Portal de Periódicos da CAPES

Lunar meteorite regolith breccias: An in situ study of impact melt composition using LA-ICP-MS with implications for the composition of the lunar crust

2010; Wiley; Volume: 45; Issue: 6 Linguagem: Inglês

10.1111/j.1945-5100.2010.01067.x

1945-5100

K. H. Joy, Ian Crawford, S. S. Russell, A. T. Kearsley,

Isotope Analysis in Ecology

Dar al Gani (DaG) 400, Meteorite Hills (MET) 01210, Pecora Escarpment (PCA) 02007, and MacAlpine Hills (MAC) 88104/88105 are lunar regolith breccia meteorites that provide sampling of the lunar surface from regions of the Moon that were not visited by the US Apollo or Soviet Luna sample return missions. They contain a heterogeneous clast population from a range of typical lunar lithologies. DaG 400, PCA 02007, and MAC 88104/88105 are primarily feldspathic in nature, and MET 01210 is composed of mare basalt material mixed with a lesser amount of feldspathic material. Here we present a compositional study of the impact melt and impact melt breccia clast population (i.e., clasts that were generated in impact cratering melting processes) within these meteorites using in situ electron microprobe and LA-ICP-MS techniques. Results show that all of the meteorites are dominated by impact lithologies that are relatively ferroan (Mg# 10), and have low incompatible trace element (ITE) concentrations (i.e., typically <3.2 ppm Sm, 10 ppm Sm), High Magnesium Suite (typically >70 Mg#) or High Alkali Suite (high ITEs, Sc/Sm ratios <2) target rocks. Instead the meteorite mafic melts are more ferroan, KREEP-poor and Sc-rich, and represent mixing between feldspathic lithologies and low-Ti or very low-Ti (VLT) basalts. As PCA 02007 and MAC 88104/05 were likely sourced from the Outer-Feldspathic Highlands Terrane our findings suggest that these predominantly feldspathic regions commonly contain a VLT to low-Ti basalt contribution. Lunar regolith is defined as a layer or mantling deposit that that has been generated from continual meteoroid bombardment of bedrock lithological units (McKay et al. 1991). This physical modification causes the shattering, disaggregation, pulverization, melting, transporting, and mixing of lithic and mineral fragments, forming a poorly consolidated layer on the lunar surface (Hörz et al. 1991). Typically this layer is much less than 1 cm in grain size, although very large boulders and cobbles are commonly strewn across the lunar surface. The lunar regolith provides a record both of the Moon's geological history and its interaction with the dynamic inner solar system space environment (Lucey et al. 2006). This archive is important for our understanding of the evolution of terrestrial planets, as few other planets (such as the Earth, Mars, and Venus) have retained such a complete record of their earliest history. Lunar regolith breccia meteorites are consolidated samples of the lunar regolith that were ejected from the Moon and transported to Earth by impact cratering processes. They have potentially been launched from anywhere on the lunar surface, although their precise launch craters are unknown. These meteorites reflect a broad range of compositional affinities from different lunar lithological terranes (Korotev 2005), including the feldspathic highlands (likely from both the far and nearside regions); from basaltic lavas; from areas formed from mixtures of basalt and feldspathic bedrock; and also from environments dominated by KREEP-rich impact melts (i.e., samples characterized by high concentrations of potassium, rare earth elements, and phosphorus). Radiogenic isotope studies indicate that the majority of known lunar meteorites have been launched from the Moon in the last 10 million years (Korotev 2005), and all have been launched in the last 20 million (Nishiizumi and Caffee 2001). As there are believed to have been no large craters formed on the Moon during this period (all craters in the last 1 million years or so are <3.6 km in diameter; Warren 1994), it is assumed that all lunar meteorites must therefore be launched from small craters only a few kilometers or less in diameter (Head et al. 2002). Moreover, evidence from studies of cosmic-ray exposure indicates that many of the regolith breccia samples have been preferentially derived from very shallow to shallow stratigraphic horizons (2 m to <100 m; Warren 1994). For this reason they can provide a valuable calibration for the composition of the upper regolith, especially for inaccessible farside geological terranes (Korotev et al. 2006). For example, lunar feldspathic regolith breccia meteorites have been used to help calibrate remote sensing information from the lunar feldspathic highlands (Warren 2005; Joy et al. 2006a; Prettyman et al. 2006; Swinyard et al., 2009), and have been used to inform discussions of lunar anorthositic primary crust petrogenesis and evolution (Palme et al. 1991; Jolliff and Haskin 1995; Shearer and Floss 2000; Korotev et al. 2003; Longhi 2003; Warren et al. 2005). Lunar regolith breccias are polymict fragmental rocks, containing components that are characteristic of the lunar soil (defined by McKay et al. 1991 as "the sub-centimeter fraction of the lunar regolith" formed of very fine grained (average grain size between 60 and 80 μm clastic and mineral material). These regolith components include impact melt spherules (melt beads), impact glass, agglutinates (unique glassy bonded soil particles containing an impact derived nanophase Fe component; Taylor et al. 1991), an exogenous meteoritic component (Rubin 1997; Zolensky 1997; Day et al. 2006), and clasts of older breccias. The presence of just one or a combination of several of these components is evidence that the breccia was fused together in a lunar regolith environment. Estimates of the maturity of this regolith environment (i.e., how long it was exposed to space) is inferred from bulk sample grain size (generally mature soils are finer grained and have more agglutinate particles) and surface exposure ages (i.e., length of time it was exposed to ionizing radiation; Lucey et al. 2006). Igneous rock and mineral fragments form a significant component of regolith breccias, and the modal mineralogy, mineral composition, and bulk composition of these clasts can be used to construct an inventory of their bedrock source lithologies. This inventory can then be used to interpret the petrological history of individual clasts and provide new information about global lunar magmatic processes and geological variation (e.g., Nyquist et al. 2006). Lunar regolith breccias commonly contain clasts of impactite rocks derived from rock that was recrystallized, partially melted or completely melted by meteorite impacts (see discussion by Cohen et al. 2004). These impactites represent a compositional mixture of all the rocks types that were located in the target area and commonly also contain a minor additional contribution from the impacting bolide body. Thus, the products of meteorite impacts (melts and metamorphites) can commonly be distinguished from indigenous igneous products (and their metamorphites) by their siderophile elements abundances, where impact material will be relatively rich in siderophiles, derived from meteoritic impactors (Warren and Wasson 1977, 1978). Impact lithologies are an important indicator of geochemical and lithological provenance as their compositions provide an insight into the nature of local geological terranes. For example, the low-K Fra Mauro ("LKFM") mafic impact melts (Spudis et al. 1991; Korotev 2000), returned from the Apollo 15 landing site (i.e., samples 15445 and 15455, Ryder and Spudis 1994), are thought to have been formed during the Imbrium basin impact (Spudis 1993). As these mafic impact melts were returned from a known locality on the lunar nearside, it can be assumed that this reflects a geochemical signature directly related to basin forming within the Procellarum KREEP Terrane ("PKT"; Jolliff et al. 2000; Haskin et al. 2000; Korotev 2000), and is perhaps indicative of the nature of the lower near-side crust (approximately 40 to 80 km depth: Spudis 1993). Impact melt and breccia clasts therefore are important micro-samples to study in regolith breccias as they preserve a compositional record of crustal terranes from regions not sampled by the Apollo and Luna sample return missions. We were loaned a polished thick section block of DaG 400 (8.5 × 6.8 × 1 mm) from the Vatican Observatory Collection. The sample had previously been used for trace element investigations by LA-ICP-MS (laser ablation inductively coupled mass spectrometry) and so there was some scarring by laser pits and ablated ejecta (see Consolmagno et al. [2004] for details of this prior study). In order to gain precise quantitative reproducible results, microprobe measurements were not taken from areas within 150 μm of these pre-existing laser pits and the section was repolished to remove as much deposited material as possible before we started our analysis. In addition, a small portion of DaG 400 (200 mg) was removed from a large slab held at the Natural History Museum London (NHM) for use in bulk composition analysis (Table 1). We also studied polished carbon coated thick-thin sections of MET 01210,21 (10.4 × 14.6 × 0.2 mm), MET 01210,27 (8.6 × 13.1 × 0.2 mm), PCA 02007,34 (14 × 6.8 × 0.2 mm), MAC 88104,47 (10 × 8 × 0.2 mm), and MAC 88105,159 (11 × 6 × 0.2 mm) (Fig. 1) and used chip sub-splits of MET 01210,25, PCA 02007,37, MAC 88104,45, MAC 88105,106, and homogenous powder sub-split MAC 88105,41 (Lindstrom et al. 1991) for bulk composition analysis (Table 1). We determined the bulk-rock major, minor, and trace-element composition of powdered chips of the meteorites by inductively coupled plasma––atomic emission spectrometry (ICP-AES) and mass spectrometry (ICP-MS) at the NHM using the techniques and instrumental setup described by Joy et al. (2008). The instruments used were a Varian VISTA PRO Axial ICP-AES and a Varian ICP-MS. Comparison between our elemental values and literature meteorite compositions are discussed in section Constraining the Source Region of DaG 400, MET 01210, PCA 02007, and MAC 88104/05. Classification of clast lithologies within the meteorite samples studied here. Classification derived from texture (Fig. 3), mineral content (Fig. 4), and bulk chemical composition in the meteorite sections (Table 1). a) DaG 400: the largest clast is elongate, has a sub-rounded rim, and is approximately 5 × approximately 2.5 mm. The section is crossed and cut by veins, probably formed through impact melt injection, and some fractures have been infilled with terrestrial CaCO3. b) PCA 02007,34: elongate thin section formed from small clasts (<1.5 mm) that are mostly derived from impact melt and impact melt breccias. Dashed outlines represent four regolith breccia clasts comprising of a similar range of material as seen in the rest of the sample. c) MET 01210,27 and d) MET 01210,21. Small clasts of mare basalt, impact melt, feldspathic granulites, and monomict mineral fragments are consolidated in a glassy matrix. The sections are cross-cut with variable sized fractures that have a preferred orientation (left to right as shown here). e) MAC 88105,159 and f) MAC 88105,47; large impact melt clasts and ferroan anorthosite clasts are consolidated into a glassy matrix; samples are cross-cut by large fractures that are not infilled with terrestrial contaminants. Scale bar applicable for all images is 2 mm. We determined the major and minor element concentrations of the mineral phases present in these sections using a Cameca SX50 wavelength dispersive electron microprobe (EMP) at the NHM. For silicate, sulfide and metal analysis the Cameca was operated at a 20 keV accelerating voltage with a 20 nA beam current with a focused beam analysis for 10 to 30 s count time per element. X-ray maps and clast bulk composition analyses were made by energy dispersive spectrometry using a JEOL 5900LV SEM at the NHM, fitted with an Oxford Instruments INCA energy dispersive spectrometer (EDS) X-ray microanalyzer system and operated at a 20 keV accelerating voltage and 2 nA beam current. All WDS and EDS investigations employed well characterized natural and synthetic standards as used in previous studies by Joy et al. (2006b, 2008). We identified and classified impact lithologies according to the textural classifications of Stöffler et al. (1980) where impact melt is defined as clast poor with a glassy or crystalline matrix, and impact melt breccias are defined as clast rich breccia with either a particulate, glassy or clastic matrix. Many impact melt and impact melt breccia clasts are cracked and/or very fine grained and it is difficult to directly measure spot points of individual mineral and glass phases using EMP analysis. As such, it is challenging to consistently calculate every bulk clast composition using a modal recombination technique or by using an average of matrix-corrected gridded spot analyses. We