of silicates at superheated temperatures also seems unlikely, although 470 impossible to rule out. 471 A much more likely source of volatiles is sulfur vaporization during impact 472 melting. Numerous previous studies have pointed to the role of sulfur vaporization 473 during metamorphic and impact processes of ordinary chondrites. Lauretta et al. (1997) 474 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 21 - showed that a small amount of sulfur vaporizes at the metamorphic temperatures of 475 ordinary chondrites. Sulfur vaporization is also a common problem in ordinary chondrite 476 melting experiments (e.g., Jurewicz et al., 1995) and has been invoked to explain the 477 formation of sulfide-rich regions in the Smyer H chondrite impact melt breccia (Rubin, 478 2002). In PAT 91501, the larger vesicles have metal-sulfide intergrowths associated with 479 them, suggestive of formation by sulfide vaporization during impact melting 480 We can calculate the amount of S gas required to create the abundance of vesicles 481 (~2 vol.%) documented with computer tomography. The formula for the bulk density ? of 482 a vesicular material is 483 1/? = n / ?g + (1 - n) / ?ng (1) 484 where ?g is the gas density, ?ng is the density of the non-gas part (i.e. solid or liquid) and n 485 is the mass fraction of gas. If the conditions are such that the gas law holds at least 486 approximately, the density of the gas is given by 487 ?g = (m P) / (R T) (2) 488 where m is the molecular weight of the gas, P is its pressure, R is the universal gas 489 constant (8314 J/kmol) and T is the gas temperature. Substituting for ?g: 490 1/? = (n R T)/(m P) + (1 - n) / ?ng (3) 491 The two terms on the right are the partial volumes of gas and non-gas, respectively, so the 492 gas volume fraction vg is given by 493 ?g = [(n R T)/(m P)] / [(n R T)/(m P) + (1 - n) / ?ng] (4) 494 which is more conveniently re-arranged as 495 n / (1 - n) = (vg m P) / [(1 - vg) ?ng R T] (5) 496 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 22 - Using the values we estimated, vg = 0.02, T = 1670 K, ?ng = 3520 kg m-3, m = 64 for S2 or 497 SO2, and P = 5 x 105 Pa for a depth of a few km in a 50 km radius asteroid, appropriate to 498 the lithostatic load likely to occur in a silicate melt at depth, we find [n / (1 - n)] = 1.4 x 499 10-5, so that n = 1.4 x 10-5 to the same precision. If we regard this as the abundance of S2, 500 it indicates that ~15 ppm of gas are necessary for formation of the vesicles. 501 Alternatively, we can estimate the amount of S2 gas produced by vaporization if 502 all the FeS in a chondrite melted using the following equation from Lauretta et al. (1997): 503 Fe1-xS ? 0.99 Fe 1-x 0.99 S + 0.005S2 (6). 504 Thus, each mole of sulfide liberates 0.005 moles of S2. So for a typical L-505 chondrite mode of FeS (~4.2vol%), it would be expected that ~210 ppm of S2 would form 506 during melting. While this may seem like a small amount, the vesicle volume produced 507 by this amount of gas would be much greater than that observed in PAT 91501. As we 508 discuss in the next section, the amount of sulfide present in the impact melt likely results 509 from gravitational segregation of the dense metal-sulfide particles in the silicate melt. It 510 is likely that the amount of sulfide that actually vaporized is much closer to about one-511 tenth that of average L chondrites, and therefore, a S2 gas abundance of ~20 ppm is 512 probably more reasonable, in excellent agreement with above calculations for the amount 513 of S2 gas required to create the volume of space occupied by the vesicles. 