Chemical modification of projectile residues and target material in a MEMIN cratering experiment


  • Matthias EBERT,

    Corresponding author
    1. Museum für Naturkunde (MfN), Leibniz-Institut für Evolutions- und Biodiversitätsforschung, Invalidenstraße 43, D-10115 Berlin, Germany
      Corresponding author. E-mail:
    Search for more papers by this author
  • Lutz HECHT,

    1. Museum für Naturkunde (MfN), Leibniz-Institut für Evolutions- und Biodiversitätsforschung, Invalidenstraße 43, D-10115 Berlin, Germany
    Search for more papers by this author
  • Alexander DEUTSCH,

    1. Institut für Planetologie, Westfälische Wilhelms-Universität Münster (WWU), Wilhelm-Klemm-Str. 10, D-48149 Münster, 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:


Abstract– In the context of the MEMIN project, a hypervelocity cratering experiment has been performed using a sphere of the iron meteorite Campo del Cielo as projectile accelerated to 4.56 km s−1, and a block of Seeberger sandstone as target material. The ejecta, collected in a newly designed catcher, are represented by (1) weakly deformed, (2) highly deformed, and (3) highly shocked material. The latter shows shock-metamorphic features such as planar deformation features (PDF) in quartz, formation of diaplectic quartz glass, partial melting of the sandstone, and partially molten projectile, mixed mechanically and chemically with target melt. During mixing of projectile and target melts, the Fe of the projectile is preferentially partitioned into target melt to a greater degree than Ni and Co yielding a Fe/Ni that is generally higher than Fe/Ni in the projectile. This fractionation results from the differing siderophile properties, specifically from differences in reactivity of Fe, Ni, and Co with oxygen during projectile-target interaction. Projectile matter was also detected in shocked quartz grains. The average Fe/Ni of quartz with PDF (about 20) and of silica glasses (about 24) are in contrast to the average sandstone ratio (about 422), but resembles the Fe/Ni-ratio of the projectile (about 14). We briefly discuss possible reasons of projectile melting and vaporization in the experiment, in which the calculated maximum shock pressure does not exceed 55 GPa.


Impact cratering has been, and still is, a major process in the origin and evolution of all the solid bodies of the solar system (French 1998). Only under rare circumstances are natural impacts observed directly, e.g., the impact at Sikhote-Alin, Russia (Krinov and Fesenkov 1959); the Carancas event, Peru (Kenkmann et al. 2009); or the collision of Shoemaker-Levy 9 with Jupiter (Weaver et al. 1995). Projectiles that cause impact events can be preserved in the form of fragmental residues or geochemical traces. For instance, small and young craters that are formed by the impact of iron meteorites like the recently discovered Kamil crater (Egypt; D’Orazio et al. 2011), Wabar crater (Saudi Arabia; Mittlefehldt et al. 1992), and even the Meteor Crater 1.2 km in diameter (Arizona, USA; Mittlefehldt et al. 2005) each contain, or lie amidst debris containing, strong deformed fragments of projectiles (shrapnel) as well as melted, disseminated residues in impact breccias and crater floor fractures. At larger craters, the nature of the impactor can usually be only deciphered by means of geochemical analysis of siderophile elements in melt-bearing rock types. The process(es) of mixing projectile matter into target and impactite materials are far from being understood (e.g., Mittlefehldt et al. 1992; Evans et al. 1994). It could be emplaced in a vapor phase, as melt droplets, or as fine-grained solid material.

Along with the exploration of terrestrial and planetary craters, and numerical modeling of cratering processes, laboratory experiments are the most important tool to study impact processes. In the past 50 yr, numerous studies on hypervelocity impact experiments were carried out, using different set-ups and different target and projectile materials. Most authors describe crater dimensions, such as size, depth, and shape, as a function of projectile size and weight, velocity, and angle of impact (Moore et al. 1963; Shoemaker et al. 1963; Hörz 1969, 2012; Gault 1973; Polanskey and Ahrens 1990; Grey et al. 2002; Schultz 2003; Cintala and Hörz 2008). The projectiles were predominantly manufactured of artificial materials like steel, aluminum, lexan and polyethylene, and were accelerated onto basalt, granite, dolomite, sandstone, loose sand, chondritic material, metal (steel, Al, Fe, Mo) blocks and plates as well as onto ice.

Projectile material and analytical methods used in previous experiments, however, neither allowed study of complex projectile-target interaction nor fractionation between siderophile elements typically used to identify and characterize a meteoritic component in nature (e.g., Palme et al. 1981; Tagle and Hecht 2006). The identification of meteoritic matter in impact melt rocks relies on the study of siderophile elements such as Fe, Ni, Co, and Cr and the platinum group elements (PGE), which are significantly enriched in meteorites compared with terrestrial rocks (e.g., Palme et al. 1981). While only a few experimental studies have dealt with the physical and chemical dissemination of the projectile during the impact process (Hörz et al. 1983; Jammes et al. 1983; Evans et al. 1994; Rowan and Hörz 1995; Wozniakiewicz et al. 2011), the results of these experiments clearly show the importance of melting, vaporization, and mixing of projectile and target in the cratering process. Jammes et al. (1983) impacted quartz sand targets vertically with aluminum projectiles. In these experiments, only a small part of the projectile melt adjacent to the fused sand reacted with silica, but the major part of the projectile neither mixed into the target material nor reacted with it. Evans et al. (1994) observed heterogeneous distributions of projectile material among the ejecta. They found that the early high-angle ejecta between 55° and 75° are most projectile rich. Rowan and Hörz (1995) conducted shock recovery experiments using silicate projectiles (e.g., dunite) and diverse metal (e.g., Al and Cu) targets. The authors described physical mixing of projectile and target melts, but did not observe significant chemical reactions between them. Wozniakiewicz et al. (2011) performed hypervelocity impact experiments with pyrrhotite as well as pentlandite projectiles and Al foils targets, as analogue to the Stardust mission. They observed fractionation processes during melting and mixing of molten projectile and target material. The sulfur of the projectiles was partially separated from the Fe and Ni, leaving nearly the original Ni- to Fe- ratio (Wozniakiewicz et al. 2011).

