Deformation and melting of steel projectiles in hypervelocity cratering experiments


Corresponding author. E-mail:


Abstract– We carried out hypervelocity cratering experiments with steel projectiles and sandstone targets to investigate the structural and mineralogical changes that occur upon impact in the projectile and target. The masses of coherent projectile relics that were recovered in different experiments ranged between 58% and 92% of their initial projectile masses. A significant trend between impact energy, the presence of water in the target, and the mass of projectile relics could not be found. However, projectile fragmentation seems to be enhanced if the target contains substantial amounts of water. Two experiments that were performed with 1 cm sized steel projectiles impacting at 3400 and 5300 m s−1 vertically onto dry Seeberger sandstone were investigated in detail. The recovered projectiles are intensely plastically deformed. Deformation mechanisms include dislocation glide and dislocation creep. The latter led to the formation of subgrains and micrometer-sized dynamically recrystallized grains. In case of the 5300 m s−1 impact experiment, this deformation is followed by grain annealing. In addition, brittle fracturing and friction-controlled melting at the surface along with melting and boiling of iron and silica were observed in both experiments. We estimated that heating and melting of the projectile impacting at 5300 m s−1 consumed 4.4% of the total impact energy and was converted into thermal energy and heat of fusion. Beside the formation of centimeter-sized projectile relics, projectile matter is distributed in the ejecta as spherules, unmelted fragments, and intermingled iron-silica aggregates.


Abundant extraterrestrial material has been delivered to Earth throughout its history via impact events; the physical and chemical processes that affect the composition and distribution of projectile matter upon impact, however, are not yet understood in sufficient detail (Kearsley et al. 2004; Koeberl 2007). Fragments of the projectile are preserved at young, simple impact craters, like Meteor (Barringer) Crater, USA (Buchwald 1975; Mittlefehldt et al. 2005); Wolfe Creek, Australia (Knox 1967); Wabar, Saudi Arabia, Monturaqui, Chile, Henbury, Australia (Gibbons et al. 1976); or Kamil, Egypt (Folco et al. 2010, 2011; D’Orazio et al. 2011). Large craters only rarely contain projectile relics (Maier et al. 2006). Traces of the projectile occur in larger and older impact structures as chemical components (e.g., nickel, chromium, osmium isotopes, platinum group elements [PGEs]) in various settings, including impact melt rocks and breccias, and fractures in the crater floor (Grieve et al. 2006; Tagle and Hecht 2006; Tagle and Berlin 2008). This “meteoritic component” allows the deduction of the type of projectile via measurement of the element abundances and ratios of the PGEs, although interelement fractionation of the PGEs during emplacement or postimpact alteration may hamper a correct identification of the projectile type. The best-known example for a meteoritic component is the iridium anomaly in the worldwide Cretaceous-Paleogene (former “K/T”) event bed that represents the distal ejecta of the approximately 200 km sized Chicxulub impact structure, Yucatán, Mexico. The Chicxulub impact event triggered the “K/T” mass extinction event (Hildebrand et al. 1991; Schulte et al. 2010).

The physical conditions during emplacement of projectile matter, modes of target-projectile mixing, and possible fractionation processes are not well constrained (e.g. Evans et al. 1994; Koeberl 1998). Projectile matter could be emplaced in condensates from a vapor phase, as melt droplets, as fine-grained solid material, or as spall fragments. Experimental cratering with well-defined projectile matter and sophisticated ejecta catcher arrangements is a good approach to study projectile dissemination (Shoemaker et al. 1963; Jammes et al. 1983; Rowan et al. 1996; Kearsley et al. 2007). Previous attempts to detect projectile matter in melt lithologies produced by hypervelocity experiments were made with scanning electron microscope (SEM) or electron microprobe (EMP) analysis (Hörz et al. 1983; Rowan and Hörz 1995). These experiments were performed with metal targets (e.g., Au, Cu, Al, Mo), and rocks (e.g., basalt glass, dunite) or Fe-Ni-PGE alloy (Evans et al. 1994) as projectiles. Only Shoemaker et al. (1963) and Jammes et al. (1983), who investigated projectiles, performed experiments in approximate analogy to nature by using quartz sand or sandstone as targets. Generally, only limited indications of mixing between projectile and target melts were reported in these experiments. For silicate projectiles, Hörz et al. (1983) and Rowan et al. (1996) observed fractionation between some major elements, which was attributed to selective oxidation/reduction and/or selective vaporization processes. Impact-related high-temperature vaporization of silicates and metals was also investigated by means of laser pulse experiments (Gerasimov et al. 2005).

