Università di Pisa

Dipartimento di Scienze della Terra

Scuola di Dottorato in Scienze di Base “Galileo Galilei” Programma in Scienze della Terra XXVII Ciclo SSD GEO/07

SHOCK METAMORPHISM AND IMPACT MELTING AT KAMIL CRATER, EGYPT

PhD Student Advisor Prof. Massimo D’Orazio Agnese Fazio Co-advisor Dott. Luigi Folco

Anno Accademico 2013-2014

Ricorda: “Quando stai per rinunciare, quando senti che la vita è stata troppo dura con te, ricordati chi sei. Ricorda il tuo sogno”. (Il Delfino - S. Bambarén)

TABLE OF CONTENTS

ABSTRACT 7

RIASSUNTO 9

PREFACE 11

1. INTRODUCTION 13 1.1. IMPACT CRATERING AS A TERRESTRIAL GEOLOGICAL PROCESS 13 1.2. IMPACT CRATERING STAGES 17 1.3. SHOCK METAMORPHISM 21 1.3.1. Quartz 24 1.3.2. Deformation in other minerals 28 1.3.3. Selective and localized melting 29 1.4. IMPACT MELTING 31 1.5. SHOCK EFFECTS IN QUARTZ-BEARING ROCKS: CRYSTALLINE VS. SEDIMENTARY TARGETS 34 1.6. REFERENCES 37

2. SHOCK METAMORPHISM AND IMPACT MELTING IN SMALL IMPACT CRATERS ON EARTH: EVIDENCE FROM KAMIL CRATER, EGYPT 41

3. TARGET-PROJECTILE INTERACTION DURING IMPACT MELTING AT KAMIL CRATER, EGYPT 89

4. MICROSCOPIC IMPACTOR DEBRIS IN THE SOIL AROUND KAMIL CRATER (EGYPT): INVENTORY, DISTRIBUTION, TOTAL MASS AND IMPLICATIONS FOR THE IMPACT SCENARIO 131

5. CONCLUSIONS 161

6. FUTURE WORK 165 6.1. COMBINED MICRO-RAMAN AND TEM STUDY OF HIGH-PRESSURE PHASES FROM KAMIL CRATER (EGYPT): IMPLICATIONS FOR THEIR FORMATION IN SMALL IMPACT CRATERS ON EARTH 165 6.2. LIQUID IMMISCIBILITY FEATURES IN IMPACT MELTS 165 6.3. REFERENCES 166

APPENDIX I. USE OF THE UNIVERSAL STAGE (U-STAGE) FOR INDEXING PLANAR DEFORMATION FEATURES IN QUARTZ 169

APPENDIX II. THE EXTREMELY REDUCED SILICATE-BEARING IRON METEORITE NORTHWEST AFRICA 6583: IMPLICATIONS ON THE VARIETY OF THE IMPACT MELT ROCKS OF THE IAB-COMPLEX PARENT BODY 175

APPENDIX III. OTHER ACTIVITIES 207

ACKNOWLEDGMENTS 209

ABSTRACT

Shock effects in small terrestrial impact craters (diameter < 300 m) have been poorly studied because small craters are rare and often deeply eroded. Kamil is a young (< 5000 yr), small (45-m-in-diameter), and well preserved impact structure caused by the hypervelocity impact of the iron meteorite Gebel Kamil on sedimentary rocks in southwestern Egypt. Its pristine state of preservation makes Kamil a natural laboratory for the study of the cratering process of small impactors (about 1-m-in-diameter) on Earth, their consequences, and their impact on the terrestrial environment for hazard assessment.

This PhD Thesis deals with the definition of the shock metamorphism and impact melting in small terrestrial impact craters through a comprehensive mineralogical, petrographic, and geochemical study of shock-metamorphosed rocks and impact melts from Kamil. This study also allows us to constrain the impact cratering process related to the impact of meter-sized iron meteorites on Earth.

The results of this PhD Thesis highlight for the first time that a meter size iron body impacting on a sedimentary target can produce a wide range of shock features. These divide into two categories as a function of their abundance at the thin section scale: i) pervasive shock features (the most abundant), including fracturing, planar deformation features, and impact melt lapilli and bombs, and ii) localized shock features including high-pressure phases and localized impact melting in the form of intergranular melt, melt veins, and melt films in shatter cones. Pervasive shock features indicate the shock pressure suffered by rocks. The most shocked samples (impact melt lapilli and bombs) indicate that the shock pressure at the contact point between the projectile and the target was between 30 and 60 GPa. Based on the planar impact approximation model, this implies that the impact velocity of Gebel Kamil was at least 5 km s-1, for an impact angle of 45°. Localized shock features formed from the local enhancement of shock pressure and temperature at pores and/or at the heterogeneities of the target rocks. Thus, it is possible to find high-pressure phases and intergranular melting in sample that suffered low or moderate shock pressures.

In small meteorite impacts, the projectile may survive the impact through fragmentation. In addition, it may melt and interact with both shocked and melted target rocks. The interaction between target and projectile liquids is a process yet to be completely understood. Impact melt lapilli and bombs from Kamil are very fresh and their study can help constrain the target-projectile interaction. Two types of glasses constitute the impact melt

7 lapilli and bombs: a white glass and a dark glass. The white glass is inclusion-free, mostly

SiO2, and has negligible amounts of Ni and Co, suggesting derivation from the target rocks with negligible interaction with the projectile liquid (<0.1 wt% of projectile contamination). The dark glass is made of a silicate glass with variable amounts of Al, Fe, and Ni. It also includes variously shocked and melted fragments from the target and projectile (Ni-Fe metal blebs). All this indicates an extensive interaction with the projectile liquid. The dark glass is thus a mixture of target and projectile (estimated projectile contamination 11-12 wt%) liquids. Based on the recently proposed models for the target-projectile interaction and for impact glass formation, we propose a model for the glass formation at Kamil. Between the contact and compression stage and the excavation stage, projectile and target liquids can chemically interact in a restricted zone. The projectile contamination affected only a shallow portion of the impacted target rocks. White glass formed out of this zone, escaping interaction with the projectile. During the excavation stage, due to a brief and chaotic time sequence and the high temperature, dark glass engulfed and coated white glass and target fragments and stuck on iron meteorite shrapnel fragments.

The microscopic impactor debris, systematically collected from the soil around Kamil, includes vesicular masses, spherules, and coatings of dark impact melt glass that is a mixture of impactor and target materials (Si, Fe, Al-rich glass), and Fe-Ni oxide spherules and mini shrapnel fragments. As a consequence of an oblique impact, this material formed a downrange ejecta curtain of microscopic impactor debris due SE-SW of the crater (extension ~300,000 m2, up to ~400 m from the crater), consistent with previous determination of the impactor trajectory. The Ni contents of the soil provided an estimate of the mass of the microscopic debris of the Gebel Kamil meteorite dispersed in the soil. This mass (<290 kg) is a small fraction of the total impactor mass (~10 t) in the form of macroscopic shrapnel. Kamil Crater was generated by a relative small impactor that is consistent with literature estimates of its pre-atmospheric mass (>20 t, likely 50-60 t).

8 RIASSUNTO

Gli effetti di shock registrati in piccoli crateri di impatto terrestri (diametro < 300 m) sono stati poco studiati perché i piccoli crateri sono rari e spesso profondamente erosi. Kamil è una struttura di impatto giovane (<5000 anni), di piccole dimensioni (45 m di diametro) e ben preservata. È stata prodotta dall’impatto iperveloce della meteorite metallica Gebel Kamil su rocce sedimentarie dell’Egitto sudoccidentale. Il suo ottimo stato di preservazione permette di considerare Kamil un laboratorio naturale per studiare il processo di craterizzazione di piccoli impattori (circa 1 m di diametro) sulla Terra, le loro conseguenze e il loro impatto sull’ambiente terrestre per la valutazione del rischio.

La presente Tesi di Dottorato contribuisce alla definizione del metamorfismo da shock e della fusione da impatto in piccoli crateri terrestri attraverso uno studio comprensivo di tipo mineralogico, petrografico e geochimico delle impattiti e dei vetri da impatto di Kamil. Questo studio permette inoltre di ampliare le conoscenze sul processo di craterizzazione legato all’impatto di meteoriti metalliche di dimensioni metriche sulla Terra.

I risultati di questa Tesi di Dottorato evidenziano per la prima volta che un corpo metallico di dimensioni metriche può produrre una vasta gamma di strutture e associazioni mineralogiche da shock impattando su un target sedimentario. Queste sono state divise in due categorie, in funzione della loro abbondanza a scala della sezione sottile: i) effetti di shock pervasivi (più abbondanti), comprendenti fratturazione irregolare, piani di materiale amorfo orientati parallelamente agli indici cristallografici del quarzo o Planar Deformation Features (PDFs), lapilli e bombe di vetro da impatto; ii) effetti di shock localizzati (meno abbondanti) comprendenti fasi di alta pressione e fusione localizzata in forma di vetro intergranulare, vene e film di vetro su shatter cones (strutture coniche caratterizzate da strie disposte a coda di cavallo). Gli effetti di shock pervasivi indicano la pressione subita dalla roccia. I campioni più shockati (lapilli e bombe di vetro) indicano che la pressione al punto di contatto tra la meteorite e il proiettile era tra i 30 e 60 GPa. Sulla base del modello dell’approssimazione planare di impatto, la velocità minima di impatto della meteorite Gebel Kamil era ~5 km s-1, assumendo un angolo di impatto di 45°. Gli effetti di shock localizzati si sono formati come conseguenza di un aumento della pressione e temperatura di shock in corrispondenza dei pori e/o di eterogeneità delle rocce del target. Per questo è possibile trovare fasi di alta pressione e vetro intergranulare in campioni che hanno subito basse o moderate pressioni di shock.

9 In piccoli impatti di meteoriti, il proiettile può sopravvivere all’impatto, frammentandosi e fondendo. Il suo fuso potrà interagire con rocce del target shockate e fuse. L’interazione tra i fusi del proiettile e del target è un processo ancora non pienamente compreso. I lapilli e le bombe di vetro da impatto di Kamil sono molto freschi e il loro studio può aiutare a vincolare il processo di interazione tra target e proiettile. Due tipi di vetri costituiscono i lapilli e le bombe di vetro da impatto: un vetro bianco e un vetro scuro. Il vetro bianco è privo di inclusioni, è principalmente costituito da SiO2 e ha contenuti trascurabili di Ni e Co, suggerendo una derivazione dalle rocce target con un’interazione trascurabile con il proiettile (<0.1 wt% di contaminazione del proiettile). Il vetro scuro è costituito da un vetro silicatico con contenuti variabili di Al, Fe e Ni. Il vetro scuro è così una mistura dei liquidi del target e proiettile (contaminazione stimata del proiettile: 11-12 wt%). Il vetro scuro inoltre include frammenti fusi e variamente shockati del target e sferule metalliche di Ni e Fe. Tutto questo indica un’estensiva interazione con il proiettile. Sulla base dei recenti modelli sull’interazione tra target e proiettile e sulla formazione dei vetri da impatto, noi proponiamo un modello per la formazione dei vetri a Kamil. Tra lo stadio di contatto e compressione e lo stadio di escavazione, i fusi del proiettile e del target possono interagire chimicamente in una zona ristretta. La contaminazione del proiettile riguarda solo la porzione più superficiale delle rocce coinvolte dall’impatto. Il vetro bianco si forma fuori da questa zona, senza interagire con il liquido del proiettile. Durante lo stadio di escavazione, a causa della breve e caotica sequenza di eventi e delle alte temperature, il vetro scuro può inglobare e avvolgere il vetro chiaro e frammenti del target e può attaccarsi ai frammenti di meteorite.

Il detrito microscopico dell’impattore, sistematicamente campionato dal suolo circostante Kamil, comprende masse di vetro vescicolare, sferule, e pellicole di vetro scuro (vetro ricco di Si, Fe e Al, analogo al vetro scuro di lapilli e delle bombe) su frammenti di target e di vetro chiaro, sferule di ossidi di Fe e Ni, e piccoli frammenti di meteorite. Come conseguenza di un impatto obliquo questo materiale ha formato una coltre di ejecta verso SE- SW del cratere (estensione ~300,000 m2, fino a ~400 m dal cratere), questo è consistente con gli studi precedenti sulla traiettoria di impatto. Il contenuto Ni del suolo fornisce una stima della massa del detrito microscopico del meteorite Gebel Kamil disperso nel suolo. Questa massa (<290 kg) è una piccola frazione della massa totale dell’impattore (~10 t) in forma di frammenti macroscopici. Il cratere Kamil è stato generato da un impattore relativamente piccolo, questo è consistente con le stime della sua massa pre-atmosferica (>20 t, probabilmente 50-60 t) riportate in letteratura.

10 PREFACE

The aim of this PhD Thesis is the definition of the shock metamorphism and impact melting in small impact craters on Earth through the study of the impactites (rocks produced by hypervelocity impacts) found at Kamil Crater (Egypt). Kamil is a small impact crater (45- m-diameter) discovered in 2008. It was produced by the hypervelocity impact of the iron meteorite Gebel Kamil most likely < 5000 yr ago. This crater is very well preserved and is located in a very simple geological setting.

This PhD Thesis is organized into six chapters and three appendices. Chapter 1 is a brief digression on all the main aspects concerning impact cratering. In this chapter the answers to the following questions can be found: How many impact craters are there on Earth? Where are they located? How large are they? Why is impact cratering a unique process? How does an impact crater form? Which are the shock effects produced by impacts? Can the target rocks influence the shock effects? This chapter has the scope to support the understanding of the following three chapters. They are structured as scientific papers and they are accepted or submitted to international journals.

Chapter 2 is the paper entitled “Shock metamorphism and impact melting in small impact craters on Earth: evidence from Kamil Crater, Egypt”, by Agnese Fazio, Luigi Folco, Massimo D’Orazio, Maria Luce Frezzotti, and Carole Cordier. It is a compendium of all the shock effects that we found studying rocks from Kamil. The aim of this paper is to establish the impact velocity of the meteorite Gebel Kamil on the basis of shock effects. Moreover, this paper highlights that, due to the wide range and the freshness of the shock features, Kamil can be considered a natural laboratory for studying impact cratering and shock deformation processes in small impact structures. This paper was submitted to the international peer- reviewed journal Meteoritics & Planetary Science on 20th March 2014, accepted on 15th August 2014, and published on the December issue (vol. 49, pp. 2175-2200).

Chapter 3, is the paper entitled “Target-projectile interaction during impact melting at Kamil Crater, Egypt”, by Agnese Fazio, Massimo D’Orazio, Carole Cordier, and Luigi Folco. It is a detailed petrographic and geochemical study of the impact melt lapilli and bombs from Kamil. Impact melt masses are the most significant samples for the study of the process of the interaction between the target and the projectile liquids during the formation of Kamil, helping to constrain the general impact scenario. This paper will be submitted to an international peer-reviewed journal.

11 Chapter 4, is a paper entitled “Microscopic impactor debris in the soil around Kamil Crater (Egypt): inventory, distribution, total mass and implications for the impact scenario”, by Luigi Folco, Massimo D’Orazio, Agnese Fazio, Carole Cordier, Antonio Zeoli, Matthias van Ginneken, and Ahmed El-Barkooky. It is a report on the microscopic impactor debris collected around the crater. Beyond presenting an inventory of all microscopic particle types, this study allows to constrain the impactor mass at the time of the impact. This paper was submitted to the international peer-reviewed journal Meteoritics & Planetary Science on 18th August 2014 and it was accepted on 11th December 2014.

The conclusions of this PhD Thesis are reported in Chapter 5. Finally, future activities and projects related to the present PhD topic are discussed in Chapter 6.

Appendix I provides detailed instructions for the use of the Universal Stage (U-Stage) in indexing planar deformation features in shocked quartz. The U-Stage is considered an outdated technique by the most part of modern geologists. However, in the last years it is back to the top among planetary geologists as a low-cost and widespread technique for the indexing of planar deformation features in quartz.

Appendix II reports of the paper entitled “The extremely reduced silicate-bearing iron meteorite Northwest Africa 6583: implications on the variety of the impact melt rocks of the IAB-complex parent body” by Agnese Fazio, Massimo D’Orazio, Luigi Folco, Jérôme Gattacceca, Corinne Sonzogni). I started the study of this meteorite for my Master Thesis (Tesi di Laurea Magistrale), and I concluded it during the first months of my PhD. This paper was published in the international peer-reviewed journal Meteoritics & Planetary Science in the volume 48, number 12 of December 2013.

Appendix III lists the other activities that I conducted during my PhD.

12 1. INTRODUCTION

1.1. Impact cratering as a terrestrial geological process

Impact cratering is the most common geological process of the Solar System. On Solar System bodies, impact cratering plays an important role modeling their surfaces and shapes. Moreover, cosmic impacts can catastrophically destroy solar system bodies or form new ones (e.g., the Moon), change their surface geochemistry, generate abrupt climate changes and mass extinctions, and allow the formation of ore deposits, hydrocarbons reservoirs, and biological niches (Osinski 2008).

Impact craters are the main geological landforms of the solid bodies of the Solar System (Fig. 1) with the exception of the Earth. On Earth, impact structures are not preserved for long because they are continually obliterated by tectonic activity, erosion, burial, weathering, volcanic resurfacing, and vegetation. The first impact crater on Earth was recognized in 1905. It was the Barringer Crater (also known as Meteor; Fig. 2). However, it was only in the late 1960s that impact cratering was recognized as an important geological process not only of the Solar System but also of the Earth. Indeed, in those years, unique petrological and geochemical features produced by the passage of intense shock waves were recognized (see section 1.3. Shock metamorphism) and new high-resolution images of the Solar System bodies were obtained by satellites and modern telescopes (for review see French 1998; Osinski and Pierazzo 2013).

Fig. 1. The surface of Mercury. Credit: Nasa/Johns Hopkins University Applied Physics Laboratory/Carnegie Institution of Washington.

13

Fig. 2. Barringer Crater, Arizona, USA. Nowadays, 184 impact structures have been recognized on Earth (Fig. 3; http://www.passc.net/EarthImpactDatabase/index.html.). They range from 13.5 m (Carancas, Peru) up to 300 km (Vredefort, South Africa). The surfaces of the Solar System solid bodies are characterized by a random distribution of impact craters. However, looking at the distribution of impact craters on Earth, the distribution is not random, but it reflects geological and social factors. This is due to three main reasons: (i) the age of the surfaces and their geological stability show that the best areas for the preservation of impact structures are continental shields and cratonic areas; (ii) the difficulties in searching for impact structure under the sea, due to the high exploration costs and the young age of oceanic crust; (iii) systematic search has been done only on specific continental areas (i.e., Canada, Russia, Ukraine, and Australia).

Fig. 3. Distribution of Earth’s craters. Small craters (< 300 m in diameter) have been highlighted by a larger circle. Modified after http://www.passc.net/EarthImpactDatabase/index.html.

14 The rate of the impacts follows an exponential trend with impactor size. Small bodies are more abundant and their collisions are more probable. Impacts producing small craters (< 300 m in diameter) on Earth occur on a decadal to a secular time scale; impacts producing craters > 200 km in diameter occur on a billion time scale (e.g., Bland and Artemieva 2006). However, Fig. 4 shows that craters between 0 and 1 km are a quarter of the impact craters between 1 and 10 km, indicating an incomplete record of small impact structures on Earth. In Fig. 3 the 17 small impact craters have been highlighted by a larger circle. Small craters are quickly eroded or buried in sediments and their features are easily lost (Fig. 5). This means that our knowledge about the formation mechanisms and the shock metamorphic features of craters of these proportions is not detailed (see section 1.3. Shock metamorphism). Moreover, this is an important limitation to the knowledge of the hazard that small impactors constitute to human populations.

100 90 80 70

60 0 - 1 km 50 1 - 10 km 40 11-100 km

Numer of Crater 30 101-1000 km 20 10 0

Fig. 4. Histogram showing the number of impact craters on Earth for each diameter class.

15

Fig. 5. Example of three terrestrial small craters (< 300 m in diameter). Source: http://www.passc.net/EarthImpactDatabase/index.html. Impact cratering is a peculiar geological process. Its differences with respect to the other more familiar geological processes contributed to the initial skepticism for this process. These differences have been summarized in six points by French (1998):

(i) Rarity. In the whole human history, only two impact cratering events were observed by humans (Sikhote-Alin, Russia, in 1947, and Carancas, Peru, in 2007). In both cases, they are small impact craters (Sikhote-Alin formed a crater field and the largest crater is 26 m in diameter; Carancas formed one single crater 13 m in diameter). For this reason, we have only indirect evidence of the regional and global risk that a large meteorite impact can cause. (ii) Immense energy. The energy of a large impact is not comparable to the energy release by earthquakes or volcanic eruptions. It has been calculated that the formation of the Barringer Crater released an energy of three orders of magnitudes higher than the Hiroshima atomic bomb, and that the impact responsible for the mass extinction of large reptiles (Chicxulub, Mexico, 180 km in diameter) was two orders of magnitude more energetic than the total annual energy release from Earth (heat flow, seismic, and volcanic eruption). (iii) Instant effects. By contrast to the other geological processes, impact cratering is a virtually instantaneous process. It has been estimated that a crater of 1 km in diameter (hence similar to Barringer Crater) forms in a few seconds, while less than 10 minutes are necessary to form impact structures 200 km in diameter, like Sudbury (Canada) or Vredefort (South Africa). However, post-shock modification

16 (i.e., geological adjustments, mostly due to gravity) in these very large impact structures can continue for many years. (iv) Concentrated energy release. Besides being immense and instantaneously released, the energy is also released virtually in a single point of Earth’s surfaces. Then, it is rapidly transferred into target rocks, atmosphere, and biosphere. (v) Extreme physical conditions. The shock waves propagate into the target rocks at supersonic velocities of about 5–8 km/s. Rocks involved in this passage suffered pressures, temperatures and strain of several orders of magnitude higher than those reached by endogenic metamorphism (for details on shock metamorphism see section 1.3. Shock metamorphism). (vi) Unique deformation effects. The effects produced by the propagation of shock waves into target rocks cause the formation of unique deformation effects. The occurrence of these unique features is nowadays the most important criterion for establishing the impact origin of a crater. For details on the shock metamorphic futures see section 1.3. Shock metamorphism.

1.2. Impact cratering stages

On Earth, impact craters form when a projectile is large enough (typically > 50 m for stony meteorites and > 20 m for iron meteorites) to pass through the atmosphere without a significant deceleration and disruption, thereby reaching the Earth’s surface at a velocity of > 11 km/s (hypervelocity). Hypervelocity impact starts to form as soon as an extraterrestrial object strikes the ground surface at its original cosmic velocity. Hypervelocity impact produces high-pressure shock waves, that radiate into the target at a velocity of 5–8 km/s. Hypervelocity impact craters are so characterized to produce unique shock metamorphic effects (Section 1.3.).

It is widely accepted that the mechanism of the formation of hypervelocity impact craters can be outlined into three main stages (Melosh 1989; French 1998; Osinski 2008; Osinski and Pierazzo 2013): (i) contact and compression; (ii) excavation; (iii) modification. For very large impact craters a fourth stage could be considered: hydrothermal and chemical alteration.

(i) Contact and compression. The first stage of impact cratering begins when the projectile hits the target (Figs. 6a and 6b). Models suggest that the projectile penetrates a solid target about one or two times its diameter. The impact pressure

17 at the contact point depends on the impact velocity. It can be estimated using the planar impact approximation equations (Melosh 1989; Melosh 2013), when both the nature of the target and of the projectile are known. Typically, at the contact point shock pressure > 100 GPa. After the contact with the ground, the kinetic energy of the impact is transferred to the target and to the projectile in form of shock waves. The shock waves travel in two opposite directions, as illustrated by the train model (Fig. 6c). Contrary to the normal compressional waves, shock waves are also characterized by the propagation of a material flow. The velocity of the shock wave and of the material flow are indicated by the symbols Us and Up in Fig. 6. Us is higher than Up. When the shock front of the shock wave travelling into the projectile reaches its free upper surface (Fig. 6d), the shock wave is reflected back initially into the projectile and immediately later into the target as a rarefaction wave (Fig. 6f). The rarefaction wave travels faster than the shock wave, because it travels in compressed material.

Fig. 6. Schematic representations showing the first two stages of impact cratering (contact and compression and excavation; sketch on the right; after Osinski 2008) and the formation and propagation of shock and rarefaction waves (sketch on the left; after Langenhorst and Deutsch (2012)). Us and Up indicate the velocities of the shock wave and the velocity of the material flow, respectively. In Fig. 7 a schematic representation of the propagation of the shock wave into the target is shown. From this image it is possible to note that the volume of rocks shocked between 1 and 5 GPa is greater than the volume of rocks shocked between 5 and >50 GPa.

18

Fig. 7. Initial shock-wave pressures and excavation flow lines around the impact point during the contact and compression stage. Image after French 1998. The duration of the contact and compression stage is very short: 2 s for a 50-km- diameter projectile travelling object travelling at 25 km s-1, and less than 0.01 s for a 100-m-diameter object travelling at the same speed (French 1998). (ii) Excavation. The passage from the contact and compression stage to the excavation stage is a continuum. The passage of the rarefaction stage decompresses the rocks causing the complete (only, for large impact events) melting and/or vaporization of the projectile and the acceleration of excavated target rocks in the opposite direction of the ground (Fig. 6e). During this stage the crater opens up, thanks to a complex interaction among shock and rarefaction waves. This interaction produces an “excavation flow-field” and generates a so-called “transient cavity” (Fig. 6f). The projectile plays no roles in the crater opening. The excavated material is ejected ballistically, following the trajectories showed in Fig. 7, to form the ejecta blanket. Transient cavities with a maximum diameter of 2 km (for sedimentary rocks) and 4 km (for crystalline rocks) do not undergo further modification, more precisely they undergo minimum modification that stabilizes the structure (e.g., landslips). These craters are called simple impact craters (Fig. 8), in contrast to the complex impact craters (see point (iii) Modification). Simple impact craters are described

by the following geometrical parameters: D = diameter, dA = apparent depth, tbr = ,

and dT = true depth (dA + tbr). The boundary between simple and complex impact craters is not fixed, it can vary on Earth according to the lithology of the target material and on other solid planetary bodies because of the different gravitational

19 acceleration (e.g., on the Moon the transition between simple and complex crater is between 15 and 27 km).

Fig. 8. Schematic representation of the final simple impact crater. Modified after Osinski 2008. Although the excavation stage is longer than the contact and compression stage, it is still brief by geological standards. For example, a 1-km-diameter transient crater can be excavated in 6 s and a 200-km-diameter crater requires about 90 s (Melosh 1989). (iii) Modification. The effects of the modification stage are governed by the size of the transient cavity and by the lithology and the features of the target rocks. Shock waves do not play a further part in crater development. For these reasons, the modification stage essentially concerns the complex impact craters. This stage is characterized by the gravitational collapse of the initial transient cavity. It determines an uplift of the crater floor and so the formation of a central peak (Fig. 9). Besides the central peak, a very large impact crater can show a peak ring or a multi-ring morphology. The modification stage does not have a defined end: processes relative to the modification of very large impact structures can involve widespread disturbances in the Earth’s crust and merge into endogenous geological processes. In some cases the modification stage can last for years.

Fig. 9. Schematic representations showing the modification stage for complex impact craters. Images after Osinski 2008.

20 In about 70 impact structures from 1.8 km (Lonar, India) to 250 km (Sudbury, Canada) in diameter impact-generated hydrothermal systems have been recognized. Due to their diffusion, it seems highly probable that any hypervelocity impact can generate a hydrothermal system, as long as sufficient H2O is present. For this reason, post-impact hydrothermal activity could be considered as a fourth stage of the impact cratering process.

1.3. Shock metamorphism Rocks involved in the impact cratering process can vaporize, melt, shock metamorphose, or deform. All rocks affected by one or more hypervelocity impact(s) resulting from collision(s) of planetary bodies are called "impactites" (Stöffler and Grieve, 2007). The classification of impactites follows the recommendation formulated by the IUGS Subcommission on the Systematic of Metamorphic Rocks. It is based on the degree of shock metamorphism, the occurrence of melting, and the geological setting of impactites. Fig. 10 shows the impactite classification suggested by the IUGS Subcommission on the Systematic of Metamorphic Rocks.

Fig. 10. Impactite classification according the IUGS Subcommission on the Systematic of Metamorphic Rocks (modified after Stöffler and Grieve 2007). Shock metamorphism is a type of metamorphism of rocks and minerals caused by shock wave compression and decompression due to the hypervelocity impact of a solid body or due to the detonation of high-energy chemical or nuclear explosives. Contrary to endogenic metamorphism, shock metamorphism is characterized by pressure > 5 GPa and temperatures

21 up to 10000 °C (Fig. 11; Table 1). These enormous differences in the pressure and temperature ranges, and the very short duration of the process cause unique irreversible changes to rocks and minerals.

Fig. 11 Pressure and temperature ranges of endogenic and shock metamorphism. Image after French 1998.

Table 1. Differences between endogenic and shock metamorphism. Modified after French 1998.

Characteristic Endogenic metamorphism Shock metamorphism

Surface or near-surface regions of Geological setting Typically to depths of 10-50 km Earth’s crust 100-400 GPa near impact point; Pressures Typically <1-3 GPa 10-60 GPa in large volumes of surrounding rocks Up to 10000°C near impact point; Temperatures Generally ≤ 1000°C typically 500-3000°C in surrounding rocks

Strain rates 10-3-10-6 s-1 104-106 s-1

Time for completion of process From 105-107 yr Instantaneous (few seconds)

Rapid: abundant quenching and Slow; minerals closely approach Reaction time preservation of metastable equilibrium minerals and glasses

The identification of shock effects is one of the main criteria for identifying impact structures. Other criteria for identifying a new impact structures are geophysical investigation and siderophile elements and isotopic anomalies, however their treatment is beyond the scope of this Thesis.

22 According to recent literature, the shock metamorphism does not include the impact melting (French 1998; Osinski and Pierazzo 2013). For this reason, in this dissertation shock metamorphism and impact melting features are presented in two different sections.

The most significant shock effect at the meso- to macro-scale is the occurrence of multiple sets of conical striated fractures surfaces. These structures are known as shatter cones and are uniquely produced by hypervelocity impacts. Hence, their occurrence is a strong clue for the identification of impact structures. They are best developed in fine-grained massive lithologies (such as limestone), but they can also occur in more coarse-grained lithologies, such as granite, in which, however, they are usually more poorly developed.

Fig. 12. Shatter cones in a very fine-grained carbonate form Steinheim (Germany). Nesting of multiple cones are visible. Image after French and Koeberl 2010. Shatter cones usually occur in the central uplifts of complex impact structures and in some cases, isolated fragments/clasts of shatter cones have been found in impact breccias, within or outside the crater. Shatter cones are penetrative features; for this reason they are different to other striated features such as wind-abrasion structures. Although shatter cones are widely accepted as unequivocal proof of a meteorite impact crater, their formation mechanism is still debated (e.g., Sagy et al. 2002; Baratoux and Melosh 2003; Wieland et al. 2006). It is generally accepted that shatter cones form at relatively low shock pressures, generally between approximately 2 and 10 GPa, and exceptional up to 30 GPa. The shock pressure of shatter cones is estimated on the basis of the shock features recorded by minerals constituent the rock. Some occurrences of melt film among conical surfaces suggest that the formation of melt could play an important role in the formation mechanism of these structures (Gay 1978; Gibson and Spray 1998; Nicolaysen and Reimold 1999; Pittarello et al. 2011; Fazio et al. 2014).

23 Shock-induced changes in minerals could be classified in four typologies of effects: (i) deformation, e.g., formation of dislocations, planar microstructures, mechanical twins, kink bands, and mosaicism; (ii) phase transformation into high-pressure phases and diaplectic glass; (iii) decomposition into a solid residue and a gaseous phase; (iv) melting and vaporization of entire mineral (Langenhorst 2002).

Shock-induced changes in minerals have been reported for various rock-forming minerals (Fig. 13). However, quartz is the mineral that shows the widest variety of shock effects over a wide range of pressures in a consecutive manner, because of its three- dimensional linked, corner-shared SiO4-tetrahedra crystal structures. Moreover, quartz is the most common mineral on the Earth’s crust. The combination of these two factors determines that shock effects in quartz are widely studied and well pressure-calibrated (e.g., experiments on single quartz crystal by Langenhorst and Deutsch 1994 and Langenhorst 1994). In the next subsections the shock features occurring in quartz (1.3.1. Quartz) and other common minerals (1.3.2 Deformation in other minerals) from crystalline rocks are presented. The comparison between the shock metamorphic effects in crystalline (non-porous) rocks and sedimentary (porous) rocks is discussed in section 1.5. Shock effects in quartz-bearing rocks: crystalline vs. sedimentary target.

Fig. 13. Shock features in some of the most common rock-forming minerals. Source: http://www.lpi.usra.edu/exploration/training/resources/. Credit Kring D. A. (1989). 1.3.1. Quartz The initial deformations occurring in quartz are the reduction of refractivity and birefringence indices, and mosaicism. All these effects are not diagnostic features for the identification of new impact structures. A crystal showing mosaicism is recognizable because

24 of the occurrence of several sub-domains with slightly different optical axes. Mosaicism is produced by the distortion of the crystal lattice into small domains that are rotated by low angles against each other. Refractivity and birefringence indices decrease increasing the shock compression, drifting towards the amorphous values. The most significant shock effect on quartz is the occurrence of planar microstructures. Planar microstructures are divided into planar fractures (PFs) and planar deformation features (PDFs). Contrary to irregular fractures (ubiquitous in shocked quartz), planar microstructures are crystallographically controlled: PFs and PDFs are oriented parallel to rational crystallographic planes. Each plane is indicated by its crystallographic index, following the four-digit notation of Miller-Bravais {hkil}. Planar fractures are open fissures parallel to the rational crystallographic plane (0001)

and {} 1011 . Planar fractures develop in sets of fractures with spacing of about 20 µm. Each fracture is generally wider than 3 µm. Planar fractures commonly control and/or limit the distribution of adjacent PDF sets; therefore PFs are formed earlier than PDFs. Planar fractures € are supposed to form between 5 and 8 GPa. Contrary to PDFs, PFs are not considered as unequivocal evidence of shock, although they are rare and differently oriented in porous sedimentary rocks (see section 1.5).

Planar deformation features are considered as one of the best criteria for the identification of new impact structures. Planar deformation features develop in sets parallel to the rational crystallographic plane as PFs, however they are generally thinner and narrower than PFs. Planar deformation features are less than 2 µm thick and between 2 and 10 µm spaced (Fig. 14). Up to ten differently oriented sets of PDFs per crystal have been counted. They form within a wide range of pressures: from 5-10 GPa to 35 GPa (Fig. 13). Planar

deformation features are differently oriented according to their pressure of formation. {} 1013

and {} 1011 planes are the two most common orientations between ~ 10 and ~ 20 GPa. Planes

oriented parallel to {} 1013 and, subsequently, {} 1012 may form between ~ 20€ and ~ 25 GPa.

€ Over ~ 25 GPa, {} 1012 planes are more common than {} 1013 ones. The orientation of planes could be done€ by TEM or by universal€ stage analyses (Langenhorst 2002; Ferrière et al. 2009; Huber et al. 2011). € €

25

Fig. 14. Planar deformation features (PDFs) in a quartz grain from Kamil crater sandstone. In altered and geologically old impact structures, the original amorphous material in PDF planes is recrystallized back to quartz, and arrays of small (1-2 µm) fluid inclusions develop along the original planes. These PDFs are called decorated PDFs (Stöffler and Langenhorst 1994). The most accepted model of formation of PDFs was proposed by Goltrant et al. (1992). This model is the result of TEM observations and of the estimated elastic properties of quartz. They observed that the magnitude of shear moduli changes as a function of pressure causing a separation of uncompressed and compressed regions in the crystal. Due to the different lattice dimensions, both crystal regions are incompatible with respect to their lattice parameters. The misfit is compensated by the formation of amorphous lamellae, namely PDFs, behind the shock front by solid-state amorphization. This mechanism operates up to 25

GPa (Fig. 15a). Above this pressure and up to 35 GPa, the shock temperature Th exceeds the melting temperature of quartz Tm, causing the formation of melt and the subsequent lengthening of PDFs, whose boundaries become wavy (Fig. 15b). Because the post-shock temperature Tr is below that of the Tm, diaplectic glass forms (glass formed by solid-state transformation). Between 35 and 50 GPa the entire grain could be transformed into diaplectic glass (Fig. 15c). Above 50 GPa, also the Tr is higher than the Tm, determining the total melting of the quartz grains, or rather the formation of lechatelierite (pure SiO2 glass; Fig. 15d). This model was proposed by Langenhorst (1994), and new evidence from Kamil impactites seems to confirm it (see Chapter 3 of this Thesis).

26

Fig. 15 Langenhorst model for the formation of PDFs, diaplectic glass and leshatelierite. Th= shock temperature, Tr= post-shock temperature, and Tm= melting temperature of quartz. Image from Langenhorst 1994. During the passage of shock waves, quartz can transform into high-pressure phases that are coesite and stishovite. Both coesite and stishovite can also form during endogenic processes, but stishovite is generally very rare, hence their combined occurrence can be considered as a univocal feature of shock metamorphism. The formation pressures of these polymorphs during endogenic processes are about 2 GPa for coesite and around 7-8 GPa for stishovite (Heaney et al. 1994). Pressures of formation of 2-8 GPa are too low for shock metamorphism (few seconds) and are possible because rocks suffer these pressures for a very long time (million years) under static equilibrium conditions. In non-porous crystalline rocks coesite forms between 30 and 60 GPa, instead stishovite between 12 and 45 GPa. The formation pressures of these phases are very different if the rock is non-porous or porous (see Section 1.5). Moreover, the crystallization order is switched compared to the endogenic environment: stishovite generally forms at a lower pressure than coesite. Coesite is thought to form during the passage of the rarefaction wave, namely during the decompression stage (Stöffler and Langenhorst 1994). Both polymorphs were found within diaplectic glass, along

27 grain boundaries or in association with PDFs. High-pressure phases can be detected by X-ray powder diffraction, µRaman or TEM. Among the phase transformations, diaplectic glass can also be considered. Diaplectic glass is a glass that formed without melting but by solid-state transformation. Diaplectic glass typically forms from framework minerals, such as quartz and feldspar (diaplectic glass from feldspar is known as maskelynite). Diaplectic glass is characterized to be amorphous, to have a very low refractive index, to preserve the pre-shock shape, morphology and texture of the parent mineral, and being devoid of flow structures and vesicles. The minimum shock formation pressure of diaplectic glass is about 35 GPa for non-porous crystalline rocks. If rocks suffer a thermal annealing at temperatures above 1200°C, diaplectic glass is not preserved. It recrystallizes forming ballen α-cristobalite or ballen α-quartz (Ferrière and Osinski 2013 and reference therein). Ballen α-cristobalite or ballen α-quartz, together called ballen silica, and toasted quartz are not primary shock features. They form because of thermal annealing. The identification of these post-shock thermal features is important for recognizing large, old and eroded impact structures. Ballen silica occurs as more or less spheroidal or rarely ovoid bodies of around a few micrometers up to 200 µm. These bodies can intersect or penetrate each other or abut each other (Ferrière and Osinski, 2013 and references therein). Toasted quartz is characterized to be orange-brown to grayish-reddish brown in color. The formation mechanism of toasted quartz is not clear. It probably results from hydrothermal or other post- shock modification or from the exsolution of water from glass, primarily along PDFs, during heat-driven recrystallization (Ferrière and Osinski, 2013 and reference therein).

1.3.2. Deformation in other minerals Feldspar crystals, being tectosilicate as quartz, show a wide range of shock metamorphic effects. Similarly to quartz, they develop fracturing, PFs, PDFs, and diaplectic glass (called maskelynite). Planar deformation features have also been observed in olivine, zircon and tourmaline, pyroxene, amphibole, sillimanite, garnet, and apatite (Ferrière and Osinski, 2013 and reference therein). Shocked mica crystals commonly show kink bands, these features also develop by tectonic deformation, so they are not diagnostic criteria for establishing an impact origin of a structure.

Zircon under high shock pressure and temperature decomposes into baddeleyite

(ZrO2) and a SiO2 phase. Zircon decomposition occurs at temperatures higher than 1750°C (French 1998). Decomposed zircon shows typical granular-textures. Zircon can also transform

28 into a high-pressure polymorph called reidite (about 20 GPa; Ferrière and Osinski, 2013 and reference therein).

Graphite transforms into diamond upon shock metamorphism. Impact diamonds are usually polycrystalline and rich in crystallographic defects. They commonly preserve some features of the precursor graphite. The origin of impact diamond is still debated. One of the widely accepted hypotheses, based on graphite-diamond textural relationship within shocked rocks from Ries Crater (Germany), suggests that diamonds are formed from graphite by shock-induced solid-state phase transformation during shock compression (El Goresy et al. 2001). Diamonds from Ries are supposed to have formed around 30-40 GPa. Shock formation pressures can vary a lot also for diamonds according to the porosity of the rock and the nature of graphite (degree of crystallinity; Ferrière and Osinski, 2013 and reference therein). Besides diamond, graphite can transform into lonsdaleite and other unknown C-phase(s). Lonsdaleite and the other C-phase(s) may represent intermediate phases before diamond formation. One of the best techniques for studying carbon phases is µRaman spectroscopy.

1.3.3. Selective and localized melting Before whole-rock melting, melting starts in correspondence of rock heterogeneities as pores or boundaries between different mineral grains (interstitial impact melt glass). Selective melting could also occur in single mineral grains (mineral glass).

Interstitial impact melt glasses have compositions that are a mixture of adjacent minerals. In meteorites these interstitial impact melt glasses are also known as melt pockets. Mineral glasses have the same composition as the pre-existing minerals. Mineral glass is different from diaplectic glass, because the latter forms by solid-state transformation and retains the shape of the mineral grain.

Localized melt can also be found in the form of melt veins and/or melt films. Melt veins are irregular veins of quenched melt produced by shock-induced localized melting in moderately to strongly shocked rocks (Stöffler and Grieve 2007).

One particular type of melt veins is represented by pseudotachylite veins. Pseudotachylite veins are specific melt veins produced by frictional melting (pseudotachylite is a genetic name). Frictional melting is controlled by the mechanical properties of a rock’s constituent minerals: mineral with the lowest fracture toughness and breakdown temperature are preferentially involved into frictional melting (Spray 2010; Fig. 16).

29

Fig. 16. Melting or breakdown temperature TMB versus fractures toughness KIC for selected minerals. Image after Spray 2010. Pseudotachylite veins vary from m-scale to µm-scale and they are characterized by a clear margin with the host rock, injection veins, fragment inclusions, and high-pressure mineral inclusions. Some authors classify pseudotachylites into two types. S-types (shock-type) pseudotachylite veins are typically thin (< 2 mm) and contain relict mineral grains and high- pressure polymorphs. They usually occur irregularly distributed within the innermost shock zone of the impact structures. E-types (endogenic-type) can be up to hundreds of meters wide, and occurs only in the periphery of the impact structures (Spray 2010; Fig. 17). Pseudotachylite veins can form networks of veins giving rise to “pseudotachylite breccias”.

30

Fig. 17. Vredefort (South Africa) is the type locality of pseudotachylite veins. The term pseudotachylite was introduced in literature in 1916 by Shand to identify and describe these veins in Vredefort impact structure. Image after Spray 2010. Other forms of localized melting in the form of veins and/or films were described at the surfaces of shatter cones from a few localities: Sudbury (Ontario, Canada) and Vredefort (South Africa; e.g., Gay 1978; Gibson and Spray 1998; Nicolaysen and Reimold 1999), Vista Alegre (Brazil; Pittarello et al. 2011), Kamil (Egypt; Fazio et al. 2014, Chapter 2 of this Thesis) and, possibly, from Santa Fe (New Mexico, USA; Fackelman et al. 2008).

1.4. Impact melting Whole-rock melting is one of the most significant pieces of evidence of hypervelocity impact events (documented at approximately half of the known terrestrial impact structures). Impact melting is different from thermal melting (such as that produced by endogenic magmatic processes), because impact melting is a function of shock pressure, porosity and the compressibility of the target rock lithology and their constituent minerals. Impact melting occurs upon decompression from high shock pressure and temperature.

Impact melting can occur in a variety of settings: i) km-scale sheets and/or isolated bodies within the crater (impact melt rocks with variable amounts of clastic debris of different degree of shock metamorphism); ii) m- to cm-scale irregular and aerodynamically sculptured glassy particles, either within impact breccias inside the crater (e.g., suevite) or in nearby ejecta deposit around the crater in the form of impact melt lapilli and bombs. iii) tens of m- to cm-scale injection dikes and sills in the crater floor and walls; iv) distal cm- to µm-scale

31 discrete particles (spherules, tektites, and microtektites) distributed regionally to globally (French 1998, Stöffler and Grieve 2007, and Osinski et al. 2013).

The main settings in which impact melt-bearing materials occur in a complex impact structure are shown in Fig. 18. The occurrence and the properties of impact melts provide a lot of information about the crater formation process.

Fig. 18. Main settings in which impact melt-bearing materials occur in a complex impact structure. Image after Osinski et al. 2013. Besides the geological setting, impact melt-bearing rocks can be classified on the basis of the textural features of the groundmass or matrix and clast content. In Fig. 19 shows the classification scheme by Osinski et al. 2013. Clast-rich impact melts are also known as impact melt breccias. The prefix particulate is suggested if there is evidence that the groundmass remained molten during and after deposition.

32

Fig. 19 Classification scheme of impact melt-bearing impactites. Image after Osinski et al. 2013. The impact melt process is essentially the same in sedimentary and crystalline rocks, although shock pressures necessary to melt those rocks are different (see next section 1.5). The mineralogical assemblage and the porosity of sedimentary rocks (e.g., the occurrence of phyllosilicate; Fig. 16) could facilitate the melting of sedimentary rocks respect to crystalline rocks. The consequence of these differences is that the products of melting in sedimentary and crystalline targets appear different. In crystalline targets, impact melts products have igneous structures. In large craters, they occur as large coherent melt bodies, with relatively homogenous compositions. In some cases, impact melt sheets can extend laterally many kilometers and can be several hundred meters thick. In sedimentary targets, impact melt products generally occur as isolated clasts within various impactite types and/or as clast-rich impact melt rocks. Impact melt products of sedimentary targets made of different lithologies result in unmixed and heterogeneous melts.

Most impact melt rocks have been produced as a consequence of large meteoritic impacts (projectile diameter larger than few meters). These rocks reflect the chemical composition of the target rocks; the contribution of the projectile is lower than 1 vol.%, because during large meteoritic impact the projectile vaporizes. For this reason, the nature of the projectile can possibly be detected only through the analysis of siderophile elements and of isotopic composition of osmium and chromium (e.g., for review Koeberl et al. 2012). During small impact events (final crater diameter < 1.5 km) impact melt rocks result both by the melting of the target and of the projectile; indeed the projectile survives the impact and partially melted impactor debris can be found in the proximity of the crater (e.g., Wabar (116

33 m in diameter), Kamil (45 m in diameter), Henbury (157 m in diameter), Aouelloul (390 m), Barringer (1.2 km in diameter); Table 7 in Fazio et al. 2014 – Chapter 2 of this Thesis). Nineteen out to 27 small impact craters are formed by the impact of iron or stony-iron projectiles. The great geochemical difference between target rocks and iron projectiles allows us to study how the projectile interacts with the target, nevertheless this mixing process is not yet very well understood. In recent years new steps forward in this field have been made thanks to new experimental and numerical simulations (e.g., MEMIN experiments; Ebert et al. 2013 and 2014) and to new evidences from Wabar impactites (Hamann et al. 2013) and Kamil impactites (Chapter 3 of this Thesis).

1.5. Shock effects in quartz-bearing rocks: crystalline vs. sedimentary targets Shock pressure calibration for single quartz crystal and for quartz-bearing crystalline was proposed for the first time by Stöffler (1971). This experimental calibration has not changed very much over the years, and it is still widely accepted (Table 2). Although quartz is the main component of several sedimentary rocks, shock effects in quartz-bearing sedimentary rocks are poorly calibrated for two main reasons: i) sedimentary porous rocks can display a very wide range in their highly mineralogical assemblage, porosity, type and amount of matrix, grain size, water contents, and fabric that make it difficult to compare rocks with a different lithology; ii) in naturally shocked sedimentary porous rocks, shock deformation effects typically attributable to very different shock pressures can coexist in crystalline rocks (Kieffer 1971; Kieffer et al. 1976; Grieve et al. 1996).