However, little is known about processes that control interelement fractionation between siderophile elements during cratering on silicate-dominated natural targets. The impact research unit MEMIN (Multidisciplinary Experimental and Numerical Impact research Network) is aimed at a better understanding of impact crater formation using fully instrumented mesoscale hypervelocity cratering experiments in combination with numerical modeling and a thorough post-mortem investigation of the crater, the ejecta, and the projectile remnants (Kenkmann et al. 2011, 2013). Here, we present results of one MEMIN hypervelocity cratering experiment using an iron meteorite sphere as projectile and a sandstone target. Sandstone and iron meteorite were chosen due to their strong chemical difference (Table 1). This set-up enables (1) explicit identification of minute particles or condensates of the projectile, and (2) determination of mixing and/or fractionation processes between projectile and target matter. Hence, the use of well-characterized projectile material is mandatory for this project of the MEMIN research unit. This study addresses a fundamental topic in impact cratering, namely the fate of the projectile upon impact, although our experimental impact velocities are lower than most planetary impacts.

Table 1. Chemical composition of the Campo del Cielo (CDC) meteorite, Seeberger Sandstein, and the phyllosilicate-bearing sandstone matrix.
(wt%±2σ)KamacitecTaenitecSchreibersitecRhabditec aCDC (whole rock)
Fe92.93 ± 0.5267.09 ± 6.7751.23 ± 0.2645.03 ± 3.0992.60
Ni6.25 ± 0.3932.95 ± 6.6034.24 ± 0.2241.18 ± 3.156.62
Co0.45 ± 0.030.02 ± 0.010.08 ± 0.010.07 ± 0.020.43
P0.07 ± 0.020.10 ± 0.0415.02 ± 0.0814.03 ± 0.620.25
Si0.01 ± 0.006b.d.l.b.d.l.b.d.l.n.a.
modal (%)b88. 
  1. aBuchwald (1975).

  2. bBunch and Cassidy (1968).

  3. cElectron-microprobe data.

  4. dXRF data; Ni, Co, Cr were measured with ICP-MS.

  5. b.d.l. = below detection limit; n.a. = not analyzed.

SiO294.77 ± 0.9046.60 ± 5.52
Al2O33.11 ± 0.1025.86 ± 9.50
Fe2O30.45 ± 0.093.25 ± 2.60 (FeO)
CaO0.05 ± 0.010.23 ± 0.12
MgO0.12 ± 0.021.13 ± 0.79
Na2O0.15 ± 0.050.1 ± 0.18
K2O0.23 ± 0.012.93 ± 2.10
TiO20.31 ± 0.040.16 ± 0.26
P2O50.02 ± 0.010.08 ± 0.07
Ni3.33 ± 0.41b.d.l.
Co0.96 ± 0.04b.d.l.
Cr63.33 ± 21.01b.d.l.

Experimental and Analytical Methods

Experimental Setup

The MEMIN experiment under discussion (D3-3298) was carried out with the extra large two-stage light-gas gun (Lexow et al. 2013). For experiment D3-3298 a spherical projectile (Ø 10 mm, weight 4.12 g) was manufactured at the EMI machine shop on a lathe from a piece of the iron meteorite Campo del Cielo. The projectile impacted along a horizontal trajectory with a velocity of about 4.56 km s−1 onto a 50 × 50 × 50 cm sized cube of air-dry Seeberger Sandstein, a sandstone from the Thuringian basin in Germany. The impact energy was calculated at about 43 kJ, the estimated peak pressure at the contact between projectile and target is 55 GPa, using the planar impact approximation with Coconino sandstone and steel parameters (Melosh 1989). The target chamber was evacuated to 0.3 bar during the shot, to reduce deceleration of the projectile in flight and turbulent deflection of the ejecta.

Material ejected from the crater was collected with a newly designed catcher system, in which bricks of phenolic foam and tiles of degassed petrolatum were assembled in a modular system (Reiser et al. 2011). The catcher was mounted parallel to the target surface at a distance of 53 cm and had an entry hole 12 cm diameter for the projectile. Impact and ejection processes were recorded with a high-speed camera at 105 frames per second. After the shot, the target chamber was flooded with air; loose particles were collected from the bottom of the target chamber. The impacted sandstone block and the catcher assemblage were carefully removed for further investigation. The ejecta fragments analyzed in this study originate from the “inner zone” of the green phenolic foam plates (Sommer et al. 2013). They were recovered from the small green foam pieces with tweezers and brushes. The position of single ejecta pieces was recorded for determination of the ejecta angle. The “inner zone” of the catcher system contains material that was ejected with angles between 70° and 85° to the target surface.

Analytical Techniques

The recovered ejecta particles were characterized by optical and scanning electron microscopy (JEOL JSM-6610LV) at the MfN Berlin. Quantitative chemical analyses were performed at the MfN Berlin with the JEOL JXA-8550F electron microprobe, equipped with a field emission gun and 5 wavelength-dispersive spectrometers. Analytical conditions varied, depending on the measured phase. Projectile residues were measured with 15 kV accelerating voltage and 50 nA beam current; sandstone melt with 15 kV and 30 nA. The sample spot-size varied between 1 μm for metallic components and 3 μm for silicate components. We used pure element standards for Fe, Ni, and Co, and mineral standards for the other elements. The raw data were processed for matrix effects using a conventional ZAF routine in the JEOL series program. Special analytical conditions were applied to trace Fe and Ni at low concentrations in the order of a few 100 ppm (e.g., in shocked quartz). In this case, Fe and Ni were analyzed at 15 kV, 60 nA, 60 s counting time on peak and background and adding the counting rates of two spectrometers. The JEOL software enables a detailed consideration of possible peak overlapping between the different emission lines of Fe and Ni (and Co). A detection limit (D.L.) of 100 ppm was achieved for Ni (D.L. of Fe about 200 ppm) using the JEOL calculation method:


where Iback is the average intensity of the background X-rays, tback is the counting time of the background signals in seconds, InetSTD is the intensity of net X-rays of the standard sample, and mass(%)STD is the mass concentration in the standard sample. The accuracy of the measurements was in the order of 5–10% tested on two NIST glass standards (NIST SRM 610 and 612; Table 2). For measurements on the NIST standards the average of 112 measurements performed with a defocused beam (20 μm) was used as Ni is distributed slightly heterogeneously in these materials (Jochum et al. 2011). Therefore, the precision of trace element analysis is most likely better than the values given in Table 2, which include the standard heterogeneity. Even at concentrations below the calculated D.L., the standard values were reproduced with a rather high accuracy of 10% (see Table 2).

Table 2. Measurements of the NIST glass standards.
NIST SRM 610MfN (this study)sigmaLiteratureasigmaNISTbsigma
  1. aJochum et al. (2011); preferred average values of the authors.

  2. bReference value determined by the National Institute of Standards and Technology (from Jochum et al. 2011).

Ni (ppm)43133458.744854
Fe (ppm)4294545894589
Ni (ppm)431738.83.7 38.80.2
Fe (ppm)5719 512 512

Whole rock geochemical analysis of the target sandstone was carried out using X-ray fluorescence spectroscopy (XRF, Siemens SRS 3000) at the MfN Berlin on glass tablets. In order to get precise values for Ni, Co, and Cr, additional measurements were conducted with ICP-MS (ALS Environmental Lab, Vancouver). The proportion of the phyllosilicate-bearing matrix on the sandstone whole rock was determined in two steps on thin sections: First, element-mapping with SEM; and second, calculation with an image analysis program (ImageJ).


Composition of Target and Projectile

Target (Seeberger Sandstein)

The Seeberger Sandstein, a sandstone of Rhaetian age, was chosen as target material. Blocks were cut from the stratigraphic layer 3, which is characterized by fine grain sizes of 76–125 μm, high quartz content, a visible but not dominant layering, and a porosity of 23 ± 1 vol% of the bulk sandstone volume. The sandstone is comprised of about 89 vol% quartz, about 10 vol% phyllosilicate (evaluated by graphical analysis of EDX element-mapping images), and traces of accessory minerals (rutile, altered ilmenite, zircon, rare iron-oxides/hydroxides). The phyllosilicates, mostly clay minerals and subordinate micas, account for the majority of the sandstone matrix forming coatings around quartz grains and occasional accumulations of fine material within the pore space between the frameworks of quartz grains (Fig. 2). Individual phyllosilicate flakes of the matrix are smaller than 10 μm. The rare detrital micas have a grain size in the order of 100 μm. The composition of the sandstone and its phyllosilicate-bearing matrix is given in Table 1.

Projectile (Campo Del Cielo Meteorite)

The Campo del Cielo (CDC) meteorite was chosen as a material that represents natural “meteorite” impacts. Furthermore, large homogenous Fe-meteorite pieces are available to produce multiple identical projectiles that are also strong enough to resist the stresses that occur during the hypervelocity acceleration. Campo del Cielo is an IAB iron meteorite that belongs to the group of coarse octahedrites (Og) with a Widmannstätten pattern bandwidth of 3.0 ± 0.6 mm (Buchwald 1975). The Campo del Cielo meteorite has a homogenous kamacite matrix, minor amounts of taenite, plessite, schreibersite, and rhabdite (Fig. 1), that are distributed irregularly, as well as some rare silicate-inclusions. The modal composition of the meteorite and the chemical compositions of each phase are listed in Table 1. Taenite and plessite appear only as degenerated comb plessite or as pearlitic fields, in sizes smaller than 3 × 1 mm (Buchwald 1975). The most elongated taenites vary in composition. They predominantly have Ni-poor (Fe-rich) cores and Ni-rich (Fe-poor) rims. Iron-nickel phosphides (Fe,Ni)3P occur in two different textural modes in the Campo del Cielo meteorite: elongate idiomorphic rhabdites of micrometer-scale width (Fig. 1B) within coarser kamacite (e.g., Oshtrakh et al. 2011), and coarser blocky schreibersite macro-inclusions (50–200 μm; Fig. 1A). The schreibersite crystals are often fractured, presumably by the same stress that formed the Neumann bands (mechanical deformation twins) in the kamacite matrix (Bunch and Cassidy 1968). The silicate inclusions occur as centimeter-sized lenses and have an average composition of 37.9 vol% forsterite, 35.3 vol% chrome-diopside, 18.4 vol% enstatite, and 8.4 vol% oligoclase (Park et al. 1966). Material with silicate inclusions was avoided during preparation of the meteorite projectiles. Micro-CT scans (EMI) have been carried out to characterize the meteorite projectiles in terms of internal fractures, and distribution of the different mineralogical components.

Figure 1.

 A) Backscattered electron image (BSE) of the Campo del Cielo meteorite with various phases. B) Rhombohedral rhabdite in kamacite matrix of the Campo del Cielo meteorite projectile.


The different ejecta fragments were distinguished according to macro- and microscopic features (Figs. 2–4). The ejected material is represented inter alia by weakly deformed, highly deformed, and highly shocked ejecta fragments. These three types represent end members of different shock-metamorphic states. Consequently the ejecta contain fragments with intermediate stages between weakly deformed and highly shocked fragments.

Figure 2.

 Photomicrograph (A) and BSE image (B) of a weakly deformed ejecta fragment; for detailed explanation see text.

Figure 3.

 Photomicrograph (A) and BSE image (B) of a highly deformed ejecta fragment; for detailed explanation see text.

Figure 4.