therefore chose to estimate clast major element composition using an EDS digitally controlled raster beam analysis (RBA). Several RBA were made for each clast, from the same region of interest (for the exact numbers of analyses per clast please refer to Table S1). X-ray spectra were collected from each digitized pixel of a selected region (polygon) of the clast, avoiding where possible pore spaces, fractures that were potentially infilled with terrestrial contaminants, surface contaminants etc. The accumulated X-ray counts were added together and in-built system matrix corrections performed on the total counts to derive element atomic abundances. Analytical errors per element for a single RBA analysis are <0.05 wt% (1σ) and typically 5 wt% (e.g., SiO2, Al2O3, CaO, MgO, FeO) is typically 0.3–2%. For oxides with 0.5–5 wt% (e.g., MgO, FeO, TiO2) the relative standard deviation ranges from 5–10%. For oxides with 20% and for K2O, Cr2O3, and MnO is sometimes >80%. Where the relative standard deviation was >100% we removed the value from the reported bulk clast composition. We acknowledge that there are problems associated with the raster beam analysis technique (i.e., the possible overestimation of elements concentrated in the less-dense phases, and underestimation of elements concentrated in the denser phases, phase edge effects and data acquired from sample cracks and other non-planar surfaces: Albee et al. 1977; Warren 1997; Lindstrom 1999) so, we have corrected the data set using the methodology described by Warren (1997) to account for unequal host phase density effects. This correction method has also been used by Arai and Warren (1999) and by Cohen et al. (2004) and Warren et al. (2005) to report the approximate bulk compositions of impact melt assemblages in QUE 94281 and Dhofar 026, respectively. The corrected RBA clast bulk compositions are shown in Table S1. We measured the concentration of minor and trace elements in individual clasts by time-resolved analysis LA-ICP-MS (laser ablation inductively coupled plasma mass spectrometer) at UCL/Birkbeck with a similar instrumental setup to that described in Joy et al. (2008). We utilized a New Wave 213 aperture imaged frequency quintupled Nd:YAG laser ablation system (213 nm) coupled to an Agilent 7500a quadrupole-based ICP-MS with a shield torch to reduce polyatomic interferences. We operated the laser source with a pulse frequency of 10 to 20 Hz set at 50 to 60% efficiency (conditions dependent on whether the sample was a thick thin section or a block). Data were collected for 55 s, during which time the abundances of 35 to 38 elements was monitored by repeatedly sweeping the mass spectrometer over the intended mass range. Instrumental background levels were established by a "gas blank"; that is, analysis of the mixed He optional gas and Ar carrier gas with the laser off for 30 s. We then ablated the sample for 25 s. The system was purged with He for 30 s between analyses and we waited for at least a further 2 min before proceeding to the next analysis. Data were reduced using the GEMOC Glitter software programme (http://www.glitter-gemoc.com/) where plots of counts per second versus time were examined for each element per analysis, and integration intervals for the gas background and the sample analysis were selected manually. There is a potential risk that in 3-dimensions the laser can penetrate to underlying clast/mineral phases or the glass slide mount, however, this problem is overcome by careful monitoring and selection of the time-resolved signal. Calcium (43Ca) was used as an internal standard, using CaO abundance in clasts by EMP/RBA (Table S1). Analyses were calibrated with NIST 612 external standard measurements (a synthetic doped glass: Pearce et al. 1997) of the same size dimensions. Repeatability of the NIST 612 standard measurements has a total relative standard deviation range of between 1 and 12% for all elements analysed and was typically between 2 and 5%. Accuracy was assessed by comparing our repeat NIST 612 measurements to the Pearce et al. (1997) NIST 612 values, where the relative error was <4.5% and typically <2% (i.e., <0.75 ppm absolute error). The LA-ICP-MS methodology, whilst typically used to measure minor and trace element concentrations in individual mineral (Joy et al. 2006b, 2008; Schnare