The analysis of the deformation inventory of projectile relics has not been considered in detail so far and is a focus of this contribution. We present results of hypervelocity cratering experiments using steel projectiles (Figs. 1 and 2) and sandstone targets. We specifically focus on two experiments that were conducted as a pilot study (P1-2808) (Schäfer et al. 2006; Kenkmann et al. 2007, 2011) and test shot (3232) in the framework of the Multidisciplinary Experimental and Modeling Impact Research Network (MEMIN), but include for comparison recent MEMIN experiments. One major goal of this study is to reach a better understanding of the parameters that control the fate of projectiles in cratering events, i.e., their dissemination into the target by injection, and ejection, vaporization, and condensation. This study resulted in the selection of appropriate projectile materials (Ebert et al. 2013) for the recent series of impact experiments (Poelchau et al. 2013). The impact velocities of the experiments presented here are comparable to those typical for small iron projectiles decelerated by the Earth’s atmosphere and, hence, are applicable to small terrestrial craters like the Kamil crater (D’Orazio et al. 2011).

Figure 1.

 Optical micrograph of steel SAE 4130 used as a projectile in the MEMIN cratering experiments P1-2808 and 3232: The homogeneous matrix is peppered with heterogeneously distributed Si-C-rich domains (dark to medium grey) of various sizes and compositions. Note that the SiC domains are primary and are not a relic of the sample grinding and polishing procedure. Reflected light, // nicols.

Figure 2.

 Recovered steel projectiles. A) front, B) rear of the initially spherical projectile of experiment P1-2808. C) front, D) rear of the projectile relic of test shot 3232. Image width: approximately 2 cm.


Experimental Setup

Cratering experiments were conducted with a two-stage light-gas gun developed at the Ernst-Mach-Institute (EMI), Efringen-Kirchen, Germany (Schäfer et al. 2006). We fired steel spheres of 10 and 12 mm diameter and masses of 4.1–4.6 g with velocities ranging between 2500 and approximately 5300 m s−1 vertically onto blocks of Seeberger Sandstein. Cratering and ejection processes were monitored by a high-speed framing camera (Fig. 3). Details of the experimental setup and the obtained craters are given in Kenkmann et al. (2011) and Poelchau et al. (2013).

Figure 3.

 High-speed shadowgraph image of dry experiment P1-2808, showing a reflected projectile relic (black arrow indicates bright particle) as it overtakes the central plume. The brightness indicates a high temperature.

Geochemical Analysis

Whole rock geochemical analysis of the target was carried out by X-ray fluorescence spectroscopy (XRF) on glass pellets at the Museum of Natural History Berlin (MfN). In addition, projectile and target materials were characterized by optical and electron microscopy. The deformed samples were analyzed using a scanning electron microscope (SEM, JEOL-JSM 6300 at MfN; LEO 1525 field emission gun microscope at Albert-Ludwigs-Universität Freiburg, ALU) in backscattered (BSE) and secondary electron mode (SE). High-resolution element mapping (4076 × 3072 pixels) was acquired with a Bruker AXS Quantax EDS system using a silicon drift detector that allows ultra-high resolution. The energy resolution of ≤125 eV at Mn-Kα, resulting in ≤48 eV at C-Kα worked at count rates of 100,000 cps.

Texture Analysis

For texture analysis of projectile relics, the technique of electron back scattered diffraction (EBSD) was applied. EBSD enables the measurement of the crystallographic orientation of minerals and metals, and the calculation of misorientation axes and angles between any two data points (Prior et al. 1999). Orientation contrast (OC) images allow the mapping of all misorientation boundaries in a specimen, and thus provide a location map for EBSD analyses. EBSD coupled to OC imaging in the SEM enables complete specimen microtextures to be determined (Prior et al. 1999). For the analysis of the texture, the projectile relic was fully impregnated in epoxy resin and sectioned using a 0.8 mm thick diamond blade mounted on a precision saw. A low rotation speed and sufficient cooling were used to minimize preparation-related deformation during cutting. Grinding and subsequent polishing were performed using silicium carbide lapping films and polycrystalline diamond suspension down to 3 μm grain size. A final surface treatment with aluminium oxide suspension was applied using a soft synthetic cloth under the sample’s own weight for 15 min.

SEM-EBSD work at the LEO 1525 field emission gun microscope (ALU) was performed under high vacuum conditions with a 120 μm aperture. Sample charging due to surface oxidation was avoided by vapor-depositing a 2–3 nm thick carbon film on the polished surface. Prior to mounting in the SEM chamber, the sample was demagnetized using an electromagnetic coil to avoid image drift during acquisition. Orientation contrast images on steel were made at 12 kV acceleration voltage and a sample-to-backscatter detector distance of less than 2 mm. Atomic contrast images were taken using the same set-up, but longer working distances. EBSD analysis was performed using a commercial Oxford Crystal system with a modified detector. The SEM was operated at 18 kV and with a 27 mm working distance. Due to hardware limitations, the minimum electron back scattered pattern acquisition time was 150 milliseconds per point. EBSD maps were constructed via beam scanning with a step size of 0.15 μm. The crystal lattice of iron bcc (body-centered-cubic) was used for automatic indexation of the captured diffraction pattern.

