The first MEMIN shock recovery experiments at low shock pressure (5–12.5 GPa) with dry, porous sandstone

Authors

  • Astrid KOWITZ,

    Corresponding author
    1. Museum für Naturkunde, Leibniz-Institut für Evolutions- und Biodiversitätsforschung, Invalidenstraße 43, 10115 Berlin, Germany
    Search for more papers by this author
  • Ralf T. SCHMITT,

    1. Museum für Naturkunde, Leibniz-Institut für Evolutions- und Biodiversitätsforschung, Invalidenstraße 43, 10115 Berlin, Germany
    Search for more papers by this author
  • W. UWE REIMOLD,

    1. Museum für Naturkunde, Leibniz-Institut für Evolutions- und Biodiversitätsforschung, Invalidenstraße 43, 10115 Berlin, Germany
    2. Humboldt Universität zu Berlin, Unter den Linden 6, 10099 Berlin, Germany
    Search for more papers by this author
  • Ulrich HORNEMANN

    1. Ernst-Mach-Institut, Am Klingelberg 1, 79588 Efringen-Kirchen, Germany
    Search for more papers by this author

Corresponding author. E-mail: astrid.kowitz@mfn-berlin.de

Abstract

Abstract– As part of the MEMIN research program this project is focused on shock deformation experimentally generated in dry, porous Seeberger sandstone in the low shock pressure range from 5 to 12.5 GPa. Special attention is paid to the influence of porosity on progressive shock metamorphism. Shock recovery experiments were carried out with a high-explosive set-up that generates a planar shock wave, and using the shock impedance method. Cylinders of sandstone of average grain size of 0.17 mm and porosity of about 19 vol%, and containing some 96 wt% SiO2, were shock deformed. Shock effects induced with increasing shock pressure include: (1) Already at 5 GPa the entire pore space is closed; quartz grains show undulatory extinction. On average, 134 fractures per mm are observed. Dark vesicular melt (glass) of the composition of the montmorillonitic phyllosilicate component of this sandstone occurs at an average amount of 1.6 vol%. (2) At 7.5 GPa, quartz grains show weak but prominent mosaicism and the number of fractures increases to 171 per millimeter. Two additional kinds of melt, both based on phyllosilicate precursor, could be observed: a light colored, vesicular melt and a melt containing large iron particles. The total amount of melt (all types) increased in this experiment to 2.4 vol%. Raman spectroscopy confirmed the presence of shock-deformed quartz grains near the surface. (3) At 10 and 12.5 GPa, quartz grains also show weak but prominent mosaicism, the number of fractures per mm has reached a plateau value of approximately 200, and the total amount of the different melt types has increased to 4.8 vol%. Diaplectic quartz glass could be observed locally near the impacted surface. In addition, local shock effects, most likely caused by multiple shock wave reflections at sandstone-container interfaces, occur throughout the sample cylinders and include locally enhanced formation of PDF, as well as shear zones associated with cataclastic microbreccia, diaplectic quartz glass, and SiO2 melt. Overall findings from these first experiments have demonstrated that characteristic shock effects diagnostic for the confirmation of impact structures and suitable for shock pressure calibration are rare. So far, they are restricted to the limited formation of PDF and diaplectic quartz glass at shock pressures of 10 GPa and above.

Introduction

A significant number of known terrestrial impact structures occur in quartz-bearing sedimentary targets that are characterized by significant porosity and, quite likely, the presence of pore water. So far, only one systematic study of shock metamorphism was carried out of a porous, water-bearing sandstone target at Meteor Crater. This work led to a shock classification scheme for this type of rock (Kieffer 1971; Kieffer et al. 1976) (Table 1). This calibration scheme is quite different from that for nonporous quartz-bearing rocks. Shock classification of porous, water-bearing sandstones is also, like that for nonporous quartz-bearing rocks, based on five progressive shock stages, but uses the complete removal of porosity in shock stage 1; varied amounts of quartz, coesite, and glass in shock stages 2–4; and the occurrence of a vesicular melt with lechatelierite as diagnostic criterion for shock stage 5 (Kieffer 1971; Kieffer et al. 1976; and modified by stöffler 1984; Grieve et al. 1996; Stöffler and Grieve 2007). Another systematic study with CaCO3-bearing sandstones was carried out at the Haughton impact structure, Canada, which uses the classification scheme of Kieffer (1971) but added a shock stage 6 for the occurrence of recrystallized SiO2 glass (Osinski 2007).

Table 1. Shock classification of porous, water-bearing sandstones (based on Kieffer 1971; Kieffer et al. 1976; modified by Stöffler 1984; Stöffler and Grieve 2007).
Shock stageShock pressure range* (GPa)Lithology
  1. *Average pressure; locally, peak pressures can be much higher.

0≤0.2–0.9Undeformed sandstone
1a0.2–0.9–∼3Compacted, deformed sandstone with remnant porosity
1b ∼3–∼5.5Compacted, deformed sandstone compressed to zero porosity
2∼5.5–∼13Dense (nonporous) sandstone with 2–5 wt% coesite, 3–10 wt% glass, and 80–95 wt% quartz
3∼13–∼30Dense (nonporous) sandstone with 18–32 wt% coesite, traces of stishovite, up to 20 wt% glass, and 45–80 wt% quartz
4>∼30Dense (nonporous) sandstone with 10–30 wt% coesite, 20–75 wt% glass, and 15–45 wt% quartz
5 Vesicular (pumiceous) rock with 0–5 wt% coesite, 80–100 wt% glass (lechatelierite), and up to 15 wt% quartz

These earlier shock pressure classification schemes are not based on shock recovery experiments; instead, pressures were estimated by interpretation of changes in the Hugoniot curve of Coconino sandstone and its deviation from the Hugoniot curve of quartz (Kieffer 1971; Kieffer et al. 1976). Previous shock recovery experiments on porous and wet quartz-bearing rocks are rare and were not part of a continuous series with increasing shock pressure, and did not provide comprehensive textural/petrographic analysis. For example, Hiltl et al. (1999, 2000) only investigated the grain size variations in shocked porous and wet sandstone as a function of shock pressure.

