Deformation of dry and wet sandstone targets during hypervelocity impact experiments, as revealed from the MEMIN Program

Authors

  • Elmar BUHL,

    Corresponding author
    1. Helmholtz-Zentrum Potsdam, Deutsches GeoForschungsZentrum (GFZ), Telegrafenberg, D-14473 Potsdam, Germany
    2. Institut für Geowissenschaften, Albert-Ludwigs-Universität Freiburg (ALU), Albertstr. 23-B, D-79104 Freiburg, Germany
    Search for more papers by this author
  • Michael H. POELCHAU,

    1. Institut für Geowissenschaften, Albert-Ludwigs-Universität Freiburg (ALU), Albertstr. 23-B, D-79104 Freiburg, Germany
    Search for more papers by this author
  • Georg DRESEN,

    1. Helmholtz-Zentrum Potsdam, Deutsches GeoForschungsZentrum (GFZ), Telegrafenberg, D-14473 Potsdam, Germany
    Search for more papers by this author
  • Thomas KENKMANN

    1. Institut für Geowissenschaften, Albert-Ludwigs-Universität Freiburg (ALU), Albertstr. 23-B, D-79104 Freiburg, Germany
    Search for more papers by this author

*Corresponding author. E-mail: buhl@gfz-potsdam.de

Abstract

Abstract– Hypervelocity impact experiments on dry and water-saturated targets of fine-grained quartz sandstone, performed within the MEMIN project, have been investigated to determine the effects of porosity and pore space saturation on deformation mechanisms in the crater’s subsurface. A dry sandstone cube and a 90% water-saturated sandstone cube (Seeberger Sandstein, 20 cm side length, about 23% porosity) were impacted at the Fraunhofer EMI acceleration facilities by 2.5 mm diameter steel spheres at 4.8 and 5.3 km s−1, respectively. Microstructural postimpact analyses of the bisected craters revealed differences in the subsurface deformation for the dry and the wet target experiments. Enhanced grain comminution and compaction in the dry experiment and a wider extent of localized deformation in the saturated experiment suggest a direct influence of pore water on deformation mechanisms. We suggest that the pore water reduces the shock impedance mismatch between grains and pore space, and thus reduces the peak stresses at grain–grain contacts. This effect inhibits profound grain comminution and effective compaction, but allows for reduced shock wave attenuation and a more effective transport of energy into the target. The reduced shock wave attenuation is supposed to be responsible for the enhanced crater growth and the development of “near surface” fractures in the wet target.

Introduction

Impact cratering is a fundamental process throughout the solar system and on Earth (Shoemaker 1977). As most of the planetary surfaces are porous, a considerable number of impacts occur on porous target material such as sedimentary rocks on Earth (Earth Impact Database 2012), or regolith on planetary bodies (Hörz and Cintala 1997). Impacts on comets can be regarded as an endmember concerning target porosity (Housen et al. 1999; Housen and Holsapple 2003; Schultz et al. 2010). Understanding the influence of pore space on impact mechanisms is hence essential for the investigation of natural impact phenomena in the solar system (Kieffer 1971; Kenkmann 2003; Osinski et al. 2005; Folco et al. 2011; Key and Schultz 2011). The role of interstitial water or water ice in porous rocks, which is common in terrestrial and Martian target materials, is not yet well understood. On one hand, the presence of pore water may reduce the shock-related compaction of a rock; on the other hand, pore water may cause mechanical disruption of the target if steam is produced during shock wave release.

Several studies, including the investigation of natural impact craters (Kieffer 1971), impact experiments (Gault and Greeley 1978; Love et al. 1993; Housen and Holsapple 2003; Kenkmann et al. 2011), and numerical simulations (Kieffer and Simonds 1980; Wünnemann et al. 2006; Collins et al. 2011) have shown that porosity and pore space saturation have a major influence on crater formation. Nevertheless, these effects are currently not fully understood and still under debate (see discussion in Baldwin et al. 2007). Schäfer et al. (2006) and Baldwin et al. (2007) have shown that experimental studies performed on solid porous targets provide further insights into the effect of porosity on crater formation. Most of the cratering experiments in the past have either been focused on porous, but loose particulate materials like sand (Gault et al. 1968; Braslau 1970; Stöffler et al. 1975; Cintala et al. 1999; Anderson et al. 2004; Lohse et al. 2004; Hermalyn and Schultz 2011), glass beads (Yamamoto and Nakamura 1997; Barnouin-Jha et al. 2007), or powdered pumice and basalt (Hartmann 1985). Experiments performed on solid geo-materials were commonly using dense or low-porous rocks like basalt and gabbro (Gault and Heitowit 1963; Moore et al. 1963; Lange et al. 1984; Polanskey and Ahrens 1990), granite (Hörz 1969; Burchell and Whitehorn 2003), or limestone (Ahrens and Rubin 1993), or utilizing glass (Field 1971) or ice (Lange and Ahrens 1987). Only sparse literature is concerned with impact cratering in solid and porous materials like sandstone (Maurer and Rinehart 1960; Shoemaker et al. 1963; Baldwin et al. 2007; Kenkmann et al. 2011), sintered glass beads (Love et al. 1993; Michikami et al. 2007), gypsum (Onose et al. 2011), or snow (Arakawa and Yasui 2011). Furthermore, only parts of these experiments have been investigated with regard to subsurface processes (Maurer and Rinehart 1960; Polanskey and Ahrens 1990; Kenkmann et al. 2011; Onose et al. 2011), such as pore collapse, shock metamorphism, target damaging, shear localization, fracture propagation, and dilatancy.