Results of the Cratering Experiments

Characterization of Projectile and Target

Our analysis mainly concentrated on the experiments P1-2808 and 3232. In these experiments, we used spherical projectiles of alloyed heat-treatable steel SAE 4130 (German Industry standard material number DIN 1.7218) whose composition is given in Table 1. Figure 1 illustrates the textural and chemical heterogeneities of this steel. The homogeneous matrix of this steel contains heterogeneously distributed Si-C-rich domains of various sizes and compositions (Si 25–70 wt%, C < 75 wt%). The melting temperature of SAE 4130 is approximately 1430 °C.

Table 1. Chemical composition of the projectile matter in experiment P1-2808 and test shot 3232 used for this study (steel SAE).
(wt%)Projectile steel SAE
Cr0.92 ± 0.17
Mn0.51 ± 0.10
Co0.14 ± 0.03
Ni0.07 ± 0.02
Mo0.16 ± 0.05

The target material is upper Triassic Seeberger Sandstein, a sandstone quarried at Seeberg near Gotha, Germany (Stück et al. 2011; TRACO-company). The bed chosen for the target block (bank 5) is an arenitic sandstone with grain size of 0.17 ± 0.01 mm (Kenkmann et al. 2011). It exhibits a centimeter-spaced layering with a porosity varying between 12 and 20 vol%. Quartz grains often contain thin coatings of iron oxides and clay minerals. Accessory minerals are feldspar, mica, zircon, and hematite. The rock has the following composition (XRF) [wt%]: 97.2 SiO2, 0.11 TiO2; 1.2 Al2O3, 0.25 Fe2O3, 0.08 CaO, 0.02 Na2O, 0.12 K2O, 0.6 LOI (loss on ignition).

In subsequent experiments, we also used the iron meteorite Campo del Cielo and the alloyed steel D290-1 as projectiles (Table 2) (Ebert et al. 2013; Poelchau et al. 2013).

Table 2. MEMIN experiments in which larger projectile relics were recovered. For experimental details, see Poelchau et al. (2013).
ExperimentProjectile materialSphere diameterImpact velocityProjectile massImpact energyTarget conditionsTarget dimensionProjectile relic massProjectile relic characteristics
(mm)(m s−1)(g)(kJ)(sandstone)(cm)(g)(%)
P1-2808SAE 4130 steel1053404.158.4Dry100 × 50 × 502.8469.27One large bowl-shaped fragment
P2-2809SAE 4130 steel1052704.156.9Wet – 44%100 × 50 × 50Tiny fragments <1 mm
3232SAE 4130 steel1034004.123.7Dry100 × 50 × 50∼3.0∼73One large bowl-shaped fragment
E1-3382D290-1 steel1245644.647.9Dry80 × 80 × 503.2370.19One mid-size bowl-shaped fragment and irregular-shaped fragments
E2-3383D290-1 steel1245784.648.2Wet – 50%80 × 80 × 502.6858.36One small bowl-shaped fragment and irregular-shaped fragments
D4-3299Iron meteorite1035074.1225.3Dry50 × 50 × 502.9571.71One large bowl-shaped fragment
D5-3300Iron meteorite1025034.1413.0Dry50 × 50 × 503.2478.33One large bowl-shaped fragment
E3-3384Iron meteorite1245904.648.5Wet – 50%80 × 80 × 504.2592.32Irregular-shaped fragments


In experiment P1-2808, a 1 cm steel sphere (4.1 g) was launched into an air-dry sandstone block with an impact velocity of 5340 m s−1. The impact energy was 58.4 kJ. A comprehensive description of the experiments is given in Kenkmann et al. (2011). Planar impact approximation using coefficients for iron and Coconino sandstone (Kieffer and Simonds 1980) yielded a peak shock pressure of 70 GPa (Poelchau et al. 2013). The air pressure in the target chamber was 500 mbar. The test shot 3232 was carried out under identical condition with an impact velocity of 3400 m s−1 (23.7 kJ). For experimental details of the experiments listed in Table 2, we refer to Poelchau et al. (2013).

Recovered Projectile

In the experiment P1-2808, 2.84 g (69 wt%) of the projectile was recovered from a fiber board mounted approximately 55 cm adjacent to the target surface. The remaining 31 wt% of the projectile were melted or vaporized, or dispersed to very fine fragments. The soft-wood fiber building board surrounding this piece was burnt, documenting the high temperature of the steel fragment. This piece was traced in one shadowgraph image (Fig. 3) as it overtook the expanding central plume at a speed of about 450 m s−1at 280 μs after the initial contact of the projectile with the target.