As part of the MEMIN (Multidisciplinary Experimental and Modelling Impact Research Network) research program (Kenkmann et al. 2011, 2013; Poelchau et al. 2013) this study is focused on shock recovery experiments to investigate the progressive shock metamorphism of porous sandstone in the low pressure range from 5 to 12.5 GPa. This pressure range is important for the recognition of impact structures in quartz-bearing porous targets because most of the shocked material within these craters belongs to this low shock pressure regime. In particular in eroded impact structures, where the crater fill has been largely or even completely removed, only low shock-deformed material remains for petrographic investigations. Another goal of this study is to work toward an improved calibration and shock classification system for porous, quartz-bearing sandstones, in this low pressure regime.

Methodology and Sample Material

Analytical Methods

All analytical work was carried out at the Museum für Naturkunde Berlin. The shock-deformed sandstone specimens were studied with a Zeiss Axioskop polarizing microscope, using high-quality polished thin sections with a thickness of 30 μm. The orientation of planar deformation features (PDF) was measured with a four axial Universal Stage (Leitz) and indexed by a method described in Stöffler and Langenhorst (1994).

Scanning electron microscopy (SEM) was performed with a JEOL JSM-6610LV instrument equipped with a LaB6-cathode and a BRUKER Quantax 800 energy-dispersive X-ray spectrometry (EDX) system. The EDX system was used for identification and qualitative analysis of minerals. Quantitative mineral analysis and some imaging were carried out with a JEOL JXA-8500F electron microprobe equipped with five wavelength-dispersive spectrometers, an EDX system, and a field-emission cathode for enhanced focusing of the electron beam. Microchemical analysis was based on a range of Smithsonian and Astimex standards. Detection limits (all given in ppm) were determined for Si at approximately 300, Fe approximately 1400, Mg approximately 300, Al approximately 400, Na approximately 300, Ti approximately 700, S approximately 500, K approximately 300, and Ca at approximately 300.

Porosity was determined on SEM and microprobe backscattered electron (BSE) images using the image analysis software JMicroVision. Additional porosity measurements on bulk sandstone specimens of approximately 1.5 cm3 size by the Archimedes method using purified water as liquid were carried out with a METTLER-TOLEDO AT 261 DeltaRange balance equipped with a density kit.

Micro-Raman spectroscopy was performed with an edge-filter-based DILOR LabRam instrument operating with an integrated HeNe-Laser (632.8 nm) and employed to identify diaplectic quartz glass and other SiO2 phases (Ferrière et al. 2009; McMillan et al. 1992). Raman spectra were collected from 100 to 1150 cm−1 wavelength; an energy of 3 mW on a 2.5 μm spot size was used. Final spectra represent accumulations of 10 single spectra recorded with a collection time of 10–20 s each. An optical lens of 100x magnification and the LapSpec 5.0 software were utilized.

Whole-rock geochemical analysis of the sandstone was carried out with X-ray fluorescence spectrometry (XRF) on a BRUKER AXS S8 Tiger instrument. Major elements were measured on glass tablets using an analytical program based on 50 international rock standards. For loss on ignition (LOI) determination about 1 g of pulverized sample material dried for 4 h at 105 °C was used. The sample material was heated for 4 h at 1000 °C, and the LOI was calculated using the weight difference before and after heating.

Target Material—Seeberger Sandstein

The sample material for the shock recovery experiments is a mostly beige but streaky brownish, well-sorted sandstone that is quarried at Seeberg near Gotha in Thuringia, Germany, and accordingly known as Seeberger Sandstein (Seidel 2003; and references therein; Stück et al. 2011). The sandstone unit contains different banks, of which bank 5 was used for the current experiments. Our samples from bank 5 match those used in the MEMIN feasibility study (Kenkmann et al. 2011), and are similar to the sandstone cubes quarried in bank 3, which have been the target material for all further MEMIN experiments (Kenkmann et al. 2013; Poelchau et al. 2013). Macroscopically, bank 5 consists of fine-grained sandstone displaying a lamination at a spacing between 2 and 15 mm, which is mostly defined by weathered, brown-stained (obviously quite iron-rich) layers and laminations (Fig. 1a). Cylinders with a diameter of 15 mm were drilled out of a sandstone block, with cylinder long axes lying in the lamination plane.

Figure 1.

 Unshocked Seeberger sandstone, bank 5. a) Macroscopic view of the well-sorted, finely laminated sandstone. b) Microphotograph in plane polarized light. c) BSE image displaying quartz (medium gray) with a selvedge filled with phyllosilicate (dark gray rims), pores (black), and the accessory minerals ilmenite (altered, upper left corner), titanite (upper right corner), and zircon (lower left corner). d) Backscattered electron (BSE) image of a single pore filled with phyllosilicate and subrounded goethite/limonite particles (light phases).

Microscopically, bank 5 sandstone is an arenitic quartz sandstone with a mean grain size of 0.17 ± 0.01 mm (Fig. 1b; Kenkmann et al. 2011). Most quartz grains are subrounded. Besides quartz, accessory minerals include ilmenite (altered), phyllosilicates, goethite/limonite, biotite, titanite, rutile, zircon, feldspar, and pyrite (Fig. 1c). The sandstone has an average porosity of 12.2 and 19.2 vol%, as estimated on backscattered electron (BSE) images and by measurements using the Archimedes method, respectively. This difference in porosity results from the two analytical approaches: image analysis was done on high-magnification images that only represent small selected areas of heterogeneous sandstone with a pore space of highly variable pore sizes (approximately 10–100 μm). In addition, there are experimental uncertainties on the results of the Archimedes method (e.g., degree of water and the relatively small bulk samples of approximately 3.5 cm3 may also not be representative for the heterogeneous sandstone). Nevertheless, the measured porosities are comparable to data by Stück et al. (2011) (15.8 vol%). They are in the range given for Coconino sandstone (10–20 vol%; Kieffer et al. 1976), which is important for comparison of the results of shock deformation studies. Phyllosilicates occur as aggregates of chaotically arranged flakes in selvedges on or around quartz grains and as partial or complete fillings of pores, often in association with goethite/limonite (Fig. 1d). Phyllosilicate content as determined by image analysis is, on average, 12.6 vol%. Electron microprobe analyses of aggregates of phyllosilicate flakes indicate that the main mineral group involved is illite/montmorillonite.