 Photomicrograph (A) and scanning electron microscope image—BSE mode (B) of a highly shocked ejecta fragment; for detailed explanation see text.

This study is focused on the geochemical interaction between projectile and target material. Although only the highly shocked ejecta contain projectile material, we also characterize weakly and highly deformed ejecta to allow comparison between the unshocked sandstone with its shocked counterparts. Detailed information about grain size distributions of the ejecta are presented in Sommer et al. (2013).

The weakly deformed ejecta fragments represent almost unstressed micrometer- to millimeter-sized sandstone fragments of yellow-beige color, with a completely preserved original pore space (Fig. 2A). The quartz grains very rarely display cracks and fractures; the phyllosilicate-bearing matrix also lacks evidence for mechanical stress (Fig. 2B). The intensity of sandstone deformation in these fragments is comparable with ejecta type 1 of Kenkmann et al. (2011).

The highly deformed ejecta fragments show a distinctive white color (Fig. 3A). They occur as micrometer- to millimeter-sized fragments, but are on average smaller than weakly deformed ejecta material. Fracturing in quartz is typical, causing disintegration of the original texture (Fig. 3B). This strong fracturing increases the surface area of the quartz, so that the fragment appears macroscopically white. The shock significantly reduced the pore space and compressed and deformed phyllosilicates of the matrix, so that individual grains are barely recognizable. Large micas were strongly deformed (Fig. 3B). The highly deformed ejecta show no evidence of melting or other shock-metamorphic features, and projectile admixture has not been found. The intensity of deformation, indicated by the microstructures, is comparable with the type 3 sandstone deformation in Kenkmann et al. (2011). Kiefer (1971) described similar features in shocked fragments of the Coconino sandstone in Meteor Crater, which are characterized by a “snow” white color.

The highly shocked ejecta fragments are represented by white-gray variegated, micrometer- to millimeter-sized fragments (Fig. 4). In these ejecta fragments, the microstructure of the unshocked sandstone is completely lost. Moreover, the highly shocked ejecta fragments show various shock-metamorphic features including multiple sets of planar deformation features (PDF) in quartz, the onset to complete transformation of quartz to fused silica glass, and partial melting of the sandstone (Figs. 5 and 6). This partial melting of the sandstone mainly comprises the phyllosilicate-bearing matrix, but involves quartz, too. The presence of vesicles in the shock-produced melt substantiates the release of volatiles from the phyllosilicates during melting (Figs. 5 and 6B). However, vesicles also exist within partly melted quartz grains where no phyllosilicates were intruded (Fig. 5).

Figure 5.

 A) Shock-metamorphic features of the sandstone and projectile droplets (BSE-image). B) Different sets of PDF in quartz and silica glass (BSE-image); for detailed explanation see text.

Figure 6.

 A, B) BSE image of a highly shocked ejecta fragment showing typical textures and components. C) Magnification of Fig. 6B shows a molten margin of a partly fused projectile fragment; due to the high contrast, the silicate phases on the right side are not visible (BSE-image). D) Typical melt pocket (center) with a rim in a partly fused projectile fragment; for detailed explanation see text.

Figure 5A illustrates differences in the degree of shock-metamorphic overprint in quartz over a narrow area in one ejecta fragment: in the interior of grain Qtz-A, we see two sets of PDF together with some irregular cracks while onset of melting occurs at the grain margins. Grain Qtz-B shows an even higher degree of melt formation: melting starts at the rim and propagates along cracks and PDF. In this grain, the PDF are slightly expanded compared with the PDF of Qtz-A. Qtz-C, however, is almost completely transformed into silica glass. Slightly brighter areas in Qtz-C are the only remnants of the original quartz. Vesicles in Qtz-C are lined up along these unmelted relics. We observed traces of turbulent movement, such as schlieren of silica glass in the upper part of Qtz-C. The small sinuous silica glass between Qtz-A and Qtz-C also indicates movements in the sandstone melt. The alignment of vesicles in Qtz-C indicates melting and healing along a former crack.

Figure 5B shows further sets of PDF in quartz as well as the formation of glass along a fracture. The density contrast between crystalline and amorphous SiO2 is clearly visible in this backscattered electron (BSE) image with amorphous parts of lower density (PDF lamellae and fused silica glass) displaying a darker gray in the BSE-mode.

The gray parts of the highly shocked ejecta fragments are caused by finely disseminated projectile matter mixed with partially molten sandstone (Figs. 4 and 6). Projectile material is mechanically injected into the sandstone to various amounts. During the impact, the Campo del Cielo projectile was transformed into different states, comprising (1) partly fused fragments containing shocked schreibersite, melt pockets, and molten margins, and (2) projectile droplets. (1) Partly fused fragments (150–500 μm) usually have molten margins, which are generally rich in vesicles (Fig. 6C). Small melt pockets (10–40 μm) occur within these projectile fragments (Fig. 6D). (2) The melt pockets consist of two coexisting melts of different composition. One melt is situated at the rim of the melt pocket and appears brighter in the BSE image due to higher mean atomic number. The other melt of lower mean atomic number is located in the center of the melt pocket. Projectile fragments that show such melt pockets and molten margins are named “partly fused projectile fragments” in this study.

In addition, a metallic melt forms immiscible droplets (spherical and spheroidal shape) in the sandstone melt. Figures 5A and 6B demonstrate that metallic droplets have entered the low-viscosity sandstone melt, but not the fused silica glass. The size of the metallic droplets ranges from a few nanometers up to 50 μm. The spheroidal shape is caused by surface tension at the contact with quartz. Mineralogical and geochemical features in the highly shocked ejecta fragments are the most interesting results of this study, and they are described in more detail in the following sections.

