Silica as a shock index in shergottites: A cathodoluminescence study

Meteoritics & Planetary Science 40, Nr 7, 967–979 (2005) Abstract available online at http://meteoritics.org

Silica as a shock index in shergottites: A cathodoluminescence study Hasnaa CHENNAOUI AOUDJEHANE,1, 2 Albert JAMBON,2* Bruno REYNARD,3 and Philippe BLANC4 1Département

de Géologie, Université Hassan II, Faculté des Sciences Casa Aïn Chock, BP 5366, Mâarif, Casablanca, Morocco MAGIE, Université Pierre et Marie Curie, CNRS UMR 7047 case 110, 4 place Jussieu, 75252 Paris Cedex 05, France 3Laboratoire des Sciences de la Terre, CNRS UMR 5570, Ecole Normale Supérieure de Lyon, 46 Allée d'Italie, 69364 Lyon Cedex 7, France 4Service MEB, Université Pierre et Marie Curie, UFR 928, 4 Place Jussieu, 75252 Paris Cedex 05, France *Corresponding author. E-mail: [email protected] 2Laboratoire

(Received 23 July 2004; revision accepted 12 April 2005)

Abstract–Silica in shergottites is a minor phase of great significance. Determining its structural state as either silica glass, quartz, cristobalite, tridymite, coesite, stishovite, or post-stishovite could provide informations about their shock history. The purpose of this work is to assess the shock intensity in shergottites using two spectroscopic methods. On a conventional polished section, a scanning electron microscope (SEM) enables us to study the cathodoluminescence (CL) of silica at variable magnification. The results were crosschecked by systematic Raman spectroscopy of the selected areas. CL spectra differ substantially from one another and enable separating stishovite, high and low pressure silica glass, quartz, and cristobalite. We studied a set of five shergottites: Northwest Africa (NWA) 480, NWA 856, Zagami, Shergotty, and Los Angeles. Stishovite is common in Shergotty, Zagami, NWA 856, and NWA 480 and absent in the studied section of Los Angeles. High-pressure glass is very common, particularly in close association with stishovite. According to the textural relationship, it may be a product of the retromorphosis (amorphization during decompression) of stishovite. Large stishovite areas result from the transformation of preexisting low-pressure silica crystals, while needles result from the high-pressure transformation of pyroxene to glass (melt) and silica. In the latter case, they are found in melt pockets and represent a small fraction of areas of overall pyroxene composition. Needles exhibit square sections of about 1 µm. Silica spots identical to those described previously as poststishovite are found in Shergotty, Zagami, NWA 480, and NWA 856. At present, the spectroscopic distinction of post-stishovite from stishovite is difficult. Post-stishovite is destroyed under the Raman beam, and CL spectra are possible mixtures of several phases (e.g., glass and post-stishovite). It is concluded that the shock intensity is highly heterogeneous, and the pressure probably exceeded 60 GPa in all shergottites studied here. INTRODUCTION The recent discovery of several specimens of Martian meteorites (SNC) in Antarctic and hot deserts has drastically increased their population to the present value of 31 (Meyer 2003). Among those, nakhlites (clinopyroxene cumulates) with six specimens are rare, and Allan Hills (ALH) 84001 (orthopyroxenite) and Chassigny (dunite) are unique. The remaining material consists of lherzolitic shergottites, picritic shergottites, and basaltic shergottites, all of which contain typical maskelynite, i.e., feldspathic glass. A major difference between shergottites and nakhlites is the significantly lower intensity of shock in the latter. Among basaltic shergottites, silica, a minor phase, is present to various extents, but its crystalline state has been specified only rarely (Sharp et al.

1999; El Goresy et al. 2000; Barrat et al. 2002; Jambon et al. 2002) probably because of the small size and difficult separation. Knowing the crystalline state of silica is important, as it should permit assessing the intensity of the shock. Among the conventional techniques used for determining the phases are X-ray diffraction and Raman spectroscopy. The latter is very convenient as it permits the characterization of domains no larger than a few microns in a polished section. In the present work, we used cathodoluminescence (CL) spectroscopy to characterize silica in shergottites at a similar scale. The feasibility of the technique and its ability to distinguish between glass, cristobalite, quartz, or stishovite is demonstrated, as all results have been crosschecked using Raman spectroscopy. Scanning electron microscopy (SEM) is

967

© The Meteoritical Society, 2005. Printed in USA.

968

H. Chennaoui Aoudjehane et al.

a standard technique for studying the petrology of meteorites; as CL spectroscopy can be performed simultaneously with minimal additional hardware, it is anticipated that CL spectroscopy will be widespread in the near future. There is some controversy about the peak shock pressure in shergottites. Stöffler et al. (1986) and Stöffler (2000) studied carefully the density of maskelynite and concluded that the peak shock pressure in Shergotty ranges between 29 and 45 GPa. Still they do not rule out the possibility of localized pressure peaks of 60 to 80 GPa, in agreement with the observation of melt pockets. Sharp et al. (1999) and El Goresy et al. (2000, 2003), from the texture and speciation of silica and other high pressure phases, argue that the pressure in Shergotty exceeded 90 GPa, in agreement with the presence of post-stishovite in a α-PbO2 structure (Dera et al. 2002). They further argue that the density of maskelynite is not appropriate to indicate the peak pressure. Due to relaxation effects in the molten state, the density observed is that of the transition temperature, a function of the rate of quenching (pressure and temperature release) and does not reflect the peak pressure precisely. Boctor et al. (2001) observed the presence of significant amounts of pristine water in post-stishovite, which may promote the transformation to high-pressure phases. Recently, a number of high pressure phases have been identified in shergottites such as Ca-majorite, wadsleyite, and hollandite (Malavergne et al. 2001; El Goresy et al. 2000), confirming the variety of high-pressure phases and the potential wealth of information they carry. From their study of NWA 856, Leroux et al (2004) derive a pressure of less than 45 GPa from the TEM study of pyroxenes. Surprisingly though, they observe that silica is under the form of cristobalite in a low-pressure phase. Similarly, using TEM, Weber et al (2000) observed cristobalite in Zagami and indicate pressures not exceeding 31 GPa in agreement with the observations of Stöffler et al (1986), Stöffler (2000). Malavergne et al (2001) using TEM observed cristobalite as well, and based on the presence of Ca-rich majorite and maskelynite, suggest pressures in excess of 60 GPa in both Shergotty and Zagami. They notice the significant heterogeneity of shock features for places very close to one another. However, the presence of cristobalite in Shergotty, Zagami, and NWA 856 could be an artifact of TEM observations, resulting from the transformation of high pressure silica phases under the electron beam or during sample preparation (ion milling). We expected that Raman and CL spectroscopy, which do not require special preparation of the samples, could permit us to check this point and further document the shock pressure supported by these samples. The samples of Shergotty, Zagami, NWA 480, and NWA 856 that we had the opportunity to study always showed that stishovite, high-pressure glass, and post-stishovite (as described by Sharp et al. [1999] and El Goresy et al. [2000])