514 The abundance of 15-20 ppm S2 required for vesicle formation and in equilibrium 515 with sulfide is small in comparison to the total mass of sulfur present. Thus, it is no 516 surprise that evidence for its condensation cannot be found. No sulfide or sulfur linings 517 have been observed on vesicle walls in PAT 91501, although moderate terrestrial 518 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 23 - weathering has occurred in the meteorite and hydrated iron oxides of terrestrial origin 519 commonly occur as vesicle linings. We considered the possibility that pentlandite found 520 in the metal-sulfide assemblages might reflect S volatilization. However, no isotopic 521 fractionation consistent with S volatilization was observed between pentlandite and 522 troilite and, as discussed later, it appears more likely that pentlandite is an equilibrium 523 phase formed during cooling in the Fe-Ni-S system. 524 5.3 Physical Setting and formation of PAT 91501 525 While PAT 91501 joins a growing list of impact-melted rocks from the L 526 chondrite parent body, its ancient age and large metal-sulfide nodules and vesicles and 527 their striking orientation are unique. Whereas it lacks the abundant unmelted clasts 528 observed in many impact melt breccias, similar clast-poor lithologies are observed in 529 Shaw and, most notably, as a 30 cm wide vein in Chico (Bogard et al., 1995). 530 Mittlefehldt and Lindstrom (2001) suggested that PAT 91501, because of its 531 homogenously melted nature and relatively slow cooling compared to other impact melts, 532 could be part of an impact melt basal layer found on the floor of a crater (Melosh, 1989) 533 or as a melt dike injected into surrounding country rock (St?ffler et al., 1991). Our work 534 provides additional constraints to distinguish between these two settings. There are 535 several reasons to question the formation of PAT 91501 in a crater floor melt sheet. On 536 Earth, these melt sheets tend to experience rapid cooling and be clast laden. Only in the 537 very largest terrestrial craters (e.g., Manicouagan, Sudbury) where impact melt sheets 538 exceed 200 m in thickness are clast-poor, igneous textured rocks observed (Keil et al., 539 1997). Likewise, fragmentation of a vesicular lava flow or impact melt sheet will occur 540 at the surface of a low-gravity, atmosphereless body. To achieve the equivalent of 541 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 24 - terrestrial atmospheric pressure, McCoy et al. (2006) calculated that a melt sheet ~130 m 542 thick would be needed on a body ~250 km in radius. These two estimates are in good 543 agreement and suggest the need for a melt sheet in excess of 100 m thickness. An impact 544 event capable of producing such a thick melt sheet on an asteroid would, instead, 545 collisionally disrupt the body (Keil et al., 1997). Further, floor melts would, at the 546 moment of crater formation, be exposed to space and, thus, the vesicles would degas. 547 Thus, we suggest that a melt dike injected into the surrounding country rock below the 548 impact crater is a more viable setting for the formation of PAT 91501. 549 Injection of molten chondritic material into the surrounding country rock provides 550 both the moderate pressure necessary for vesicle retention and a thermal environment 551 conducive to rapid cooling without quenching. We have been unable to constrain the 552 cooling rate. Although zoning within the large taenite particles might normally be taken 553 as indicative of cooling, we argue instead that the taenite(?)-troilite-pentlandite 554 assemblage is an equilibrium assemblage formed during cooling at temperatures between 555 ~300-500 ?C, consistent with phase relations in the Fe-rich portion of the Fe-Ni-S system 556 (Ma et al., 1998). With the surrounding country rock cooler than the melt, solidification 557 could have occurred in a matter of hours. The absence of distinct clasts in PAT 91501 is 558 not inconsistent with such a model. While most impact melt breccias, by definition, 559 contain unmelted clastic material from the country rock, subcrater melt dikes on Earth 560 exhibit a range of widths (Keil et al., 1997) and it is reasonable to assume that PAT 561 91501 sampled one of the wider, clast-free portions of a dike. Indeed, Bogard et al. 562 (1995) argue that a 30 cm wide zone of clast-poor impact melt in Chico samples such an 563 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 25 - intrusive dike. At a maximum dimension of ?20 cm, PAT 91501 would not be 564 extraordinary in this regard. 