In the experiment 3232, 73% of the projectile was recovered. Projectiles were also recovered from other MEMIN experiments (Fig. 4; Table 2). The masses of coherent projectile relics range between 58% and 92% of their initial projectile masses. A significant trend between impact energy and mass of the recovered projectile relic could not be found (Fig. 5). Likewise, no systematic correlation could be determined between the water content of the target and the recovery rate of the projectile. However, the chance of fragmentation of the projectile is enhanced if target water is present. For instance, in experiment E3-3384 (Table 2; Fig 4D), numerous irregularly formed shrapnel fragments developed. In experiment P2-2809 (Table 2), no larger fragments could be collected.

Figure 4.

 Morphologies of recovered projectile relics of MEMIN experiments. Front and rear sides are shown for each projectile relic. Shrapnel-like splinters developed in experiment E3-3384 that was performed with a partly water-saturated target. Experimental details are listed in Table 2 and in Poelchau et al. (2013).

Figure 5.

 The ratio of the projectile relic mass to the initial mass of the projectile strongly scatters and shows no significant correlation with the impact energy. Data are taken from Table 2.

Surface Features of the Deformed Projectile

The recovered projectile relics all have the shape of a deformed bowl with a convex outer and an irregularly formed or concave inner side, a serrated, sometimes fragmented brim, and a knobby region in the center of the inner side (Figs. 2 and 4). The overall shape points to intense plastic deformation of the projectiles. The detailed surface analysis of the projectile relic of experiment P1-2808 shows that the serrated rim is formed by brittle fracturing under tensile stresses, yet the fracturing is intimately linked with melting (Figs. 6A–D). Bubbles and droplets (Figs. 6A, 6C, 7C, and 7D) indicate a temperature near the liquid–vapor transition of the steel. Delicate melt filaments bridge the open fissures (Figs. 6C and 6D). They indicate a rather low viscosity of the melt. The convex-shaped part of the projectile relic contains arrangements of parallel trending grooves and striae, which are similar to slickenside surfaces that develop on shear planes (Fig. 7A). The orientation of striae and grooves are perpendicular to the serrated brim. Some shear planes contain tiny spheroids (Fig. 7B). Foam textures of burst spheroids indicate boiling (Figs. 7C–D) as a result of shock and shear heating. The boiling temperature of iron at ambient pressure is approximately 2750 °C. The SEM analyses of the composition of melt and spheroids suggest melting of projectile and target (Fig. 8). Mixing of iron and silica occurs in various proportions, although Fe analysis in Si-melt is complicated by the presence of tiny Fe-droplets (Ebert et al. 2013). A minor fraction of the Si-melt may have originated from the SiC-rich domains in the steel (Fig. 1). EDX analyses show that the Fe-spherules contain traces of Ca, Al, and Ti on their surfaces. Sources of these elements are likely the phyllosilicate coatings around the quartz grains and Ti-oxide minerals in the target sandstone. Deposition of these coatings may be due to selective absorption or condensation.

Figure 6.

 A–D) SEM-SE micrographs of the surface of the recovered projectile relict (experiment P1-2808) showing bubbles of various sizes and amorphous films decorating the projectile surface. Thin filaments bridging tensional cracks indicate low viscosities of the melt films.

Figure 7.

 A–D) SEM-SE micrographs of the surface of the recovered projectile relict (experiment P1-2808). A–B) The smooth and striated surface is interpreted as a shear plane. Tiny spherules indicate melting by shear heating. C–D) Boiling bubbles and melt films on a shear plane of the projectiles’ surface.

Figure 8.

 Element mapping of the projectile surface (experiment P1-2808). Iron (green) of the projectile and silicon (blue) show intense mixing. Both phases indicate melting: iron forms spherules, whereas silicon forms a thin melt film and delicate apophyses. (Width of micrographs approximately 400 μm.)

Mosaics of atomic contrast BSE images provide overviews of cross-sections of the projectile relics of experiment P1-2808 and 3232 (Figs. 9A and 9B). Their surfaces are partly covered by highly deformed discontinuous layers with thickness varying between a few micrometers and several millimeters. They occur along the convex and concave sides of the projectiles and, in the case of P1-2808, also along fractures within the main body of the projectile relic. In both experiments, these layers are composed of highly shocked and fragmented quartz crystals impregnated by a matrix consisting of a composite silica-oxide and steel melt (Fig. 9D) that was welded to the projectile surface. Evidence of melting is given by the presence of quenched Fe-droplets and a foam structure indicative of degassing in both experiments. The amorphous nature of the melt is confirmed by the absence of an electron diffraction pattern in these areas. In the case of experiment P1-2808 in which a higher level of shock was achieved, all quartz grains show multiple sets of planar deformation features (PDF) (Fig. 9D) and the grains are welded by lechatelierite. In experiment 3232, PDFs are less frequent.