The chemical composition of Seeberger sandstone, bank 5 is presented in Table 2. In keeping with the most important mineral (/-groups), quartz and phyllosilicate, the silica and alumina contents, with some minor potassium, dominate the composition.

Table 2. Chemical composition of the Seeberger sandstone, bank 5.
Major element oxideswt%Trace elementsppm
  1. XRF data; detection limits: 1.0 wt% for SiO2; 0.5 wt% for Al2O3; 0.05 wt% for Fe2O3; 0.01 wt% for TiO2, MnO, MgO, CaO, Na2O, K2O, and P2O5; 15 ppm for Zn; 10 ppm for Zr and Ba; and 5 ppm for V, Cr, Co, Ni, Rb and Sr. Standard errors: 0.5 wt% for SiO2; 0.1 wt% for Al2O3; 0.05 wt% for Fe2O3, MgO, CaO, Na2O, and K2O; 0.01 wt% for TiO2, MnO, and P2O5; 30 ppm for Ba; 20 ppm for Zn; and 5 ppm for V, Cr, Co, Ni, Rb, Sr, and Zr.

  2. *Total Fe as Fe2O3, b.d. = below detection limit, LOI = loss on ignition.

SiO296.30V<15
TiO20.13Cr13
Al2O31.50Co  b.d.
Fe2O3*0.31Ni7
MnOb.d.Zn27
MgO0.09Rb  b.d.
CaO0.04Sr13
Na2Ob.d.Zr164
K2O0.11Ba50
P2O5b.d.  
LOI0.80  
Total99.28  

Experimental Set-up

These shock recovery experiments were carried out at the Efringen-Kirchen facility of the Fraunhofer Institut für Kurzzeitdynamik (Ernst-Mach-Institut), with dry Seeberger sandstone under ambient conditions on sample cylinders of 15 mm diameter and 20 mm length. The experimental set-up consists of an explosively driven flyer plate that impacts a cylindrical ARMCO iron container, in which the sandstone cylinder is shielded (Fig. 2) (for a full description of experimental set-up see, e.g., Langenhorst and Hornemann 2005).

Figure 2.

 Experimental set-up for the shock recovery experiments. a) Complete experimental assembly with high explosives, outer steel block (momentum trap), ARMCO iron flyer plate, spacing ring of acrylic glass (diameter 64 mm, height 10 mm, wall thickness 4 mm), and inner ARMCO iron cylinder (diameter 48 mm, height 50 mm) containing the sandstone sample. b) Detailed view of flyer and driver plates with thicknesses d and D, respectively, and the sandstone sample.

Variations of shock pressure in the sandstone cylinders are achieved through different combinations of the thicknesses of the flyer and driver plates, and by the use of different explosives (see experimental parameters in Table 3). The precision of shock pressure determination in the ARMCO iron driver plate is in the order of ±4%. We used the impedance method for our experiments to achieve relatively low shock pressures. The shock pressure in the sample is determined by graphic impedance matching (e.g., Langenhorst and Hornemann 2005). In the absence of Hugoniot data for Seeberger sandstone above 5 GPa, we used the Hugoniot data of Coconino sandstone (Ahrens and Gregson 1964; compiled by Stöffler [1982] with additional data from Shipman et al. [1971]) for shock pressure determination. This is thought to lead to an additional error estimated at about 1–2 GPa in shock pressure with respect to the actual sample material, because the curves do not match well above 5 GPa.

Table 3. Experimental parameters and calculated data for shock recovery experiments.
Container/experiment No.1–11–21–31–4
  1. *Based on Hugoniot data for Coconino sandstone.

Sandstone cylinderDiameter 15 mm, length 20 mm
High explosiveTNTComp.BComp.BComp.B
Thickness (d) of flyer plate (mm)4443
Thickness (D) of driver plate (mm)6.515108.5
Shock pressure at the base of the ARMCO iron driver plate (GPa)18.425.033.542.5
Calculated pressure at the top of the sandstone cylinder (GPa)*5.07.510.012.5
Estimated maximum pressure-pulse duration at the top of the sandstone cylinder (μs)1.110.660.880.55

This experimental set-up allows complete recovery of shocked samples, although sandstone may have lost coherence. ARMCO containers were carefully opened with a lathe. For subsequent petrographic investigations, polished thin sections were prepared; they were cut perpendicular to the shock front, parallel to the long axis, and through the middle of the shocked sandstone cylinder.

The pressure-pulse duration for the four experiments was estimated with the help of time–distance plots for the shock and rarefraction waves (Fig. 3) using the method described in Stöffler and Langenhorst (1994). For these plots, we have used shock (U) and particle (up) velocities calculated from pressure-particle velocity graphs for ARMCO iron and Coconino sandstone. It is remarkable that the rarefraction wave catches the shock wave somewhere in the middle of the sandstone cylinder; at this depth the initial shock wave is attenuated. This applies to all four shock experiments reported here. Nevertheless, the initial shock wave also propagates through the surrounding ARMCO iron cylinder but with a much higher velocity, and it triggers nonplanar shock waves at the interface between the ARMCO iron cylinder and the sandstone sample. The maximum pressure-pulse duration (Table 3) was estimated from the time difference between the arrival time of the shock and the rarefraction waves at the upper interface between ARMCO iron and sandstone. The pressure-pulse duration decreases in the sandstone cylinder with depth, until that depth is reached where the rarefraction wave catches up with the shock wave.

Figure 3.

Time–distance plot for the shock recovery experiment at 5 GPa based on the Hugoniot curve for Coconino sandstone. The plot displays the distances of shock and rarefraction waves with time in the flyer (light gray) and driver plate (medium gray), and in the sandstone cylinder (dark gray). Note that the rarefraction wave catches the shock wave within the sandstone cylinder and, thus, cancels the shock wave. vfp = velocity of flyer plate, UFe = shock wave velocity within ARMCO iron, UFe + upFe = rarefraction wave velocity in ARMCO iron, Usst = shock wave velocity within sandstone, Usst + upsst = rarefraction wave velocity in sandstone, upsst = particle velocity in sandstone.