To gain further insights into these processes and to overcome the lack of experimental data, MEMIN (Multidisciplinary Experimental and Modeling Impact research Network) was founded in 2009 by the DFG (German Research Foundation) as a delocalized, multidisciplinary research unit for experimental hypervelocity impact cratering (Schäfer et al. 2006; Kenkmann et al. 2011, 2013; Poelchau et al. 2013). The project focuses on impact cratering experiments into geological materials to understand details of the cratering process through in situ measurements (Hoerth et al. 2013), extensive postimpact analysis (Dufresne et al. 2013; Ebert et al. 2013; Sommer et al. 2013), and numerical modeling (Güldemeister et al. 2013). For this study, one unsaturated (“dry,” experiment number A6-5126) and one saturated (“wet,” experiment number A11-5181) sandstone block were impacted by steel spheres, using acceleration facilities of Fraunhofer Institute for High-Speed Dynamics, Ernst-Mach-Institut (EMI), in Freiburg, Germany. Here, we present a comprehensive multidisciplinary analysis of the crater floor and its subsurface in direct comparison with the pre-impact material properties.

Methods

Experimental Set-up

The two impact cratering experiments (dry target experiment A6-5126 and wet target experiment A11-5181) were conducted at EMI, using the so-called SLGG (“space” two-stage light-gas gun) for acceleration of the projectile. The SLGG with an 8.5 mm caliber launch tube is a two-stage light-gas gun, mainly used for satellite vulnerability tests (Schneider and Schäfer 2001). This facility has a horizontal launch tube of 1.5 m length and depending on the projectile size impact velocities up to 9 km s−1 are possible. The projectiles in the two experiments were spheres of D290-1 high-speed steel with a diameter of 2.5 mm and a mass of 0.0671 and 0.0670 g, respectively. These projectiles were accelerated to 4.8 and 5.3 km s−1 corresponding to kinetic impact energies of 773 and 941 J, respectively. The size of the projectiles was chosen to obtain a high ratio between projectile diameter and the grain size of the target. This is of great importance for the scaling between experimental results and observations within natural impact sites. Also, this set-up produces a shock pressure plateau with a length of several grain diameters. The atmospheric pressure in the target chamber was kept at 100 mbar. For in situ observations the impact was filmed by high-speed cameras (Hoerth et al. 2013) and the sandstone block was equipped with an ultrasound sensor system (Moser et al. 2013). The targets were sandstone cubes of 20 cm side length. Samples were large enough to prevent complete target failure.

Target Characterization

The target material used within the MEMIN experiments is Seeberger Sandstein (Layer 3), quarried in central Germany near Gotha, Thuringia, by the TRACO Company (Fig. 1). This well-sorted fluvial quartz sandstone has an uppermost Triassic age (Rhaetian approximately 200 Ma) and a mean grain size of 100 ± 25 μm. It was chosen as target material due to its quartz-dominated mineralogical composition. The sandstone comprises approximately 89 vol.% quartz; approximately 10 vol.% phyllosilicate; and traces of rutile, altered ilmenite, zircon, rare iron-oxides/hydroxides (Ebert et al. 2013). This well-sorted sandstone shows a centimeter-spaced layering and has an overall porosity of 23 ± 1 vol.%. Quartz grains are subrounded and cemented by quartz, phyllosilicates, and iron mineral coatings. Liesegang rings developed as reddish-brownish bands show a variation in iron oxidation throughout the sample. Uniaxial compression tests showed strength values of about 70 MPa for dry samples and about 57 MPa for wet samples (Poelchau et al. 2013). All blocks were cut from the same stratigraphic layer. We selected a domain within the stratigraphic sequence, which stands out by the highest homogeneity. Bedding planes are oriented ±5° to the surface of the cubes used in our experiments. Due to modest cross-bedding, typical for fluviatile sandstones, a more precise orientation is hindered.

Figure 1.

 Target sandstone. a) Backscatter electron (BSE) photomicrograph of a thin section of unshocked Seeberger Sandstein (layer 3). b) Modal composition of the thin section. Note the relatively homogeneous grain-size distribution of quartz grains.

The blocks were carefully oriented in the target chamber to obtain an impact perpendicular to the target surface, and in the center of the blocks. Temperature and moisture of the dry sample were controlled by storage under room conditions. The wet sample was saturated with tap water by applying a vacuum onto the free surface, while the other five sides of the cube were covered by water. For a high saturation, the atmospheric pressure was reduced down to 30 mbar, and the procedure of applying vacuum and refilling the water reservoir was repeated several times. To control the progress of water infiltration during this process, the wet sample block was weighed repeatedly. Therefore, the actual saturation S is given by

image(1)

where mtotal is the weighed mass of the saturated sandstone, mdry is the mass of the dry sample weighed before watering, ρw is the density of water, Vblock is the total volume of the sample block, and φsandstone is the total porosity of the sandstone. In the wet target experiment (A11), a maximum saturation of about 91% was achieved. Thus, 91% of the bulk pore volume was filled with water. The residual 9% of unsaturated pore space is assumed to be dead pore space that has no or insufficient connectivity to the water-saturated porosity. The actual distribution of water-filled and empty pores is not known. Some minor uncertainty about saturation during impact arises due to water spill during the time of the applied vacuum in the target chamber. As the atmospheric pressure of 100 mbar in the target chamber is considerably higher than the vacuum of 30 mbar used for saturation, we assume that the water spill in the target chamber is relatively small and thus the saturation at the time of impact is about 90%.

Preparation

Immediately after the experiment, the sandstone blocks were weighed to determine the excavated masses and thus the energy-related cratering efficiency (Poelchau et al. 2013). Afterward, the blocks and the craters were photographed. High-resolution 3-D point clouds were obtained by laser scanning techniques (eScan3DTM by Digital Corp). These point cloud data were used to model the crater cavity as 3-D grid-structures, which in turn were used for detailed morphological investigations and crater efficiency estimations (Dufresne et al. 2013). For subsurface investigations the crater cavities and the less stable crater floor of the experiments A6 and A11 were impregnated with low-viscosity epoxy. We applied a combination of vacuum and heat to achieve a deep penetration of the epoxy and to stabilize the crater subsurface. However, during cutting with a water-lubricated rock saw, some of the highly fragmented, weak material beneath the impregnated section was completely disintegrated and had the tendency to be washed out by the cooling water. Hence, a second phase of soaking with epoxy and stabilization became necessary for further preparation. Thereafter, the surfaces were polished and photo-documented for a macroscopic crack mapping. For closer inspection of each block, three standard thin sections were prepared of the zones directly underneath the crater floor. Part of the remaining block was used for a preliminary analysis by X-ray tomography.

