Impact melting in sedimentary target rocks: An assessment
Published:January 01, 2007
- PDF LinkChapter PDF
G.R. Osinski, J.G. Spray, R.A.F. Grieve, 2007. "Impact melting in sedimentary target rocks: An assessment", The Sedimentary Record of Meteorite Impacts, Kevin R. Evans, J. Wright Horton, Jr., David T. King, Jr., Jared R. Morrow
Download citation file:
Despite being present in the target sequence of ∼70% of the world's known impact structures, the response of sedimentary rocks to hypervelocity impact remains poorly understood. Of particular significance is the relative importance and role of impact melting versus decomposition in carbonate and sulfate lithologies. In this work, we review experimental evidence and phase equilibria and synthesize these data with observations from studies of naturally shocked rocks from several terrestrial impact sites. Shock experiments on carbonates and sulfates currently provide contrasting and ambiguous results. Studies of naturally shocked materials indicate that impact melting is much more common in sedimentary rocks than previously thought. This is in agreement with the phase relations for calcite. A summary of the criteria for the recognition of impact melts derived from sedimentary rocks is presented, and it is hoped that this will stimulate further studies of impact structures in sedimentary target rocks. This assessment leads us to conclude that impact melting is common during hypervelocity impact into both crystalline and sedimentary rocks. However, the products are texturally and chemically distinct, which has led to much confusion in the past, particularly in terms of the recognition of impact melts derived from sedimentary rocks.
An extensive literature review indicates that sedimentary rocks are present in the target sequence of ∼70% of the world's known impact structures (Table 1) 102 103 104. Despite this large percentage, the response of sedimentary rocks to hypervelocity impact remains poorly understood. Sedimentary rocks differ from most igneous and metamorphic rocks in that they are typically rich in volatiles, be it CO2 in carbonates, SOx in evaporites, or H2O in hydrous minerals, such as clays. The high porosities of many sedimentary rocks and, thus, the potential to be water-saturated, and the ubiquitous presence of layering, further sets these lithologies apart from metamorphic and igneous rocks. It has been believed for some time that porosity, the presence of volatiles, and the heterogeneity of layered targets exert a considerable influence on the details of processes and, ultimately, the products of hypervelocity impact (e.g., Kieffer and Simonds, 1980). An understanding of the response of sedimentary rocks to impact is, therefore, needed in order to assess the effect of the cratering process on these lithologies and the potential environmental influences such impacts may have had on Earth.
For example, it is widely held that carbonates and evaporites decompose during impact and release massive amounts of CO2 (e.g., O'Keefe and Ahrens, 1989) and sulfur-bearing gases (e.g., Pope et al., 1994), respectively. Impacts into sedimentary targets may, therefore, be more environmentally damaging than impacts of the same size into igneous or metamorphic terrains. This is widely assumed to be the case with the ∼180 km diameter Chicxulub impact structure, Mexico, which impacted into a thick sequence of carbonates and evaporites (e.g., Penfield and Camargo, 1981). Chicxulub coincides with the Cretaceous-Tertiary (K-T) boundary, and quantification of the amount of CO2 and sulfur-bearing gases released during the impact process—from the decomposition of carbonates and evaporites, respectively—is required to assess their role in the resultant K-T mass extinction event (Alvarez et al., 1980). It is notable that the relative climatic importance of CO2 and sulfur-bearing gases (e.g., SO2 and SO3) in the wake of the Chicxulub impact event is still actively debated (Pierazzo et al., 1998).
One of the outstanding questions in impact cratering studies is the relative importance and role of impact melting versus decomposition for impacts into sedimentary rocks. This question cannot be completely addressed through experimentation in the laboratory, which is limited to impact velocities generally below that required for extensive melting (Grieve and Cintala, 1992). The duration of the shock state is also much shorter in experiments than in nature. Numerical and computer-based modeling may offer some important information; however, as Pierazzo et al. (1998) note, “there is no good model for melt production from impact craters in sedimentary targets.” Very few detailed, systematic studies of naturally shocked sedimentary rocks have been carried out to date, yet such studies offer the only true ground-truth data on the response of sedimentary rocks to impact. The aim of this paper is to provide an up-to-date assessment of the importance of impact melting versus decomposition during impacts into sedimentary target rocks, based on studies carried out by the authors and a review of the existing literature.
