Hypervelocity impacts on dry and wet sandstone: Observations of ejecta dynamics and crater growth


Corresponding author. E-mail: tobias.hoerth@emi.fraunhofer.de


Abstract– This study deals with the investigation of highly dynamic processes associated with hypervelocity impacts on porous sandstone. For the impact experiments, two light-gas accelerators with different calibers were used, capable of accelerating steel projectiles with diameters ranging from 2.5 to 12 mm to several kilometers per second. The projectiles impacted on dry and water-saturated Seeberger Sandstone targets. The study includes investigations of the influence of pore water on the shape of the ejecta cloud as well as transient crater growth. The results show a significant influence of pore water on ejecta behavior. Steeper ejecta cone angles are observed if the impacts are conducted on wet sandstones. The transient crater grows at a faster rate and reaches a larger diameter if the target is water saturated. In our experiments, target porosity leads to smaller crater sizes compared with nonporous targets. Water within the pore space reduces porosity and counteracts this process. Power law fits were applied to the crater growth curves. The results show an increase in the scaling exponent μ with increasing pore space saturation.


Hypervelocity impacts are a fundamental process in our solar system. A possible avenue to better understand this extremely complex and rapid geologic process is to conduct impact experiments in the laboratory, which are of particular importance for unraveling the ejection process. The ejection process of impact cratering depends on a great number of factors including: (1) the target properties like density, strength, friction coefficient and porosity; (2) the presence of target volatiles; (3) the presence and properties of an atmosphere; (4) the impact conditions (impact velocity, angle); and (5) gravity. Laboratory experiments to investigate ejecta behavior were mainly conducted with loose granular target materials. Cintala et al. (1999) investigated the influence of impact velocity on the functional relationship between launch position and speed or angle of ejection, respectively, for impacts on coarse-grained sand. They found a power law decrease of the ejection speeds as proposed by Housen et al. (1983) as well as a decline of the ejection angle with increasing radial distance from the impact point. Furthermore, an increase of the ejection angle nearby the crater rim could be observed in this study. Yamamoto et al. (2005) measured the ejecta velocity distribution for impacts on regolith targets with variable impact angle. Further investigations of ejecta behavior using a PTV (particle tracking velocimetry) technique were carried out in which individual ejecta particles were tracked to determine ejection speeds as well as ejection angles as a function of time (Hermalyn and Schultz 2010, 2011). Furthermore, 3-D particle image velocimetry (3D PIV) was used to measure ejecta particle positions and velocities within the ejecta curtain (Anderson et al. 2003, 2004).

Compared with granular material, experimental data from hypervelocity impacts into consolidated rocks are sparse. Amongst others, gabbro (e.g., Lange et al. 1984; Polanskey and Ahrens 1990), granite (e.g., Hörz 1969; Burchell and Whitehorn 2003) and basalt (Gault et al. 1963; Gault 1973) as well as a basalt grout (Housen 1992) were used as target materials. Impact experiments on porous sintered targets were conducted by Michikami et al. (2007). Hypervelocity impacts on porous solid rocks like sandstone have been carried out by Shoemaker et al. (1963) and the influence of pore water on the cratering process was studied by Schäfer et al. (2006), Baldwin et al. (2007), and Kenkmann et al. (2011). It was found that water in the pore space significantly influences the crater morphology as well as ejecta velocities.

The dynamic growth of the crater cavity associated with the excavation flow field is of specific interest. The so-called quarter-space technique is a promising tool to gain direct insights into the formation of the growing cavity (Piekutowski 1980; Schmidt and Housen 1987; Schultz et al. 2005). This technique, however, possibly influences the crater formation process due to shock reflections or sliding friction at the Plexiglass window (Barnouin-Jha et al. 2007). Yamamoto et al. (2006) investigated the crater formation process in vertical impacts on soda-lime glass spheres by observing the crater formation from the top using a high-speed camera. Formation of the transient crater, i.e., the crater which forms during the excavation process, as well as ejecta curtain angles, was studied by Barnouin-Jha et al. (2007). These authors used a loose granular target and a laser sheet technique to document material transport in the velocity range of up to several hundred meters per second.