et al. 2008; Haloda et al. 2009) and glass phases (Norman et al. 1998; Neal and Kramer 2006), also provides an opportunity to measure the bulk composition of polymineralic clasts, multiphase matrix and mesostasis components within small samples mounted into polished thick thin-sections or thick section blocks (Consolmagno et al. 2004; Bland et al. 2005; Joy et al. 2008). Although partially destructive in nature, the technique does not require the whole sample to be disaggregated, ensuring that any untargeted phases in the sample can be studied in the future. Analyses of clast trace element concentrations were performed in situ. Ablated sample material was removed in either several elongated tracks (100–200 μm long) or circular pits depending on the clast size (Fig. 2). Track and circular pit width was 55–80 μm. Ablated masses were removed from different portions of the same clast (within the area previously studied by EMP RBA techniques), to provide a representative study of the clast's bulk composition rather than just the composition of a limited spatial area (i.e., see Fig. 2). Results for each track were averaged together and the errors listed in Table S1 represent the two standard deviation variation of all data collected within a single clast, and do not represent analytical errors. Thus, clasts with large element errors represent rocks that were more mineralogically heterogeneous or coarsely grained (i.e., one measurement may be rather unrepresentative of the total clast composition, increasing the elemental variability measured by this study), and smaller errors indicate finer grained (i.e., all phases were easily sampled) or more homogenous material. These standard deviations are typically much greater than analytical errors (see above for the range of system precision and accuracy). Impact melt clast in PCA 02007. a) Backscatter electron image before laser ablation analysis showing heterogeneous texture of plagioclase grains included in a matrix of plagioclase and mafic melt. b) Reflected light image of the same clast after trace element investigation by LA-ICP-MS, clearly showing laser tracks of ablated material. Scale bar is 300 μm in both images. L1, L2 and L3 illustrate the three laser ablation pits made in this clast. G = granulized clast, I = impact melt clast, Plag = relict plagioclase clast. We note that TiO2, MnO, and Cr2O3 were measured by both the RBA technique and the LA-ICP-MS technique (Tables 2 and S1). These three elements generally occur at low levels (<1.2 wt% and typically 500 μm in size (Fig. 1a). This large grain size is proportionally higher than that of material in the Apollo 16 breccias studied by McKay et al. (1986) and is evidence that DaG 400 represents material derived from a comparatively more immature burial environment. DaG 400 comprises a mixture of lithic and mineral fragments (Fig. 1a) that are fused in a glassy pale gray matrix. Two predominant types of rock fragments occur in DaG 400 (1) clasts that formed as a result of impact melting processes and (2) lithic clasts that crystallized in a non-impact igneous process, which have then been subsequently altered by secondary thermal processes (either shock or regional metamorphism causing granulisation, annealed grain boundaries or shock deformation textures). These igneous lithic clasts have mineral compositions (Fig. 4) indicative of originating from ferroan anorthosite (according to the classification scheme of James et al. [1989] and Floss et al. [1998]) and more magnesian anorthosite bedrock environments. Mineral compositions of lithic clasts and mineral fragments in DaG 400, MET 01210, and PCA 02007. a) Pyroxene compositions in the meteorite samples. b) Histogram of Fo (forsterite) content of olivine grains. c) Anorthite composition range in the stones. Impact clasts contribute to approximately 40% of the DaG 400 section. Of these, most are identified as microporphyritic crystalline impact melt breccias containing a clast component (dominantly of plagioclase with minor olivine) formed from either disaggregated grains or grains that have a rounded/sub-rounded grain boundary as a result of resorption into the impact melt (Simonds et al. 1973; Powell et al. 1975). These clasts (Figs. 3a and 3b, see also Cohen et al. 1999) are bound by an amorphous mafic melt matrix. Impact