Results

Significant shock-induced effects in the sandstone cylinder are generated in the experiments (Table 4). These include changes in porosity, optical parameters, the generation of intra- and intergranular fractures, different kinds of melt, and the formation of diaplectic quartz glass. These observations are restricted to the uppermost one millimeter of the shocked sandstone cylinder. Due to interaction of shock and rarefraction waves traveling in the sandstone and in the surrounding ARMCO iron container, the shock pressure determination is problematic in the deeper parts of the sandstone cylinder, so that these parts were excluded from most of the investigation. However, so-called “local effects” occur also in the deeper part of the shocked cylinders, and these are discussed here as well.

Table 4. Deformation and transformation effects observed in the shocked sandstone samples.
Shock pressure*
Experiment
Unshocked
(1–0)
5 GPa
(1–1)
7.5 GPa
(1–2)
10 GPa
(1–3)
12.5 GPa
(1–4)
  1. *Calculated pressure at the top of the sandstone cylinder in contact with the driver plate; + = effect is present, (+) = effect is rarely present, further investigations necessary, − = effect not observed.

Observations     
Pores ∼19 vol%crushedcrushedcrushedcrushed
Extinction of quartz grains sharpundulatoryweak mosaicismweak mosaicismweak mosaicism
Fractures (f) ∼2 f/mm∼134 f/mm∼171 f/mm∼203 f/mm∼208 f/mm
 Large intergranular fractures and/or shear zones+++
 Intragranular fracturesextremely rare++++
Melt (glass) ∼1.6 vol%∼2.4 vol%∼4.8 vol%∼4.8 vol%
 Dark vesicular melt++++
 Light vesicular melt+++
 Melt + large iron particles+++
 Molten ARMCO iron++
Diaplectic quartz glass ++
Local shock effects      
 Planar deformation features
(PDF)
++
 Cataclastic microbreccia+++
 Diaplectic quartz glass(+)++
 SiO2 melt(+)++

Porosity

In the shock experiments all pores are entirely closed or, respectively, filled with a microbreccia, a melt (now glass), or a mixture of both (Fig. 4). Already at the lowest shock pressure of 5 GPa all pores are closed completely and discrimination of individual quartz grains on BSE images is difficult to impossible because now grain boundaries and fractures are indistinguishable from one another. Moreover, individual quartz grains are surrounded by brownish and partially opaque or isotropic regions, which strongly increase in abundance and size with increasing shock pressure. They are mostly made up of a highly vesicular melt containing molten phyllosilicates and/or molten iron oxide minerals.

Figure 4.

 Comparison of unshocked and shocked Seeberger sandstone samples with optical micrographs (transmitted light, left side, a–e) and BSE images (right side, f–j). Unshocked sandstone (a, f), and sandstone shocked at 5 GPa (b, g), 7.5 GPa (c, h), 10 GPa (d, i), and 12.5 GPa (e, j). Note the total closure of pore space as well as the difficulty to discriminate individual quartz grains even at 5 GPa on BSE images. Quartz grains are surrounded by grayish regions that increase with shock pressure in amount and width (black arrows = propagation direction of the shock front).

Optical Extinction of Quartz Grains

In the unshocked sandstone nearly all quartz grains show sharp extinction. In contrast, already at a pressure of 5 GPa quartz grains display undulatory extinction, and at pressures of 7.5 GPa and above, the quartz grains show weak but prominent mosaicism.

Fracturing Phenomena

Numerous fractures, both intragranular and intergranular, occur throughout the shocked samples. In the sample shocked to 5 GPa large intergranular fractures were not observed (Fig. 5a). At pressures of 7.5, 10, and 12.5 GPa (Figs. 5b–d) large intergranular fractures originated at the upper edges of the sample cylinders, at contacts to the ARMCO iron container, and traversed diagonally through the samples converging in the center of the sandstone cylinder. These fractures are characterized by wall displacement, obviously representing shear zones. Several other large intergranular fractures extend radially toward the centers of the cylinders. Their number increases with increasing shock pressure from 7.5 to 12.5 GPa (Figs. 5b–d). The degree of pulverization of samples and, thus, loss of material during recovery, increases with increasing shock pressure.

Figure 5.

 Cross-sections through the recovered samples cut perpendicular to the impacted surface (left sides), and optical scans produced in transmitted light of the corresponding thin section (right sides) (black arrows = propagation direction of the shock front). a) 5 GPa; b) 7.5 GPa; c) 10 GPa; and d) 12.5 GPa. Note the increasing number of intergranular fractures and the increasing loss of the shocked and pulverized material during preparation in the sample shocked at higher pressures.

At the microscale, many irregular intragranular as well as numerous subplanar intragranular fractures are present in quartz. The subplanar fractures appear as single sets oriented at approximately 50° (Fig. 6a), 0° (Fig. 6b), or 35° (Fig. 6c) to the shock front (± parallel to the surface of the sandstone specimen). They also appear as directional sets where the larger one is oriented at approximately 50° (Fig. 6d) and the shorter one at approximately 22° to the shock front, and at approximately 65° to each other. Spacings of subplanar fractures are highly variable and range from <10 to 60 μm.

Figure 6.

 Optical micrographs (transmitted light) of subplanar, intragranular fractures within single quartz grains displaying different orientations in relation to the shock front as indicated by the arrows parallel to the propagation direction of the shock front. Fracture orientation at a) about 50°, NW–SE, 7.5 GPa, b) about 0°, E–W, 5 GPa, c) about 35°, NE–SW, 12.5 GPa. d) Directional fracture sets with a larger one (orientated about 50°, NW–SE) and shorter ones (orientated at about 22°, ENE–WSW) enclosing an angle of about 65°, 5 GPa.

Backscattered electron imaging shows that in all samples all quartz grains are entirely crossed by infinitesimal fractures (microfractures) and that, therefore, some grain boundaries are no longer readily identifiable. There is a huge number of short, irregular, vermicular microfractures forming a network across the whole area, especially in the samples shocked at 7.5 GPa and higher pressures (Fig. 7a). These fractures are similar to those described by Reimold and Hörz (1986), in Hospital Hill Quartzite experimentally shocked to 8 GPa, who termed these features “shock extension fractures.”

Figure 7.