Table 2. Progressive shock metamorphism of quartz-feldspathic crystalline rocks. Table after Stöffler and Grieve 2007, modified after Stöffler 1971.

Modified after Stöffler 1971 Shock Pressure Temperature stage Shock produced phenomena (GPa) (°C)

0 Fractured minerals

Ia ~ 10 ~ 100 Quartz and feldspar with planar deformation features Quartz and feldspar with planar deformation features Ib ~ 20 ~ 170 and reduced refractive index stishovite and minor coesite Diaplectic SiO glass and feldspar glass; coesite and II ~ 35 ~ 300 2 traces of stishovite; cordierite glass Normal feldspar glass (vesiculated) and diaplectic SiO III ~ 45 ~ 900 2 glass; coesite; cordierite glass Rock glasses or crystallized melt rocks (quenched from VI ~ 60 ~ 1500 liquids) V ~ 80-100 > 2500 Rock glasses (melts condensed from silicate vapor)

34 Recently, new experiments and numerical simulations have been carried out to better understand how sedimentary porous rocks behave under the passage of a shock wave (e.g., Wünnemann et al. 2006; Schade and Wünnemann 2007; Wünnemann et al. 2008; Güldemeister et al. 2013; Kowitz et al. 2013). They demonstrate that shock pressure and temperature are enhanced in correspondence of pores and more in general in correspondence of heterogeneities of the rocks (i.e., grain boundary, pre-existing fractures, weakness planes). The shock pressure peak is recorded close to the pore, calculations by Güldemeister et al. (2013) show that here the shock pressure can be two to four times higher than elsewhere in the same sample. Hence, in rocks shocked at the nominal pressure of few GPa (2.5-17.5 GPa) it is possible to find shock features consistent with shock pressures in excess of 30 GPa (Kowitz et al. 2013) localized in correspondence of pores and eventually in correspondence of other heterogeneities.

The first shock pressure calibration for porous quartz-rich sandstone was proposed by Kieffer et al. (1976) for the Coconino Sandstone Formation. This classification is based on observations on progressively shocked Coconino Sandstone rocks from Barringer Crater. This classification was widely accepted, but it was never calibrated. Recently, Kowitz et al. (2013) calibrated and improved this classification scheme as shown in Table 3.

Table 3. Progressive shock metamorphism of porous sandstone (modified after Kieffer et al. 1976 and Kowitz et al. 2013).

Modified after Kieffer et al. 1976 Modified after Kowitz et al. 2013 Shock stage Pressure Temperature Nominal shock Glass and high- Shock produced phenomena (GPa) (°C) pressure (GPa) pressure phases

0 ≤ 0.2-0-9 25 Undeformed sandstone

Deformed sandstone with remnant 1a 0.2-0.9 – ~3 2.5 0 porosity ~250 5 0.03 vol.% Deformed sandstone compressed to 1b ~3 – 5.5 7.5 0.4 vol.% zero porosity 10 2.2 vol.% Dense sandstone with 2-5 vol.% 2 ~5.5 – ~13 ~350 coesite, 3-10 vol.% glass, and 80-95 12.5 7.4 vol.% vol.% quartz Dense sandstone with 18-32 vol.% 3 ~13 – ~30 ~950 coesite, traces of stishovite, 0-20 vol.% 15 24.5 vol.% glass, 15-45 vol.% quartz Dense sandstone with 10-30 vol.% 4 >~30 >1000 coesite, 20-75 vol.% glass, and 15-45 17.5 80.6 vol.% vol.% quartz Vesicular (pumiceous) rock with 0-5 5 vol.% coesite, 80-100 vol.% glass (lechatelierite), and 0-15 vol.% quartz

35 The main differences between the classification for quartz-bearing crystalline rocks and for quartz-bearing sedimentary rocks relate to the absence of PDFs in quartz in sedimentary rocks and to the appearance of high-pressure silica polymorphs and glass at pressure > 5 GPa. More precisely, rare PDFs could be found in quartz of sedimentary rocks, for example in Coconino Sandstone from Barringer Crater, PDFs constitute about 5 vol.% (Robertson 1980). Moreover, when PDFs occur in sedimentary rocks, their poles have orientations dominated by high angles to c-axis (> 45°), like {} 1122 and {} 1011 (e.g., in Oasis (Libya), Aorounga (Chad), Avak (Alaska, USA), Tookoonooka (Australia) impact structures; Robertson 1980; Grieve and Therriault 1995; Grieve et al. 1996). € € The lack of PDFs and their anomalous orientations in sedimentary targets have been attributed to the different impedance of the porous rocks and crystalline rocks: at shock pressure where the low angle to c-axis PDFs form (e.g. (0001) and {} 1013 ), strain was taken up to close voids without PDFs formation (Robertson 1980). Planar deformation features can develop only when the porosity is zero. When sandstone rocks have a low porosity (e.g., € silicified sandstone) they behave as crystalline rocks, hence PDFs are common and their poles are oriented at low angles to c-axis (Milton et al. 1972; Fazio et al. 2014).

Coesite and stishovite can form starting from 5.5 GPa. High-pressure polymorphs are typically localized within so-called “symplectic regions”. These areas are roundish and opaque. They are made of a microscopic to sub-microscopic intergrowth of quartz, diaplectic glass, high-pressure polymorphs, and lechatelierite. The formation of these regions is the result of complex interactions during the collapse of the pore spaces (Kieffer et al. 1976). Diaplectic glass in porous sedimentary rocks also forms starting from 5.5 GPa.

In non-porous crystalline rocks, individual minerals start to melt around 50 GPa. Around 60 GPa the melting of the whole rocks can be considered concluded (Stöffler and Langenhorst 1994). In porous sedimentary rocks, individual quartz grains start to melt at around 20 GPa, the whole rock melting is concluded around 30-35 GPa (Kieffer et al. 1976). In order to understand how the shock pressure necessary to melt a rock changes with the rock porosity, Wünnemann et al. (2008) carried out numerical simulations for quartz and calcite (Fig. 20). As expected, the critical pressure for melting decreases with increasing porosity of the target rock. Calcite (dashed line) has a linear curve line, whereas quartz (solid line) shows a kink in the curve due to the solid-state phase transition. Gray lines represent the extrapolation of simulation results at porosity > 50% (Wünnemann et al. 2008).

36

Fig. 20 Critical pressure for melting vs. porosity for quartz-rich and calcite-rich rocks. Image after Wünnemann et al. (2008). 1.6. References http://www.lpi.usra.edu/exploration/training/resources/ http://www.passc.net/EarthImpactDatabase/index.html Baratoux D. and Melosh H. J. 2003. The formation of shatter cones by shock wave interference during impacting. Earth and Planetary Science Letters 216:43-54. Bland P. A. and Artemieva N. A. 2006 The rate of small impact on Earth. Meteoritics & Planetary Science 4:607–631. Ebert M., Hecht L., Deutsch A. and Kenkmann T. 2013. Chemical modification of projectile residues and target material in a MEMIN cratering experiment. Meteoritics & Planetary Science 48, 134–149. Ebert M., Hecht L., Deutsch A., Kenkmann T., Wirth R. and Berndt J. 2014. Geochemical processes between steel projectiles and silica-rich targets in hypervelocity impact experiments. Geochimica et Cosmochimica Acta 133, 257–279. El Goresy A., Gillet P., Chen M., Künstler F, Graup G., Stähle V. 2001. In situ discovery of shock-induced graphite-diamond phase transition in gneisses from the Ries Crater, Germany. American Mineralogy 86:611-621. Fackelman S. P., Morrow J. R., Koeberl C., and McElvain T. H. 2008. Shatter cone and microscopic shock-alteration evidence for a post-Paleoproterozoic terrestrial impact structure near Santa Fe, New Mexico, USA. Earth and Planetary Science Letters 270:290-299. Fazio A., Folco L., D’Orazio M., Cordier C., and Frezzotti M. L. 2014. Shock metamorphism and impact melting in small impact craters on Earth: Evidence from Kamil Crater,

37 Egypt. Meteoritics & Planetary Science DOI: 10.1111/maps.12385 (Chapter 2 of this Thesis). Ferrière L., Morrow J. R., Amgaa T., and Koeberl C. 2009. Systematic study of universal- stage measurements of planar deformation features in shocked quartz: Implications for statistical significance and representation of results. Meteoritics & Planetary Science 44:925-940. French B. M. 1998. Traces of catastrophe: A handbook of shock- Metamorphic effects in terrestrial meteorite impact structures. Houston, Texas, USA. LPI Contribution No. 954, Lunar and Planetary Institute. 120 p. French B. M. and Koeberl C. 2010. The convincing identification of terrestrial meteorite impact structures: What works, what doesn’t work, and why. Earth-Science Reviews 98:123-170. Gay N. C. 1978. The composition of spherules and other features on shatter cone surfaces from the Vredefort structure, South Africa. Earth and Planetary Science Letters 41:3:372-380. Gibson H. M. and Spray J. G. 1998. Shock-induced melting and vaporization of shatter cone surfaces: Evidence from the Sudbury impact structure. Meteoritics & Planetary Science 33:2:329-336. Goltrant O., Leroux H., Doukham J. C., and Cordier P. 1992. Formation mechanism of planar deformation features in naturally shocked quartz. Physics of the Earth and Planetary Interiors 74:219-240. Grieve R. A. F. and Therriault A. M. 1995. Planar deformation features in quartz: target effects (abstract). 26th Lunar and Planetary Science Conference 515-516. Grieve R. A. F., Langenhorst F., and Stöffler D. 1996. Shock metamorphism of quartz in nature and experiment: II. Significance in geoscience. Meteoritics & Planetary Science 31:6-35. Güldemeister N., Wünnemann K., Durr N., and Hiermaier S. 2013. Propagation of impact- induced shock waves in porous sandstone using mesoscale modelling. Meteoritics & Planetary Science 48:1:115-133. Hamann C., Hecht L., Ebert M. and Wirth R. 2013. Chemical projectile–target interaction and liquid immiscibility in impact glass from the Wabar craters, Saudi Arabia. Geochiica et Cosmochimica Acta 121, 291–310.

38 Heaney P. J., Prewitt C. T., and Gibbs G. V. 1994. Silica: Physical behavior, geochemistry and materials applications. Reviews in Mineralogy vol. 29. Mineralogical Society of America, Washington, DC. Huber M. S., Ferrière L., Losiak A., and Koeberl C. 2011. ANIE: A mathematical algorithm for automated indexing of planar deformation features in quartz grains. Meteoritics & Planetary Science 46:9:1418-1424. Kieffer S. W. 1971. Shock metamorphism of the Coconino Sandstone at Meteor Crater, Arizona. Journal of Geophysical Research 76:5449-5473. Kieffer S. W., Phakey P.P., and Christie J. M. 1976. Shock processes in porous quartzite: Transmission electron microscope observations and theory. Contribution to Mineralogy and Petrology 59:41-93. Koeberl C., Claeys P., Hecht L. and McDonald I. 2012. Geochemistry of impactites. Elements 8:37-42. Kowitz A., Güldemeister N., Reimold W. U., Schmitt R. T. and Wünnemann K. 2013.

Diaplectic quartz glass and SiO2 melt experimentally generated at only 5 GPa shock pressure in porous sandstone: Laboratory observations and meso-scale numerical modeling. Earth and Planetary Science Letters 384:17-26. Langenhorst F. 2002. Shock metamorphism of some minerals: Basic introduction and microstuctural observations. Bulletin of the Czech Geological Survey 77:265-282. Langenhorst F. 1994. Shock experiments on pre-heated α- and β-quartz: II. X-ray and TEM investigations. Earth and Planetary Science Letters 128:683-698. Langenhorst F. and Deutsch A. 2012. Shock metamorphism of minerals. Elements 8:31-36. Langenhorst F. and Deutsch A. 1994. Shock experiments on pre-heated α- and β-quartz: I. Optical and density data. Earth and Planetary Science Letters 125:407-420. Melosh 2013. The contact and compression stage of impact cratering. In Impact Cratering (eds. G. R. Osinski and E. Pierazzo). Wiley-Blackwell, Chichester, West Sussex, U. K., pp. 32–42. Melosh 1989. Impact Cratering: A Geologic Process. Oxford University Press, New York. 245 pp. Milton D. J., Barlow, B. C., Brett R., Brown A. R., Glikson A. Y., Manwaring E. A., Moss F. J., Sedmik E. C. E., Van Son J. and Young G. A. 1972. Gosses Bluff impact structure, Australia. Science 175:1199-1207. Nicolaysen L. O. and Reimold W. U. 1999. Vredefort shatter cones revisited. Journal of Geophysical Research 104:B3:4911-4930.

39 Osinski G. R. 2008. Meteorite impact structures: the good and the bad. Geology Today 24:13- 19. Osinski G. R. and Pierazzo E. 2013. Impact cratering: Processes and Products. In Impact Cratering (eds. G. R. Osinski and E. Pierazzo). Wiley-Blackwell, Chichester, West Sussex, U. K., pp. 1–20. Pittarello L., Koeberl C. and Crósta A. P. 2011. Melt on the striated surface of shatter cones in basalt (abstract). 74th Annual Meteoritical Society Meeting #5080. Robertson P. B. 1980. Anomalous development of planar deformation features in shocked quartz of porous lithologies (abstract). 11th Lunar and Planetary Science Conference 938-940. Sagy A., Reches Z., and Fineberg J. 2002. Dynamic fracture by large extraterrestrial impacts as the origin of shatter cones. Nature 418:310-313. Schade S. and Wünnemann K. 2007. Numerical modelling of pore space collapse due to shock wave compression (abstract #2029). 38th Lunar and Planetary Science Conference. Spray J. G. 2010. Frictional melting processes in planetary materials: From hypervelocity impact to earthquakes. Annual Review of Earth and Planetary Science 38:221-254. Stöffler D. 1971. Progressive metamorphism and classification of shocked and brecciated crystalline rocks at impact craters. Journal of Geophysical Research 76:5541-5551. Stöffler D. and Grieve R. A. F. 2007. Impactites, Chapter 2.11 in Fettes, D. and Desmons, J. (eds.) Metamorphic Rocks: A Classification and Glossary of Terms, Recommendations of the International Union of Geological Sciences, Cambridge University Press, Cambridge, UK, pp. 82–92, 111–125, and 126–242. Wieland F., Reimold W.U. and Gibson, R.L. 2006. New observations on shatter cones in the Vredefort impact structure, South Africa, and evaluation of current hypotheses for shatter cone formation. Meteoritics & Planetary Science 41:1737–1759. Wünnemann K., Collins G. S., and Melosh H. J. 2006. A strain-based porosity model for use in hydrocode simulations of impacts and implications for transient crater growth in porous targets. Icarus 180:514-527. Wünnemann K., Collins G. S., and Osinski G. R. 2008. Numerical modelling of impact melt production in porous rocks. Earth and Planetary Science Letters 269:530-539.

40 2. SHOCK METAMORPHISM AND IMPACT MELTING IN SMALL IMPACT CRATERS ON EARTH: EVIDENCE FROM KAMIL CRATER, EGYPT

1* 1 1 2 Agnese FAZIO , Luigi FOLCO , Massimo D’ORAZIO , Maria Luce FREZZOTTI , and Carole 3,4 CORDIER 1Dipartimento di Scienze della Terra, Università di Pisa, Via S. Maria 53, 56126 Pisa, Italy 2Dipartimento di Scienze dell'ambiente e del Territorio e di Scienze della Terra (DISAT), Sezione di Scienze Geologiche Geotecnologie, Piazza della Scienza 4, 20126 Milano, Italy 3Université de Grenoble Alpes, ISTerre, BP 53, 38041 Grenoble CEDEX 9, France 4CNRS, ISTerre, BP 53, 38041 Grenoble CEDEX 9, France

In Meteoritics & Planetary Science 2014, vol. 49, issue 12, pp. 2175-2200.

Abstract – Kamil is a 45-m-diameter impact crater identified in 2008 in southern Egypt. It was generated by the hypervelocity impact of the Gebel Kamil iron meteorite on a sedimentary target, namely layered sandstones with subhorizontal bedding. We have carried out a petrographic study of samples from the crater wall and ejecta deposits collected during our first geophysical campaign (February 2010) in order to investigate shock effects recorded in these rocks. Ejecta samples reveal a wide range of shock features common in quartz-rich target rocks. They have been divided into two categories, as a function of their abundance at thin section scale: 1- pervasive shock features (the most abundant), including fracturing, planar deformation features, and impact melt lapilli and bombs, and 2- localized shock features (the least abundant) including high-pressure phases and localized impact melting in the form of intergranular melt, melt veins and melt films in shatter cones. In particular, Kamil Crater is the smallest impact crater where shatter cones, coesite, stishovite, diamond, and melt veins have been reported. Based on experimental calibrations reported in the literature, pervasive shock features suggest that the maximum shock pressure was between 30 and 60 GPa. Using the planar impact approximation we calculate a vertical component of the impact velocity of at least of 3.5 km s-1. The wide range of shock features and their freshness make Kamil a natural laboratory for studying impact cratering and shock deformation processes in small impact structures.

INTRODUCTION

Shock metamorphism and impact melting in terrestrial rocks during small-scale meteorite impacts is still not fully known. This is mainly due to the rarity of very small impact craters on Earth (only 17 terrestrial craters out of 184 have diameters < 300 m; Earth

41 Impact Database, accessed on August 2014), and, more importantly, to the generally low degree of preservation of very small craters. One exception is Kamil Crater (southern Egypt). This 45-m-diameter crater shows all its original features in an almost pristine state of preservation due to its very young age (most likely < 5000 yr), the favorable climatic conditions of the Sahara Desert, and the mechanical strength of the local target rocks. After its discovery in 2008, Kamil has been explored in 2010 by our first Italian–Egyptian geophysical campaign (Folco et al. 2010, 2011; D'Orazio et al. 2011; Urbini et al. 2012). During this campaign a series of target rock samples representing the crater wall and ejecta were collected with the aim of investigating the shock metamorphism and impact melting induced in the local rocks by the impact of the small (5-10 t) Gebel Kamil Ni-rich ataxite (D'Orazio et al. 2011). In this work we present a detailed description of the shock features found within these samples. We will show that the target rocks from Kamil preserve an astonishingly wide record of shock conditions, and we will use petrographic observations on these shock features to constrain peak shock pressure and impact velocity of the iron meteorite projectile. Because the target rocks of Kamil area are dominantly siliciclastic sedimentary rocks, our data will be useful for comparing the better known shock effects on crystalline rocks (e.g., Stöffler and Langenhorst 1994) with those affecting sedimentary rocks, which are characterized by highly variable porosity, type and amount of matrix, grain size, water content, fabric and mineralogical assemblage (e.g., Kowitz et al. 2013a,b).

GEOLOGICAL SETTING

A detailed account of the geology and geophysics of Kamil is given in Urbini et al. (2012). The Kamil impact structure is situated in a rocky desert area in the East Uweinat district of southern Egypt (Fig. 1). It occurs in a simple geological context: flat, rocky desert surface, and target rocks comprising subhorizontally layered sandstones probably belonging to the Cretaceous Gilf Kebir Formation. The Gilf Kebir Formation unconformably overlies the Precambrian Basement, with a peneplaned surface marking a considerable stratigraphic gap. No rocks from the basement were observed in the crater and ejecta. The impact did not involve the crystalline basement.

Kamil is a simple, approximately circular crater, with a diameter of 45 ± 2 m (Figs. 1 and 2a), a depth of ~ 10 m, an upraised rim about 3 ± 0.7 m above the pre-impact surface as modeled by Urbini et al. (2012). In situ, uplifted sandstone strata are observed along the upper part of the crater walls (e.g., Fig. 2b). Tear faults separate sectors of the crater wall with distinct inclinations of the upturned strata. The lower part of the crater walls are covered by

42 slump debris (Fig. 2c). The bottom of the crater is occupied by a breccia lens with a maximum thickness of ~ 6 m. A wind-blown sand deposit covers part of the northern crater wall and crater floor. The overall stratigraphic sequence consists of pale (top) and reddish brown (bottom; Fig. 2b), coarse to very coarse, gritty, and ferruginous quartzarenite. The topmost layers are whitish, kaolinitic, and fine-grained.

Fig. 1. Enhanced true color QuickBird satellite scene (22 October 2005; courtesy of Telespazio) of the Kamil area (Egypt; see inset) showing the location of the studied samples.

43

Fig. 2. Field photos of Kamil wall rocks. a) Photomosaic of the interior of the crater viewed from southwest. The location of sample L13 is shown. The white rectangle indicates the outcrop featured in (b). b) Upturned rim strata exposed on the north wall of the crater. Samples M26 and M27 were collected from the top pale layer and the underlying reddish layer, respectively. The thickness of the top pale layer is about 70 cm. c) Slump debris covering the lower part of the southern crater wall. The location of sample L14 is shown. The Kamil structure is characterized by a radial pattern of bright ejecta (Fig. 1). Three nearly straight major ejecta rays trend to the north (approximately 355°), southeast (approximately 130°), and southwest (approximately 210°). The ray trending to the southeast (Fig. 3a) is also the longest (~ 300 m from the crater rim) and the largest one in terms of volume of ejected material, and gives rise to a positive topography close to the crater rim. A nearly continuous ejecta blanket extends for ~ 50 m from the crater rim. Overall, the bulk of the ejecta material is preferentially concentrated between the two main ejecta rays trending to the north and southwest. Within ~ 50 m from the crater rim, ejecta consist of pale and reddish target rocks (Figs. 3b and 3d) plus impact melt lapilli and bombs (Fig. 3c). Further away, the pale target rocks dominate. The masses of the ejected debris range from ~ 4 t to dust (Fig. 3b; Urbini et al. 2012).

44

Fig. 3. Field photos of Kamil ejecta. a) Panoramic view from the southeast side of the crater rim showing the main ejecta ray trending 130°E. The arrow indicates the largest ejecta boulder (sample L07; 2.1 x 0.7 x 1.1 m, equivalent to ~ 4 t), lying about 90 m outside from the crater rim. b) Close-up-view of the same boulder. The boulder is overturned with pale (sample L07a) and reddish (sample L07b) sandstone layers at its stratigraphic top and bottom, respectively. c) Sample L08: impact melt bomb (long axis ~ 10 cm) consisting of white pumiceous glass partially coated by a magnetic dark glass. d) Sample R01 (long axis ~ 12 cm): ejected sandstone showing a shatter cone surface (see text for details). Thousands of shrapnel pieces up to 34 kg in mass, formed through fragmentation upon hypervelocity impact of the Gebel Kamil iron meteorite, were found concentrated due southeast of the crater with a broad concentration maximum at ~ 200 m from the crater rim (D'Orazio et al. 2011).

With the exception of the eolian sand deposit that covers part of the northern crater wall, the Kamil structure is well-preserved. This is evidenced —for instance— by the lack of erosion features on the crater rim, and the essentially pristine distribution of bright ejecta and meteorite fragments (D’Orazio et al. 2011; Urbini et al. 2012). The pristine state of preservation is in agreement with the young age of the crater (< 5000 yr; Folco et al. 2011) estimated on the basis of archeological evidence.

Due to the high depth-to-diameter ratio of the transient cavity, Urbini et al. (2012) suggested that Kamil formed by the impact of a single iron mass (or a tight cluster of fragments). Based on the geometry of the crater and asymmetries in shrapnel and ejecta distribution Folco et al. (2011), D'Orazio et al. (2011) and Urbini et al. (2012) concluded that the Gebel Kamil iron meteorite arrived from the NW with an impact angle between 30° to 45°. The minimum projectile mass inferred from systematic meteorite search and geomagnetic data is ~ 5 t (Urbini et al. 2012).

45 SAMPLES AND ANALYTICAL METHODS

Geological samples studied in this work were collected during the first Italian– Egyptian geophysical campaign to Kamil in February 2010 (Tables 1 and 2; Fig. 1). Due to severe restrictions imposed by the local authorities, only a limited selection of rock samples from the crater wall and ejecta rays totaling less than 10 kg was allowed for export and research abroad. Nonetheless, these samples are a representative selection of the target enabling a first study of shock features at Kamil.

Two to three thin sections from each sample were made for petrographic observations by optical and electron microscopy. An optical microscope of the Zeiss Axioplan type and Scanning Electron Microscope (SEM) Philips XL30, operating at 20 kV coupled with an energy-dispersive X-ray detector (EDX), were used at Pisa University’s Dipartimento di Scienze della Terra. A Field Emission Scanning Electron Microscope (FE-SEM) Jeol JSM 6500F (upgraded to 7000 series), operating at 10 kV, was used at the Istituto di Geofisica e Vulcanologia (INGV) of Rome. Modal abundances of phyllosilicate minerals and iron-oxide matrix, as well as the porosity of the rock samples from the crater rim and ejecta were determined by digital image analysis of back-scattered electron images and X-ray element maps using the JMicroVision software.

We determined the crystallographic orientations of planar deformation features (PDFs) in the quartz crystals of sample L23 following procedures described in Langenhorst (2002) using a five-axis U-stage mounted on a Zeiss Axioplan optical microscope. For PDF indexing, the Automated Numerical Index Executor (ANIE, Huber et al. 2011) was used.

Quantitative chemical microanalyses were carried out with a Jeol JXA 8230 electron microprobe fitted with five wavelength dispersive spectrometers at the Institut des Sciences de la Terre (ISTerre) of Grenoble. Running conditions were 15 kV accelerating voltage, 12 nA beam current and 1 µm nominal beam spot. The ZAF procedure was employed for raw data reduction. Standards used for instrumental calibration were SiO2-rich glass (USNM 72854), ilmenite (USNM 96189), and hornblende (USNM 143965). Average detection limits are 0.04 wt% for Na2O and ZrO2, 0.02 wt% for Al2O3, Cr2O3, and V2O3, and 0.01 wt% for

MgO, CaO, K2O, FeO, MnO, TiO2, NiO, and P2O5. Typical precisions are better than 1% (relative standard deviation = 1 x standard deviation/average*100) for oxides concentration >1 wt% and between 1 and 10% for oxide concentrations between 1 and 0.1 wt%. The accuracy is typically better than 5%.

46 In order to concentrate and identify high-pressure silica polymorphs through X-ray Powder Diffraction (XRPD), aliquots of samples L23 and M25 were powdered, treated with dilute (~ 8 vol.%) hydrofluoric acid for 12 hours at room temperature, and then partitioned using a high-density liquid (2.675 g cm-3 sodium polytungstate aqueous solution). X-ray Powder Diffraction of particles denser that 2.675 g cm-3 was conducted using a Philips PW1830 instrument at Pisa University’s Dipartimento di Scienze della Terra operating with a Bragg-Brentano geometry and Ni filtered Cu Kα radiation. Each 2θ step (0.2°) was counted for 1 s.

In situ identification of silica phases was conducted through Laser-induced Raman microspectroscopy using a Horiba LabRAM HRVIS instrument of the “Centro Scansetti” at the Torino University’s Dipartimento di Scienza della Terra. The studied sample was a double-polished 100 µm-thick section from sample L23. Spectra were excited using the 532.11 nm emission of an argon-ion laser. The Raman microprobe system (focal length 800 mm) was equipped with an Olympus BX41 optical microscope and Peltier-cooled charge- coupled device (CCD) detector and with an Olympus 100X objective lens (numerical aperture 0.9), the focal spot was ~ 1-2 µm in diameter. Silica phase spectra accumulation times varied between 30 and 90 s, while diamond spectra were acquired in 1-5 s. The wavenumbers of the Raman lines were calibrated daily using the silicon band at 520.6 cm-1.

47

Table 1. Location, description and shock features of Kamil Crater target rocks collected from the crater walls and ejecta.

Shock features Incipient Melt Reduction High- Decom- Inter- Latitude Longitude Fractu Undulose PF/PDF PDF in melting of veins Sample Location Lithology Pressure of birefrin- pressure posed granular N E -ring extinction in quartz tourmaline accessory and (GPa) gence phases zircon melt minerals films Crater wall S sector, M to C pale quartzarenite with pervasive L02 22°1'5.58" 26°5'15.94" ------in situ UL syntaxial quartz. PI = 0 vol.%. < 1 S sector, VF pale wacke with level of C pale L13 22°1'5.47" 26°5'15.80" ------in situ UL siltstone, > 20 vol.% KM. PI = 0 vol%. < 1 M to C pale quartzarentite, ~ 6 vol.% KM S sector, and ~ 3 vol.% IM with levels of C < 1 L14 22°1'5.59" 26°5'15.79" slump ------siltstone to very F wacke, ~ 40 vol.% debris KM. PI = 4 vol%. N sector, C pale quartzarenite with pervasive < 1 M26 22°1'6.44" 26°5'15.45" in situ, top syntaxial quartz, < 5 of both KM and IM. ------of UL PI= 0 vol.%. N sector, in situ, VC reddish quartzarenite, ~ 6 vol.% KM < 1 M27 22°1'6.44" 26°5'15.45" ------bottom of and ~ 2 vol.% IM. PI = 17 vol%. UL Ejecta: sandstone blocks 50 m due M to F pale wacke, ~ 30 vol.% KM. PI = R01 22°1'8.02" 26°5'17.28" x x x - x n.a. x - - x NE of CR 0 vol.%. <5 54 m due R02 22°1'7.13" 26°5'18.16" NE of the VC to F pale quartzarenite. PI = 0 vol%. <5 x x x ------crater L07a M pale quartzarenite. ~ 15 vol.% < 1 90 m due KM, bottom of the ejecta. PI= 15 vol.%. L07 22°1'4.22" 26°5'18.35" ESE of L07b M to VC reddish quartzarentite. Up ------CR to 5 vol.% KM and IM, top of the ejecta. PI = 24 vol.%. M to F pale wacke, ~ 40 vol.% KM with 100 m due <5 M24 22°1'9.89" 26°5'16.04" levels of M- to C siltstone, ~ 20 vol.% x x x - - n.a. - - - - N of CR KM. PI = 0 vol.%. 400 m due L23 22°1'1.23" 26°5'30.13" F to C pale quartzarenite. PI = 0 vol.%. x x x x x x x x x x SE of CR 20-25 740 m due VF pale wacke to M pale siltstone. PI = 0 <5 M25 22°0'45.16" 26°5'1.73" SSW of x x x ------x vol.%. CR

Abbreviations: UL = upturned layer; CR = crater rim; VF = very fine; F = fine; M = medium; C = coarse; VC = very coarse; KM = clay minerals in matrix; IM= iron-oxide matrix; PI= porosity index; x= observed, - = not detected, n.a. = not analyzed.

Table 2. Location and description of impact melt lapilli and bombs from Kamil Crater.

Sample Latitude Longitude Location Description

Ejecta: impact melt lapilli and bombs Magnetic dark glass with inclusions of sandstone clasts, diaplectic glass and lechatelierite. Metallic blebs are scattered in the dark L15 22°01'6.05" 26°05'15.77" Inside the crater glass.

M23 22°01'5.69" 26°05'16.27" Inside the crater Gray lechatelierite. Vesicles are rounded and generally micrometric. Single grain of decomposed zircon.

52 m due ENE of L08 22°01'4.05" 26°05'16.53" White lechatelierite partially coated by a < 2 mm thick magnetic dark glass envelope. the CR 52 m due ENE of White lechatelierite with relicts of highly shocked quartz grains (melted PDFs, amorphization) stained by reddish iron-oxi- L09 22°01'4.01" 26°05'16.57" the CR hydroxides. 57 m due E of Magnetic dark glass with inclusions of sandstone clasts, diaplectic glass and lechatelierite. Metallic blebs are scattered in the dark L05 22°01'5.29" 26°05'18.17" the CR glass. 57 m due E of Magnetic dark glass with inclusions of sandstone clasts, diaplectic glass and lechatelierite. Metallic blebs are scattered in the dark L06 22°01'5.27" 26°05'18.14" the CR glass. Rare occurrence of fragments of the iron projectile Gebel Kamil. 57 m due ESE of Magnetic dark glass with inclusions of sandstone clasts, diaplectic glass and lechatelierite. Metallic blebs are scattered in the dark E30 22°01'3.39" 26°05'16.44" m.s. the CR glass. Rare occurrence of fragments of the iron projectile Gebel Kamil. Found stuck to the surface of iron-shrapnel specimens. Magnetic dark glass found stuck to the surface of a shrapnel of the iron projectile Gebel Kamil. Inclusions of sandstone clasts, 61 m due SSE of 22°01'4.51" 26°05'18.11" diaplectic glass and lechatelierite are abundant. Metallic blebs are scattered in the dark glass. Rare occurrence of meteorite MNA07m.s. the CR fragments. 120 m due ENE L03 22°01'7.91" 26°05'19.88" White lechatelierite coated by a < 2 mm thick magnetic dark glass envelop. of the CR 120 m due ENE L04 22°01'7.93" 26°05'19.92" Pumiceous white lechatelierite. Vesicles are elongated and can be up to 3 mm in size. of the CR

Abbreviations: CR = crater rim; m.s. = meteorite shrapnel.

RESULTS

Petrography of crater wall rocks

The layered rocks cropping out at the Kamil walls are sandstones ranging from very coarse quartzarenite (sample M27; Fig. 4a) to coarse siltstone with intercalated levels of very fine wacke (sample L13; Fig. 4b). The location and description of crater wall samples are summarized in Table 1.

The mineralogy of the crater wall rocks is dominated by sub-rounded quartz grains. Accessory minerals up to 400 µm in size include fine intergrowths of Fe-Ti oxides, besides zircon, tourmaline and rutile (Table 3). The sum of accessory minerals usually constitutes about 1 - 2 vol.% of the crater wall rocks. They are more common in fine-grained rocks.

The phyllosilicate matrix is dominated by kaolinite (Table 3). In some of the fine- grained samples, the matrix can reach up to 40 vol.% (e.g., L13 and L14, Figs. 4b and 4c and Table 2). Minor amounts (~ 2 vol.%) of iron oxides occur in the matrix of sample M27 (Fig. 4a). Samples M26 and L02 are devoid of matrix material, and show polygonal texture due to extensive overgrowth of syntaxial quartz cement (Fig. 4d).

The porosity of these rocks is generally lower than 4 vol.% (Figs. 4b and 4c), and close to zero in matrix-free samples M26 (Fig. 4d) and L02. The exception is sample M27 where porosity is 17 vol.% (Fig. 4a).

Table 3. Representative electron microprobe analyses (oxide wt%) of the matrix clay minerals and accessory minerals of unshocked samples.

Phyllosilicate mineral matrix Fe-Ti oxides Rutile Zircon Tourmaline

SiO2 42.0 45.9 46.1 38.9 1.38 0.57 0.03 34.3 36.8

Al2O3 33.1 36.8 36.6 30.2 1.97 1.27 0.04 b.d.l. 35.2

Na2O b.d.l. 0.07 b.d.l. 0.11 0.06 b.d.l. b.d.l. b.d.l. 1.46 MgO 0.10 0.13 0.12 0.07 0.07 0.19 b.d.l. b.d.l. 2.98 CaO 0.04 0.19 0.17 0.01 0.30 0.12 b.d.l. b.d.l. 0.05

K2O 0.05 0.05 0.05 0.03 0.02 b.d.l. b.d.l. b.d.l. 0.03

FeOtot 2.43 1.51 1.44 1.73 12.8 25.1 0.05 b.d.l. 9.63 MnO b.d.l. b.d.l. b.d.l. b.d.l. 0.06 0.12 b.d.l. b.d.l. 0.10

TiO2 0.46 1.93 2.17 0.76 73.6 64.0 93.5 b.d.l. 0.10

Cr2O3 0.03 0.03 b.d.l. b.d.l. 0.83 0.37 0.24 b.d.l. b.d.l. NiO b.d.l. b.d.l. b.d.l. b.d.l. 0.07 b.d.l. b.d.l. b.d.l. 0.03

V2O3 0.08 0.09 0.09 0.04 3.22 2.76 4.98 b.d.l. b.d.l.

P2O5 b.d.l. 0.18 0.16 0.08 0.22 0.09 0.01 b.d.l. b.d.l.

ZrO2 b.d.l. b.d.l. 0.05 0.11 0.44 0.21 0.11 66.8 b.d.l. Total 78.5 87.0 87.0 72.4 95.1 94.8 99.0 101.1 86.4

Abbreviation: b.d.l. = below detection limit. FeOtot : Total iron as FeO. Detection limits are 0.04 wt% for Na2O and ZrO2, 0.02 wt% for Al2O3, Cr2O3, and V2O3, and 0.01 wt% for MgO, CaO, K2O, FeO, MnO, TiO2, NiO, and P2O5.

51

Fig. 4. Photomicrographs of target rocks collected from the crater wall. a) Sample M27: very-coarse quartzarenite collected in situ from the upturned layers in the northern sector of the crater wall (crossed polarizers, XPL). The white arrows show the iron-oxide matrix. b) Sample L13: coarse siltstone to very fine wacke collected in situ from the southern sector of the crater wall (plane polarized light, PPL). c) Sample L14: layered sandstone collected from the slump debris on the southern sector of the crater wall (PPL). d) Sample M26: coarse quartzarenite collected in situ from the upturned layers in the northern sector of the crater wall (XPL). The inset shows a detail of the extensive overgrowth of syntaxial quartz cement (black arrows). Petrography of ejecta

Sandstone blocks

Six ejected sandstone samples were collected at different distances from the crater in order to explore possible variations of the degree of shock metamorphism with increasing ejection distance (Table 1).

In terms of overall texture and mineral composition, the ejected sandstone samples are indistinguishable from those of the crater wall rocks (Table 1; Fig. 5). They range from very- coarse pale quartzarenite to very fine pale wacke. Quartz is the most abundant mineral, followed by heavy accessory minerals (1 - 2 vol.%) including fine intergrowths of Fe-Ti oxides, zircon, tourmaline and rutile. The kaolinite-rich matrix is more abundant in the most fine-grained lithologies, totaling up to 40 vol.%. The matrix of the very-coarse sandstone L07b contains abundant iron oxides, constituting up to 5 vol.% of the bulk rock (Fig. 5b). The porosity is negligible in all samples but L07, in which it ranges from 15 vol.% (L07a) to 24 vol.% (L07b) (Table 1; Figs. 5a and 5b).

52

Fig. 5. Photomicrographs of four ejected target rock samples (L07, M24, L23 and M25), ordered according to their increasing distance from the crater rim (same magnification as in Fig. 4). a) Sample L07a: undeformed pale quartzarenite (PPL). b) Sample L07b: undeformed reddish quartzarenite (PPL). c) Sample M24: pale wacke with highly fractured quartz grains (PPL). d) Sample L23: shocked pale quartzarenite (PPL). e) Sample L23: same image of (d) in XPL. f) Sample M25: pale wacke (PPL). The black arrow points to a melt vein. See Table 1 for detailed sample description. Impact melt lapilli and bombs

Eight impact melt lapilli and bombs, ranging from a few centimeters to a few tens of centimeters in size, were collected within 100 m from the crater and inside the crater (Table 2; Fig. 1). They are divided into two end-member compositions: a white silica-rich glass and a Fe-Ni-rich dark glass (Fig. 6). White glass lapilli and bombs may be coated by a more or less continuous rim of dark glass up 2 mm thick (Fig. 3c; Table 2). Clasts of dark glass have also been discovered stuck to the surface of iron-shrapnel specimens (D’Orazio et al. 2011; Table 2). All samples are well preserved and they are not affected by secondary alteration, with the only exceptions being some portions of sample L09 that are stained by reddish iron oxi- hydroxides (Tables 2 and 4).

53

Fig. 6. Impact melt bombs (cut surfaces). a) Sample L04: white, pumiceous, lechatelierite. b) Sample L06: dark, magnetic glass with abundant sandstone clasts, lechatelierite clasts, and meteorite fragments. Abbreviations: MF = meteorite fragment; SC = shocked sandstone clast; LG = lechatelierite. White glasses typically show pumiceous texture with vesicle sizes ranging from a few tens of micrometers (Fig. 7a) to several millimeters (Fig. 6a). Sample L09 (Table 2) contains millimetric relicts of quartz grains. Silica content in white glass ranges from 99.2 wt% to 96 wt%. The glass is, thus, referred as lechatelierite. FeO content is lower than 0.04 wt%, with the exception of some portions of sample L09 where FeO content can reach ~ 4 wt%. The

Al2O3 content is lower than 0.26 wt%.

Dark glass lapilli and bombs show scoriaceous texture. They contain several inclusions (Fig. 7b); in order of abundance they are: clasts up to 5 mm in size of shocked sandstone (grain size up 700 µm) with planar deformation features, fragments up to 3 mm in size of vesicular white glass (lechatelierite), fragments up to 1 mm in size of diaplectic glass, µm-sized metal blebs and fragments up to 200 µm in size of the Gebel Kamil iron meteorite

(Figs. 6b and 7b). Dark glass has higher contents of Al2O3 and FeO than white glass (Table

4). The average of Al2O3/SiO2 and FeO/SiO2 ratios are 0.23 and 0.50, respectively. The maximum NiO content of dark glass is 2.51 wt%, but it generally ranges from 0.2 to 0.4 wt%. The inclusions of lechatelierite glass have significantly higher NiO content (0.05 wt%) than other lechatelierite glasses (≤0.01 wt%; Table 4). Metal blebs are significantly enriched in Ni (from 56.1 wt% to 69.8 wt%) and Co (from 1.56 wt% to 2.07 wt%) relative to the Gebel Kamil meteorite (Ni 20.6 wt% and Co 0.76 wt%, D’Orazio et al. 2011).

54

Table 4. Representative electron microprobe analyses (oxide wt%) of glass of impact melt lapilli and bombs.

White glass (lechatelierite) Dark glass L09 L03 L04 L06* L04 L09 L03 L06 L06 L06 L06

SiO2 99.2 98.5 98.5 98.5 98.2 96.0 60.2 57.4 54.3 54.3 49.8

Al2O3 0.03 0.02 0.03 0.03 0.26 0.11 14.8 11.9 12.5 11.4 12.4

Na2O b.d.l. 0.04 b.d.l. b.d.l. b.d.l. 0.03 b.d.l. 0.04 0.12 0.07 0.05 MgO b.d.l. 0.01 0.01 0.03 b.d.l. b.d.l. 0.29 0.19 0.20 0.22 0.17 CaO 0.01 b.d.l. b.d.l. 0.01 b.d.l. 0.01 0.25 0.87 1.04 0.74 0.76

K2O b.d.l. 0.01 b.d.l. 0.02 b.d.l. 0.01 0.18 0.20 0.22 0.28 0.11

FeOtot 0.03 0.04 0.02 0.02 0.01 3.93 19.9 25.4 28.6 29.7 32.2 MnO 0.00 0.01 0.00 0.00 0.00 0.00 0.27 0.37 0.66 0.55 0.66

TiO2 b.d.l. 0.02 b.d.l. b.d.l. b.d.l. b.d.l. 0.57 0.93 1.13 0.87 1.09 NiO b.d.l. b.d.l. b.d.l. 0.05 0.01 0.01 0.40 0.43 0.26 b.d.l. 2.51

P2O5 b.d.l. b.d.l. b.d.l. b.d.l. 0.02 b.d.l. 0.17 b.d.l. 0.01 0.02 0.18

ZrO2 0.04 b.d.l. b.d.l. b.d.l. b.d.l. 0.04 0.22 b.d.l. b.d.l. b.d.l. b.d.l. Total 99.4 98.7 98.6 98.6 98.5 100.2 97.2 97.6 99.0 98.0 100.0 Abbreviation: b.d.l. = below detection limit. *: white glass in inclusion in dark glass lapilli and bombs

Detection limits are 0.04 wt% for Na2O and ZrO2, 0.02 wt% for Al2O3, and 0.01 wt% for MgO, CaO, K2O, FeO, MnO, TiO2, NiO, and P2O5. FeOtot : Total iron as FeO

Fig. 7. Backscattered electron (BSE) images of sectioned impact melt bombs. a) White, pumiceous, lechatelierite (sample L04). b) Dark, magnetic, impact melt glass (sample L06). The latter is laden with clasts of target and projectile materials. Abbreviations: MF = meteorite fragment; SC = shocked sandstone clast; LG = lechatelierite; DG = diaplectic glass; MB = Ni-Fe metallic bleb. Shock metamorphic features

Fracturing, reduction of birefringence and undulose extinction

Fracturing, undulose extinction and reduction of birefringence in quartz are common to all ejecta sandstone samples with the exception of sample L07.

Concussion fractures, i.e., fractures due to the collision of two or more quartz grains, are common in samples containing relatively large quartz grains up to hundreds of

55 micrometers in size embedded in a more fine-grained mineral matrix (samples M24 and R02; Fig. 8).

Fig. 8. BSE images of ejected sandstone. a) Kaolinite-rich matrix between quartz grains in sample M24. b) Concussion fractures in quartz grains in sample R02. The number of irregular fractures per millimeter (f/mm) estimated according to Kowitz et al. (2013a) decreases in the following order: sample L23 (450 f/mm), R02 (~ 115 f/mm), R01 (~ 110 f/mm), M25 (~ 100 f/mm), and M24 (~ 60 f/mm). Note that sample R01 bears striated shattered surfaces decorated by discontinuous µm-thick silica-rich melt films (Fig. 3d; see section “Melt in shatter cones section for further details”). In this sample, fractures are generally parallel to the melt films and their abundance increases to ~ 170 f/mm towards the µm-thick silica-rich melt films.

Planar deformation features

Quartz

Planar deformation features (PDFs) in quartz grains occur in sample L23 and in several sandstone clasts embedded in the dark impact melt glasses (hereafter sandstone clasts). Irrespective of the grain size of the sandstone, PDFs occurring within individual crystals develop in domains up to 200 µm across (Fig. 9a).

A maximum of 4 sets of PDFs has been recognized in a single quartz grain (Fig. 9a). Overall, an average of 2.1 sets of PDFs per quartz grain is observed; PDFs occur in about 70% of quartz grains. The spacing between PDFs belonging to the same set in a quartz grain is not uniform and it ranges from 0.2 µm to 2 µm. Their thickness is rather uniform, varying from 0.08 µm to 0.15 µm (Fig. 9a).

We measured the crystallographic orientations of 149 sets of PDFs in 70 quartz grains in two representative thin sections of sample L23, and of 59 sets of PDFs in 29 quartz grains in three thin sections of dark impact melt glass containing sandstone clasts. The results of

56 PDF indexing are summarized in Table 5 and in Fig. 10. The histograms in Figs. 10a and 10b report the frequency distribution of PDF poles to c-axis angle within 5° bins (i.e., with a 5° error) for sample L23 and sandstone clasts, respectively. The histograms in Figs. 10c and 10d show the absolute frequency percent of indexed PDFs within a 5° error for sample L23 and sandstone clasts, respectively.

Fig. 9. BSE images of features in sandstone ejecta sample L23. a) Portion of a quartz grain showing four sets of PDFs. The four PDF orientations have been highlighted with four white-dashed lines. b) Quartz grain showing two main sets of enlarged PDFs, which underwent preferential melting. The inset shows an enlarged PDF containing a tiny elongated vesicle (~ 0.3 µm). The bright aggregate on the left-side of the image is a fine intergrowth of baddeleyite (ZrO2) and a SiO2 phase, resulting from the decomposition of a zircon crystal. c) Tourmaline grain surrounded by vesicles up to 50 µm in diameter as evidence of incipient melting at the contact with the surrounding quartz grains. White rectangle indicates the area of the close-up view of Fig. 9d. d) Close- up view of the rectangular area highlighted in (c) showing PFDs in tourmaline. e) Intergranular SiO2 melt (lechatelierite) with a highly vesicular central region. f) Close-up view of the glass area in (e) showing the textural relationships between the intergranular lechatelierite and PDFs in adjacent quartz grains.

57 The crystallographic orientations of PDFs in sample L23 and sandstone clasts are quite similar; however, sandstone clasts show a larger number of PDFs with poles to c-axis

angles > 40°. In sample L23, the most frequent orientations are {} 1013 23%, {} 1012 14%,

{} 1014 13%, {} 1011 11%, and {} 1122 7% (Table 5; Fig. 10c). Instead, in sandstone clasts the

orientations {} 1013 and {} 1122 are most common at 17%,€ followed by€ {} 1011 , 15%, {} 1012 ,

€ 12%,€ and {} 1014 , 10%€ (Table 5; Fig.10d). The unindexed planes represent the 15.4% and 15.3% of the total number of measured sets of PDFs in sample L23 and sandstone clasts, € € € € respectively. €

Fig. 10. Histograms showing the results of PDF indexing for quartz in quartzarenite sample L23 (a) and in sandstone clasts occurring in dark glass (b). Indexed PDFs are marked in grey, while unindexed PDFs are marked in black. (c-d) Histograms for the same samples with polar angle values of PDFs binned at 5° intervals (i.e., the estimated measurement error). The measured sets of PDFs that are within the area of overlap between the index and the index are shown in grey. {} 1013 {} 1014

€ €

58

Table 5. Crystallographic orientations of PDFs in quartz from sample L23 and from the sandstone clasts embedded in dark impact melt lapilli and bombs.