are pervasive phases. Concerning post-stishovite, its intimate relationship with HP silica glass or stishovite on one hand, its reduced size on the other, and the absence of Raman spectra make the interpretation of CL measurements difficult. So far, only HP silica glass and stishovite spectra have been unambiguously recorded. Notice that, according to SEM images and CL spectra, suspected post-stishovite spots were not transformed by the electron beam. This contrasts with its instantaneous transformation under the laser beam of Raman spectroscopy. ANALYTICAL TECHNIQUES Previous studies using CL in shergottites or terrestrial rocks were mostly devoted to imaging studies (Boggs et al. 2001). Here, besides imaging, much effort was devoted to the spectral analysis of emission with the aim of identifying speciation of silica to obtain information on the shock intensity. Cathodoluminescence We used two cathodoluminescence systems (Kalceff et al. 2000): • The first and simple one is a cold cathode system (Technosyn mark II): in a small chamber under vacuum (0.1 Torr), a HV discharge produces plasma, which promotes the luminescence of the sample surface. An optical microscope permits us to observe and eventually record the cathodoluminescence of the sample through a window. This system is very convenient for making color pictures. • The second one is more sophisticated: in a scanning electron microscope (SEM), a mirror permits the collection of CL photons either to obtain an image (50 sec) in scanning mode or to derive the spectrum of the emitted light with a fixed probe. The SEM used in this study is a JSM-840A equipped with a conventional W filament working at 25 kV and a sample current of between 1 to 100 nA. Cathodoluminescence is collected with a rhodium-coated paraboloidal mirror: Blanc-Perray (UPMC) OPEA system. When shifted over the sample, it permits focusing of the emission to an optical fiber and a spectrometer. We used either one of two spectrometers. The first one, H10 UV, is equipped with a photomultiplyer (PM) and is cooled with a thermoelectronic system cooled by water. It has a spectral resolution of 1 nm and a dispersion of 8 nm per mm of slit width. Both panchromatic images in its zero position or monochromatic images in any other fixed position can be recorded. When working in imaging mode, the spot size at 25 kV is about 5 µm, and the area scanned is variable from 100 to 106 µm2. Over this area, the luminescence image can be stored.

Silica as a shock index in shergottites This first spectrometer permits to record spectra between 300 and 900 nm with a significant sensitivity at low wavelength. The second spectrometer TRIAX (180)® JY, equipped with three gratings permits to record spectra from 200 to 1200 nm with a CCD detector cooled with liquid nitrogen and working at 140 K. The CCD efficiency drops dramatically below 300 nm, while the PM is four times more sensitive in the blue than in the red. The transfer functions of the spectrometers have been determined using a xenon lamp, whose light was collected by the paraboloidal mirror of the optical system. The raw data yield wavelength spectra (between 200 and 1200 nm). All recorded spectra have been corrected for the transfer function of the spectrometer and recast in wavenumber (cm−1). A blank is recorded before every new setting of the analytical conditions and used for correcting the new spectrum. Its contribution is usually negligible. The maximum intensity of the recorded spectra is highly variable from a few hundred counts to a few thousands counts per channel. Tuning the sample current can control the emission level. In spectral mode, the scanning area is fixed (magnification of ×400 whenever possible), three scans of 10 sec each permit to average the instrumental noise. The system is periodically calibrated using a Nd-YAG as a reference sample, which also permits to position the sample to the focus of the paraboloidal mirror (±0.05 mm). All CL spectra have been recorded at 25 kV with a sample current of 100 nA but varied magnifications to minimize mixed spectra as much as possible. Raman Spectroscopy Raman spectra were recorded on the polished sections with a Dilor XY spectrometer equipped with confocal optics and a nitrogen-cooled CCD detector. A microscope is used to focus the excitation laser beam (488 nm and 514 nm lines of a Spectra Physics Ar+ laser) to a 2 µm spot and to collect the Raman signal in the backscattered direction. Accumulations lasted from 120 to 300 sec. The laser power was restricted from 2 to 50 mW to avoid deterioration of the samples. SAMPLES This work was motivated by the recent discovery of Moroccan shergottites, in which shock features are conspicuous. The set of shergottites samples was obtained from two sources. The Theodore Monod Consortium for NWA 480 and NWA 856, and the Museum National d’Histoire Naturelle in Paris (MNHN) for Zagami, Shergotty, and Los Angeles. All samples were polished thick sections, except for Shergotty, which was a thin section. The areas of interest have been selected using backscattered electron (BSE) images at magnifications between ×40 and ×200 coupled with EDS analysis. Microprobe analyses (Cameca