565 Although cooling and crystallization may have occurred relatively rapidly in this 566 dike, we suggest that it was far from a quiescent environment. Despite the preservation 567 of metal and sulfide as mm- to cm-sized nodules, it is clear from the bulk elemental 568 composition that metal and sulfide were lost from the system. Comparison of metal and 569 sulfide abundances in PAT 91501 (~0.3 and ~0.5 vol.%, respectively) with those reported 570 for average L chondrites (3.7 and 4.2 vol.%, respectively; McSween et al., 1991) suggests 571 that the melt from which PAT 91501 crystallized lost ~90% of the metal and sulfide 572 component prior to solidification. This loss is not surprising, given the marked density 573 contrast between molten metal, sulfide and silicate. In fact, a similar density contrast 574 exists between molten silicates and the vesicles, leading to rapid rise within the melt. The 575 velocity, u, of settling or rising is determined using the Stoke?s velocity equation 576 (Turcotte and Schubert, 2002) 577 u = 1 3 ?? r2 g ? (7) 578 where ?? is the difference in density between the metal, sulfide or vesicles and the silicate 579 melt; r is the radius of the grain of metal or sulfide or vesicle; g is gravity for an assumed 580 50 km radius parent body (0.012 m/s2); and ? is the viscosity of the silicate melt through 581 which the metal, sulfide or vesicle is moving. We estimated the liquidus temperature of 582 bulk L-chondrite composition to be between 1400 and 1600 ?C, which affects the 583 viscosity of the melt. Using the maximum size of the metal, sulfide and vesicles 584 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 26 - determined from the CT scan, we calculate that the largest vesicle would rise at ~3 m/hr 585 while the largest metal particle would sink at ~2 m/hour. 586 One of the most astonishing results from the CT scans is the orientation of the 587 metal/sulfide intergrowths. These orientations, reflected in the relative orientation of 588 metal to sulfide (Fig. 3) and the orientations of metal-sulfide contacts and vesicle 589 elongation ? are consistent with formation in a gravitational field. In this respect, PAT 590 91501 is exceptional in that we know which way was up while it was on the asteroid (Fig. 591 3). To the best of our knowledge, only one other meteorite can claim such a distinction. 592 In the Cape York meteorite (Kracher et al., 1977; Buchwald, 1987), elongate troilite 593 inclusions contain chromites concentrated at one end and phosphates at the other, which 594 may be indicative of formation in a gravitational field (Kracher and Kurat, 1975), but 595 have also been attributed to melt migration in a thermal gradient (Buchwald, 1987). 596 In practice, calculated velocities probably represent theoretical maximums, as the 597 vesicles likely coalesced during rise while the metal particles typically contain significant 598 amounts of attached, less-dense sulfide. Nonetheless, these calculations suggest that 599 metal-sulfide particles and vesicles should have rapidly segregated from the volume of 600 melt that eventually crystallized to form PAT 91501. 