PDF Reference # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 e - 2+e

Unin- Miller-Bravais Indices (0001) {} 1013 {} 1012 {} 1011 {} 1010 {} 1122 {} 1121 {} 2131 {} 5161 {} 1120 {} 2241 {} 3141 {} 4041 {} 5160 {} 1014 dexed

Total number of sets 0 35 21 17 0 11 0 3 1 2 3 7 1 0 19 23 6 Sample L23 € € € € € € € € € € € € € € Absolute frequency 0 23% 14% 11% 0 7% 0 2% 1% 1% 2% 5% 1% 0 13% 15% 5%

Sandstone Total number of sets 0 10 7 9 0 10 0 2 0 0 1 1 0 1 6 9 3 clasts Absolute frequency 0 17% 12% 15% 0 17% 0 3% 0 0 2% 2% 0 2% 10% 15% 6%

Amorphous planar to subplanar and subparallel lamellae, thicker than common PDFs, were found in some quartz grains abutting melt pockets and melt veins (Fig. 9b). They show undulose margins with the host grain. They are in optical and compositional continuity with the surrounding melt and they can contain tiny vesicles (see inset in Fig. 9b). We interpret them as original PDFs that enlarged as consequence of the ongoing quartz melting along them. Actually, they represent an intermediate stage of the melting of quartz grains. We refer to these lamellae as enlarged PDFs. A similar occurrence of enlarged PDFs in quartz grains has been reported by Ebert et al. (2013; Fig. 5a, Qtz-B). Quartz grains with enlarged PDFs were not used for the PDF crystallographic orientation measurements and statistics.

Tourmaline

Planar deformation features were also observed in tourmaline (Figs. 9c and 9d). They are generally short, namely < 10 µm in length, and thin, namely < 0.1 µm in thickness. At least three PDF sets were identified in a single tourmaline grain. Some of them are decorated by tiny vesicles.

High-pressure phases

We found three high-pressure phases in sample L23: coesite, stishovite and diamond. Coesite is the most abundant. It was positively identified by micro-Raman and X-ray Powder Diffraction (XRPD) and imaged in the Field Emission Scanning Electron Microscope (FE- SEM); stishovite was identified only by XRPD; diamond only by micro-Raman.

Figure 11 shows the XRPD spectrum of the denser than 2.675 g cm-3 powder of sample L23. Besides the peaks of quartz, 2 additional peaks occur at 2θ = 28.9° and 30.4°. They are the 100 intensity peaks of coesite (040) and stishovite (110), respectively. No other peaks related to these phases are detectable, due to their low intensities and overlap with the peaks of quartz.

61

Fig. 11. X-ray powder diffraction pattern for sample L23 (fraction denser than 2.675 g cm-3). All X-ray peaks are from quartz, with the exception of the peaks at 2θ values of 28.9° and 30.4°, which are the 100 intensity peaks of coesite (040) and stishovite (110), respectively. Coesite was confirmed both as single phase and associated with diaplectic glass/ SiO2 melt by Raman microspectroscopy (Fig. 12a). Coesite peaks are all slightly shifted towards higher Raman shifts by ~ 2 cm-1 with respect to literature values (e.g., Frezzotti et al. 2012). Full width at half maximum of the coesite peak at 523 cm-1 ranges from 6.2 cm-1 to 11.6 cm-1. Optical microscope observations indicate that coesite is present within roundish domains of silica glass set between highly deformed quartz grains (Fig. 12a). They were observed in ~ 60 different occurrences for each thin section of sample L23. The average size of these coesite- bearing domains is 400 µm in diameter. Their cores are colorless in transmitted light and may host a few vesicles. Their peripheries are turbid brown in transmitted light. Back-scattered FE-SEM images show that these areas are composed of myriads of cryptocrystals (from 0.1 to 2 µm in size) of coesite, some of them showing resorbed crystal boundaries (Fig. 12d). They are embedded in a diaplectic glass/ SiO2 melt. The abundance of diaplectic glass/ SiO2 melt increases toward the colorless vesicle-bearing core. The chemical composition of one of these roundish aggregates is reported in Table 6.

62

Fig. 12. High-pressure silica phases occurring in quartzarenite sample L23. a) Raman spectra for coesite and for coesite + diaplectic glass/ SiO2 melt of sample L23. Intensity of the spectrum for coesite + diaplectic glass/ SiO2 melt was enhanced five times with respect to the spectrum for coesite. b) Photomicrograph of intergranular colorless SiO2 melt surrounded by brownish cryptocrystalline and amorphous material (PPL). c) BSE image of the area of photomicrograph (b). The arrows in (b) and (c) indicate the same vesicles within the colorless SiO2 melt. d) Detail of the outer zone (white rectangle in (c)) made up of sub-micrometric coesite grains (C) embedded in a glassy matrix (G).

63

Table 6. Representative electron microprobe analyses (oxide wt%) of glasses and mineral phases of ejecta sandstone samples L23 and M25.

L23 M25

Coesite Intergra- Intergra- Fe-Ti oxides Glass 1a Glass 2b Melt veins aggregate nular glass nular glass

SiO2 98.7 98.6 45.2 0.49 1.34 74.7 98.7 46.1 53.7 46.9

Al2O3 b.d.l. 0.05 36.1 0.53 0.52 0.68 b.d.l. 27.8 32.9 39.3

Na2O b.d.l. b.d.l. 0.25 b.d.l. b.d.l. b.d.l. b.d.l. 0.15 0.67 0.26 MgO b.d.l. b.d.l. 0.07 0.36 0.41 0.07 b.d.l. 0.09 0.17 0.11 CaO 0.02 0.02 0.18 0.20 0.14 0.06 0.02 0.08 0.23 0.20

K2O b.d.l. 0.01 0.12 0.01 b.d.l. 0.03 b.d.l. 0.10 0.18 0.41

FeOtot 0.01 0.03 0.99 23.1 21.5 4.21 0.01 1.22 1.39 1.27 MnO b.d.l. b.d.l. 0.01 1.02 1.00 0.18 b.d.l. b.d.l. 0.04 b.d.l.

TiO2 0.01 b.d.l. 3.13 68.5 67.9 12.1 0.01 1.62 2.06 3.16

Cr2O3 n.d. n.d. n.d. 0.21 n.d. n.d. n.d. n.d. n.d. n.d. NiO b.d.l. b.d.l. b.d.l. 0.02 b.d.l. b.d.l. b.d.l. b.d.l. 0.02 0.02

V2O3 n.d. n.d. n.d. 3.02 n.d. n.d. n.d. n.d. n.d. n.d.

P2O5 b.d.l. b.d.l. 0.14 0.13 0.09 0.06 b.d.l. 0.25 0.17 0.08

ZrO2 b.d.l. b.d.l. 0.14 b.d.l. b.d.l. b.d.l. b.d.l. 0.14 0.01 0.11 Total 98.8 98.7 86.3 97.6 93.0 92.1 98.8 87.3 91.5 91.9

Abbreviations: b.d.l. below detection limit. Detection limits are 0.04 wt% for Na2O and ZrO2, 0.02 wt% for Al2O3 Cr2O3, and V2O3, and 0.01 wt% for MgO, CaO, K2O, FeO, MnO, TiO2, NiO, and P2O5; n.d.= not determined. a Analysis of the glass from ~ 10 µm from the fine intergrowths of Fe-Ti oxides (see text for explanation) b Analysis of the glass from ~ 130 µm from the fine intergrowths of Fe-Ti oxides (see text for explanation)

FeOtot : Total iron as FeO

Micro-Raman spectra of a melt pocket within sample L23 further revealed the occurrence of a peak centered at 1336 cm-1 (Fig. 13), and although shifted towards higher Raman shifts (~ 4 cm-1 respect to literature values; e.g., Frezzotti et al. 2012), it corresponds to diamond. Full width at half maximum of the diamond peak ranges from 3.2 cm-1 to 6.4 cm- 1 . Diamonds are up to 1 µm in size and are commonly associated with diaplectic glass/ SiO2 melt.

Fig. 13. Raman spectra for diamond and for diamond + diaplectic glass/ SiO2 melt found in diaplectic glass/ SiO2 melt pocket of sample L23. Diaplectic glass/ SiO2 melt peaks are those with the asterisk. Intensity of the spectrum for diamond + diaplectic glass/ SiO2 melt was enhanced five times respect the spectrum for diamond. Zircon decomposition

Zircon is one of the most common accessory phases in the sandstones of Kamil. Some of zircon grains in the shocked sandstone L23 and in shocked sandstone clasts are decomposed into baddeleyite (ZrO2) + SiO2 fine intergrowths (Fig. 9b). A single decomposed zircon was also found in the impact melt bomb M23.

Intergranular melt

Thin films (from a few micrometers up to 200 µm) of glass occupy the intergranular space between some quartz grains showing PDFs and between quartz and accessory minerals in sample L23. The composition of these intergranular melts varies from nearly pure lechatelierite (SiO2 = 98.6 wt%; Table 6) to a “kaolinite” glass (Al2O3/SiO2 = 0.8; Table 6).

Figures 9e and 9f show details of an intergranular glass film with nearly pure SiO2

67 composition (SEM-EDS analysis) characterized by highly vesicular central portion. In quartz crystals adjacent to glass films, planar amorphous lamellae occur; they could be interpreted as enlarged PDFs or as injected melt in fractures. They are in optical and compositional continuity evidencing an intimate genetic relationship.

Figure 9c shows a tourmaline crystal whose boundaries abutting quartz grains underwent melting. Melting is documented by the occurrence of vesicles up to 50 µm in diameter.

The glass surrounding a Fe-Ti oxide crystal shows a compositional variation that we interpret as the mixing between a Fe-Ti-rich melt and a SiO2 melt (Table 6). The contents of

FeO and TiO2 are, respectively, 23.1 wt% and 68.5 wt% in the unmelted oxide, 21.5 wt% and 67.9 wt% in the glass analyzed at ~ 10 µm from the oxide, and 4.21 wt% and 12.1 wt% in the glass analyzed at ~ 130 µm from the oxide. The TiO2/FeO ratio is, however, almost unchanged (3.0 in the unmelted portion, 3.1 in the glass at ~ 10 µm from the oxide boundary, and 2.9 in the glass at ~ 130 µm from the oxide boundary). Intergranular melts have been also observed between quartz and zircon grains in sample L23.

Melt veins

A few melt veins crosscut sample M25. They are typically a few tens of micrometers wide with some up to 0.2 mm (Figs. 5f and 14). The contact with the host rock is sharp and no shock features occur in the latter (Fig. 14a). Occasionally, orthogonal or quasi-orthogonal injection veins propagate from a main vein (Fig. 14a). Several relict minerals occur in the veins. Quartz is the most common and sometimes it shows planar or subplanar amorphous lamellae (Fig. 14b). These structures resemble the enlarged PDFs occurring in shocked quartz grains of sample L23 (Fig. 9b). The chemical composition of glass in melt veins is the result of the mixing between kaolinite-derived, quartz-derived, and Fe-Ti oxide-derived melts. The

Al2O3/SiO2 ratio of the glasses of melt veins varies between 0 and 0.84 (Table 6; Fig. 14c;

Al2O3/SiO2 ratio of pure kaolinite end-member = 0.85), with most values above 0.5 indicating that the kaolinite fraction in the melt is generally above 50 %. The veins brighter in the SEM-

BSE images contain up to 3.16 wt% and 1.39 wt% of TiO2 and FeO, respectively, as a result of the melting of Fe-Ti oxides (Table 6; Figs. 14a and 14b). These compositional variations give rise to schlieren, which emphasize flow textures (Figs. 14a and 14b). No high-pressure phases were identified by Raman microspectroscopy in the melt veins.

68

Fig. 14. BSE images of ejected wacke sample M25. a) Finely vesicular melt vein. The bright material is enriched in Fe and Ti. Note the straight contact with the undeformed host rock, and the arrowed injection vein on the right side of the vein. b) Close-up view of the white rectangular area in (a) showing a finely vesicular portion with schlieren and relict quartz grains with planar amorphous lamellae. c) Highly vesicular silica- and alumina- rich melt forming a melt vein. It is the result of the preferential melting of the phyllosilicate mineral matrix and quartz of the host rock.

69 Melt in shatter cones

The external surfaces of the pale wacke sample R01 show mesoscopic fractures with striae that are arranged in poorly defined cm-scale horse-tail patterns in hierarchical branched structures (Figs. 3d, 15a and 15b). Some striated surfaces have semiconical morphology (Fig. 15b). Overall, we identify these structures as shatter cones.

The striated surfaces of the cones are coated by thin white films (< 200 µm in thickness). They are indicated by black arrows in Fig. 15b. Petrographic investigations (optical and electron microscopy) suggest that these films are made by amorphous material. Sections orthogonal to the cones (Figs. 15c and 15d) show undulating sample morphology. In thin section (Fig. 15e), the thin films of glass are nearly opaque. Backscattered electron images reveal that the glass is highly vesicular and clast-laden (Figs. 15f and 15g). Clasts include mainly relicts of silica mineral grains. Back-scattered electron images of sections broadly parallel to the direction of the striae show the fracture number in the host wacke increases by about 1.5 x on approaching the contact with the melt (see section “Fracturing, reduction of birefringence and undulose extinction”), and that vesicles in the melt are coherently stretched defining a shear fabric (Fig. 15g). The major axes of the vesicles form an angle of ~ 45° relative to the contact with the host wacke. The glass is silica- and alumina- rich with Al2O3/SiO2 = 0.48 (SEM-EDS analyses), very close to the Al2O3/SiO2 ratio (0.46; SEM-EDS analyses) of the whole rock.

70

Fig. 15. Mesoscopic and microscopic features of wacke sample R01. a) Shatter cone structures with striae arranged in a horse-tail patterns. b) Close-up view of the rectangular area in image (a). Striations on the shatter cone surface radiate from a common apex. They are discontinuously coated with by a white film (100s of µm thick) of silica-rich glass (black arrows). The white arrow indicates where images (c) and (d) were taken, and where the thin section featured in (e) was cut. c) A cross sectional view of the shatter cone surface coated by silica rich glass. d) Close-up view of the same feature shown in (c). e) Mosaic of photomicrographs of a thin section of sample R01 cut perpendicular to the shattered surface. The silica-rich melt appears as a discontinuous brown coating. The black arrow indicates where the BSE image shown in (f) was taken. f) BSE image of the silica-rich glass coating. The glass is highly vesicular and contains quartz relics (see white arrows) of the host rock. g) BSE image of the silica-rich glass coating of a section broadly parallel to the direction of the striae; vesicles in the melt are coherently stretched defining a shear fabric forming an angle of ~ 45° (see dashed white lines) relative to the contact with the host wacke.

71 DISCUSSION

All samples from the crater wall of Kamil and the largest ejecta boulder (sample L07) do not show any shock feature; these rocks, thus, experienced shock pressures < 1 GPa (Kieffer 1971; Table 1). The only impact-related macroscopic structures observed on crater wall samples are striated surfaces on tear fault planes described by Urbini et al. (2012).

All other ejecta samples exhibit a wide range of shock features from fracturing to impact melting. Shock features have been divided into two categories, as a function of their abundance at the thin section scale: 1- pervasive shock features are the most abundant and can represent up to the 100 vol.% of the sample: they include fracturing, PDFs, and impact melt lapilli and bombs, and 2- localized shock features occur in less than 1 vol.% of the sample: they include high-pressure phases and localized impact melting in the form of intergranular melt, melt veins and melt films on shatter cones.

Pervasive shock features also allow to estimate shock pressures suffered by the target rocks, whereas localized shock features are consequence of local enhancement of shock pressure and temperature in correspondence to heterogeneities of the target rock.

Pervasive shock features and shock pressures

According to Kowitz et al. (2013a) the fracture density is directly proportional to the shock pressure suffered by rocks. As the porosity of sandstone used for the experimental calibration by Kowitz et al. (2013a) is higher than the porosity of sandstone rocks of Kamil area, we can not apply their calibration to our sample. We can only constrain the shock pressure suffered by sample M24 (~ 60 f/mm), M25 (~ 100 f/mm), R01 (~ 110 f/mm), and R02 (~ 115 f/mm) to below 5 GPa and suggest that pressure increased from sample M24 to sample R02. Shock pressure below 5 GPa for sample R01 showing shatter cones is consistent with the most common low-pressure formation regime for shatter cones, ranging from ~ 2 to ~ 10 GPa (French 1998). Sample L23, which has a fracture density of ~ 450 f/mm suggests the highest shock pressures (>> 5 GPa), that is confirmed also by the occurrence of PDFs.

The overall distribution of crystallographic orientations of PDFs in quartz crystals of sample L23 and of the sandstone clasts embedded in impact melt is similar. They are both dominated by {} 1013 orientations and by the lack of (0001) orientations (Table 5; Fig. 10). In sedimentary targets, the formation of PDFs with (0001) and {} 1013 orientations is usually prevented because the shock energy is initially used to close the pores of the target rocks €

€ 72 (Robertson 1980). As a result, PDFs are generally rare in sedimentary targets (e.g., 5 vol.% in Coconino Sandstone, Barringer Crater, Arizona, USA; Robertson 1980) or their orientations

are dominated by high angles to the c-axis (e.g., {} 1122 , {} 1011 ; Grieve and Therriault 1995; Grieve et al. 1996). Some target rocks at Kamil have low porosity (e.g., porosity in crater wall sandstones is < 17 vol.% and typically < 4 vol.%; Table 1). It is thus likely that only a small € € fraction of the shock energy was used for pore collapse, thereby preventing formation of

PDFs with (0001) orientation and allowing formation of PDFs with {} 1013 orientation. In other words, due to their low porosity, some of the Kamil target rocks behaved like crystalline rocks. This behavior allows us to infer shock pressures using experimental calibrations on € single quartz crystals reported in the literature (e.g., Stöffler and Langenhost 1994). The distribution of PDF orientations in sandstone clasts exhibits a higher abundance of high angles to c-axis orientations (> 40°) relative to sample L23 (Table 5; Fig. 10). The distribution

in sandstone clasts is dominated by the {} 1013 , {} 1122 , and {} 1011 orientations, indicating shock pressures between 10 and 20 GPa. The distribution in sample L23 is dominated by index {}1013 and {}1012 , indicating shock pressures between 20 and 25 GPa. Planar € € € deformation features are distributed in domains within individual crystals; their maximum length is ~ 200 µm; their spacing ranges from 0.2 µm to 2 µm; the maximum number of PDF € € sets occurring within a single crystal domain is 4.

In non-porous crystalline rocks, melting of individual minerals starts at ~ 40 GPa, whereas whole-rock melting starts at ~ 60 GPa (Stöffler 1971). In sandstones (e.g., Coconino Sandstone, porosity: 10-20 vol.%), shock melting of individual quartz grains starts at pressures as low as ~ 5 GPa and whole-rock melting occurs above ~ 30 - 35 GPa (Kieffer et al. 1976; Kowitz et al. 2013b). Recent experiments on porous sandstones (porosity: 25-30 vol.%) show that ~ 80 vol.% of the target material melts or transforms into high-pressure phases at 17.5 GPa (Kowitz et al. 2013b). Whole rock melting at Kamil is documented by impact melt lapilli and bombs (Figs. 3c, 6 and 7; Table 2). These consist of white lechatelierite and black Fe-Ni-rich glass (Figs. 6 and 7; Table 4). The former is the result of bulk melting of the target rocks. The latter testifies to melting of the iron meteorite projectile and mixing with the target-derived melts (see also Folco et al. 2011). In particular, the nearly

pure SiO2 composition (SiO2 > 96 wt% and Al2O3 < 0.24 wt%; Table 4) of the white glass indicates that the most likely precursor material amongst the target rocks studied in this work was similar to the quartzarenite almost free or devoid of kaolinite-rich matrix (e.g., samples

73 L02 and M26 collected from the crater wall at the top of the stratigraphic sequence). The porosity of these quartzarenite rocks is extremely low (typically < 4 vol.%; Table 1) due to pervasive growth of syntaxial quartz (e.g., Fig. 4d). Sedimentary target rocks of this nature are expected to behave like crystalline rocks. We thus infer that shock pressures experienced by the precursor rocks of the white glass were between ~ 30 and ~ 60 GPa. More likely, shock pressures were > 50 GPa, on the basis numerical modeling by Wünnemann et al. (2008) for rocks with a very low initial porosity (~ 4 vol.%).

Impact velocity

The planar impact approximation (Melosh 2013) provides a theoretical basis for constraining a fundamental parameter for the definition of the Kamil impact scenario: the impact velocity at the contact with the target. The vertical component of the impact velocity was calculated using the linear shock-particle velocity equation of state parameters of an iron meteorite projectile and a sedimentary target with the physical properties of the Coconino Sandstone (Melosh 2013). Input shock pressures were 30 GPa and 60 GPa, i.e., the range of maximum shock pressures recorded by the target rocks at Kamil that generated impact melt lapilli and bombs.

The results of the planar impact approximation indicate that the vertical component of the impact velocity is 3.5 km s-1 for 30 GPa and of 5.5 km s-1 for 60 GPa. Assuming an impact angle of 45°, as inferred by Urbini et al. (2012), the Gebel Kamil iron projectile impacted the ground with a face-on impact velocity between 5.0 km s-1 (30 GPa) and 7.5 km s-1 (60 GPa).

Localized shock features

The closure of the pores upon shock loading of porous and heterogeneous target rocks can generate local amplification of shock pressure and temperature (e.g., Grady 1980; Wünnemann et al. 2006; Schade and Wünnemann 2007; Wünnemann et al. 2008; Ogilvie et al. 2011; Kowitz et al. 2013b). For instance, experimental work and modeling by Güldemeister et al. (2013) showed that the initial shock pressure in porous and heterogeneous sandstones may experience a four-fold increase. Pressure amplification due to pore collapse can, thus, determine the localized occurrence of high-pressure phases and melting in rocks that underwent low to intermediate shock pressures (5 GPa - 20 GPa).

A wide range of localized shock features are recorded in the sandstone target rocks at Kamil (Table 1). They can be divided into two main groups: 1- high-pressure phases, namely

74 coesite, stishovite and diamond; 2- localized shock melt, namely intergranular shock melt, melt veins and shock melt films on shatter cones. Some of these localized shock features have never been reported before from small impact structures (Table 7).

Three high-pressure mineral phases were identified in sample L23 (Table 1): coesite, stishovite and diamond. Coesite occurs in roundish area, described in Figs. 12b and 12c. Figure 12d shows that coesite occurs as irregularly-shaped sub-micrometric crystals more abundant towards the periphery of intergranular diaplectic glass/ SiO2 melt pockets. Silica glass pockets are often vesicular towards the center (Figs. 12b and 12c). The microstructural setting of coesite in sample L23 is very similar to that of coesite found by Kieffer et al. (1976; Fig. 7a, p. 56) in Coconino Sandstones shocked between 20 and 30 GPa and described as symplectic regions. Kieffer et al. (1976) proposed that these microstructures were formed by a complex multistage process of phase changes (see Fig. 25, p. 89, in Kieffer et al. 1976), summarized as follows: (i)- a hot, amorphous water-bearing silica material (called “jet”) is injected into the pores during the shock wave passage; (ii)- during the initial decompression crystals of coesite nucleate and grow; water present in the initial jet concentrates in the residual melt; (iii)- upon further pressure release, crystals of coesite in the central region of the original pore begin to melt; (iv)- subsequently, water vapor exsolves from the melt forming vesicles within the glass. Due to the similarity between the coesite-melt microstructures observed in shocked quartzarenite rocks from Kamil and in shocked Coconino Sandstone from Barringer Crater, we suppose that the above-described process probably occurred also during the shock metamorphism of Kamil target rocks.

The full width at half maximum for the coesite main peak and diamond peak indicate that these phases are crystalline. Peaks for coesite and diamond are slightly shifted towards higher Raman wavenumbers, suggesting a moderate confining pressure (e.g., Hemley 1987) probably exerted by the diaplectic glass/ SiO2 melt that has a higher volume than other SiO2 phases (Figs. 12a and 13).

Films of intergranular melt occurring in quartzarenite sample L23 at the contact between quartz grains have a variable composition from pure SiO2 to pure kaolinite. This indicates that the precursor material of sample L23 contained some (likely small) amount of phyllosilicate mineral matrix. The films of silica glass are often in continuity with enlarged PDFs (Fig. 9b). The sequence of images shown in Fig. 9a and Fig. 9b, features some analogies with the PDF formation model proposed by Langenhorst (1994). According to this model, PDFs start to form by solid-state amorphization at pressures between 5 GPa and 10

75 GPa, i.e., at a shock pressure regime unable to produce shock temperatures and post-shock temperatures above the melting temperature of quartz (Fig. 9a). Between 25 GPa and 35 GPa small bands of melt form in correspondence to PDFs, because shock temperature exceeds melting temperature. Over 35 GPa both shock and post-shock temperatures are higher than melting temperature and the quartz crystals are more extensively consumed by melting (Fig. 9b). Finally, over 50 GPa lechatelierite forms (see inset Fig. 9b).

Localized shock melting due to amplification of shock pressures and temperatures associated to pore collapse is also documented at the crystal boundaries between accessory mineral and quartz in sample L23 (Fig. 9c). Note that when hydrated minerals like tourmaline are involved, the melt volume and the vesicle size are greater than that produced by anhydrous minerals, documenting a role of volatiles in shock melting.

The chemical composition of the melt veins of sample M25 (Table 6) indicates that they derived from the preferential melting of the phyllosilicate mineral matrix of the host rock. Phyllosilicate minerals have lower melting and breakdown temperatures (up to 650°C), Mohs number (up to 4), indentation hardness (up to 2 GPa), yield strength (up to 0.66 GPa), and shear yield strength (up to 30 GPa) than other common rock-forming minerals (Spray 2010). We thus conclude that coarse siltstone levels in sample M25 played an important role in the formation of these melt veins. Sample L13 show alternating sandstone and siltstone levels and it could represent the protolith of sample M25 (Table 1; Fig. 4b).

The hundred-micrometers-thick glass films observed on ejecta sandstone sample R01 on shatter cone surfaces (Fig. 15) is characterized by a highly vesicular texture and high silica and alumina composition (Al2O3/SiO2 = 0.48) consistent with localized melting of the host wacke (Al2O3/SiO2 = 0.46). Melt films at the surface of shatter cones were also reported from the Sudbury (Ontario, Canada) and Vredefort (South Africa) impact structures (e.g., Gay 1978; Gibson and Spray 1998; Nicolaysen and Reimold 1999) and, possibly, from the Santa Fe (New Mexico, USA) impact structure (Fackelman et al. 2008). Although it is generally accepted that shatter cones typically develop at low shock pressures between ~ 2 and ~ 10 GPa (French 1998), the precise mechanism of shatter cone formation is still debated (e.g., Sagy et al. 2002 and 2004; Baratoux and Melosh 2003; Wieland et al. 2006). Nonetheless, evidence from Kamil sample R01 confirms that melting conditions can be attained at the interface between adjacent shatter cone surfaces. Furthermore, the micrometric-scale shear fabric observed in the melt films (namely, parallel to the major axis of the shatter cones; Fig. 15g) suggests that frictional melting contributed, at least in part, to their formation, as

76 previously suggested by Nicolaysen and Reimold (1999) based on the occurrence of microdisplacements on the surface of some shatter cones from the Vredefort impact structure.

Comparison of small (< 1.5 km in diameter) impact structures

Kamil is the sixth smallest impact structure (45 m-in-diameter) known on Earth. In common with most of very small (< 300 m-in-diameter) and small (< 1.5 km-in-diameter) impact structures, Kamil was formed by the impact of an iron or stony-iron meteorite (19 out of 27) on a sedimentary target (17 out of 27; Table 7). The age of the impact is still uncertain. It is, however, likely younger than 5000 yr based on archeological evidence (Folco et al. 2011) and, therefore one of the ten youngest impact craters of Earth.

As discussed in the previous sections, ejecta at Kamil show a wide range of shock features, some of them never reported for impact craters with sizes comparable to Kamil (Table 7). Urbini et al. (2012) estimated that the volume of rock excavated by the impact of the Gebel Kamil meteorite was about 3800 m3. Not all the excavated rocks suffered shock pressure sufficient to produce shock effects, for example the large quartzarenite boulder L07. During impact cratering, the volume of rock that suffered shock pressures between 1 and 5 GPa is much larger than the volume of rock shocked over 5 GPa (French 1998). Moreover, incoherent deposits are easily eroded and weathered. Thus, the common occurrence of shock metamorphosed rocks and impact melt lapilli and bombs is a further proof of the pristine state of preservation of Kamil. There are three reasons for the exceptional state of preservation of Kamil: 1- the young age of the impact (most likely < 5000 yr; Folco et al. 2011); 2- the mechanical strength of the target rocks, namely quartzarenite (Table 1); and 3- the low erosion rates of arid desert areas (e.g., 1-16 m Ma-1 for the Namib Desert; Bierman and Caffee 2001).

Table 7 shows that shock features have been reported only for 19 small impact craters out of 27, and often in abstract form only. In very small impact craters < 300 m diameter, observations about shock features are reported almost exclusively for impact craters younger than 5000 yr. Besides Kamil, significant shock features have been reported only from Wabar (Saudi Arabia). Extending the comparison to impact craters up to 1.5 km diameter, extensive shock features have been reported for Aouelloul (Mauritania), Monturaqui (Chile), and Barringer (Arizona, USA).

77

Table 7. Summary table of shock features in small terrestrial impact craters with diameters ranging from 13.5 m to 1.5 km. Data about Kamil Crater are highlighted in bold. PF Whole-rock Shock in Diameter Age Shatter HP Diaplec- Melt Interstit- Projectile Crater Location Target Projectile and/or impact other Notes Ref. (km) (Ma) cones phases tic glass veins ial glass impact meltb PDF glassa minerals Carancas Peru 0.0135 S chondrite 0.000007 n.r. x deb. n.r. n.r. n.r. n.r. n.r. n.r. n.r. melt deb. (1), (2), (3) Kansas, Haviland 0.015 S pallasite 0.001 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1) USA dens. of Dalgaranga Australia 0.024 C stony-iron 0.02 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. Qtz (1), (4) (r.a.f.) iron - Sikhote Alin Russia 0.027 C 0.000067 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1) IIAB Alberta, iron - Whitecourt 0.036 S <0.0011 n.r. x deb. n.r. n.r. n.r. n.r. n.r. x n.r. (1), (5) Canada IIIAB x (Fe-Ti iron - *Coe, Stv, Kamil Egypt 0.045 S <0.005? x x x* x x x x L and B x oxide, Tur, (1), (6) ung. and Dia Zrn) iron - Campo del Argentina 0.05 C IAB- <0.004 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1) Cielo complex Sobolev Russia 0.053 C iron <0.001 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1)

Ilumetsä Estonia 0.08 S unk. 0.0066 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1) iron - Veevers Australia 0.08 S <1 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1) IIAB Morasko Poland 0.1 S iron <0.1 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1) iron - 0.004 ± Kaalijärv Estonia 0.11 S IAB- x n.r. n.r. n.r. n.r. n.r. x (r.a.f.) n.r. n.r. (1), (7), (8) 0.01 complex (1), (9), (10), Saudi iron - Wabar 0.116 S 0.00014 n.r. x x coe n.r. n.r. n.r. x L and B x n.r. (11), (12), Arabia IIIAB (13) iron - 0.0042 ± (1), (13), Henbury Australia 0.157 S n.r. n.r. n.r. n.r. n.r. n.r. x L x n.r. IIIAB 0.019 (14) (continued)

Table 7. Continued PF Shock in Diameter Age Shatter HP Diaplec- Melt Interstit- Whole-rock Projectile Crater Location Target Impactor and/or other Notes Ref. (km) (Ma) cones phases tic glass veins ial glass impact glassa impact meltb PDF minerals iron - Texas, Odessa 0.168 S IAB- <0.0635 n.r. n.r. n.r. n.r. n.r. n.r. x (r.a.f.) x n.r. (1), (15) USA complex iron - 0.0054 ± Boxhole Australia 0.17 C n.r. n.r. n.r. n.r. n.r. n.r. n.r. x n.r. (1), (16) IIAB 0.0015 Only one glassy 5 α-iron Macha Russia 0.3 S iron < 0.007 n.r. x x Stv n.r. n.r. n.r. spherule (0.12 particles, only n.r. (1), (17) mm) one spherule Aouelloul Mauritania 0.39 S iron 3.0 ± 0.3 n.r. x n.r. n.r. n.r. n.r. x x x (Zrn) (1), (18) x (planar Amguid Algeria 0.45 S unk. <0.1 n.r. elements; n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1), (19) r.a.f.) iron - (1), (13), Monturaqui Chile 0.46 C IAB- <1 n.r. x x Coe x n.r. n.r. x x x (Pl, Bt, Ap) (20) complex South 0.25 ± x s.c.- suevite Kalkkop 0.64 S unk. x n.r. x n.r. n.r. x n.r. n.r. (1), (21) Africa 0.05 like breccia dens. of iron - Wolfe Creek Australia 0.875 S <0.3 n.r. x (r.a.f.) n.r. n.r. n.r. n.r. n.r. n.r. n.r. Qtz (1), (22) IIIAB (r.a.f.) Tswaing (ex- South 0.220 ± x sulfides suevite Pretoria 1.13 C chondrite n.r. x n.r. x n.r. n.r. x X (Afs) (1) (23) Africa 0.052 spherules breccia Saltpan) iron - Arizona, 0.049 ± x Coe *suspec- (1), (9), (24), Barringer 1.19 S IAB- n.r.* x n.r. n.r. x x x x USA 0.003 Stv ted (25), (26) complex Tabun-Khara- Mongolia 1.3 C unk. 150 ± 20 n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. n.r. (1) Obo x x (Bt, Afs) Karikkoselkä Finland 1.5 C unk. 230 x (r.a.f.) n.r. n.r. n.r. n.r. n.r. n.r. (1), (27) (r.a.f.) (r.a.f.) Saarijärvi Finland 1.5 C unk. >600 x x n.r. n.r. n.r. n.r. n.r. n.r. x (Bt, Chl) (1), (28)

Abbrevations: Ref. = references; S = sedimentary; C = crystalline; ung. = ungrouped; unk. = unknown; x = observed; n.r. = not reported; r.a.f. = reported in abstract form; deb. = debated; dens. = densification; L = lapilli; B = bombs; s.c. = shatter cones; Coe = coesite; Stv = stishovite; Dia = diamond; Qtz = quartz; Tur = tourmaline; Zrn = zircon; Pl = plagioclase; Bt = biotite; Ap = apatite; Afs = alkalifeldspar; Chl = chlorite a Whole-rock impact glass indicates a glass resulting only by the melting of the target rocks. b Projectile impact melt includes both whole-rock impact glass contaminated by meteoritic components and metal spherules. (1) http://www.passc.net/EarthImpactDatabase/saarijarvi.html; (2) Tancredi et al. 2009; (3) Kenkmann et al. 2009; (4) Miura and Kato 1991; (5) Kofman et al. 2010; (6) This work; (7) Dietz 1968; (8) Smith and Hodge 1993; (9) Bunch and Cohen 1964; (10) Chao et al. 1961; (11) Hörz et al. 1989; (12) Mittlefehldt et al. 1992; (13) Gibbons et al. 1976; (14) Taylor 1967; (15) Smith and Hodge 1997; (16) Hodge and Wright 1970; (17) Gurov and Gurova 1998; (18) Koeberl et al. 1998; (19) McHone et al. 1990; (20) Bunch and Cassidy 1972; (21) Reimold et al. 1998; (22) Miura 1995; (23) Reimold and Koeberl 2014; (24) Mark 1995; (25) Kieffer et al. 1976; (26) Mittlefehldt et al. 2005; (27) Personen et al. 1999; (28) Öhman and Preeden 2013.

The most common shock features in small impact craters are PDFs and impact melt lapilli and bombs. Planar deformation features are, however, rare in very small impact craters (5 out of 17 very small impact craters). Shatter cones and high-pressure phases have been reported only from 4 small impact craters (including Kamil). Melt veins have been reported only from Kamil (Table 7), although, in Kalkkop a possible pseudotachylite veinlet was found (Reimold et al. 1998).

The shock features found in ejecta, from the Kamil indicate that the formation of these features is possible also in very small impact structures formed by meter-sized projectiles impacting the ground with a high velocity, i.e., > 3 km s-1 (hypervelocity impact). Their detection requires a combination of special circumstances like the young age of the crater, the high mechanical strength of target rocks, and minimal burial and low erosion rates.

CONCLUSIONS This is a detailed report of the petrography and some chemical observations of samples from the crater wall and ejecta deposits from Kamil Crater (Egypt) collected during the first Italian-Egyptian geophysical expedition in February 2010. Data allow us to draw the following conclusions in terms of shock effects recorded by target rocks, impact scenario, and unique state of preservation of the ejecta rocks: 1. A broad set of shock features is recorded in the target rocks at Kamil (45-m-in- diameter), ranging from fracturing to whole-rock impact melting. 2. Shock features are classified into two categories: 1- pervasive shock features, including fracturing, planar deformation features, and impact melt lapilli and bombs, and 2- localized shock features including high-pressure phases and localized impact melts occurring as intergranular melt, melt veins and melt films enveloping shatter cones. 3. Kamil is the smallest impact structure where shatter cones, coesite, stishovite, diamond, and impact melt (target and projectile) have been reported. 4. Because of the occurrence of impact melt lapilli and bombs, the maximum shock pressure during the impact of the iron meteorite Gebel Kamil was between 30 GPa and 60 GPa. 5. The maximum shock pressures recorded at Kamil can be achieved through face-on impact velocities from 5.0 km s-1 (30 GPa) and of 7.5 km s-1 (60 GPa), assuming an impact angle of 45°.

81 6. Localized shock features, namely high-pressure phases (coesite, stishovite, and diamond) and localized impact melts (intergranular melt, melt veins and melt films on shatter cones), are the result of enhanced shock pressure and temperature related to pores collapse and heterogeneities of the target rocks. 7. Shatter cones from Kamil are coated by thin (< 200 µm) films of glass, confirming that melting conditions can be attained at the surfaces of shatter cones, and evidence for shear shows that frictional melting contributed, at least in part, to their formation. 8. The hypervelocity impact of meter-sized iron meteorite projectiles can produce shock effects similar to those observed in high velocity, larger impacts. The young age of the crater (most likely < 5000 yr), the mechanical strength of target rocks and the low erosion rates of the hot-desert area played a crucial role in the preservation of all these shock features.

Acknowledgements - This work was supported by the Italian Ministero degli Affari Esteri - Progetti di Grande Rilevanza, Protocollo Esecutivo ITALIA-EGITTO. The 2010 geophysical expedition work was carried out within the framework of the 2009 Italian- Egyptian Year of Science and Technology. We thank Prof. M. Alsherbiny (former President of the Egyptian National Academy for Scientific Research and Technology) and Prof. F. Porcelli (Scientific Attaché, Italian Embassy, Egypt) for diplomatic and institutional support; the Egyptian Army for logistical support. Agnese Fazio was supported by the PhD School Scuola di Dottorato in Scienze di Base Galileo Galilei program on Earth Science of the University of Pisa, and by the Barringer Family Fund for Meteorite Impact Research 2014. Luigi Folco and Massimo D'Orazio are also supported by the University of Pisa Fondi di Ateneo. Institut des Sciences de la Terre (ISTerre) of Grenoble is part of Labex OSUG@2020 (ANR10 LABX56). Astrid Kowitz and John G. Spray are thanked for constructive reviews, and Uwe Reimold for editorial handling. The authors are grateful to Valentina Batanova, Andrea Cavallo and Cristian Biagioni for their assistance during electron microprobe analysis at ISTerre (Grenoble), field emission scanning electron microscope analysis at Istituto di Geofisica e Vulcanologia (Rome), and X-ray powder diffraction analysis at Pisa University’s Dipartimento di Scienze della Terra, respectively. Raman facilities by the Center ‘G. Scansetti’ for studies on asbestos and other toxic particulates, were funded by the Compagnia di San Paolo, Torino. PNRA is acknowledged for preliminary Raman microspectroscopy analyses.

82 REFERECES

Baratoux D. and Melosh H. J. 2003. The formation of shatter cones by shock wave interference during impacting. Earth and Planetary Science Letters 216:43-54. Bierman P. R. and Caffee M. 2001. Slow rates of rock surface erosion and sediment production across the Namib Desert and escarpments, southern Africa. American Journal of Science 301:326-258. Bunch T. E. and Cassidy W. A. 1972. Petrographic and electron microprobe study of the Monturaqui impactite. Contributions to Mineralogy and Petrology 36:95-112. Bunch T. E. and Cohen A. J. 1964. Shock deformation of quartz from two meteorite craters. Geological Society of America Bulletin 75:12:1263-1266. Chao E. C. T., Fahey J. J., and Littler J. 1961. Coesite from Wabar Crater, near Al Hadida, Arabia. Science 133:882-883. Dietz R. S. 1968. Shatter cones in cryptoexplosion structures. In Shock metamorphism of natural materials, edited by French B. M. and Short N. M. Baltimore, Maryland: Mono Book Corp., USA. pp. 267-285. D’Orazio M., Folco L., Zeoli A., and Cordier C. 2011. Gebel Kamil: The iron meteorite that formed the Kamil crater (Egypt). Meteoritics & Planetary Science 46:8:1179-1196. Earth Impact Database. http://www.passc.net/EarthImpactDatabase/index.html. Accessed March 14, 2014. Ebert M., Hecht L., Deutsch A., and Kenkmann T. 2013. Chemical modification of the projectile residues and target material in a MEMIN cratering experiment. Meteoritics & Planetary Science 48:134-149. Fackelman S. P., Morrow J. R., Koeberl C., and McElvain T. H. 2008. Shatter cone and microscopic shock-alteration evidence for a post-Paleoproterozoic terrestrial impact structure near Santa Fe, New Mexico, USA. Earth and Planetary Science Letters 270:290-299. Folco L., Di Martino M., El Barkooky A., D’Orazio M., Lethy A., Urbini S., Nicolosi I., Hafez M., Cordier C., van Ginneken M., Zeoli A., Radwan A. M., El Khrepy S., El Gabry M., Gomaa M., Barakat A. A., Serra R., and El Sharkawi M. 2011. Kamil Crater (Egypt): Ground truth for small-scale meteorite impacts on Earth. Geology 39:179-182. Folco L., Di Martino M., El Barkooky A., D’Orazio M., Lethy A., Urbini S., Nicolosi I., Hafez M., Cordier C., van Ginneken M., Zeoli A., Radwan A. M., El Khrepy S., El

83 Gabry M., Gomaa M., Barakat A. A., Serra R., and El Sharkawi M. 2010. The Kamil Crater in Egypt. Science 329:804. French B. M. 1998. Traces of catastrophe: A handbook of shock- Metamorphic effects in terrestrial meteorite impact structures. Houston, Texas, USA. LPI Contribution No. 954, Lunar and Planetary Institute. 120 p. Frezzotti M. L., Tecce F., and Casagli A. 2012. Raman spectroscopy for fluid inclusion analysis. Journal of Geochemical Exploration 112:1-20. Gay N. C. 1978. The composition of spherules and other features on shatter cone surfaces from the Vredefort structure, South Africa. Earth and Planetary Science Letters 41:3:372-380. Gibbons R. V., Hörz F., Thompson T. D., and Brownlee D. E. 1976. Metal spherules in Wabar, Monturaqui and Henbury impactites. Proceedings of the 7th Lunar and Planetary Science Conference pp. 863-880. Gibson H. M. and Spray J. G. 1998. Shock-induced melting and vaporization of shatter cone surfaces: Evidence from the Sudbury impact structure. Meteoritics & Planetary Science 33:2:329-336. Grady D. E. 1980. Shock deformation of brittle solids. Journal of Geophysical Research 85:B2:913-924. Grieve R. A. F. and Therriault A. M. 1995. Planar deformation features in quartz: target effects (abstract). 26th Lunar and Planetary Science Conference 515-516. Grieve R. A. F., Langenhorst F., and Stöffler D. 1996. Shock metamorphism of quartz in nature and experiment: II. Significance in geoscience. Meteoritics & Planetary Science 31:6-35. Güldemeister N., Wünnemann K., Durr N., and Hiermaier S. 2013. Propagation of impact- induced shock waves in porous sandstone using mesoscale modelling. Meteoritics & Planetary Science 48:1:115-133. Gurov E. P. and Gurova E. P. 1998. The group of Macha craters in western Yakutia. Planetary and Space Science 46:2/3:323-328.

Hemley R. J. 1987. Pressure dependence of Raman spectra of SiO2 polymorphs: α-Quartz, coesite, and stishovite. In High-Pressure Research in Mineral Physics, edited by Manghnani M. H. and Syono Y. Tokio, Japan. Terra Scientific Publishing Company and Washington, D. C. American Geophysical Union pp. 347-359. Hodge P. W. and Wright F. W. 1970. Meteoritic spherules in the soil surrounding terrestrial impact craters. Nature 255:717-718.

84 Hörz F., See T. H., Murali A. V., and Blanchard D. P. 1989. Heterogeneous dissemination of projectile materials in the impact melts from Wabar Crater, Saudi Arabia. Proceedings of the 19th Lunar and Planetary Science Conference pp. 697-709. Huber M. S., Ferrière L., Losiak A., and Koeberl C. 2011. ANIE: A mathematical algorithm for automated indexing of planar deformation features in quartz grains. Meteoritics & Planetary Science 46:9:1418-1424. Kenkmann T., Artemieva N. A., Wünnemann K., Poelchau M. H., Elbeshausen D., and Núñez del Prado H. 2009. The Carancas meteorite impact crater, Peru: Geologic surveying and modeling of crater formation and atmospheric passage. Meteoritics & Planetary Science 44:7:985-1000. Kieffer S. W. 1971. Shock metamorphism of the Coconino Sandstone at Meteor Crater, Arizona. Journal of Geophysical Research 76:5449-5473. Kieffer S. W., Phakey P.P., and Christie J. M. 1976. Shock processes in porous quartzite: Transmission electron microscope observations and theory. Contribution to Mineralogy and Petrology 59:41-93. Koeberl C., Reimold W. U., and Shirey S. B. 1998. The Aouelloul crater, Mauritania: On the problem of confirming the impact origin of a small crater. Meteoritics & Planetary Science 33:513-517. Kofman R. S., Herd C. D. K., and Froese D. G. 2010. The Whitecourt meteorite impact crater, Alberta, Canada, Meteoritics & Planetary Science 45:1429-1445. Kowitz A., Schmitt R. T., Reimold W. U., and Hornemann U. 2013a. The first MEMIN shock recovery experiments at low shock pressure (5-12 GPa) with dry, porous sandstone. Meteoritics & Planetary Science 48:1:99-114. Kowitz A., Güldemeister N., Reimold W. U., Schmitt R. T. and Wünnemann K., 2013b.

Diaplectic quartz glass and SiO2 melt experimentally generated at only 5 GPa shock pressure in porous sandstone: Laboratory observations and meso-scale numerical modeling. Earth and Planetary Science Letters 384:17-26. Langenhorst F. 2002. Shock metamorphism of some minerals: Basic introduction and microstructural observations. Bulletin of Czech Geologicsl Survey 77:265-282. Langenhorst F. 1994. Shock experiments on pre-heated α- and β- quartz: II. X-ray and TEM investigations. Earth and Planetary Science Letters 128:683-698. Mark K. 1995. Meteorite craters. Tucson, Arizona, USA: The University of Arizona Press. 288 p.