 BSE images of the sample shocked to 10 GPa. a) Network of short, irregular, vermicular microfractures in quartz. b) Microfractures occurring together with microbands (upper microband indicated by an arrow) in quartz. The smaller microfractures are oriented at about 65° (NW–SE) in relation to the shock front and seem to be displaced by larger homogeneous microbands, oriented at about 10° (NNW–SSE). c–d) Enlargements of (b). c) Microbands with internal subplanar microstructures that are oriented at about 34° (NE–SW) and about 65° (NW–SE) in relation to the shock front. d) A third internal subplanar microstructure (white bar) within the homogenous microbands is oriented at about 10° (WNW–ESE) to the shock front; black arrows = propagation direction of the shock front.

Microfractures also occur together with microbands (Fig. 7b). The smaller microfractures seem to be displaced by the larger microbands. These microbands are not open fractures but represent homogeneous, subplanar, approximately 2 mm wide features that are cut only sporadically by younger fractures, although the interspaces between them are strongly affected by microfracturing. In BSE images these homogeneous, subplanar microbands contain at least three internal subplanar microstructures (Figs. 7c and 7d) appearing darker than the surrounding crystalline quartz. This indicates that they have a lower density and, hence, are diaplectic or silica glass (Langenhorst and Deutsch 2012). Therefore, these fine microstructures show strong similarities to planar deformation features (PDF). It is noticed that they are often oriented parallel to the surrounding open fractures.

Tensional fractures were observed in all shock experiments. For example, Fig. 8a shows a quartz grain broken into two fragments due to tensional fracturing. At pressures of 7.5 GPa, additional networks of fractures are sporadically observed that form angles of 60°, respectively, 120°, to each other (Fig. 8b), and therefore have a strong similarity to Riedel shears (Katz et al. 2004).

Figure 8.

 a) Tensional fracture (about 40 μm in width), along which the whole upper part of the sample has lifted up and the large central quartz grain has been broken into two fragments (BSE image, 5 GPa). b) Two systems of microfractures forming an angle of 60°, respectively, 120° (BSE image, 7.5 GPa); (black arrows = propagation direction of the shock front).

To obtain quantitative data for the density of fractures close to the surface of the sandstone cylinders, the intersections with fractures were counted on SEM images, along three different profiles of 1000 μm length each. Averages of the three individual measurements were calculated and plotted with their standard deviations which show that the number of fractures varies considerably over short distances. However, in general, the number of fractures (f) per millimeter increases with increasing shock pressure (Fig. 9).

Figure 9.

Mean fracture densities and standard deviations versus shock pressure. The average number of fractures (f) per millimeter increases from 134 f/mm in the sandstone shocked to 5 GPa, via 171 f mm−1 (7.5 GPa) to 203 f/mm and 208 f/mm at 10 and 12.5 GPa, respectively. Note the apparent attainment of a plateau at 10 and 12.5 GPa.

Melt Provenance

Four different types of melt were generated at pores and fractures in the shocked samples, of which the first three are melts that are generated in situ and the fourth is a result of the response of the experimental assembly to shock. The following characterization is based on BSE images and semiquantitative EDX analyses.

  • 1 The first type of melt is highly vesicular and occurs in pockets. In BSE images it is much darker than the surrounding quartz. It shows a flow texture with schlieren. Sizes of the melt pockets vary from <1 to approximately 5 μm (Fig. 10a). The approximate average composition of the melt is 54 wt% SiO2, 26 wt% Al2O3, 2.1 wt% MgO, 1.7 wt% FeO, 1.3 wt% K2O, 0.4 wt% TiO2, 0.3 wt% Na2O, 0.5 wt% CaO, and 0.04 wt% SO3. This dark melt was formed in all shock experiments, but the proportion of melt increases considerably with increasing shock pressure.
  • 2 The second type of melt is identical in appearance to the first one and displays a similar chemical composition, with the exception of a typically higher iron content resulting in a comparatively light color in BSE images (Fig. 10b). This melt was not observed in the 5 GPa experiment but in all experiments at higher shock pressures. The abundance of this melt type increases with higher pressure as well.
  • 3 The third type of melt has a similar chemical composition to types (i) and (ii) and shows also a flow texture with schlieren. However, type 3 contains additional iron particles with grain sizes of 0.5–2.0 μm (Fig. 10c), and rarely displays vesicles. This melt is present only near the surface of a sample. It appears rarely in the sandstone shocked to 7.5 GPa but is more abundant in the experiments at 10 and 12.5 GPa. The iron particles probably originate from the ARMCO iron cylinder.
  • 4 The fourth type of melt (Fig. 10d) represents iron injected from the ARMCO iron driver plate into fractures at the surface of the samples shocked to 10 and 12.5 GPa. This melt is completely crystallized.
Figure 10.

 Microprobe BSE images of different types of melt generated in shocked sandstone (black arrows = propagation direction of the shock front). a) Dark, highly vesicular melt (type i) at 10 GPa. b) Comparison of dark (type i, left side) and light (type ii, right side) vesicular melt (12.5 GPa). c) Al-Fe-Mg-rich melt (type iii) with large iron particles. One feldspar grain is molten at the edge in contact to melt (gray arrow), causing enrichment of Al2O3 in this region (12.5 GPa). d) Injection of molten ARMCO iron (melt type iv) into fractures in sandstone (12.5 GPa).

Image analysis shows that the total combined amount of melt types i–iv increases from approximately 1.6 vol% in the sandstone shocked to 5 GPa to approximately 2.4 vol% (7.5 GPa), and approximately 4.8 vol% (both 10 and 12.5 GPa).

Diaplectic Quartz Glass

Isolated quartz grains at the surface of the sandstone cylinders shocked to 7.5, 10, and 12.5 GPa appear partly blurred in the optical microscope under plane polarized light (Fig. 11a); they are isotropic when viewed with crossed polarizers. Such areas are homogeneous with no fractures, although the rest of the grain is microfractured (Fig. 11b). The number of fractures within single grains increases with increasing distance from the surface but does not show any fractures in this particular area. Furthermore, there is a slight grayscale difference, especially at the margins of the homogeneous zones, between the crystalline quartz (slightly lighter) and the amorphous phase (slightly darker due to lower density, Langenhorst and Deutsch 2012). Raman spectra (Fig. 11c) indicate the presence of crystalline quartz with its well-developed peak at 463 cm−1 (spectrum a5) at the outer perimeter, followed by a spectrum (a4) with this peak shifted to 458 cm−1, which is characteristic for shock-deformed crystalline quartz (Ferrière et al. 2009). In spectrum (a1) this particular peak is absent, but several smaller peaks occur (e.g., at 495 cm−1, and 604 cm−1) that are characteristic for diaplectic quartz glass (Ferrière et al. 2009). The Raman spectra clearly document a transition from totally crystalline quartz over deformed crystalline quartz to diaplectic quartz glass with decreasing distance to the sample surface. This phenomenon could be observed in quartz of the samples shocked to 10 and 12.5 Gpa, and a transition from unshocked quartz to only deformed quartz was noted in quartz grains of the sample shocked to 7.5 Gpa.