The current study was conducted in the framework of the research group MEMIN (“Multidisciplinary Experimental and Modeling Impact Research Network”; Kenkmann et al. 2011; Poelchau et al. 2013), which was established to investigate hypervelocity impacts into solid geologic materials as well as the associated ejection process with experimental and numerical methods. Here, we report on data of the shape and evolution of the ejecta cloud, the crater sizes, as well as the transient crater growth for strength-dominated crater formation as a function of water saturation and impact energy.


The impact experiments were conducted at two different light-gas accelerators at the Ernst-Mach-Institut in Freiburg and Efringen-Kirchen, Germany. The smaller facility in Freiburg uses a launch tube with a caliber of about 8.5 mm, while the much larger facility in Efringen-Kirchen uses a 38.7 mm caliber launch tube (Lexow et al. 2013). There are several reasons for using two different accelerators: The smaller gun allows a more detailed study of the impact and the ejection process because high-speed cameras and high-power flashes can be set up closer to the target yielding better illuminated high-speed images and a better resolution. The larger facility, on the other hand, allows acceleration of much larger projectiles and thus studying impacts at significantly higher impact energies. The impact experiments conducted in the framework of MEMIN cover a large range of impact energies, which is of particular importance for the investigation of scaling.

We used a sandstone named Seeberger Sandstein from Thuringia, Germany, as target material. The material is fine grained with an average grain size of about 100 μm. The sandstone has a bulk density of about 2.1 g cm−3 and a porosity of about 23%. The rock has a high purity (high amount of SiO2) and shows a stratification (Kenkmann et al. 2011; Poelchau et al. 2013). For the experiments at the small and the large accelerator, we used target blocks with edge lengths of 20 × 20 × 20 cm and 80 × 80 × 50 cm, respectively.

The experimental set-up is shown in Fig. 1. A high-speed camera with frame rates of 50, 54, 75, and 100 kfps (kiloframes per second) was used to record the impact and the ejection process. The camera was placed perpendicular to the trajectory of the projectile. Incident and transmitted high-power flashes were used for illuminating the ejection process. For the small-scale experiments, we used an exposure time of 1 μs, whereas for the large-scale experiments, an exposure time of 3.4 μs was used because in those experiments, it was much more difficult to illuminate the target chamber. Because of the very short exposure times, a strong light source is needed during the recording process. The small-scale experiments were set up with one 500 J flash (incident light, duration about 2 ms referred to 50% of light output) combined with one 250 J flash (transmitted light, duration about 12 ms referred to 50% of light output). In addition, incident light (500 J, 2 ms) from the top of the target chamber was used. The large-scale experiments were carried out with two 6000 J flashes (incident and transmitted light, respectively) capable of illuminating the ejection process for more than 20 ms. Because the high-power flashes feature rise times of several tens of microseconds, they were triggered shortly before the impact to make sure that the first stages of the impact process are recorded with the largest available amount of light.

Figure 1.

 Experimental set-up in the target chamber of the small accelerator in top view. In all experiments, projectiles were launched horizontally and impacted on an upright surface of the target. The projectile enters the bunker from the blast tank via an aperture. Three triggered flashes illuminate the ejection process from the sides and from the top. The high-speed framing camera films the ejection process in a plane perpendicular to the target surface. Specially devised witness plates consisting of Vaseline and phenolic foam, referred to as ejecta catchers, were placed opposite the target surface.

The experimental parameters are compiled in Table 1. Most of the impact experiments were conducted with spherical steel projectiles (high alloy steel D290-1), while experiment E3-3384 at the large accelerator was conducted with a sphere made of the Campo del Cielo iron meteorite (Ebert et al. 2013). Ambient pressure in the target chamber was set to about 100 mbar (small accelerator), and to about 300 mbar for the experiments at the large facility. A further reduction of the gas pressure in the experiments at the large accelerator is prevented since separation of the sabot from the projectile requires a certain atmospheric pressure in the blast tank and target chamber and blast tank have an interconnected gas medium.