969

SX50) have been performed on the same spots, which permitted us to establish the low level of major impurities in silica, in particular Al, Na, and K (on the average 97.96% SiO2, 0.91% Al2O3, 0.10% K2O, 0.08% CaO, 0.41% Na2O, 0.16% TiO2, 0.27% FeO). We noticed the absence of correlation between CL spectra and compositions. The first step in our study was to demonstrate that stishovite exhibits the same spectrum independently of extrinsic factors such as rock type, composition, and temperature of crystallization. At the same time, spectra for quartz (Capinha granite, Portugal), synthetic optical silica glass, and synthetic cristobalite (Collection de Minéralogie de l’Université Pierre et Marie Curie, Paris VI) have been determined for comparison. In this work, we put a specific emphasis on the sample sections, in which stishovite was clearly identified, i.e., NWA 480, NWA 856, Zagami, and Shergotty. We found no evidence of stishovite in the studied section of Los Angeles; this possibly means that stishovite in Los Angeles is significantly less abundant. RESULTS SEM Images All samples have been examined by backscattered electron (BSE) imaging, first at low magnification (×40), to identify the presence of silica. In all shergottites studied, the silica occurs in various forms: • Geometrical patches interstitial in the texture and in close association with maskelynite. In this situation, the silica is often mixed with maskelynite (Jambon et al. 2002). When in contact with pyroxenes, the latter are usually affected by cracks radiating from silica. • Within impact melt pockets where they appear as polygonal spots of up to several hundred micrometers, rounded clots of variable size usually cracked, and tiny needles of square section of about 1 µm across. Observations at high magnification (×6000) by BSE, secondary electron and cathodoluminescence images were recorded before spectral analysis. In some cases, secondary electron images have been recorded after spectral analysis, which permits to precisely visualize the area covered by CL spectral analysis. NWA 480 (Barrat et al. 2002)

The surface area of the section is nearly 1 cm2. It shows a few impact melt pockets smaller than 1 mm. They are distributed randomly, with no apparent connection to the major fractures. Fifteen areas with significant luminescence have been analyzed. One such area is presented in Fig. 1; it is the only area of this sample where stishovite could be identified, all other areas consisting of high-pressure silica glass.

970

H. Chennaoui Aoudjehane et al.

Fig. 1. a) Optical image of CL by the Technosyn in NWA 480. The low magnification permits us to localize an impact melt pocket with silica (stishovite and glass); b) backscattered electron image (BSE) of the same region. The area is surrounded with maskelynite and pyroxene. MP = melt pocket, M = maskelynite, St = stishovite, Px = pyroxene. Notice the positions of Fig. 2 and Fig. 3.

In Fig. 1a and Fig. 1b, we report the largest occurrence of high-pressure silica (about 1 mm) within an impact melt pocket among all our observations. The silica is surrounded by a matrix of pyroxene composition with numerous small needles of silica oriented against the boundary of the pure silica area (longitudinal sections) and randomly distributed further away (Fig. 2). They exhibit nice square sections, appear to be hollow at high magnification, and were later identified as stishovite. They appear fractured. In the middle of the large silica area, we observed a network of tiny

fractures (Fig. 3). We found this textural feature to be characteristic of the high-pressure silica in shock-melt pockets and also corresponding to the most luminescent spots. The large spots are likely to be a low-pressure silica phase transformed by the shock. The tiny silica needles, however, suggest a high-pressure transformation of pyroxene to yield stishovite needles and melt since we found no Ca, Mg, Fe oxides associated with the needles. This particular texture, described here for the first time, is common in shergottites.

Silica as a shock index in shergottites

971

Fig. 2. Backscattered electron (BSE) image of a typical impact melt pocket in NWA 480. Surrounded by pyroxene, one stishovite needle at the bottom appears cut across in longitudinal section, others show a “hollow” square core when sectioned across (detail from Fig. 1).

Fig. 3. BSE image in the luminescent pure stishovite zone in NWA 480 showing a characteristic pattern of numerous cracks inside the mineral. (detail from Fig. 1). NWA 856 (Jambon et al. 2002)

According to the density of fractures and the significant areal fraction of impact melt pockets (Figs. 4a–4b), it is a shergottite section with pervasive evidence of shock. Sixteen areas, rich in silica, have been studied (imaging, CL

Fig. 4. a) BSE image in an impact-melt pocket of NWA 856. Dark stishovite needles appear in a pyroxene matrix surrounded by pyroxene and maskelynite. Notice the concentric fractures about the melt pocket. MP = melt pocket, M = maskelynite, Px = pyroxene; b) CL panchromatic image of the same zone in the melt pocket; areas with stishovite needles and patches luminesce within the dark pyroxene matrix.

spectroscopy, and Raman spectroscopy). Three such areas exhibit stishovite, while silica in all others appears to be highpressure silica glass. Nevertheless, the textural properties are very similar to those observed in NWA 480. Impact melt pockets contain clots of stishovite in a systematic way surrounded by a partly

972

H. Chennaoui Aoudjehane et al.

glassy matrix of mixed pyroxene and maskelynite, with floating needles of hollow stishovite with square sections (Fig. 5), and somehow connected to the coherent silica areas. An internal network of fractures is also observed in the most luminescent silica spots. BSE images at high magnification show that silica density is heterogeneous. The high-pressure glass could be stressed or homogeneous (Fig. 6).

Fig. 5. BSE image of a detail in NWA 856 (Fig. 4) showing the square sections of stishovite needles nucleated against the compact stishovite.