Chemical Modification of the Sandstone

As described above, the sandstone melt is produced by preferential melting and mixing of the phyllosilicate-bearing sandstone matrix and quartz. In rare cases, rutile is involved. Melting of the sandstone resulted in melts of quite different compositions with SiO2 contents of 50.6 to 89.1 wt.%, Al2O3 of 2.51 to 16.4 wt.%, FeO of 5.78 to 32.3 wt.%, and NiO of 0.01 to 0.33 wt.% (Table 3). The chemical composition of the sandstone melt is variable at the scale of micrometers. Especially remarkable is the highly variable FeO content in the sandstone melt (Fig. 7). Figures 7 and 8 show metal contents instead of metal oxides allowing inclusion of Campo del Cielo meteorite data (nonoxide) into the diagrams. Although a certain fraction of iron in the sandstone melt obviously originated from the phyllosilicate matrix, the amount of iron-oxides/hydroxides (forming the occasional staining of the sandstone) is simply too low to contribute to the Fe budget of the melt (see Table 1; 0.45% Fe2O3 in Seeberger Sandstein). The FeO content in most of the analyzed sandstone melt patches is higher than the average FeO content of the phyllosilicate matrix (3.25 ± 2.60 wt.%; Table 1). The FeO concentration roughly shows a positive correlation with the Al2O3 content of the sandstone melt. NiO is also slightly enriched in the sandstone melt compared with the bulk sandstone (Fig. 8). Within the sandstone melt, Fe/Ni remains generally one to two orders of magnitude below that ratio in the projectile; yet it is shifted in some cases to lower Fe/Ni (Fig. 8). The shocked quartz with PDF and silica glass also show FeO and NiO enrichment compared to unshocked Qtz grains (Fig. 8). Quartz with PDF, and silica glass both contain minor FeO (<1.1 wt.%) and NiO (<0.06 wt.%). The average FeO content of shocked Qtz is 0.48 wt.%. The average Fe/Ni of Qtz with PDF (about 20) and of silica glass (about 24) is in significant contrast to the average Fe/Ni in the sandstone (about 422), but resembles Fe/Ni of the projectile (about 14). Representative microprobe data are given in Table 3.

Table 3. Electron-microprobe data for various phases in the highly shocked ejecta fragments.
Element (wt%)SiO2TiO2Al2O3FeONiOMgOCaONa2OK2OFe/NiTotal
  1. b.d.l. = below detection limit; n.a. = not analyzed.

Sandstone melt
Mean (n = 55)61.850.9211.1224.940.050.380.120.030.55500 
Silica glass
Mean (= 44)98.87 0.040.520.02    24 
1.36 0.080.460.02      
Quartz with PDF
Mean (= 47)99.07 0.050.440.02    20  
Unshocked quartz
Qtz.0198.86n.a.b.d.l.b.d.l.b.d.l.n.a.n.a.n.a.n.a. 98.86
Qtz.0298.74n.a.b.d.l.b.d.l.b.d.l.n.a.n.a.n.a.n.a. 98.74
Qtz.0398.73n.a.b.d.l.b.d.l.b.d.l.n.a.n.a.n.a.n.a. 98.73
Qtz.0499.09n.a.b.d.l.b.d.l.b.d.l.n.a.n.a.n.a.n.a. 99.09
Mean (n = 25)98.57                   
Figure 7.

 Fe versus Si for various materials of the highly shocked ejecta fragments. The insert shows in detail the “shocked quartz”-field (Qtz with PDF, silica glass). d.l. = detection limit.

Figure 8.

 Fe versus Ni for various materials of the highly shocked ejecta fragments. The inset represents the “shocked quartz”-field in detail (Qtz with PDF, silica glass). Unshocked quartzes are below the detection limits and not shown here.

Chemical Modification of the Fe-Meteorite Projectile

Microprobe analyses of each projectile material are compiled in Fig. 9 and Table 4 gives some representative analyses. The chemical modifications observed in shocked projectile material are discussed with respect to the unshocked CDC meteorite (Table 1).

Figure 9.

 Fe/Ni and Fe/Co for different phases of the projectile before and after the impact. The curved line represents the mixing line between unshocked kamacite and rhabdite. The dots on the mixing line are a subdivision in 20% steps. The inset shows in detail the data array close to the CDC meteorite ratio.

Table 4. Electron-microprobe data for various projectile residues (after impact).
Element (wt%)CoFeNiOPSiTotalFe/NiSphere Ø
  1. n.a. = not analyzed.

Partly fused projectile fragment
Mean (n = 63)0.4692.786.210.440.100.08100.0714.9   
Schreibersite (shocked)
Mean (n = 8)0.0953.3529.280.9014.990.1298.731.8   
Melt pocket (r = rim; c = center)
Mean (n = 7)0.3276.6721.440.321.110.0399.893.5 
Mean (n = 4)0.2964.9823.860.2710.590.02100.002.7 
Molten margins of partly fused projectile fragments
Mean (n = 28)0.4691.096.980.600.200.2399.5813.1   
Projectile droplets
Mean (n = 92)0.5988.738.730.390.400.3999.1410.212

The matrix of partly fused projectile fragments is oxidized to some extent and shows slightly higher P and Si values, but in general, resembles the original kamacite matrix. Schreibersite areas in the matrix of the partly fused projectile fragments have higher Fe/Ni due to an increase in Fe and decrease in Ni, but their P content remained unchanged. Shock metamorphism caused complete melting of the rhabdites and some melting of the kamacite host crystal producing melt pockets. The melt in the pocket shown in Fig. 6D is separated into a P-rich melt in the center, and a P-poor melt at the rim (Fig. 6D) that has a slightly higher Ni-content (Table 4). Microprobe line-scans through the melt pockets show Ni depletion in the surrounding kamacite.