Table 1. Parameters of the conducted experiments.
Target and Exp.#Projectile materialProjectilediameter (mm)Projectile mass (g)TargetmaterialWatersaturation (%)Projectile
velocity (km s−1)
Pressure in target chamber (mbar)
A5-5125Steel2.50.0672Seeberger SandstoneDry5.1100
A6-5126Steel2.50.0671Seeberger SandstoneDry4.8100
A11-5181Steel2.50.0670Seeberger Sandstone∼905.3100
A13-5182Steel2.50.0667Seeberger Sandstone∼505.2100
A12-5183Steel2.50.0676Seeberger Sandstone∼905.360
E1-3382Steel127.31Seeberger SandstoneDry4.6280
E2-3383Steel127.32Seeberger Sandstone∼504.6310
E3-3384Campo del Cielo meteorite127.09Seeberger Sandstone∼504.6292


High-speed images were evaluated to obtain the ejecta curtain angles and the transient crater growth as a function of time. Before each high-speed recording, a scale bar was installed directly in front of the target and filmed to convert pixels to millimeters. The angle φ between the ejecta cone and target surface (Fig. 2a) was measured manually using an image editing software. We estimate a measurement error of ±3°. Note that φ is not equivalent to the ejection angle of individual particles within the ejecta cone. The growth of the transient crater could not be determined directly because it was not possible to get direct insight into the crater. Using image editing software, we measured the width of the neck of the ejecta cone near the target surface as an approximation for the diameter of the crater cavity with time (Fig. 2b). This approach is based on the assumption that the narrowest part corresponds to the width of the transient crater. We estimate a measurement error of ±3 pixels, which corresponds to about 1–2 mm for the small-scale experiments and to about 4 mm for the large-scale experiments.

Figure 2.

 Methodology for the determination of various cratering characteristics. a) The ejecta cone angle is given by a tangential fit to the straight ejecta cone. b) The transient crater diameter is obtained by measuring the diameter of the neck of the ejecta cone. For explanation, see text.


Impact Craters

In Fig. 3, two craters from impacts on dry and water-saturated sandstones are shown (Exp. A5-5125 and Exp. A12-5183). The main features of the “dry” craters (Fig. 3a) are (1) a white central depression consisting of highly crushed target material, (2) an outer spallation zone, and (3) areas of incipient spallation where spall plates are partially detached from the target material, but are still fixed. The sizes and shapes of the impact craters strongly depend on the water saturation of the target material. Craters formed in water-saturated sandstone targets (Fig. 3b) are larger in diameter and volume. In addition, they show a different spallation behavior. The morphology of the spallation zone is “terrace-shaped.” A detailed description of the crater sizes and morphologies is provided by Dufresne et al. (2013).

Figure 3.

 Impact craters for impact on a) dry (A5-5125) and b) wet (A12-5183) sandstone target. Note the difference in scale bars.

Ejection Behavior and Shape of the Ejecta Curtain

A typical evolution of the ejection process is depicted in Fig. 4 where twelve high-speed video frames at different time steps are shown for an impact on dry sandstone (Exp. A6-5126, see Table 1). In the first frame, taken about 38 μs after the impact, a cone with high-speed ejecta has formed. In the central part, an orange-colored ejecta plume is observed, which is also cone-shaped. Although we did not measure any spectral wavelengths of this central plume, the color indicates that the material is luminescent and radiates heat. The second and third frame, about 158 and 278 μs after the impact, show the fully developed ejecta cone. The ejecta plume is no longer observable in these time frames. The lower part of the ejecta cone becomes increasingly steeply inclined with respect to the target surface, and finally develops into a “tube”-like shape. This tube keeps its shape for a comparatively long time (several milliseconds). At later time steps, the ejecta tube has increased in length and has developed a “neck.” A ring vortex has caused turbulences in the upper part of the ejecta cone (938–1648 μs). Spallation fragments leaving the target surface at comparatively low velocities appear at later time steps. These plate-shaped fragments are ejected if target material fails when its tensile strength is exceeded by rarefaction waves, which are reflected from free surfaces (Melosh 1984; Dufresne et al. 2013).

Figure 4.