Shergotty (Tschermak 1872; Duke 1968; Smith and Hervig 1979; Stolper and McSween 1979; Stöffler et al. 1986)

The mineralogy of Shergotty is very similar to that of NWA 480 and NWA 856. The level of shock in this sample is probably more intense than in NWA 480 and NWA 856. Indeed, Sharp et al. (1999) described a new high-pressure form of silica or post-stishovite. Thirty-three areas have been selected (imaging, CL spectroscopy, and Raman spectroscopy) and a total of 50 CL spectra have been recorded, 12 of which being stishovite and four being a mixture of stishovite and other phases. Among those, three are impact melt pockets, the fourth one corresponding to a single crystal associated with silica glass. The remaining 34 are silica glass. The stishovite of the melt pockets in Shergotty is quite different from what is observed in the above samples: no needles are observed and instead small areas of a few µm with a cracked aspect are frequent and characteristic of this sample. This cracked silica is composed by the close association of stishovite and HP silica glass (Fig. 7a and Fig. 7b), suggesting that glass results from amorphization upon decompression. The texture of this area and some other areas of HP silica glass (Fig. 7c) are similar to that described by El Goresy et al. (2000) as post-stishovite. The intensity of luminescence in Shergotty is significantly higher compared to all other shergottites studied. Zagami (Stolper et al. 1979; McCoy et al. 1992)

The textural properties of Zagami are very similar to those observed in Shergotty, NWA 480, and NWA 856. Twenty-nine silica areas have been studied in the section (imaging, CL spectroscopy, and Raman spectroscopy). Seven melt pockets similar to those of Shergotty have been identified with a mixture of stishovite and other phases. In the MNHN section, no stishovite needles were observed, unlike in NWA 480 and NWA 856. Still, in another section (courtesy El Goresy, MPI Mainz) we found typical needles in pyroxene glass. Twenty-four silica CL spectra have been recorded. Fourteen of them are stishovite spectra, the remaining 10 are H.P silica glass according to their Raman spectra (Okuno et al. 1999) Los Angeles (Rubin et al. 2000; Mikouchi 2001; Xirouchakis et al. 2002)

Fig. 6. BSE image of a homogeneous high-pressure silica glass in NWA 856 without stress and the tweed pattern.

Thirteen silica areas present in the MNHN section have been studied. The section exhibits a single melt pocket devoid of stishovite. Eleven CL spectra have been recorded. No occurrence of stishovite has been observed. According to the Raman spectra, all areas exhibit HP silica glass (Okuno et al. 1999). Interestingly, in a recent work, Walton and Spray (2003) show a nice picture of stishovite needles in pyroxene glass in a melt pocket even though they were not recognized as such.

Silica as a shock index in shergottites

973

Fig. 7. a) BSE image of a silica zone with high-pressure silica glass and stishovite in Shergotty. St = stishovite, G = high-pressure silica glass; b) CL image of the same zone showing luminescence of stishovite (bright) and silica glass. The relationship of the two phases suggests retromorphosis of stishovite or post stishovite to the surrounding glass; c) secondary electron (SE) image of a HP silica glass (G) with typical texture of post-stishovite as described by El Goresy et al. (2000), surrounded by fractured maskelynite (M).

Spectroscopic Measurements In NWA 480, NWA 856, Zagami, Shergotty, and Los Angeles, CL spectra were recorded (Figs. 8–9), stishovite, when present, or glass was identified by Raman spectroscopy (Figs. 10–11) from the same spot, which permits us to assign the spectra to one or the other phase. In Shergotty, no stishovite Raman spectra could be recorded because of the

exceedingly high stray luminescence under the laser beam. Stishovite was suggested based on its CL spectrum (Fig. 8) after comparison with the other shergottite samples. In Los Angeles, no stishovite CL spectra could be recorded probably because there was no stishovite in this section. Typical stishovite CL spectra are presented in Fig. 8, and typical HP silica glass in the studied shergottites are presented in Fig. 9. The CL spectra of quartz, synthetic silica glass, and

974

H. Chennaoui Aoudjehane et al.

Fig. 8. Stishovite CL spectra in NWA 480, NWA 856, Shergotty, and Zagami. All CL spectra corrected for the transfer function, background, and recast against the wave number (cm−1), two peaks are shown near 11,500 and 22,500 cm−1. Notice the presence of a third peak near 25,000 cm−1 in some spectra for NWA 856 and Shergotty.

Fig. 9. CL Spectrum of high-pressure silica glass in NWA 856, NWA 480, Shergotty, Zagami, and Los Angeles showing two peaks near 15,100 and 18,000 cm−1.

cristobalite are presented in Figs. 12–14 for comparison. The intensity of the luminescence is highly variable. Since intensity is strongly sensitive to a number of factors (abundance of the phase of interest, magnification, current, position of the sample), this variability will not be discussed in detail, as no quantitative estimate is presently possible. More interesting is the identification of peaks. In stishovite, clearly two peaks are easily recognized: one near 11,500 cm−1 the other near 22,500 cm−1. The intensities of the two peaks vary independently from one another. The Raman spectra of high-pressure silica have been described previously by Gillet et al. (1990), in particular the

spectra for stishovite, coesite and quartz for the crystalline phases and glasses of both low and high pressure are available. Raman spectra obtained for stishovite from NWA 480 and NWA 856 are reported in Fig. 10. The characteristic peaks of stishovite at 231, 589, 753, and 967 cm−1 are observed in both samples. A broad feature appears between 400 and 500 cm−1. This region is characterized by the vibration of rings of tetrahedra in glass. The peak position (vibration frequency) depends on the number of tetrahedra involved in cycles as this number, on the average, decreases with density. The wide feature in Raman spectra corresponds to a contribution of several frequencies.

Silica as a shock index in shergottites

975

Fig. 10. Two typical Raman spectra obtained for both NWA 480 and NWA 856. The spectra are not corrected for the transfer function. The stishovite peaks are easily identified (231, 589, 752, and 967 cm−1). The broad peak between 400 and 500 cm−1 corresponds to high-pressure silica glass probably due to the retromorphosis of stishovite or post stishovite (see Fig. 11).