Mapping

The photographs were utilized for determination of macroscopic damage structures and fractures. Systematic electron photomicrographs were taken with a Zeiss Leo 1525 field emission scanning electron microscope (FE-SEM, ALU Freiburg) in secondary electron (SE) and backscattered electron (BSE) mode (20 kV, 16 mm working distance). These conditions provide strong phase contrasts and are optimized for an easy detection of pore space. The Oxford Instruments software tool INCA was used for automated image recording. For detailed but area-wide microscopic analysis of the central region of the crater subsurface, high-resolution photomosaics were compiled from a set of about 300 BSE images of identical magnification (160×). These composite maps have the same width as the thin sections (approximately 2.8 cm). The maps were used to determine grain-scale damage in detail. Mapping was carried out with the geographic information system ArcGis (ESRI Inc.).

Porosity Determination

Determination of the pre-impact porosity and the postimpact porosity was carried out by applying quantitative image analysis. The free image analysis software tool ImageJ (http://rsbweb.nih.gov/ij/) was used to apply gray-scale intensity thresholds for pore space determination. Using the high contrast BSE electron micrographs enabled a clear discrimination between grains and epoxy-filled pore space.

We also estimated sample pore space using X-ray tomography. This nondestructive method allows a full 3-D acquisition of the porosity. For the dry experiment, a thick section was investigated by means of a Phoenix NanoTom (GE, General Electric Company) μ-CT X-ray scanner (GFZ, Potsdam). The maximum nominal resolution is given as <1 μm, but the best we could achieve for our specimen size is 13 μm per voxel.

Fracture Analysis

To analyze the ubiquitous fractures beneath the crater floor, numerous fractures were mapped in both experiments as polylines (objects of one or more line segments) in ArcGis. The fractures were mapped at a distance to the impact point source, to allow for distinction of individual fractures. The impact point source (Melosh 1989; after Birkhoff et al. 1948) is an assumed depth db below the target surface where the projectile is assumed to come to a halt. Its formula was found as the depth-of-burst for an explosion crater with the same energy and volume as the equivalent impact crater. This approximation is very simple, but seems appropriate for our analysis. A more detailed study of the depth-of-burst can be found in Holsapple (1980). Depth db is expressed as:

image(2)

where rp is the projectile radius, ρp is the projectile density, and ρt is the target density (Melosh 1989). This depth is the point of origin of the z-model (Maxwell 1977), which corresponds to the depth of burial in craters produced by explosives. Within the applied Cartesian coordinate system, the x,y-coordinates and the measured orientation (azimuth) of each polyline center were determined via the Field Calculator tool for ArcGis. Additionally, the theoretical orientation for a perfect radial inclination of each fracture was calculated by means of the obtained coordinates as:

image(3)

where α is the inclination angle and x, y are the coordinates of the fracture center. To estimate the difference between the true fracture orientation and the radial direction, angular deviations were determined (Fig. 2). Clockwise deviation angles are assumed negative, and complementary counterclockwise deviation angles are assumed positive. A detailed description of the procedure to determine concentric deviations is given by Poelchau and Kenkmann (2008).

Figure 2.

 Methodology for the determination of the angular deviation of concussion fractures. For each position of a fracture, a theoretical radial orientation toward the impact point source can be calculated (dashed lines). If this orientation differs from the measured orientation of the fracture, the angular deviation can be calculated.

Results

Impact on Dry Sandstone—Experiment A6-5126

Crater Geometry

The diameter of the experimentally produced crater cavity was measured every 10 degrees (18 measurements) resulting in a mean diameter of 5.76 cm with a standard deviation of 2.79 mm. The depth is 1.10 cm, and the depth/diameter ratio is 0.19 (Table 1). The 3-D laser scanning resulted in a volume of 7.6 cm3 for the excavated cavity. A detailed description of the excavation process and the deformation and emplacement of ejecta material is given in Sommer et al. (2013) and Hoerth et al. (2013). As radial symmetry could be expected for impacts perpendicular to the surface of a homogeneous material, the irregularity of the crater shape is caused by random fracture connectivity that led to ejection of spall plates at some places, while in other cases, spall remained incipient (Dufresne et al. 2013). Minute material inhomogeneities may also control this process (Fig. 3b). With Equation 2, the depth of the impact point source is calculated to be 4.97 mm beneath the target surface (Fig. 3a) using a projectile radius of 1.25 mm, a density of 8.1 g cm−3 for the projectile, and a target density of 2.05 g cm−3. The estimated depth db corresponds to roughly two projectile diameters.

Table 1. Overview of experimental parameters and resulting volumes.
 Exp. A6-5126
Dry sandstone
Exp. A11-5181
Wet sandstone
  1. 1Zone of tensile failure.

  2. 2Zone of pervasive grain crushing and compaction.

  3. 3Zone of localized deformation.

Grain size (μm)∼70–125∼70–125
Porosity (%)23 ± 123 ± 1
 Dry90%
Projectile mass (g)0.06710.0670
Velocity (km s−1)4.85.3
Kinetic energy (J)773941
Crater diameter (mm)57.6101.6
Crater depth (mm)11.014.3
Crater volume (mm3)760031100
Volume zone A1 (mm3)/(% crater vol.)107/1.4608/2
Volume zone B2 (mm3)/(% crater vol.)659/9−/−
Volume zone C3 (mm3)/(% crater vol.)1209/161960/6
Max. thickness zone A1 (mm)/(dp)0.9/0.361.9/0.76
Max. thickness zone B2 (mm)/(dp)3.4/1.36−/−
Max. thickness zone C3 (mm)/(dp)2.4/0.966.0/2.4
Figure 3.

 a) Map of 180 merged BSE electron micrographs along a cross-section through the crater center (shown in b). The area of the crater cavity is hatched and the calculated position of the impact point source is shown in combination with the initial projectile diameter. Four different zones were characterized and discriminated: yellow—zone of tensile failure, orange—zone of pervasive grain crushing and comminution, red—zone of localized deformation, magenta—zone of incipient spallation. Marked areas are magnified in (c–f). b) Image of the crater cavity at oblique illumination. c) Tensile fractures directly underneath and subparallel to the crater floor. d) Pervasive grain crushing of quartz grains is accompanied by kinked mica and a phyllosilicate aggregation. e) Shear-enhanced compaction in a steeply inclined shear zone. f) Spallation fractures occur with shallow inclination as tensile fractures beneath the crater rim. See text for further explanation.