 Typical evolution of the ejecta at different time steps after the impact (Exp. A6-5126). The first image shows a presumably hot central plume and the initial ejecta cone. The cone rapidly expands and steepens. Subsequent stages show that the upper part of the cone is increasingly affected by turbulences. Simultaneously, the cone transforms to a tube in which more and more large-sized spall plates (red arrow points at a spall plate) become entrained.

Table 2. Experimental results.
Target and Exp.#Water
saturation (%)
Ejecta cone angles (°) ±3° R/a
A5-5125Dry    5615

Figure 5 compares the ejecta evolution of two experiments that were conducted with roughly the same impact energy (Table 1), but with different target water saturation at the large accelerator with an ambient pressure of 300 mbar in the target chamber. The upper half shows ejecta from the impact on the dry sandstone (Exp. E1-3382), the lower one from the impact on sandstone with 50% water saturation (Exp. E3-3384). The ejecta cone for the wet sandstone is straight and displays a larger angle to the target surface. This behavior is reflected in the imprints on the ejecta catcher (i.e., specially designed witness plates placed opposite the target surface), which have a distinct outer limit for the wet target experiments, but a more blurry imprint for the dry target experiments (Kenkmann et al. 2011; Sommer et al. 2013). Both the cone and the presumably hot ejecta plume move faster than in the dry target experiment. The luminous effect in the center of the ejecta cone is brighter for the dry sandstone compared with the wet sandstone. In the wet target experiment, the luminescent ejecta plume is obscured by the cone that contains only nonluminescent debris. Ejecta–atmosphere interactions are different in both experiments. In the dry target experiment, turbulences affect the geometry of the curtain right from the beginning so that the ejecta could not develop the straight cone geometry. In the wet target experiments, vortices develop at a later stage, but hardly change the straight cone shape.

Figure 5.

 Comparison between ejecta clouds for impacts on dry (Exp. E1-3382, top) and wet sandstone (Exp. E3-3384, bottom). While the illumination technique is kept constant, the ejecta of the dry sandstone experiment appear brighter. The hot central ejecta plume of the wet target experiment is obscured by the darker and presumably cooler cone particles and is visible only in its uppermost part. The maximum ejection velocity of cone particles is slightly higher than that of the dry experiment. The central plume of the wet target experiment also has a higher ejection velocity than the plume of the dry experiment.

Temporal Evolution of Ejecta Cone Angles

All cratering experiments show a similar evolution of ejecta cone angles, which can be subdivided into three subsequent phases. Figure 6 illustrates the temporal development of the ejecta cone angles for a) dry and b) wet sandstone targets. In the first phase, the ejecta curtain is initially inclined at about 45° with respect to the target surface regardless of the water saturation of the sandstone target. A rapid steepening of the cone angle occurs until a plateau is reached at about πt = 100. πt denotes the nondimensional time, which is calculated by multiplying the absolute time by the impact velocity and dividing the result by the projectile radius. In the second phase, ejecta cone angles range from about 55 to 60° for dry targets to more than 70° for wet targets. In this period, the ejecta cones develop a straight geometry. After that, a third phase of steepening starts, which results in the formation of a tube (see Fig. 4) in which maximum values of more than 80° are achieved.

Figure 6.

 Ejecta cone angles for impact experiments on dry (a) and wet sandstones (b). Larger angles are observed for wet sandstone.

Transient Crater Growth

The diameter of the growing crater cavity was measured as a function of time. We used the nondimensional variable R(t)/a in which R(t) denotes the current crater radius as a function of time and a denotes the projectile radius, respectively, and plotted them as a function of the nondimensional time πt = vit/a in which vi denotes the impact velocity. The point-source concept (Holsapple and Schmidt 1987; Holsapple 1993), which applies to the growth of the transient crater (Housen and Holsapple 2011), leads to a power-law behavior of the growing cavity in the central area of the growth curve until target strength reduces the crater growth.

We used a scaling law given by Holsapple and Housen (2007) to determine the scaling parameter μ for transient crater growth:


ρp and ρt denote the density of the projectile and the target, respectively, ν is another scaling exponent, and K is a constant.