Fig. 11. Typical Raman spectrum of HP silica glass in Los Angeles in the same conditions as above. Stishovite peaks (arrows) do not appear. The broad glass peak between 400 and 500 cm−1 is conspicuous. The dashed line is for easier comparison with the Raman spectrum of HP silica glass in Shergotty. When compared with Los Angeles, the peak at 510 cm−1 presents a significant shift to higher frequencies, corresponding to the higher density of the silica glass in Shergotty.

The average wave number has been shown to increase with the density of the glass at the expense of the smaller values, which tend to disappear as the pressure is increased. Notice that in Raman spectroscopy the peaks derived from the crystalline structures are more intense and well defined than those deriving from amorphous material. At this stage, we can conclude that stishovite is always shown to coexist with HP pure silica glass. This glass is interpreted as a result of amorphization during decompression

of stishovite (see Figs. 7a–7b). In Shergotty, the pressure at which the glass was quenched is variable from one spot to the next possibly because of the variable P, T, t paths during quenching. Shergotty exhibits the highest glass density among the samples studied (Fig. 11). In NWA 480, NWA 856, Zagami, Shergotty, and Los Angeles, CL spectra of HP silica glass have been recorded (Fig. 9). In agreement with Raman spectra (Fig. 11), they exhibit two peaks of HP silica glass, one being at about

976

H. Chennaoui Aoudjehane et al.

Fig. 12. Reference CL spectrum of quartz in Capinha granite (Portugal) collected under the same conditions as above, showing two peaks near 14,900 and 20,100 cm−1.

intense near 15,900 cm−1, the other two near 19,500 cm−1, and 26,500 cm−1. CL spectra for silica phases differ significantly from one another and permit us to recognize easily each silica phase. High-pressure post-silica phases have been obtained experimentally and observed in Shergotty by El Goresy et al. (1998 ), Sharp et al. (1999), Dera et al. (2002). A number of areas exhibiting the specific features described have been recognized in Shergotty, Zagami, and NWA 856. As already mentioned by El Goresy (personal communication), we noticed that the putative post-stishovite is immediately destroyed under the Raman beam. CL spectra could be obtained and left us with some perplexity. Diverse types of spectra were recorded which might be interpreted as mixtures of glass, stishovite, and possibly post-stishovite. In the absence of a crossreference by Raman spectrometry for the same area, a clean post stishovite spectrum could not be derived. An additional difficulty could also result from the existence of several post-stishovite phases (baddeleyite, CaCl2 and α-PbO2 structures have been described: German et al. 1973; Belonoshko et al. 1996; Dubrovinsky et al. 1997; Karki et al. 1997; Teter et al. 1998). Clearly, this point requires further investigation. DISCUSSION Stishovite

Fig. 13. Reference CL spectrum of a synthetic low pressure silica glass (optical glass) collected under the same conditions as above, showing two peaks near 15,500 and 21,500 cm−1.

Fig. 14. Reference CL spectrum of synthetic cristobalite (Université Pierre et Marie Curie, collection of mineralogy), collected under the same conditions as above, with three peaks near 15,900, 19,500, and 26,500 cm−1.

15,100 cm–1; the others near 18,000 cm−1. These spectra are noisier because the intensities of luminescence are lower. CL spectrum of quartz (Fig. 12) shows two peaks. One is near 14,900 cm−1, the other, less intense, is near 20,100 cm−1. For synthetic low pressure silica glass, the CL spectrum (Fig. 13) presents two peaks, one near 15,500 cm−1 and another more intense near 21,000 cm−1. The measurement of CL spectrum of cristobalite shows three peaks, the more

The CL spectrum of stishovite (Fig. 8) consists of two peaks at 11,500 cm−1 and 22,500 cm−1 of variable relative intensities. According to the experiments of Kaus (2002) this may be due to polarization effects. Our setup, however, does not permit to check this point. In Shergotty and NWA 856, the CL spectra could present a third peak near 25,000 cm−1 (Fig. 8); the simplest interpretation is the presence of poststishovite, probably in relation with the intensity of the shock and the sample history (Gillet et al. 1990) and in agreement with the diffraction patterns obtained independently by El Goresy et al. (1998), Sharp et al. (1999), and Dera et al. (2002). Therefore, this point needs confirmation. Raman spectra (Fig. 10) show that stishovite never occurs alone; it is always associated with silica glass, which we interpret as the result of either stishovite or post-stishovite retromorphosis upon decompression. The luminescence of HP silica glass associated with stishovite is so much lower that it never appears clearly in the CL spectra and also because we tried as much as possible to avoid mixed areas. HP Silica Glass In NWA 480, NWA 856, Shergotty, Zagami, and Los Angeles (Fig. 9), the CL spectra of HP silica glass differ from those of stishovite. They present two peaks (15,100 and 18,000 cm–1). The peak at 15,100 cm−1 is the highest one. The luminescence of HP silica glass is not as intense as that of