Figure 9.

 A) BSE overview of the studied sample cross-section of the projectile relic P1-2808 (bright areas: steel, dark grey areas: sandstone remnants). B) 172 stitched BSE images display a cross-section of the projectile relic 3232. Note the discontinuous thin silicate layer at the outer rim of the sample and the rounded sandstone/steel imbricate next to the inner rim. Stars indicate areas investigated by means of Orientation Contrast imaging and EBSD mapping. SF is a spall fracture. C) Orientation contrast (OC) image of a deformed zircon single crystal adjacent to projectile relic 3232 (see B for location). D) BSE image of vesicular lechatelierite and shocked quartz grains with abundant planar deformation features. This type of shock metamorphosis is typical in the matter that is glued to projectile P1-2808.

In experiment 3232, a single crystal (80 μm) of zircon (Fig. 9C) was found in this outer layer. It shows intersecting planar to curvi-planar and lensoid microstructures of a few micrometer width. BSE orientation contrasts change abruptly and are either lower or higher than the host grain, suggesting that the lamellae may represent micro-twins (Timms et al. 2012) or deformation bands. The diffraction patterns of the lamellae are weak and vanish with progressive radiation. This suggests that these areas are strongly deformed domains that locally lead to amorphization. Granular fabrics formed by amorphization and decomposition of ZrSiO4 could not be detected (Fiske et al. 1994; Wittmann et al. 2006).

The two cross-sections of the projectile relics of experiments P1-2808 (Fig. 9A) and 3232 (Fig. 9B) display convex–concave sickle shapes. However, the projectile relic of P1-2808 (Fig. 9A) has a more complex geometry and is subdivided into several splinters. The projectile relic of experiment 3232 shows a knobby region near the center of its concave side. This is occupied by a mass of fragmented quartz grains and steel splinters that most likely were detached from the projectile relic during spallation at the rear of the projectile. A thin film of steel cements the individual elements together. Melting of quartz grains in this zone is rare.

Deformation of the Recovered Projectile

Orientation contrast imaging was carried out for the experiments 3232 and P1-2808 (Fig. 10). The variable intensity of the contrast here does not depend on differences in the atomic masses, but on the orientation of the crystal lattice planes with respect to the incident electron beam due to the electron channeling mechanism. Orientation contrast within the recovered projectile of experiment 3232 reveals a strong grain shape preferred orientation (GSPO). Size, orientation, and shape of recrystallized grains correlate with their position within the deformed projectile and indicate that the microstructures are the result of impact deformation (Figs. 10A–C). In the vicinity of the convex side of the sample, grain long axes are parallel to the curvature of the rim. The diameters of the crystal domains are usually <2 μm and their aspect ratios (ratio of long and short axis) may exceed 10. In the concave portion of the projectile, and in the central area, a more irregular to chaotic pattern with no clear GSPO is found. Due to their extremely small size, several domains do not show a clear diffraction pattern. Orientation contrast imaging of the recovered projectile of experiment P1-2808 also displays domains with fine grain sizes and strong GSPO parallel to the curvature of the projectile surface (Fig. 10D). The grain sizes are even finer grained and range in the submicrometer size. However, these domains alternate with areas of large uniform grain aggregates that are not strained and that formed perfectly equilibrated triple point networks (Figs. 10D–F). This is an unequivocal evidence for grain growth under static conditions and indicates postshock thermally induced annealing of the microstructure.

Figure 10.

 EBSD coupled to OC imaging in the SEM enables complete specimen microtextures to be determined. Orientation contrast (OC) images of areas situated along the outer rim of the steel projectile (see location of positions in Fig. 9B). Note the intense grain shape-preferred orientation and grain elongation dipping NE/SW (A) for position 1, horizontal (B) for position 2, and dipping NW/SE (C) for position 3 in projectile relic 3232. Long grain axes are oriented parallel to the projectile surface indicating strain elongation. Orientation contrast images (D) through (F) were carried out on the projectile relic of experiment P1-2808. They display very fine-grained domains with a grain shape-preferred orientation alternating with annealed domains.