Subsurface Structure

A macroscopic inspection revealed that a large-scale (macroscopic) fracture system was not developed beneath the crater floor except for a few spallation fractures beneath the outer crater that are supposed to result from an interaction of the compressive stress wave with the free surface (Melosh 1984). Different levels of damage correspond to the appearance of different deformation modes in the rock. In the more detailed BSE image analysis, we distinguished and mapped four distinct zones, characterized by different deformation microstructures. The resulting deformation zones are illustrated in Fig. 3a. Although the transitions between the zones are more gradual than depicted in Fig. 3a, a clear discrimination is possible. Three of the zones are located directly underneath the crater floor and form a concentric succession; the fourth zone is located underneath the crater rim.

Zone of Tensile Failure

This zone (yellow in Fig. 3a) is located directly underneath the central crater floor and has a maximum thickness of 0.9 mm (equal to 0.36 projectile diameters dp) and flattens out to both sides. The microstructure of the deformed sandstone is dominated by pervasive grain crushing and numerous irregular fractures. Grains are mostly disaggregated or crushed forming fine (<5 μm) fragments. Phyllosilicates are mobilized and surround fine-grained fragments and fill virtually all gaps. No shock metamorphic overprint of quartz grains has been recognized. This absence of PDFs and diaplectic glass indicates that shock pressures >10 GPa were not reached in the crater's subsurface. This fabric is cut by larger tensile fractures, which are subparallel to the crater surface (Fig. 3c). The fractures have mean length of 0.3 mm (0.14 dp) and a length-to-width ratio of 6.5, determined by ArcGis mapping tools. These tensile fractures show a maximum spacing of about 80 μm, and they developed by unloading due to the release wave. Hence, these features postdate compaction and grain crushing. Assuming a radial symmetry the calculated volume of this zone is 107 mm3 (about 0.014 crater volume).

Zone of Pervasive Grain Crushing and Compaction

This zone (orange in Fig. 3a) directly beneath the zone of tensile failure is characterized by pervasive grain crushing and by a nearly complete loss of porosity in terms of bulk pore volume and size of the remaining pores. As in the upper zone, the structure is dominated by numerous cracks in the few, still recognizable “original” grains, and a “matrix” build-up of fine debris. Indications of shock metamorphism could not be found. The difference to the zone of tensile failure presented above only concerns the lack of tensile cracks. The nearly complete loss of porosity is probably related to contact fragmentation along grain boundaries, rotation of single grains, and filling of the residual pore space with a mixture of very fine quartz fragments and phyllosilicates. The latter show aggregations with structures that resemble flow structures, documenting some mobilization during impact. In the 2-D section, this zone has a sickle-shaped geometry with an approximate maximum thickness of 3.4 mm (1.36 dp) and an overall lateral extent of 14 mm. This gives a calculated volume of 659 mm3 (Table 1; about 0.09 crater vol.). At most, 10% of the grains remained intact or show only minor damage.

Zone of Localized Deformation

The zone of localized deformation (red in Fig. 3a) is the outermost zone where strong deformation and comminution processes are still visible. The maximum thickness of the zone is 2.4 mm (0.96 dp). The calculated volume of this zone is 1209 mm3 (about 0.16 crater vol.). In contrast to the zone of pervasive grain crushing and compaction, fracturing of quartz grains in this zone is restricted to limited areas. These areas of localized and intense deformation are enclosed by mostly intact parts of the sandstone (Fig. 3e). The deformed zones contain fragmented grains and mobilized phyllosilicates. The overall porosity is slightly reduced due to alternation between compacted and uncompacted domains. In some of the deformed domains, shear-enhanced compaction seems to be the dominant comminution mechanism, whereas the development of the majority of the domains is not visibly related to shearing. The localized zones of deformation typically have a length-to-width ratio of 2.2 ± 0.8 (gray in Fig. 4). Their length is on average 0.39 mm (0.16 dp). There is no clear trend concerning the orientation of the long axis of the damaged zones with respect to the impact point source. In the 2-D section, one subset runs roughly concentric, while a second subset shows a 40–50° angle to the concentric orientation.

Figure 4.

 Detailed mapping of comminution domains (black/gray) within the zone of localized deformation for both the dry (right) and wet (left) target cratering experiments.

Zone of Incipient Spallation

Spallation is restricted to the outer parts of the crater cavity and is observed to a depth of about 1 cm beneath the original target surface (Fig. 5). In contrast to two thirds of the crater circumference where all spall plates were ejected during the excavation process, a 4.7 cm long and 2 cm wide spall plate is still attached at the lower right-hand side of the crater (Fig. 5a). To distinguish this behavior from the spallation of sandstone fragments, this phenomenon is referred to as incipient spallation (Dufresne et al. 2013). Consistent with observations in other cratering experiments (e.g., Moore et al. 1963; Hörz 1969; Evans et al. 1978), the fragment shapes are usually tabular. Three major fractures are visible in the subsurface and show apparent inclination angles between 14 and 17° with respect to the target surface (Fig. 3f). Their length varies between 12 and 22 mm (4.8–8.8 dp). A 3-D shape analysis of ejected spall plates is given in Sommer et al. (2013). In thin sections, the fractures look slightly different from their macroscopic appearance due to attrition during preparation. Spallation fissures are intergranular, mode I fractures in hardly damaged host rock. Their maximum aperture is about 120 μm. Disruption preferably occurred along grain-to-grain contacts. Spall plates are prevented from being ejected if the spall fractures did not fully propagate to the free upper surface. Spallation is suggested to be developed as tensional fractures, created by complex interaction of the shock wave, and the tensile rarefaction wave reflected from the free surface (Melosh 1984; Polanskey and Ahrens 1990). However, the shape of the fractures we found is generally concave toward the target surface, which differs from the convex shape of the near-surface zone described by Melosh (1984). This suggests that the mechanics behind the processes of spallation are more complex than described to date. More work on the spallation features and the mechanisms is planned within the MEMIN network.