Silica as a shock index in shergottites stishovite. Raman spectra show clearly a broad peak between 400 and 500 cm−1 corresponding to HP silica glass (Okuno et al. 1999). A comparison between HP silica glass Raman spectra in Los Angeles and Shergotty (Fig. 11) shows a shift of the peak from 495 cm−1 in Los Angeles to 510 cm−1 in Shergotty. According to Okuno et al. (1999), this indicates a lower density and therefore lower shock intensity in Los Angeles compared to the stishovite-bearing shergottites that could explain the absence of stishovite in Los Angeles in spite of the similarity with the other samples studied. Still, this point is a matter of discussion as we notice (see also Malavergne et al. 2001) that shock is very heterogeneously revealed in all specimens for which more than one section was available (NWA 856, Shergotty, Zagami). Indeed, Walton and Spray (2003), in their section of Los Angeles, have recorded (but not recognized) the occurrence of stishovite needles, which are absent in the MNHN section. Two different habits of HP silica glass are observed in maskelynite. One has the typical texture of post-stishovite (Fig. 7c; below) in Shergotty and Zagami. The other type is featureless, like maskelynite (Fig. 6), and is probably pristine glass and not a product of retromorphosis. The presence of plagioclase-silica eutectic compositions in NWA 856, and more generaly, compositions derived from chemical diffusion between silica and plagioclase indicate that these phases were liquids, not glasses, or diffusion would have been impossible over the time allowed for cooling. Post-Stishovite The presence of post-stishovite in Shergotty has been attested by electron and X-ray diffraction patterns. In all specimens studied, the areas which exhibit a texture identical to that of post-stishovite revealed glass or stishovite, or both. Other spectra have been recorded, which could be mixtures of the above with some additional unidentified peaks. Because of the significant variations from one spectrum to the other, those could not be assigned to a specific phase. We could not benefit from Raman spectroscopic information since poststishovite is instantaneously destroyed under the laser beam. The electron beam is not as destructive when used in a scanning mode and CL spectra could be recorded. The absence of identified post-stishovite can be interpreted in one or several different ways: 1) post-stishovite does not exhibit cathodoluminescence. As it is always associated with silica glass, CL records only the presence of glass despite its weak luminescence; 2) post-stishovite is only exceptionally preserved. As we analyzed a limited number of areas, we were not lucky enough to find post-stishovite; 3) the spectra of post-stishovite differ marginally from those of stishovite, and in the absence of a reference we were unable to make the distinction; and 4) several post-stishovite phases are present in the different areas leading to an apparent non reproducibility. The presence of a third peak in some stishovite CL spectra in Shergotty and NWA 856 and the

977

exceptionally high luminescence suggest the presence of post-stishovite, which would reconcile the present results with the diffraction data. The fractures radiating into maskelynite from the areas where post-stishovite is suspected (Fig. 7c), indicate volume expansion due to retromorphosis upon decompression and point out where a high-pressure phase was formed but not preserved. Other Silica Phases Quartz, cristobalite, and synthetic low-pressure glass have been studied in reference samples for comparison. The CL spectra (Figs. 12–14) differ substantially from those of stishovite and HP silica glass. They also differ from each other. The luminescence of all the other silica phases is less intense than that of stishovite. We never observed, either with CL or Raman spectroscopy, the presence of cristobalite, low pressure glass, coesite, or quartz in shergottites. We therefore infer that cristobalite identified by TEM is an artifact of sample preparation. This seems reasonable since cristobalite is a low-pressure phase that should not coexist with either stishovite or high-pressure glass. More interesting is the systematic absence of coesite. This high-pressure phase has been identified as a high-pressure polymorph of silica in shocked rocks, either meteorites or suevites. Because of the heterogeneous distribution of shock, one should expect that some silica could be transformed to coesite. If we interpret this in terms of the equilibrium phase diagram of silica, the shock pressure should be beyond the field of coesite (significantly more than 15 GPa), and upon relaxation and cooling, the P-T path should be maintained in the field of liquid silica and not coesite, with temperatures above 3000 to 4000 K (Luo et al. 2002). This is in agreement with the presence of numerous occurrences of silica glass and also with the occurrence of eutectic composition in the plagioclase-silica system as shown by Jambon et al. (2002) in NWA 856. However, the preservation of dense silica glass revealed by Raman spectroscopy is not in favor of such a high temperature path. Okuno et al. (1999) have shown that densified silica glass relax to low-pressure silica glasses within minutes at temperatures as low as 900 K. Even if we consider the time scales of shock events recently inferred for Martian meteorites of 10–100 ms (Beck et al. 2005), relaxation of densified glass should occur at temperatures of the order of 1200 K. Thus, no densified silica glass should be preserved with the high temperature path proposed from equilibrium phase diagram. The observation of dense silica glass thus constrains the shock and residual temperature to low values, and the absence of coesite and low-pressure polymorphs of silica is more likely due to kinetic hindrance of the back-transformation, also allowing the preservation of stishovite and post-stishovite phases. Depending on the shock pressure, temperature, and cooling rate, the transition temperature will freeze a more or less dense glass, as recognized in the Raman spectra.

978

H. Chennaoui Aoudjehane et al.

Post-stishovite is a highly unstable form of silica and despite the rapid quench did not survive in most cases. It is however revealed by textural features such as radial fractures and tweed pattern. If correct, this indicates a temperature in excess of 3000 K, which according to the shock experiments of Lyzenga et al. (1983), and Luo et al. (2002) corresponds to a pressure of at least 60 GPa. Heterogeneity of Shock The form of silica is not the same everywhere in the rock indicating that the peak shock pressure was highly variable. In addition, different sections do not exhibit the same phases. In Shergotty, only one thin section and none of the polished thick sections of the MNHN exhibited stishovite, while poststishovite was described in other sections. In the MNHN section of Los Angeles, stishovite was not observed either; still Walton and Spray (2003, Fig. 3) show stishovite needles identical to those observed in NWA 480 and NWA 856. In the Zagami section of MPI (Mainz), stishovite needles and grains were observed, while they are absent in the MNHN section. This could also explain the possible presence of diverse poststishovite phases. CONCLUSION In NWA 480, NWA 856, Zagami, and Shergotty, stishovite was recognized by Raman and/or CL spectroscopy. The comparison of the spectra obtained by CL and Raman spectroscopy permits us to identify the characteristic CL peaks for stishovite and high-pressure silica glass. Stishovite is always associated with high-pressure silica glass. The proportion of stishovite and glass is highly variable even in the same sample. The variable position of the glass peak (Raman spectroscopy) corresponds to its variable density in relation with its cooling-decompression history. When decomposed as the sum of two components (stishovite and glass), all spectra look alike and differ significantly from the spectra obtained for low-pressure silica phases (quartz, cristobalite and silica glass). According to our observations of various silica phases in shergottites, we can contribute to the discussion of the shock pressure in shergottites. Stöffler et al. (1986), Stöffler (2000) and Malavergne et al. (2001) noticed that shock is strongly heterogeneous in Shergotty, Zagami, NWA 480, NWA 856, and Los Angeles. This is well confirmed by the present study. The occurrence of several well-localized melt pockets with stishovite needles growing in molten pyroxene, poststishovite associated with massive stishovite and HP glass, radiating cracks into both surrounding pyroxene and maskelynite, and the absence of coesite indicate that the shock pressure probably exceeded 60 GPa, which is in agreement with the observations of El Goresy et al. (2000) and Sharp et al. (1999). Shock intensity is related to the highly anisotropic impedance of the rock related to the highly variable elasticity