Figure 11.

 Formation of diaplectic quartz glass in the sandstone shocked to 12.5 GPa (black arrows = propagation direction of the shock front). a) Transition of quartz to diaplectic quartz glass in a single grain (optical micrograph, plane polarized light); a1–a5 annotations correspond to Raman measurements displayed in (c). b) BSE image of the area shown in (a) with diaplectic quartz glass (homogeneous zone without fractures) close to the upper sample surface. c) Raman spectra of crystalline quartz (a5), shocked quartz (a4), and diaplectic quartz glass (a1) demonstrating the transition from crystalline quartz to diaplectic quartz glass with decreasing distance to the upper sample surface.

Localized Shock Effects

In addition to the above described deformation features, several shock-induced local phenomena are distributed locally within the entire sandstone cylinder (e.g., Fig. 12a). They include PDF, cataclastic microbreccia, the presence of diaplectic quartz glass, and SiO2 melt formation. Most of these features develop exclusively in the vicinity of large intergranular fractures or shear zones (described in the Fracturing Phenomena section).

Figure 12.

 a) Sketch of the distribution of localized deformation effects in the sandstone sample shocked to 10 GPa. Note that most of these effects are restricted to large curved intergranular fractures and shear zones (black arrows = propagation direction of the shock front). b) Three quartz grains with one or two sets of PDF (optical micrograph; transmitted light). Orientations in relation to the shock front are approximately 58° (NW–SE) in the upper left grain (PDF parallel to (10i3)), about 45° (NE–SW) in the upper right grain (PDF parallel to (10i2), and about 60° (NNE–SSW) and 70° in the upper and lower parts, respectively, of the lower middle grain (PDF parallel to (10i3). c) Quartz grain with two sets of PDF, the first one orientated ±90° (N–S) to the shock front, and the second about 52° (NNE–SSW) to the shock front (BSE image). d) Quartz grain with two perpendicular sets of PDF orientated at about 62° (NE–SW) and about 43° (NW–SE) in relation to the shock front (BSE image).

Figure 12a shows exemplary the occurrence of localized shock effects in the sandstone cylinder shocked at 10 GPa. Relatively close to the initiation points of shear zones, starting from the upper edges of the sample cylinders in contact with the ARMCO iron container, half a dozen quartz grains display PDF parallel to (10inline image3) and (10inline image2) with spacings of <1 to approximately 3 μm (Fig. 12b). Interestingly, they are slightly curved in direct contact to the shear zone. BSE images (Figs. 12c and 12d) show quartz grains with two sets of PDF. In Fig. 12d both sets are nearly perpendicular to each other. Both BSE images show a slightly curved nature of the subplanar features in some areas near the contact to the shear zone, which is not typical for PDF, but has been observed occasionally in nature (Trepmann and Spray 2005).

The shear zones locally show a zonation. The outer part consists of a cataclastic microbreccia, followed by a zone of partly fused quartz grains, and a SiO2-rich melt in the center. In many parts of the shear zones this sequence is incomplete. Figure 13a displays microbrecciated quartz at the rim of such a shear zone and a SiO2-rich melt in the center including slightly elongated vesicles indicative of lateral movement. There are also schlieren visible in the melt due to material contrast indicating inhomogeneous mixing of the different molten minerals of the sandstone. Figure 13b shows a melt vein filled with a fine-grained SiO2-rich matrix including elongated vesicles and schlieren of molten rutile and iron minerals that are derived from the adjacent areas. At the rim of the melt vein adjacent quartz grains are partly molten. These localized SiO2-rich melts consist—in contrast to the previously described melts (see the Melt Provenance section)—more or less of pure SiO2 and display significantly fewer vesicles. The localized SiO2 melts occur at shock pressures of 10 and 12.5 GPa, exclusively within intergranular fractures of variable size, and at their intersections in the central part of a sample where larger melt pockets were generated. In addition, some quartz grains within these intergranular fractures show, as confirmed by Raman spectroscopy, a transition to diaplectic quartz glass.

Figure 13.

 a) SiO2-rich melt with flow texture containing elongated vesicles, partially molten quartz, and cataclastic microbreccia at the margins of a wide shear zone (10 GPa, BSE image). b) A melt vein trending at NW–SE, which is filled with a fine-grained SiO2-rich matrix including schlieren and droplets of molten rutile and iron minerals (lighter phases), elongated vesicles, and partly molten quartz grains along the edges of the vein (12.5 GPa, BSE image); (black arrows = propagation direction of the shock front).

Discussion

Influence of the Experimental Set-Up

Although the time–distance plots for all four shock experiments suggest that the initial shock wave does not reach the bottom of the sandstone cylinders, the entire samples have undergone shock deformation. For example, microscopic and BSE observations of the entire sandstone cylinders have demonstrated a total closure of pores and formation of melt and fractures. This could be explained by plastic waves, unloading after propagation of the initial shock pulse, or additional spherical shock waves that originate from the ARMCO iron-sandstone interface at the sandstone cylinder jacket. The shock wave propagates much faster through the ARMCO iron and releases, therefore, such spherical shock waves at every point of the interface before the initial planar shock wave reaches these points.

In addition, the shock pressure caused by the initial planar shock wave generally decreases significantly with depth due to pore crushing that consumes energy from the shock wave and attenuates shock pressure (Güldemeister et al. 2013). Therefore, we can use only the uppermost part of the shocked sandstone cylinders for the investigation of regular (i.e., compression/decompression-related) shock effects. The spherical shock waves originating from the cylinder jacket were also reflected at the bottom of the sandstone cylinder at contact to ARMCO iron, leading to further interferences in the lower parts of the sample cylinders until the arrival of the rarefraction wave. This causes strong fragmentation and at higher shock pressures pulverization (cf. Fig. 5).