85 McHone J. F. jr., Lambert P., Dietz R. S., and Briedj M. 1990. Impact structures in Algeria (abstract). Meteoritics 15:331-332. Melosh H. J. 2013. The contact and compression stage of impact cratering. In Impact Cratering, edited by Osinsky G. R. and Pierazzo E., UK-USA: Wiley-Blackwell. pp. 336. Mittlefehldt D. W., Hörz F., See T. H., Scott E. R. D., and Mertzman S. A. 2005. Geochemistry of target rocks, impact-melt particles, and metallic spherules from Meteor Crater, Arizona: Empirical evidence on the impact process. In Large Meteorite Impacts III edited by Kenkmann T., Hörz F., and Deutsch A. Boulder, Colorado. Geological Society of America Special Paper 384 pp. 367-388. Mittlefehldt D. W., See T. H., and Hörz F. 1992. Dissemination and fractionation of projectile materials in the impact melts from Wabar Crater, Saudi Arabia. Meteoritics 27:361- 370. Miura Y. 1995. New shocked quartz with high density from wolf creek impact crater (abstract). Meteoritics 30:5:551. Miura Y. and Kato T. 1991. Anomalous shocked quartz in Australian impact craters (abstract). Meteoritics 26:4:373 Nicolaysen L. O. and Reimold W. U. 1999. Vredefort shatter cones revisited. Journal of Geophysical Research 104:B3:4911-4930. Ogilvie P., Gibson R. L., Reimold W. U., Deutsch A., and Hornemann U. 2011. Experimental investigation of shock metamorphic effects in a metapelitic granulite: the importance of shock impedance contrast between components. Meteoritics & Planetary Science 46:10:1565-1586. Öhman T. and Preeden U. 2013. Shock metamorphic features in quartz grains from the Saarijärvi and Söderfjärden impact structures Finland. Meteoritics & Planetary Science 48:6:955-975. Personen L. J., Elo S., Lehtinen T., Jokinen T., Puranen R., and Kivekäs L. 1999. Lake Karikkoselkä impact structure, central Finland: New geophysical and petrographic results. In Large Meteorite Impacts and Planetary Evolution II, edited by Dressler B.O. and Sharpton V. L. Boulder, Colorado. Geological Society of America Special Paper 339 pp. 131-147. Reimold W. U. and Koeberl C. 2014. Impact structures in Africa: a review. Journal of African Earth Sciences 93:57-175.

86 Reimold W. U., Koeberl C., and Reddering J. S. V. 1998. The 1992 drill core from the Kalkkop impact crater, Eastern Cape Province, South Africa: Stratigraphy, petrography, geochemistry and age. Journal of African Earth Sciences 26:4:573-592. Robertson P. B. 1980. Anomalous development of planar deformation features in shocked quartz of porous lithologies (abstract). 11th Lunar and Planetary Science Conference 938-940. Sagy A., Reches Z., and Fineberg J. 2002. Dynamic fracture by large extraterrestrial impacts as the origin of shatter cones. Nature 418:310-313. Sagy A., Fineberg J., and Reches Z. 2004. Shatter cones: Branched, rapid fractures formed by shock impact. Journal of Geophysical Reserach 109:B10209:1-20 Schade S. and Wünnemann K. 2007. Numerical modelling of pore space collapse due to shock wave compression (abstract #2029). 38th Lunar and Planetary Science Conference. Smith T. and Hodge P. 1997. Discovery of impactite at the Odessa meteorite Crater. Meteoritics & Planetary Science 32:S4:A122. Smith T. and Hodge P. 1993. Microscopic meteoritic material surrounding meteorite craters (abstract). Meteoritics 28:3:439. Spray J. G. 2010. Frictional melting processes in planetary materials: From hypervelocity impact to earthquakes. Annual Review of Earth and Planetary Science 38:221-254. Stöffler D. 1971. Progressive metamorphism and classification of shocked and brecciated crystalline rocks at impact craters. Journal of Geophysical Research 76:5541-5551. Stöffler D. and Langenhorst F. 1994. Shock metamorphism of quartz in nature and experiment: I. Basic observation and theory. Meteoritics 29:155-181. Tancredi G., Ishitsuka J., Schultz P. H., Harris R. S., Brown P., ReVelle D. O., Antier K., Le Pichon A., Rosales D., Vidal E., Varela M. E., Sánchez L., Benavente S., Bojorquez J., Cabezas D., and Dalmau A. 2009. A meteorite crater on Earth formed on September 15, 2007: The Carancas hypervelocity impact. Meteoritics & Planetary Science 44:12:1967-1984. Taylor S. R. 1967. Composition of meteorite impact glass across the Henbury strewnfield. Geochimica et Cosmochimica Acta 31:961-968. Urbini S., Nicolosi I., Zeoli A., El Khrepy S., Lethy A., Hafez M., El Gabry M., El Barkooky A., Barakat A., Gomaa M., Randwan A. M., El Sharkawi M., D’Orazio M., and Folco L. 2012. Geological and geophysical investigation of Kamil Crater, Egypt. Meteoritics & Planetary Science 47:1842-1868.

87 Wieland F., Reimold W. U. and Gibson, R.L. 2006. New observations on shatter cones in the Vredefort impact structure, South Africa, and evaluation of current hypotheses for shatter cone formation. Meteoritics and Planetary Science 41:1737–1759. Wünnemann K., Collins G. S., and Melosh H. J. 2006. A strain-based porosity model for use in hydrocode simulations of impacts and implications for transient crater growth in porous targets. Icarus 180:514-527. Wünnemann K., Collins G. S., and Osinski G. R. 2008. Numerical modelling of impact melt production in porous rocks. Earth and Planetary Science Letters 269:530-539.

88 3. TARGET-PROJECTILE INTERACTION DURING IMPACT MELTING AT KAMIL CRATER, EGYPT

1* 1 2,3 1 Agnese FAZIO , Massimo D’ORAZIO , Carole CORDIER , and Luigi FOLCO 1Dipartimento di Scienze della Terra, Università di Pisa, Via S. Maria 53, 56126 Pisa, Italy 2Université de Grenoble Alpes, ISTerre, BP 53, 38041 Grenoble CEDEX 9, France 3CNRS, ISTerre, BP 53, 38041 Grenoble CEDEX 9, France

Abstract – In small meteorite impacts, the projectile may survive the impact through fragmentation; in addition it may melt and chemically and physically interact with both shocked and melted target rocks. However, the mixing between projectile and target melts is a process that is still not completely understood. Kamil Crater (45 m in diameter; Egypt), generated by the hypervelocity impact of the Gebel Kamil Ni-rich ataxite on sandstone target, allows to study the target-projectile interaction in a very simple and fresh geological setting. We conducted a petrographic and geochemical study of macroscopic impact melt lapilli and bombs ejected from the crater, which were collected during our geophysical campaign in February 2010. Two types of glasses constitute the impact melt lapilli and bombs: a white glass and a dark glass. The white glass is inclusion-free, it is mostly made of SiO2, and it has negligible amount of Ni and Co, suggesting derivation from the target rocks with negligible interaction with the projectile melt (<0.1 wt% of projectile contamination). The dark glass is a silicate melt with variable contents of Al2O3 (0.84-18.7 wt%), FeOT (1.83-61.5 wt%) and NiO (<0.01-10.2 wt%). The dark glass typically includes fragments (from few µm to several mm in size) of shocked sandstone, diaplectic glass, lechatelierite, and Ni-Fe metal blebs. The metal blebs are enriched in Ni compared to iron meteorite Gebel Kamil. The dark glass is thus a mixture of target and projectile melts (11-12 wt% of projectile contamination). Based on recently proposed models for target-projectile interaction and for impact glass formation, we suggest a scenario for the glass formation at Kamil. Between the contact and compression stage and the excavation stage, projectile and target liquids can chemically interact in a restricted zone. The projectile contamination affected only a shallow portion of the impacted target rocks. White glass formed out of this zone, escaping interaction with the projectile. During the excavation stage, due to a brief and chaotic time sequence and the high temperature, dark glass engulfed and coated white glass and target fragments and stuck on iron meteorite shrapnel fragments.

89 INTRODUCTION Impact melting is a common process occurring during meteorite impacts on solid bodies of the Solar System. On Earth, evidences of this process have been found as impact melt rocks (kilometer- to centimeter-scale) forming layered bodies in the crater or in the ejecta blanket, as dykes, veins, and vein networks (kilometer to centimeter-scale) in the crater basement, or as centimeter- to micrometer-sized masses and spherules in distal strewn fields called tektites and microtektites (e.g., Stöffler and Grieve, 2007; Osinski et al. 2013). Most impact melt rocks have been produced by large meteoritic impacts of projectiles more than tens of meters in diameter or larger. Impact melt rocks reflect the chemical composition of the target rocks; indeed the contribution of the projectile is typically lower than 1 vol%, because the projectile is entirely vaporized during large meteoritic impacts. Thus, the nature of the projectile can be detected only through the analysis of few geochemical fingerprints such as the distribution of siderophile elements or the isotope composition of osmium and chromium (e.g., Koeberl et al. 2012; Goderis et al. 2013). During small impact events (final crater diameter < 1.5 km) impact melt rocks are formed by the melting of both the target and the projectile; the projectile survives to the impact and its partially melted and fragments can be found in the proximity of the crater (e.g., Wabar (largest crater 116 m in diameter), Kamil (45 m in diameter), Henbury (157 m in diameter), Aouelloul (390 m), Barringer (1.2 km in diameter)). Nineteen out of 27 small impact craters were generated by the impact of iron or stony-iron projectiles (Earth Impact Database, accessed on December 2014). Target rocks and iron projectiles have extremely different chemical compositions, thereby facilitating our understanding of the mixing-mingling processes between projectile and target. In recent years, new steps forward in this field have been done thanks to new experiments carried out in the framework of the MEMIN (Multidisciplinary Experimental and Modeling Impact Research Network) program (Ebert et al. 2013 and 2014) and thanks to new evidence from natural impactites (e.g., Wabar crater; Hamann et al. 2013).

In this paper we present the results of the petrographic and geochemical study of impact melt lapilli and bombs from Kamil Crater (southwester Egypt; Folco et al. 2010; 2011). Impact melt lapilli and bombs are made by two types of glasses. These glasses have different petrographic features (glass color, type and number of inclusions) and show a bimodal distribution of the projectile contamination: white glass (white in color, almost inclusion free, and < 0.1 wt% of projectile contamination) and dark glass (grayish to greenish in color, inclusion rich, and 11-12 wt% of projectile contamination). These results will be

90 used to constrain the mechanism of formation of impact melt rocks in small terrestrial impact craters.

BACKGROUND

Kamil is a small impact structure (45 m in diameter) in southwestern Egypt (22°01’06’’N, 26°05’16’’E). It was discovered in 2008 (Folco et al. 2010 and 2011). In 2010, an Italian-Egyptian geophysical campaign was organized with the aim to conduct a geological and geophysical survey of the crater, a systematic collection of macroscopic meteorite specimens, a systematic collection of microscopic impactor debris and an explorative sampling of macroscopic ejecta, including impact melt lapilli and bombs and target blocks. The results of the geological and geophysical survey are reported in Urbini et al. (2012). The petrography and the geochemistry of the iron meteorite Gebel Kamil are discussed in D’Orazio et al. (2011). The main petrographic and geochemical features of shocked and melted ejecta are presented in Fazio et al. (2014). The study of the microscopic impactor debris is reported in Folco et al. (2015). In the following paragraphs we summarize the main features of the target rocks, of the iron projectile, and of the ejecta.

Target rocks Kamil Crater is located in a rocky desert area in the East Uweinat district in southwestern Egypt. The impact of Gebel Kamil affected only the sedimentary rocks of the Gilf Kebir Formation, without involving the Precambrian crystalline basement, cropping out due northwest of the crater (> 150 m). In the crater area, the Gilf Kebir Formation consists of sandstone with subhorizontal bedding consisting of reddish brown rocks topped by a meter- thick level of pale rocks (Urbini et al. 2012).

Pale and reddish brown rocks are both made of up to 98 vol% quartz; Fe-Ti oxides, tourmaline and zircon are the most common accessory phases. Pale rocks are coarse quartzarenite to medium-fine-grained wacke with siltstone levels. The matrix is mainly composed by kaolinite (Table 1), and its abundance increases with decreasing grain-size, ranging from negligible contents in coarse quartzarenite up to 40 vol% in medium-fine- grained wacke. Porosity is usually lower than 4 vol%. In coarse quartzarenite rocks, pores are filled by an extensive overgrowth of syntaxial quartz cement. Reddish rocks are very coarse- to coarse-grained, gritty and ferruginous quartzarenite. Iron oxides and hydroxides are important components of the matrix together with kaolinite (Table 1). The porosity of reddish quartzarenites is generally higher than that of pale rocks (up to 24 vol%). Target rocks from Kamil area show values of loss on ignition (L.O.I.) ranging from 0.65 and 3.91. These

91 contents are mainly attributable to the presence of H2O from the kaolinite of the matrix. The contribution of other volatile compounds is negligible due to the mineral composition of the target rocks.

Table 1. Chemical composition of target rocks at Kamil Crater. Major elements (wt%) were determined by X-ray fluorescence analyses, except where otherwise specified. Trace elements (ppm) were determined by ICP-MS. Modified after Folco et al. (2015).

Pale quartzarenite, Reddish Pale wacke with Kaolinite-rich matrix negligible matrix quartzarenite, ~ 5 siltstone levels (n=18)a contents vol.% matrix

SiO2 99.8 87.8 95.7 44.5

TiO2 0.07 1.15 0.09 0.55

Al2O3 0.25 10.1 1.98 34.8

FeO T 0.10 0.66 1.54 1.18 MnO <0.01 <0.02 0.06 <0.01 MgO 0.18 0.14 0.25 0.06 CaO 0.05 0.11 0.31 0.04

Na2O 0.05 0.13 0.08 0.04

K2O <0.01 0.04 <0.01 0.05

P2O5 <0.01 0.06 0.04 0.09 sum 100.5 100.2 100.1 81.3b L.O.I. 0.65 3.91 1.19

Trace elements Co 0.5 2.1 10 Ni 6.9 7.3 13 Cu 5.2 8.9 15 V 3.0 53 32 Cr 20 57 9.5

Ni/Fe 0.009 0.001 0.001 Ni/Co 13.2 3.51 1.19

a Average of EPMA analyses. b The low total of the average of the EPM analyses of kaolinite is assumed to be due to the water content of this mineral (~19 wt%).

FeOT: total iron as FeO. Abbreviation: L.O.I.: Loss On Ignition. Projectile

Kamil was formed by the hypervelocity impact of the iron meteorite Gebel Kamil. Gebel Kamil is an ungrouped ataxite (Ni = 20.6 wt%) characterized by a very fine-grained duplex plessite metal matrix (D'Orazio et al. 2011). Accessory minerals are schreibersite, troilite, daubréelite and native copper, in order of decreasing abundance.

Thousands of meteorite fragments were found around the crater. They are all shrapnel in the <1 g to 34 kg mass range with the exception of a single regmaglypted individual of 83 kg.

92 The systematic visual search of specimens > 10 g (D'Orazio et al. 2011) and the geomagnetic survey (buried shrapnel > 100 g; Urbini et al. 2012) indicate that the minimum projectile mass is close to 5 t. However, the total mass of the Gebel Kamil meteorite is close to 10 t taking into account the estimate of the mass of shrapnel < 10 g, buried shrapnel < 100 g, and impactor debris fraction (Folco et al. 2015). This value is in agreement with estimates of its pre-atmospheric mass, namely >20 t, most likely 50-60 t (Ott et al. 2014). The absence of companion craters suggests that Gebel Kamil meteorite impacted the ground as a single mass or a very tight cluster of fragments, and that it underwent only minor fragmentation and separation during atmospheric flight (Folco et al. 2010 and 2011).

Ejecta

A radial pattern of ejecta characterizes the surrounding of Kamil Crater (Urbini et al. 2012). It is formed by meteorite shrapnel fragments, shocked and unshocked sandstone blocks, and impact melt lapilli and bombs. The bulk of the ejecta material is preferentially concentrated between the two main ejecta rays trending to the north and southwest, respectively, indicating that the projectile arrived from NW. Within ~ 50 m from the crater rim unshocked sandstone clasts and impact melt lapilli and bombs are more common; further away, the shocked sandstone clasts dominate.

Shocked sandstone blocks show an almost complete set of shock metamorphic features including fracturing, planar deformation features (PDFs) in quartz, high-pressure

SiO2 polymorphs, diamond, melt veins and melt in shatter cones (Fazio et al. 2014). Some of these features have never been reported before from impact craters of comparable size.

Impact glass lapilli and bombs are constituted by two types of glasses: the white glass (Figs. 1a and 1b) and the dark glass (Figs. 1c and 1d; Table 2). Dark glass is dark-gray to brown in color, magnetic, and contains several target fragments, whereas white glass is white to light gray in color, non-magnetic, and nearly clast free. White glass can be completely or partially enveloped by dark glass (Fig. 1b), whereas the reverse has never been observed. Similar relations are common also among the micrometric impact melt particles sampled in the soils around the crater (Folco et al. 2015). Dark glass fragments were also found stuck onto the meteorite shrapnel (Fig. 1d).

93

Fig. 1. Impact glasses. a) Cut surface of the white impact melt bomb L04. b) Cut surface of the white impact melt bomb L03 coated by dark glass. c) Cut surface of the dark impact melt bomb L06. The dark glass contains abundant fragments of sandstone and lechatelierite, and rare fragments of meteorite. Abbreviations: MF = meteorite fragment; SC = shocked sandstone clasts; L = lechatelierite. d) Dark glass stuck onto a meteorite shrapnel fragment (see white arrows). The side length of the scale cube is 1 cm.

94

Table 2. Location and description of impact melt lapilli and bombs from Kamil Crater (Table modified after Fazio et al., 2014).

Sample Latitude N Longitude E Location Description Ejecta: impact melt lapilli and bombs Lapillus made of dark glass with inclusions of sandstone clasts, diaplectic glass and L15 22°01'6.05" 26°05'15.77" Inside the crater lechatelierite. Metallic blebs are scattered in the dark glass.

M23 22°01'5.69" 26°05'16.27" Inside the crater Bomb made of white lechatelierite. Vesicles are rounded and generally micrometric.

Bomb made of white lechatelierite partially coated by < 2 mm thick dark glass L08 22°01'4.05" 26°05'16.53" 52 m due ENE of CR envelope. Lapillus made of white lechatelierite stained by reddish-brown material (iron-oxi- L09 22°01'4.01" 26°05'16.57" 52 m due ENE of CR hydroxides). Relicts of highly shocked quartz grains (melted PDFs, amorphization) are common in this sample. Lapillus made of dark glass with inclusions of sandstone clasts, diaplectic glass and L05 22°01'5.29" 26°05'18.17" 57 m due E of CR lechatelierite. Metallic blebs are scattered in the dark glass. Bomb made of dark glass with inclusions of sandstone clasts, diaplectic glass and L06 22°01'5.27" 26°05'18.14" 57 m due E of CR lechatelierite. Metallic blebs are scattered in the dark glass. Rare occurrence of fragments of the iron projectile Gebel Kamil. Dark glass found stuck to the surface of a shrapnel of the iron meteorite Gebel Kamil. E30m.s. 22°01'3.39" 26°05'16.44" 57 m due ESE of CR Inclusions of sandstone clasts, diaplectic glass and lechatelierite are common. Metallic blebs are scattered in the dark glass.

Dark glass found stuck to the surface of a shrapnel of the iron meteorite Gebel Kamil. MNA07m.s. 22°01'4.51" 26°05'18.11" 61 m due SSE of CR Inclusions of sandstone clasts, diaplectic glass and lechatelierite are common. Metallic blebs are scattered in the dark glass. Bomb made of white lechatelierite completely coated by < 2 mm thick dark glass L03 22°01'7.91" 26°05'19.88" 120 m due ENE of CR envelope. Bomb made of white lechatelierite. Vesicles are elongated and can be up to 3 mm in L04 22°01'7.93" 26°05'19.92" 120 m due ENE of CR size. Abbreviations: CR: crater rim; m.s.: meteorite shrapnel.

ANALYTICAL METHODS

The petrographic study of the impact melt lapilli and bombs was carried out by optical and electron microscopy observations. Optical microscope Zeiss Axioplan and Scanning Electron Microscope (SEM) Philips XL30, operating at 20 kV and coupled with an energy- dispersive X-ray fluorescence spectrometer (EDX), were used at Pisa University’s Dipartimento di Scienze della Terra (Italy). Field Emission Scanning Electron Microscope (FE-SEM) Jeol JSM 6500F (upgraded to 7000 series), operating at 10 kV, was used at the Istituto di Geofisica e Vulcanologia (INGV) of Rome (Italy).

The chemical analyses of glasses and minerals were obtained by electron microprobe. For the analyses of the dark and white glasses we used the electron microprobe Jeol JXA 8230 fitted with five wavelength dispersive spectrometers at the Institut des Sciences de la Terre (ISTerre) of Grenoble (France). Running conditions were 15 kV accelerating voltage, 12 nA beam current and 1 µm nominal beam spot. The ZAF procedure was employed for raw data reduction. Standards used for instrumental calibration were SiO2-rich glass (USNM 72854), ilmenite (USNM 96189), and hornblende (USNM 143965).

For the analyses of the Ni-Fe metal blebs embedded in glass we used the electron microprobe Cameca SX50 fitted with four wavelength dispersive spectrometers at the Istituto di Geoscienze e Georisorse (IGG) of the Consiglio Nazionale delle Ricerche in Padova (Italy). Running conditions were 15 kV accelerating voltage, 20 nA beam current and 1 µm nominal beam spot. The manufacturer-supplied PAP procedure was applied for raw data reduction. Standards used for instrumental calibration were natural minerals (diopside, apatite and sphalerite), and pure elements (Fe, Ni, and Co).

Whole-rock major elements of unshocked target materials and impact melt lapilli and bombs from Kamil Crater were determined on glass beads by X-Ray Fluorescence (XRF; ARL 9400 XP spectrometer) at Pisa University’s Dipartimento di Scienze della Terra (Italy), using the procedure described by Tamponi et al. (2003). The bulk chemical composition of sample L09 was obtained by Hand-Held X-Ray Fluorescence (HH-XRF) spectrometer NITON XL3t GOLDD+ at Pisa University’s Dipartimento di Scienze della Terra (Italy). Inductively coupled plasma-mass spectrometry (ICP-MS; Perkin Elmer Nexion 300X) was used to determine trace element concentrations of target material and impact melt lapilli and bombs. Analyses were carried out at Pisa University’s Dipartimento di Scienze della Terra.

97 Table 3. Chemical composition of impact melt lapilli and bombs from Kamil Crater. Major elements (wt%) were determined by X-ray fluorescence, except where otherwise specified. Trace elements (ppm) were determinated by ICP-MS.

M23 L04 L09a L03 L08 L06 White glass White glass White glass White glass White glass stained Fe-oxi- coated by dark coated by dark Dark glass hydroxides glass glass

SiO2 99.9 99.4 97.1 96.2 98.7 74.2

TiO2 0.08 0.06 0.02 0.15 0.05 0.57

Al2O3 0.46 0.69 1.12 1.76 0.64 5.80

FeO T 0.40 0.42 1.38 0.77 0.84 14.6 MnO 0.02 0.02 b.d.l. 0.03 0.02 0.18 MgO 0.22 0.43 b.d.l. 0.30 0.31 0.22 CaO 0.12 0.65 b.d.l. 0.34 0.34 0.37

Na2O 0.10 0.11 b.d.l. 0.09 0.10 0.07

K2O <0.01 0.04 0.06 0.04 0.02 0.06

P2O5 <0.01 0.02 0.18 0.03 0.02 0.22 sum 101.3 101.8 99.9 99.7 101.0 96.3 L.O.I. 0.53 1.89 0.99 1.21 -1.38

Trace elements Co 1.8 1.2 1.0 7.2 11 908 Ni 34 18 5.6 121 233 22950 Cu 12 24 30 15 21 87 V 11 11 16 16 20 227 Cr 24 43 16 18 21 357

Ni/Fe 0.011 0.005 0.001 0.020 0.035 0.203 Ni/Co 19.3 14.6 5.6 16.8 22.1 25.3 aMajor-element composition of sample L09 was obtained by Hand-Held X-Ray Fluorescence. FeOT: total iron as FeO Abbreviation: L.O.I.: Loss On Ignition. RESULTS

White glass

Impact melt lapilli and bombs made of white glass can be classified as vesicular clast- poor to clast-free impact melt rocks, according to the classification scheme proposed by Stöffler and Grieve (2007) and Osinski et al. (2013). At hand-specimen scale they generally appear white or light gray (Figs. 1a and 1b); only one sample (labeled L09) shows some reddish-brown portions. Some masses are similar to volcanic pumice (Fig. 1a), with rounded vesicles ranging from few tens of micrometers to few millimeters (Figs. 2a-c); they occupy up to ~ 30 vol% of the samples. Several, impact melt lapilli and bombs made of white glass are coated by dark glass (Fig. 1b).

The bulk chemical compositions of impact melt lapilli and bombs made of white glass characteristically contain more than 96 wt% of SiO2 (Table 3). Samples M23 and L04, that are not coated by dark glass, have less than 0.5 wt% of FeOT. Nickel and Co concentrations of

98 these samples are of the same order of magnitude of the target rocks (Tables 1 and 3). Sample

L09 has a moderate content of FeOT (1.4 wt%), but very low contents of Co and Ni (1 and 6 ppm, respectively). Sample L03 and L08, that are coated by dark glass (Figs. 1b and 2a), have significant concentrations of FeOT (~ 0.8 wt%), Ni (121 – 233 ppm) and Co (7 – 11 ppm).

In thin section, the white glass is generally colorless (Fig. 2a). Backscattered-SEM images show that the composition of the white glass is homogenous (Fig. 2b). However, the glass of the impact melt lapillus L09 is opaque because it is stained by a reddish-brown material (Fe-oxi-hydroxides; Fig. 2c). For this reason the white glasses have been divided into lechatelierite and ferruginous lechatelierite. Lechatelierite has >98 wt% on average of SiO2.

Ferruginous lechaterierite is composed dominantly by SiO2 (average 92.2 wt%) with minor contents of FeOT (average 3.5 wt%; Table 4). Lechatelierite and ferruginous lechatelierite are both characterized by low Al2O3 contents (up to 1.4 wt% in ferruginous lechatelierite) and negligible contents of NiO (Table 4).

99

Fig. 2. Photomicrographs of white impact melt lapilli and bombs. a) Detail of the white impact melt bomb L03 coated by dark glass (optical microscope planar polarized light, PPL). b) Backscattered electron (BSE)-SEM image of the white impact melt bomb L04. The white glass (lechatelierite) is chemically homogenous. Abbreviation: V = vesicle. c) Impact melt bomb L09 made of white glass stained by reddish-brown material (iron-oxi-hydroxides). Due to its relatively higher FeOT content, it is here called ferruginous lechatelierite. A former quartz grain (dominantly made of amorphous silica with crystalline domains) occurs in the upper right portion of the photomicrograph (optical microscope crossed polars, XPL).

100 Table 4. Representative electron microprobe analyses (wt%) of white glass of impact melt lapilli and bombs.

SiO2 TiO2 Al2O3 FeO T MnO NiO MgO CaO Na2O K2O P2O5 Sum

Lechatelierite L03_10 98.5 0.02 0.02 0.04 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 98.6 L03_11 99.3 b.d.l. 0.03 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 99.3 L04_02 98.6 b.d.l. 0.29 0.25 b.d.l. b.d.l. b.d.l. 0.02 0.06 0.04 b.d.l. 99.3 L04_11 98.5 b.d.l. 0.03 0.02 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 98.6 Avg. (n=8) 98.3 0.02 0.14 0.07 0.02 0.02 0.02 0.05 0.02 0.02 98.6 ±1σ 1.44 0.15 0.08 0.01 0.01 0.01 Ferruginous lechatelierite L09_03 93.5 b.d.l. 0.29 4.73 b.d.l. b.d.l. 0.03 0.03 b.d.l. b.d.l. 0.04 98.6 L09_04 95.6 b.d.l. 0.22 1.23 b.d.l. 0.02 0.14 0.02 b.d.l. b.d.l. b.d.l. 97.3 L09_07 92.3 b.d.l. 0.18 0.65 b.d.l. 0.05 0.15 0.03 0.04 b.d.l. 0.04 93.4 L09_15 87.4 0.01 0.25 9.89 b.d.l. b.d.l. b.d.l. 0.03 b.d.l. 0.03 0.01 97.7 Avg. (n=11) 92.2 0.04 0.53 3.48 0.04 0.08 0.03 0.04 0.03 0.04 96.3 ±1σ 4.11 0.02 0.86 3.30 0.01 0.05 0.01 0.02 0.02 Former quartz grains L09_01 99.2 b.d.l. 0.03 0.03 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 99.3 L09_16 99.1 b.d.l. b.d.l. 0.07 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 99.2 L09_20 98.9 b.d.l. 0.08 0.17 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 0.02 0.03 99.2 Avg. (n=8) 98.6 0.15 0.11 0.02 0.03 0.03 0.02 98.9 ±1σ 0.50 0.14 0.08 0.01 0.01 0.00 Fe-rich silicate melt L04_4 23.7 0.06 0.85 58.8 0.87 b.d.l. 3.86 5.22 b.d.l. b.d.l. 0.27 93.6 L04_7 18.9 0.02 0.28 64.5 1.19 b.d.l. 3.97 2.71 b.d.l. b.d.l. 0.10 91.6 L04_8 8.24 b.d.l. 0.36 80.3 0.05 b.d.l. 0.03 0.19 b.d.l. b.d.l. 0.32 89.5 L04_12 37.1 0.04 0.65 60.5 0.16 0.02 0.22 0.25 0.05 b.d.l. 0.38 99.4 Avg. (n=6) 21.3 0.04 0.54 65.3 0.70 2.62 2.46 0.05 0.23 93.2 ±1σ 9.35 0.02 0.23 7.82 0.48 2.02 1.94 0.11 Abbreviations: avg., average; b.d.l., below detection limit; n, number of analyses. Detection limits are 0.04 wt% for Na2O, 0.02 wt% for Al2O3, and 0.01 wt% for MgO, CaO, K2O, FeO, MnO, TiO2, and P2O5. FeOT: total iron as FeO. Former quartz grains were found in the ferruginous lechatelierite impact melt lapillus L09 (Fig. 2c; Table 4). These grains are colorless and easily recognizable from the ferruginous lechatelierite. They are dominantly made of amorphous SiO2 material (probably diaplectic glass) with birefractive domains (Figs. 3a and 3b). They contain micrometric vesicles (Fig. 3c). These domains show planar deformation features (PDFs; Fig. 3c). In Fig. 3c, at least two sets of PDFs occur, their orientations are marked by dashed black lines. In Fig. 4, for different BSE-FESEM images of Kamil PDFs showing four stages of the process of quartz amorphization are reported. The first two images are from the shocked sandstone sample L23 and show well-formed PDFs (Fig. 4a) and slightly enlarged PDFs (Fig. 4b). The second two images are from the sample L09 and show that PDFs are very enlarged and wavy (Fig. 4c), and in some portions they have almost completely substituted the crystalline quartz

101 (Fig. 4d). In Fig. 4d the process of quartz amorphization is close to the melting, as indicated by the occurrence of small vesicles. PDFs in sample L09 could, therefore, be considered the final stages of the process of quartz amorphization and melting. Due to their features, PDFs of sample L09 were not indexed with the Universal Stage. Using JMicroVision software for image analyses, we quantified the portion occupied by amorphous and melted SiO2 materials

(amorphous and melted SiO2 materials are darker than crystalline quartz, because they have a lower density) in the four images of Fig. 4. We obtained that the portion occupied by amorphous material is ~12 vol%, ~44 vol%, ~73 vol%, and ~85 vol% from Fig. 4a to Fig. 4d.

In Fig. 4d, a bright 1 µm-in-size object can be observed. EDX-SEM analysis indicates that it is composed of 100 wt% SiO2. Due to its brightness, we suppose that it is a high- pressure polymorph of SiO2. The occurrence of surrounding radial fractures propagating in the host phase indicates a transformation from a phase with a small molar volume to a phase with a higher molar volume (e.g., from stishovite to coesite). These objects are quite common in former quartz grains almost completely substituted by amorphous SiO2 in sample L09.

102

Fig. 3. Photomicrographs of a former quartz grain (dominantly made of amorphous silica with crystalline domains) within the ferruginous lechatelierite in sample L09. a) Colorless former quartz grain (PPL). The dashed line outlines a slightly darker domain. b) XPL view of the same former quartz grain in (a) Dashed line outlines the same domain in (a) to show the portion with the highest birefringence. c) Detail of a former quartz grain where al least two (dashed black lines) sets PDFs and vesicles occur (PPL). d) BSE-FESEM image of the contact area separating a former quartz grain and lechatelierite. Small elongated vesicles/cracks occur in lechatelierite. They are parallel among them, parallel to the contact, and arranged geometrically. e) A detailed view of these geometric vesicles/cracks arranged following the orientation of the PDF sets. f) BSE-FESEM image of another contact area separating a former quartz grain and lechatelierite. Former quartz grains show two different types of contacts with the surrounding lechatelierite (Figs. 3d-f). The first type is characterized by a distinct boundary marked by small elongated vesicles/cracks on the lechatelierite side (Figs. 3d and 3f). These vesicles/cracks are geometrically arranged, parallel to the contact, and their size decreases approaching the contact (Figs. 3d and 3e). Due to their geometric arrangement, these vesicles/cracks in the glass could be developed parallel to crystallographic planes of former quartz crystals. The second type of contact between the former quartz grains and the surrounding lechatelierite is represented in Fig. 3f. In this case there is not a distinct boundary

103 between quartz and lechatelierite, roundish vesicles decrease in size approaching the quartz grain (Fig. 3f).

Fig. 4. Sequence of four BSE-FESEM images showing the progressive enlargement of PDFs till the almost complete melting of the quartz (amorphous SiO2 material is darker than quartz because its density is lower). a) Well-formed PDFs in shocked sandstone sample L23; b) slightly enlarged PDFs in shocked sandstone sample L23; c) enlarged PDFs occurring in a former quartz grain of the impact melt lapillus sample L09; d) very enlarged PDFs occurring in a former quartz grain. With respect to (c), the amorphous SiO2 phase constituting the PDFs has almost completely substituted the quartz grain. The arrow points to a bright, 1 µm in size object which is probably a high-pressure phase of SiO2. Vesicles also occur. Accessory minerals occurring in the protolith of the white glass were found in the form of decomposed crystals, quenched crystals, or transformed into glass. For example, within the white impact melt bomb M23, we found a single zircon grain partially decomposed into baddeleyite (ZrO2) and SiO2 (Fig. 5a). Sample L09 contains vesicle-poor portions characterized by the occurrence of quenched skeletal crystals. One of these portions (about 100 µm x 140 µm in size) is shown in Fig. 5b. At least two different skeletal minerals occur. They are recognizable for their different shape and reflectance in BSE images. The darker mineral shows elongated zoned crystals with rhombic basal sections and generally larger sizes than the lighter mineral (Fig. 5c). The darker mineral is composed of SiO2, Al2O3, and TiO2. Liquid immiscibility texture occurs in the interstitial space between skeletal crystals.

Iron-rich Ni-free silicate glasses are found in confined portions of the white impact melt lapilli and bombs and generally show liquid immiscibility textures. Figs. 5d and 5e show elongated (up to tens of micrometers), curved, and thin (few micrometers) veinlets surrounded by finely dispersed micro- to nanoblebs. In another sample (L04), irregular masses of Fe-rich

104 silicate glass were found close to the vesicles (Fig. 5f). Electron microprobe analyses of this glass show that FeOT varies between 58.8 wt% and 80.3 wt%, SiO2 varies between 8.24 wt% and 37.1 wt% (Table 4). CaO and MgO are respectively the third and the fourth most abundant component of this melt, varying between 0.19 - 5.22 wt% and 0.03 - 3.97 wt%, respectively (Table 4).

Fig. 5. Accessory minerals in the form of decomposed crystals, quenched crystals, or glass within the white glass. a) BSE-SEM image of a partially decomposed zircon embedded in lechatelierite in sample M23. b) BSE- FESEM image of a 100 x 140 µm area within ferruginous lechatelierite of the sample L09 made of quenched skeletal crystals (Al-Ti-rich). c) Close-up view of the rectangular area highlighted in (b) showing a detail of the quenched skeletal crystals. d) BSE-FESEM image of curved needles and microblebs of Fe-rich Ni-free silicate melt in sample L08. e) Close-up view of the rectangular area highlighted in (d) showing microblebs of Fe-rich Ni-free silicate melt at the end of curved veinlet. Note the fractal distribution of nano-blebs around the bleb and the veinlet. f) BSE-SEM image of Fe-Ca-Mg-rich Ni-free silicate melt around a vesicle of the sample L04. In this picture, Fe-Ca-Mg-rich Ni-free silicate melt shows irregular shapes.

105 Dark glass

Impact melt lapilli and bombs made of dark grayish to greenish glass are vesicular clast-rich impact melt rocks with scoriaceous textures (Fig. 1c), according to the classification scheme proposed by Stöffler and Grieve (2007) and Osinski et al. (2013). The dark glass occurs as isolated masses, as well as coatings of the white glass impact melt lapilli and bombs (Figs. 1b and 2a), and as small masses stuck on the surfaces of iron meteorite fragments (Fig. 1d). Among the impact melt lapilli and bombs from Kamil, dark glass sample L06 has higher

concentrations in FeOT (14.6 wt%), Al2O3 (5.8 wt%), Ni (2.3 wt%), and Co (908 ppm) than white glass (Table 3).

In thin section, the dark glass is from old-yellow to brown in color and rarely opaque (Figs. 2a and 6a). Vesicles are commonly micrometric and roundish, however irregular vesicles up to 5 mm also occur. Dark glass contains abundant variably shocked target fragments and projectile materials mainly in form of Ni-Fe metal blebs (Figs. 1c, 6a and 6b).

Target fragments are, in order of abundance, clasts up to 5 mm in size of shocked

sandstone, clasts up to 1 mm in size of diaplectic glass and/or poorly-vesicular SiO2 glass, and clasts up to 3 mm in size of vesicular white glass (lechatelierite; Figs. 6a and 6b). Target inclusions are generally monotypic clasts.

Fig. 6 (next page). Photomicrograph of sample L06, dark glass impact melt bomb. a) PPL image showing the main optical features of the dark glass: old-yellow to opaque glass; fragments of shocked sandstone (SC), lechatelierite (L), and diaplectic glass (D); Ni-Fe metal blebs (white arrows). b) BSE-SEM image of the same area in (a). Two white rectangles indicate the areas of the close-up views of Figs. 6d and 6f. c) Detail of structure of the diaplectic glass showing a small brighter circular object surrounded by radiating fractures similar to that shown in Fig. 3d. It is probably a high-pressure phase of SiO2. d) Close-up BSE-FESEM view of a white rectangular area in (b) showing the features of the dark glass between a lechatelierite and a diaplectic glass. The margins are irregular, and the dark glass appears brighter at the contact with the target fragments. The white rectangle indicates the area of the close-up view of Fig. 6e. e) Close-up BSE-FESEM view of a white rectangular area in (d) showing the microstructures of the diaplectic glass at the contact with the dark glass. f) Close-up BSE-FESEM view of the other white rectangular area in (b) showing another view of the contact area between lechatelierite and diaplectic glass. Two white rectangles indicate the areas of the close-up views of Fig. 6g and 6h. g) Close-up BSE-FESEM view of a white rectangular area in (f) showing a SiO2 phase with lobated margins at the contact with dark glass. On the right side of the image, the dark glass shows an immiscibility texture. A micrometric Ni-Fe bleb is also shown in this image; it is indicated by a white arrow. h) Close-up BSE-FESEM view of the other white rectangular area in (f) showing the immiscibility texture of the dark glass (brighter) within the lechatelierite (dark gray). A sub-micrometric Ni-Fe bleb occurring in the dark glass is indicated by a white arrow.

106

Shocked sandstone clasts are fractured quartzarenite or quartzarenite with PDFs. Quartzarenite grain size never exceeds 700 µm. In shocked sandstone clasts with PDFs, the most common crystallographic orientations of the PDFs are {} 1013 and {} 1122 amounting to

17%, followed by {} 1011 , 15%, indicating formation under shock pressures between 10 and 20 GPa (Fazio et al. 2014; Figs. 7a and 7b). Intergranular glass occurs in the interstices of € € shocked quartz grains. This glass is similar to that described in highly shocked sandstone €

107 ejecta in Fazio et al. (2014). The chemical composition of shocked quartz grains is reported in Table 5.

Fig. 7. Detail images of opaque glass of sample L06. The opaque glass (dark glass NiO-rich) is localized either around target fragments (a and b) or around vesicles (d and e). a) PPL photomicrograph of a shocked sandstone clast coated by opaque glass. b) BSE-FESEM image of the contact between the shocked sandstone clast and the opaque glass. Shocked sandstone grains show PDF and are set in an intergranular melt. Near the contact, the intergranular melt and the Ni-rich glass are partially mixed. c) BSE-FESEM detail of the PDFs of a quartz grain of a sandstone clast. d) PPL photomicrograph of opaque glass around a vesicle. Opaque glass occurs also in the upper right margin of this photomicrograph around smaller vesicles. In this photomicrograph the color variation of the dark glass are better represented. e) BSE-FESEM image showing the features of the opaque border. It is made by a thin (up to 5 µm) homogeneous NiO-rich border and by a zone with dendritic minerals (probably magnetite). Small fragments of SiO2 phases occur, they are probably made of diaplectic glass because of the absence of vesicles and PDFs.

Several target inclusions are roundish, made of SiO2, vesicle-poor, low relief, always (or nearly so) extinct under crossed polars, and devoid of any particular internal structure, we suppose that they are made of diaplectic glass. The clast marked with “D” in Figs. 6a and 6b is the largest example of this type of target inclusions. It shows neither vesicles nor schlieren textures. At high magnification it looks homogenous, but by increasing the magnification, bright micrometric objects made of SiO2 can be detected (Fig. 6c). These objects are similar to those described in the former quartz grains of the sample L09 (previous section; Fig. 4d).

They also show radial fractures. It could be a high-pressure SiO2 phase. Despite having slightly higher FeOT and NiO contents, the lechatelierite fragments and the shocked quartz grains show the same petrographic features of white glass and shocked sandstone ejecta

108 (Tables 4 and 5; Figs. 2a, 2b, 6a, and 6b; Fazio et al. 2014 for shocked sandstone ejecta features). Although target inclusions display shock features typical of different shock pressure regimes, they are found very close one to another and they do not show any preferential distribution in the dark glass (Figs. 6a and 6b).

Table 5. Representative electron microprobe analyses (wt%) of shocked quartz and dark glass of impact melt lapilli and bombs.

SiO2 TiO2 Al2O3 FeO T MnO NiO MgO CaO Na2O K2O P2O5 Sum Shocked quartz L06_34 99.8 0.04 b.d.l. 0.04 0.19 0.02 b.d.l. b.d.l. 0.05 b.d.l. b.d.l. 100.1 L06_35 100.6 b.d.l. b.d.l. 0.03 0.04 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 100.6 L06_113 99.4 b.d.l. b.d.l. b.d.l. 0.26 0.08 b.d.l. 0.04 b.d.l. b.d.l. b.d.l. 99.8 Avg. (n=5) 99.5 0.04 0.05 0.04 0.19 0.04 0.03 0.03 0.05 99.8 ±1σ 0.88 0.01 0.01 0.15 0.03 0.01 Lechatelierite inclusions L06_38 98.5 b.d.l. 0.03 0.05 0.02 b.d.l. 0.03 b.d.l. b.d.l. 0.02 b.d.l. 98.6 L06_51 99.4 b.d.l. b.d.l. b.d.l. 0.15 0.04 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 99.6 L06_130 99.1 b.d.l. b.d.l. 0.03 b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. b.d.l. 99.2 L06_135 100.0 0.06 b.d.l. b.d.l. 0.07 0.03 b.d.l. 0.02 b.d.l. b.d.l. b.d.l. 100.1 Avg. (n=25) 99.1 0.04 0.06 0.04 0.19 0.03 0.02 0.02 0.02 99.4 ±1σ 0.47 0.02 0.07 0.02 0.14 0.01 0.01 Dark glass sample L06 L06_074 90.0 0.15 2.74 6.23 0.06 0.05 0.05 0.09 b.d.l. 0.07 b.d.l. 99.4 L06_108 96.2 0.44 0.84 1.83 0.04 0.09 b.d.l. 0.04 b.d.l. 0.00 b.d.l. 99.5 L06_057 79.9 0.35 3.46 14.4 0.08 0.58 0.04 0.13 b.d.l. 0.02 0.13 99.0 L06_095 73.8 0.84 18.7 5.91 0.02 0.61 0.08 0.06 b.d.l. 0.06 0.02 100.1 L06_009 51.0 1.17 14.2 30.3 0.67 0.08 0.27 1.01 b.d.l. 0.18 0.07 99.0 L06_037 49.8 1.09 12.4 32.2 0.66 2.51 0.17 0.76 0.05 0.11 0.18 100.0 L06_063 48.1 1.07 11.0 32.4 0.28 5.77 0.16 0.85 b.d.l. 0.10 0.26 100.1 L06_012 37.1 0.94 8.43 46.0 0.23 2.33 0.09 0.24 b.d.l. 0.13 0.86 96.4 L06_019 30.0 0.83 9.76 41.1 0.70 1.26 0.23 1.08 0.06 0.12 15.5 100.6 L06_062 11.9 0.44 4.85 61.5 1.29 10.2 0.38 1.62 0.14 0.10 0.05 92.5 Avg. (n=102) 66.9 0.70 7.84 21.3 0.39 0.7 0.14 0.54 0.08 0.13 0.41 98.8 ±1σ 22.2 0.49 5.33 15.0 0.28 1.5 0.09 0.43 0.03 0.09 1.91 Dark glass sample L03 L03_007 91.6 0.15 3.88 2.4 0.03 b.d.l. b.d.l. 0.02 b.d.l. 0.02 0.03 98.2 L03_012 84.0 0.40 6.19 7.6 0.12 b.d.l. 0.08 0.11 b.d.l. 0.20 0.20 98.9 L03_018 76.3 0.46 9.59 12.3 0.43 b.d.l. 0.22 0.24 0.04 0.07 0.19 99.9 L03_010 70.1 0.57 12.6 15.6 0.64 b.d.l. 0.29 0.42 b.d.l. 0.09 0.18 100.5 L03_013 63.3 0.63 15.2 19.3 0.36 b.d.l. 0.29 0.23 0.05 0.10 0.21 99.6 Avg. (n=22) 79.0 0.46 8.50 9.8 0.21 0.16 0.17 0.06 0.13 0.14 98.5 ±1σ 12.1 0.20 5.24 6.8 0.18 0.11 0.14 0.02 0.10 0.08 Abbreviations: avg., average; b.d.l., below detection limit; n, number of analyses. Detection limits are 0.04 wt% for Na2O, 0.02 wt% for Al2O3, and 0.01 wt% for MgO, CaO, K2O, FeO, MnO, TiO2, and P2O5. FeOT: total iron as FeO. Projectile materials within the dark glass are represented by common Ni-Fe metal blebs and rare fragments, up to 200 µm in size, of the Gebel Kamil iron meteorite (Figs. 1c, 6a, 6b, and 8). Nickel-Fe metal blebs can reach the maximum size of 200 µm and show

109 fractal size distribution: by reducing the scale of investigation, increasingly smaller Ni-Fe metal blebs appear (see for example the small blebs in Figs. 6g and 6h). Blebs larger than 10 µm are nearly spherical and are made of rounded Ni-Fe metal particles (cellular intergrowth texture). The interstitial space between the metal particles is filled by P-rich (darker) and S- rich (lighter) materials (Fig. 8). Often P-rich material forms a rind separating the metal blebs from the enclosing glass (white arrows in Fig. 8a). The chemical composition of Ni-Fe metal blebs of sample L06 is quite homogenous: Ni is between 60.3 wt% and 69.8 wt% (the percent Relative Standard Deviation, RSD% = 6%), Fe is between 28 wt% and 36 wt% (RSD% = 11%) and Co is between 1.56 and 1.95 wt% (RDS% = 8%; Table 6; Fig. 9). Ni-Fe metal blebs do not show variations of the Ni content as function of their diameter (Table 6). In Fig. 9 are plotted the analyses of Ni-Fe metal blebs from the dark glass of eleven microparticles collected from the soil surrounding Kamil (Folco et al. 2015). Most of them have lower Ni content than metal blebs in sample L06.

Fig. 8. Ni-Fe metal blebs in dark glass of sample L06. a) BSE-FESEM image of a bleb, the brightness was reduced to check the presence of internal structures. The bleb has a cellular intergrowth texture, made of rounded Ni-Fe metal particles and an interstitial space filled by P-rich (darker) and S-rich (lighter) materials. P-rich (darker) material forms a rind around the bleb. b) BSE-FESEM image of another bleb. It shows the same features of (a), but rounded Ni-Fe metal particles are smaller and more space is occupied by P- and S-rich material.