PHYSICS OF IMPACT MELT GENERATION
Theoretical considerations of the impact process have revealed important results regarding the generation of impact melt:
It is widely understood that impact melting occurs upon decompression. Shock compression deposits a large amount of energy in the target, which remains as heat following decompression. If the shock is strong enough and sufficient heat remains, the released material may be left in the form of a liquid (i.e., melt) or vapor (Melosh, 1989, p. 42–44).
A large amount of compression and shock heating occurs in porous target rocks (Kieffer, 1971; Ahrens and Cole, 1974). High porosity significantly increases the amount of pressure-volume work in the target rocks resulting from the shock wave, which results in greater amounts of postshock waste heat, raising temperatures. However, the crushing of pore space reduces the overall shock pressures in the target.
The volumes of target material shocked to pressures sufficient for melting are not significantly different in sedimentary or crystalline rocks (Kieffer and Simonds, 1980).
Model calculations indicate that both wet and dry sedimentary rocks yield greater volumes of melt on impact than crystalline targets (Kieffer and Simonds, 1980).
Thus, impacts into sedimentary targets should produce as much melt as do impacts into crystalline targets. However, it has been a generally accepted observation that impact melt rocks are not generated in impact structures formed in sedimentary targets (e.g., Dressler and Reimold, 2001). This anomaly has been attributed to the formation and expansion of enormous quantities of sediment-derived vapor (e.g., H2O, CO2, SO2), resulting in an unusually wide dispersion of shock-melted sedimentary rocks (Kieffer and Simonds, 1980).
METEORITE IMPACT INTO CARBONATES
Carbonates are present in the target rocks of approximately one third of the world's known impact structures (Table 1) 102 103 104. Despite the many uncertainties regarding the response of carbonates to impact, it is commonly accepted that these lithologies decompose after pressure release due to high residual temperatures and that subsequent fast back-reactions trap a significant part of the gaseous species (Kieffer and Simonds, 1980; Martinez et al., 1994a; Agrinier et al., 2001). It should be noted that decomposition is a chemical reaction in which a compound or molecule breaks down into smaller, simpler compounds, molecules, or elements. Vaporization, on the other hand, is the process of converting a substance from a liquid or solid state to the gaseous (i.e., vapor) state.
Shock Experiments and Thermodynamic Calculations
A compilation by Agrinier et al. (2001), and updated here, reveals that shock experiments provide contrasting and ambiguous results regarding the onset of decomposition of carbonates (Table 2). This may be due, in part, to differences between experimental techniques (single shock versus reverberation), and/or properties of the sample material (e.g., porosity) (Martinez et al., 1994a), and/or the duration of the shock state. To complicate matters, substantial differences exist between experimental observations and thermodynamic calculations (e.g., Martinez et al., 1994a; Agrinier et al., 2001; Skála et al., 2002, and references therein). Early experimental studies suggested that calcite undergoes significant decomposition (>10%–50%) at pressures as low as 10–20 GPa (e.g., Lange and Ahrens, 1986). However, recent shock experiments, coupled with optical and X-ray diffraction studies, suggest that decomposition of calcite and dolomite only occurs at pressures >65 GPa and >70 GPa, respectively (Gupta et al., 2002; Langenhorst et al., 2000; Skála et al., 2002). Furthermore, recent dynamic loading and fast unloading experiments have produced complete shock melting of CaCO3 at pressures of ∼25 GPa and temperatures of ∼2700 K (Langenhorst et al., 2000).
Phase Relations of CaCO3
The phase diagram for CaCO3 has recently been reevaluated in relation to shock compression and decompression by Ivanov and Deutsch (2002), (Fig. 1). The main outcome of their work has been the extension of the liquid field of CaCO3. For example, isentropic release paths for calcite shocked to a pressure of >10 GPa first enter the liquid field, with decomposition only possible after pressure has dropped to <0.003 GPa (30 bar) at temperatures of ∼1500 K (Fig. 1). Decomposition is terminated at temperatures of <1200 K at atmospheric pressure (Ivanov and Deutsch, 2002). Thus, the phase relations of CaCO3 suggest that the expected result of hypervelocity impact into calcite is melting, with decomposition only occurring during postshock cooling. Indeed, the phase relations would suggest that CaCO3, shocked to pressures >1 GPa and temperatures >1500–2000 K, should undergo melting (Fig. 1). Ivanov and Deutsch (2002) also note that complex reaction kinetics, in particular the rate of diffusion, may limit the amount of CO2 released during impact events.