The plot in Fig. 9 represents the ideal mixing line between the rhabdite and kamacite matrix. The curved line lies above the chemical compositions of the average melt pocket rim and center. This is due to Ni-enrichment in the rim and the center of the melt pocket, which originates by diffusion from the kamacite matrix. Furthermore, the molten margins of partly fused fragments slightly differ in its chemical composition compared with the unshocked CDC meteorite. These melts have on average 0.4 wt.% more Ni, 400 ppm more Co, and 1.7 wt.% less Fe than the CDC meteorite, yielding a decrease in Fe/Ni and Fe/Co. In addition, we observed P-values of the molten margins which are similar to that of the CDC meteorite, but the oxygen content is higher by about 0.6 wt.% suggesting some oxidation. These molten margins also contain up to 2300 ppm Si. Enrichment of Ni, Co, P, and Si and depletion of Fe are most extensive for the projectile droplets that have entered the sandstone melt. These metallic droplets vary significantly in composition (cf. Table 4). The enrichment of Ni and Co over Fe correlates negatively with the size of the projectile droplets, and is most prominent in spheres with a diameter smaller than 3 μm (Fig. 10). For this correlation only metallic spheres (no spheroids) were analyzed. The chemical compositions of the molten margins of projectile fragments and the metallic droplets generally range between the CDC whole rock ratio and its kamacite ratio (Figs. 9 and 11).

Figure 10.

 Ni content versus grain size of projectile droplets.

Figure 11.

 Fe/Ni and Fe/Co for different metallic spheres from natural craters compared with results of this study. Iron meteorite composition and ratio include Campo del Cielo and Canyon Diablo. The gray-scale gradient in the “projectile droplets” field reflects the numbers of measurements—with a high number at dark gray and low number at light gray (cp. Fig. 9).


Chemical Interaction between Projectile and Sandstone Target

As mentioned above, the partly fused projectile fragments contain metallic melts at their margins (Fig. 6C). Microprobe analyses indicate that these melts represent a mixture of the CDC meteorite components. The dominant contribution is from the kamacite matrix, with an admixture of P-rich phases. We assume that the rhabdites are the main P-donators as they are more homogeneously distributed than schreibersite. The dissemination of projectile material in the sandstone is a highly dynamic and turbulent process. It appears that some portion of the molten rims were separated from the projectile fragments, injected into the low viscosity sandstone melt, and disseminated as small metallic droplets. In these projectile droplets Ni and Co are enriched to a greater degree than Fe, compared with the molten margins of partly fused projectile fragments and the CDC meteorite composition. The positive slope of this trend, shown in Fig. 9, is a result of the removal of Fe (see above). The range of projectile material compositions observed in the ejecta cannot be explained as simple mixtures of taenite and kamacite (Fig. 9). The Fe/Ni and Fe/Co decrease, due to loss of iron, but Ni/Co remains nearly constant. Obviously the Fe depletion in the molten margins of projectile fragments and the metallic droplets is associated with the enrichment of Fe and to a lesser degree of other siderophile elements in the sandstone melt and in shocked quartz.

Summarizing these petrographic and geochemical observations, two coexisting but largely immiscible melts (projectile melt and sandstone melt) exist in the highly shocked ejecta fragments. Several studies of Meteor Crater (e.g., Hörz et al. 2002; Kearsley et al. 2004; Mittlefehldt et al. 2005), Wabar crater (e.g., Gibbons et al. 1976; Hörz et al. 1989), Monturaqui crater (e.g., Bunch and Cassidy 1972; Kearsley et al. 2004), and Kamil crater (D’Orazio et al. 2011) described similar petrographic features in siliceous impactite material. The authors observed numerous metallic droplets with meteoritic origin in vesicular glasses, comparable to our observations shown in Figs. 5A and 6B. In impact melts of Meteor Crater (Arizona, USA), Hörz et al. (2002) and Kearsley et al. (2004) described a decrease in Fe/Ni in meteoritic droplets compared with the composition of the Canyon Diablo meteorite, and a complementary increase in this ratio in the surrounding target melt. Based on microprobe data, Hörz et al. (1989) and Mittlefehldt et al. (1992) demonstrated that impact melt materials from Wabar crater are generally enriched in Fe over Ni compared with the iron meteorite impactor. The results of our study (Fig. 8; Table 4) are consistent with these data. In our study the average Fe/Ni in sandstone melt of highly shocked ejecta fragments is generally above the Fe/Ni in the bulk Campo del Cielo. The Fe/Ni also varies strongly in the sandstone melt between 120 and 1406. Thus, during the formation of Meteor Crater, Wabar crater, and the laboratory impacts of the MEMIN research unit, comparable selective element partitioning processes between the projectile and target occurred.

To quantify such fractionation processes, it is important to understand the behavior of the moderate siderophile elements Fe, Co, and Ni. Partitioning of these elements between silicate and metallic melts has been studied extensively in laboratory experiments at varying conditions of pressure (P), temperature (T), oxygen fugacity (fO2), as well as metal and silicate composition. All experimental data demonstrated that independent of P, T, fO2, and compositional variations the sequence is always DNimet/sil >> DComet/sil > DFemet/sil (e.g., Schmitt et al. 1989; Walker et al. 1993; Hillgren et al. 1994, 1996; Ohtani et al. 1997; Righter et al. 1997), where Dxmet/sil is the metal/silicate partitioning coefficient of element x. The experimental results clearly indicate that the partitioning coefficient of Fe is always lower than the coefficient for Ni. Therefore, iron partitions preferentially into the silicate melt compared with Ni and Co. The siderophile character of Fe, Ni, and Co is linked to the selective oxidation behavior of these elements (White 2011) which is based on differences in the affinity to form an oxide. The Gibbs free energies (ΔG) for oxidation of pure Fe, Ni, and Co are as follows: ΔG°Fe/FeO << ΔG°Co/CoO < ΔG°Ni/NiO for all temperatures. Thus, Fe oxidizes preferentially over Co and Ni. An iron meteorite can be considered as a metallic alloy whose oxidation is more complex than oxidation of pure metals. When oxidation of alloys (e.g., Fe-Ni-Co) proceeds with selective oxidation of a less noble metal like Fe, a depletion zone of this less noble metal is formed in the underlying alloy (Seo and Sato 1983). In our experiment, the molten margins of larger projectile fragments, which are enriched in Ni, could be considered as such a depletion zone, caused by an oxidation process. We suggest that the projectile material, especially the material that later forms droplets, is injected into the sandstone melt and immediately reacts as described above. During this process, the projectile material was slightly oxidized (cf. Tables 1 and 4). We consider gas trapped/originally present in the pore space of the sandstone, and the dissociation of H2O (Brett 1967) originating from the phyllosilicate-bearing matrix as potential sources for oxygen. This view is supported by the highly vesicular sandstone melt (Figs. 5 and 6), which implies vaporization of certain amounts of H2O, which in turn, may have been dissociated. Brett (1967), Gibbons et al. (1976), Kearsley et al. (2004), Kelly et al. (1974), and Mittlefehldt et al. (2005) considered selective oxidation as the main reason for enrichment of Ni (and Co) observed in metallic droplets in impactites from the Wabar, Monturaqui, Henbury, and Meteor Craters. D’Orazio et al. (2011) described similar reaction margins in the ablation crust of Gebel Kamil, the iron meteorite that formed the Kamil crater in Egypt.