110

Fig. 9. Diagram Ni (wt%) vs Fe (wt%) showing the composition of Ni-Fe metal blebs occurring in the sample L06 in relationship with the bulk composition of the meteorite Gebel Kamil. In addition, compositional data of H-Ni and VH-Ni metal blebs occurring in the dark glass of the microparticles studied by Folco et al., (2015) are reported. The analyses of H-Ni are 28 points from 9 samples. The analyses of VH-Ni are 4 point from 2 samples.

Table 6. Bulk composition of the Gebel Kamil iron meteorite (HH-XRF; Gemelli et al. 2015), representative electron microprobe analyses (wt%), and average analyses of Ni-Fe metal blebs within dark glass (this work). The second column of the table lists the diameters of the blebs (µm).

Diameter (µm) Si P S Fe Co Ni Sum Gebel Kamil <0.05 0.04 0.02 78.5 0.69 20.6 99.8 L06_03 16 0.08 0.34 <0.04 32.3 1.56 65.7 100.0 L06_11 13 <0.08 1.32 0.11 34.4 1.81 62.7 100.3 L06_14 22 0.08 1.14 0.15 34.6 1.70 62.1 99.8 L06_19 41 <0.08 1.28 0.20 34.6 1.78 62.6 100.5 L06_26 111 0.07 <0.08 <0.04 31.7 1.99 66.3 100.1 L06_29 68 0.09 0.64 <0.04 34.2 1.86 63.2 100.0 L06_32 169 0.07 0.35 <0.04 28.2 1.63 69.1 99.3 L06_34 45 0.10 0.55 <0.04 33.2 1.86 64.5 100.1 L06_35 61 <0.08 0.33 0.06 28.0 1.85 69.8 100.0 L06_36 84 <0.08 0.15 <0.04 36.0 1.95 62.1 100.2 Avg. (n=42) 0.11 0.62 0.31 34.1 1.84 63.5 100.2 ±1σ 0.05 0.74 0.86 3.73 0.14 4.01 Abbreviations: avg., average; n, number of analyses. Details of the dark glass structures are shown in Figs. 6d-h. At high magnification the dark glass shows an emulsion texture: micrometric droplets (0.2-1 µm) of Si-rich glass are finely disseminated in a matrix of Fe-rich glass. At the contact with target fragments, the emulsion texture is absent over a thickness of about 2- 3 µm. The contacts between the target fragments and the dark glass are irregular. In some cases, they bear vesicles/cracks similar to those described in Figs. 3d and 3e and partially filled by dark glass (see white arrows in Fig.

111 6e). In another case, small SiO2 fragments show cuspate-lobated margin in contact with the dark glass (Fig. 6g).

More rarely the contact between target fragments and dark glass is characterized by a thin film (up to 100 µm) of opaque glass (Fig. 7a). Opaque glass was also found lining the vesicle walls (Figs. 6a and 7d). In BSE-FESEM images the opaque borders are made of very bright glass (Fig. 7b) and dendritic crystals (Fig. 7e). Due to their shape, these crystals could be magnetite. The occurrence of microcrystals of magnetite can be responsible for the opaque appearance of this glass.

Electron microprobe analyses of the dark glass show a great variability in Si, Al, Fe and Ni contents (Table 5). In sample L06, Al2O3, FeOT, and NiO are between 0.8-18.7 wt%,

1.8-61.5 wt%, and <0.01-10.2 wt%, respectively. Some analyses show anomalous high P2O5 values (up to 15.5 wt%). The analysis L06_062 is the one with the highest Fe and Ni contents

(FeOT 61.5 wt%; NiO 10.2 wt%). This analysis represents the composition of the opaque glass represented in Fig. 7e. Dark glass coating in sample L03 tend to have higher SiO2 (~60 wt%) and lower FeOT and NiO contents (<20 wt% and < b.d.l., respectively) in respect to sample L06. Both in dark glass of samples L06 and L03, the Al2O3 contents increase with increasing of FeOT contents and with decreasing of SiO2 contents (up to SiO2 ~50 wt%).

From SiO2 contents lower than ~50 wt%, the Al2O3 contents of dark glass of samples L06 and L03 decrease (Table 5). Quartz and lechatelierite inclusions in dark glass and dark glass form an almost continuous trend in the diagram Si (wt%) vs Fe (wt%) (Fig. 10).

112

Fig. 10. Silicon (wt%) vs. Fe (wt%) diagrams showing the geochemical features of (a) lechatelierite and dark glass of samples L06 and L03, and of (b) lechatelierite and dark glass of the microparticles studied by Folco et al., (2015). The chemical composition of Gebel Kamil, unshocked quartz, kaolinite, and the dark glass experimentally produced by MEMIN experiment (Ebert et al., 2013) are reported on both diagrams for comparison. The dashed line is the mixing line between Si and Fe.

DISCUSSION

White glass: a nearly pure target melt

Major- and trace-element compositions indicate that the white glass is a nearly pure impact melt of the target rocks. Due to the very low Al2O3 and FeOT contents in the white

113 glass (bulk analyses Al2O3 < 0.7 wt% and FeOT < 0.4 wt%, Table 2; spot analyses Al2O3 <

0.3 wt% and FeOT < 0.3 wt%, Table 3), we propose that the pale quartzarenite with negligible matrix (bulk analysis Al2O3 = 0.3 wt% and FeOT = 0.1 wt%, Table 1) is the most plausible protolith for impact melt lapilli and bombs made of white glass. Sample L09 has significant amounts of Al2O3 and FeOT (bulk analysis 1.4 wt% and 1.1 wt.%, respectively, Table 3), and we infer that its protolith was the reddish quartzarenite with a more abundant matrix composed of kaolinite and iron oxides and hydroxides (bulk analysis Al2O3 = 2.0 wt% and

FeOT = 1.5 wt%, Table 1). Due to the occurrence of the dark glass coating, bulk compositions of samples L03 an L08 have higher FeOT, Ni, and Co contents (L03: 0.77 wt% FeOT, 121 ppm Ni, and 7 ppm Co; L08: 0.84 wt% FeOT, 233 ppm Ni, and 11 ppm Co; Table 3).

The bulk chemical composition of the impact melt lapilli and bombs and target rocks from Kamil are plotted in the diagrams Fe vs Ni and Co vs Ni (Fig. 11). All the white glass samples (including those coated by dark glass) have Ni/Fe ratios significantly lower than Gebel Kamil Ni/Fe ratio (0.26; Table 3). The Ni/Co ratios of white glass samples are also lower than Gebel Kamil Ni/Co ratio (~30; Folco et al. 2015). However, samples L04 and M23 have a slightly higher ratio than L09 and target rock samples, indicating a possible minimum contamination by the projectile (Table 3; Fig. 11). However, on the bases of Ni and Co of these samples and of the impactor, the projectile contamination of white glass samples, including those coated by dark glass, was quantified to be less than 0.1 wt%.

114

Fig. 11. Diagrams (a) Ni (ppm) vs Fe (wt%) and (b) Ni (ppm) vs Co (ppm) showing the composition variations of target rocks and impact melt lapilli and bombs from Kamil. Dashed lines represent the Ni/Fe and Ni/Co of Gebel Kamil. In white glass impact melt lapilli and bombs, different examples of mineral shock and melting can be found. Accessory minerals within target rocks of the Kamil area represent about 1-2 vol%. They are Fe-Ti-oxides, tourmaline and zircon (Fazio et al. 2014). Depending on their melting temperatures, these minerals can melt or be shocked within impact melt glass. For example, a single partially decomposed zircon crystal was found within the white impact melt bomb M23 (Fig. 5a). The partial decomposition of zircon into SiO2 and ZrO2 (baddeleyite) indicates post-shock temperatures in excess of 1676 °C followed by the rapid quenching of the glass (Wittmann et al. 2006). Mineral melting produces melts with physical properties (e.g., viscosity) very different from the SiO2 melt, limiting or preventing their

115 mixing. Hence, the melts produced by mineral melting are found localized in small portions of the rocks, indicating a strong disequilibrium with the surrounding SiO2 melt. In sample L09, the melt formed by the melting of accessory minerals is quenched into Ti-bearing skeletal crystals and two immiscible glasses (a poorly polymerized TiO2-Al2O3-rich silicate melt and a highly polymerized SiO2-rich melt). In samples L08 and L04 no skeletal crystals occur, but only glassy material in the form of veinlets, micro- to nanoblebs, or with irregular shapes (Figs. 4c-f). The chemical composition of a Fe-rich Ni-free glass occurring in sample L04 is not clearly attributable to any accessory phases occurring in unshocked or low shocked sandstone from Kamil, although we suppose an important contribution from tourmaline and iron oxides (Table 4). The occurrence of quenched skeletal crystals in sample L09 indicates that it quenched at slower cooling rate than samples showing only glassy material. Quartz relicts in sample L09 are the result of the incomplete melting of quartzarenite grains and rapid quenching of the target melt (Figs. 3a-c and 4c-d). Melting usually starts in correspondence of rock heterogeneities, where it is possible to reach a local enhancement of pressure and temperature (e.g., Güldemeister et al. 2013), or in correspondence of mineral phases with lower melting temperatures, as the matrix (e.g., Ebert et al. 2013). In sample L09 the melting started in the matrix to continue into quartz grains. Due to the rapid quenching, the rocks did not arrive to total melting leaving a highly shocked residue of amorphous quartz grains. The occurrence of former quartz grains in sample L09 indicates that the maximum temperature reached to melt this sample was lower than 1515°C, that is the disequilibrium melting temperature of β-quartz (Petzold and Hinz 1976).

Former quartz grains provide information on how the melting of target sandstone proceeded. From well-formed PDFs (Fig. 4a) to the almost complete melting of the precursor grain (Fig. 4d), the volume of amorphous SiO2 (from 12.1 vol% up to 84.6 vol%) and the number of vesicles increase. The four images shown in Figure 4 are a sequence of snapshots of the process of quartz melting; an analogous process was described by Stöffler and Langenhorst (1994). Different stages could be found in the same sample indicating once more that shock pressure could be heterogeneously distributed at the millimetric to micrometric scale (Kowitz et al. 2013).

116 Dark glass: a mixture of target and projectile melts

Due to its high Fe and Ni contents, the dark glass has two precursors: Kamil target rocks and the iron meteorite Gebel Kamil (Tables 1, 5-6; Fig. 10). Due to the high Al contents of dark glass, we suppose that the target protolith was the pale wacke with siltstone levels (Table 1 and 3). Ratios Ni/Fe (0.20) and Ni/Co (25.3) of sample L06 are the closest to Gebel Kamil (Ni/Fe = 0.26 and Ni/Co = ~30; Folco et al. 2015). The Ni and Co enrichments of the dark glass of sample L06 compared to white glasses (Fig. 11) are compatible with a projectile contamination of ~11-12 wt%. As Kamil glasses, Barringer glasses and Wabar glasses also show different percentage of projectile contamination. The glasses from Barringer crater described by Hörz et al. (2002) show a range of meteoritic component from ~5 wt% to 15-20 wt%. Wabar glasses are divided, similarly to Kamil glasses, into white and black or dark glasses. The projectile contamination of the white glass is less than 1 wt% (Hörz et al. 1989; <0.1% for Kamil white glass, see previous section). Instead, the projectile contamination of black glass masses larger than 1 cm in size is ~4 wt% (Hörz et al. 1989) and of black glass particles smaller than 1 cm in size is generally > 10 wt% (Mittlefehldt et al. 1992).

Microprobe analyses reveal a very variable composition of the dark glass forming an almost continuous compositional trend between the unshocked quartz and lechatelierite and Gebel Kamil (Fig. 10). In the diagram Si (wt%) vs Fe (wt%), the electron microprobe analyses of dark glass show an almost continuous mixing trend between SiO2 phases (quartz and lechatelierite) and the projectile (Fig. 10a; Table 5). This trend is the same followed by microparticles found in Kamil soil (Fig, 10b; Folco et al. 2015) and by glass produced experimentally by shooting a projectile made of Campo del Cielo iron meteorite against a quartz-rich sedimentary target (Seeberger Sandstein; fine-grained sandstone (~89 vol% quartz and ~10 vol% of phyllosilicate; porosity ~23 vol%; Ebert et al. 2013). The dark glass of sample L06 is that showing the highest FeOT (up to 61.5 wt%) and NiO (up to 10.2 wt%) enrichment (Table 5; Fig. 10). The Fe-Ni-richest glass is localized at opaque borders shown in Fig. 7. Black-rimmed margins around large glassy target fragments were described also by Hamann et al. (2013) for the Wabar glass. In the other glass types, the moderate amount of Ni are likely due to the presence of micrometric to nanometric Ni-Fe metal blebs finely scattered into the dark glass (Table 5; Figs. 6g and 6h). Some analyses show a moderate enrichment in

TiO2 (up to 2 wt%), indicating that also accessory Fe-Ti oxides contributed to the chemical composition of the dark glass, hence also a minimal part of the Fe of the dark glass could have a terrestrial origin (Fig. 11a). The dark glass of sample L03 has a maximum FeOT

117 content of 19.3 wt%. All the analyses of this glass have negligible amounts of Ni. Due to the occurrence of clay minerals matrix both in Kamil target rocks (and in Seeberger Sandstein), the dark glass compositions both from Kamil and laboratory experiments are slightly shifted to the left hand side of the mixing curve, i.e. toward kaolinite composition (Fig. 10; Ebert et al. 2013).

In Kamil impact glasses, liquid immiscibility textures are very common in the dark glass (Figs. 6d-h), but can also be found in some restricted areas of white glass (Fig. 5c). Liquid–liquid phase-separation of the dark glass produces the formation of a poorly polymerized FeO-rich silicate melt and of a highly polymerized, SiO2-rich melt, similarly to the emulsions of Wabar black glass (Hamann et al. 2013). This is further evidence that liquid immiscibility process should be a common and an important process in impact glass formation, because impact glasses form at very high temperatures, they have very diverse compositions as a function of the percentage of projectile contamination, and they are rapidly quenched. Further studies on liquid immiscibility textures of impact glasses from Kamil and other craters are necessary to better constrain this process and the thermal history of impact glass formation.

Petrographic observations also confirm that the dark glass is the result of the interaction between the target and the projectile. Target fragments within the dark glass are very common; they are generally monotypic inclusions of shocked sandstone clasts, diaplectic glass, and lechatelierite fragments. They are randomly dispersed in the dark glass (Figs. 1c, 6a and 6b). They represent three different regimes of shock metamorphic pressures: PDFs in shocked sandstone clasts, 10-20 GPa (Fazio et al. 2014); diaplectic glass, >35 GPa (Stöffler and Langenhorst 1994); lechatelierite, 50-60 GPa (Stöffler and Langenhorst 1994).

Two possible mechanisms can explain the occurrence of target fragments within the dark glass: (i) target fragments are undigested relicts of the process of melting and mixing involving target and projectile during the end of contact and compression stage; (ii) target fragments were engulfed by the dark glass during the excavation stage. The first mechanism is similar to that described in the previous section to explain the occurrence of quartz relicts in sample L09. This mechanism is thought to be the most plausible mechanism to explain the occurrence of target inclusions within the experimentally produced dark glass by Ebert et al. (2013; 2014). An evidence of this mechanism could be represented by the occurrence of lobated margins of silica glass inclusions within dark glass (Fig. 6d and 6e). This type of margin is indicative of a high temperature process. In addition, the occurrence of dark glass

118 within the geometrical vesicles/cracks of the same inclusions is compatible with the same mechanism. On the other hand, the second mechanism could more easily explain the occurrence of dark glass coating centimeter-size lechatelierite fragments (Fig. 1b) and microparticles of target materials, both shocked quartz grains and lechatelierite (Figs. 6a, 6b, 6d, 6f, and 7a; Folco et al. 2015). White glass masses coated by dark glass were described also from Wabar (e.g., Mittlefehldt et al. 1992; Hörz et al. 1989), Barringer (e.g., Nininger 1954), and lunar samples (e.g., Apollo 16 lunar samples; Grieve and Plant 1973). Both these mechanisms could be considered plausible to explain the occurrence of variously shocked target fragments within the Kamil dark glass.

Nickel-Fe metallic blebs are very common in the dark glass and their sizes are very variable, ranging from few nm up to 200 µm. Their enrichment in Ni relative to Fe compared to Gebel Kamil meteorite derives from the different redox properties of Ni and Fe. Iron is 0 preferentially oxidized relative to Ni, because the Gibbs free energy variation (ΔG Fe/FeO) of 0 the reaction of oxidation of Fe (Fe + 0.5 O2 = FeO) is higher than ΔG Ni/NiO of the reaction of oxidation of Ni (Ni + 0.5 O2 = NiO). Moreover, the partition coefficient between a metallic and a silicate melt of Ni is higher than that of Fe (DNi met/sil >> DFe met/sil; e.g., Righter et al. 1997). Due to these reasons the silicate liquid is enriched in Fe, and the metallic liquid is enriched in Ni. We thus propose that the Ni-Fe metallic blebs are result of the chemical fractionation of the metallic liquid of the projectile. Nickel-Fe metal blebs from Kamil show a wide range of chemical compositions as shown in Fig. 9, however within the same sample their composition is quite homogeneous (Table 6). Metal blebs were described in several impactites (e.g., Wabar (Saudi Arabia), Gibbons et al. 1976; Hamann et al. 2013; Monturaqui (Chile), Bunch and Cassidy 1972; Gibbons et al. 1976; Henbury (Australia), Gibbons et al. 1976; Barringer Crater (Arizona) Kearsley et al. 2004) and in impact experiments (Ebert et al. 2013; Ebert et al. 2014). Metal blebs from these impact craters are also enriched in Ni, and show a wide range of chemical composition: from values comparable to their iron meteorite impactor (Ni = 8 wt%, 20 wt%, and 13 wt% for Wabar, Henbury, and Monturaqui, respectively; Gibbons et al. 1976) up to very high Ni concentration (Ni = 64 wt%, 94 wt%, and 75 wt% for Wabar, Henbury, and Monturaqui, respectively; Gibbons et al. 1976). The P- rich and S-rich material, occurring in the interstices and as rind of Fe-Ni metal particles forming the metal blebs, represent the residual liquid of the process of crystallization/quenching of the molten meteoritic liquid depleted in Fe into rounded Ni-Fe metal particles (bulk P and S contents of Gebel Kamil are 0.04 wt% and 0.02 wt%,

119 respectively; Gemelli et al. 2015) Similar metal bleb textures were also described by Bunch and Cassidy (1972) for the Monturaqui impactites and by Kearsley et al. (2004) for the Barringer Crater impactites.

Impact melting scenario

The model for the formation of the impact melt at Kamil is based on the geochemical and petrographic evidence discussed in the previous paragraphs, on the study of the microparticles scattered in the soil around the crater (Folco et al. 2015), on the recent model proposed by Ebert et al. (2014), and on the widely accepted general model for impact cratering (e.g., Osinski and Pierazzo 2013).

Impact cratering is a process characterized by a strong disequilibrium. The duration of impact events of the size of Kamil is supposed to be very short, on the scale of few seconds.

Experiments demonstrate that the projectile is heated (up to 1800°C; Berry et al. 2007) and can melt before the contact with the target surface (stage I in Ebert et al. 2014). A possible evidence of this process could be represented by the Fe-Ni oxide spherules collected in Kamil soil and described by Folco et al. (2015). These spherules are made of oxidized Fe and Ni (FeOT 71.6 ± 1.92 wt% and NiO 20.9 ± 1.90 wt%), without any significant target contaminations (SiO2 0.77 ± 1.05 wt% and Al2O3 0.06 ± 0.05 wt%; Folco et al. 2015). Thus, we supposed that their formation could happen in the atmosphere, before of the contact and compression stage.

The contact and compression stage starts when the projectile reaches the target surface and penetrates in it for 1-2 times its diameter (Osinski and Pierazzo 2013). At the contact point the maximum shock pressure is reached. For Kamil, the maximum shock pressure was at least of 30-60 GPa (Fazio et al. 2014), likely around 50 GPa due to the low porosity (0-4 vol%) of the target rocks (Wünnemann et al. 2008). Based on planar impact approximation (Melosh 2013), this range of shock pressures corresponds to impact velocities between 5.0 km s-1 (30 GPa) and of 7.5 km s-1 (60 GPa), assuming an impact angle of 45° (Fazio et al. 2014). During this stage, the kinetic energy of the projectile is converted into heat and shock waves propagating both in the target and in the projectile (Fig. 12a; Osinski and Pierazzo 2013). Shock pressure rapidly decreases from pressures of the order of 30-60 GPa at the contact point down to pressures < 1GPa at the crater wall (Fig. 12a).

According to Ebert et al. (2014), between the contact and compression stage and the excavation stage there is an intermediate stage, which is the most important for the glass

120 formation and for the physical-chemical interaction between the target and the projectile. This intermediate stage is indicated in Fig. 12a as “end of contact and compression stage”. During this intermediate stage, the projectile is intensely plastically deformed, as evidenced by the shear bands observed in fragments of Gebel Kamil meteorite (D’Orazio et al. 2011; Kenkmann et al. 2013) and a rarefaction wave, produced by the reflection of the shock wave at the rear surfaces of the projectile, moves downward through the projectile and then through the target. This propagation produces decompression, heating, melting both of the target and the projectile, and the ejection of meteorite shrapnel and target fragments (Fig. 12b; Stage II in Ebert et al. 2014).

At the interface between the projectile and the target, before the ejection, melt from the projectile is injected into the highly shocked, partially melted target rocks. Target and projectile melts chemically mix producing a silicate glass rich in Al and Fe and containing numerous Ni-Fe metallic blebs: the dark glass (Figs. 6 and 10). Nickel-Fe metallic blebs are enriched in Ni and represent the residue of the projectile oxidation and melting (Figs. 8 and 9). The projectile-target interaction zone has a very small volume compared to the total volume of rocks affected by the impact (Ebert et al. 2014). Hence, melted and shocked rocks formed just below the interaction zone had a negligible interaction with the projectile melt (Fig. 12b).

The smallest is the crater the shortest is the time for melts homogenization (Hörz et al. 1989). The formation of different types of glass characterized by different percentage of target rock melts and/or projectile contamination is thus more likely in small craters (e.g., Wabar glass, Hörz et al. 1989; Barringer glass, Hörz et al. 2002). Moreover, the composition of the target material could also have played a role in the formation of two different types of glasses

(Hörz et al. 1989). The nearly pure SiO2 liquid produced by the melting of quartzarenite (SiO2

> 95 wt% and Al2O3 < 2 wt%; Table 1) has a high degree of polymerization and unlikely mixes with a FeO-rich liquid. Instead, the liquid produced by the melting of pale wacke with siltstone levels containing ~ 10 wt% of Al2O3 has a lower degree of polymerization and can mix with the melt from the projectile.

The ejection (excavation stage; Fig. 12c) is the final consequence of the passage of the rarefaction wave: the ejection starts form the rear surface of the projectile proceeding into the projectile and then into the target (Langenhorst and Deutsch 2002). Highly deformed projectile fragments are the first objects to be ejected. They are followed by the ejection of still hot and plastic masses of dark and white glass and then of fragments of target rocks

121 variously shocked (Fig. 12 c). Ejection is a chaotic and rapid stage and it is likely that a significant mechanical interaction between impact melt glasses, and target and projectile fragments occurred during this stage, producing the engulfment of target fragments of various sizes and shock degrees into the dark glass (Figs. 1c, 6, and 7; Folco et al. 2015) and the sticking of the dark glass onto meteorite shrapnel fragments (Fig. 1d). Due to the high temperature of the system, minor local chemical interaction at the contact between the dark glass and the inclusions could also have occurred.

122

Fig. 12. Schematic sketch of the impact melting process. a) At the end of the contact and compression stage the shock wave passes thought the target and the projectile. The target is highly compressed and the projectile is heated and highly deformed. Shock pressure rapidly decrease from 50 GPa at the contact point up to <5 GPa at the crater wall (isobars are indicated by dashed gray curved lines). b) Close-up view of the black rectangular area in (a) showing the formation of a projectile and a target liquid. The projectile and the target liquid chemically and physically interact at the contact zone. c) During the excavation stage meteorite shrapnel fragments, dark glass, white glass, and shocked target fragments are ejected, because of the passage of the rarefaction wave. Ejection trajectories are indicated by solid black arrowhead curved lines. This is a very chaotic and rapid stage. During this stage the white glass and shocked target fragments can be engulfed or coated by the dark glass. Images modified after French (1998) and Ebert et al. (2014).

123 CONCLUSIONS

Impact melt lapilli and bombs from Kamil crater (45 m in diameter) represent an invaluable natural material for the study of impact melting process and target-projectile chemical-physical interaction in small terrestrial impact craters.

The main features of the impact melt lapilli and bombs can be summarized in the following:

(i) Two distinct glass types were found at Kamil: white glass and dark glass.

(ii) The white glass is highly vesicular and almost exclusively made of SiO2. Thus, we call it lechatelierite. Ni and Co concentrations are low and close to target concentrations. One sample is stained by iron-oxi-hydroxides, probably because of the different lithology of the target rock. Former quartz grains with planar deformation features (PDFs) occur in just one sample. (iii) The dark glass is dominantly made of a silicate melt with variable contents of

Al2O3 (0.84-18.7 wt%), FeOT (1.83-61.5 wt%) and NiO (<0.01-10.2 wt%). The dark glass typically includes fragments (from few µm to several mm in size) of shocked sandstone, diaplectic glass, lechatelierite, and Ni-Fe metal spherules. (iv) The dark glass coats the white glass in some samples and it was also found stuck onto meteorite shrapnel fragments.

The white and dark glasses show a bimodal distribution of the projectile contamination. The white glass experienced a negligible interaction with the projectile. Based on Ni and Co concentrations, we estimate a contamination of projectile material <0.1 wt%. Conversely, we estimate that the dark glass contains about 11-12 wt% of projectile material.

Based on the above evidence and on previous models on impact cratering and melting processes we suggest a model for the formation of the bimodal distribution of the projectile contamination in the impact melt lapilli and bombs at Kamil. The melting and the chemical interaction between target and projectile melts occurred in an intermediate stage between the contact and compression stage and the excavation stage. The interaction zone is supposed to have a very small volume compared to the total volume of rocks affected by the impact; hence, the dark glass formed at the contact interaction zone and the white glass formed immediately out this zone. The mixing was mainly prevented by the extremely rapid time sequence of the impact process. Kamil impact melt lapilli and bombs are further confirmation

124 that the impact melting process at small impact crater produced by iron meteorites is a process in strong disequilibrium.

The occurrence of numerous target inclusions in dark glass has a dual origin: i) they are in part relicts of the precursor rocks that suffered shock melting; ii) they are target fragments engulfed by the dark glass during the excavation stage. The first group of inclusions formed at the contact interaction zone, during the chemical mixing between the projectile and the target liquids (end of the contact and compression stage), hence during the dark glass formation. The second group of inclusions was engulfed in the dark glass during the ejection stage. This process was dominantly mechanical, even though minor and local chemical interactions could have occurred at the contact between the dark glass and the inclusions, indicating interaction at high temperatures in the ejecta curtain. Moreover, during the excavation stage, still hot masses of dark glass could be stuck onto iron meteorite shrapnel fragments close to the contact and interaction zone.

Acknowledgements – This work was supported by the Italian Ministero degli Affari Esteri - Progetti di Grande Rilevanza, Protocollo Esecutivo ITALIA-EGITTO. The 2010 geophysical expedition work was carried out within the framework of the 2009 Italian- Egyptian Year of Science and Technology. We thank Prof. M. Alsherbiny (former President of the Egyptian National Academy for Scientific Research and Technology) and Prof. F. Porcelli (Scientific Attaché, Italian Embassy, Egypt) for diplomatic and institutional support; the Egyptian Army for logistical support. Agnese Fazio is supported by the University of Pisa PhD School on Earth Science Scuola di Dottorato in Scienze di Base Galileo Galilei, and by the Barringer Family Fund for Meteorite Impact Research 2014. Luigi Folco and Massimo D'Orazio are also supported by the University of Pisa Fondi di Ateneo. Institut des Sciences de la Terre (ISTerre) of Grenoble is part of Labex OSUG@2020 (ANR10 LABX56). The authors are grateful to Valentina Batanova, Andrea Cavallo, Raul Carampin, and Gabriele Paoli for their assistance during electron microprobe analysis at ISTerre (Grenoble), field emission scanning electron microscope analysis at INGV (Rome), electron microprobe analysis at CNR-IGG (Padova), and ICP-MS analysis (Pisa), respectively.

REFERENCES Berry S., Chen F. J., Wilder M. C. and Reda D. C. 2007. Boundary layer transition experiments in support of the hypersonics program. American Institute of Aeronautics and Astronautics Thermophysics Conference, 2007–4266.

125 Bunch T. E. and Cassidy W. A. 1972. Petrographic and electron microprobe study of the Monturaqui impactite. Contribution to Mineraogy and Petrology 36:95–112.

D’Orazio M., Folco L., Zeoli A. and Cordier C. 2011 Gebel Kamil: The iron meteorite that formed the Kamil crater (Egypt). Meteoritics & Planetary Science 46:1179–1196.

Earth Impact Database. http://www.passc.net/EarthImpactDatabase/index.html. Accessed December 23, 2014.

Ebert M., Hecht L., Deutsch A. and Kenkmann T. 2013. Chemical modification of projectile residues and target material in a MEMIN cratering experiment. Meteoritics & Planetary Science 48:134–149.

Ebert M., Hecht L., Deutsch A., Kenkmann T., Wirth R. and Berndt J. 2014. Geochemical processes between steel projectiles and silica-rich targets in hypervelocity impact experiments. Geochimica et Cosmochimica Acta 133:257–279.

Fazio A., Folco L., D’Orazio M., Frezzotti M. L. and Cordier C. 2014. Shock metamorphism and impact melting in small impact craters on Earth: Evidence from Kamil Crater, Egypt. Meteoritics & Planetary Science 49:2175-2200.

Folco L., Di Martino M., El Barkooky A., D’Orazio M., Lethy A., Urbini S., Nicolosi I., Hafez M., Cordier C., van Ginneken M., Zeoli A., Radwan A. M., El Khrepy S., El Gabry M., Gomaa M., Barakat A. A., Serra R. and El Sharkawi M. 2010. The Kamil Crater in Egypt. Science 329:804.

Folco L., Di Martino M., El Barkooky A., D’Orazio M., Lethy A., Urbini S., Nicolosi I., Hafez M., Cordier C., van Ginneken M., Zeoli A., Radwan A. M., El Khrepy S., El Gabry M., Gomaa M., Barakat A. A., Serra R. and El Sharkawi M. 2011. Kamil Crater (Egypt): Ground truth for small-scale meteorite impacts on Earth. Geology 39:179– 182.

Folco L., D’Orazio M., Fazio A., Cordier C., Zeoli A., van Ginneken M., and El-Barkooky A. 2015. Microscopic impactor debris in the soil around Kamil Crater (Egypt): inventory, distribution, total mass and implications for the impact scenario. Meteoritics & Planetary Science (accepted on 11.12.2014). Gemelli M., D’Orazio M., and Folco L. 2015. Chemical analyses of iron meteorites by hand- held X-ray fluorescence. Geostandard Geoanalytical Research 39:1:55-69.

126 Gibbons R. V., Hörz F., Thompson T. D. and Brownlee D. E. 1976. Metal blebs in Wabar, Monturaqui and Henbury impactites. In 7th Proceeding on Lunar Science Conference pp. 863–880.

Goderis S., Wittmann A., Zaiss J., Elburg M., Ravizza G., Vanhaecke F., Deutsch A., and Claeys P. 2013. Testing the ureilite projectile hypothesis for the El'gygytgyn impact: Determination of siderophile element abundances and Os isotope ratios in ICDP drill core samples and melt rocks. Meteoritics & Planetary Science 48:1296-1324.

Grieve R. A. F. and Plant A. G. 1973. Partial melting on the lunar surface as observed in glass coated Apollo 16 samples. In 4th Proceeding of Lunar Science Conference Supplement 4, Geochimica et Cosmochimica Acta 1:667–679.

Güldemeister N., Wünneman K., Durr N., and Hiermaier S. 2013. Propagation of impact- induced shock waves in porous sandstone using mesoscale modeling. Meteoritics & Planetary Science 48:115–133.

Hamann C., Hecht L., Ebert M. and Wirth R. 2013. Chemical projectile–target interaction and liquid immiscibility in impact glass from the Wabar craters, Saudi Arabia. Geochimica et Cosmochimica Acta 121:291–310.

Hörz F., See T. H., Murali A. V. and Blanchard D. P. 1989. Heterogeneous dissemination of projectile materials in the impact melts from Wabar Crater, Saudi Arabia. In Proceeding 19th Lunar & Planetary Science Conference pp. 697–709.

Hörz F., Mittlefehldt D. W., See T. H. and Galindo C. 2002. Petrographic studies of the impact melts from Meteor Crater, Arizona, USA. Meteoritics & Planetary Science 37:501-531.

Langenhorst F. and Deutsch A. 2012. Shock metamorphism of minerals. Elements 8:31-36.

Kearsley A., Graham G., McDonnell T., Bland P., Hough R. and Helps P. 2004. Early fracturing and impact residue emplacement: Can modeling help to predict their location in major craters? Meteoritics & Planetary Science 39:247–265.

Kenkmann T., Trullenque G., Deutsch A., Hetcht L., Ebert M., Salge T., Schäfer F., and Thoma K. 2013. Deformation and melting of steel projectiles in hypervelocity cratering experiments. Meteoritics & Planetary Science 48:150–164.

Koeberl C., Claeys P., Hecht L. and McDonald I. 2012. Geochemistry of impactites. Elements 8:37–42.

127 Kowitz A., Güldemeister N., Reimold W. U., Schmitt R. T., and Wünnemann K. 2013.

Diaplectic quartz glass and SiO2 melt experimentally generated at only 5 GPa shock pressure in porous sandstone: Laboratory observations and meso-scale numerical modeling. Earth and Planetary Science Letters 384:17–26.

Melosh H. J. 2013. The contact and compression stage of impact cratering. In Impact Cratering, edited by Osinsky G. R. and Pierazzo E., UK-USA: Wiley-Blackwell. pp. 336.

Mittlefehldt D. W., See T. H. and Hörz F. 1992. Dissemination and fractionation of projectile materials in the impact melts from Wabar Crater, Saudi Arabia. Meteoritics 27:361– 370.

Nininger H. H. (1954) Impactite slag at Barringer Crater. American Journal of Science 252:277–290.

Osinski G. R. and Pierazzo E. 2013. Impact cratering: processes and products. In Impact Cratering (eds. G. R. Osinski and E. Pierazzo). Wiley-Blackwell, Chichester, West Sussex, U. K., pp. 125–145.

Osinski G. R., Grieve R. A. F., Marion C. and Chanou A. 2013. Impact melting. In Impact Cratering (eds. G. R. Osinski and E. Pierazzo). Wiley-Blackwell, Chichester, West Sussex, U. K., pp. 125–145.

Ott U., Merchel S., Herrmann S., Pavetich S., Rugel G., Faestermann T., Fimiani L., Gomez- Guzman J. M., Hain K., Korschinek G., Ludwig P., D’Orazio M. and Folco L. 2014. Cosmic ray exposure and pre-atmospheric size of the Gebel Kamil iron meteorite. Meteoritics & Planetary Science 49:1365–1374.

Petzold A., and Hinz W. 1976. Silikatchemie: Einführung in die Grundlagen. Deutscher Verlag für Grundstoffindustrie, Leipzig, 219 pp..

Righter K., Drake M. J., and Yaxley G. 1997. Prediction of siderophile element metal–silicate partition coefficients to 20 GPa and 2800 °C: the effects of pressure, temperature, oxygen fugacity, and silicate and metallic melt composition. Physics of the Earth and Planetary Interiors 100:115–134.

Stöffler D. and Grieve R. A. F. 2007. Impactites, Chapter 2.11. In Metamorphic Rocks: a Classification and Glossary of Terms, Recommendations of the International Union of

128 Geological Sciences (eds. D. Fettes and J. Desmons). Cambridge University Press, Cambridge, pp. 82–91, 111–125, and 126–242.

Stöffler D. and Langenhorst F. 1994. Shock metamorphism of quartz in nature and experiment: I. Basic observation and theory. Meteoritics 29:155–181.

Tamponi M., Bertoli F., Innocenti F. and Leoni L. 2003. X-ray fluorescence analysis of major elements in silicate rocks using fused glass discs. Atti della Società Toscana di Scienze Nataturali, Memorie Serie A CVII:73–80.

Urbini S., Nicolosi I., Zeoli A., El Khrepy S., Lethy A., Hafez M., El Gabry M., El Barkooky A., Barakat A., Gomaa M., Randwan A. M., El Sharkawi M., D’Orazio M. and Folco L. 2012. Geological and geophysical investigation of Kamil Crater, Egypt. Meteoritics & Planetary Science 47:1842–1868.

Wittmann A., Kenkmann T., Schmitt R. T. and Stöffler D. 2006. Shock-metamorphosed zircon in terrestrial impact craters. Meteoritics & Planetary Science 41:433–454.

Wünnemann K., Collins G. S., and Osinski G. R. 2008. Numerical modelling of impact melt production in porous rocks. Earth and Planetary Science Letters 269:530-539.

129

4. MICROSCOPIC IMPACTOR DEBRIS IN THE SOIL AROUND KAMIL CRATER (EGYPT): INVENTORY, DISTRIBUTION, TOTAL MASS AND IMPLICATIONS FOR THE IMPACT SCENARIO

1 1 1 2,3 4 Luigi FOLCO , Massimo D’ORAZIO , Agnese FAZIO , Carole CORDIER , Antonio ZEOLI , 5 6 Matthias VAN GINNEKEN , and Ahmed EL-BARKOOKY 1Dipartimento di Scienze della Terra, Università di Pisa, Via S. Maria 53, 56126 Pisa, Italy 2Université de Grenoble Alpes, ISTerre, BP 53, 38041 Grenoble CEDEX 9, France 3CNRS, ISTerre, BP 53, 38041 Grenoble CEDEX 9, France 4Museo Nazionale dell’Antartide, Università di Siena, Via Laterina 8, 53100 Siena, Italy 5Department of Earth Science and Engineering, Imperial Collage, Prince Consort Road, London SW 2BP, United Kingdom 6Department of Geology, Faculty of Science, Cairo University, Giza, Egypt

In Meteoritics & Planetary Science, Accepted on 12th December 2014

Abstract – We report on the microscopic impactor debris around Kamil Crater (45 m in diameter, Egypt) collected during our 2010 geophysical expedition. The hypervelocity impact of Gebel Kamil (Ni-rich ataxite) on a sandstone target produced a downrange ejecta curtain of microscopic impactor debris due SE-SW of the crater (extending ~300,000 m2, up to ~400 m from the crater), in agreement with previous determination of the impactor trajectory. The microscopic impactor debris include vesicular masses, spherules, and coatings of dark impact melt glass which is a mixture of impactor and target materials (Si, Fe, Al-rich glass), plus Fe-Ni oxide spherules and mini shrapnel, documenting that these products can be found in craters as small as few tens of meters in diameter. The estimated mass of the microscopic impactor debris (<290 kg) derived from Ni-concentrations in the soil is a small fraction of the total impactor mass (~10 t) in the form of macroscopic shrapnel. That Kamil Crater was generated by a relatively small impactor is consistent with literature estimates of its pre-atmospheric mass (>20 t, likely 50-60 t).

INTRODUCTION

The characterization of microscopic projectile debris occurring at small-scale meteorite craters (≤1500 m in diameter) can improve our understanding of the physical- chemical interaction between projectile and target, including impactor fragmentation and dispersion, impact melt (and vapor) production, ejecta and plume evolution in small impact events (e.g., Hörz et al. 2002; Artemieva and Pierazzo 2009; 2011; Ebert et al. 2013; Hamann et al. 2013; Kenkmann et al. 2013; Shuvalov and Dypvik 2013).

131 Kamil Crater (Egypt; Fig. 1) is a pristine (or nearly so) type structure for very small impacts on Earth (Folco et al. 2010; 2011) and it thus offers the possibility to document the characteristics of the microscopic impactor debris, and its expansion and distribution in very small-scale meteorite impacts.

Kamil Crater is a 45-m-diameter simple crater, produced by the hypervelocity impact of the Ni-rich (~20 wt%; D'Orazio et al. 2011) Gebel Kamil iron meteorite on layered sandstones with subhorizontal bedding, interbedded with minor siltstones and wacke (Urbini et al. 2012; Fazio et al. 2014). Due to its very young age (probably <5,000 yr; Folco et al. 2011) and the dry climatic conditions in the Sahara (e.g., Kuper and Kröpelin 2006), it is exceptionally well-preserved with an unaltered crater structure, rayed pattern of bright ejecta, various types of shock-metamorphosed and impact melt rocks (from cm-sized bombs to microscopic impact spherules), and a nearly intact assemblage of fragments of the projectile that fragmented into thousands of shrapnel (i.e., unmolten projectile fragments; O’Keefe and Ahrens 1985) upon impact and were deposited in and around the crater (Folco et al. 2011; D'Orazio et al. 2011; Urbini et al. 2012; Fazio et al. 2014). The concentration of large shrapnel fragments (from ~10 g to ~30 kg) due SE of the crater indicates an oblique impact from the NW (D'Orazio et al. 2011). The mass of shrapnel found through visual search (D'Orazio et al. 2011) and geomagnetic survey (Urbini et al. 2012) indicates that the mass of the projectile was at least ~5,000 kg. Both noble gases and radionuclides in Gebel Kamil point to a pre-atmospheric mass >20 t, with a preferred estimate of 50-60 t (Ott et al. 2014). Shock metamorphic and melting features in the target indicate peak pressure between 30 and 60 GPa and impact velocities of 3.5 to 5.5 km s-1 for vertical impact or 5.0 to 7.5 km s-1 for an impact angle of 45° (Fazio et al. 2014).

In this paper, we investigate the microscopic impactor debris of Gebel Kamil. Soil samples around Kamil Crater were collected during our 2010 geophysical expedition (Folco et al. 2010). Following similar studies, mainly conducted at Barringer Crater (e.g., Niniger 1956; Rinehart 1958), this work provides an inventory of the various types of microscopic impactor material that can be produced in small-scale impact craters and constrains the trajectory of the Gebel Kamil iron meteorite, providing an estimate of its total mass at the time of impact.

SAMPLES AND ANALYTICAL METHODS

In order to study the microscopic impactor debris at Kamil Crater, we collected 44 soil samples around the crater (Fig. 1; Table 1). Samples were collected at incremental distances

132 from the crater rim along eight radial traverses (45° apart, starting from the north) extending up to 1.6 km from the crater rim (i.e., at about the distance from the crater rim where the most distant macroscopic shrapnel was found (D'Orazio et al. 2011). Each sample consisted of a 30 x 30 x ~5 cm soil volume, i.e. we assumed that this volume contained all the microscopic impactor debris deposited on the considered surface. This assumption is based on the fact that the fine grained ejecta deposit is typically less than some centimeters in thickness anywhere around the crater; even within the ejecta blanket, i.e., within ~50 m from the crater rim (Urbini et al. 2012), the bulk and the thickness of the deposit is given by boulders and blocks with thin fine-grained ejecta layers in between. Samples were collected using shovel, paint brush, blade and measuring tape (Fig. 2a). Samples were dry sieved in the field to obtain the <5 mm size fraction. Exploiting the contrasting magnetic properties of projectile and target rocks, the magnetic particles were subsequently extracted using a field extractor (Fig. 2b) and weighed. To complete magnetic separation, we used the Extractor-SE in-line magnetic separator, with a stainless steel housing and three Nd magnet rods arranged in a triangular matrix producing a magnetic force of 0.6 T (S+S Separation and Sorting Technology GmbH, Schönberg, Germany). A separation efficiency >80% was derived through a number of random field tests at various locations.

Fig. 1. Kamil Crater, southern Egypt, and location map of the collected soil samples. Base map: enhanced true color QuickBird satellite image (22 October 2005; courtesy of Telespazio S.p.A.).

133 Table 1. Magnetic extract <5 mm of soil samples from Kamil Crater: location and mass. Samples are listed according to the bearing of the profiles starting from the north, and to their increasing distance from crater rim. Mass of Profile/ Distance from Altitude Lable Latitude S Longitude E magnetic Location crater rim (m) a.s.l. (m) extract (g) S12 N 15 26° 05' 18" 22° 01' 06" 599 4.5 S13 N 100 26° 05' 19" 22° 01' 05" 602 6.6 S14 N 200 26° 05' 23" 22° 01' 06" 602 7.8 S15 N 400 26° 05' 29" 22° 01' 07" 609 8.8 S16 N 800 26° 05' 42" 22° 01' 06" 615 13.3 S35 NE 15 26° 05' 59" 22° 01' 03" 597 2.8 S36 NE 100 26° 05' 14" 22° 01' 07" 596 10.5 S37 NE 200 26° 05' 11" 22° 01' 06" 608 18.4 S38 NE 400 26° 05' 07" 22° 01' 05" 604 11.0 S39 NE 800 26° 05' 00" 22° 01' 03" 605 9.0 S01 E 15 26° 04' 47" 22° 01' 07" 603 15.2 S02 E 100 26° 05' 15" 22° 01' 07" 601 11.7 S03 E 200 26° 05' 15" 22° 01' 10" 605 13.3 S04 E 400 26° 05' 16" 22° 01' 13" 604 4.0 S05 E 800 26° 05' 14" 22° 01' 19" 615 4.0 S06 E 1200 26° 05' 13" 22° 01' 33" 606 23.1 S23 SE 15 26° 05' 15" 22° 01' 04" 597 12.1 S24 SE 100 26° 05' 15" 22° 01' 02" 595 21.7 S25 SE 200 26° 05' 15" 22° 00' 58" 600 18.2 S26 SE 400 26° 05' 14" 22° 00' 58" 596 17.1 S27 SE 800 26° 05' 15" 22° 00' 36" 592 9.8 S28 SE 1220 26° 05' 22" 22° 00' 24" 588 10.1 S17 S 15 26° 05' 16" 22° 01' 04" 600 10.3 S18 S 100 26° 05' 17" 22° 01' 02" 594 25.3 S19 S 200 26° 05' 20" 22° 01' 00" 596 13.7 S20 S 400 26° 05' 25" 22° 00' 56" 595 15.6 S21 S 800 26° 05' 35" 22° 00' 47" 591 7.6 S22 S 1280 26° 05' 44" 22° 00' 36" 592 7.5 S29 SW 15 26° 05' 14" 22° 01' 05" 599 11.7 S30 SW 100 26° 05' 12" 22° 01' 03" 597 6.4 S31 SW 200 26° 05' 10" 22° 01' 01" 602 12.7 S32 SW 400 26° 05' 04" 22° 01' 01" 596 5.3 S33 SW 800 26° 04' 58" 22° 00' 46" 596 6.1 S34 SW 1160 26° 04' 46" 22° 00' 46" 592 13.7 S07 W 15 26° 05' 17" 22° 01' 07" 597 2.9 S08 W 100 26° 05' 18" 22° 01' 09" 593 5.9 S09 W 200 26° 05' 21" 22° 01' 11" 592 5.7 S10 W 400 26° 05' 24" 22° 01' 16" 596 7.9 S11 W 800 26° 05' 33" 22° 01' 25" 603 8.6 S41 NW 100 26° 05' 12" 22° 01' 09" 596 3.0 S43 NW 400 26° 05' 06" 22° 01' 11" 600 5.0 S44 NW 800 26° 04' 53" 22° 01' 27" 596 7.6

134

Fig. 2. Field photos of steps in the collection of soil samples around Kamil Crater (A, B) and stereomicrographs of representative samples of their magnetic components (C, D, E). A) A surface soil volume measuring 30 x 30 x ~5 cm (arrowed) was collected at each site and sieved (<5 mm). Inset: a detail of the sampling surface. B) The sieved sample was then let through a powerful field magnetic separator (arrowed) to obtain the magnetic extract studied in this work. The picture features one of the preliminary extraction test. C) The magnetic extract of sample S23 (<5 mm; 15 m due SE of crater rim) showing abundant dark impact melt glass particles mainly in the shape of vesicular clastic masses, mini-to-microscopic individual or compound spheroids, and coatings of pale target rock particles (mainly shocked sandstone fragments and or lechatelierite, as revealed by electron microscopy and microanalyses). Subordinate mini-to-microscopic iron meteorite shrapnel and microscopic Fe- Ni-oxide spherules also occur. D) The magnetic extract of sample S44 (<5 mm; 800 m due NW of crater rim) dominated by Precambrian crystalline basement rock fragments (mainly amphibolite gneiss). E) Close up view of the magnetic extract of sample S27 (<400 µm; 800 m due SE crater rim) showing abundant Fe-Ti oxide plus garnet grains of terrestrial origin. Abbreviations: CS: cosmic spherule; DG: individual masses of dark impact melt glass; DGc: dark glass coatings; DG sph: dark glass spheroids; Fe-Ti ox: Fe-Ti oxides; GK: Gebel Kamil iron meteorite shrapnel; Grt: garnet; Fe-Ni sph: Fe-Ni spherule; Qtz: quartz. The magnetic extracts from each soil sample were first observed under the stereomicroscope in order to identify the various types of magnetic particles, including impactor debris and terrestrial particles, and their overall distribution around the crater.