Decomposition of Carbonates During Impact: The Record in the Rocks
The decomposition of carbonates releases CO2 and produces residual solid oxides: CaO and MgO from the decomposition of calcite and dolomite, respectively. Thus, if carbonates decompose to any great degree during impact, we would expect to detect CaO and MgO in impactites within and around terrestrial impact structures. However, CaO has not been documented from any terrestrial impact structure to date (Martinez et al., 1994a; Osinski and Spray, 2001). For impacts into limestone (i.e., predominantly CaCO3), this anomaly has been attributed to subsequent rapid back-reactions of CO2 with the initially produced CaO (e.g., Martinez et al., 1994a; Agrinier et al., 2001). While such reactions are possible, it is important to note that 100% reconversion to calcite was not achieved in any of the experiments of Agrinier et al. (2001). Reconversion is also dependent on grain size, with a reduction in efficiency with increasing grain size (Agrinier et al., 2001). Small amounts of CaO may, therefore, be expected to survive. A single case of evidence for back-reactions at terrestrial impact structures has been presented. Martinez et al. (1994a) described unshocked calcite present within vesicles and holes in silicate impact glasses in naturally shocked rocks from the Haughton impact structure as evidence for back-reactions. However, the evidence presented by these authors has been reinterpreted as liquid immiscible textures between coexisting carbonate and silicate melts (Graup, 1999; Osinski and Spray, 2001). Thus, the character of some impactites from Haughton should no longer be quoted as displaying unequivocal evidence for back-reactions.
Evidence for very minor decomposition of carbonates has currently only been noted at the Ries impact structure, Germany. Baranyi (1980) noted that a <1.5 mm thick orange-brown crust or rim, comprising clays and X-ray amorphous material, surrounds 3%–10% of the limestone clasts within the Ries suevites. Baranyi (1980) inferred that the rims were formed by the thermal decomposition of carbonate clasts during postimpact contact metamorphism and that the CaO was removed by H2O circulating in the suevite: CaO is unstable in the presence of H2O, rapidly reacting to form calcium hydroxide (Ca(OH)2), which itself can dissolve in H2O to release Ca2+ and OH− ions (Chang et al., 1998). Recent studies have confirmed the findings of Baranyi (1980). In particular, it is clear that the decarbonated rims neither are cut by nor diffuse into the suevite groundmass, indicating that decomposition occurred after deposition of the suevite at temperatures of >900 °C (Osinski et al., 2004). Preliminary data, only available in abstract form, suggest the presence of similar clasts at the Chicxulub impact structure, Mexico (Deutsch et al., 2003).
Dolomite (CaMg(CO3)2) is an important component of the target stratigraphy of many terrestrial impact sites, such as Haughton. Unlike CaO, MgO is a stable mineral (i.e., periclase), which should be preserved in impactites. Indeed, Agrinier et al. (2001) note that “similar experiments with dolomite and magnesite show that residual Mg oxides do not react significantly at the 103 s timescale and may, therefore, survive as a witness of degassing in impact breccias.” To the knowledge of the authors, periclase has been documented at two terrestrial impact sites. However, it is clear, in these cases, that the this mineral was formed during contact metamorphism of carbonates, present either as clasts within a silicate impact melt layer (Clearwater Lake; Rosa, 2004) or in target rocks directly beneath a melt sheet (Manicouagan; Spray, 2006). In other words, the decomposition of carbonates at these two sites was a postshock phenomenon due to the high temperatures of juxtaposed silicate impact melt (cf. limestone at the Ries structure). For example, at Manicouagan, the carbonates beneath the melt sheet attained temperatures of >900 °C, resulting in a series of contact-metamorphic minerals: diopside + labradorite + periclase + brownmillerite + spurrite + perovskite + mayenite (Spray, 2006).