Another effect appears at the upper edge of the sandstone cylinders. In these ring-like areas the shock wave propagating from the ARMCO iron into the sandstone is reflected both forward and backward several times, so that pressure peaks are generated. Also shear is stronger in these areas due to the large difference in shock impedance between ARMCO iron and the sandstone, which enhances both pressure and postshock temperature.

In addition to the effect of shock pressure attenuation in lower parts of the sample cylinders referred to above, the interaction of the initial planar shock wave in the sandstone cylinder with additional spherical shock waves originating from the cylinder jacket makes shock pressure determination for the middle and lower parts of the sample cylinder difficult. Generally, the experimental set-up cannot generate ideal uniaxial compression. Moreover, the set-up leads to a slight lateral extension of the sample cylinder. This extension leads to the formation of large, nonplanar, intergranular fractures observed in all experiments (Figs. 5 and 12a). Also, this lateral extension has caused the formation of fractures in single quartz grains, which are oriented at approximately 45° to the shock front (Figs. 6a and 6c).

Progressive Shock Metamorphism

Shock compression of porous sandstone is distinctly different from that of nonporous rocks, especially at low shock pressures. This is obvious from the difference between the Hugoniot curve of quartz or quartzite and sandstone (e.g., Kieffer 1971; Kieffer et al. 1976; Güldemeister et al. 2013), which for the low pressure regime strongly depends on porosity. The large contrast in the shock impedances of quartz and pores leads to a distinctly heterogeneous distribution of shock pressures and temperatures until the pores are completely closed. This causes heterogeneous distribution of shock features at the microscopic scale, as observed in nature (e.g., Kieffer 1971; Kieffer et al. 1976; Grieve et al. 1996; Osinski 2007) and in experimentally shocked samples (e.g., Hiltl et al. 1999, 2000; this work).

The shocked Seeberger sandstone samples display progressive shock metamorphism with increasing shock pressure as indicated by decreasing pore volume, characteristic changes in optical extinction of quartz, increased fracturing, increased melting, and the formation of diaplectic quartz glass. For pressure calibration an average shock pressure is used in this work, which is based on the Hugoniot curve of porous Coconino sandstone. This average shock pressure neglects the strong differences in shock pressure between the different minerals, and especially between mineral grains and pores.

Pores are entirely closed or filled with melt in all shock experiments. Modeling of the effects of pore crushing by Güldemeister et al. (2013) resulted in total closure of pores at a shock pressure of 6 GPa—in excellent agreement with our observation that at 5 GPa pore space had been experimentally closed completely. The amount and the width of brownish and partially opaque regions (also described by Kieffer [1971] in naturally shocked Coconino sandstone) that surround single quartz grains and represent fully or partially molten areas strongly increases with increasing shock pressure due to pore collapse and the resultant high temperatures. Calculations of such pressure peaks by Güldemeister et al. (2013) show that shock pressures at pores locally can be two to four times higher than elsewhere in the same sample.

The extinction of quartz grains changes from mainly sharp in the unshocked sandstone over undulatory extinction in the sandstone shocked to 5 GPa to weak but still prominent mosaicism at 7.5 GPa and above. This observation ought to be checked against naturally shocked sandstone samples.

The number of all kinds of fractures (intergranular in sandstone and intragranular in individual quartz grains) also increases with higher shock pressure. At a pressure of 10 GPa, a saturation of fracture density is reached. Above 10 GPa fracturing is most likely replaced by melting due to higher shock pressures and significantly higher postshock temperatures. The orientation of both types of fractures depends on the location of a quartz grain within the sample and its orientation to the shock front. Within individual quartz grains the sequence of features produced is (1) formation of short microfractures (Fig. 7a), (2) microbands containing PDF (Fig. 7b), and, finally (3) extension fractures (Fig. 8) that originate at the surface of all samples.

The combined amount of all kinds of melt increases with increasing shock pressure, due to enhanced postshock temperatures, up to 10 GPa. The ternary MgO + K2O + Na2O − Al2O3 − FeO diagram (Fig. 14) shows that the analyzed unshocked phyllosilicate and the dark vesicular melt both plot into the Al2O3-rich area around the montmorillonite locus. The lighter melt phases are mixtures of Al2O3 and FeO, i.e., of phyllosilicate and iron oxide minerals. Figure 14 demonstrates that the dark vesicular melts, the light vesicular melt, as well as the melt containing the large iron particles, can all be derived from the melting of phyllosilicates and phyllosilicate-limonite/goethite-mixtures. The dark and light vesicular melt most likely were generated in situ, due to the high temperatures associated with pore collapse, at so-called “hot spots.” Both vesicular melt types occur already at low shock pressures, the dark vesicular melt as of 5 GPa, and the light vesicular melt as of 7.5 GPa. Both types of melt fill mainly pores and display mobilization only over short distances into their surroundings.

Figure 14.

Ternary MgO + K2O + Na2O − Al2O3 − FeO-diagram displaying the chemical compositions of phyllosilicates and phyllosilicate-goethite/limonite-mixtures of unshocked Seeberger sandstone, selected phyllosilicate minerals (calculated from their stoichiometric formulae), and melt compositions from the sample shocked at 10 GPa. Note that the iron-rich unshocked phyllosilicate likely represents analyses of mixtures of montmorillonitic phyllosilicate + Fe-oxides/hydroxides.

At higher shock pressures, additional melt with large iron particles occurs, which is injected into already existing fractures. It is reasonable to assume that the iron is derived from the driver plate. The enrichment of melt with molten iron, in contrast to the dark, vesicular melt, and its increased mobility with increasing shock pressure can be related to the higher shock temperatures caused by initial higher shock pressure. At shock pressures above 10 GPa, nearly all the phyllosilicates and phyllosilicate-goethite/limonite mixtures of the sandstone are molten. Therefore, the amount of melt reaches a plateau until the quartz portions of the sandstone begin to melt. The temperatures generated by shock pressures up to 12.5 Gpa do not cause melting of quartz within the entire sandstone sample; this would require temperatures above 1610 °C (Weast 1976) that are only reached locally at shear zones (see below).