Figure 5.

 Incipient spallation in the dry experiment (A6-5126). a) Plan view of the crater cavity. The spall sliver flake on the lower right-hand side is cross-hatched. b) Cross-section through the crater cavity along the dashed line in (a). Incipient spallation was only detected at the left crater rim. Thickness of the spallation flakes is about 0.3–0.4 mm.

Porosity

The porosity as determined from SEM micrographs varies with distance to the crater floor (Fig. 3). Please note that we use the term porosity here in a broad sense, which refers to all open voids and fractures within the target, regardless of their generation. The area with the highest density and thus the lowest porosity is the zone of pervasive grain crushing and compaction (Fig. 3d). Above this zone, the porosity is enhanced by the tensile fractures that run parallel to the crater floor (Fig. 3c). Below the zone of compaction, the porosity gradually approaches the porosity of the starting material (approximately 23%). As can be seen in Fig. 6 (gray curves), these values drop from around 40% directly at the crater floor to about 12% within a distance of about 2 mm. Beneath this point of maximum compaction, the porosity values slowly increase to values of the initial porosity at a distance of about 8 mm from the crater floor.

Figure 6.

 Postimpact porosity with increasing distance from the assumed impact point source (light gray A6-5126, dark gray A11-5181). Deviations between min and max values are caused by applying different gray-scale thresholds for the image analyses. We estimate the error in porosity values to be ±2%.

A preliminary X-ray tomography analysis of a thick section from the central part of this crater (exp. A6-5126) confirmed the above results. In contrast to image analysis performed on micrographs, X-ray tomography provides a full 3-D estimate of the pore space. The present resolution of 13 μm per voxel allows the larger pores to be resolved (Fig. 7). Despite this limited resolution, some qualitative observations can be made. Enhanced porosity, induced by tensional fractures, can clearly be seen at the crater floor. Compaction, and thus dissipation of large pores, is obvious in the zone beneath the tensional failure. On the right-hand side of the crater, spallation fractures can be seen, which are not visible in thin section.

Figure 7.

 X-ray 2-D cross-section through the dry crater (exp. A6-5126) shows first a decrease and then an increase in porosity with increasing distance to the impact point source, consistent with the image analysis methods.

Fracture Analysis

During microstrucural mapping, conspicuous intragranular fractures have been recognized throughout a larger area. The fractures show a geometry reminiscent of Hertzian fractures (Frank and Lawn 1967; Wilshaw 1971; DiGiovanni et al. 2007). Similar fractures were first described by Kieffer (1971) for Meteor Crater as concussion fractures. These fractures are a common phenomenon within the damaged part of our sandstone targets, as well as in natural impact sites. Concussion fractures are intragranular conchoidal fractures that emanate from grain-to-grain contacts of neighboring quartz grains. They often extend into both contacting grains branching from the grain-to-grain contact; in other cases, fractures can only be seen in one of the affected grains. Linear concussion fractures mostly occur in force chains of quartz grains (Fig. 8a), or the fractures occur in circular orientation around initial voids enclosed by quartz grains (Fig. 8b). These fractures are frequent within the localized zones of comminution, contributing to pore-space reduction. These fractures were mapped in the zone of localized deformation and its proximity (Fig. 3a). The analysis (Fig. 8d) reveals a preferred orientation of cracks in a radial direction from the impact point source.

Figure 8.

 a) Linear concussion fractures emanate from grain-to-grain contacts and spread through four quartz grains. These tightly bound grains act as a force chain along which impact energy is transported inside the target. b) Nearly circular concussion fractures around a crushed sedimentary pore space of the initial sandstone. Splinters and small particles of the fractured quartz grains are moved inward and partly fill the void. c) Detail of the concussion fractures mapped as poly-lines. d) Histogram for the measurements (Ndry = 1451, Nwet = 2421) of radial deviation.

Impact on Wet Sandstone—Experiment A11-5181

Crater Geometry

In the wet target experiment (A11-5181), water saturation of the pore space during the experiment was about 90%. The crater cavity is almost twice that of the dry target experiment. It has a geometry dominated by a central cavity surrounded by a flat terrace (Fig. 9a) which, in turn, is encircled by a steep slope that forms the crater rim. The crater diameter is about 10.2 cm with a standard deviation of 6 mm for the 18 taken measurements; the maximum depth is 1.4 cm and the depth/diameter ratio is 0.14. The total crater volume of 31.1 cm3, calculated from a 3-D grid model, is about four times the crater volume produced in the dry experiment (Table 1). The kinetic energy of the projectile, however, differed only by about 22%. This behavior substantiates the strong influence of water saturation on the cratering efficiency. The observed volume increase is mainly due to the enhanced lateral growth of the crater (Δradius = 22 mm), while the depth of the craters (Δdepth = 3 mm) produced in wet and dry tests is very similar.

Figure 9.

 Cross-section and details of the wet target experiment (A11-5181). a) Sketch of cross-section with microscopic maps. Positions of three different deformation zones are depicted in yellow (zone of tensile failure), red (zone of localized deformation), and purple (zone of incipient spallation). The hatched area corresponds to the crater cavity. b) Plan view on the crater cavity. The cutting plane of (a) is marked as a dashed line. c) SEM-BSE micrograph of open tensile fractures. d) SEM-BSE micrograph of localized deformation domain. e) SEM-BSE Micrograph of steep spallation fracture near the crater rim.

According to the equation (Equation 2) for the approximation of depth of burial (Melosh 1989), the impact velocity has no influence on penetration depth. For practical reasons, we assume this to be valid, and use this model for our calculation of the impact point source. Thus, the input parameters are unchanged except for the target density that, by water infiltration, is increased to 2.26 g cm−3. That, in turn, yields a slightly shallower calculated depth of the impact point source at 4.73 mm.

