Preserved glass-rich impactites on Mars

Quenched glass formed by hypervelocity impacts can encapsulate and preserve biosignatures on Earth, demonstrating the fossilization potential of glass-rich impactites on Mars. However, definitive spectral signatures of impact glass have not been identified on the martian surface from orbital remote sensing. Here we present a remote compositional survey of probable impactites in well-preserved craters, using data from the Compact Reconnaissance Imaging Spectrometer for Mars. These units are composed of mafic glasses mixed with crystalline phases including olivine and pyroxene, determined by radiative transfer Hapke modeling followed by spectral mixture analysis. This glassy material likely formed from impactinduced melting of the target rock with rapid quenching and minor subsequent devitrification or chemical alteration. The metastable glass has been preserved by the cold and dry martian climate during the Amazonian period, and this preservation—as confirmed here across the planet—provides a means to trap signs of ancient life on the accessible martian surface. Our results lend concrete support to theoretical arguments suggesting that impact glass has formed in abundance on Mars, both inside of craters and as spherules in distal strewnfields. Contrary to previous ideas, martian impact products are not destroyed by interaction with volatiles during the impact process. INTRODUCTION Hypervelocity impacts of asteroids and comets melt target rocks on planetary surfaces; this is a fundamental process in the solar system (e.g., Melosh, 1989; French, 1998). Resulting silicate liquids mix with pulverized, unmelted, shocked rock to various degrees, cooling to form impactites including breccias and impact glass (Stöffler and Grieve, 1994). The liquid component of these mixtures rapidly quenches if exposed to cold air and rock, and quenching is likely enhanced on Mars because of differential scaling between impact melt and crater volumes (Grieve and Cintala, 1997), a colder surface environment, and volatile-rich sedimentary targets that limit the extent of thick clast-poor melt sheets (Kieffer and Simonds, 1980; Pope et al., 2006; Boyce et al., 2012; Tornabene et al., 2012). On Mars, impactites should have mafic to ultramafic compositions, and it is expected that copious melt products formed from 4.5 Ga to 3.0 Ga when impact rates were much higher than at present (Newsom, 1980; Lorenz, 2000; Schultz and Mustard, 2004; Wrobel and Schultz, 2007; Schultz and Wrobel, 2012). Possible glassy impact spherules have been imaged by landed campaigns (Minitti et al., 2013), and martian impact melts and breccias have been identified from orbit based on their morphology (Tornabene et al., 2010; Osinski et al., 2011; Skok et al., 2012). However, the spectral signature of glass has not been uniquely identified in these materials. Identifying a quenched glass component would be significant because these amorphous phases can preserve biosignatures (Howard et al., 2013; Schultz et al., 2014) and serve as a substrate for microbial life (Izawa et al., 2010; Sapers et al., 2014). Partially glassy impactites on Mars may have been strongly eroded or chemically altered (Newsom, 1980; Tornabene et al., 2013), or simply may not have been detected from orbit because of the subtle spectral signature of mafic glasses at visible and near-infrared (VNIR) wavelengths. METHODS We investigated VNIR spectral signatures of geologic units inside impact craters across Mars. Targeted units are located in well-preserved craters, and we focused on those with excellent spectral exposures (i.e., low dust cover from manual inspection of spectral data) and strong geomorphic evidence for an impact origin. Some of the criteria used to identify these impactites included high nighttime temperatures, smooth textures, and entrained breccia blocks visible with orbital imagery from the High Resolution Imaging Science Experiment (HiRISE; McEwen et al., 2007). These deposits are commonly associated with crater central uplifts, and they show evidence for draping topography, ponding, and flow features. In some cases these specific geologic units have been previously mapped as impact melt rocks or melt breccias (e.g., Tornabene et al., 2010; Osinski et al., 2011), especially those at Alga Crater (Skok et al., 2012), Ritchey Crater (Marzo et al., 2010; Sun and Milliken, 2014), and Toro Crater (Marzo et al., 2010). Without petrographic analysis it is difficult to confidently classify these units (i.e., impact melts versus suevites; Stöffler and Grieve, 1994) so the non-specific “impactite” is used here, although impact breccias can be recognized with HiRISE imagery (e.g., Tornabene et al., 2012). We determined the mineralogical composition of the impactite units using reflectance data from the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM; Murchie et al., 2007) onboard the Mars Reconnaissance Orbiter spacecraft. Many primary and secondary minerals have been identified with CRISM data based on the presence of their diagnostic absorption features (e.g., Mustard et al., 2008; Skok et al., 2012), but not mafic glasses. The spectral signature of these glasses at VNIR wavelengths is broadly similar to that of other Fe-bearing phases like pyroxene and olivine: they have wide crystal field absorptions near 1.15 mm and 2.0 mm, and unaltered glass has a positively sloped spectral continuum. The presence of glass in a VNIR spectrum can be obscured when pyroxene and olivine are also present because of highly nonlinear mixing behavior; therefore, we used the Hapke radiative transfer model (Hapke, 1981) to account for nonlinearity by converting CRISM data to single-scattering albedo. We then unmixed spectra of the impactite units using a set of spectral end members to determine the presence of mafic glass based on its diagnostic spectral shape (see the GSA Data Repository1 for additional details on the modeling). Model outputs are relative spectral fractions of each spectral end member, including mafic glass. We mapped the spatial distribution of modeled glass, then targeted apparent glass-rich areas to investigate their spectral signatures and their detailed morphology with HiRISE imagery. Our methods differ from those of previous studies that have inferred the possible presence of glass (e.g., Skok et al., 2012) based qualitatively on how it modifies absorption band shapes/positions in a given spectrum. Importantly, the model can be used to spatially map the strength of the glass signature across a CRISM scene.