A selection of the magnetic particles from a representative number of samples was then studied under the Scanning Electron Microscope (SEM). Particles were first observed as whole specimens to characterize their morphology and structure. They were then sectioned in order to define their texture and overall composition. Analyses were performed using a Philips XL30 SEM, operating at 20 kV coupled with an energy-dispersive X-ray fluorescence spectrometer (EDX), available at Pisa University’s Dipartimento di Scienze della Terra, and

135 Field Emission Scanning Electron Microscope (FE-SEM) Jeol JSM 6500F (upgraded to 7000 series), available at the Istituto di Geofisica e Vulcanologia (INGV) of Rome.

The mineral chemistry of the various phases in a selection of the above particles was obtained by electron probe micro-analyses (EPMA) using a Cameca SX50 electron microprobe at IGG-CNR, Padua (Table 2). Running conditions were 15 kV accelerating voltage, 15 nA beam current, and 1 µm nominal beam spot. The manufacturer-supplied PAP procedure was employed for raw data reduction. Standards used for instrumental calibration were natural minerals (quartz, diopside, apatite and sphalerite), and pure elements (Fe, Ni, and Co). Table 2. Average major element bulk composition by EPMA of dark glass impact melt (in the form of vesicular particles, spherules and coatings), Fe-Ni oxide blebs in dark glass, Fe-oxide spherules, lechatelierite in dar glass, and Fe-Ni metal blebs in dark glass (distinguished in high-Ni and very high-Ni compositional types). Limits of detection (LOD) for the analyses of silicates and oxides are: Mg = 0.03 wt%, Al = 0.04 wt%, Si = 0.04 wt%, P = 0.05 wt%, S = 0.04 wt%, K = 0.04 wt%, Ca = 0.04 wt%, Ti = 0.04 wt%, Cr = 0.06 wt%, Mn = 0.07 wt%, Fe = 0.04 wt%, Co = 0.08 wt%, Ni = 0.04 wt%. LOD for the analyses of the metal phases are: Si = 0.03 wt%, P = 0.04 wt%, S = 0.04 wt%, Cr = 0.05 wt%, Fe = 0.04 wt%, Co = 0.06 wt%, Ni = 0.04 wt%.

Glass - Glass - Glass - Fe-oxide Fe-Ni-oxide La Fe-Ni metal blebs vesicular spherules coatings blebs spherules H- VH-

Nid Nie n. p.b 1 6 3 2 11 9 2 n. an.c 9 9 48 5 29 28 2 ox wt% σ ox wt% σ ox wt% σ ox wt% σ ox wt% σ el wt% σ el wt% MgO 0.18 0.07 0.23 0.04 0.30 0.04 0.14 0.01 0.04 0. bdl Si 0.23 0.62 0.17 01 Al2O3 10.4 2.8 12.3 1.9 16.1 1.4 4.79 2.48 0.06 0.05 0.08 P 0.67 0.57 bdl

SiO2 57.0 10.1 49.2 4.0 47.3 3.2 4.81 1.65 0.77 1.05 99.5 S 2.19 1.71 2.38

P2O5 bdl 0.09 0.05 0.13 0.03 0.57 0.62 0.10 0.11 bdl Cr bdl bdl

SO3 0.20 0.05 0.12 0.07 0.08 0.10 0.02 0.02 bdl 0.09 Fe 69.1 4.7 4.45

K2O 0.13 0.03 0.23 0.07 0.38 0.06 na na bdl Co 0.91 0.18 0.10 CaO 0.57 0.14 0.49 0.09 0.50 0.28 0.27 0.03 bdl 0.00 bdl Ni 26.3 4.5 94.2

TiO2 1.06 0.31 0.76 0.05 0.81 0.05 0.44 0.15 bdl bdl tot 99.4 101.3

Cr2O3 0.10 0.05 bdl bdl 0.07 0.05 bdl bdl MnO bdl 0.79 0.30 1.16 0.09 0.08 0.04 bdl 0.00 bdl FeO* 27.6 7.5 30.9 3.7 29.2 3.0 79.6 4.32 71.6 1.92 0.23 CoO bdl 0.10 0.05 bdl bdl 0.82 0.09 bdl NiO 0.16 0.10 2.71 1.08 1.54 0.63 0.08 0.02 20.9 1.90 bdl tot 97.4 97.9 97.5 90.9 94.3 99.9 a Lechatelierite. b Number of analysed particles. c Number of total analyses in the average. d High-Ni blebs. e Very high-Ni blebs. *All Fe as FeO. Abbreviations: na = not analyzed; bdl = below detection limit. Whole-rock major elements of unshocked target material of Kamil area (namely, sandstones ranging from very coarse quartzarenite to coarse siltstone with intercalated levels of very fine wacke; Table 3) were determined for geochemical comparison with impact glass.

136 Major elements were determined on glass beads by X-Ray Fluorescence (XRF, ARL 9400 XP spectrometer) at Pisa University’s Dipartimento di Scienze della Terra (Italy), using the procedure described by Tamponi et al. (2003). Nickel and Co were determined by Inductively Coupled Plasma Mass Spectrometry (ICPMS; Perkin Elmer NexION 300x spectrometer) available at the Dipartimento di Scienze della Terra of the University of Pisa. Table 3. Chemical composition of Kamil Crater target rocks (X-ray fluorescence analyses, except where otherwise specified). Samples were collected from the crater walls (Urbini et al. 2012; Fazio et al. 2014). Target rocks consist of layered sandstones with subhorizontal bedding belonging to the Gilf Gebir Formation. A 70 cm thick, pale quartzarenite layer (sample M26) is at the top of the impacted sedimentary sequence. The underlying rocks consists of reddish, ferruginous quartzarenite layers up to some tens of cm in thickness (sample M27). The pale quartzarenite sporadically contains pale wake with siltstone levels. A kaolinite-rich matrix occurs in the reddish quartzarenite.

Pale Pale wacke with Reddish Kaolinite-rich quartzarenite silstone levels quartzarenite, ~ 5 matrix (n=18)a vol.% matrix sample M26 M25 M27 M27

oxide wt% SiO2 99.8 87.8 95.7 43.9 TiO2 0.07 1.15 0.09 0.55 Al2O3 0.25 10.1 1.98 34.7 Fe2O3 T 0.11 0.74 1.71 1.88 MnO <0.01 0.02 0.06 b.d.l. MgO 0.18 0.14 0.25 0.22 CaO 0.05 0.11 0.31 0.05 Na2O 0.05 0.13 0.08 0.12 K2O <0.01 0.04 <0.01 0.04 P2O5 <0.01 0.06 0.04 0.09 sum 100.5 100.3 100.2 81.5 L.O.I. 0.65 3.91 1.19

element µg g-1 Nib 7 15 Cob <1 12

a Average of EPMA analyses. b Analyses by ICP-MS.

Fe2O3 T: total iron as Fe2O3. L.O.I.: Loss On Ignition. In order to quantify the amount of the iron impactor debris and to define its distribution around the crater, we determined the concentration of a number of elements in aliquots of the magnetic extracts (obtained using a small stainless steel splitter), including the expected tracers of the iron impactor Fe, Ni, Co, P (Table 4). Elemental concentrations were determined through the total dissolution method (HF + HNO3 + HClO4 + HCl) followed by Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) analyses at ACTALABS. Due to the high concentration of Ni (>10000 µg g-1) in samples S23 and S24, they were re-run for siderophile elements at an appropriate dilution using ICPS-MS, at the Dipartimento di Scienze della Terra of the University of Pisa. Note that prior to grinding and

137 powdering of the magnetic extract for geochemical analyses, we separated, counted and weighed Gebel Kamil shrapnel down to 1 mm in size from each magnetic sample under the stereo microscope (the two fractions are denominated "shrapnel fraction" and "soil fraction" in Table 4). This procedure was adopted in order to avoid problems in powdering and dissolving relatively large metal particles, and discuss the distribution and mass of the impactor debris in the soil with and without mini shrapnel.

RESULTS

Petrography of magnetic particles

A first survey of the magnetic extracts under the stereomicroscope and the microanalytical SEM-EDS revealed a systematic difference in the distribution around the crater of the various types of magnetic particles. The magnetic extracts from samples collected in the E to SW sector are dominated by dark Fe-Ni glass in a variety of structures and occurrences, plus minor Fe-Ni oxide spherules and Gebel Kamil shrapnel (Fig. 2c). Magnetic extracts from the NW sector are dominated by amphibolite gneiss particles (Fig. 2d) belonging to the Precambrian Basement that crops out 100-150 m due NW of the crater rim (Urbini et al. 2012). In the other sectors, the magnetic extracts are dominated by crystals or crystal fragments of terrestrial Fe- and Fe-Ti oxides, garnet and minor cosmic spherules mainly belonging to the barred olivine - type (Fig. 2e). Although aeolian quartz grains and sandstone particles are ubiquitous, they are more abundant in the samples collected due W of the crater. Although we cannot rule out that these particles could result from inefficient magnetic extraction, note that they often contain tiny inclusions of ferro-magnetic minerals or they are coated by ferruginous deposits. Furthermore, we shall see that the external surfaces of the aeolian quartz grains and sandstones particles from the E-to-the-S sectors are speckled with tiny impactor splashes of dark Fe-Ni dark impact melt.

In the following two sections, we will focus on Fe-Ni particles, i.e. dark Fe-Ni impact melt glass and Fe-Ni oxide spherules. Due to their Ni-rich composition, we shall see that they are the only microscopic tracers of the Ni-rich impactor debris. For details on Gebel Kamil shrapnel the reader should refer to D'Orazio et al. (2011).

Dark Fe-Ni impact glass

Dark Fe-Ni glass occurs in the form of highly vesicular particles (Fig. 3), spherules (Fig. 4) and coatings (Fig. 5).

138 Dark glass vesicular particles are irregular masses characterized by delicate reticulite to pumiceous microstructures (Figs. 2 and 3). Under the stereomicroscope they are brown to dark green. The composition of the glass (Table 2) is slightly variable and dominated by SiO2

(57.0 ± 10.1 wt%), FeO (27.6 ± 7.5 wt%) and Al2O3 (10.4 ± 2.8 wt%), with subordinate TiO2 (1.06 ± 0.31 wt%) and NiO (0.16 ± 0.10 wt%). Abundant oxide blebs up to tens of micrometers in size float within the glass (Fig. 3). They consist mainly of micrometer-sized Fe-oxides and are here referred to as "Fe-oxide blebs". Their bulk composition (Table 2) is dominated by FeO (79.6 ± 4.32 wt%), SiO2 (4.81 ± 1.65 wt%) and Al2O3 (4.79 ± 2.48 wt%) with traces of NiO (0.08 ± 0.02 wt%) and P2O5 (0.57 ± 0.62 wt%). Fe-Ni metal blebs, here referred to as "Fe-Ni metal blebs", are sometimes also observed in the dark glass. They contain abundant Ni (26.3 - 94.2 wt%; Table 2), 4.45 - 69.1 wt% Fe, minor S (~2 wt%), and traces of P (~0.6 wt%), Co (<1.0 wt%) and Si (<0.2 wt%), suggesting that there was some mixing with the target material. Both types of blebs show nanometer-sized cellular intergrowths of - as yet - unidentified phases indicative of eutectic melts (Fig. 3a, inset). Submillimeter-sized, bedrock clasts (mainly sandstones) may also be found embedded in the dark glass (Fig. 3a).

Fig. 3. Back scattered electron images of dark impact melt glass in the form of the vesicular masses from sample S23. A) A highly vesicular particle consisting of a tenuous reticular network of glass. Oxide blebs up to tens of microns in size enriched in Fe, Ni, ± P, ± S float within the host glass (inset). They consist mainly of micron- sized oxides and are here referred as "Fe-oxide blebs". Less frequently, Ni-rich metal blebs occur and are here referred as "Fe-Ni-metal blebs". A target rock clast (shocked sandstone) embedded in glass is arrowed. B) A detail of another pumiceous particle showing abundant Fe-oxide and Fe-Ni metal blebs. Abbreviations: CL: target rock clast; DG: dark glass, Fe-Ni bl: Fe-oxide and Fe-Ni metal blebs. The dark glass spherules (Fig. 4) are shiny and black under the stereomicroscope and show a number of splash form morphologies similar to those observed at Barringer Crater (e.g., Hörz et al. 2002): from ovoids and irregular dumbbell to nearly perfect spherules, in order of decreasing abundance. Some are delicately attached to each other and form

139 compound spherules (Fig. 4b). Others are partially covered by splashes of other spherules (Fig. 4c). The vesicularity of the dark glass varies from moderate to negligible (compare Fig. 4a and Fig. 4c). The external surfaces of the most vesicular particles are characterized by the occurrence of boiling and degassing microstructures including quenched bubbles and glass filaments (Fig. 4d). The filaments originate from the vesicles and are hollow. The external surfaces of the spherules are also finely decorated by dendrites of unidentified Fe-rich oxides <1 µm in size embedded in glass, often arranged in schlieren (Figs. 4a, 4d, 4e). Sectioned spherules (Fig. 4f) show the characteristic vesicularity of the dark glass, its compositional heterogeneity and the occurrence of several mineral and lithic inclusions. Inclusions mainly consist of bedrock fragments (quartzarenite), high temperature (T >1,500 °C) silica glass or lechatelierite and oxide and metal blebs (as described above). The heterogeneity of the glass is highlighted by schlieren and haloes with variable Si/Fe ratios around the inclusions. The composition of the glass (Table 2) is slightly variable, dominated by SiO2 (49.2 ± 4.0 wt%),

FeO (30.9 ± 3.7 wt%), and Al2O3 (12.3 ± 1.9 wt%), with subordinate NiO (2.71 ± 1.08 wt%) and TiO2 (0.76 ± 0.01 wt%).

Fig. 4. Back scattered electron images of dark glass spherules from sample S23. A) An ovoid spheroid showing abundant vesicles and a large glass bubble. B) Three spheroids stuck together to form a compound particle. C) A Fe-rich liquid splashed onto a glass spherule. D) A detail of the typical glass spheroid surface showing, trails (schlieren) of Fe-rich oxides, glass bubbles and glass filaments. The insets show that the filaments originate from the vesicles and are hollow. E) A sectioned spherule showing the characteristic abundant vesicularity of the dark glass, its compositional heterogeneity and the occurrence of several mineral and lithic inclusions. Inclusions mainly consists of fragments quartzite, lechatelierite (vesicular) and Fe-oxide and Fe-Ni-metal blebs. The heterogeneity of the glass is highlighted by schlieren and haloes around the inclusions. Abbreviations: B: glass bubble; F: hollow glass filament; Fe-Ni B: Fe-oxide and Fe-Ni metal blebs; L: lechatelierite; Ox: oxides; Qtz: quartz, Sch: schlieren. Dark glass coatings (Fig. 5) consist of mantles of shiny black glass continuously (or nearly so) enveloping small target lithic fragments (mainly quartzarenite) and lechatelierite.

140 They are up to some 100 µm in thickness. The morphological, textural, microstructural and compositional characteristics of their external surfaces are similar to those observed in dark glass spherules (see above). Sectioned particles show that the glass coating is vesicular and envelops lithic fragments made by variably shocked and melted target quartzarenite, i.e. lechatelierite (Fig. 5d). The composition of the glass (Table 2) is slightly variable, dominated by SiO2 (47.3 ± 3.2 wt%), FeO (29.2 ± 3.0 wt%), and Al2O3 (16.1 ± 1.4 wt%), with subordinate NiO (1.54 ± 0.63 wt%) and TiO2 (0.81 ± 0.05 wt%). Similar glass coatings have been reported from Wabar (e.g., Mittlefehldt et al. 1992) and lunar samples (e.g., Apollo 16 lunar samples; Grieve and Plant 1973) and, at a cm-scale, in impact melt lapilli and bombs at Barringer Crater (e.g., Niniger 1954) and Kamil Crater (Fazio et al. 2014).

141

Fig. 5. Stereomicrographs (A, B) and back scattered electron images (C, D) of dark impact melt coatings from sample S23. A) A whole particle coated with a shiny dark bubbly glass. B) A broken particle showing their typical structure: a dark glass coating few tens of micron in thickness continuously envelops a target rock particle (in this case an aggregate dominated by quartz and colorless glass). C) A similar broken particle characterized by abundant glass bubbles at surface. Inset: a detailed view of the contact between the glass coating and the internal target rock particle (in this case, a quartz grain). D) A sectioned particle showing that the glass coating is higly vesicular and envelops a particle made by a highly shocked quartzarenite. Inset: a close up view of a quartz grain bearing 3 sets of PDF (at least one partially melted) and interstitial lechatelierite. Abbreviations: B: glass bubbles; DG: Dark impact melt glass; PDF: planar deformation features; L: lechatelierite; Qtz: quartz.

142 Fe-Ni spherules

Fe-Ni spherules (Fig. 6) are nearly perfect spheres typically ranging from 70 to 150 µm in diameter. Under the stereomicroscope they are black with metallic luster, and characteristic polygonal ridges on the external surface (Fig. 6a). Sectioned particles reveal that Fe-Ni spherules consist mainly of interlocking crystals of at least two Fe-Ni-rich oxides yet to be determined. The less Ni-rich oxide, which appears darker in backscattered electron images (Fig. 6b), is more abundant; it is most likely Ni-rich magnetite hosting very fine dendritic intergrowths of another phase yet to be determined. The more Ni-rich oxide, which appears brighter in backscattered electron images (Fig. 6b), is more abundant towards the cores of the spheres. Relatively large voids are ubiquitous at the core of the spherules. They resemble contraction voids commonly observed in I-type cosmic spherules (i.e., molten micrometeorites consisting mainly of Fe-oxides; Genge et al. 2008) likely resulting from the rapid crystallization of the melt from the surface inwards (Feng et al. 2005), rather than vesicles. The bulk composition of the Fe-Ni oxide spherules (Table 2) is dominated by FeO (71.6 ± 1.92 wt%) and NiO (20.9 ± 1.90 wt%).

143

Fig. 6. Back scattered electron images of Fe-Ni oxide spherules from sample S23. A) A whole sphere showing characteristic polygonal ridges on the external surface. B) A section of the same sphere showing intergrowth of a more Ni-rich and a less Ni-rich oxides (pale and dark grey, respectively). Note the high vesicularity. Inset: fine intergrowth textures within the less Ni-rich oxide. Abbreviations: Fe-Ni Ox: Fe-Ni oxides; I: fine intergrowth; V: voids. Soil particles speckled with dark glass splashes and Fe-Ni oxide spherules

An important fraction of the magnetic extract consists of soil particles (including quartzarenite and wacke fragments, aeolian sand, etc.) whose external surfaces are speckled with tiny droplets and splashes of dark glass, Fe-oxide blebs, Fe-Ni metal blebs and Fe-Ni oxide spherules (Fig. 7). These magnetic particles are abundant in the proximity of the crater and farther away in the E to SSW sectors.

144

Fig. 7. Back scattered electron images of a soil fragment from sample S24. A) The soil fragment is speckled with tiny droplets of dark glass and Fe-Ni spherules (plus splashes). The dashed area is the surface detailed in B). B) A close up view of the same particle. Abbreviations: CM: mostly clay material; DG sph: dark glass spheres; DG spl: dark glass splashes; Fe-Ni bl: Fe-Ni blebs; Fe-Ni sph: Fe-Ni spherules. Main geochemical features of the dark glass

Although a detailed geochemical analysis of the impact melt glasses at Kamil Crater will be discussed in a separate paper, we briefly focus here on three main geochemical results.

The composition of the dark glass is heterogeneous at the sub-micrometer scale but its bulk composition is similar in the three occurrences (Table 2). It is essentially a mixture of three components: SiO2 (from ~47 to ~57 wt%), FeO (from ~28 to ~31 wt%) and Al2O3 (from ~10 to ~16 wt%). The Fe content, which is negligible in the target, derives from the iron impactor. The Si and Al contents, elements which are absent in the impactor, derive from the target. Bedrock quartzarenite and wacke may contain cm-thick levels of siltstones enriched in Al and kaolinite-rich matrix (Table 3). Figure 8a shows that the bulk composition of the dark glass in the various occurrences tends to form a compositional continuum between the two source materials: the target (Si-rich, plus Al) and the projectile (Fe-rich).

Figure 8b shows that the Ni/Fe ratio in the dark glass is lower than that of Gebel Kamil, whereas the Ni/Fe ratio tend to be higher in coexisting Fe-Ni-metal blebs. We explain this trend through selective fractionation of Fe and Ni into the dark glass and coexisting Fe- Ni-metal blebs, respectively, as a result of preferential Fe oxidation of a melt containing a 0 Fe,Ni projectile fraction. Iron is preferentially oxidized relative to Ni as ΔG Fe/FeO << 0 0 0 ΔG Ni/NiO, where ΔG Fe/FeO and ΔG Ni/NiO are the Gibbs free energy variations in the reactions

Fe + 0.5 O2 = FeO and Ni + 0.5 O2 = NiO; furthermore, regardless of P, T and fO2, their partition coefficients between a metallic and a silicate melts are DNi met/sil >> DFe met/sil (e.g., Righter et al. 1997). Similar geochemical trends were observed in impact melt glass at Barringer and Wabar (Mittlefehldt et al. 1992; Hörz et al. 2002; Hamann et al. 2013), emphasizing the importance of this chemical process during hypervelocity impacts. Figure 8b also shows that the very high-Ni metal blebs have distinctly higher Ni/Fe ratios relative to the

145 coexisting high-Ni oxide blebs, indicating extreme Ni fractionation into the metallic phases. More systematic EPMA work is required to verify a possible compositional continuum between the two bleb populations. The lack of Ni in Fe-oxide blebs suggests that they formed through the melting of accessory minerals like Fe-oxides, described by (Fazio et al. 2014) in the target rocks.

The Fe-Ni spherules are the product of the melting of the iron projectile, as they have similar Ni/Fe and Ni/Co ratios to that of Gebel Kamil (Figs. 8b and 8c). Their high Ni concentrations (NiO = 20.9 ±1.9 wt%; Table 2) are different from those typically observed in I-type micrometeorites (NiO = 2.75 ± 0.54; Folco and Cordier 2014, in press) and are thus not related to the sporadic flux of micrometeorites. Fe-Ni spherules from Kamil Crater are similar to Fe-Ni spherules found at the Sikhote Alin (Russia) crater field and interpreted as ablation spherules (Badyukov and Rajtala 2012). According to Badyukov and Rajtala (2012) the spherules formed late during the flight of the Sikhote Alin iron meteorite through the melting of debris produced during its disintegration. Further investigation is required to establish whether the Fe-Ni spherules from Kamil Crater are impact melts of the projectile or ablation debris formed during low-altitude atmospheric flight (i.e., shortly before the impact). In any case, they are impactor debris.

Fig. 8 (next page). Compositional diagrams of the dark glass and Fe-Ni spheres. A) Aluminum-Si-Fe (wt%) triangular plot. Dark glass is essentially a mixture of target material (mainly sandstones dominated by quartz) and the iron projectile Gebel Kamil. The vesicular glass tends to be enriched in target material. B) Ni versus Fe showing Fe-Ni fractionation trends between dark glass and coexisting Fe-oxide and Fe-Ni metal blebs (see text for details). Fe-Ni spheres plot close to the Gebel Kamil Ni/Fe ratio, documenting their parentage with the impactor. Also their high Ni contents rules out a origin as I-type micrometeorites (data from Folco and Cordier, 2014, and references therein). C) Ni versus Co. The Ni/Co ratios of the Fe-Ni spheres is distinct from those of I- type micrometeorites and similar to that of Gebel Kamil confirming their parentage.

146

147 Bulk chemistry of the magnetic extract

The concentrations of Fe, Ni, Co, P, Mg, Ca, Ti and V in the magnetic extracts (without shrapnel) are given in Table 4.

Figure 9a shows an excellent correlation between Ni and Co with a Ni/Co ratio that matches that of the iron impactor Gebel Kamil (Ni/Co in Gebel Kamil = 29.8 ± 2.5; average Ni/Co in magnetic extracts 25.8; Table 4). This indicates that the source of Ni and Co in the magnetic extracts is Gebel Kamil (D'Orazio et al. 2011) and can be satisfactorily used as tracer of the microscopic impactor debris around Kamil Crater. There is no correlation between Ni and other major elements present in Gebel Kamil, like Fe and P, even if the samples with higher Ni concentrations contains higher Fe and P (Figs. 9b and 9c). The lack of correlation indicates that Fe and P have an additional source other than Gebel Kamil, most likely the terrestrial magnetic minerals and target rock fragments mentioned above (Fe-Ti oxides, garnet, etc.) The lack of correlation between Ni and Mg (Fig. 9d) rules out an additional source of Ni from magnetic minerals from the crystalline basement (which crops out due NW of the crater; Urbini et al. 2012), as well as a contribution from Mg-rich olivine and glass from micrometeorites (e.g., Cordier et al. 2011).

Fig. 9. Major, minor and trace elements variation diagrams of the magnetic extract (< 5 mm) of soil samples (aliquots without mini shrapnel) from Kamil Crater. A-D) Ni versus Co, Fe, P, Mg, respectively. The dashed line in A-C is the Gebel Kamil ratio. The composition of the target rock (mainly sandstones) is shown for comparison. Note the very good correlation between Ni and Co with a Ni/Co ratio that is similar to that of the iron impactor Gebel Kamil, thereby indicating that the source of Ni and Co in the magnetic extracts is Gebel Kamil.

148 Table 4. Magnetic extract (<5 mm) of soil samples from Kamil Crater: geochemical composition by ICP-AES except for Ni in samples S23 and S24 which was determined by ICP-MS (see text for details), and contents of impactor equivalent mass. ICP-AES LOD for Fe, Ni, Co, P, Mg, Ca, Ti and V are: Fe = 0.01 wt%, Ni = 1 µg g- 1, Co = 1 µg g-1, P = 0.001 wt%, Mg = 0.01 wt%, Ca = 0.01 wt%, Ti = 0.01 wt%, V = 2 µg g-1. Samples are listed according to the bearing of the profiles starting from the north, and to their increasing distance from crater rim as in Table 1. The bulk chemical composition of Gebel Kamil by Hand-Held X-Ray Fluorescence (HH-XRF; Gemelli et al. 2015) is reported for comparison.

Label Shrapnel Soil Fe Ni Co P Mg Ca Ti V Impactor Total Ni Impactor fraction fraction wt% µg g-1 µg g-1 wt% wt% wt% wt% µg g-1 equivalent µg g-1 equivalent mass (g)a mass (g)b (soil) (soil) (soil) (soil) (soil) (soil) (soil) (soil) mass (g) (soil plus mass (g) e (soil) c shrapnel) d

N profile - S12 - 4.53 18.5 293 65 0.023 1.85 1.34 0.36 118 0.007 293 0.007 S13 - 6.62 10.8 115 39 0.030 1.26 2.01 0.17 41 0.004 115 0.004 S14 - 7.77 9.99 85 32 0.030 1.08 2.01 0.20 53 0.003 85 0.003 S15 - 8.77 9.03 70 26 0.029 1.09 2.26 0.15 48 0.003 70 0.003 S16 - 13.27 8.53 86 29 0.035 1.28 2.30 0.15 50 0.006 86 0.006 NE profile S35 - 2.77 10.9 227 39 0.041 1.39 2.74 0.17 56 0.003 227 0.003 S36 - 10.49 7.75 69 25 0.031 1.04 1.87 0.26 118 0.004 69 0.004 S37 - 18.37 10.9 102 53 0.041 1.67 2.76 0.19 98 0.009 102 0.009 S38 - 10.97 9.86 96 34 0.030 1.60 2.25 0.13 57 0.005 96 0.005 S39 - 9.10 20.7 118 33 0.012 1.05 0.48 0.20 148 0.005 118 0.005 E profile S01 0.359 14.87 8.15 4830 200 0.032 0.59 1.21 0.27 65 0.359 9430 0.718 S02 - 11.75 9.14 355 48 0.036 1.04 1.87 0.21 91 0.021 355 0.021 S03 - 13.27 10.6 136 34 0.032 1.19 1.87 0.13 65 0.009 136 0.009 S04 - 3.97 12.6 112 40 0.030 1.35 1.86 0.19 66 0.002 112 0.002 S05 - 3.97 10.4 113 34 0.022 1.47 1.62 0.19 61 0.002 113 0.002 S06 - 23.07 11.1 105 27 0.017 0.89 1.54 0.21 106 0.012 105 0.012 SE profile S23 1.388 10.68 12.4 15871 624 0.060 0.21 0.35 0.44 116 1.032 37058 2.236 S24 0.838 20.83 9.28 10927 428 0.050 0.50 1.17 0.37 82 1.149 18238 1.976 S25 0.864 17.30 17.7 1290 68 0.025 0.71 0.77 0.22 154 0.112 10743 0.976 S26 - 17.10 15.1 218 48 0.028 1.28 1.33 0.21 126 0.019 218 0.019 S27 - 9.77 16.4 179 50 0.016 1.87 0.83 0.23 140 0.009 179 0.009 S28 - 10.09 11.4 121 33 0.027 1.26 1.82 0.15 97 0.006 121 0.006 S profile S17 0.036 10.23 22 6200 305 0.078 0.33 0.33 0.49 256 0.317 6877 0.353 S18 2.749 22.55 9.36 4870 205 0.046 0.98 1.96 0.19 73 0.549 26071 3.298 S19 1.775 11.89 16.4 8840 367 0.070 1.04 1.04 0.42 205 0.526 33668 2.301 S20 0.142 15.42 14.4 3000 161 0.037 0.60 0.34 1.49 352 0.231 4802 0.374 S21 0.113 7.45 12 342 47 0.058 1.16 2.29 1.74 373 0.013 3328 0.126 S22 - 7.47 12.9 344 44 0.019 1.12 0.55 0.20 134 0.013 344 0.013 SW profile S29 - 11.67 15.5 898 73 0.046 1.06 1.89 0.15 67 0.052 898 0.052 S30 - 6.40 10.9 2030 109 0.033 1.06 1.59 0.22 74 0.065 2030 0.065 S31 0.414 12.25 9.45 586 46 0.037 1.14 2.63 0.12 69 0.036 7096 0.449 S32 0.005 5.26 11.3 184 48 0.039 1.79 1.73 0.19 80 0.005 382 0.010 S33 - 6.07 11.2 247 39 0.034 1.20 2.29 0.14 53 0.007 247 0.007 S34 - 13.67 11.5 145 42 0.046 0.96 0.67 2.41 403 0.010 145 0.010 W profile S07 - 2.87 20.1 980 76 0.048 0.50 0.76 0.51 256 0.014 980 0.014 S08 - 5.93 16.4 241 61 0.029 1.08 1.38 0.18 65 0.007 241 0.007 S09 - 5.67 10.8 227 37 0.021 0.94 1.49 0.16 41 0.006 227 0.006 S10 - 7.87 7.57 65 30 0.061 1.03 2.41 1.79 266 0.003 65 0.003 S11 - 8.57 12.2 113 37 0.030 1.49 2.09 0.12 87 0.005 113 0.005 NW profile S41 - 3.20 17.1 121 51 0.031 1.39 2.05 0.14 45 0.002 121 0.002 S43 - 5.20 10.7 88 30 0.031 1.23 2.72 0.18 74 0.002 88 0.002 S44 - 7.60 10.4 128 38 0.034 1.97 2.50 0.19 56 0.005 128 0.005

Fe s Ni s Co s P s S s Gebel Kamil 78.5 0.2 20.6 0.2 0.69 0.05 0.04 0.01 0.02 0.07

a Total mass in grams of shrapnel (1 to 5 mm in size) separated from the magnetic extract under the stereomicroscope. b Total mass in grams of the soil fraction of the magnetic extract without shrapnel found under the stereomicroscope. c Total equivalent mass in grams of the impactor in the soil from the magnetic extract, i.e., it does not account for the contribution of shrapnel. d Total Ni concentration in µg g-1 in the bulk sample, calculated from the contributions of shrapnel and the soil fractions of the magnetic extract. For the latter we assumed a concentration of Ni = 20 wt% (D'Orazio et al. 2011) in the impactor. e Total equivalent mass in grams of the impactor in the bulk sample calculated from "d", assuming a concentration of Ni = 20 wt% in the impactor (Gemelli et al. 2015).

149 Ni distribution in the soil around Kamil Crater and mass of the impactor debris

We shall present here the Ni distribution in the soil around Kamil Crater distinguishing the fractions with and without mini shrapnel (Table 4). The separation of mini shrapnel resulting from the analytical issues discussed above, enabled us to study possible differences in the two types of fractions. The Ni content of the fraction with mini shrapnel was calculated by summing the Ni contribution of the mini shrapnel, assuming a Ni content of ~20 wt% in Gebel Kamil (D'Orazio et al. 2011).

The Ni contents in the fractions without mini shrapnel ranges from 65 to 19,322 µg g-1 (Table 4). Fig. 10a shows that Ni concentrations >1000 µg g-1 occur in the samples from the E, SE, S and SW profiles with concentrations decreasing with distance from the crater rim. Samples from the other profiles have Ni concentrations <200 µg g-1. Given that the typical Ni concentration in the continental crust is <105 µg g-1 (Rudnick and Gao 2004), we conservatively assume that the background Ni content in the target rocks at Kamil Crater is likely <200 µg g-1, whereas higher concentrations are the signature of the impactor debris. Note that this background threshold implies a negligible loss of <1% of the total Ni content in the studied samples. The maximum Ni contents were observed in samples S23 and S24 (Table 4) collected within 100 m of the crater rim along the SE profile. Other high Ni concentrations are observed in the S profile with a maximum 200 m from the crater rim (sample S19; Table 4) and extending up to ~800 m from the crater rim (sample S20; Table 4).

By adding the Ni contribution of the separated mini shrapnel (Fig. 10b; Table 4), the distribution does not change substantially, but important Ni concentrations are observed due S and SW of the crater with a maximum ~200 m from the crater rim.

150

Fig. 10. Ni concentrations of the magnetic extracts versus distance from crater rim along the different sampling profiles. A) Trend lines for aliquot without mini shrapnel. B) Trend lines for the same samples including mini shrapnel. Concentration maps (impactor mass per unit soil surface) of the microscopic impactor debris are shown in Fig. 11. Impactor debris is mainly dispersed due SW-SE of the crater in an area of ~300,000 m2 that tails off southwestwards for up to ~400 m from the crater rim. The integrated mass of Gebel Kamil debris in the form of dark glass and Fe-Ni spherules is ~120 kg (with an error of 20% relative due to extraction efficiency). The addition of the mini shrapnel totals ~290 kg. Note that due to extraction efficiency discussed above, these mass estimates should be considered as a first-order approximation. Comparison between Fig. 11a and Fig. 11b highlights a significant shift from SE to SW in the location of the maximum concentration of the microscopic impactor debris around Kamil Crater with the addition of mini shrapnel in the count. This difference may be explained assuming a more heterogeneous

151 distribution of mini shrapnel relative to that of the dark glass due their different origins, i.e., solid-state fragmentation at the rear of the projectile versus impact melting at the front of the projectile during contact and excavation, respectively.

Fig. 11. Concentration maps of the microscopic impactor debris around Kamil Crater. A) Calculated mass distribution of the microscopic impactor debris of the soil without mini shrapnel. The total equivalent mass of the impactor equivalent is ~120 kg. B) Calculated mass distribution of the microscopic impactor debris of the soil with mini shrapnel. The total equivalent mass of the impactor is ~290 kg.

152 DISCUSSION AND CONCLUSIONS

Ejecta curtain

The hypervelocity impact of the Gebel Kamil iron meteorite produced an ejecta curtain containing microscopic impactor debris due SE-SW of the crater. Such dispersion is consistent with a downrange expansion since it extends along almost the same direction as the macroscopic shrapnel scatter field, which is considered the strongest indicator of a moderately oblique impact from the NW (D'Orazio et al. 2011; Urbini et al. 2012).

The microscopic impactor debris (i.e, the <5 mm-sized magnetic fraction of the ejecta curtain) consist mainly of shock melt particles. The shock melt is a mixture of a Si- and Al- rich target and Fe-Ni-rich projectile materials. The shock melt consists mainly of dark glass and occurs in the form of vesicular masses, spherules and coatings enfolding variably shocked fragments of the target rocks. Subordinately, the microscopic impactor debris in the ejecta curtain contains Fe-Ni oxide spherules. Whether the latter are impact melt products or ablation spherules is yet to be determined. The microscopic particles formed a hot (since the melting of silica glass and iron meteorite requires temperatures higher than ~1,700 °C), dense and turbulent spray in the ejecta curtain, as testified by the abundant dark glass particles found stuck to soil particles (Fig. 7), and by the abundant compound glass spherules (Fig. 4). The ejecta curtain containing microscopic shock melt particles was mainly dispersed due SW-SE of the crater in an area of ~300,000 m2 extending up to ~400 m from the crater rim (Figs. 10 and 11).

Although the ejecta has an asymmetric, downrange distribution, we can generalize that the microscopic impactor debris in the ejecta of small scale impact craters such as Kamil Crater is dispersed for only some crater diameters, and that the distribution is one order of magnitude smaller than that of the macroscopic shrapnel (i.e., masses >10 g), which at Kamil were found up to 1.6 km E of the crater rim (D'Orazio et al. 2011). The latter difference could be explained on considering that atmospheric drag decelerates small masses more efficiently than large masses, in agreement with recent models (Shuvalov and Dypvik 2013).

The impact melt products observed at Kamil Crater are similar to those observed at Wabar (e.g., Mittlefehldt et al. 1992) and Barringer Crater (e.g., Niniger 1956; Rinehart 1958) which were also generated by the hypervelocity impacts of iron projectiles on sedimentary targets including sandstone, dolostone and loose sand. This documents that these impact

153 cratering products can be found around small terrestrial craters with diameters ranging from several kilometers down to tens of meters.

Our estimate for the mass of the microscopic impactor debris at Kamil Crater (~120 - 290 kg, without and with mini shrapnel, respectively) is a small fraction (~1.5 - 3.5%) of the estimated impactor mass in the form of macroscopic shrapnel (~8,000 kg; see below: Table 5). Numerical models for the larger Barringer Crater (~1.2 km diameter; iron impactor; sedimentary target; oblique impact; Artemieva and Pierazzo 2011) indicate much higher fractions (45%). We thus infer that molten microscopic impactor debris decreases significantly with decreasing crater diameter, whereas the production of macroscopic shrapnel increases. This is likely a consequence of projectile deceleration in the atmosphere. While in large craters impact velocities are the same as preatmospheric velocities, in small craters impact velocities are defined mainly by atmospheric entry (e.g., Melosh 1989). Table 5. Impactor mass.

Mass source Method m (kg) Reference ç Regmaglypted specimens (individual) visual search 83 D'Orazio et al. (2011)

Shrapnel mass recovered on surface >10 g visual search 700 D'Orazio et al. (2011)

Shrapnel mass weighed on surface (>10 g) visual search 1500 D'Orazio et al. (2011) ç Shrapnel mass on surface ( >10 g) estimate from density maps 3400 D'Orazio et al. (2011) ç Shrapnel mass buried in ejecta (>100 g) magnetic survey 1700* Urbini et al. (2012) ç Impactor mass in impact melt debris in soil (<5 mm) geochemistry 120 This work

Impactor mass in impact melt masses (>5 mm) nd ç Shrapnel mass on surface (<10 g) estimate from mass distribution 2250 This work ç Shrapnel mass buried in ejecta (<100 g) estimate from mass distribution 730 This work Total 8283

ç Masses in the total. * This is the maximum value of the 1100 - 1700 kg mass range estimated by Urbini et al. (2012). Abbreviations. nd: not determined Impactor trajectory

The distribution of the microscopic impact debris around Kamil Crater is concentrated due SWSE of the crater. Overall, this asymmetric distribution is consistent with other findings, most notably the distribution of the macroscopic shrapnel (D'Orazio et al. 2011) indicating that the iron bolide approached the Earth’s crust from the northwest (305-340° N), travelling along a moderately oblique trajectory. However, the distribution of the microscopic impactor debris is shifted south-southwestwards relative to that of macroscopic shrapnel. This deviation suggests that the microscopic impactor debris in the ejecta curtain was somewhat wind-blown by dominant north-northeasterly winds during deposition. In other words, after

154 initial ballistic ejection from the launch point, air drag decelerated the microscopic impactor debris in the ejecta curtain until its motion was controlled by air flow.

Alternatively, the microscopic impactor debris distribution is, at least in part, secondary, possibly due to some south-southwestward aeolian transport after deposition or occasional water runoff (see wadis in Fig. 11) after deposition. The lack of erosion features in the delicate microscopic dark glass particles (e.g., Figs. 3-5) favors the former hypothesis.

In any case, evidence from Kamil Crater documents that, compared to the distribution of macroscopic shrapnel, the distribution of microscopic impactor debris is a weaker criterion for establishing the direction of impact.

Impactor mass in microscopic impact melt and total impactor mass

The present study allows us to further constrain the impactor mass at the time of impact (Table 5). The mass of the impactor in the form of <5 mm impact melt debris (impact melt particles without mini shrapnel) estimated through concentration maps is ~120 kg. The total mass of the impactor in the form of <5 mm debris (impact melt particles plus mini shrapnel) estimated through concentration maps is ~290 kg. The total mass of macroscopic shrapnel >10 g inferred from concentration maps of specimens found on the surface around the crater during systematic search is 3,400 kg (D'Orazio et al. 2011). The total mass of shrapnel >100 g buried in the ejecta blanket inferred through a geomagnetic survey is in the 1,100 - 1,700 kg range (Urbini et al. 2012).

Two size fractions of shrapnel must be now included in the count: shrapnel <10 g on the surface and those <100 g buried in the ejecta. In order to provide an estimate of the contributions of the first two shrapnel fractions above, we determined the shrapnel mass distribution (Fig. 12). The distribution is based on counts of specimens >10 g found in the field by visual search. Assuming that the logarithmic distribution applies to the two missing mass fractions, they should total ~2,250 kg and ~730 kg, respectively (Table 5). In addition, the count should include the total mass of the impact melt bombs and lapilli of dark glass (impactor-rich glass) scattered in and around the crater (Folco et al. 2011; Fazio et al. 2014). Field observations indicate that this fraction should amount to a total mass comparable to that of microscopic impact melt impactor debris.

155

Fig. 12. Shrapnel mass distribution used in this work for an estimate of the total shrapnel mass <10 g on the surface and <100 g buried in the ejecta. The distribution is based on counts of specimens >10 g found in the field by visual search. We thus speculate that the total mass of Gebel Kamil on impact with the target was close to 10 t. This is consistent with estimates of its pre-atmospheric mass, namely >20 t, most likely 50-60 t (Ott et al. 2014). The estimated mass of the projectile at the contact is also consistent with calculations by Folco et al. (2011) and Urbini et al. (2012) by applying cratering models from literature (i.e., Collins et al. 2005) to Kamil Crater. Scaling relations predict that a crater approximately 45 m in diameter can be generated in a sedimentary target by iron projectiles ranging from approximately 7 to 20 t in mass, assuming meteoroid entry velocities ranging from 12 to 18 km s-1 and entry angles ranging from 30° to 45°.

Acknowledgements – This work was supported by the Italian Ministero degli Affari Esteri – Progetti di Grande Rilevanza, Protocollo Esecutivo ITALIA-EGITTO. The 2010 geophysical expedition work was carried out within the framework of the 2009 Italian- Egyptian Year of Science and Technology. We thank Prof. M. Alsherbiny (former President of the Egyptian National Academy for Scientific Research and Technology) and Prof. F. Porcelli (Scientific Attaché, Italian Embassy, Egypt) for diplomatic and institutional support; the Egyptian Army for logistical support. Luigi Folco and Massimo D'Orazio are also supported by the University of Pisa Fondi di Ateneo, and Agnese Fazio by the Barringer Family Fund for Meteorite Impact Research 2014. ISterre is part of Labex OSUG@2020 (ANR10 LABX56). We thank F. Hörz and M. Ebert for constructive reviews, and N. Artemieva for editorial handling.

156 REFERENCES

Artemieva N. and Pierazzo E. 2009. The Canyon Diablo impact event: 2. Projectile motion through the atmosphere. Meteoritics & Planetary Science 44:25–42. Artemieva N. and Pierazzo E. 2011. The Canyon Diablo impact event: Projectile fate and target melting upon impact. Meteoritics & Planetary Science 46:805–829. Cordier C., van Ginneken M., Folco L. 2011. Nickel abundance in stony cosmic spherules: constraining precursor material and formation mechanisms. Meteoritics & Planetary Science 46:1110–1132. D’Orazio M., Folco L., Zeoli A., Cordier C. 2011. Gebel Kamil: the iron meteorite that formed the Kamil Crater (Egypt). Meteoritics & Planetary Science 46:1179–1196. Earth Impact Database. http://www.passc.net/EarthImpactDatabase/index.html. Accessed August 14, 2014. Ebert M., Hecht L., Deutsch A. and Kenkmann T. 2013. Chemical modification of projectile residues and target material in a MEMIN cratering experiment. Meteoritics & Planetary Science 48: 134–149. Fazio A., Folco L., D'Orazio M., Frezzotti M. L., Cordier C. 2014. Shock metamorphism and impact melting in small impact craters on Earth: Evidence from Kamil Crater, Egypt. Meteoritics & Planetary Science (available on-line as early view; DOI: 10.1111/maps.12385). Feng H., Jones K. W., Tomov S., Stewart B., Herzog G. F., Schnabel C., Brownlee D. E. 2005. Internal structure of type I deep-sea spherules by X-ray computed microtomography. Meteoritics & Planetary Science, 40:195–206. Folco L. and Cordier C. 2014. Micrometeorites. In: Planetary Mineralogy (M. Lee, editor). EMU Notes in Mineralogy, 15, European Mineralogical Union and the Mineralogical Society of Great Britain & Ireland, London (in press). Folco L., Di Martino M., El Barkooky A., D’Orazio M., Lethy A., Urbini S., Nicolosi I., Hafez M., Cordier C., van Ginneken M., Zeoli A., Radwan A. M., El Khrepy S., El Gabry M., Gomaa M., Barakat A. A., Serra R. and El Sharkawi M. 2010. The Kamil crater in Egypt. Science 329:804. Folco L., Di Martino M., El Barkooky A., D’Orazio M., Lethy A., Urbini S., Nicolosi I., Hafez M., Cordier C., van Ginneken M., Zeoli A., Radwan A. M., El Khrepy S., El Gabry M., Gomaa M., Barakat A. A., Serra R. and El Sharkawi M. 2011. Kamil crater

157 (Egypt): Ground truth for small-scale meteorite impacts on Earth. Geology 39:179– 182. Gemelli M., D'Orazio M., Folco L. 2015. Chemical analyses of iron meteorites by hand-held X-ray fluorescence. Geostandards and Geoanalytical Research 39:1:55-69. Genge M. J., Engrand C., Gounelle M., Taylor S. 2008. The classification of micrometeorites. Meteoritics & Planetary Science 43:497–515. Grieve R. A. F. and Plant A. G. 1973. Partial melting on the lunar surface as observed in glass coated Apollo 16 samples. Proceedings of the fourth Lunar Science Conference (Supplement 4. Geochimica et Cosmochimica Acta 1:667–679. Hamann C., Hecht L., Ebert M., R. Wirth 2013. Chemical projectile–target interaction and liquid immiscibility in impact glass from the Wabar craters, Saudi Arabia. Geochimica et Cosmochimica Acta 121:291–310. Hörz F., Mittlefehldt D. W., See T., Galindo 2002. Petrographic studies of the impact melts from Meteor Crater, Arizona, USA. Meteoritics & Planetary Science 37:501–531. Kenkmann T., Trullenque G., Deutsch A., Hecht L., Ebert M., Salge T., Schnfer F., Thoma K. 2013. Deformation and melting of steel projectiles in hypervelocity cratering experiments. Meteoritics & Planetary Science 48:150-164. Kuper R. and Kröpelin S. 2006. Climate-controlled Holocene occupation in the Sahara: Motor of Africa’s evolution. Science 313:803–807. Melosh H. J. 1989. Impact cratering: A geologic process. Oxford Monographs on Geology and Geophysics 11. Oxford: Oxford University Press. 245 p. Mittlefehldt D. W., See T. H., Hörz F. 1992. Dissemination and fractionation of projectile materials in the impact melts from Wabar Crater, Saudi Arabia. Meteoritics 27:361– 370. Nininger H. H. 1954. Impactite slag at Barringer Crater. American Journal of Science 252:277–290. Nininger H. H. 1956. Arizona’s meteorite crater. Past - present - future. Denver, CO: American Meteorite Laboratory. 232 p. O’Keefe J. D. and Ahrens T. J. 1985. Sampling of planetary surfaces by oblique impact jet entrainment (abstract). 16th Lunar and Planetary Science Conference. pp. 629–630. Ott U., Merchel S., Herrman S., Pavetich S., Rugel G., Faestermann T., Fimiani L., Gomez- Guzman J. M., Hain K., Korschinek G., Ludwig P., D’Orazio M., Folco L. 2014. Cosmic ray exposure and pre-atmospheric size of the Gebel Kamil iron meteorite. Meteoritics & Planetary Science 49:1365-1347.