Brett (1967) suggested that a combined oxidation-fractionation process occurred in a very short time interval prior to the projectile melt injecting into the target melt. This implicates formation of an iron-oxide layer in the projectile prior to the incorporation of FeO into the sandstone melt. In this model, an FeO “skin” around each metallic droplet partitions into the target glass, enriching it in iron, and causing an FeO-free surface of the droplet. In our samples, we did not observe a distinct oxide layer on projectile material. Small droplets have relatively greater surface areas for oxidation, thus possibly explaining the Ni- (and Co-) enrichment with decreasing size (Gibbons et al. 1976), as shown in Fig. 10.

It is apparent from Fig. 11 that projectile droplets at terrestrial craters differ from the respective original meteorite composition. For example, some spheres of meteoritic origin from the Wabar crater have Ni contents of about 90 wt.% (e.g., Gibbons et al. 1976). As previously mentioned, Fe oxidizes more preferentially than Co, which in turn oxidizes more preferentially than Ni. We did not observe this chemical behavior in the projectile material of our study. For most droplets, Ni/Co ranges between the ratios for the bulk CDC and kamacite. Only the most fractionated (Fe-poor) meteoritic droplets show some Ni/Co fractionation (Bunch and Cassidy 1972; Gibbons et al. 1976; Kearsley et al. 2004). At Fe/Co < 15 it is clearly shown that the Ni/Co-ratio of Wabar, Monturaqui, and Meteor Crater spheres differ significantly from their initial meteorite ratio (Fig. 11). In these cases Ni is on average more enriched than Co resulting in a higher Ni/Co. Another process causing selective oxidation of iron meteorite material occurs by interaction of the incoming projectile with air (Mittlefehldt et al. 2005). These authors show that ballistically dispersed metallic spherules from Meteor Crater are generally enriched in Ni and Co over Fe. Projectile droplets from the Meteor Crater that directly interacted with molten target material, however, show a much more pronounced fractionation between Fe and Ni + Co (Kearsley et al. 2004) than the spheres selectively oxidized in air (Fig. 11). Although the projectile droplets in our experiment are surrounded by a sandstone melt, we do not observe such an extreme inter-element fractionation as recorded in samples from Meteor Crater (Kearsley et al. 2004). We suggest that the difference in the intensity of interelement fractionation between nature and experiment is due to differences in physical conditions like maximum pressure, temperature, and time. Especially differences in cooling rates may influence the intensity of chemical interactions between projectile and target. In the case of the MEMIN experiments, composition and thus structure of the siliceous sandstone melt affect the amount of projectile elements dissolved. This melt varies considerably in Al content as a result of different proportions of molten quartz and Al silicates (e.g., phyllosilicates). The Al-rich melts are considerably more depolymerized than Si-rich and Al-poor melts (Mysen and Richet 2005). The relative enrichment of Fe in Al-rich melts is facilitated by the charge compensation of Al3+ by divalent cations.

We have documented in our highly shocked ejecta fragments a slight enrichment of meteoritic Fe and Ni in quartz with PDF and silica glass (Fig. 8; Table 3). This corresponds to an addition of 0.1–0.7% meteorite components to shocked quartz. In contrast to Fe/Ni in sandstone melts, the average Fe/Ni of these shocked SiO2 phases lies close to the ratio for the original Campo del Cielo (Tables 3 and 4). This result is unusual, and rather unexpected. Superficial contamination of the quartz by matter from adjacent projectile droplets as well as partly fused projectile fragments during polishing of the thin sections cannot be completely excluded. We intend to apply other sensitive preparation and analytical methods to test whether this observation can be confirmed.

P–T-Conditions During the Impact

Poelchau et al. (2011, 2013) calculated a maximum shock pressure of 55 GPa for our experiment using the planar impact approximation and material parameters for Coconino sandstone and steel. Such a maximum shock pressure, however, is insufficient to cause melting let alone vaporization of the meteoritic projectile. Therefore it is necessary to consider spatially localized factors that may boost local heating beyond the bulk temperatures calculated here to levels sufficient to melt projectile and target material. For instance, the melting point of an iron meteorite is about 1497 °C, its boiling point is about 3235 °C (Remo 1994), which is similar to pure iron with melting and boiling points of 1536 and 2860 °C, respectively. Several observations that are at odds with the calculated peak pressures need to be explained.