601 Unless this rock happened to capture a snapshot of metal-sulfide particles sinking 602 and vesicles rising, the retention of any vesicles or metal-sulfide requires another 603 explanation. Far from being dominated by gravitational settling or rising under the 604 influence of buoyancy alone, we suggest that the system was also influenced by the 605 movement of melt within the fracture and the binding of dense metal-sulfide and buoyant 606 vesicles to produce particles of near-neutral buoyancy. When the melt was injected into 607 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 27 - the cold country rock, it began to rise due to the marked thermal difference, and hence 608 ~10% density difference, between the melt and country rock. Using the method of 609 Wilson and Head (1981), we calculate that the melt rose at a velocity of 0.028 m/s 610 through the dike and solidified due to cooling after migration of ~220 m (McCoy et al., 611 2006). At this rate, the magma solidified after rising through the dike for ~2 hours. 612 Importantly, the rate of rise of the melt through the dike was roughly an order of 613 magnitude faster than the rate of metal-sulfide settling or vesicle rise. Thus, settling of 614 metal and sulfide to the bottom of the dike was inhibited by the rapid rise of melt through 615 the dike. 616 It also bears noting that the gravitational vector inferred from the metal-sulfide 617 contacts is inconsistent with the orientation of the vesicle long axes (Fig. 4). This slight 618 offset may result from minor turbulence in the rising magma, or possibly an additional 619 lateral component of melt movement that would be reflected in the vesicle shapes but not 620 the gravitational settling of the metal. Alternatively it may reflect the lower solidus 621 temperature of metal/sulfide compared to silicate melt, perhaps permitting time for a 622 vector change between solidifications. 623 Finally, the preservation of metal-sulfide-vesicle assemblages may result from the 624 offsetting differences in density and buoyancy. It is interesting to note that the upward 625 velocities of the average-sized gas bubbles responsible for the vesicles and the downward 626 velocity of the average-sized metal-sulfide grains are very similar at all temperatures. If 627 surface tension forces bind bubbles and grains of comparable size together, offsetting 628 buoyancy may be created that would cause the linked bubbles and sulfide grains to be 629 suspended, or at least to move only very slowly, in the melt. Neutral buoyancy has been 630 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 28 - suggested for magnetite and vesicles in the Bishop Tuff, where a vesicle either scavenged 631 magnetite crystals from the melt or served as a nucleation point for magnetite growth, in 632 the pre-eruptive magma (Gualda and Anderson, 2007). The attainment of neutral 633 buoyancy in the upward moving melt from which PAT 91501 crystallized might explain 634 the retention of even large metal-sulfide particles. 635 6. CONCLUSIONS 636 Among the abundant impact melt rocks and breccias from the L chondrite parent 637 body, PAT 91501 is unique in exhibiting cm-sized metal-sulfide particles and vesicles, 638 for the remarkable alignment of these particles, and for its ancient age. Sulfur 639 volatilization must have been a ubiquitous process during impact melting of chondritic 640 materials and other meteorites (e.g., Chico) are known that reasonably sample impact 641 melt dikes injected into the crater basement. These other meteorites do not exhibit the 642 large vesicles seen in PAT 91501. This sample must have formed during an early impact 643 on the L chondrite parent body that was large enough to form wide, clast-free melt veins. 644 This process involved relatively slow cooling and crystallization, coalescence and rise of 645 vesicles, coalescence and sinking of metal-sulfide particles, formation of metal-sulfide-646 vesicle aggregates that were neutrally buoyant, and upward flow of magma in the dike. 647 Although similar processes must have occurred in the formation of other chondritic 648 impact melt rocks, they did not combine in the unique combination that formed PAT 649 91501. 650 Acknowledgements ? We thank Robbie Score, Cecilia Satterwhite, the Meteorite 651 Processing Laboratory at Johnson Space Center, and the Meteorite Working Group for 652 providing samples. Tim Gooding (Smithsonian) provided expert technical assistance. 