EBSD mapping of the projectile relic 3232 (Figs. 11A and 11B) shows that the GSPO correlates with a crystallographic preferred orientation (CPO) (Figs. 11C and 11D). The maxima of the pole figures suggest that several symmetrically equivalent glide planes along {111} and (100) were activated and that slip occurred in the [110] direction. The CPO indicates that deformation mechanisms involving dislocation glide and creep were active and that the projectile was plastically deformed during its collision with the target rock. Further evidence of plasticity is observable in the sub grain and grain boundaries maps shown in Figs. 11E and 11G. Some intensively elongated grains show an internal development of subgrain boundaries. The cumulated internal misorientations reach values up to 50° (Fig. 11F) as a result of elongation. The presence of low-angle grain boundaries indicates that dislocations have migrated into low-angle boundaries. Further grain deformation is accommodated by subgrain rotation and leads to subgrain rotation recrystallization and the building of high angle boundaries. As a result of dynamic recrystallization, fine grains developed. Indications for subsequent annealing could not be found. In addition to plasticity, the projectile shows several internal spall fractures (Fig. 9A).

Figure 11.

 A, B) EBSD maps from locations 1 and 3 (see Fig. 9B) of sample 3232 with step sizes of 0.15 μm. The color coding represents the preferential orientation of the horizontal direction in the given inverse pole figure look-up table. C, D) (100), (110), and (111) pole figure distributions; equal area projection, lower hemisphere, linear scale, and relative intensity. Orientation data extracted from maps 1 and 2, respectively. Note the strong CPO related to the plastic deformation of steel. E) Grain/subgrain boundary maps extracted from a selected region of map 1 (see location A). Image in the background is the EBSD pattern quality map. Green line: local misorientation between 5 and 10 degrees; red line: local misorientation higher than 10°. F) Misorientation profile across a highly elongated grain. The value of misorientation is calculated from the first orientation value calculated at point X. G) Grain/subgrain boundary map extracted from map 2. Same color coding as E).

EBSD mapping of the projectile relic of experiment P1-2808 (Figs. 12A and 12B) is shown in Fig. 12. The extreme fine grain sizes (<1 μm) do not allow a proper EBSD mapping and indexing of crystallographic orientations. Where grain sizes allow indexing, the GSPO also correlates with a CPO as in projectile relic 3232. In contrast, the domains with large grains and stable triple points do not show signs of a CPO. The grains are internally strain-free (Fig. 12C). The misorientation between adjacent grains is random (Fig. 12B). The selective annealing of the projectile relic of experiment P1-2808 is most likely a result of the heterogeneous distribution of SiC inclusions within the steel (Fig. 1). In domains where SiC inclusions occur, grain boundaries are pinned and grain growth is prohibited. In domains that are free of impurities, static grain growth could take place.

Figure 12.

 EBSD mapping results obtained on sample P1-2808 (see Fig. 9A for map localization) using a beam acceleration voltage of 20 kV and a step size of 180 nm. A) Pattern quality map. The pattern quality depends on crystal perfection and phase orientation as well as on experimental parameters (Humphreys et al. 2001). B) Inverse pole figure map (Z direction, normal to the plane of measurement) overlain on the pattern quality map in semi-transparency mode. Note the color uniformity within the large grains indicating the lack of internal misorientations. C) Texture component map overlain on the pattern quality map in a semi-transparency mode. The rainbow color coding represents the deviation from an arbitrary reference orientation (RO; Euler angles: Psi1 = 21.93°, PHI = 40.32°, Psi2 = 32.29°). The blue color represents no deviation from the reference, the red color a maximal deviation of 60°. The large angular deviation spread visible in the histogram to the right of the map indicates the lack of any crystallographic-preferred orientation.

Projectile Residues in the Crater Floor and Dispersed in the Ejecta

In addition to the large bowl-shaped projectile relics, fine-grained metallic particles were collected in the MEMIN experiments with a bar magnet from (1) the ejecta catcher, and (2) the pulverized material decorating the crater floor surface. The particles, which are interpreted as residues of the projectiles, comprise of (1) angular fragments, (2) spheroids that are similar in shape to spheres produced in gas welding, and (3) delicate carbon-rich paramagnetic apophyses (Fig. 13). The apophyses from experiment P1-2808 intensively pervade and impregnate the fine-grained sandstone target and weld the quartz grains (Fig. 13). Injection along early formed fracture planes would theoretically form two-dimensional injection planes. The dendroid shape appearance of these apophyses does not support the assumption that the metallic injections occurred along fracture planes. The compositional variety of the residues is large and exceeds the range defined by the compositional end-members steel, quartz, and phyllosilicate; carbon is locally enriched. A possible source of carbon could be the carbon-rich inclusions in the steel (Fig. 1) or carbon from the fiber board.

Figure 13.

 SEM-SE micrographs of a paramagnetic fragment collected from the crater floor of the dry experiment P1-2808. A–C) Apophyses permeate a deformed sandstone aggregate. The branches are enriched in C, Fe, and Si. D) Polished cut surface of one of the apophyses (SEM-BSE micrograph).