In our case, “time” denotes the absolute time after the impact. The determination of the crater radius R(t) by measuring the width of the neck of the ejecta cone leads to a systematic overestimation of the actual transient crater radius at the target surface level. This systematic error should apply to each measurement at any value of time t within the excavation phase, where strength and gravity do not affect cratering. This means that the rate of growth given by the slope of the fitted curve (and thus μ) is not affected by this error. The error does lead to an intercept value that is too high, and the position of the curve relative to literature values should be interpreted with some caution and should be seen as an upper limit.

For the dry sandstones (Fig. 7), no significant difference in transient crater growth appears for the small-scale and large-scale experiments shortly after the impact. The time when the growth curves start to flatten out, however, is different for the individual curves. This time denotes the end of transient crater formation. For experiment E1-3382, crater formation is finished earlier (πt ∼ 300) than for experiments A5-5125 and A6-5126 (πt ∼ 500). Furthermore, larger scaled crater radii are reached in the small-scale experiments. An example for a power-law fit to experiment A6-5126 is shown in Fig. 7. We compared our results with a formula for transient crater growth in the strength-regime (wet soils and rock, Holsapple and Housen 2007). The constant value 1.9 was calculated by the projectile density δ ∼ 8 g cm−3 and target density ρ ∼ 2.1 g cm−3 (see Holsapple and Housen [2007] for detailed description of the formula). The slope μ/1 + μ for our dry sandstone targets (μ/1 + μ ∼ 0.36 which leads to μ = 0.56) and the target materials to which this formula applies is approximately equal. However, the values for the scaled crater radii are smaller in our experiments compared with the values for wet soils and rock.

Figure 7.

 Comparison of the transient crater growth for impacts on dry sandstones for two small-scale and one large-scale experiment. Flattening of the data denotes the end of transient crater formation. Comparison with literature data (upper solid line, wet soils and rock, Holsapple and Housen 2007) indicates an equal growth rate, but smaller scaled crater radii, in our porous sandstones. The lower line is a fit to data from experiment A6-5126. Values for μ lie between 0.49 and 0.56 for these experiments.

For the wet sandstones (Fig. 8), similar growth curves can be observed. Again, for the large-scale experiments, the scaled crater radii are smaller and crater growth is finished earlier than for the small-scale sandstones. Furthermore, the more water saturated the sandstone is, the faster the crater grows and the larger the scaled transient craters are. Compared with “wet soils and rock” (Holsapple and Housen 2007), craters in our water-saturated sandstones grow at a faster rate. The scaled crater radii for the wet sandstones are closer to the line for wet soils and rock than the radii for the dry sandstones. The constant value 1.8 for this formula was calculated by the projectile density δ ∼ 8 g cm−3 and target density for wet sandstone ρ ∼ 2.2 g cm−3.

Figure 8.

 Comparison of the transient crater growth for impacts on partially and fully water-saturated sandstones for three small-scale and two large-scale experiments. Flattening of the curves denotes the end of transient crater formation. The growth rate increases with increasing water saturation. Larger craters are formed in wet sandstones compared with dry sandstones. The lower line is a fit to data from experiment A13-5182. Values for μ lie between 0.56 and 0.64 for 50% saturated sandstones and around 0.66 for 90% saturated sandstones.

The values for the scaling exponent μ for the dry sandstones lie between 0.49 and 0.56; for the partially saturated sandstones, the values are between 0.56 and 0.64. For the fully saturated targets, our values are close to 0.66.


In our hypervelocity impact experiments on dry and wet sandstones, we successfully recorded the dynamic ejecta behavior as well as the temporal development of the transient crater. It was shown that the ejection process can be subdivided into different stages and lasts a comparatively long time. The ejecta cone changes its angle as a function of time with respect to the target surface and finally develops into a “tube,” which ejects a significant amount of target material as shown in ejecta catcher imprints by Sommer et al. (2013). In all conducted experiments, a sequence of successive stages of ejection was observed.