653 Larry Nittler and Jianhua Wang provided invaluable assistance with ion microprobe 654 analyses of sulfides and the insights of Joe Goldstein and Jijin Yang helped us understand 655 the formation of zoning within the metal-sulfide particles. We thank Bill Carlson for his 656 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 29 - early collaboration on the computed tomography aspects of this project and Ralph Harvey 657 and Roy Clarke for their insights into the terrestrial and asteroidal history of PAT 658 91501/91516/91528. Aspects of this work were supported by the NASA 659 Cosmochemistry Program (DDB, TJM, GFH) and the Becker Endowment to the 660 Smithsonian Institution (TJM). Facility support and software development at the 661 University of Texas High-Resolution X-ray CT Facility were provided by NSF grants 662 EAR-0345710 and EAR-0113480. Reviews by Henning Haack, Guy Consolmagno, and 663 Anders Meibom (AE) improved the manuscript and are much appreciated. 664 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 30 - REFERENCES 665 666 Bogard D.D. (1995) Impact ages of meteorites: A synthesis. Meteoritics 30, 244-268. 667 Bogard D.D. and Hirsch W.C. (1980) 40Ar /39Ar dating, Ar diffusion properties, and 668 cooling rate determinations of severaly shocked chondrites. Geochim. Cosmochim. 669 Acta 44, 1667-1682. 670 Bogard D.D. and Garrison D.H. (2003) 39Ar-40Ar ages of eucrites and thermal history of 671 asteroid 4 Vesta. Meteoritics & Planetary Science, 38, 669-710 672 Bogard D.D., Garrison D.H., Norman M., Scott E.R.D., and Keil K. (1995) 39Ar-40Ar age 673 and petrology of Chico: Large-scale impact melting on the L chondrite parent body. 674 Geochim. Cosmochim. Acta 59, 1383-1399. 675 Bogard D. D., Garrison D. H., and McCoy T. J. (2000) Chronology and petrology of 676 silicates from IIE meteorites: Evidence of a complex parent body evolution. 677 Geochim. Cosmochim. Acta 64, 2133-2154. 678 Buchwald V.F. (1987) Thermal migration III: Its occurrence in Cape york and other iron 679 meteorites. Meteoritics 22, 343-344. 680 Clarke R.S. Jr. (1994) In Ant. Meteorite News. (Score R. and Lindstrom M.M., Eds.) Ant. 681 Meteorite News. 17, #1, 15-16. NASA Johnson Space Center, Houston, Texas, 682 USA. 683 Dixon, E. T.; Bogard, D. D.; Garrison, D. H.; Rubin, A. E. (2004) 39Ar-40Ar evidence for 684 early impact events on the LL parent body. Geochim. Cosmochim. Acta, 68, 3779-685 3790. 686 Eberhardt P., Eugster O., Geiss J., and Marti K. (1966) Rare gas measurements in 30 687 stone meteorites. Naturforsch. 21A, 414-426. 688 Eugster O. (1988) Cosmic-ray production rates for He-3, Ne-21, Ar-38, Kr-83, and Xe-689 126 in chondrites based on Kr-81/Kr exposure ages. Geochim. Cosmochim. Acta 690 52, 1649-1662. 691 Garrison D. H. and Bogard D. D. (2001) 39Ar-40Ar and space exposure ages of the unique 692 Portales Valley H-chondrite. Lunar Planet. Sci. 32, #1137. 693 Graf Th. and Marti K. (1995) Collisional records in LL chondrites. J. Geophys. Res. 694 (Planets) 100, 21247-21263. 695 Graf Th., Signer P., Wieler R., Herpers U., Sarafin R., Vogt S., Fieni Ch., Pellas P., 696 Bonani G., Suter M., and W?lfli W. (1990a) Cosmogenic nuclides and nuclear 697 tracks in the chondrite Knyahinya. Geochim. Cosmochim. Acta 54, 2511?2520. 698 Graf Th., Baur H., and Signer P. (1990b) A model for the production of cosmogenic 699 nuclides in chondrites. Geochim. Cosmochim. Acta 54, 2521?2534. 700 Gualda G.A.R. and Anderson A.T. Jr. (2007) Magnetite scavenging and the buoyancy of 701 bubbles in magma. Part 1: Discovery of a pre-eruptive bubble in Bishop rhyolite. 702 Contrib. Mineral. Petrol. 153, 733-742. 703 Haack, H., Farinella, P., Scott E.R.D., and Keil, K (1996) Meteoritic, asteroidal, and 704 theoretical constraints on the 500 MA disruption of the L chondrite parent body. 705 Icarus 119, 182-191. 706 Haack H., Pedersen T. P., and Rasmussen K. L. (2000) Portales Valley ? thermal history 707 of a unique meteorite. Meteoritics & Planetary Science, 35, A67. 