The melting of ARMCO iron at the top of the container and the cylinder jacket is most likely caused by small cavities and the collapse of pores located at or close to this interface, as well as plastic work. These features lead to shock pressure and temperature peaks that can cause the melting of ARMCO iron within the contact area to the sandstone sample.

Diaplectic quartz glass could be observed locally in the samples shocked to 10 and 12.5 GPa, near the top of the sandstone cylinders where highest shock pressure is experienced. Therefore, the formation of diaplectic quartz glass within the upper part of these shocked sandstone cylinders is interpreted as a real pressure-dependent shock feature. The shock pressure necessary for the formation of diaplectic quartz glass in these porous samples is obviously much lower than that produced in shock experiments with quartz single crystals and quartzite (e.g., Langenhorst and Deutsch 1994; Stöffler and Langenhorst 1994; Grieve et al. 1996), but locally, due to pore crushing, pressure peaks are generated reaching up to 2 to 4 times of the initial shock pressure and also much higher temperatures. A strong temperature dependency for the formation of diaplectic quartz glass has also been recognized in shock experiments with preheated quartz, quartzite, and granite samples (Langenhorst and Deutsch 1994; Huffman et al. 1993; Huffman and Reimold 1996).

Localized Shock Effects

The localized shock effects are not a direct consequence of the initial shock compression, but rather a corollary of the experimental set-up. Nevertheless, the appearance of these localized shock effects also increases with increasing shock pressure.

PDF observed in the samples shocked to 10 and 12.5 GPa are restricted exclusively to the upper edges of the sandstone cylinders. Their crystallographic orientations indicate that the sandstone in these areas is moderately (>10 GPa) to strongly (>20 GPa) shocked (Stöffler and Langenhorst 1994). Locally, pressure peaks produced by multiple reflections of the shock wave at the container-sample interface occur. Therefore, exact pressure determination within these regions is difficult. The slightly curved nature of some of these PDF is probably the result of (1) a rotation of the individual quartz grain during their formation, which is caused by the simultaneous formation of the large shear zone from the edges of the sandstone cylinder, or (2) a later offset of the individual quartz grain caused by movement along the large shear zone.

The large nonplanar intergranular fractures and shear zones host microbreccia, diaplectic quartz glass, and SiO2 melt. Cataclastic microbreccias are generated within these shear zones due to high mechanical strain. They are, therefore, only indirectly related to shock pressure. Diaplectic quartz glass and SiO2 melt occur within these shear zones in the 10 and 12.5 GPa samples. Both features could be the result of significantly higher temperatures within these shear zones in comparison to the surrounding sandstone. These high temperatures originate from the combination of shock heating and additional frictional heating effective only in the shear zones. Indicators for extremely high temperatures, locally occurring within the shear zones, are SiO2 melts including schlieren of molten titania after rutile, which has a melting point of 1835–1840 °C (Weast 1976).

Shock Classification

Based on the shock classification system for porous sandstone, our experimentally shocked samples belong to shock stages 1b (5 GPa) and 2 (7.5, 10, 12.5 GPa) (Table 1) (Kieffer 1971; Kieffer et al. 1976). For these shock stages average shock pressures of approximately 3–5.5 GPa and approximately 5.5–13 GPa were estimated by these authors. Our 5 GPa experiment shows a near-complete closure of pore space, which is the basic criterion for shock stage 1b, thus confirming the original pressure calibration. Nevertheless, the common occurrence of PDF and up to 10 vol% of quartz glass, and small amounts of high-pressure phases, observed in shock stage 2 samples of Coconino sandstone (Kieffer 1971; Kieffer et al. 1976), are not seen in our 7.5 to 12.5 GPa samples. In contrast, only a small amount of diaplectic quartz glass was noted in our 10 and 12.5 GPa samples. This discrepancy, especially the absence of the high-pressure phases, between the shock experiments and the naturally shocked Coconino sandstone samples might be an effect related to differences in pressure-pulse duration and postshock temperature between experiment and nature. Nevertheless, a first step for the calibration and improvement of the shock classification system for porous sandstone is achieved with these new shock experiments.

Conclusion

The numerous local effects, caused by the experimental set-up, include formation of PDF and shear zones associated with cataclastic microbreccias, diaplectic quartz glass, and SiO2 melt.

Our first shock experiments with dry, porous Seeberger sandstone have produced shock features in quartz as known from naturally shocked porous sandstone and shock experiments with quartz single crystals and quartzite at low shock pressures (e.g., Kieffer 1971; Kieffer et al. 1976; Reimold and Hörz 1986; Langenhorst and Deutsch 1994). The progressive shock deformation stages for the Seeberger sandstone are:

  • 1 At 5 GPa, pores are totally closed or filled with melt and quartz grains show undulatory extinction. On average, 134 fractures per mm are observed. Dark vesicular melt occurs and its amount is, on average, 1.6 vol%.
  • 2 At 7.5 GPa, quartz grains show weak mosaicism and the number of fractures per mm increases up to 171. Two additional kinds of melt, a lighter, vesicular melt and a melt containing large iron particles, occur. The volume of all types of melt amounts to 2.4 vol%.
  • 3 At 10 and 12.5 GPa, quartz grains also show weak mosaicism and the number of fractures per millimeter increases to a mean value of 200. The volume of melt averages at 4.8 vol%. Diaplectic quartz glass is present.

In summary, effects exclusively characteristic for shock loading are rare, and are at the moment restricted to the formation of diaplectic quartz glass at shock pressures of 10 GPa and above.

Acknowledgments— This work is supported by Deutsche Forschungsgemeinschaft (DFG) grants FOR 887 and Re 528/8-1 and 8-2. High-quality polished thin sections of the shocked samples were prepared by U. Heitmann, WWU Münster. The Raman investigations were carried out with the expert assistance of J. Fritz. Modeling of the experimental set-up was carried out by Nicole Güldemeister, which led to many fruitful discussions. We appreciate technical assistance for sample and container preparation and the subsequent investigations of the shocked samples by K. Born, P. Czaja, A. Ueno, H.-R. Knöfler, H. Schneider, K. Krahn, C. Radke, A. Yener, and H. Götz. We want to thank the reviewers, F. Langenhorst and G. Osinski, and the editor A. Deutsch for their very constructive comments that resulted in significant improvement of the manuscript.

Editorial Handling— Dr. Alex Deutsch

Ancillary