Melting of Carbonates During Impact: The Record in the Rocks
Evidence for the impact melting of carbonates has now been recognized at five terrestrial impact structures (Table 3), as well as in distal impact ejecta from undetermined source craters in West Greenland (Jones et al., 2005), and the Western Desert of Egypt (Osinski et al., 2007). Based on our observations of impactites from several impact sites, and a review of the existing literature, three main occurrences of carbonate impact melts are distinguished: (1) groundmass-forming phases within impact melt-bearing crater-fill and proximal ejecta deposits, (2) globules and irregularly shaped masses within impact glass clasts from proximal ejecta deposits, and (3) individual particles and spherules within the proximal and distal ejecta deposits. Textural and chemical evidence for the impact melting of carbonates during hypervelocity impact is provided by:
Liquid immiscible textures. Liquid immiscibility is an important mechanism of magmatic differentiation in some igneous systems (Roedder, 1978). It is important to note that liquid immiscibility sensu stricto describes the process whereby an initially homogeneous melt reaches a temperature where it can no longer exist stably and so unmixes into two liquids of markedly different composition and density (Roedder, 1978). Textural evidence for the presence of immiscible impact-generated melts has been documented at several terrestrial impact sites (Table 3). This evidence includes ocellar or emulsion textures of globules of carbonate in silicate glass, sharp menisci and budding between silicate and carbonate glasses, and deformable and coalescing carbonate spheres within silicate glass (Figs. 2A–2D). These provide unequivocal evidence for carbonates and silicate glasses being in the liquid state at the same time. However, in contrast to magmatic systems, it is possible, and is to be expected, that impact-generated melts from target rocks at different stratigraphic levels will not completely mix or homogenize. This is particularly true for melt derived from a crystalline basement with overlying sedimentary rocks. Thus, it is suggested that the term “carbonate-silicate liquid immiscibility” be avoided unless there is unequivocal evidence for the unmixing of an originally homogeneous impact melt.
Quench textures. Fragments of calcite displaying a distinctive feathery texture are common in proximal ejecta at Chicxulub (Fig. 2E) (Jones et al., 2000). These are well-understood quench crystal morphologies, which indicate rapid crystallization from a melt (Jones et al., 2000).
Carbonate spherules. Individual calcite spherules are present within a variety of impactites at several impact structures (Table 3; Figs. 2F and 2G). As noted by French (1998), the production of spherules appears to be a typical process in impact events. Carbonate spherules may be misinterpreted as secondary vesicle fillings; however, in several cases there is clear evidence indicating that these features are quenched droplets of carbonate melt (Figs. 2F and 2G). This evidence includes the presence of carbonate spherules embedded in unaltered impact glass and in samples where no other secondary carbonate minerals are present (Jones et al., 2005; Osinski and Spray, 2001).
Euhedral calcite crystals within impact glass clasts. Micrometersized euhedral crystals of calcite have been described from impact glass clasts within crater-fill impactites from the Haughton structure (Osinski et al., 2005). Their euhedral nature indicates that these crystals grew while the surrounding silicate glass was still in a fluid state.
Carbonates intergrown with CaO-MgO-rich silicates.Figure 2I shows euhedral crystals and pockets of calcite-dolomite within a groundmass of pyroxene in glassy clasts from the proximal ejecta at Meteor Crater. These textures are difficult to reconcile with an origin through alteration but are consistent with an impact melt origin for the carbonates (Osinski et al., 2003). The melting of carbonates at Meteor Crater is also supported by the unusual composition of associated pyroxenes and olivines: coexisting Ca-rich pyroxene (diopside/wollastonite) and Mg-rich olivine (forsterite) are common in carbonatitic igneous rocks (Barker, 2001; Gittins, 1989). The assemblage of Ca-Mg-rich clinopyroxene (diopside/augite) + Mg-rich olivine (forsterite) + calcite ± dolomite has also been produced in experiments using the simplified system of CaO-MgO-SiO2-CO2 (Lee et al., 2000; Lee and Wyllie, 2000).
CaO-MgO-CO2-rich glasses. Unusual silicate glasses have been documented at several terrestrial impact sites (Table 3). At the Haughton structure, groundmass-forming glasses possess extremely high MgO contents of up to ∼35 wt% (Osinski et al., 2005). The only MgO-rich target lithology at Haughton is dolomite (∼21 wt% MgO and ∼31 wt% CaO), which indicates that this lithology must have been a major component in the melt zone. It is also notable that many glasses at Haughton yield consistently low EDS (energy-dispersive spectrometry) analytical totals, typically ranging from ∼50 wt% to ∼65 wt%. Qualitative analyses suggest that the predominant volatile species in these glasses is CO2. Similar glasses are present at Meteor Crater (Figs. 3A and 3B).