158 Rinehart J. S. 1958. Distribution of meteoritic debris about the Arizona meteorite crater. Smithsonian Contributions to Astrophysics 2:145–159. Rudnick, R.L. and Gao, S. 2004. Composition of the Continental Crust. In: Treatise on Geochemistry. Holland, H.D. and Turekian, K.K. (Editors), Elsevier, Amsterdam 3:1- 64. Shuvalov, V. and Dypvik H. 2013. Distribution of ejecta from small impact crters. Meteoritics & Planetary Science 48:1034-1042. Tamponi M., Bertoli F., Innocenti F. and Leoni L. (2003) X-ray fluorescence analysis of major elements in silicate rocks using fused glass discs. Atti Società Toscana di Scienze Naturali, Memorie Serie A CVII:73–80. Urbini S., Nicolosi I., Zeoli A., El Khrepy S., Lethy A., Hafez M., El Gabry M., El Barkooky A., Barakat A., Gomaa M., Radwan A. M., El Sharkawi M., D’Orazio M., Folco L. 2012. Geological and geophysical investigation of Kamil crater, Egypt. Meteoritics & Planetary Science 47:1842–1868.

159

5. CONCLUSIONS

Kamil Crater is a small simple crater of 45 m in diameter. It was generated by the impact of the Gebel Kamil iron meteorite on Cretaceous sandstones of the Gilf Kebir Formation. On the basis of archaeological evidence, the impact occurred likely < 5000 yr ago. Kamil can be considered a natural laboratory to study the cratering process of small impactors (about 1 m in diameter) on Earth and their impact on terrestrial environment for impact hazard assessment, because of the simple geological setting in which it occurred and its pristine state of preservation.

The petrographic and geochemical study of shock features of ejected sandstone blocks (Chapter 2), impact melt lapilli and bombs (Chapter 2 and Chapter 3), and microscopic impactor debris collected in the soil around Kamil Crater (Chapter 4) allow to drawn the following conclusions:

1. Ejected rocks collected at Kamil show a broad set of shock features ranging from fracturing to whole-rock impact melting. They were classified into two categories: i) pervasive shock features, including fracturing, planar deformation features, and impact melt lapilli and bombs, and ii) localized shock features including high- pressure phases and localized impact melts occurring as intergranular melt, melt veins and melt films enveloping shatter cones.

2. Pervasive shock features occupy ~100 vol.% of the sample at the thin section scale. They provide information on the shock pressure suffered by the target rocks. Localized shock features (<1 vol.% of the sample at the thin section scale) are the result of enhanced shock pressure and temperature related to pores collapse and heterogeneities of the target rocks.

3. Kamil is the smallest impact structure where shatter cones, coesite, stishovite, diamond, and impact melt (target and projectile) have been reported.

4. The maximum shock pressure recorded by Kamil rocks is between 30 and 60 GPa, likely around 50 GPa. Hence, the minimum impact velocity of Gebel Kamil was of 5.0 km-1, assuming an impact angle of 45°.

5. The hypervelocity impact of meter-sized iron meteorite projectiles can produce shock effects similar to those observed in high velocity, larger impacts. The young age of the crater (most likely < 5000 yr), the mechanical strength of target rocks

161 and the low erosion rates of the hot-desert area played a crucial role in the preservation of all these shock features.

6. The study of Kamil glasses help to constrain impact melting process and target- projectile interaction in small craters.

7. Two glass end-members were found at Kamil: white glass and dark glass. The

white glass is highly vesicular and almost exclusively made of SiO2. Nickel and Co concentrations are close to target concentrations. The dark glass is dominantly

made of a silicate melt with variable contents of Al2O3, FeO, and NiO. The dark glass typically includes fragments of shocked sandstone, diaplectic glass, lechatelierite, and Ni-Fe metal spherules.

8. The white glass derived from the melting of the target rocks with negligible projectile contamination (<0.1 wt%). The dark glass is a mixture of target and projectile liquids. The projectile contamination of the dark glass is ~11-12 wt%.

9. Dark glass formed at the contact between the target and projectile liquid (target- projectile interaction zone) during an intermediate stage between the contact and compression stage and the excavation stage. White glass formed immediately out the interaction zone, indicating that the projectile contamination rapidly decreases.

10. During the excavation stage, due to short and chaotic time sequence and to high temperature, dark glass can engulf and coat white glass and target fragments and stick on macroscopic iron meteorite shrapnel fragments.

11. Kamil soil contains numerous Ni-rich microparticles microparticles in the form of glassy fragments, oxide spherules, metallic blebs, and microscopic shrapnel pieces. These particles were distributed in a downrange ejecta curtain of microscopic impactor debris due SE-SW of the crater (extension ~300,000 m2, up to ~400 m from the crater). This is a further confirmation of previous determination of the impactor trajectory.

12. Ni contents of Kamil soil indicate that the mass of the impactor dispersed in soil is of ~120 - 290 kg. This is a small fraction (~1.5-3.5%) of the estimated impactor mass in the form of macroscopic shrapnel. At Barringer (1.2 km-in-diameter), the amount of impactor dispersed in soil was of the order of 45% the mass of the impactor. This difference confirms that the smaller the crater, the smaller the production of molten microscopic debris relative to the projectile mass.

162

13. The total impactor mass at the contact with the target estimated from our study of macroscopic and microscopic impactor debris is about 10 t. That Kamil Crater was generated by a relatively small impactor is consistent with literature estimates of its pre-atmospheric mass (>20 t, likely 50-60 t).

163

6. FUTURE WORK

6.1. Combined micro-Raman and TEM study of high-pressure phases from Kamil Crater (Egypt): implications for their formation in small impact craters on Earth

In the framework of the Barringer Family Company award, I am studying the high- pressure phases occurring in Kamil sandstone rocks. The aims of my work are to constrain the processes of formation of diamond and coesite-bearing regions (symplectic regions) carrying out micro-Raman and Transmission Electron Microscope (TEM) study.

Preliminary results of Raman investigations (~30 single Raman spectra) on a diamond-bearing SiO2 melt pocket show that at least three C-phases occur. They are diamond (sp3-type C-C bonding; tetrahedral structure), an amorphous phase with two broad bands at ~1580 cm-1 and ~1350 cm-1 (structurally similar to graphite; sp2-type C-C bonding), and an amorphous phase with a broad band at ~1430 cm-1 (C-phase with a structure intermediate to diamond sp3 and graphite sp2; Núñez-Regueiro et al. 1992). The occurrence of these phases together with diamond suggests that Kamil diamond formed by a two-steps-process of amorphization from graphite to an amorphous carbon (sp2) and then to an amorphous intermediate sp2-sp3 phase. This latter phase would represent the precursor phase of diamond, which likely crystallizes during the shock pressure unloading.

Preliminary results of Raman and TEM investigation (~30 single Raman spectra and one session at TEM) on symplectic regions confirm our previous observations carried out with Field Emission Scanning Electron Microscopy (FE-SEM). Coesite is in form of submicrometric crystals distributed in a glassy matrix in the external part of the symplectic regions. Symplectic regions are surrounded by shocked quartz with planar deformation features. TEM investigation also reveals that coesite is usually twinned.

In the next months, Raman maps will be collected to understand (i) how C-phases are spatially distributed in the SiO2 melt pocket, (ii) which are their sizes, and (iii) which are their textural relationships. Moreover, we will focus on the structural characterization of SiO2 phases through TEM investigation. These further data will help up to constrain the process of diamond and symplectic region formation.

6.2. Liquid immiscibility features in impact melts

Kamil impact glasses show emulsion textures that are the result of silicate liquid immiscibility. These features are common in dark glass, but can also be found in some

165 restricted areas of white glass. The dark glass was formed by the mixing of two very diverse liquids in terms of chemistry and physical properties: one from the target (Si-Al-rich) and another from the projectile (Fe-rich). Liquid–liquid phase-separation produces the formation of a poorly polymerized FeO-rich silicate melt and of a highly polymerized, SiO2-rich melt. The melting of accessory phases in white glass determines the formation of liquids very different from the SiO2 melt in terms of chemistry and physical properties. An example of liquid–liquid phase-separation in white glass is the formation of a poorly polymerized TiO2-

Al2O3-rich silicate melt and of a highly polymerized, SiO2-rich melt.

Emulsion textures can be found in numerous fresh impact glasses, as for example Wabar, Barringer, and Tenoumer glasses. Moreover, these textures were also found in impact glass experimentally produced in the framework of the MEMIN (Multidisciplinary Experimental and Modeling Impact Research Network) project (Ebert et al. 2013 and 2014). The pilot study on Wabar emulsions show that liquid immiscibility is a major process in the formation of glassy impactites and allows the reconstruction of the thermal history of impact glass formation (Hamann et al. 2013).

In order to better understand the process of liquid immiscibility in impact glasses and to constrain the thermal history of impact glass formation, the pilot study on Wabar glass will be extended to Kamil, Barringer, Tenoumer, and experimentally produced glasses. For this study, I am collaborating with researchers of Museum für Naturkunde of Berlin. This study is based on high-resolution analytical techniques: Field Emission Electron Microprobe (FE- EMPA) and Transmission Electron Microscope (TEM) on FIB foils. In December 2014, we completed the acquisition of electron microprobe data and the cutting of FIB foils. TEM investigations are planned for January 2015.

6.3. References

Ebert M., Hecht L., Deutsch A. and Kenkmann T. 2013. Chemical modification of projectile residues and target material in a MEMIN cratering experiment. Meteoritics & Planetary Science 48:134–149. Ebert M., Hecht L., Deutsch A., Kenkmann T., Wirth R. and Berndt J. 2014. Geochemical processes between steel projectiles and silica-rich targets in hypervelocity impact experiments. Geochimica Cosmochimica Acta 133:257–279.

166 Hamann C., Hecht L., Ebert M. and Wirth R. 2013. Chemical projectile–target interaction and liquid immiscibility in impact glass from the Wabar craters, Saudi Arabia. Geochimica Cosmochimica Acta 121:291–310. Núñez-Regueiro M., Abello L., Lucazeau G., and Hodeau J.-L. 1992. Diamond from fullerenes: Evidence from Raman measurements. Physical Review B 46:9903-9905.

167

APPENDIX I. USE OF THE UNIVERSAL STAGE (U-STAGE) FOR INDEXING PLANAR DEFORMATION FEATURES IN QUARTZ

Quartz is one of the most common minerals on Earth and shows the greatest variety of shock effects (see Section 1.3.1 Quartz). Quartz is a uniaxial (trigonal) mineral. Because of the similarity with the hexagonal system, it is possible to use four axes to describe the crystals a1, a2, a3 and c, where a1=a2=a3, a1, a2, a3 intersect at an angle of 60° and lie on a plane, and c is perpendicular to the plane (Fig. 1).

Fig. 1. Schematic representation of the crystallographic axes of quartz. Image from http://www.quartzpage.de/

The most common and diagnostic shock features of quartz are the formation of planes of amorphous material oriented parallel to rational crystallographic planes. These planes are called planar deformation features (PDFs). Each plane is indicated by its crystallographic index, following the four-digit notation of Miller-Bravais {hkil}. The indexing of PDFs is important to establish the shock pressure suffered by the sample: at different shock pressures, planes are oriented parallel to different crystallographic indexes.

Three main techniques can be used to index PDFs: Universal Stage (U-Stage), Transmitted Electron Microscope (TEM), and Electron Backscattered Diffraction in the SEM (EBSD). TEM and EBSD are expensive techniques. TEM is also time-consuming, as the obtainment of a statistically meaningful data set requires may and long analytical sessions. U- stage, although outdated (the production ceased in the late 1990’s), remains a useful instrument to index hundreds of planes in a relatively little time and it is very down market.

U-stage is a device that can be mounted on an optical light microscope (Fig. 2a). Note that short objectives are necessary to use U-stage (Fig. 2). Thanks to U-stage, a thin section could be oriented in all space directions to quantify the optical and geometric features of its

169 crystals. The thin section is mounted between two glass hemispheres (upper and lower) surrounded by a system of graduated circles (axes), which measure the angular positions of the tilted section (Fig. 2b). U-stage has four or five axes. The five axes are shown in Fig. 2b. The A5 axis is the microscope table (Fig. 2a). In the past, U-stage was largely used to determine the optical orientation of crystals, 2V, optical sign, direction of pleochroism, dispersion, refractive index, orientation of twinning.

Fig. 2. Optical Microscope of the Zeiss Axioplan type equipped with a Zeiss five-axes (A1-A5) U-stage at the Dipartimento di Scienze della Terra (Univeristy of Pisa). Dashed line indicates the direction of tilt of A2. Two arcs, indicated with A2, are used to read the dip direction.

In order to index, PDFs in quartz, appropriate glass hemispheres must be used, namely those that have a refractive index of the studied mineral: 1.555 for quartz (Fig. 2b). Between the thin section and glass hemispheres 1-2 droplets of oil for microscopy (with the same refractive index of the quartz) have to be applied.

In this Thesis, we follow the PDF indexing procedure described by Langenhorst (2002) and Voorn (2010). After selecting a quartz crystal with at least two crossing PDF sets, the PDF indexing procedure could be summarized into 4 steps: i) locate the c axis; ii) locate the normals to PDF planes relative to the c-axis; iii) plot the determined direction on the stereonet; iv) compare the stereonet with the standard stereographic projection of quartz.

i) Locate the c axis. First check if A5 is in “0-point” position, then turn on the cross- polarizer and insert the gypsum plate. If color is added (commonly the grain appears blue), the operator has to rotate A1 clockwise until the grain goes into extinction. Instead, if color is subtracted (commonly the grain appears yellows)

170 the operator has to rotate A1 counter-clockwise until the grain goes into extinction. On A1 it is possible to read the azimuth of c (Fig. 3).

Fig. 3. The azimuth of the c-axis of a quartz crystal can be read on the A1 axis. The white circle highlights the value of the azimuth.

Then, the dip direction has to be measured. To do this operation A4 has to be rotated until the grain lights up again, then rotate A2 until the grain goes into extinction again. The grain has to remain dark for further rotating of A4. Using the two arcs it is possible to read the dip direction (Fig. 4). Note to indicate if the c axis dips west (to the left) or east (to the right).

Fig. 4. The A2 axis is tilted to measure the dip of the - axis. The dip angle can be read with the A2 arcs. The arrow indicates the point of reading.

171 Now we must verify if the c-axis is horizontal (equatorial) or vertical (polar), rotating A4 in 0° position. If it is horizontal, the crystal is not under extinction for any further rotation of A5. If it is vertical, for further rotation of A5 the crystal remains extinct. Horizontal positions are more common than vertical ones. ii) Locate the normals to PDF planes. The procedure for locating the normals (poles) to PDF planes is similar to that described for locating the c-axis. The operator has to return the A5 to 0° position without moving the grain, because this is a measure relative to c-axis. After selecting a set of PDFs, A1 is rotated until the lines lie parallel to NS crosshair: the measurement read on A1 is the azimuth. Then, the operator has to tilt the A2 until lines appear the thinnest, namely the PDF planes have to be vertical. It is possible to test if it is the correct position by defocusing the PDF. If a PDF is vertical, the PDF trace does not move. Using the A2 arc (right or left as a function of the dip direction), the operator can read the dip. This procedure has to be repeated for all sets of PDFs occurring in the selected grain. iii) Plot the determined direction on the stereonet. At this point the measurement (three at least: c axis, and 2 poles of PDFs) have to be plotted on the Wulff stereonet. Hence, the measurement must be rotated putting the c-axis at the center of the net. For this operation, it is advisable to use transparent paper sheets. iv) Compare the stereonet with the standard stereographic projection of quartz. The transformed stereonet can be compared with the standard stereographic projection of quartz (Fig. 5) by superimposing the transparent paper sheet on it. In Fig. 5 the most recent version of the stereographic projection of quartz is shown (Ferriére et al. 2009). This version introduces a new PDF orientation indicated with the letter “e”. “e” corresponds to the Miller-Brevais

index {} 1014 . It partially overlaps the 5° envelope of the index {} 1013 , indicated by the number 2. All PDF planes which plot in the overlapping zone between

€ {} 1013 and {} 1014 crystallographic orientations are indicated€ as 2+e orientations (Ferrière et al. 2009 and Huber et al. 2011).

€ €

172

Fig. 5. Standard stereographic projection of quartz. Image from Ferriére et al. 2009.

The PDF indexing needs of a large number of measurements to be as statistically meaningful as possible. Hence the four steps described above have to be repeated for tens of quartz grains.

Step (iii) could be done using software for measure data processing measurement commonly used in Structural Geology, such as, for examples Stereo32. In this way the comparison with the standard stereographic projection of quartz can be done using a software for image processing.

Another way to process data has been proposed by Huber et al. (2011). They created a program for the automatic indexing of PDFs. This program is called ANIE (Automated Numerical Index Executor). ANIE is designed to run in Microsoft Excel (in Windows environment). It is very simple to use and time-saving and it makes possible to skip steps (iii) and (iv) to avoid introducing manual graphical errors. Moreover, histograms and summary tables are automatically generated.

173 References http://www.quartzpage.de Huber M. S., Ferrière L., Losiak A., and Koeberl C. 2011. ANIE: A mathematical algorithm for automated indexing of planar deformation features in quartz grains. Meteoritics & Planetary Science 46:9:1418-1424. Langenhorst F. 2002. Shock metamorphism of some minerals: Basic introduction and microstuctural observations. Bulletin of the Czech Geological Survey 77:265-282. Voorn M. H. 2010. A new way to confirm meteorite impact produced planar features in quartz: Combing Universal Stage and Electron Backscatter Diffraction techniques. MSc Thesis Utrecht University (Netherlands) – Earth Sciences Department – Structural Geology and Tectonic Research Group. Ferrière L., Morrow J. R., Amgaa T., and Koeberl C. 2009. Systematic study of universal- stage measurements of planar deformation features in shocked quartz: Implications for statistical significance and representation of results. Meteoritics & Planetary Science 44:925-940.

174 APPENDIX II. THE EXTREMELY REDUCED SILICATE-BEARING IRON METEORITE NORTHWEST AFRICA 6583: IMPLICATIONS ON THE VARIETY OF THE IMPACT MELT ROCKS OF THE IAB-COMPLEX PARENT BODY

During the first months of my PhD, I focused on studying the iron meteorite Northwest Africa 6583, that was the objective of my Master Thesis (Laurea Magistrale in Scienze e Tecnologie Geologiche). Follow the integral version of the paper “The extremely reduced silicate-bearing iron meteorite Northwest Africa 6583: implications on the variety of the impact melt rocks of the IAB-complex parent body” by Agnese Fazio, Massimo D’Orazio, Luigi Folco, Jérôme Gattacceca, Corinne Sonzogni, it was published on the international peer-reviewed journal Meteoritics & Planetary Science volume 48, number 12 of December 2013.

Abstract–Northwest Africa (NWA) 6583 is a silicate-bearing iron meteorite with Ni = 18 wt%. The oxygen isotope composition of the silicates (Δ′17O = -0.439 ‰) indicates a genetic link with the IAB-complex. Other chemical, mineralogical, and textural features of NWA 6583 are consistent with classification as a new member of the IAB-complex. However, some unique features, e.g., the low Au content (1.13 µg g-1) and the extremely reducing conditions of formation (approximately -3.5 ΔIW), distinguish NWA 6583 from the known IAB-complex irons and extend the properties of this group of meteorites. The chemical and textural features of NWA 6583 can be ascribed to a genesis by impact melting on a parent body of chondritic composition. This model is also consistent with one of the most recent models for the genesis of the IAB-complex. Northwest Africa 6583 provides a further example of the wide lithological and mineralogical variety that impact melting could produce on the surface of a single asteroid, especially if characterized by an important compositional heterogeneity in space and time like a regolith.

INTRODUCTION

Silicate-bearing iron meteorites carry important information about both nebular and asteroidal processes. They are extremely diverse, particularly in terms of texture and chemical composition of the (Fe,Ni) metal, as well as in terms of texture and mineral composition of the silicate fraction. The possibility of measuring the oxygen isotope composition of the silicates in silicate-bearing irons has established a genetic relationship between several iron meteorite groups and known stony meteorite groups. A possible genetic link has been invoked

175 between IIE irons and H chondrites and between IAB-complex irons and winonaites (Clayton and Mayeda 1996; Folco et al. 2004).

Models elaborated to explain the occurrence of silicate minerals (most commonly olivine, pyroxene, plagioclase, and silica polymorphs) within an (Fe,Ni) metal matrix are extremely complex and may involve impact melting on the surface of their parent asteroids, incomplete metal-silicate fractionation during core formation, oxidation/reduction processes, mixing of silicate and metallic melts, complete destruction and reassembling of asteroidal bodies, etc. (e.g., Mittlefehldt et al. [1998] and the references therein, for a review).

In this paper, we report on the structure, chemistry, mineralogy, and oxygen isotope composition of a unique, silicate-bearing iron meteorite named Northwest Africa 6583 (Meteoritical Bulletin no. 100, unpublished data). Based on the collected data, we establish its relationship with a known meteorite group, we define its petrogenesis (including its crystallization history and oxygen fugacity), and discuss the implications for the parent body of the IAB-complex.

SAMPLES AND ANALYTICAL METHODS

Northwest Africa 6583 meteorite (hereafter NWA 6583) was purchased from a Moroccan dealer in October 2010 by Mr. Mirko Graul (Bernau, Germany). Before sampling, NWA 6583 was a single 1825 g iron meteorite with an irregular flattened shape, measuring 105 × 100 × 38 mm (Fig. 1). The upper portion of the meteorite is glossy-brown and presents approximately 40 rounded depressions from 0.5 to 1 cm in diameter (Fig. 1a). The portion originally embedded in soil (approximately 40% of the surface) is oxidized and partially covered by thin crusts of light-brown calcareous material (mostly soluble calcium salts) of secondary origin, i.e., caliche or calcrete (Figs. 1b and 1c). We did not observe any remnant of the fusion crust.

176

Fig. 1. The iron meteorite Northwest Africa 6583 (NWA 6583) before cutting. a) Top view; b) bottom view; c) side view. The side length of the scale cube is 1 cm. Images from Mirko Graul. A 60.5 g etched end cut is on deposit at the Pisa University’s Museo di Storia Naturale, while M. Graul holds the main mass. A further 30 g full slice was donated by the owner of the meteorite for further studies. A sample of 580 mg was extracted from the end cut for inductively coupled plasma-mass spectrometry (ICP-MS) analysis. Subsequently, the end cut and the slice were polished for metallographic, scanning electron microscope coupled with an energy-dispersive X-ray fluorescence analysis (SEM-EDX), and electron probe microanalysis (EPMA) studies. The total examined surface was approximately 48 cm2.

Backscattered and secondary electron (BSE, SE) images and semiquantitative chemical analyses were obtained at Pisa University’s Dipartimento di Scienze della Terra with a Philips XL30 SEM-EDX, operating at 20 kV. Mineral compositions were determined with a CAMECA SX50 electron microprobe fitted with four wavelength-dispersive spectrometers at the Istituto di Geoscienze e Georisorse (IGG) of the Consiglio Nazionale delle Ricerche (CNR) in Padua. Running conditions were 15 kV accelerating voltage, 20 nA beam current, and 1 l m nominal beam spot. Counting times for the determined elements were

177 20 s and 10 s at peak and background, respectively. The manufacturer supplied PAP procedure was employed for raw data reduction. Standards used for instrumental calibration were minerals (diopside, apatite, and sphalerite), synthetic compounds (MnTiO3, Cr2O3), and pure elements (Fe, Ni, Co, and Cu) for metal, sulfide, and phosphide analysis and minerals

(diopside, orthoclase, albite,) and synthetic compounds (Al2O3, MnTiO3, Cr2O3, Fe2O3, NiO) for silicate analysis.

The bulk chemical analysis of the metal of NWA 6583 was performed at the Pisa University’s Dipartimento di Scienze della Terra by conventional ICP-MS (Thermo PQII Plus), following the procedure described in D’Orazio and Folco (2003). To maximize analytical accuracy, the sample solutions were measured using the standard additions method instead of an external calibration. At the concentration levels of NWA 6583, the analytical uncertainties (RSD%) ranged from 2% to 5% for all elements but Re, for which RSD% ranges from approximately 10 to 20%.

Oxygen three-isotope analyses of NWA 6583 silicate inclusions were carried out at the Stable Isotopes Laboratory of CEREGE (Centre Européen de Recherche et d’Enseignement de Géosciences del’Environnement) in Aix-en-Provence. Molecular oxygen was extracted using the IR-laser fluorination technique (Alexandre et al. 2006; Suavet et al. 2010). About 3 mg of pyroxene grains was extracted from the metal matrix to produce two powdered aliquots (labeled Sample A and Sample B), each weighing approximately 1.5 mg.

The two aliquots were heated with a 30 W CO2 IR-laser in the presence of BrF5. The released gas was purified through cryogenic nitrogen traps and one KBr trap. Molecular oxygen was passed through a –114 °C slush and trapped in a nitrogen-cooled molecular sieve before being expanded at 100 °C directly to a dual-inlet mass spectrometer (ThermoQuest Finnigan Delta Plus). The isotope compositions are expressed in standard d -notation, relative to Vienna 18 18 16 18 16 standard mean ocean water (‰ versus V-SMOW): δ O = ([ O/ O]sample/ [ O/ O]standard – 1) 17 17 16 17 16 17 × 1000 and δ O = ([ O/ O]sample /[ O/ O]standard – 1) × 1000. Δ′ O was calculated as 1000 ln (1 + [δ17O /1000]) – λ 1000 ln(1 + [δ18O /1000]) where λ = 0.5247 (Miller 2002). Measured δ18O and δ17O values of the samples were corrected on a daily basis using the quartz laboratory standard Boulangé e itself calibrated against the international standard NBS28 (quartz sand) with δ18O = 9.6 ± 0.123 ‰, δ17O = 5.026 ± 0.075 ‰, and Δ′17O = 0 ± 18 17 17 0.026 ‰ (±1σ, n = 23). Reproducibility of δ OBoulangé, δ OBoulangé, Δ′ OBoulangé analyses (±1σ) is 0.119‰, 0.061‰, and 0.024‰, respectively (n = 63).

178 RESULTS

Texture and Chemistry of the Main Minerals

Etched sections show that NWA 6583 is devoid of macroscopic Widmanstätten pattern. Northwest Africa 6583 shows, instead, a polycrystalline texture (Fig. 2) consisting of irregular, subequant crystals of (Fe,Ni) ataxitic metal with different orientations and sizes ranging from 0.5 to 22 mm (approximately 75% of crystals have sizes between 1 and 6 mm). Troilite, graphite, and schreibersite are the most abundant accessory minerals and they are easily identifiable with the naked eye (Fig. 2).

Fig. 2. a) Polished and etched surface of the NWA 6583 type specimen. b) Sketch of the same surface showing the boundaries of (Fe,Ni) metal grains and the crystals of troilite (gray hatched), graphite (black) and schreibersite (gray). The side length of the scale cube on the left-hand side of the images is 1 cm. The bulk composition of the NWA 6583 (Fe,Ni) metal is characterized by high content of Ni (18.0 wt%) and moderately low Co content (0.43 wt%). The trace element distribution is close to the cosmic abundance. In greater detail, we observe relatively higher concentrations of the moderately volatile, chalcophile, and siderophile incompatible elements Cu, Ge, Ga, Sn, As, and Sb (Table 1), and lower concentrations of the refractory, highly siderophile, and compatible elements Ir, Re, and W, as evidence of a limited trace element fractionation.

179 Table 1. ICP-MS analysis of bulk metal of NWA 6583.

Ni 18 Pd 9.6 Co 0.43 Sn 32 Cu 1350 Sb 2.5 Ga 52 W 0.5 Ge 125 Re 0.03 As 11.9 Ir 0.19 Mo 2.78 Pt 1.68 Ru 2.46 Au 1.13 Rh 0.65 Ni and Co are in wt%, the other elements are in µg g-1.

The (Fe,Ni) ataxitic crystals have homogeneous Ni-rich composition (Ni = 17.5 ± 0.4 wt%, Co = 0.40 ± 0.04 wt%, n = 100; Table 2). Their microstructure consists of ragged platelets (10–15 µm in size) arranged in octahedral orientation (Fig. 3a). In literature, such microstructure in Ni-rich (Fe,Ni) metal is known as martensitic structure (Buchwald 1975; Goldstein et al. 2009). The martensitic matrix is speckled with numerous small (from 3 to 6 µm wide and up to 200 µm long) spindles of kamacite (Ni = 6.9 ± 0.5 wt%, Co = 0.59 ± 0.05 wt%, n = 11; Table 2). Kamacite spindles are nucleated on tiny schreibersite crystals and are surrounded by a very thin (typically < 0.5 l m) rim of taenite (Fig. 3b). In the outer portion (up to approximately 250 l m from the grain boundary) of the largest metal grains, the amount of kamacite spindles decreases. The (Fe,Ni) metal grains are separated from each other by small schreibersite crystals enveloped in kamacite. Table 2. Electron microprobe average compositions of martensite, kamacite, troilite, and schreibersite of NWA 6583 (units in wt%). In parentheses are reported the standard deviations (1σ).

Martensite Kamacite Troilite Schreibersite

N. of analyses 100 11 37 2 Mg 0.05 (0.01) 0.03 (0.02) 0.03 (0.01) < 0.03 Si 0.13 (0.03) 0.12 (0.03) 0.03 (0.03) < 0.03 P 0.30 (0.37) < 0.08 < 0.08 14.7 (0.06) S 0.04 (0.01) < 0.04 36.9 (0.51) 0.08 (0.02) Ca < 0.03 < 0.03 < 0.03 < 0.03 Ti 0.03 (0.00) < 0.03 0.13 (0.02) < 0.03 Cr < 0.06 < 0.06 0.83 (0.07) < 0.06 Mn < 0.06 0.08 (0.01) 0.15 (0.04) < 0.06 Fe 81.8 (1.66) 93.1 (1.06) 62.0 (0.93) 65.9 (1.33) Co 0.39 (0.04) 0.59 (0.05) < 0.06 0.30 (0.02) Ni 17.7 (1.32) 6.9 (0.47) 0.09 (0.03) 18.9 (0.36) Cu 0.14 (0.04) < 0.06 < 0.06 < 0.06 Zn < 0.10 < 0.10 < 0.10 < 0.10 Sum 100.7 100.8 100.3 99.9

180

Fig. 3. a) Back-scattered electron (BSE) image showing details of an (Fe,Ni) ataxitic grain (slightly etched surface) of the NWA 6583 metal. Note the chemical homogeneity and martensitic structure. b) Secondary electron image of some kamacite spindles nucleated on tiny schreibersite crystals (kamacite spindles appear excavated due to deep etching). The (Fe,Ni) metal of NWA 6583 shows an anomalously high content of Si (martensite, average = 0.13 ± 0.02 wt%, n = 100; kamacite, average = 0.12 ± 0.03 wt%, n = 11; Table 2). These contents are much higher than those commonly observed in iron meteorites, where Si is generally below 35 µg g-1 (0.0035 wt%; Wai and Wasson 1969). Two notable exceptions are Horse Creek and Tucson, which contain 2.5 wt% and 0.8 wt% Si, respectively (Wai and Wasson 1969). High concentrations of Si (from approximately 0.3 to 4 wt%) are typical of the (Fe,Ni) metal of enstatite chondrites and aubrites (Brearley and Jones 1998).

Troilite is the most abundant accessory mineral (approximately 4.5 vol%); it occurs as rounded blebs up to 5 mm in maximum length. The troilite crystals close to the external surface of the meteorite (< 2 mm) show lamellar twinnings, melting textures, and are polycrystalline (Fig. 4a). The composition of troilite is homogeneous and characterized by moderately high contents of Cr, Mn, and Ti (0.83 ± 0.07 wt%, 0.15 ± 0.04 wt%, and 0.13 ± 0.02 wt%, respectively, n = 37; Table 2). Troilite crystals often include, preferentially at their margins, small crystals of different species of Mn-Fe-Mg-Zn-Cu-Ni sulfides, Mg-silicates, and rare Fe-Ni phosphide (see next section). Native copper has been occasionally found at the troilite-(Fe,Ni) metal interface.

Graphite is the second most abundant accessory mineral (approximately 1.5 vol%). It usually occurs as subspherical rosettes 50 – 500 µm in diameter, occasionally reaching 10 mm (Figs. 2, 4b, and 5a,b), and as isolated lamellar crystals up to 0.7 mm in length (Fig. 4c).

Schreibersite (approximately 1 vol%) occurs as three different types: (1) large skeletal crystals up to 8 mm in maximum length, sometimes enveloping troilite crystals; (2) tiny (1– 30 µm) crystals at the core of the kamacite spindles (Fig. 3b); (3) slightly larger euhedral crystals at the (Fe,Ni) metal grain interface. Two analyses of the large skeletal crystals show a

181 Ni/Fe atomic ratio of 0.27 (Table 2). Some large schreibersite crystals located close (< 2 mm) to the external surface of the meteorite show evidence of melting. This is documented by the occurrence of rims with very fine-grained eutectic textures, which are thicker on the crystal side closer to the external surface of the meteorite.

Terrestrial weathering products are represented by iron hydroxides that are concentrated along the (Fe,Ni) grain boundaries (Fig. 2a), around troilite crystals (Fig. 4d), and on the external surface (Fig. 1).

Fig. 4. Optical microscope views (reflected plane polarized light) of polished NWA 6583 slices. a) Troilite crystals with a melted rim and inclusions of enstatite and alabandite. b) Graphite rosette and troilite crystal with an inclusion of alabandite. A small elongated crystal of enstatite occurs between the graphite and the troilite. c) Graphite crystal surrounded by Fe-hydroxides. d) Detail of a troilite crystal boundary with an inclusion of a wedge-shaped niningerite crystal. The troilite-(Fe,Ni) metal boundary is marked by Fe-hydroxides. e) Detail of a troilite crystal boundary with inclusions of a wedge-shaped alabandite crystal and of elongated enstatite crystals. f) Detail of a cluster of silicate crystals associated to troilite, located close to the external surface of the meteorite.

182 Accessory Sulfide Inclusions in Troilite

About 40% of troilite crystals contain small inclusions of various sulfide minerals. The most common sulfides belong to the alabandite–niningerite series (Figs. 4d,e, 5, and 6a; Table 3). These crystals are wedge-shaped and are located very close to the troilite crystal boundaries. The variable compositions may be distinguished on the basis of the different reflectivity in SEM-BSE images (the higher the Mg/Mn ratio, the lower the BSE reflectivity). Some crystals are compositionally zoned, showing an irregular decrease of the Mg/Mn ratio from core to rim (Fig. 5c).

Fig. 5. BSE images of accessory sulfides of NWA 6583. a) Rounded crystal of troilite including two wedge- shaped alabandite crystals. b) Complex assemblage of alabandite-niningerite, enstatite and Fe-Zn-Mn sulfide (most likely buseckite) included in troilite. c) Skeletal crystal of niningerite included in troilite. d) Detail of the melted rim of a troilite crystal showing tiny crystals of troilite rimmed by Fe-Ni sulfide. In the FeS-MnS-MgS compositional triangle (Fig. 6a; Keil 2007), the data points form a continuous trend from Mg-free alabandite (Mg = 0.07 wt% and Mn = 52 wt%) to Mn- bearing niningerite (down to Mg = 15.5 wt% and Mn = 19.7 wt%) with less variable Fe content (from Fe = 9.40 wt% to Fe = 19.8 wt%). These sulfides contain very low or negligible Ca (< 0.32 wt%) and Cr (< 0.37 wt%; Table 3). Niningerite usually has higher Fe, Ca, and Cr contents than alabandite (Table 3).

183 Table 3. Representative electron microprobe analyses of Fe-Mn-Mg-Zn sulfides of NWA 6583 (units in wt%).

Alabandite Niningerite Buseckite (?) Mg 0.07 3.84 7.17 8.01 11.3 13.4 12.1 14.3 14.7 15.5 <0.03 <0.03 0.06 <0.03 Si <0.03 0.07 <0.03 <0.03 0.07 <0.03 <0.03 0.06 <0.03 0.06 0.06 <0.03 0.07 0.06 S 36.8 38.5 39.7 40.5 41.7 42.8 42.1 43.2 43.9 44.4 34.5 35.0 35.9 35.6 Ca <0.03 <0.03 <0.03 <0.03 0.09 0.14 0.20 0.16 0.18 0.32 <0.03 <0.03 <0.03 <0.03 Cr <0.06 0.09 0.06 0.20 0.14 0.12 0.22 0.29 0.21 0.37 <0.06 <0.06 <0.06 <0.06 Mn 52.0 44.5 42.0 38.1 35.7 31.1 27.4 27.9 21.1 19.7 13.7 14.5 15.9 14.7 Fe 9.40 12.0 10.5 12.6 10.6 11.8 17.5 13.8 19.9 19.8 25.3 25.6 22.8 24.0 Ni 0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 <0.07 Cu 0.57 <0.06 0.08 <0.06 <0.06 <0.06 <0.06 <0.06 <0.06 <0.06 0.27 0.11 0.28 0.41 Zn <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 24.7 24.4 24.9 25.2 Sum 98.9 99.0 99.6 99.4 99.6 99.4 99.5 99.7 100.0 100.1 98.5 99.6 100.0 100.0

Cation formula (a.p.f.u.) based on 2 atoms Mg 0.002 0.132 0.237 0.262 0.355 0.413 0.379 0.435 0.444 0.464 0.002 Si 0.002 0.002 0.002 0.002 0.002 0.002 0.002 S 1.007 1.006 0.994 1.002 0.995 1.000 0.996 0.997 1.005 1.005 0.995 0.996 1.013 1.007 Ca 0.002 0.003 0.004 0.003 0.003 0.006 Cr 0.001 0.001 0.003 0.002 0.002 0.003 0.004 0.003 0.005 Mn 0.831 0.678 0.614 0.551 0.497 0.424 0.378 0.376 0.282 0.261 0.230 0.240 0.262 0.243 Fe 0.148 0.179 0.151 0.179 0.146 0.158 0.237 0.182 0.261 0.257 0.418 0.418 0.370 0.390 Ni 0.001 Cu 0.008 0.001 0.004 0.002 0.004 0.006 Zn 0.349 0.340 0.345 0.350

MnS 84.6 68.5 61.3 55.5 49.8 42.6 38.0 37.9 28.6 26.5 23.0 24.0 26.8 24.7 FeS 15.1 18.1 15.1 18.1 14.6 15.9 23.9 18.3 26.5 26.2 41.9 41.8 37.9 39.7 MgS 0.24 13.3 23.6 26.4 35.5 41.5 38.1 43.8 45.0 47.3 35.0 34.0 35.3 35.6 The concentrations of P, Ti, and Co are below the detection limit of 0.08, 0.03, and 0.06, respectively. In association with sulfides of the alabandite–niningerite series, we also found a Fe- Zn-Mn sulfide mineral. This mineral forms 50– 200 µm anhedral crystals characterized by (Mn+ Fe)/Zn atomic ratio close to 2 (Figs. 5b and 6b). Normalizing the cations to a total of one, the formula can be written as: (Fe0.42,Zn0.34,Mn0.23)S. Recently, Ma et al. (2012) described a new hexagonal mineral species, buseckite (Fe0.46,Zn0.32,Mn0.16,Mg0.04)S1.01, found within the Zakłodzie ungrouped enstatite-rich achondrite (Fig. 6b). The Fe-Zn-Mn sulfide we found in NWA 6583 could be, in fact, buseckite; however, X-ray diffraction data are needed to confirm this hypothesis.

184

Fig. 6. a) Compositional triangle FeS-MnS-MgS (mol. %) for the alabandite-niningerite crystals of NWA 6583. b) Lower part of the compositional triangle FeS-MnS-ZnS (mol. %) for the unknown (most likely buseckite) Fe- Zn-Mn sulfide found in NWA 6583. The white star represents the average composition of buseckite found in the Zakłodzie meteorite (Ma et al. 2012). Two more accessory sulfides have been sporadically found in NWA 6583 in association with troilite: an Fe-Cu sulfide with a Fe/Cu atomic ratio close to 1 (most likely chalcopyrite), and an Fe-Ni sulfide (both analyzed by SEM-EDX only). The first has been found in the external portions of troilite, while the second has been found in the melted rim of a large troilite where it forms the very thin (from 1 to 5 l m) rim of small troilite crystals (Fig. 5d).

Silicate Inclusions

Northwest Africa 6583 contains crystals of Mg and Mg-Ca silicates that may be found as isolated euhedral crystal embedded in the (Fe,Ni) metal matrix or included within troilite (Figs. 7a–c), or as clusters of tens of individuals with euhedral to subhedral textures even when they abut the metal matrix. In both the studied slices, the clusters of silicates occur close to the external surface of the meteorite, in correspondence with the bottom of the small depressions (Figs. 1, 2, 4f, and 7d). The size of the silicate crystals ranges from a few tens of micrometers up to 1 mm. Their composition is homogeneous. The most common silicate is

185 enstatite, very close to the Mg2Si2O6 endmember composition (Fs0.44±0.22, Wo1.57±0.15; n = 17; Table 4). Some crystals, especially those forming the clusters, show complex intergrowths with diopside and troilite (Figs. 7b–f and 4f). Diopside never forms isolated crystals. Its composition (Fs1.52±0.75, Wo44.34±5.98; n = 9; Table 4) is close to the CaMgSi2O6 endmember composition, even though it is slightly more iron-rich than the coexisting enstatite. Olivine is much less abundant. The few euhedral crystals found within troilite, or in the metal matrix, are very close to the Mg2SiO4 endmember composition, Fa1.3 (SEM-EDX analyses only).

Table 4. Electron microprobe average compositions of pyroxenes of NWA 6583 (units in wt%). In parentheses are reported the standard deviations (1σ).

Enstatite Diopside N. of analyses 17 9

SiO2 60.3 (0.4) 56.7 (1.6)

TiO2 <0.03 <0.03

Al2O3 0.07 (0.04) 0.27 (0.21)

Cr2O3 <0.05 <0.05 FeO 0.36 (0.19) 0.98 (0.44) MnO <0.06 <0.06 NiO <0.07 0.08 (0.01) MgO 39.5 (0.3) 19.7 (3.0) CaO 0.88 (0.10) 22.5 (2.7)

Na2O <0.09 0.14 (0.03) Sum 101.2 100.4

En 98.0 (0.2) 53.8 (6.2) Fs 0.44 (0.25) 1.55 (0.75) Wo 1.57 (0.15) 44.3 (6.0)

186

Fig. 7. BSE images of silicate inclusions of NWA 6583. a) Euhedral, homogeneous enstatite crystal embedded in the (Fe,Ni) metal matrix. b) Subhedral crystal of diopside with a strongly resorbed and irregular enstatite core. The tiny white particles included in the diopside are troilite. c) Small cluster of enstatite crystals showing inclusions of diopside and troilite. d) Cluster of tens of silicate crystals associated to troilite and schreibersite, located close to the external surface of the meteorite. e) Detail of an elongate euhedral crystal of pyroxene showing complex intergrowth between diopside and enstatite. f) Detail of an anhedral enstatite crystal, found within the silicate cluster of Fig. 7d, showing an overgrowth of fine-grained diopside-enstatite-troilite crystals. Two aliquots of pyroxene extracted from NWA 6583 were analyzed to obtain the oxygen isotope composition, yielding an average δ17O = 2.059 ‰, δ18O = 4.767 ‰, Δ′17O = – 0.439 ‰ (Table 5).

187 Table 5. Oxygen isotope composition of two sub-samples of pyroxene crystals separated from NWA 6583.

d17O ‰ 1 σ‰ d18O ‰ 1 σ‰ ∆’17O ‰ 1 σ‰

Sample A 2.178 0.061 4.942 0.119 -0.411 0.024

Sample B 1.939 0.061 4.591 0.119 -0.466 0.024

Average 2.059 4.767 -0.439 The precision (1s) was estimated from repeated measurements of an in-house quartz standard (see text).

DISCUSSION A new member of the IAB-complex

Northwest Africa 6583 was previously classified as an ungrouped iron meteorite (Meteoritical Bulletin, no. 100, in preparation). Our subsequent data on the oxygen isotope composition of the silicate inclusions of NWA 6583 plot below the terrestrial mass fractionation line within the field of the silicate inclusions of the IAB-complex iron meteorites and winonaites (δ17O = 2.059 ‰, δ18O = 4.767 ‰, and ∆'17O = -0.44 ‰; Fig. 8 and Table 5). Several lines of evidence (e.g., oxygen isotope composition, chemistry, mineralogy) suggest that IAB-complex irons and winonaites are genetically related (Clayton and Mayeda 1996; Benedix et al. 2000), i.e., that they formed in parent bodies within the same oxygen reservoir of the early Solar System and possibly in the same parent body. The oxygen isotope composition of NWA 6583 thus establishes a strong petrogenetic relationship with the related IAB-complex iron meteorites and winonaites.

Fig. 8. Plot of ∆’17O vs d18O for the pyroxenes of NWA 6583 (mean of two analyses). The compositional fields of other meteorite groups are plotted for comparison (data from Clayton et al. 1984; Clayton and Mayeda 1996 and 1999; Wang et al. 2003; Franchi 2008).

188 Only the 6.1% of the entire iron meteorite population (66 out of the 1074 classified irons; Meteoritical Bulletin Database, accessed March 2013) has a Ni contents in the 13.5 - 22.5 wt% range, i.e., a range defined by the NWA 6583 Ni content (18 wt%) ± 25%. Most of these meteorites belong to the group IVB (14 irons) and to the IAB-complex (27 irons), while 23 are ungrouped irons and only one belongs to the group IIF (Repeev Khutor; Kracher et al. 1980). Northwest Africa 6583 is very different from IVB irons for its two-to-three orders of magnitude higher Ga, Ge, and Cu contents and for its one-to-two orders of magnitude lower Re, Ir, W, and Mo (Walker et al. 2008; Fig. 9). In turn, NWA 6583 is similar to members of the IAB complex as well as to some ungrouped irons like Gebel Kamil and Yamato 791076 (Wasson et al. 1989; Wasson and Kallemeyn 2002; D’Orazio et al. 2011). The high Ge/Ga ratio and the low Ni/Ir ratio of Repeev Khutor, IIF iron group, (16.6 and 4.8, respectively) exclude any relationship between NWA 6583 (Ge/Ga = 2.4 and Ni/Ir = 94.7) and the IIF irons group.

Fig. 9. Bulk concentration of siderophile and chalcophile elements of NWA 6583 normalized to the CI chondrite (McDonough and Sun 1995). In the diagram are also plotted the compositional fields of the IVB iron group and of the IAB-complex irons with Ni contents between 13.5 wt% and 22.5 wt% (see the text). Elements are ordered from left to right by decreasing 50% condensation temperature. Source of data: group IVB irons, Walker et al. (2008); IAB-complex irons, Wasson and Kallemeyn (2002); the Pd contents of Dayton and Gay Gulch (IAB- complex), Hoashi et al. (1993). According to Wasson and Kallemeyn (2002), irons belonging to the large IAB- complex should have Au > 1.3 µg g-1, As > 10 µg g-1, Co > 0.39 wt%, Sb > 0.18 µg g-1 and 0.4 ≤ Ge/Ga ≤ 7. Northwest Africa 6583 matches this classification scheme for all elements but Au (1.13 µg g-1; Table 1). Despite its Au contents the iron meteorite NWA 5804 (1.04 µg

189 g-1; Meteoritical Bulletin, no. 100, in preparation) has been classified as IAB-ungrouped. Wasson and Kallemeyn (2002) in their work about the IAB-complex reported that despite their Au contents Zacatecas 1792, NWA 176 and Bocaiuva could be considered as IAB- related (?). Northwest Africa 6583, NWA 5804, NWA 176, Bocaiuva, and Zacatecas 1792 define a positive trend in the diagram Ni vs Au and it is parallel to the overall trend defined by the IAB-complex irons (Fig. 10). However, the creation of a new low-Au IAB-subgroup is premature at this stage, also because the oxygen isotope composition of Bocaiuva and NWA 176 (∆'17O = -4.39 ‰ and ∆'17O = -5.21 ‰, respectively, Malvin et al. 1985 and Liu et al. 2001) is very different from that majority of IAB-complex members (NWA 6583 inclusive).