Planar impact approximation suggests that the peak shock pressure in experiment P1-2808 may have locally reached 70 GPa (Poelchau et al. 2013). The peak shock pressure in experiment 3232 may have reached 40 GPa. The dynamic response of steel and iron to shock wave loading was investigated, e.g., by Barker and Hollenbach (1974) or Brar and Rosenberg (1996). The Hugoniot elastic limit of almost pure iron was found to be 1.37 GPa. This provides a limit for the brittle response of the recovered projectile. At shock pressures >13 GPa, the passage of the shock front causes iron-bcc to transform into a close-packed phase where most of the structure is isotropic hexagonal close-packed (hcp) and, depending on shock strength and grain orientation, some fraction is austenite face-centered-cubic (fcc) (Kadau et al. 2007).

The observed melting and even vaporization of steel in experiment P1-2808 are in stark contrast to the expected shock and postshock heating of iron. For instance, iron shocked to 39 GPa experiences a volume compression of 19% and temperature rise of only approximately 320–340 °C (Kadau et al. 2007). The postshock temperature rise of steel upon unloading from a peak pressure of 60–70 GPa is on the order of only 400 °C. This temperature is clearly insufficient to melt considerable amounts of the steel projectile, which has a melting temperature of 1430 °C. Significant differences in shock impedance between the iron phase and the <40 μm-sized Si-C-rich domains (Fig. 1) may, however, explain the presence of solidified Si-rich shock-melt filaments. Multiple shock reverberation at the iron-SiC phase boundaries could lead to a shock pressure equilibration between iron and SiC and would cause a strong temperature increase in SiC.

A possible source for the strong projectile heating in addition to shock and postshock heating is the strong temperature increase in sandstone upon shock loading. Porous sandstone achieves a shock temperature of 1000 °C at shock pressures of 15 GPa (Ahrens and Gregson 1964; Stöffler and Langenhorst 1994). Extraordinary heat pulses in sandstone are caused by the high shock impedance contrasts between the quartz grains and the interstitial pores and the abrupt closure of pore space (Kieffer 1971). In dense quartz, 45 GPa are necessary to reach this temperature. The temperature pulse within the porous target may specifically affect the immediate projectile–target interface and its surrounding where the heat could be conducted. Shoemaker et al. (1963) performed an impact experiment with a steel projectile and Coconino sandstone, and estimated that the total heat flow across the interface is about 4 J cm−2 leading to a net temperature increase of only 10–20 °C in the steel projectile. Thus, heat conduction from a hot target to the projectile is limited and cannot account for the high temperature within the projectile.

More important heat sources stem from dissipation of internal energy as heat during plastic deformation and frictional shearing (Kenkmann et al. 2000; Langenhorst et al. 2002; Van der Bogert et al. 2003; Spray 2010). Both processes can be observed in the investigated samples and are spatially separated. Whereas shear heating predominantly affected the surface of the projectiles and even led to localized boiling of iron at the immediate projectile–target interface, plastic deformation pervasively heated up the interior of the recovered projectile by crystal plasticity. The texture analysis of the projectile relics revealed that both dislocation glide and dislocation creep played important roles for the ductile response, whereas twinning is of subordinate importance. Dislocation glide led to a strong grain shape and crystallographic-preferred orientation. In the projectile relic of experiment 3232, this is particularly observed near the convex outer part of the recovered projectile and the grain elongation is parallel to the surface (Figs. 10B and 11A–C). Dislocation creep resulted in dynamic recrystallization and the formation of fine grain sizes with low aspect ratios. In the projectile relic of experiment P1-2808, the dynamically recrystallized grain size is finer than in experiment 3232, indicating a larger amount of accommodated strain in experiment P1-2808. The creep behavior of iron alloys strongly depends on the C-content (Lesuer et al. 2001). High-temperature creep with dynamic recrystallization is expected at homologous temperatures T/TM of 0.7 and higher, where TM is the melting temperature. Dislocation creep and dynamic recrystallization are associated with atom diffusion through the crystal lattice. As diffusion is a rate-dependent process, the period of deformation plays a critical role. High-temperature creep is restricted to the period of the contact and compression phase that lasted approximately 1–2 μs.