Subsurface Structure

At the macroscopic scale, spallation fractures can be traced beneath the steeper inclined slopes of the crater (Fig. 10b). In addition, small fractures were detected beneath the central crater floor that might be related to a conical fracture system. The microscopic analysis of the wet target experiment revealed distinct differences in comparison with the dry experiment (A6-5126). We have identified three different zones of characteristic deformation (Fig. 9a). The main structural difference between the dry and the wet target experiments is that the zone of pervasive grain crushing and compaction is missing below the crater of the wet block. However, damage manifested in the zones of tensile failure and localized deformation shows significantly larger extent. Besides the higher impact energy, the pore-space filling is the only reasonable difference in the experimental set-up that can be responsible for this change in the resulting deformation. Thus, deformation mechanisms will be strongly influenced by the presence of water in the pore space (“pore space saturation”).

Figure 10.

 Areas of incipient spallation from the wet target experiment (A11-5181). a) The larger fragment reaching into the crater from the lower left-hand side is the most prominent part of incipient spallation in this example. As localization of incipient spallation from the surface is complex, only the flakes assured from cross-section (dashed line) are depicted. b) Cross- section along dashed line in (a). Smaller flakes of incipient spallation can be seen on both sides at the crater rim. Note the bifurcation of both fractures. Also, downward bending fractures (conical fractures [blue]) are present at the crater rim.

Zone of Tensile Failure

This zone (yellow in Fig. 9a) is again situated directly beneath the central crater floor. Its matrix, consisting of a mixture of crushed quartz fragments and intact quartz grains, is pervaded by larger fractures subparallel to the crater floor. Although the sandstone is pervasively disintegrated and nearly all grain-to-grain contacts are broken, more unaffected and partly intact quartz grains occur in comparison with the dry experiment. The sandstone of the wet target experiment (A11-5181) shows otherwise a less severe damage than its dry counterpart (A6-5126). Again, no shock metamorphism of quartz grains was detected. The zone of tensile failure is about 23 mm (9.2 dp) wide and <2 mm (0.76 dp) thin, with a calculated volume of 609 mm3 (about 0.02 crater vol.). The largest open tensile fractures are distributed evenly over the whole area and are up to 2 mm long and 200 μm wide. The fractures have a mean length of about 0.6 mm (0.23 dp). They are thus larger than in the dry experiment and show a smaller length-to-width ratio of 5.7.

Zone of Localized Deformation

This zone (red in Fig. 9a) is the largest of the traceable zones. It is very similar to the corresponding zone in the dry experiment (A6-5126) and has a similar sickle-like shape. Its lateral extent is approximately 26 mm (10.4 dp) and its maximum thickness is 6 mm (2.4 dp). The calculated volume is 1960 mm3 (about 0.06 crater vol., Table 1). A minor deformed host rock engulfs distinct, strongly deformed domains with grain crushing and pore collapse (Fig. 9d). Just as in the dry case some of these domains show compaction and a shear displacement component, determined by kinked micas. The rest of the domains show geometries that do not suggest shear displacement while compaction and porosity reduction is observed. The localized zones of deformation typically have a length-to-width ratio of 2.3 ± 0.9, determined by ArcGis mapping tools. Their length is, on average, 1.3 mm (about 0.5 dp). This means, despite the enhanced length of the deformed domains in the wet target experiment (A11-5181), the aspect ratio of these areas is consistent with the results of the dry experiment (A6-5126) (Fig. 11).

Figure 11.

 Length-to-width ratio of the deformed domains of the dry target (gray triangles) and the wet target (black squares) experiment plotted against their length. The presence of interstitial water seems to enhance the mean size of the domains, while the length-to-width ratio appears to be unaffected.

Zone of Incipient Spallation

Incipient spallation fractures occur on both sides of the crater rim. Assuming radial symmetry for the cratering process, a larger volume of incipient spallation can be expected on the lower left-hand side of the crater where a larger fragment was not ejected (Fig. 10a). The geometries of the fractures on both sides are characterized by a strong upward kink near their center (Fig. 10b). This kink is related to a bifurcation of the fracture, where one branch is extending toward the free surface and the other branch is slightly bent downward, propagating into the host rock (Fig. 10b). The latter one can be related to a fracture population referred to as conical— or “near surface”— fractures.

Porosity

The porosity beneath the impact point source is strongly increased to about 57% directly beneath the crater floor due to the presence of tensile fractures parallel to the crater floor. This value drops within 3 mm distance to a porosity of about 20% (black lines in Fig. 6). With increasing distance to the crater floor the porosity varies between 17% and 22%, close to the porosity determined for the unshocked sandstone. A zone of massive compaction is absent in the wet target experiment.

Fracture Analysis

In thin sections, the same numerous grain-to-grain contact fractures were found as in the dry experiment (A6). To compare fracture orientation between both experiments about 1800 concussion fractures were mapped and analyzed by the described method yielding results in good agreement with that of the dry experiment (Fig. 8d). Preferred orientation of cracks in a radial direction strongly dominates also in the wet target experiment

Discussion

The presence of pore water in impact experiments on sandstone has been shown to have a major influence on the crater topography (Schäfer et al. 2006; Baldwin et al. 2007; Kenkmann et al. 2011). Our experiments have shown a decrease in the depth-to-diameter ratios from 0.19 for the dry case (Exp A6-5126) to 0.14 in the wet case (Exp. A11-5181). This result is in good agreement with the larger-scale experiments of the MEMIN pilot study (Kenkmann et al. 2011) where ratios of 0.23 and 0.16 have been found for a dry target and wet target, respectively. The values are also in the range published by Baldwin et al. (2007), although the variance in that study is higher, probably due to a smaller projectile/grain-size ratio.

This study demonstrates that the different topographies of craters formed in wet and dry targets (Dufresne et al. 2013) result from major effects of pore water on the damage behavior of the target rocks. The study of the deformation microstructures in the crater subsurface did not reveal a single shocked quartz grain in the target subsurface of both the wet and dry target craters, in contrast to the ejecta (Sommer et al. 2013).