Carbonate chemistry. It has become clear from work at the Ries and Haughton structures that calcite displaying textures indicative of an impact melt origin is also often distinctly different in composition than carbonates from the target rocks and postimpact hydrothermal and/or diagenetic settings, at the same structure. For example, groundmass-forming calcite in crater-fill impact melt breccias at Haughton displays higher MgO, FeO, SO3, Al2O3, and SiO2 contents than the calcite developed in sedimentary target rocks and postimpact hydrothermal products (Osinski et al., 2005). The relatively high Al2O3 and SiO2 component is particularly notable as carbonatitic (i.e., igneous) calcites are the only other known carbonates to contain elevated levels of SiO2 and Al2O3. The possibility that these anomalous compositions are due to problems during analysis has also been ruled out (Osinski, 2005a; Osinski et al., 2005) (e.g., the euhedral calcite crystals in Figure 2H contain up to 8 wt% Al2O3 and 2 wt% SiO2; no Al2O3 is present in the host glass). Experiments have shown that rapid crystallization (quenching) of a high-temperature SiO2-rich carbonate melt can produce SiO2-rich carbonates (∼3–10 wt% SiO2; Brooker, 1998). This is not to say that secondary carbonates do not occur at these sites; however, due consideration of textures and chemistry allows the discrimination of igneous hydrothermal calcite, even within individual clasts (e.g., Figs. 3C–3E).
METEORITE IMPACT INTO EVAPORITES
Despite the suggested environmental consequences of the release of sulfur species during impact events (Pierazzo et al., 1998), the response of sulfates to hypervelocity impact has been little studied. Previous work focused primarily on experimental studies and computer-based simulations, with a view to estimating the threshold for the vaporization and decomposition of sulfate minerals (Chen et al., 1994; Gupta et al., 2001; Ivanov et al., 1996; Pierazzo et al., 1998; Yang and Ahrens, 1998). It is apparent that shock experiments currently provide equivocal results, with estimates for incipient vaporization ranging from ∼32.5 ± 2.5 GPa (Gupta et al., 2001) to >85 GPa (Langenhorst et al., 2003). Preliminary attempts at producing a phase diagram and equation of state for anhydrite in relation to impact events were presented by Ivanov et al. (2004). These authors noted that non-porous anhydrite has two solid-solid polymorphic phase transitions at high pressure and should theoretically undergo melting at >80–90 GPa, with incipient decomposition at pressures of ∼60–70 GPa, for venting of gas products, and 100–110 GPa for gas products of the reaction in equilibrium. Little is known about the phase relations for porous anhydrite and gypsum, which would be more applicable for impact into natural sedimentary targets.
Evaporites, comprising predominantly gypsum (CaSO4·2H2O) and anhydrite (CaSO4), are present at only a handful of terrestrial impact craters (Table 1) 102 103 104, the majority of which are buried or eroded. The only published study available of naturally shocked sulfate rocks and minerals from a terrestrial impact structure is the recent work on the Haughton structure (Osinski and Spray, 2003). Only the main observations and conclusions will be discussed here.
Detailed SEM (scanning electron microscope) studies reveal that in addition to shock-melted calcite and silicate glass (Osinski and Spray, 2001), anhydrite also constitutes an important component of the groundmass of the crater-fill melt breccias at Haughton. Textural and chemical evidence presented in Osinski and Spray (2003) indicates that the anhydrite represents a primary impact melt phase. Evidence for this includes (1) liquid immiscible textures developed between celestite and/or anhydrite and silicate glass and/or calcite (Figs. 4B and 4C), (2) possible quench-textured groundmass-forming anhydrite, (3) the presence of carbonate melt globules within anhydrite (Fig. 4A), (4) flow textures developed between anhydrite and silicate glasses, (5) elevated (trapped) concentrations of Si, Al, Mg, and H2O in groundmass-forming anhydrite, and (6) clasts of anhydrite-quartz lithologies exhibiting evidence for (partial) shock melting (Figs. 4D–4F). Evidence from clasts of shocked anhydrite-quartz lithologies in the crater-fill deposits at Haughton suggests that anhydrite undergoes shock melting at pressures of >50–60 GPa (Figs. 4D–4F). Preliminary evidence from Chicxulub suggests that clasts of anhydrite within impact melt rocks and impact melt breccias underwent postshock annealing, melting, and possible decomposition (Deutsch et al., 2003; Langenhorst et al., 2003).