708 Hartmann W.K., Ryder G., Dones L., and Grinspoon D. (2000) The time-dependent 709 intense bombardment of the primordial Earth/Moon system. In Origin of the Earth 710 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 31 - and Moon (R.M. Canup and K. Righter, eds.), University of Arizona Press, Tucson, 711 pp 493-512. 712 Harvey R.P. and Roedder E. (1994) Melt inclusions in PAT 91501: Evidence from 713 crystallization from an L chondrite impact melt. Lunar Planet. Sci. 25, 513. 714 Herzog G.F. (2005) Cosmic-ray exposure ages of meteorites, pp. 347-380. In Meteorites, 715 Comets, and Planets (ed. A.M. Davis) Vol. 1 Treatise on Geochemistry (eds. H.D. 716 Holland and K.K. Turekian), Elsevier-Pergamon, Oxford. 717 Herzog G. F., Vogt S., Albrecht A., Xue S., Fink D., Klein J., Middleton R., 718 Weber H. W., and Schultz L. (1997) Complex exposure histories for meteorites 719 with "short" exposure ages. Meteoritics, 32, 413-422 720 Jarosewich, E. (1990) Chemical analyses of meteorites: A compilation of stony and iron 721 meteorite analyses. Meteoritics 25, 323-337. 722 Jurewicz A.J.G., Mittlefehldt D.W., and Jones J.H. (1995) Experimental partial melting 723 of the St. Severin (LL) and Lost City (H) chondrites. Geochim. Cosmochim. Acta 724 59, 391-408. 725 Keil K., St?ffler D., Love S.G., and Scott E.R.D. (1997) Constraints on the role of impact 726 heating and melting in asteroids. Meteoritics and Planet. Sci. 32, 349-363. 727 Ketcham R. A. (2005) Computational methods for quantitative analysis of three-728 dimensional features in geological specimens. Geosphere 1(1), 32-41. 729 Ketcham R. A. and Carlson W. D. (2001) Acquisition, optimization and interpretation of 730 X-ray computed tomographic imagery: Applications to the geosciences. Computers 731 and Geosciences 27, 381-400. 732 Koorochantseva E. V., Trieloff M., Lorenz C. A., Buykin A. I., Ivanova M. A., Schwartz 733 W. H., Hopp J., and Jessberger E. K. (2007). L-chondrite asteroid breakup tied to 734 Ordovician meteorite shower by multiple isochron 40Ar-39Ar dating. Meteoritics and 735 Planet. Sci. 42, 113-130. 736 Kracher A. and Kurat G. (1975) An unusual phosphate-sulfide assemblage in the Cape 737 York iron meteorite. Meteoritics 10, 429. 738 Kracher A., Kurat G., and Buchwald V.F. (1977) Cape York: The extraordinary 739 mineralogy of an ordinary iron meteorite and its implication for the genesis of 740 IIIAB irons. Geochem. J. 11, 207-217. 741 Kring D.A., Swindle T.D., Britt D.T., and Grier J.A. (1996) Cat Mountain: A meteoritic 742 sample of an impact-melted asteroid regolith. J. Geophys. Res., 101, 29,353-29,371. 743 Kring D. A., Hill D. H., Gleason J. D., Britt D. T., Consolmago G. J., Farmer M., Wilson 744 S., and Haag R. (1999) Portales Valley: a meteoritic sample of the brecciated and 745 metal-veined flor of an impact crater on an H-chondrite asteroid. Meteoritics and 746 Planet. Sci. 34, 663-669. 747 La Croix L.M. and McCoy T.J. (2007) Shock Classification of Antarctic Ordinary 748 Chondrites. Lunar and Planet. Sci. Conf. XXXVIII, abst# 1601. 749 Lauretta D.S., Lodders K., Fegley B., and Kremser D.T. (1997) The origin of sulfide-750 rimmed metal grains in ordinary chondrites. Earth and Planet. Sci. 151, 289-301. 751 Leya I., Lange H.-J., Neumann S., Wieler R., and Michel R. (2000) The production of 752 cosmogenic nuclides in stony meteoroids by galactic cosmic ray particles. Meteorit. 753 Planet. Sci. 35, 259?286. 754 Leya I., Wieler R., Aggrey K., Herzog G. F., Schnabel C., Metzler K., Hildebrand A. R., 755 Bouchard M., Jull A. J. T., Andrews H. R., Wang M.-S., Ferko T. E., Lipschutz M. 756 E., Wacker J. F., Neumann S., and Michel R. (2001) Exposure history of the St-757 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 32 - Robert (H5) fall. Meteorit. Planet. Sci. 36, 1479?1494. 