of the different minerals. Pyroxenes were not molten, except marginally in melt pockets where they transform to melt and stishovite. Plagioclase was molten as revealed by textural evidences (El Goresy et al. 1998) as well as mixed with molten silica to give eutectic compositions (Jambon et al. 2002). Silica is transformed to high-pressure phases (stishovite and probably post-stishovite), which retromorphose to silica liquid and glass. The presence of post-stishovite is revealed by the significant volume increase after decompression underscored by radiating cracks into surrounding maskelynite and pyroxenes and the specific tweed pattern. In shergottites, pyroxenes were only marginally molten and thus the textures were preserved. The density of silica glass is variable: it probably depends on the local shock intensity and the cooling path. If a silica-rich melt cools under pressure, then the density of its glass will be significantly higher than that formed at ambient pressure. The same obviously applies for its structure. The density of maskelynite is an important and interesting quantitative measurement of shock intensity Stöffler et al. (1986), Stöffler (2000). Since its ability to record the peak shock pressure has been challenged (El Goresy et al. 2000; Malavergne et al. 2001), every other index of shock intensity is welcome. The transformation kinetics of crystalline silica phases differs from that of glasses but the great instability of stishovite and post-stishovite remains a serious difficulty. Unlike Raman spectroscopy, CL is apparently harmless to post-stishovite and easy to implement; it therefore appears as a promising technique. Acknowledgments–We are grateful to M. Bourot Denise (Muséum National d’Histoire Naturelle) for a loan of sections of Shergotty, Zagami, and Los Angeles, A. El Goresy (MPI, Mainz, Germany) for the loan of a Zagami section, C. Derré (Université Pierre et Marie Curie, Paris VI, France) for the Capinha granite sample, and the “Collection de Minéralogie” (Université Pierre et Marie Curie, Paris VI) for the sample of synthetic cristobalite. H. Chennaoui Aoudjehane benefited from grants from UPMC and Agence Universitaire de la Francophonie. The CNRS (France) and the CNRST (Morocco) supported this work. G. Montagnac (ENS, Lyon, France) is thanked for help in Raman spectroscopy. This paper has greatly benefited from the constructive reviews of Dr. Valérie Malavergne and Dr. Friedrich Horz, and the careful editorial handling and helpful suggestions of the associate editor, Dr. Allan Treiman. Editorial Handling—Dr. Allan Treiman REFERENCES Barrat J. A., Gillet Ph., Sautter V., Jambon A., Javoy M., Göpel C., Lesourd M., Keller F., and Petit E. 2002. Petrology and chemistry of the basaltic shergottite Northwest Africa 480. Meteoritics & Planetary Science 37:487–501. Beck P., Gillet Ph., El Goresy A., and Mostefaoui S. 2005. Timescales of shock processes and the size of martian meteorites

Silica as a shock index in shergottites

source craters, constrained from shock metamorphism (abstract #1333). 36th Lunar and Planetary Science Conference. CDROM. Belonoshko A. B., Dubrovinsky L. S., and Dubrovinsky N. A. 1996. A new high pressure silica phase obtained by molecular dynamics. American Mineralogist 81:785–788. Boctor N. Z., D’Alexander C. M. O., Wang J., and Hauri E. 2001. Hydrogen isotope studies of water-bearing post-stishovite silica phase and feldspathic glass in the martian meteorites Shergotty and Zagami (abstract #1309). 32nd Lunar and Planetary Science Conference. CD-ROM. Boggs S., Krinsley D. H., Goles G. G., Seyedolali A., and Dypvik H. 2001. Identification of shocked quartz by scanning cathodoluminescence imaging. Meteoritics & Planetary Science 36:783–791. Dera P., Prewitt C. T., Boctor N. Z., and Hemley R. J. 2002. Characterization of a high-pressure phase of silica from the martian meteorite Shergotty. American Mineralogist 87:1018– 1023. Dubrovinsky L. S., Saxena S. K., Lazor P., Ahuja R., Eriksson O., Wills J. M., and Johansson B. 1997. Experimental and theoretical identification of a new high-pressure phase of silica. Nature 388: 362–365. Duke M. B. 1968. The Shergotty meteorite: Magmatic and shock metamorphic features. In Shock metamorphism of natural materials, edited by French B. M and Short N. M. Baltimore: Mono Book Corporation. pp. 612–621. El Goresy A., Dubrovinsky L., Sharp T. G., Saxena S. K., and Chen M. 2000. A monoclinic post-stishovite polymorph of silica in the Shergotty meteorite. Science 288:632–634. El Goresy A., Graup G., and Chen M. 2003. The assemblage maskelynite-post-stishovite silica polymorphs in the Shergotty meteorite: New stringent constraints to peak-shock pressures (abstract). Meteoritics & Planetary Science 38:A50. El Goresy A., Sharp T. G., Wopenka B., and Chen M. 1998. A new very high-pressure silica mineral in the Shergotty SNC meteorite: Implications for shock metamorphism and the Earth’s lower mantle (abstract #1707). 29th Lunar and Planetary Science Conference. CD-ROM. German V. N., Podurets M. A., and Trumin R. F. 1973. Synthesis of a high-density phase of silicon dioxide in shock wave. Soviet Physics JETP 37:107. Gillet Ph., Le Cléac’h A., and Madon M. 1990. M. High-temperature Raman spectroscopy of SiO2 and GeO2 polymorphs: Anharmonicity and thermodynamic properties at high temperatures. Journal of Geophysical Research 95:21,635– 21,655. Graham A. L., Bevan A. W. R., and Hutchison R. 1985. Catalogue of meteorites, 4th edition. Tucson: The University of Arizona Press. Jambon A., Barrat J. A., Sautter V., Gillet Ph., Göpel C., Javoy M., Joron J. L., and Lesourd M. 2002. Djel Ibone, the new basaltic shergottite Northwest Africa 856: Petrology and chemistry. Meteoritics & Planetary Science 37:1147–1164. Kalceff M. A. S., Matthew R. P., Anthony R. M., and Kalceff W. 2000. Cathodoluminescence microcaracterisation of silicon dioxide polymorphs. In Cathodoluminescence in the geosciences, edited by Pagel M., Barbin V., Blanc P., and Ohnenstetter D. Berlin: Springer-Verlag. 193 p. Karki B. B., Warren M. C., Stixrude L., Acland G. J., and Crain J. 1997. Ab initio studies of high-pressure structural transformations in silica. Physical Review B 55:3465–3471. Kaus A. 2002. Kathodolumineszenz in der Planetologie: Methodik und Anwendung. Ph.D thesis, Universität Münster, Germany. Leroux H., Devouard B., Cordier P., and Guyot F. 2004. Pyroxene microstructure in the Northwest Africa 856 martian meteorite.