METEORITE IMPACT INTO TERRIGENOUS CLASTIC ROCKS
Terrigenous clastic sedimentary rocks, including sandstones, conglomerates, and mudrocks (i.e., siltstone, claystone, shale), are present at many impact sites (Table 1) 102 103 104. Unlike carbonates and evaporites, there has been a limited acceptance that these lithologies can melt during the impact process, although it is generally assumed that the majority of the resultant melts are dispersed due to the expansion of large amounts of H2O originally present in these lithologies (Kieffer and Simonds, 1980).
Fragments of shock-melted sandstones in ejecta were first studied in detail at Meteor Crater, Arizona (Kieffer, 1971; Kieffer et al., 1976) (Table 4). These studies revealed the dramatic effects of porosity, grain characteristics, and volatiles on the response of quartz to impact in sedimentary targets. For example, in crystalline rocks, quartz will typically be transformed to diaplectic glass at 32–50 GPa, with melting at >50–60 GPa (Grieve et al., 1996). At Meteor Crater, however, diaplectic glass is present in rocks shocked to pressures as low as ∼5.5 GPa, with whole-rock melting occurring at >30–35 GPa (Kieffer, 1971; Kieffer et al., 1976). Shock-melted sandstones have now been recognized at a number of terrestrial impact structures in a variety of different stratigraphic settings (Fig. 5; Table 4). Shock-melted shales have also been recognized from the Haughton structure (Redeker and Stöffler, 1988; Osinski et al., 2005). The high SiO2 contents of the Libyan Desert glass (>95 wt% SiO2; Weeks et al., 1984) and urengoites from Siberia (89–96 wt% SiO2; Deutsch et al., 1997) also indicate an origin through the impact melting of sandstones, although no source craters for these two glass types have been found to date.
Impact breccias at the Gosses Bluff (Milton et al., 1996) and Goat Paddock (Milton and Macdonald, 2005) impact structures, comprise a groundmass of predominantly silica glass, indicating that considerable amounts of sandstone-derived impact melts can also be preserved within structures developed in sandstone-rich targets. It has also been shown that sandstone-derived impact melts form a volumetrically minor, yet ubiquitous component of the groundmass of Ries surficial suevites (Osinski et al., 2004).
DISCUSSION AND CONCLUSIONS
It is apparent that impact melting in sedimentary targets is much more common than previously realized. In terms of carbonates, this is consistent with the recently reappraised phase relations of CaCO3 (Fig. 1), which indicate that melting is the dominant response of carbonates to hypervelocity impact. Very high postshock temperatures close to the point of impact will result in vaporization of carbonates, as is the case for all rock types; however, this is not the same as chemical decomposition. The phase relations of CaCO3 suggest that limited decomposition from CaCO3 melt may be possible following decompression, although evidence for this has not yet been observed in naturally shocked rocks. For impact into limestones, this absence of evidence may be due, in part, to the recombination of CO2and CaO during fast back-reactions (e.g., Agrinier et al., 2001). However, Ivanov and Deutsch (2002) note that complex reaction kinetics, in particular the rate of diffusion of CO2, may limit the amount of this gas released from carbonates during impact events.
It is also important to note that decomposition of CaCO3can only occur after pressure release due to high residual temperatures during postshock cooling (>1200 K; Ivanov and Deutsch, 2002). It is suggested that this thermal constraint is important in determining if, and when, decomposition can occur (cf. Deutsch et al., 2003). For example, it is notable that all silicate melt phases in the groundmass of the crater-fill deposits at Haughton were quenched to glasses (Osinski et al., 2005), indicating rapid cooling that would have inhibited the decomposition of CaCO3 melt and clasts. At the Ries impact structure, however, postimpact temperatures in the suevite deposits were sufficiently high, in some regions, to result in minor decomposition at the edge of limestone clasts (Baranyi, 1980). This is consistent with preliminary results from Chicxulub, which suggest that minor decomposition of limestone clasts occurred only in impact melt rocks and impact melt breccias, and not in the lower-temperature lithic breccias or melt-bearing breccias (Deutsch et al., 2003). Thus, this decomposition is a postimpact contact-metamorphic process, which also occurs in igneous rocks. In other words, the decomposition of limestone during the impact process is not, therefore, due to shock sensu stricto, but is governed by the postimpact temperature of the melt-clast mixture: rapid quenching and/or low post-shock temperatures will inhibit carbonate decomposition.