758 Ma L., Williams D.B. and Goldstein J.I. (1998) Determination of the Fe-rich portion of 759 the Fe-Ni-S phase diagram. J. Phase Equilibria 19, 299-309. 760 Masarik J., Nishiizumi K., and Reedy R. C. (2001) Production rates of 3He, 21Ne and 761 22Ne in ordinary chondrites and the lunar surface. Meteorit. Planet. Sci. 36, 643?762 650. 763 Marlow R., Score R. and Mason B. (1992) In Ant. Meteorite News. (Score R. and 764 Lindstrom M.M., Eds.) 15, #2, 30. NASA Johnson Space Center, Houston, Texas, 765 USA 39pp. 766 McCoy T.J., Keil K., Bogard D.D., Garrison D.H., Casanova I., Lindstrom M.M., 767 Brearley A.J., Kehm K., Nichols R.H. Jr. and Hohenberg C.M. (1995) Origin and 768 history of impact-melt rocks of enstatite chondrite parentage. Geochim. 769 Cosmochim. Acta 52, 161-175. 770 McCoy T.J., Ketcham R.A., Wilson L., Benedix G.K., Wadhwa M., and Davis A.M. 771 (2006) Formation of vesicles in asteroidal basaltic meteorites. Earth and Planet. 772 Sci. 246, 102-108. 773 McSween H.Y., Bennett M.E., and Jarosewich E. (1991) The mineralogy of ordinary 774 chondrites and implications for asteroid spectrophotometry. Icarus 91, 107-116. 775 Melosh H.J. (1989) Impact Cratering: A Geologic Process. Oxford Univ. Press, New 776 York, New York, USA. 245pp. 777 Mittlefehldt D.W and Lindstrom M.M. (2001) Petrology and geochemistry of Patuxent 778 Range 91501 and Lewis Cliff 88663. Meteoritics and Planet. Sci. 36, 439-457. 779 Middleton R. and Klein J. (1986) A new method for measuring 10Be/9Be ratios. Proc. 780 Workshop Tech. Accel. Mass Spectrom., (Eds. R.E.M. Hedges and E.T. Hall) June 781 30-July 1, 1986, Oxford, England, 76-81. 782 Middleton R., Klein J., Raisbeck G.M. and Yiou F. (1983) Accelerator mass spectrometry 783 with aluminum-26. Nucl. Instru. Meth. Phys. Res. 218, 430-438. 784 Oppenheimer C. (2004) Volcanic degassing, in: R. Rudnick (Ed.), The Crust, Elsevier-785 Pergamon, Oxford, pp. 123-166. 786 Pellas P. and Fi?ni C. (1988) Thermal histories of ordinary chondrite parent asteroids. 787 Lunar and Planet. Sci. Conf. XIX, 915-916 (abs). 788 Rubin A.E. (1985) Impact melt products of chondritic material Rev. Geophys. 23, 277-789 300 790 Rubin A.E. (2002) Smyer H-chondrite impact-melt breccia and evidence for sulfur 791 vaporization. Geochim. Cosmochim. Acta, 66, 699-711. 792 Scott, E.R.D. (2002) Meteorite evidence for the accretion and collisional evolution of 793 asteroids, Chapter in Asteroids III, W. Bottke, A. Cellino, P. Paolicchoi, and R. 794 Binzel, (eds.), University of Arizona Press, p. 697-709. 795 Scott E. R. D., Haack H., and Love S. G. (2001) Formation of mesosiderites by 796 fragmentation and reaccretion of a large differentiated asteroid. Meteoritics and 797 Planet. Sci. 36, 869-881. 798 St?ffler D., Keil K., and Scott E.R.D. (1991) Shocke metamorphism of ordinary 799 chondrites. Geochim. Cosmochim. Acta 55, 3845-3867. 800 Taylor G.J., Keil K., Berkley J.L., Lange D.E., Fodor R.V., and Fruland R.M. (1979) The 801 Shaw meteorites: History of a chondrite consisting of impact-melted and 802 metamorphis lithologies. Geochim. Cosmochim. Acta. 43, 323-337. 803 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 33 - Turcotte D. L. and Schubert G. (2002) Geodynamics 2nd Ed. Cambridge Univ. Press, 804 New York, New York, USA 456pp. 805 Turner G., Enright M.C., Cadogan P.H. (1978) The early history of chondrite parent 806 bodies inferred from Ar-40-Ar-39 ages. Proc. 9th Lunar and Planet. Sci. Conf., 807 989-1025. 808 Vogt S. and Herpers U. (1988) Radiochemical separation techniques for the 809 determination of long-lived radionuclides inmeteorites by means of accelerator 810 mass spectrometry. Fresenius Z. Anal. Chemie 331, 186-188. 811 Wilson L. and Head J.W. (1981) Ascent and eruption of basaltic magma on the Earth and 812 Moon. J. Geophys. Res. 86, 2971-3001. 813 Yamaguchi A., Taylor G. J., Keil K. Floss C., Crozaz G., Nyquist L. E., Bogard D. D., 814 Garrison D. H., Reese Y. D., Wiesmann H., and Shih C.