Laboratory investigations of target damage by hypervelocity impacts have been carried out for dense solid rocks (Hörz 1969; Polanskey and Ahrens 1990) and for porous sandstone (Kenkmann et al. 2011). In agreement with Kenkmann et al. (2011), zones of increasing deformation could be defined with increasing proximity to the point of impact. The most severe target deformation in both experiments is found directly at the central crater floor.

Deformation Microstructures Related to Shock Compression

The development of the deformation microstructures with increasing strain is illustrated schematically in Fig. 12. The onset of deformation is characterized by grain boundary breakup along the relatively weak phyllosilicate cementation. First fractures occur along force chains of quartz grains aligned perpendicular to the shock front (maximum principal stress) (compare Figs. 8a and 12b). These fractures are commonly referred to as Hertzian or concussion fractures. These fractures are observed during quasi-static formation of deformation and compaction bands (Frank and Lawn 1967; Wilshaw 1971; Wong and Baud 1999; DiGiovanni et al. 2007) as well as in impact structures (Kieffer 1971).

Figure 12.

 Sketch of damage evolution with increasing strain. The initial composite of quartz grains and phyllosilicates is increasingly fractured and comminuted. Note the damage localization (c) before total failure (d). Compare with Fig. 8. Interestingly, the microstructural features such as Hertzian cracks, extensive comminution, and effective closure of initial pore space by grain fragments are very similar to experimentally induced compaction bands.

The preferred orientation of concussion fractures is radial (Fig. 8d). Generally, radial fracture populations have been reported for numerous impact experiments (Moore et al. 1963; Evans et al. 1978; Lange and Ahrens 1987; Polanskey and Ahrens 1990). Radial fractures are usually subdivided into a spoke-like and a conical-shaped population (e.g., Evans et al. 1978; Ahrens and Rubin 1993). A correlation of the reported concussion fractures with one of these configurations is not yet possible due to the location of the thin sections directly underneath the impact point and the lack of 3-D fracture information. The general mode of formation of radial fractures in impact is supposed to occur when high compressive stress (σ1) is in radial orientation and the minimum stress component (σ3) is oriented concentrically (circumferential stress) (see fig. 8 in Ai and Ahrens 2004). This orientation of the principal stresses is assumed to occur when the compressive wave front reaches the grain (Shibuya and Nakahara 1968; Field 1971; Ahrens and Rubin 1993). These fractures are then developed as tensional mode I fractures during shock loading.

With increasing strain, grains are crushed and the initial pore space collapses. New fractures occur along the contacts and bands of strong deformation form along the initial force chain (Fig. 12c). At even higher strain, these bands gain width and finally coalesce to a complete zone of pervasive grain crushing and compaction (Figs. 3d and 12d). This development is in good agreement with the formation of compaction bands (Menéndez et al. 1996; Stanchits et al. 2009). In contrast to shear localization, the formation of localized compaction bands requires enhanced effective mean stress (Issen and Rudnicki 2001; Olsson et al. 2002; Aydin et al. 2006; Fossen et al. 2007).

Pervasive and Localized Deformation

Interestingly, the degree of disruption and pore space collapse in the proximity to the crater floor is much higher in the dry target experiment than in the wet target experiment. The zone of pervasive grain crushing and comminution described for the dry target experiment (Fig. 3d) has not been recognized in the wet target experiment. It is conceivable that pore water effectively prevented intensive pore collapse and grain comminution. However, it is also possible that the crushed debris was completely removed during the subsequent excavation of the crater. Grain-size distribution of ejected material (Sommer et al. 2013) shows a higher fraction of very fine-grained fragments for the wet target experiments, thus supporting the latter assumption. This would imply that grain comminution was equal or even enhanced in comparison with the dry experiment, but all of the fine fragmented debris was excavated and thus cannot be seen in the subsurface. A structural cause for a more effective ejection of comminuted material in the wet target experiment is most likely the volumetric expansion of interstitial water upon pressure release. If peak pressures are ≤5 GPa, water remains a liquid upon pressure release (Riney et al. 1970); at pressures in excess of approximately 10 GPa, partial vaporization occurs, producing a liquid–vapor mixture (Kieffer and Simonds 1980), and at pressures ≥70 GPa, water is completely vaporized on pressure release (Butkovich 1971). Based on numerical modeling (Güldemeister, personal communication), a volume of 950 mm3 has experienced a shock pressure in excess of 5 GPa in the wet target experiment. Assuming equilibration of the shock pressure between quartz grains and the interstitial pore water by multiple reverberations, at least the central fraction of the water should vaporize upon pressure release under these minimum conditions. Steam production leads to volumetric expansion up to a factor 1700. This expansion possibly intensifies comminution and leads to the enhanced ejection of material. Additionally, the over-pressurized pore water could enhance the breakup of grain-to-grain contacts and thereby lead to enhanced tensile failure.

In contrast, results from shock-recovery experiments suggest that the presence of pore water actually inhibits the collapse of pore space and reduces the degree of fragmentation, at least in a confined experimental set-up (Hiltl et al. 1999). Hiltl et al. (1999) explain this observation with the ability of water to transfer and distribute compressive stress. The presence of water is thus reducing the contact forces. A further indication for a priori reduced fragmentation in the wet target experiment are yet uncalibrated shock pressure measurements 20 cm behind the target surface in larger scale experiments (Hoerth, personal communication). For the wet target experiments, higher pressure pulses were measured with respect to dry target experiments indicating reduced shock wave attenuation due to a reduced energy dissipation by pore space collapse and grain breakage. Likewise, the distinctly higher cratering efficiency in the wet target experiment with respect to the dry target experiment suggests that the shock wave attenuation was less in the wet target.