The presence of deformation features indicating high-temperature creep suggests that both projectile relics were heated to at least 900–1000 °C. For the projectile relic of experiment P1-2808, the real temperature was most likely higher than this. The corresponding thermal energy Q, required for a temperature change ΔT of the projectile with mass m, is given by Q = cpmΔT. The specific heat capacity cp of steel SAE 4130 is 0.48 KJ−1 kg−1 K−1. Thus, 2.3% of the total impact energy (P1-2808: 58.4 kJ) is consumed for heating the projectile. Assuming that the entire missing mass (1.26 g) of the projectile in experiment P1-2808 was melted (31%), a minimum thermal energy of 850 J is required to achieve this temperature and about 365 J to convert the solid at its melting point into a liquid without increase in temperature (latent heat of fusion of iron 288 J g−1), equivalent to approximately 2.1% of the total impact energy. This value may represent an upper limit as it is not sure that the entire missing projectile was melted. Even boiling of iron was locally found to occur on shear planes at the projectile target interface and in residues. However, the quantity of vaporized iron and the input of heat for vaporization cannot be estimated here, but may counterbalance the overestimated thermal energy for melting. To summarize, heating plus melting of the projectile consumed about 4.4% of the total impact energy in the case of experiment P1-2808. The projectile experienced high temperatures for at least 250 μs (Fig. 3). However, high-temperature creep was restricted to the period of the contact and compression phase, as the differential stress required for dislocation motion may have dropped to zero during the ejection process. After this time, the projectile relic left the presumably very hot central ejecta plume (Fig. 3). Static annealing of steel SAE 4130 occurs above approximately 850 °C. The observed annealing of the projectile relic of experiment P1-2808 indicates that this temperature must have persisted for a while after deformation, whereas in experiment 3232, the temperature should have rapidly dropped below 850 °C. In the smaller scale impact experiment of Shoemaker et al. (1963) who also used a steel projectile and a sandstone target (impact energy 3.7 kJ), about 10% of the projectile material was melted.

The recovered projectiles of our MEMIN experiments were strongly distorted. Their convex sides lined the growing crater cavity. The projectile relic of experiment P1-2808 was more severely stretched and strained. When the shock wave was reflected at the inner side of the projectile, the deformed projectile was subjected to brittle tensile fragmentation. Subsequently, the splinters were stacked and pushed together (Fig. 9A). To a lesser degree, this is also seen in the knobby and rough region in the inner part of the concave, bowl-shaped projectile relic of experiment 3232 that contains numerous steel splinters that were originally located at the rear of the projectile, but became intermingled with target material that most likely intruded from the sides, when the projectile fully penetrated into the target. The recovered material shows a variety of complex textures and chemical peculiarities. For example, the origin of the silica in the recovered projectile matter cannot be constrained with certainty. Textural and chemical homogeneity of the projectile material are important to avoid unbalanced mass distribution, and a clear chemical distinction from the target sandstone. Ebert et al. (2013) give a detailed geochemical analysis of projectile–target interaction using Campo del Cielo iron meteorite material as a projectile.


Small terrestrial impact craters of several tens to hundreds of meters in diameter are often formed by iron meteoroids. While traversing the Earth’s atmosphere, iron meteoroids with sizes of a few meters to decameters are often capable of withstanding the aerodynamic stresses without complete fragmentation. The bodies decelerate to a few kilometres per second and finally form meteorite craters like Kamil, Egypt, or Wabar, Saudi Arabia. To study the fate of these projectiles, hypervelocity cratering experiments at impact velocities ranging between 2500 and approximately 5300 m s−1 with centimeter-sized steel projectiles and sandstone targets were carried out. The masses of coherent projectile relics that were recovered in different experiments ranged between 58% and 92% of their initial projectile masses. A systematic relationship between the impact energy and the relative proportion of projectile relic could not be found.

A detailed microstructural investigation of projectile relics showed that their surfaces indicate shearing and brittle fracturing. Melting and boiling of iron and silica are evident from the formation of bubbles, droplets, and melt filaments. For the experiment P1-2808 that was carried out at an impact velocity of 5300 m s−1 and an impact energy of 58.4 kJ, it is estimated that at least 4.4% of the impact energy is dissipated by heating and melting of the projectile. Dislocation glide and creep along with dynamic recrystallization are the main deformation mechanisms during plastic deformation of the projectile, followed by a later annealing of the microstructure. Small-scale residues of the projectile are distributed in the ejecta as spherules, unmelted fragments, and intermingled iron-silica aggregates.

Acknowledgments— We thank R. T. Schmitt (MfN) and I. Domke (WWU) for providing selected geochemical data, M. Poelchau (ALU) for coordination, H. Knöfler (MfN) and H. Ickler (ALU) for technical support, and the other MEMIN group members for numerous fruitful discussions. The manuscript benefitted from constructive reviews by Dr. Axel Wittmann, Dr. C. Van der Bogert, and the associate editor Dr. N. Artemieva. This work is funded by the German Research Foundation DFG, Research Unit FOR-887, grants DE 401/23-1, HE-2893/8-1, KE-732/18-1.

Editorial Handling— Dr. Natalia Artemieva