Stage 1

A luminescent, presumably hot central plume emerges very soon from the impact point at an angle of 80 to 90° from the target surface. This plume becomes more voluminous with increasing impact energy. The central plume expands with the highest ejection velocities recorded. The plume shows a large amount of variation and modification in shape and size, which, we believe, depends on several factors, including porosity and target saturation, projectile energy, and target chamber pressure. Plumes like this were observed in several other studies (e.g., Schultz 1996; Schultz et al. 2005; Ernst and Schultz 2007). Simultaneously with the central plume, the ejecta cone develops and rapidly steepens its cone angle from 45° to a stable angle that is characteristic for the target properties.

Stage 2

The central plume disappears. A straight cone is ejected at a stable cone angle. The cone angle is dependent on the water saturation of the target (Fig. 6). This period correlates with the steady growth of the crater cavity.

Stage 3

This period is characterized by an increase in the cone angle. This steepening in the cone angle can be a consequence of: (1) a steepening of the ejection angle of the excavated particles while the velocity profile of particles with respect to their launch position is kept constant, (2) a change in the velocity profile of launched particles while the ejection angle is kept constant, (3) atmospheric turbulences and ring vortices that shift the cone to steeper angles, or (4) a combination of the three processes. Our analysis indicates that all three processes may operate simultaneously.

The ring vortex shown in Fig. 4 exists because the experiments were conducted in an atmosphere (air pressure 100 mbar). The vortex is most likely due to particle deflection caused by aerodynamic drag as well as forces resulting from reduced pressures behind the “upwards” moving ejecta cone (an impact experiment recently conducted at EMI under low atmospheric pressure shows that no ring vortex forms if the target chamber is evacuated). Such ring vortices are described by Schultz (1992) and Barnouin-Jha and Schultz (1996) where the vortex is inside the cone because of an “outwards” moving ejecta curtain.

Stage 4

The onset of stage 3 correlates with the completion of the transient crater cavity. Now spall fragments become entrained and are by volume the dominant ejecta at stage 4 (Dufresne et al. 2013). Spall fragments are typical for impacts into brittle solid bodies like sandstone where the cratering process is strength-dominated. An explanation for this behavior is given by Melosh (1984). As can be seen in Fig. 4, these spall fragments move much more slowly than the other ejecta (compare also Polanskey and Ahrens 1990).

Stage 3 continuously passes into stage 4 in which a tube-shaped stream of debris and spall fragments become ejected perpendicular to the target surface. The amount of target material excavated during this late stage greatly exceeds the mass excavated by the high-speed, cone-shaped ejecta (Dufresne et al. 2013; Sommer et al. 2013). The ejecta entrained into the tube have a velocity on the order of several tens of meters per second and consist of larger particles of up to several centimeters in size. The almost vertical ejection angle suggests that these late ejecta are caused by release waves reflected back from the rear surface of the sandstone targets, which arrive at the crater floor after the transient crater formation has finished and force shattered material to be ejected. Using a compression wave velocity of 2915 m s−1 (Moser et al. 2013), the two-way travel time for a wave within a sandstone cube with an edge length of 20 cm is about 137 μs, which is in agreement with the onset of the “ejecta-tube” (Fig. 4). Such an ejecta tube was also observed in a study by Michikami et al. (2007) in which impact experiments on targets with varying porosity were conducted.

Effect of Target Water on Ejecta Behavior and Crater Growth

Craters formed in water-saturated targets are wider than craters formed in dry targets (Figs. 3, 7, 8; Dufresne et al. 2013) which is possibly due to less inelastic effects (pore crushing) and thus less energy loss of the shock wave caused by incompressible water in the pore space. Furthermore, the compressive strength of the wet sandstone (58.4 ± 0.8 MPa, Poelchau et al. 2013) is lower than for the dry sandstone (67.3 ± 2.7 MPa, Poelchau et al. 2013). In addition, an influence of vaporization of water on crater size and shape is possible. Vaporization of water leads to a sudden volume increase obviously influencing both crater formation and ejecta dynamics (Kieffer and Simonds 1980).