The zone of localized deformation reveals abundant localized deformation bands, which are comparable to the compactional deformation bands described by Okubo and Schultz (2007) for the Upheaval Dome impact structure that was formed in a porous sandstone target (Kenkmann et al. 2005). This zone constitutes a significantly larger volume in the wet target experiment compared with the dry target (Fig. 4). This also strongly suggests that shock wave attenuation was less effective in the wet target. Indeed, more and often larger crushed domains can be seen closer to the impact point source, indicating enhanced growth at higher strain. In advance, a principal reduction of yield strength for wet sandstones is reported for quasistatic compaction tests (Zang et al. 1996; Zhu and Wong 1997; Stanchits et al. 2009). In our case, the uniaxial compressive strength (UCS) of the sandstone was reduced from about 70 MPa for dry samples to about 57 MPa for wet samples. There is good reason to believe that this is also valid for dynamic fragmentation (Lomov et al. 2001), possibly leading to the enhanced comminution.

At greater distance (4.9 dp for the dry, and 7.2 dp for the wet target experiment) to the impact point source, comminution is absent and concussion fractures in addition to grain boundary breakup are the dominant deformation mechanisms in this outermost zone of impact-induced alteration.

The presumably most far-reaching damage mechanism is grain boundary breakup, caused by the relative low tensile strength of the phyllosilicate cement. Even though this behavior is not detectable in macroscopic view and hardly recognizable in microscopic analysis, the breakup of grain boundaries may play an important role for the determination of target damage evolution by P-wave analysis. A velocity reduction was described in detail by Ahrens and Rubin (1993) and Xia and Ahrens (2001) for experimental impact craters. Further descriptions of reduced wave velocities are given for natural impact sites as Meteor Crater in Arizona (Ackermann et al. 1975) or the Ries crater in southern Germany (Pohl et al. 1977). Moser et al. (2013) provide preliminary results of ultrasound velocity reduction in the subsurface of the MEMIN experiments. The extent of reduced P-wave velocities roughly correlates with the zone of localized deformation.

On the basis of numerical simulation of MEMIN experiments, Kenkmann et al. (2011) display the distribution of total plastic strain in the subsurface of the crater. The mapped zones of the different degrees of deformation roughly follow the isobars of the numerical models. In contrast to an ideal hemispherical propagation of the shock wave (e.g., see Melosh 1989; fig. 5.2), the zones of shock deformation (e.g., zone of localized deformation) are not spherical. The elongation of the isobars along the impact axis is in good agreement with observations in highly porous targets (Colombo et al. 2003; Onose et al. 2011) where also U-shape forms for compaction zones can be seen.

Deformation Microstructures Related to Pressure Release

The zone of tensile fracturing beneath the central crater can clearly be attributed to dilatancy upon pressure release. Likewise, fractures that can be seen at the crater rims in both experiments are a consequence of spallation. Interestingly, their geometries are different for the dry and the wet target experiments. In general, the spall fractures emerge from the crater walls, propagate outward, and in all cases, at least one branch reaches the target surface. In contrast to the dry experiment, the spallation fractures in the wet target experiment show strong kinks. These kinks seem to be related to the bifurcation of the fractures. The secondary branches propagate with a shallow inclination into the target block. The geometry of the latter strongly resembles the fractures described by Evans et al. (1978), Polanskey and Ahrens (1990) and Ai and Ahrens (2004) as “near surface” or conical fractures in nonporous materials (Fig. 13). These fractures are supposed to develop along the near-surface zone, as described by Melosh (1989). Deviations from this simple geometry are possibly related to the nonspherical geometry of the shock deformation zones. We suggest that these fractures propagate outward while the shock wave interferes with the release wave that is reflected obliquely from the target surface. Due to a more effective transport of the shock wave energy into the target and maybe enhanced pore pressures in the wet target case, conical fractures could propagate to greater distance to the impact and are thus observable beyond the zone of spall excavation. This interpretation is in good agreement with the fracture patterns of higher energy experiments in sandstone (Kenkmann et al. 2011).

Figure 13.

 Schematic unscaled comparison of impact damage in wet sandstone (this work; Kenkmann et al. 2011) and basalt (Polanskey and Ahrens 1990). Despite differences in scale and target rock, similarities in deformation patterns can be seen. Spallation fractures and conical fractures are developed in the same way. The shear fracture zone, described by Polanskey and Ahrens (1990), may correspond to the damage zones found in the sandstone. In both types of experiments, radial fractures are present, but on a different scale. We suggest that both types of radial fractures are developed by the same mechanism of a maximum compressive stress (σ1) applied radially and a minimal compressive strength (σ3) applied concentrically. Concentric fractures have not been recognized in the sandstone experiment.

Conclusions

The aim of this study is a better understanding of deformation mechanisms in porous solid rock during hypervelocity impacts. To determine the effect of pore-space saturation on the deformation mechanisms, a “dry” sandstone block and a “wet” sandstone block with 20 cm side length were impacted with 2.5 mm steel spheres (4.8 and 5.3 km s−1, respectively), and subsequently comparatively studied by microstructural analysis methods. Different modes of brittle deformation have been recognized and mapped for both cases. Differences between the experiments are in good agreement with planar shock-recovery experiments on dry and saturated sandstone (Hiltl et al. 1999). The presence of water inhibits pore-space collapse and comminution, at least in the crater's subsurface, and thereby causes decreased shock wave attenuation. This phenomenon results in the absence of a zone of complete compaction and a greater extent of the zone of localized deformation in the wet target experiments. The development of concussion fractures is independent of the presence of water, as they are ubiquitous in both wet and dry target rocks. Their radial orientation is consistent with their supposed development as tensional Mode I fractures along grain-to-grain contacts and force chains.

Acknowledgments— We appreciate funding by the German Research Foundation (DFG), grant KE732/15-1 and DR213/15-1. These projects are part of the DFG research unit FOR-887: “Experimental Impact Cratering: The MEMIN program.” We are grateful to Tobias Hoerth and the technicians of the EMI for conducting the experiments, Herbert Ickler for preparing thin sections, and the entire MEMIN team for stimulating discussions. We greatly appreciate the very thoughtful reviews of F. Hörz and W. B. McKinnon that greatly helped to improve our manuscript. We thank Dr. A. Deutsch for the editorial handling.

Editorial Handling— Dr. Alexander Deutsch

Ancillary