Fig. 10. Log Ni-log Au diagram for NWA 6583, the IAB main group (MG), the five IAB-complex subgroups (IAB-sLL, IAB-sLM, IAB-sLH, IAB-sHH, and IAB-sHL), the IAB-ungrouped NWA 5804 and the IAB- related irons NWA 176, Bocaiuva and Zacatecas 1792 (data for IAB-complex, NWA 176, Bocaiuva and Zacatecas 1792 are from Wasson and Kallemeyn 2002, data for NWA 5804 are from Meteoritical Bulletin Database). Silicate inclusions in iron meteorites are reported in the IAB, IIE, and IVA iron groups (e.g., Haack and McCoy 2003). The occurrence of silicate inclusions in NWA 6583 is consistent with IAB-complex iron classification. Their overall mineral composition is most similar to the reduced assemblages found in silicate inclusions in IAB-complex irons, in winonaites, enstatite chondrites and aubrites. However, the whole-rock oxygen isotope compositions of enstatite chondrites and aubrites plot on the terrestrial fractionation line (Δ’17O from 0.15‰ to -0.37‰, Clayton et al. 1984) ruling out a possible link between NWA 6583 with enstatite chondrites (EH and EL) and aubrites (Fig. 8). The mineral compositions of olivine and pyroxene in NWA 6583 (olivine Fa1.3, enstatite Fs0.44 ± 0.22, Wo1.57 ± 0.15, diopside Fs1.52 ± 0.75, Wo44.3 ± 6.0; Table 4) matches the compositions of the silicate inclusions of the most reduced IAB-complex irons like Burkhala, Pine River, Kendall County, Elephant Moraine-84300, and of the winonaites Pontlyfni, Queen Elizabeth Range-945335, Yamato-8005, -75300, -74025, -75303, -75261 (Table 6).

190

Table 6. Main features of NWA 6583 and IAB-complex irons and winonaites containing low-Fe pyroxenes and olivine, and Fe-Mn-Mg-Zn sulfides.

Meteorite Classif. Ni (wt%) ref.a olivine Ca-poor pyroxene Ca-rich pyroxene ref. alabandite niningerite Zn-sulfide ref. Δ'17Ob ref.

Fac Fsc Woc Fs Wo

NWA 6583 18 (1) 1.3d 0.4 1.6 1.5 44.3 (1) xe x x (1) -0.44 (1)

EET 84300 IAB-ungr 10.1 (2) 0.8 6 - - - (3,4) -0.52 (5) Kendall IAB-ungr 5.6 (2) - 1 0.7 0.8 44.8 (6) -0.32 (5) County Pine River IAB-sLL 8.0 (2) 1.0 4 1.2 1.9 44.3 (6) -0.54 (5)

Burkhala IAB - ungr 0.01 - - 1.5 48.3 (7) x - x (7)

Y-8005 Winonaite 6.4 (8) 1.2-2.1 2.0-2.2 1.5 1.5 46.1 (9) -0.53 (5)

Pontlyfni Winonaite 0.7-1.1 0.5-1.2 1.2-1.6 0.4-1.1 45.5-46.1 (9) -0.54 (10)

QUE 94535 Winonaite 1-3 1-2 - - - (9) -0.67 (5)

Mt. Morris Winonaite 7.0 (10) 1.3-3.6 4.1 1.6 1.7 45.5 (10,9) x - x (11) -0.50 (10)

Y-75300 Winonaite 6.4 (30.4) (12) 1.7-1.8 1.9-2.4 1.5-1.6 0.9-1.1 46.6-46-7 (9,13) -0.50 (10) 6.1-6.5 Y-74025 Winonaite (9,13) 1.8 2.3 1.7 0.8 45.9 (9,13) -0.75 (10) (41.6) Y-75305 Winonaite 5.8 (40.6) (12) 1.8 1.9 1.7 0.9 47.7 (9,13) x - - (12) -0.53 (5)

Y-75261 Winonaite 0.3 0.3 - - - (9) -0.42 (10) aReference: (1) this work, (2) Wasson and Kallemeyn 2002, (3) Benedix et al. 2000, (4) Antarctic Meteorite Newsletter, 9, 1986, (5) Clayton and Mayeda 1996, (6) Bunch et al. 1970, (7) Yaroshevskiy et al. 1989, (8) Yugami et al. 1996, (9) Benedix et al. 1998, (10) Bild 1977, (11) Prinz et al. 1980, (12) Kimura et al. 1992. bΔ'17O = 1000 ln (1+ (δ17O/1000)) – λ1000 ln (1+ (δ18O/1000)) where λ = 0.5247 (Miller 2002) cMolar fraction of Fa, fayalite; Fs, ferrosilite; Wo, wollastonite. dTrace amounts confirmed by SEM-EDS analyses. ex = observed; - = not observed.

Northwest Africa 6583 contains Fe-Mn-Mg-Zn sulfides as frequently observed in the most reduced inclusions of the IAB-complex. However, while in IAB-complex irons Fe-Mn- Mg sulfides occur as almost Mg-free alabandite, in NWA 6583 they give rise to an almost complete alabandite-niningerite solid solution (Fig. 6a). Note that this is the first report of such an extended solid solution in the alabandite-niningerite series. The Mn-rich composition of the (Fe,Zn,Mn)S in NWA 6583 (Mn ~ 25 wt%; Table 3), although atypical of the IAB- complex, is similar to that observed in the Waterville IAB-complex iron (Mn ranges from 2.31 wt% to 23.4 wt%; Weinke et al. 1979) (Fig. 6b). Thus, the overall mineralogical composition of the inclusions is also consistent with the classification of NWA 6583 as a new member of the IAB-complex.

Petrogenesis of NWA 6583 Textural and chemical constraints

The silicate inclusions in the NWA 6583 metal consists of euhedral to subhedral individual crystals and polycrystalline aggregates with subhedral textures (Fig. 7a). The contacts between silicate inclusions and the surrounding metal matrix are igneous and defined by the euhedral to subhedral crystal boundaries of the silicates (Fig. 7b,c). Complex textural relationships are observed between enstatite and troilite (Fig. 7f). Their contacts are decorated by fine intergrowths of diopside plus sulfide that require further studies. No angular silicate inclusions (evidence of clastic origin) commonly reported in silicate inclusions in iron meteorites (e.g., Benedix et al. 2000) were observed in NWA 6583. Textural relationship thus indicates that NWA 6583 formed through the crystallization of a melt composed of two immiscible liquids of silicate and metallic compositions. The occurrence of silicate inclusions consisting exclusively of individual crystals of enstatite suggests very high melt temperatures close or above the enstatite melting point ~ 1570°C.

The metal composition of NWA 6583 is close to the cosmic abundances (from 0.4 x CI for Ir to 18.8 x CI for Sb) and shows a modest enrichment of incompatible elements (Fig. 9). The compositions of the silicates are broadly chondritic. These lines of evidence indicate the formation of the parent melt from an unfractionated precursor and limited parent-body processing.

The homogeneous composition of the silicate crystals suggests equilibrium crystallization at very high temperatures whereas the subliquidus history recorded by the surrounding metal is compatible with a relatively fast cooling. Northwest Africa 6583 is made

193 of 0.5-20 mm precursor g-iron crystals (Fig. 2). They are much smaller than the g-iron crystals of most iron meteorite groups; as an example IIIAB irons have their largest g-iron crystals > 2 m (Buchwald 1975). The small size of the NWA 6583 ataxitic (Fe,Ni) metal crystals suggests a rapid cooling of the metal liquid through the g-iron + liquid and the g-iron stability fields. Polycrystalline textures with small size (from 2 cm to 40 cm) of g-iron crystals are a common feature of IAB-complex irons (Wasson and Kallemeyn 2002). Polycrystalline textures of similar grain-size (namely, similar precursor g-iron crystals) are observed in Mundrabilla (grain size 2 – 5 cm, Buchwald 1975) and interpreted by Scott (1982) as the result of cooling rates of the order of ~ 6.3°C yr-1 at solidification temperatures. The lack of Widmanstätten pattern in the (Fe,Ni) metal grains suggests cooling rates faster than ~ 0.01°C yr-1 at temperatures between 500°C and 700°C (Goldstein et al. 2009). Their martensitic structure documents very fast cooling rates at even lower temperatures between

650°C and 290°C for a metal composition Fe82Ni18. Below the temperature of the onset of the reaction γ  γ + α of 650 °C relatively high cooling rates are indeed required to preserve γ- iron in a metastable state down to the α2 (martensite) inversion temperature of ~290°C (Goldstein et al. 2009). Redox formation conditions

The very low Fe content of silicate inclusions, the occurrence of graphite and Mg-Mn sulfides, the occurrence of Cr, Mn, and Ti in troilite and the high Si content of (Fe,Ni) metal indicate that NWA 6583 equilibrated under very low oxygen fugacity conditions. In this section we will examine our data to constrain the redox conditions under which NWA 6583 formed.

In most meteorites Fe is partitioned among metal, sulfide and silicates. In highly reducing environments, Fe is virtually excluded from silicates (Reid and Cohen 1967), and its siderophile and chalcophile behavior increases. The oxygen fugacity under which an assemblage containing Fe-Ni-(Si) metal and pyroxenes equilibrated may be estimated on the basis of the FeO content of enstatite following the equation:

ln fO = 2/ 3 ΔG 0 / RT + ln a pyroxene − ln a metal − ln a metal 2 [ R ()FeSiO3 ()Fe ()Si ]

0 where f is the oxygen fugacity, ΔGR is the Gibbs free energy difference of the reaction O2

€ Fe (metal) + Si (metal)+ 3/2 O2 (gas)= FeSiO3 (pyroxene)

€ €

194 pyroxene metal metal and a , a and a are the activities of FeSiO3 in enstatite, Fe in the metal and Si FeSiO3 Fe Si in the metal, respectively. Evidence for equilibrium conditions of the NWA 6583 phases involved in the reaction includes the prevailing euhedral shapes of the enstatite crystals and € € € their homogeneous composition (Tables 2 and 4).

An equilibrium temperature of ~ 900°C has been estimated using the two pyroxene

geothermometer of Wells (1977). The activity of FeSiO3 in pyroxene has been calculated

metal metal using the ideal site mixing model of Ghiorso and Carmichael (1980), while a Fe and a Si have been assumed to be equal to their mole fractions (e.g., Kilburn and Wood 1997), respectively. The value of ∆G0 (P,T) was calculated at 1 bar and 900 °C using the R € € thermodynamic values from Berman (1988) for ferrosilite, and from Pankratz (1982) for iron, silicon and oxygen. The value of oxygen fugacity calculated in this way is -3.8 log units € below the Iron-Wüstite (IW) buffer (∆IW = -3.8). A similar approach was used by Benedix et al. (2005) for IAB-complex irons and Righter and Drake (1996) for IAB-complex and IIE irons, winonaites, acapulcoites, SNC and ordinary chondrites.

Experimental studies indicate that monosulfides of the lithophile elements Mg and Mn are stable only at very low f , i.e., ∆IW < -3; (Siebert et al. 2004; Berthet et al. 2009). O2 The stability of graphite is controlled by oxygen fugacity, pressure and temperature.

At a given oxygen€ fugacity, graphite stability increases with pressure and decreases with temperature (Rubin 1997). However, as the parent bodies of meteorites generally have small diameters, graphite may be stable only at low values of oxygen fugacity. As an example, within an asteroidal body with a mean density of 3340 kg m-3 and a diameter of 100 km, graphite could be stabilized only at ∆IW < 1 at a depth of 10 km and a temperature of 900 °C.

Kilburn and Wood (1997) showed how the siderophile behaviour of Si increases as oxygen fugacity decreases. In their experiments, the (Fe,Ni) metal phase contained 0.37 wt% Si at ∆IW = -4.0. Extrapolating the data of Kilburn and Wood (1997) to the Si content of the metal of NWA 6583 (0.13 wt%) we estimate a ∆IW value of -3.5.

Chromium, like other lithophile elements (e.g., Mn and Ti) shows an increase in its chalcophile behaviour as the oxygen fugacity decreases. Thus, the relatively high Cr content of troilite of NWA 6583 (~ 0.8 wt%; Table 2) is again suggestive of highly reducing conditions.

195 All these data indicate that the mineral assemblage of NWA 6583 formed under very low oxygen fugacity conditions (~ -3.5 ∆IW; Fig. 11). These conditions are intermediate between those characterizing enstatite chondrites and aubrites (from -6 ∆IW to -4 ∆IW; Righter and Neff 2007), and IAB-complex irons and winonaites (from -2.8 ∆IW to -1.5 ∆IW; Righter and Drake 1996; Righter and Neff 2007).

Fig. 11. Histogram of ΔIW values for NWA 6583, IAB-complex irons, winonaite-acapulcoites and EH and EL chondrites and aubrites (modified after Righter and Drake 1996 and Righter and Neff 2007). The number of samples within each group of meteorites is represented by (n).

Formation process and geological setting

The fact that NWA 6583 is the product of the crystallization of a very high temperature melt consisting of a mixture of metal and subordinate silicate liquids with virtually undifferentiated composition suggests formation through total melting of a chondritic precursor. Total melting is characteristically attained through impact melting. Consistently, the textural and mineralogical records of a relatively rapid cooling history at subliquidus temperatures suggests that NWA 6583 crystallized in a surface or subsurface environment of the parent body. Furthermore, silicate inclusions are frequently observed in impact melt irons (e.g., Buchwald 1975; Schrader et al. 2010).

Impact melt rocks on asteroids may occur in a variety of geological settings including dikes and veins in the basements of craters, melt lumps and melt sheets in the breccia layer at the crater floor, melt splashes at the top or within the ejecta blanket, (e.g., Stöffler et al. 1991; Taylor et al. 1993; Keil et al. 1997), or effectively differentiated metal and silicate melt sheets at the crater floor (e.g., Vickery and Melosh 1983; Gaffey and Gilbert 1998; Folco et al. 2004; Ruzicka et al. 2005). Molten debris liberated during catastrophic break up of the parent

196 asteroid, and eventually reassembled in an offspring rubble-pile is an additional possibility (e.g., Benedix et al. 2000; Asphaug et al. 2011).

The polycrystalline texture and the modest enrichment in incompatible elements in the metal favours crystallization of NWA 6583 in a subsurface environment, e.g., veins and dikes in the basements of craters, or melt lumps and melt sheets within the breccia layer at the crater floor, where slow enough cooling rates could allow some chemical fractionation of the metal and the formation of its polycrystalline texture. The lack of metal-silicate fractionation in NWA 6583 rules out crystallization in large differentiated melt sheets at crater floor (e.g., Folco et al. 2004).

Lastly, the martensitic structure within metal grains indicative of very fast cooling rates at in the 650°C - 290°C temperature range (see Textural and chemical constraints) suggests, however, final cooling in a surface environment. This could be explained through excavation during a later impact, most likely by spallation to account for the lack of deformation in NWA 6583.

Some troilite crystals are polycrystalline and show melting textures and lamellar twinning. Some schreibersite crystals also show melting texture. These effects are localized within 2 mm from the external surface of the meteorite and are due to reheating during ablative flight through the Earth atmosphere of NWA 6583. The melting of schreibersite indicates that the temperature was higher than 1000 °C (Axon 1963).

Implication for the IAB-complex parent body

The IAB-complex is the second largest iron meteorite group. It is characterized by a large variability in the composition of the metal (Wasson and Kallemeyn 2002), type and composition of the silicate inclusions (Benedix et al. 2000), and oxygen fugacity conditions of formation (Righter and Drake 1996; Righter and Neff 2007). Because of this large variability, the origin and evolution of the IAB-complex iron meteorites is a matter of debate (e.g., Haack and McCoy 2003).

Wasson and Kallemeyn (2002) argued that the main group was not formed by fractional crystallization due to their unfractionated (or nearly so) geochemistry. Rather, its formation involved modest crystal segregation with solid and melt essentially in equilibrium in isolated impact melt pools on the surface of a porous chondritic body, and multiple impact- induced melting events created the different subgroups. Other formation models include crystallization of a sulfur- and carbon-rich core in a partially differentiated object (Kracher

197 1985; McCoy et al. 1993), or catastrophic breakup and reassembly of a partially differentiated object at its peak temperature (namely, 1200 - 1400°C) 4.4 - 4.54 Ga ago (Benedix et al. 2000). These clustered ages would support the catastrophic breakup model for IAB-complex iron parent body. Alternatively, they would document a period of intense bombardment in the history of the IAB-complex iron parent body.

The impact melting scenario envisaged for NWA 6583 in the previous section fits the model by Wasson and Kallemeyn (2002). Furthermore, it is difficult to explain the intimate coexistence in NWA 6583 of metal and silicate liquids (which are characterized by contrasting densities) within the context of planetary differentiation.

Remarkably, NWA 6583 shows a number of peculiarities relative to other members of the IAB-complex group of meteorites. Northwest Africa 6583 is extremely reduced: the record of its unusual mineral composition widens the range of redox formation conditions of the IAB-complex irons down to ∆IW ~ -3.5 (Fig. 11). Furthermore, NWA 6583 has a chemical composition characterized by high Ni and very low Au (i.e., Ni = 18 wt %, Au = 1.13 µg g-1), which deviates somewhat from the classification compositional scheme proposed by Wasson and Kallemeyn (2002) for the IAB-complex irons. The unusual geochemical composition and redox formation condition of NWA 6583 thus significantly extends the properties of the IAB-complex. This, once again, poses the question whether the members of the IAB-complex irons formed in a single parent asteroid or in several similar parent asteroids.

Wasson and Kallemeyn (2002) suggested that the IAB-complex formed in different impact melt pools on the surface of a single parent asteroid. This model finds support on their similar metal composition and the overlapping oxygen isotope composition of the IAB- complex main group and subgroups sLH, sLM, and sLL. Sombrerete, member of the sHL subgroup, is however an exception having an oxygen isotope composition (∆17O = -1.39‰; Clayton and Mayeda 1996) which is much more enriched in 16O than the typical IAB composition. Therefore, sHL could have formed in a distinct parent asteroid (Wasson 2011).

The oxygen isotope composition of NWA 6583 is typical of the IAB-complex irons (Fig. 8). Thus, the unusual geochemical composition and redox formation conditions of NWA 6583 provide a further example of the lithological and mineralogical variety that impact melting could produce on the surface of a single asteroid. Such variety requires that the surface of the parent body of the IAB-complex irons was characterized by an important compositional heterogeneity in space and time (e.g., a regolith) and by an important

198 heterogeneous distributions of reducing agents (e.g., C- or S-rich materials?). For example, carbonaceous chondrite-like xenoliths are reported in HED meteorites and H chondrites by Gounelle et al. (2003) and Briani et al. (2012) respectively, and interpreted as fossil micrometeorites. Similarly to what suggested for the parent bodies of the HED meteorites, Vesta, and H chondrites, the parent body of the IAB-complex irons might have formed in a region of the Solar System subject to a flux of exotic reducing material like materials with carbonaceous chondritic composition.

SUMMARY AND CONCLUSIONS

Northwest Africa 6583 is a silicate-bearing iron meteorite. In terms of texture and chemical composition NWA 6583 can be classified as a Ni-rich (18 wt%) polycrystalline (grain-size up to 22 mm) ataxite. The oxygen isotope composition of its silicates indicates a genetic link with the IAB-complex irons. The bulk metal composition of NWA 6583 matches the classification scheme for the IAB-complex with the exception of lower Au content (1.13 µg g-1). The overall mineralogical composition of the inclusions is consistent with several inclusions of the IAB-complex irons. Thus, NWA 6583 has been classified as a new member of the IAB-complex.

The textural features of the silicate inclusions of NWA 6583 indicate that the meteorite is the product of the crystallization of a very high temperature melt (≥ 1570°C) consisting of a mixture of metal and subordinate silicate liquid. The chemical composition of the metal is close to the cosmic abundance suggesting that the metallic liquid formed through total melting of a chondritic precursor. Total melting is considered to result from impact melting. An impact melting origin is also consistent with the occurrence of silicate inclusions and with the rapid cooling history at subliquidus temperatures recorded by the meteorite texture. According to Scott (1982), polycrystalline textures with similar grain size in Mundrabilla record cooling rate of ~ 6.3°C yr-1. Furthermore, the lack of the Widmanstätten pattern in NWA 6583 indicates cooling rates faster than ~ 0.01°C yr-1 at temperatures comprised between 500°C and 700°C; Goldstein et al. 2009). It is probable that the crystallization of NWA 6583 occurred in a subsurface environment to allow the nucleation of numerous g-iron crystals and a modest enrichment in incompatible elements. The martensitic structure of the metal indicates a very fast final cooling in a surface environment.

The proposed model for NWA 6583 fits the model by Wasson and Kallemeyn (2002) for the petrogenesis of the IAB-complex irons. Nevertheless, some unique elements distinguish NWA 6583 from known IAB-complex irons: the low Au content of the metal, the

199 extremely reducing condition of formation (~ -3.5 ΔIW), and the occurrence of an almost complete alabandite niningerite solid solution of the Fe-Mn-Mg sulfides. All these features extend the properties of the IAB-complex. Probably NWA 6583 and the other IAB-complex irons with similar oxygen isotope composition could from in the same parent body. Indeed, impact melting can generate a wide variability of rocks, especially if the parent body is characterized by an important compositional heterogeneity in space and time (e.g., a regolith). Impactors with different contents of reducing agents could also play an important role in determining the large range of the redox conditions of the IAB-complex.

Acknowledgments-This work was supported by the Italian Ministry of Education, University and Research (MIUR-PRIN 2008 project, code 008222KBS_005). Luigi Folco is supported by the Italian Programma Nazionale delle Ricerche in Antartide (PNRA) and the Italian Ministero degli Affari Esteri Progetti di Grande Rilevanza. The authors are grateful to Raul Carampin for assistance during electron microprobe analyses at CNR (Consiglio Nazionale delle Ricerche) Istituto di Geoscienze e Georisorse in Padua, and to Mirko Graul for generously providing us the study samples of NWA 6583. S. Kissin and D. Schrader are thanked for constructive reviews, and Nancy Chabot for editorial handling.

REFERENCES Alexandre A., Sonzogni C., Basile I., Sylvestre F., Parron C., Meunier J. D., and Colin F. 2006. Oxygen isotope analyses of fine silica grains using laser-extraction technique: comparison with oxygen isotope data obtained from ion microprobe analyses and application to quartzite and silcrete cement investigation. Geochimica et Cosmochimica Acta 70:2827–2835. Antarctic Meteorite Newsletter 1986. 9:1. Asphaug E., Jutzi M., and Movshovitz N. 2011. Chondrule formation during planetesimal accretion. Earth and Planetary Science Letters 308:369-379. Axon H. J. 1963. Destruction of the Widmanstätten structure in iron meteorites by laboratory heat treatment. Nature 197:1291. Benedix G. K., McCoy T. J., Keil K., Bogard D. D., and Garrison D. H. 1998. A petrologic and isotopic study of winonaites: Evidence for early partial melting, brecciation, and metamorphism. Geochimica et Cosmochimica Acta 62:2535-2553. Benedix G. K., McCoy T. J., Keil. K., and Love S. G. 2000. A petrologic study of the IAB iron meteorites: Constraints on the formation of the IAB-Winonaite parent body. Meteoritics & Planetary Science 35:1127-1141.

200 Benedix G. K., Lauretta D. S., McCoy T. J. 2005. Thermodynamic constraints on the formation conditions of winonaites and silicate-bearing IAB irons. Geochimica et Cosmochimica Acta 69:5123-5131. Berman R. G. 1988. Internally-consistent thermodynamic data for minerals in the system

Na2O-K2O-CaO-MgO-FeO-Fe2O3-Al2O3-SiO2-TiO2-H2O-CO2. Journal of Petrology 29:455-522. Berthet S., Malavergne V., and Righter K. 2009. Melting of the Indarch meteorite (EH4 chondrite) at 1 GPa and variable oxygen fugacity: Implication for early planetary differentiation processes. Geochimica et Cosmochimica Acta 73:6402-6420. Bild R. W. 1977. Silicate inclusions in group IAB irons and a relation to the anomalous stones Winona and Mt Morris (Wis). Geochimica et Cosmochimica Acta 41:1439-1456. Brearley A. J. and Jones R. H. 1998. Chondritic Meteorites. In Planetary Materials, edited by Papike J. J. Reviews in Mineralogy 36:3/1 – 3/398. Briani G., Gounelle M., Bourot-Denise M., and Zolensky M. E. 2012. Xenoliths and microxenoliths in H chondrites: Sampling the zodiacal could in the asteroid Main Belt. Meteoritics & Planetary Science 47:880-902. Buchwald V. F. 1975. Handbook of Iron Meteorites. Berkeley, CA: University of California Press. 1416 p. Bunch T. E., Keil K., and Olsen E. 1970. Mineralogy, and petrology of silicate inclusions in iron meteorites. Contributions to Mineralogy and Petrology 25:297-340. Clayton R. N. and Mayeda T. K. 1996. Oxygen isotope studies of achondrites. Geochimica et Cosmochimica Acta 60:1999-2017. Clayton R. N. and Mayeda T. K. 1999. Oxygen isotope studies of carbonaceous chondrites. Geochimica et Cosmochimica Acta 63:2089-2104. Clayton R. N., Mayeda T. K., and Rubin A. E. 1984. Oxygen isotopic compositions of enstatite chondrites and aubrites. Journal of Geophysical Research 89:C245-C249. D’Orazio M. and Folco L. 2003. Chemical analysis of iron meteorites by inductively coupled plasma-mass spectrometry. Geostandards Newsletter: The Journal of Geostandards and Geoanalysis 27:215-225. D’Orazio M., Folco L., Zeoli A., and Cordier C. 2011. Gebel Kamil: The iron meteorite that formed the Kamil crater (Egypt). Meteoritics & Planetary Science 46:1179-1196. Folco F., Bland P. A., D’Orazio M., Franchi I. A., Kelley S. P., and Rocchi S. 2004. Extensive impact melting on the H-chondrite parent asteroid during the cataclysmic

201 bombardment of the early solar system: Evidence from the achondritic meteorite Dar al Gani 896. Geochimica et Cosmochimica Acta 68:2379–2397. Franchi I. A. 2008. Oxygen isotopes in asteroidal materials, In Oxygen in the Solar System, edited by Rosso J. J. Reviews in Mineralogy and Geochemistry 68:345-397. Gaffey M. J. and Gilbert S. L. 1998. Asteroid 6 Hebe: The probable parent body of the H-type ordinary chondrites and the IIE iron meteorites. Meteoritics & Planetary Science 33:1281–1295. Ghiorso M. S. and Carmichael I. S. E. 1980. A regular solution model for met-aluminous silicate liquids: Applications to geothermometry, immiscibility, and the source regions of basic magmas. Contributions to Mineralogy and Petrology 71:323-342. Goldstein J. I., Scott E. R. D., and Chabot N. L. 2009. Iron meteorites: Crystallization, thermal history, parent bodies, and origin. Chemie der Erde - Geochemistry 69:293-325. Gounelle M., Zolensky M. E., Jer-Chyi L., Bland P. A., and Alard O. 2003. Mineralogy of carbonaceous chondritic microclasts in howardites: Identification of C2 fossil micrometeorites. Geochimica et Cosmochimica Acta 67:507:527. Haack H. and McCoy T. J. 2003. Irons and Stony Irons. In Meteorites, comets, and Planets , edited by A.M. Davis. Vol. 1 Treatise on Geochemistry edited by H. D. Holland and K. K. Turekian, Elsevier Pergamon, Oxford, 325-345. Hoashi M., Brooks R. R., Reeves R. D. 1993. Palladium, platinum and ruthenium in iron meteorites and their taxonomic significance. Chemical Geology 106:207-218. Keil K. 2007. Occurrence and origin of keilite, (Fe<0.5,Mg>0.5)S, in enstatite chondrite impact-melt rocks and impact-melt breccias. Chemie der Erde - Geochemistry 67:37-54. Keil K., Stöffler D., Love S. G., and Scott E. R. D. 1997. Constraints on the role of impact heating and melting in asteroids. Meteoritics & Planetary Science 32:349-363. Kilburn M. R. and Wood B. J. 1997. Metal–silicate partitioning and the incompatibility of S and Si during core formation. Earth and Planetary Science 152:139-148. Kimura M., Tsuchiyama A., Fukuoka T., and Iimura Y. 1992. Antarctic primitive achondrites, achondrites, Yamato-74025, -75300, and -75305: Their mineralogy, thermal history and the relevance to winonaite. Proceedings NIPR Symposium Antarctic Meteorites pp. 165- 190. Kracher A. 1985. The evolution of partially differentiated planetesimals: Evidence from iron meteorite groups IAB and IIICD. Proceedings, 15th Lunar and Planetary Science Conference. pp. C689-C698.

202 Kracher A., Wills J., and Wasson J. T. 1980. Chemical classification of iron meteorites—IX. A new group (IIF), revision of IAB and IIICD, and data on 57 additional irons. Geochimica et Cosmochimica Acta 44:773-787. Liu M., Scott E. R. D., Keil K., Wasson J. T., Clayton R. N., Mayeda T., Eugster O., Crozaz G., and Floss C. 2001. Northwest Africa 176 a unique iron meteorite with silicate inclusion related to Bocaiuva (abstract #2152). 32nd Lunar and Planetary Science Conference. Ma C., Beckett J. R., and Rossman G. R. 2012. Buseckite, (Fe,Zn,Mn)S, a new mineral form the Zakłodzie meteorite. American Mineralogist 97:1226-1233. Malvin D. J., Wasson J. T., Clayton R. N., Mayeda T. K., and da Silva Curvello W. 1985. Bocaiuva – A silicate-inclusion bearing related to the Eagle-Station pallasites. Meteoritics 20:259-273. McCoy T. J., Keil K., Scott E. R. D., and Haack H. 1993. Genesis of the IIICD iron meteorites: Evidence from silicate-bearing inclusions. Meteoritics 28:552:560. McDonough W. F. and Sun S. S. 1995. The composition of the Earth. Chemical Geology 120: 223-254. Meteoritical Bulletin Database, http://www.lpi.usra.edu/meteor. Meteoritical Bulletin no. 100, in preparation. Miller M. F. 2002. Isotopic fractionation and the quantification of 17O anomalies in the oxygen three-isotope system: an appraisal and geochemical significance. Geochimica et Cosmochimica Acta 66:1881-1889. Mittlefehldt D. W., McCoy T. J., Goodrich C. A., and Kracher A. 1998. Non-Chondritic meteorites from asteroidal bodies. In Planetary Materials, edited by Papike J. J. Reviews in Mineralogy 36:4/1 – 4/195. Pankratz L. B. 1982. Thermodynamic Properties of Elements and Oxides. U. S. Bureau of Mines Bulletin 672. Prinz M., Waggoner D. G., and Hamilton P.J. 1980. Winonaites: A primitive achondritic group related to silicate inclusions in IAB irons (abstract). Lunar and Planetary Science 11:902-904. Reid A. M. and Cohen A. J. 1967. Some characteristics of enstatite from enstatite achondrites. Geochimica et Cosmochimica Acta 31:661-672. Righter K. and Drake M. J. 1996. Core formation in Earth’s Moon, Mars and Vesta. Icarus 124:513-529.

203 Righter K. and Neff K. E. 2007. Temperature and oxygen fugacity constraints on CK and R chondrites and implications for water and oxidation in early solar system. Polar Science 1:25-44. Rubin A. E. 1997. Igneous graphite in enstatite chondrites. Mineralogical Magazine 61:699- 703. Ruzicka A., Killgore M., Mittlefehldht D. W., and Fries M. D. 2005. Portales valley: Petrology of a metallic-melt meteorite breccia. Meteoritics & Planetary Sciences 40:261-295. Schrader D. L., Lauretta D. S., Connolly jr. H. C., Goreva Y. S., Hill D. H., Domanik K. J., Berger E. L., Yang H., and Downs R. T. 2010. Sulfide-rich metallic impact melts from chondritic parent bodies. Meteoritics & Planetary Sciences 45:743-758. Scott E. R. D. 1982. Origin of rapidly solidified metal-troilite grains in chondrites and iron meteorites. Geochimica et Cosmochimica Acta 46:813-823. Siebert J., Malavergne V., Guyot F., Combes R., and Martinez I. 2004. The behaviour of sulphur in metal–silicate core segregation experiments under reducing conditions. Physics of the Earth and Planetary Interiors 143–144:433–443. Stöffler D., Keil K., and Scott E. R. D. 1991. Shock metamorphism of ordinary chondrites. Geochimica et Cosmochimica Acta 55:3845-3867. Suavet C., Alexandre A., Franchi I. A., Gattacceca J., Sonzogni C., Greenwood R. C., Folco L., and Rochette P. 2010. Identification of the parent bodies of micrometeorites with high-precision oxygen isotope ratios. Earth and Planetary Science Letters 293:313-320. Taylor G. J., Keil K., McCoy T., Haack H., and Scott E. R. D. 1993. Asteroid differentiation: Pyroclastic volcanism to magma oceans. Meteoritics 28:34-52. Vickery A. M. and Melosh H. J. 1983. The origin of SNC meteorites: An alternative to Mars. Icarus 56:299-318. Wai C. M. and Wasson J. T. 1969. Silicon concentrations in the metal of iron meteorites. Geochimica et Cosmochimica Acta 33:1465-1471. Walker R. J., McDonough W. F., Honesto J., Chabot N. L., McCoy T. J., Ash R. D., and Bellucci J. J. 2008. Modeling fractional crystallization of group IVB iron meteorites. Geochimica et Cosmochimica Acta 72:2198–2216. Wang P., Rumble D. III, McCoy T. J. 2003. Oxygen isotopic compositions of IVA iron meteorites: Implications for the thermal evolution derived from in situ ultraviolet laser microprobe analyses. Geochimica et Cosmochimica Acta 68:1159-1171.

204 Wasson J. T. 2011. Relationship between iron-meteorite composition and size: compositional distribution of irons from North Africa. Geochimica et Cosmochimica Acta 75:1757- 1772. Wasson J. T. and Kallemeyn G. W. 2002. The IAB iron-meteorite complex: A group, five subgroups, numerous grouplets, closely related, mainly formed by crystal segregation in rapidly cooling melts. Geochimica et Cosmochimica Acta 66:2445-2473. Wasson J. T., Ouyang X., Wang J., and Eric J. 1989. Chemical classification of iron meteorites: XI. Multi-element studies of 38 new irons and the high abundance of ungrouped irons from Antarctica. Geochimica et Cosmochimica Acta 53:735-744. Weinke H. H., Kiesl W., and Clarke R. S. jr. 1979. Mineralogical and chemical investigation of the Waterville Iron Meteorite (abstract). Meteoritics 561-564. Wells P. R. A. 1977. Pyroxene thermometry in simple and complex systems. Contribution to Mineralogy and Petrology 62:129-139. Yaroshevskiy A. A., Migdisova L. F., and Konokova N. N. 1989. The mineral assemblage in the sulfide nodules of the Burkhala iron meteorite and its formation conditions. Geokhimiya 6:825-837. Yugami K., Takeda H., Kojima H., and Miyamoto M. 1996. Distribution of opaque minerals in primitives achondrites. Lunar and Planetary Science 27: 1483-1484.

205

APPENDIX III. OTHER ACTIVITIES

• Educational support

A.A. 2012-2013 and A.A. 2013-2014. Lecture “Laboratorio di Petrografia” (Optical microscope) for BSc students (4 hours per week).

• Fieldwork Activity

20th November 2012 – 20th January 2013. Member of the XXVIII Italian Expedition in Antarctica (Programma Nazionale di Ricerche in Antartide – Project “Meteoriti Antartiche”).

• Grants and awards

Barringer Award 2014 (5000 $) recipient for a project entitled “Combined micro- Raman and TEM study of high-pressure phases from Kamil Crater (Egypt): implications for their formation in small impact craters on Earth”.

Travel grant “Brian Mason Awards” (1000 $) recipient for the international conference of the Meteoritical Society (8th – 13th September 2014, Casablanca, Morocco) by the IMCA – International Meteorite Collectors Association.

• Conference participations

21st – 24th September 2014. GeoFrankfurt Conference, Frankfurt, Germany. Contribution on shock effects recorded at Kamil Crater (oral communication).

8th – 13th September 2014. Metoritical Society Conference (MetSoc), Casablanca, Morocco. Two contributions on shock effects recorded at Kamil Crater (oral communication) and on target-projectile interaction during impact melting at Kamil Crater (poster session).

16th – 18th September 2013. Chairman at the IX Forum Italiano di Scienze della Terra – FIST Geoitalia 2013 – session of Planetary Geology. Two contributions on shock metamorphism at Kamil Crater (oral communication) and on Northwest Africa 6583 meteorite (poster session).

• School and workshop participations

25th August – 4th September 2014. EMU School in Planetary Mineralogy, Glasgow, United Kingdom.

207 19th September 2013. Workshop on “Earth Science Communication at Museums: Strategies and Tools”, organized in the framework of the IX Forum Italiano di Scienze della Terra – FIST Geoitalia 2013.

1st – 6th September 2013. 7th Scuola Nazionale Scienza, Comunicazione, Società SCS (School on Science Communication): Il ricercatore visibile –Strumenti per comunicare la Scienza, Pollenzo (CN), Italy. The school was organized by Agorà Scienze – Centro Interuniversitario del Piemonte.

24th – 25th October 2012. 1st Scuola di applicazione della Spettroscopia Raman alle Scienze della Terra. Università Bicocca, Milano Italy.

• Science communication activities

15th February 2015. Educational seminar entitled “Le meteoriti” at the Municipality of Alfianello (BS) on the occasion of the 132th anniversary of the fall of the meteorite Alfianello.

January 2015. Informative paper on Meteorite-Times online magazine entitled “Kamil Crater (Egypt) a natural laboratory to study shock metamorphism and impact melting”.

20th August 2014. Participation in the live radio broadcast of Radio3Scienza of the station Rai Radio 3 to talk about meteorite, Antarctica, and Agnesearth blog.

25th October 2013. Educational seminar entitled “Antartide: un continente laboratorio. A caccia di meteoriti” at the Municipality of Cascina (PI), Italy, addressed to high-school students.

Since September 2013. During the practical hours of the School on Science Communication I lunched a scientific blog about meteorites and my academic activities. http://agnesearth.wordpress.com/

19th – 20th April 2013. Two educational seminars about meteorites and meteorite search in Antarctica within the project “Adotta una scuola” of the Programma Nazionale di Ricerche in Antartide at the Istituto Geotecnico Minerario "U. Follador” of Agordo (BL), Italy, and at the Municipality of Pedavena (BL), Italy.

208 ACKNOWLEDGMENTS

Questa è la parte della Tesi che mi viene più difficile da scrivere. Ma sento di doverla scrivere. La devo scrivere perché non l’ho fatto né alla Triennale né alla Magistrale. La devo scrivere perché è giusto che chi mi ha aiutata, consigliata, supportata, sopportata e ascoltata durante questi anni trovi spazio in questo lavoro.

Non è facile e non so bene neanche da dove iniziare. Mi sono trasferita a Pisa a 19 anni per studiare Geologia carica di sogni e di speranze. Non avevo, a dire il vero, piena coscienza di quello che mi sarebbe aspettato nei mesi e negli anni seguenti, sapevo però che stavo facendo la cosa giusta. Sapevo che stavo inseguendo un sogno. Un sogno che mi portavo dentro da quando avevo 10 anni: studiare la Terra. E come in una gara, esame dopo esame, ho raggiunto il primo traguardo, la Laurea Triennale, poi meno di due anni dopo è arrivato il secondo, la Laurea Magistrale, e adesso 8 anni e 6 mesi dopo la mia immatricolazione all’UniPi è arrivato il terzo traguardo, quello più alto, quello più ambito: il Dottorato. Questi 8 anni e mezzo sono trascorsi tra alti, altissimi, bassi e bassissimi. Sono accadute cose che non avrei mai immaginato potessero accadermi, ho realizzato sogni prima che potessi iniziare a sognarli, mi sono presa tante piccole e grandi soddisfazioni, ho pianto di gioia e disperazione, ho sbagliato, mi sono sentita sola, ho avuto paura, ho rischiato di perdere la strada. Non so cosa sarà di me dopo il raggiungimento di questo Traguardo con la T maiuscola. Non è facile… alterno momenti in cui so quello che voglio a momenti in cui mi sento sola, persa in mezzo ad un oceano senza sapere in che direzione nuotare… Quello che è certo è che devo nuotare in una direzione, senza nuotare il rischio di affogare è alto! Qualunque sarà la direzione che sceglierò nei prossimi mesi, spero che mi permetta di crescere, migliorare e arricchirmi personalmente, culturalmente e scientificamente.

E’ giunto il momento di ringraziare tutti coloro che hanno fatto parte in vario modo e in varia misura di questo lungo cammino. Spero di non dimenticare nessuno.

Il primo grazie è per chi mi ha seguito durante tutti e tre i miei traguardi universitari: Prof. Massimo D’Orazio. Il secondo grazie è per il Dott. Luigi Folco. Senza di voi non avrei avuto modo di affacciarmi a questo mondo così affascinante fatto di sassi che cadono dal cielo e di strutture geologiche che ci ricordano che la Terra è parte di un sistema più grande e ancora per la sua grande parte inesplorato. Grazie perché senza di voi non avrei mai potuto vivere tutte le esperienze indimenticabili di questi anni. Grazie per i numerosi consigli, la

209 disponibilità e la pazienza. Grazie per i rimproveri e per i complimenti. Scusatemi se non sono sempre stata all’altezza del mio ruolo. Grazie ai revisori di questa Tesi, Prof.ssa Maria Chiara Domeneghetti e Dott. Gabriele Giuli, coinvolti con poco preavviso e inaspettatamente in questo lavoro. Grazie per i vostri commenti, correzioni e soprattutto per la vostra disponibilità e puntualità. Grazie ai miei compagni di “stanza”, a quelli che sono ancora presenti e anche a quelli che hanno proseguito per altre strade. Grazie per le dritte informatiche e geologiche, grazie per i consigli estetici su figure, presentazioni e poster, grazie per avermi aiutata nella preparazione dei campioni, al SEM e all’ICP-MS, grazie per avermi sostituita alle esercitazioni, grazie per aver ascoltato le mie lamentele. Grazie per non avermi fatto sentire più sola di quanto spesso non mi sia sentita. Grazie a tutto il DST-UniPi, una sorta di grande e strana famiglia. Grazie a tutti i professori e i tecnici che ho incontrato durante questi anni di analisi per la mia tesi. In particolare ringrazio la Prof.ssa Maria Luce Frezzotti per avermi introdotta nel mondo della microscopia Raman, per i consigli e per le lunghe giornate trascorse insieme a Milano. Grazie a Carole Cordier per il supporto analitico e logistico durante la mia permanenza a Grenoble e per avermi inviato correzioni e commenti alle bozze di abstract e articoli sempre in tempi record.

Grazie a chi mi ha sempre amata e non smetterà mai di farlo, grazie a chi è stato sempre dalla mia parte e lo sarà per sempre, grazie ai miei sostenitori morali ed economici: grazie alla mia famiglia, alla mia Mamma Loredana e al mio Papà Pietro. Grazie a mia sorella, migliore amica e coinquilina Ludovica. Nonostante ogni tanto mi fai arrabbiare, devo ammettere che sono di più le volte che mi hai aiutata anche involontariamente. Forse senza di te in questi anni avrei pianto di più, avrei mangiato di meno, avrei avuto il frigo più vuoto di quello che normalmente è e sarei arrivata più volte al punto di non sapere che mettermi perché era tutto nel cesto della biancheria da lavare. Grazie Mamma, Papà, Ludo because we’re the happy family…. Grazie ai miei nonni Salvatore, Graziella, Paolo e Dora (ordine strettamente anagrafico!) per pensarmi sempre, per i pranzi e le cene, per chiedermi ancora come va a scuola, per essere orgogliosi di me. Senza offesa per nessuno, questa tesi è dedicata a Te che per tutta l’estate del 2009 mi hai chiesto insistentemente “quanto manca”, a Te che nelle ultime telefonate non mi dicevi neanche ciao, ma solo “quanto manca”, a Te che ti sei aggrappato a quel pensiero per restare in vita, a Te che hai atteso quel mio primo traguardo

210 per andartene sereno, a Te che ne sei andato esattamente un mese dopo il mio primo traguardo e 15 giorni dopo la mia festa di laurea palazzolese, quando ormai io avevo realizzato il mio primo grande sogno. Grazie ovviamente anche a tutto il resto della mia famiglia zii e cugini.

A cavallo tra famiglia e amici non potevi che non esserci tu: Giovanni. Ormai sono più di 7 anni che sei il fidanzato di Agnese (cit.) e una dozzina che ci conosciamo. Grazie per le tue poche ma buone parole, per spegnere la macchina infernale dei pensieri negativi del mio cervello, per ricordarmi che le cose brutte della vita sono altre, per spronarmi, per farmi ridere e per essermi ancora accanto.

Grazie alle “scarpe col tacco & Co.”, gli amici di una vita. Negli anni siamo cambiati, ci siamo allontanati, ma la cosa più bella è che ci ritroviamo sempre o quasi come se il tempo non fosse mai passato. Grazie agli amici pisani che di pisano hanno o hanno avuto solo il libretto dell’UniPi per aver i numerosi weekend e serate pisane. Grazie a quelli della “pausa pranzo” che ogni giorno per 45/60 minuti mi allontanano dalla mia scrivania tra risate, gossip e lezioni di economia domestica. Grazie agli amici milanesi che di milanese hanno solo la carta d’identità e forse neanche quella per aver allietato tra cene e uscite in bici e per negozi questi ultimi mesi di dottorato. Un grazie particolare va a Davide, per le lunghe chiamate su Skype, per gli SOS linguistici e per aver contribuito alla mia scelta di continuare a provarci. Grazie Fiorella per le lunghe chiacchierate virtuali, per credere nelle mie possibilità, per il tifo spassionato e per l’ottima cena che prima o poi ricambierò. Grazie ad Umberto Repetti e all’associazione di Meteoriti Italia per l’opportunità che mi avete dato con le conferenze a Pedavena e Agordo e per tifare sempre per i miei successi. Grazie anche agli amici che furono e che non sento da tanto tempo. Ogni tanto vi penso, spero di poter rincontrare molti di voi come se il tempo non fosse passato. In un modo o un altro anche voi avete contribuito a rendermi quella che sono.

Grazie a Agorà Scienze e agli studenti della 7a Scuola di Comunicazione della Scienza per avermi fatto riscoprire la gioia della Scienza. Grazie a voi per aver fatto nascere Agnesearth!

211 Grazie ad Agnesearth, un blog nato per gioco senza il quale non potrei più immaginarmi e che mi ha permesso di resistere e andare avanti in questo ultimo anno e mezzo. Grazie all’EMU School e alla MetSoc 2014 per avermi schiarito l’orizzonte, per avermi fatto capire che dovevo, potevo e soprattutto volevo ancora provarci.

Un grazie che forse molti riterranno stupido, ma per me non lo è affatto va a Internet e tutte le tecnologie e tutti gli strumenti ad esso correlati: MSN Messenger, Skype, Facebook, WhatsApp e Wordpress. Grazie a loro ho ritrovato e mantenuto amicizie, mi sono sentita meno sola al mio arrivo a Pisa e nelle serate e momenti più tristi e ho potuto condividere la mia felicità con tutti.

Non so che sarà di me nei prossimi mesi e quali avventure mi attenderanno… ma so che sono circondata da belle persone che tifano per me e io tifo per loro! Grazie a tutti di cuore.

P.S. “Righe difficili da scrivere”, ma una volta preso il via non sono più riuscita a fermarmi… Chiedo venia per la lunghezza (inaspettata) di questa sezione.

212