Despite the growing body of evidence for the melting of carbonates during hypervelocity impact, the pressure-temperature conditions at which this occurs are poorly understood. Appraisal of Figure 1 predicts that calcite, shocked to >0.1 GPa and post-shock temperatures >1500 K, should undergo impact melting during decompression (Ivanov and Deutsch, 2002) (i.e., the melting of calcite requires low pressures but high temperatures). In this respect, it is interesting to note that calcite underwent melting at >10 to <20 GPa in porous sandstones at the Haughton structure (Osinski, 2005b), corresponding to postshock temperatures of >1250 K (Kieffer, 1971). Thus, temperature is obviously the limiting factor and explains why melting of calcite did not occur in the Haughton sandstones at pressures <10 GPa (i.e., the corresponding shock temperatures were not high enough). Further studies are required to assess the pressure-temperature conditions at which calcite undergoes melting and the influence of porosity.
With respect to dolomite, the lack of periclase (MgO)—which unlike CaO is stable—in impactites suggests that the decomposition of dolomite is not an important process during impact events. Complications also arise because accurate data for the melting of dolomite are not available (i.e., there are no known binary equilibria in which dolomite does not form solid solutions [Treiman, 1989]). It is also interesting to note that primary dolomite (i.e., impact melt phases) is extremely rare in terrestrial impact structures. This is compatible with observations of carbonatites, the vast majority of which are calcitic (K. Bell et al., 1998). Indeed, the phase relations of the systems CaO-MgO-CO2-H2O (Lee et al., 2000) and CaO-MgO-SiO2-CO2-H2O (Otto and Wyllie, 1993) indicate that calcite is the liquidus phase for a wide range of compositions and pressure-temperature conditions. Thus, for impacts into dolomite-rich target rocks, a CaO-MgO-rich melt will be generated and calcite will typically be the first phase to crystallize out of the melt, with dolomite only forming at lower temperatures upon slow cooling. This is consistent with observations from the Haughton structure in which the groundmass of crater-fill impact melt breccias comprises calcite and MgO-rich glasses (Osinski et al., 2005). This also appears to be the case for impact melt breccias from Chicxulub, although the primary MgO-rich glass has been altered to MgO-rich clay (Nelson and Newsom, 2006). These observations also suggest that for impacts into carbonate-dominated targets, such as Chicxulub and Haugh-ton, calcite and MgO-rich glasses are the typical product. For target sequences richer in siliciclastic sedimentary rocks, such as at Meteor Crater, it appears that SiO2-CaO-MgO-rich melts are produced, from which a variety of silicate phases can crystallize (e.g., Ca-rich clinopyroxene and Mg-rich olivine) (Osinski et al., 2003) in addition to, or in place of, MgO-rich glasses.
To complicate matters further, it has been known for decades that natural systems are several orders of magnitude more complex than the simple one- or two-component systems used in experiments and theoretical calculations. This is the norm in natural carbonates that are typically “impure” (e.g., containing quartz, alkali feldspar, clays, etc.). Thus, impact melting of large volumes of carbonate strata will yield melts that are not pure CaCO3 or CaMg(CO3)2. To the knowledge of the authors, there are no studies that have addressed this problem with respect to decomposition. The very existence of carbonatite lava flows indicates that “impurities” play a major role in carbonate melts (Gittins and Jago, 1991). According to the phase relations of CaCO3, the existence of carbonatite lava flows is not possible. However, it has been shown that the presence of “impurities” in the melt, such as H2O (Wyllie and Tuttle, 1960) and F (Gittins and Jago, 1991), dramatically lowers the minimum melting temperature of calcite, which allows carbonate melts to be stable at atmospheric pressure at temperatures <900 °C (Gittins and Jago, 1991).
In comparison to carbonates, fewer studies have been conducted into the response of sulfates to hypervelocity impact. Studies of naturally shocked rocks and minerals from the Haughton structure suggest that the anhydrite undergoes shock melting at pressures of >50–60 GPa. This is lower than preliminary estimates of >80–90 GPa derived from the phase relations of Ivanov et al. (2004); however, this estimate is for nonporous anhydrite, and it is widely known that melting of minerals in porous targets occurs at much lower pressures (e.g., Kieffer, 1971). These results are consistent with recent shock experiments, which suggest that anhydrite does not decompose below pressures of ∼85 GPa (Langenhorst et al., 2003).