The power law fits to the crater growth curves resulted in values ranging from 0.49 (dry target) to 0.66 (fully saturated target) for the scaling exponent μ. According to Holsapple and Housen (2007), this exponent has to lie in the range of 1/3 to 2/3. These authors describe the exponent μ to be dependent on porosity: For porous and nonporous materials, the exponent μ is closer to 1/3 and 2/3, respectively. They obtained values of about 0.55 for nonporous materials and about 0.41 for materials with moderate porosity (30–35%). In our study, μ increases with increasing water saturation (i.e., decreasing porosity) of the target material. Poelchau et al. (2013) determined μ = 0.57 ± 0.10 for the same experiments (dry sandstones) by crater volume scaling, which is in accordance with our values. The faster growth of the transient crater is most likely due to the faster passage of the shock wave. As described by Butkovich (1971) and Erskine et al. (1994) for tuffs, shock wave velocities are higher if the material is water saturated.

Comparison of our crater growth curves with a scaling law given by Holsapple and Housen (2007) shows that larger crater radii are reached for wet soils and rock compared with the dry sandstone used in our impact experiments. The crater growth curves for the wet sandstones, however, are closer to the values obtained for wet soils and rock. As a possible explanation, the “damping effect” of the pores due to pore crushing reduces the energy of the shock wave. Water within the pore space, however, counteracts this effect.

The weaker luminous effect of the ejecta plume for impacts on wet sandstones could have its origin in the reduction of target porosity by means of water. Porosity strongly influences the magnitude of the shock- and post-shock temperature, i.e., at a given pressure, much higher temperatures are reached in the porous material (e.g., Stöffler and Langenhorst 1994). The crushing of the pores as an adiabatic process leads to a temperature rise. Interstitial water prevents the pores from being crushed. We assume that reduction of the impedance mismatch between sandstone and pores by water saturation is the cause for a lower temperature and, hence, a weaker luminous effect.

Although there is no physical model describing the link between ejecta angles and the coefficient of internal friction, experimental and numerical studies show that higher friction leads to smaller ejection angles (Barnouin-Jha et al. 2007; Collins and Wünnemann 2007). In our experiments, the water in the pore space possibly reduces internal friction and thus increases the ejecta cone angles. Numerical calculations conducted by Senft and Stewart (2008) for hypervelocity impacts on rock targets with and without an icy layer (on the surface or buried) showed that ejection angles increase if an icy layer, which has a low coefficient of friction compared to the rock target, exists. Furthermore, impact experiments on target materials with different viscosity (e.g., water, wet clay slurry, stiff clay slurry, and dry sand) conducted by Greeley et al. (1980) showed that the ejection angle depends on viscosity of the target material. The higher the viscosity was, the smaller the ejecta cone angles were. In these experiments, impacts on wet clay slurry resulted in significantly steeper ejection angles than impacts on dry sand. These results also correspond to our results in which larger cone angles for water-saturated targets were measured.


Hypervelocity impact experiments in the strength-dominated regime on dry and wet sandstone targets at different scales were conducted and the ejecta behavior and the crater formation process were investigated. Optical high-speed recordings provided insights into the highly dynamic processes associated with the impact and the ejection. Phenomena such as an “ejecta tube,” which develops at late time steps after the impact and a hot central plume were observed. The results show a significant dependency of the temporal development of the ejecta cone angles as well as transient crater growth on pore space saturation: Wet sandstones show larger ejecta cone angles than dry sandstones. It was shown that the greater the water saturation is, the faster the transient crater develops. Furthermore, craters formed in wet sandstone targets have larger radii and volumes. Power law fits applied to the crater growth curves resulted in a dependency of the scaling exponent μ on pore space saturation. Larger values for μ are obtained for water-saturated targets. Comparison with a scaling law for nonporous rocks shows that porosity reduces crater sizes, whereas interstitial water counteracts this process.

Acknowledgments— We appreciate the funding of the MEMIN research unit FOR-887 and the DFG project TH 805/4-1 by the German Research Foundation (DFG). Furthermore, we thank B. Hermalyn and an anonymous reviewer for their detailed comments which helped to improve the manuscript, as well as N. Artemieva for the editorial work. We also thank the EMI technicians for their support.

Editorial Handling— Dr. Natalia Artemieva