Highly shocked ejecta fragments such as illustrated in Fig. 5A display different shock-metamorphic features on a very small scale: melting of quartz seems to start at the quartz grain rims and along the PDF. Hence, it is possible that a transformation from diaplectic quartz to fused quartz (silica glass) has occurred within the PDF. This ongoing transformation is well illustrated in Qtz-B and Qtz-C (Fig. 5A). Vesicles within the Qtz-C and the schlieren-like rims indicate a predominantly molten state of this grain (fused Qtz). The PDF are isotropic glass lamellae along defined crystallographic planes in crystalline quartz; these PDF form by the transformation of quartz in the solid state in the shock pressure interval of 13–35 GPa. The width of the amorphous lamellae increases with pressure to finally merge at a shock pressure > 35 GPa into patches of diaplectic glass that finally replace the entire Qtz grain in a pseudomorphic manner (e.g., Langenhorst and Deutsch 1994; Langenhorst 2002). All these shock features occur at a distance of about 20 μm (Fig. 5) next to fused silica glass, substantiating a very heterogeneous PT history.

Concerning the projectile fragments, tiny vesicles in the molten rim of a partly fused projectile fragment (Fig. 6C) may indicate boiling of the projectile. This finding concurs with data by Kenkmann et al. (2007, 2013) for the MEMIN pilot study (P1-2808) with a steel projectile and a target of Seeberger Sandstein (impact velocity 5.3 km s−1; calculated peak shock pressure of 50–60 GPa). These authors report droplets with vesicles on the surface of recovered pieces of the projectile and foamy textures of projectile spheroids, indicative of temperatures near the liquid-vapor transition of steel.

In the numerical modeling of the formation of Meteor Crater, Artemieva and Pierazzo (2011) used the ANEOS for pure Fe as equation of state for the Canyon Diablo IAB iron meteorite projectile. They report an onset of melting at about 162 GPa, and of vaporization at about 320 GPa. The peak pressure of 55 GPa as calculated for our experiment is obviously far too low to explain the observed phase changes in the CDC projectile. To overcome this problem, we suggest the formation of locally restricted areas/volumes of significantly enhanced pressure and/or temperature conditions; such areas may be at or near the contact between projectile and sandstone target. Three principal mechanisms are envisaged.

(1) As suggested by Kenkmann et al. (2007, 2013) friction and deformation (i.e., plastic work) of the projectile may play an important role in the increase in temperature.

(2) The effect of shock heating in porous material may be the other process for substantial temperature increase on a macroscopic and microscopic level. Macroscopically, in porous targets, the kinetic energy of the projectile is transferred more effectively into target heating (Ahrens and Gregson 1964). The amount of plastic work during compaction and compression of a porous material is large compared to processes in a shock front passing through nonporous material, leading to higher temperatures after the shock wave has passed (Wünnemann et al. 2006). On the microscopic scale, “pore collapse” results in local pressure and temperature peaks. Recent developments of new material models for use in hydrocodes treat the collapse of pore space during the impact process (Güldemeister et al. 2013). These models provide a good approximation to understand the P–T differences documented by mineralogical changes. Güldemeister et al. (2011, 2013) showed that despite a general decrease in the average shock pressure due to porosity, a significant local increase in shock pressure (up to four times of the initial pressure) occurs in the vicinity of pores. Therefore, we suggest enhanced post shock temperatures (cp., e.g., Stöffler and Langenhorst 1994). This pore collapse mechanism can explain our observations in the highly shocked ejecta fragments, with low and high shock-metamorphic features occurring very close to each other.

(3) The target chamber was evacuated to about 0.3 bar during the impact experiment. The residual atmosphere of the chamber is shocked and compressed in front of the accelerated projectile. The shock compression leads to heating of the air and consequently also of the projectile surface. Berry et al. (2007) measured surface temperatures of up to 1800 °C during a free flight experiment of a hemispherical graphite projectile, accelerated to 4.5 km s−1 at 0.32 bar experimental gas pressure. As the experimental conditions of Berry et al. (2007) are similar to our experiments we also expect some heating and maybe melting of the projectile front prior to impact.


We have observed complex mixing and shock features in ejecta fragments collected from a MEMIN experiment with a 1 cm sized projectile of Campo del Cielo iron meteorite and a sandstone target. These features include the occurrence of PDF in quartz, transformation of quartz into diaplectic glass, as well as the formation of fused vesicular silica glass. For the first time meteoritic Fe and Ni were detected in shocked quartz. Regarding the projectile, various amounts of material are mechanically injected (as partly fused fragments) into the sandstone. Heterogeneous melting of sandstone and projectile occurred, followed by mixing of both melts. This process is connected with significant interelement fractionation. In sandstone melt, meteoritic Fe is selectively enriched over Ni and Co, whereas in coexisting meteoritic droplets and molten margins of projectile fragments, Ni and Co are enriched over Fe. This fractionation most likely results from differences in the siderophile character, or in other words, from differences in the reactivity of the three elements with oxygen. Data from the Meteor and Wabar craters show trends similar to those observed in the mesoscale laboratory craters of the MEMIN project. The variability and strong deviation of Fe/Ni, and Fe/Co in the impact melts from the respective ratios in the meteorite cause a severe problem for identification of the projectile types in impact craters as already noted by Attrep et al. (1991) and Mittlefehldt et al. (1992). In accordance with these authors, we recommend caution in assigning specific projectile types without detailed knowledge of projectile-target mixing and interelement fractionation processes involved.

Acknowledgments— We thank A. Kearsley and D. W. Mittlefehldt for their reviews and their constructive comments. We also thank N. Artemieva for helpful comments on the manuscript. We thank P. Czaja (MfN) for help with the microprobe analyses, K. Born (MfN) for the support at the scanning electron microscope, R.T. Schmitt (MfN) for the whole-rock XRF analysis, and H. R. Knöfler for the sample preparation. L. Scharfe (Universität Potsdam) and I. Domke (WWU) assisted with petrographical and chemical analysis. MEMIN project coordination by M. Poelchau (ALU) and numerous fruitful discussions with the other MEMIN team members are greatly appreciated. This study is part of the MEMIN program supported by the German Science Foundation DFG (Research Unit FOR-887; De 401/23-1, He-2893/8-1, Ke-732/18-1).

Editorial Handling— Dr. Natalia Artemieva