-Y. (2001) Post-815 crystallization reheating and partial melting of eucrite EET90020 by impact into the 816 hot crust of asteroid 4Vesta ~4.50 Gyr ago. Geochim. Cosmochim. Acta, 65, 3577-817 3599. 818 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 34 - 819 820 Table 1. Noble gas abundances in two samples of PAT 91501 3He 4He 20Ne 21Ne 22Ne 36Ar 38Ar 40Ar 10-7 10-6 10-8 10-8 10-8 10-9 10-9 10-5 Sample ,109 500oC 2.33 8.58 0.32 0.30 0.38 0.53 0.47 1.02 1550oC 2.21 36.60 9.12 9.97 10.76 9.42 10.06 2.62 Sample ,106 500oC 2.70 5.73 0.31 0.029 0.38 0.48 0.48 1.07 900oC 1.68 24.46 3.30 3.56 3.88 1.53 0.87 1.29 1200oC 0.24 0.34 3.24 3.51 3.83 4.50 5.36 0.62 1550oC 0.03 0.12 1.15 1.26 1.37 1.68 2.35 0.17 821 Table 2. Abundances of cosmogenic species and cosmic-ray exposure ages of PAT 91501. Sample ,106 ,109 3He 46.5 45.4 21Ne 8.62 10.27 38Ar 0.86 0.99 20Ne/22Ne 0.845?0.015 0.847?0.005 22Ne/21Ne 1.097?0.003 1.084?0.003 T3 28.7 28.0 T21 24.6 27.5 T38 20.6 23.1 T10-21 29.6 T26-21 25.5 Sample ,34 ,38 ,40 ,42 10Be 20.8 20.6 20.3 26Al 64.9 61.9 60.6 55.2 Noble gas concentrations in 10-8 cm3 STP/g. Cosmic-ray exposure ages, T, in Myr. 10Be and 26Al activities in dpm/kg; uncertainties are estimated to be ?7%. T10-21 and T26-21 after Graf et al. (1990a). 822 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 35 - Figure Captions 823 Figure 1. Photograph of PAT 91501 ,50. Numerous vesicles and metal-sulfide grains (up 824 to cm-sized) are visible on the cut surface. Cracks throughout sample are likely due 825 to terrestrial weathering. Scale cube is 1cm on a side. 826 Figure 2. A) Reflected light optical photomicrograph of an intergrown metal-sulfide 827 particle in contact with a vesicle in PAT 91501 (,111). The particle has been etched 828 to show the metallographic texture consisting of mainly taenite (t) and martensite 829 (m). Troilite (tr) with small particles of embedded pentlandite (p) rims the entire 830 particle. B) Nickel composition (following traverse illustrated in A) across the two 831 domains showing high-Ni inclusion-free (or poor) taenite rims grading into 832 intermediate-Ni martensitic cores. 833 Figure 3. A single frame from the 3 dimensional rotation visualization made from the CT 834 scan of PAT 91501 (,50), in which vesicles and metal-sulfide intergrowths are 835 highlighted. Metal (yellow), sulfide (magenta) and vesicles (blue bubbles) are set 836 in a semi-transparent outline of the specimen pictured in Figure 1. Arrow points to 837 prominent, large vesicle seen in Fig. 1. The specimen is oriented as it would have 838 been at the time of crystallization as suggested by the metal-sulfide orientations 839 (sulfide above metal in all instances). Note, however, that long axes of vesicles and 840 metal-sulfide masses are offset somewhat to the left. 841 Figure 4. Stereo plots from PAT 91501 (,50) of a) the normals to the planes defined by 842 the contact between metal and sulfide with the size of the each circle proportional 843 to the area of the contact and b) orientations of the vesicle long axes with the circle 844 areas proportional to vesicle volume. Clustering of orientations are observed for 845 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 36 - both metal-sulfide contacts and vesicle elongation. See text for discussion of 846 orientation directions. 847 Figure 5. Ar-Ar ages (Gyr, rectangles, left scale) and K/Ca ratios (stepped line, right 848 scale) as a function of cumulative release of 39Ar for temperature extractions of a 849 melt sample of PAT 91501. Seven extractions releasing 30-78% of the 39Ar give an 850 age of 4.461 ?0.008 Gyr, which we interpret to be the formation time of the PAT 851 91501. 852 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 37 - 853 854 Fig.1 855 856 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 38 - 857 Fig.2 858 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 39 - 859 Fig.3 860 861 Fig.4 862 ACCEPTED MANUSCRIPT AC CE PT ED M AN US CR IP T - 40 - 863 Fig.5 864 865