Impact-generated melt occurs in two main forms within terrestrial impact structures (Grieve and Cintala, 1992): (1) as crystallized and/or glassy coherent impact melt sheets or layers, within allochthonous crater-fill deposits, and (2) as discrete glassy clasts within so-called “suevites” in allochthonous craterfill deposits and ejecta deposits. It has long been accepted that the volume of impact melt recognized in sedimentary and in mixed sedimentary-crystalline targets is about two orders of magnitude less than for crystalline targets in comparably sized impact structures (Kieffer and Simonds, 1980). However, recent work has shown that the crater-fill deposits at Haughton are impact melt breccias or clast-rich impact melt rocks (Osinski et al., 2005) and that they are stratigraphically equivalent to coherent impact melt layers developed at craters in crystalline targets (Grieve, 1988; Osinski et al., 2005). Importantly, the present and probable original volumes (∼7 km3 and ∼22.5 km3, respectively) of the crater-fill impact melt breccias at Haughton are analogous to characteristics of coherent impact melt sheets developed in comparably sized structures formed in crystalline targets (Osinski et al., 2005). For example, the volume of crater-fill impact melt rocks and melt-bearing breccias at the 24 km diameter Boltysh and 28 km diameter Mistastin impact structures are ∼11 and 20 km3, respectively (Grieve, 1975; Grieve et al., 1987).
It should be noted that the clast content of crater-fill impact melt breccias at Haughton (up to ∼40–50 vol%; Osinski et al., 2005) is higher than in comparably sized structures developed in crystalline targets (e.g., ∼20–30 vol% at Mistastin; Grieve, 1975). Based on the work of Kieffer and Simonds (1980), Osinski et al. (2005) suggested that this observation might be explained by the effect of mixing “wet” sediments or carbonates into a melt as opposed to dry crystalline rocks. The enthalpies of H2O-bearing and carbonate systems are such that a much smaller proportion of admixed sedimentary rocks than of anhydrous crystalline rock is required to quench the melt to subsolidus temperatures (Kieffer and Simonds, 1980). Thus, a lower percentage of sedimentary rocks will be assimilated than crystalline rocks, before a melt is quenched, resulting in higher final clast contents for melts derived from impacts into sedimentary as opposed to crystalline targets (Osinski et al., 2005).
In summary, synthesizing observations from terrestrial impact structures with recent experimental results, computer simulations, and phase relations, it is clear that previous assumptions about the response of sedimentary rocks during hypervelocity impact events are inaccurate. Impact melting appears to be the dominant response of hypervelocity impact into carbonates, evaporites, and terrigenous sedimentary rocks. Limited decomposition from the melt phase may be possible following decompression if the melt remains at high temperatures long enough for this to occur. However, in the limited cases where decomposition of carbonates has been recognized, this was clearly a postimpact contact-metamorphic process, which is governed by the postimpact temperature of the melt-clast mixture. The apparent “anomaly” between the volumes of impact melt generated in sedimentary versus crystalline targets in comparably sized impact structures may, therefore, be due to difficulties in recognizing impact melts derived from sedimentary rocks.
The authors thank the many individuals and institutions that have facilitated and funded fieldwork at various terrestrial impact structures, which forms the basis for this study. This study is based, in part, on the Discussion chapter of the Ph.D. thesis of G.R.O. and was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) through research grants to J.G.S. G.R.O. was supported by Canadian Space Agency (CSA) Space Science Research Project 05P-07. Giovanni Pratesi and Adrian Jones are thanked for supplying several of the images in Figure 2. Jayanta Kumar Pati and Kai Wünnemann are thanked for their helpful and constructive reviews. Planetary and Space Science Centre contribution 54. Contribution from the Earth Sciences Sector 2006519.
The Sedimentary Record of Meteorite Impacts
- calcium carbonate
- carbonate rocks
- chemically precipitated rocks
- clastic rocks
- experimental studies
- glass materials
- hypervelocity impacts
- impact craters
- impact features
- impact melts
- phase equilibria
- sedimentary rocks
- shock metamorphism
- target rocks