979

Meteoritics & Planetary Science 39:711–722. Luo S. N, Mosenfelder J. L, Asimow P. D., and Ahrens T. J. 2002. Direct shock wave loading of stishovite to 235 GPa: Implications for perovskite stability relative to an oxide assemblage at lower mantle conditions. Geophysical Research Letters 29:36-1–36-4. Lyzenga G. A., Ahrens T. J., and Mitchell A. C. 1983. Shock temperatures of SiO2 and their geophysical implications. Journal of Geophysical Research 88:2431–2444. Malavergne V., Guyot F., Benzerara K., and Martinez I. 2001. Descriptions of new shock-induced phases in the Shergotty, Zagami, Nakhla, and Chassigny meteorites. Meteoritics & Planetary Science 36:1297–1305. McCoy T. J., Taylor G. J., and Keil K. 1992. Zagami: Product of a two-stage magmatic history. Geochimica et Cosmochimica Acta 56:3571–3582. Meyer C. 2003. Mars Meteorite Compendium. http:// curator.jsc.nasa.gov/antmet/mmc/mmc.htm. Mikouchi T. 2001. Mineralogical similarities and differences between the Los Angeles basaltic shergottites and the Asuka881757 lunar mare meteorite. Antarctic Meteorite Research 14: 1–20. Okuno M., Reynard B., Shimada Y., Syono Y., and Willaime C. 1999. A Raman spectroscopic study of shock-wave densification of vitreous silica. Physics and Chemistry of Minerals 26:304–311. Rubin A. E., Warren P. H., Greenwood J.P ., Verish R. S., Leshin L. A., Hervig R. L, Clayton R. N., and Mayeda T. K. 2000. Los Angeles: The most differentiated basaltic martian meteorite. Geology 28:1011–1014. Sharp T. G., El Goresy A., Wopenka B., and Chen M. 1999. A poststishovite polymorph in the meteorite Shergotty: Implications for the impact events. Science 284:1511–1513. Smith J. V., and Hervig R. L. 1979. Shergotty meteorite: Mineralogy, petrography, and minor elements. Meteoritics 14:121–142. Stöffler D. 2000. Maskelynite confirmed as diaplectic glass: Indication for peak shock pressures below 45 GPA in all martian meteorites (abstract #1170). 31st Lunar and Planetary Science Conference. CD-ROM. Stöffler D., Ostertag R., Jammes C., Pfannschmidt G., Sen Gupta P. R., Simon S. B., Papike J. J., and Beauchamp R. H. 1986. Shock metamorphism and petrography of the Shergotty achondrite. Geochimica et Cosmochimica Acta 50:889–913. Stolper E. M., and McSween H. Y., Jr. 1979. Petrology and origin of the shergottite meteorites. Geochimica et Cosmochimica Acta 43: 1475–1498. Stolper E. M., McSween H. Y., Jr, and Hays J. F. 1979. A petrogenetic model of the relationships among achondritic meteorites. Geochimica et Cosmochimica Acta 43:589–602. Teter D. M., Hemley R. J., Kresse G., and Hafner J. 1998. High pressure polymorphism in silica. Physical Review Letters 80: 2145–2148. Tschermak G. 1872. Die Meteoriten von Shergotty und Gopalpur. Sitzungsberichte der Mathematisch-Naturwissenschaftlichen Classe der Kaiserlichen Akademie der Wissenschaften 65:122– 146. Walton E. L. and Spray J. G. 2003. Mineralogy, microtexture, and composition of shock-induced melt pockets in the Los Angeles basaltic shergottite. Meteoritics & Planetary Science 38:1865– 1875. Weber I., Greshake A., and Bischoff A. 2000. Low-cristobalite in the martian meteorite Zagami (abstract #1342). 31st Lunar and Planetary Science Conference. CD-ROM. Xirouchakis D. M., Draper D. S., Schwandt C. S., and Lanzirotti A. 2002. Crystallization conditions of Los Angeles, a basaltic martian meteorite. Geochimica et Cosmochimica Acta 66:1867– 1880.