melt with a crystalline texture is rare in our section (Fig. 3c), although has been reported in other DaG 400 studies (Cohen et al. 1999; Warren 2005). Regardless of texture, all of the impact melt and breccia clasts in DaG 400 have bulk major element compositions that can be classified as feldspathic (with 26–31 wt% Al2O3; Fig. 5 Table S1) consistent with normative gabbroic/noritic/troctolitic anorthosite. These clasts are within the compositional range of the feldspathic lunar meteorites (Fig. 6) and have a limited range of bulk Mg# compositions (Mg#63-67: Table S1). Backscatter electron (BSE) microscope images of impact melt clasts. DaG 400. a) Clastic melt with large inclusions of plagioclase and minor olivines with rounded grain boundaries from thermal erosion in the melt. b) Clastic melt small clasts of plagioclase in a much finer melt matrix with small plagioclase and crypto-pyroxene melt. c) Non-clastic quenched clast with very small melt crystals radiating from nucleation points. MET 01210. d) Clastic impact melt breccia with different sized inclusions of plagioclase grains. e) Devitrified microcrystalline impact-melt clast. f) Intersertal melt clasts with plagioclase laths ophitically enclosing mafic glassy melt pockets. PCA 02007. g) Non-clastic melt clast with tabular plagioclase laths in a mafic glass melt matrix. h) Large intersertal melt clast with plagioclase laths enclosing mafic glassy melt pockets similar to the "reannealed feldspathic" clast shown in figure 8n of Day et al. (2006). i) Large quenched feldspathic impact melt glass fragment. MAC 88104/05. j) and k) Basaltic microcrystalline melt clasts. l) Clastic single generation melt with different size inclusions of plagioclase chadacrysts. Plag = plagioclase, Pyx = pyroxene, Dotted outline denotes clast boundary. Scale bar represents 100 μm. Bulk Sc versus Al2O3 compositions of impact melt and melt breccia clasts in this study (Tables 1, 2, and S1) compared with bulk composition of lunar meteorite samples (taken from range of literature sources including the online supporting material table A3.9 of Wieczorek et al. [2006]; Sokol et al. [2008]; and data compiled by Kevin Righter for the "Lunar Meteorite Compendium"http://curator.jsc.nasa.gov/antmet/lmc/index.cfm) and mean composition of Apollo impact melt groups (mafic groups listed in Jolliff [1998] and feldspathic groups listed in Korotev [1994]). Meteorite bulk sample compositions (Table 1) are shown as large symbols with white text and black background (D = Dag 400, T = MET 01210, P = PCA 02007 and M = MAC 88104/05). Average meteorite impact melt breccia compositional groups (Table 2) are shown as large symbols with black text and white background (Av. IMB). Individual impact melt data points (Table S1) are shown as small versions of the same symbol with black text and white background. Error bars for Sc are 1σ error as listed in Table 2. Al2O3 values of individual impact melt clasts (Table S1) and compositional melt groups (Table 2) are "corrected" raster beam analysis values to account for the unequal host-phase density effect described by Warren (1997). The Apollo 16 compositional groups (grey bars at right of image) show the range of the four impact melt compositional groups as listed in Fig. 1 of Korotev (1994). KREEP symbol denotes high-K KREEP concentrations listed by Warren (1989). MAN symbol = Magnesian Anorthosite Suite as defined by the bulk composition of Dhofar 489 lunar meteorite (Korotev et al. 2006). The trend of rock fragments from sample 65713 is taken from Jolliff and Haskin (1995). The Apollo 16 melt trend is a linear fit through the Apollo 16 impact melt group average compositions listed in Korotev (1994) and Jolliff (1998). Normative mineralogies are based on CIPW-norms of impact melt clasts in this study where An = anorthosite normative; G/N/T An = gabbroic/noritic/troctolitic anorthosite normative; An N = anorthositic norite normative; and Norite = norite normative. Bulk compositions of meteorites and impact melt and melt breccia clasts in DaG 400, MET 01210, PCA 02007 and MAC 88104/05 (Tables 2 and S1). Symbols are the same as Fig. 5. Error bars on the composition "groups" are the 1σ error as listed in Table 2 and represent clast compositional heterogeneity. a) Plot of Sc/Sm versus TiO2 (where TiO2 is taken from LA-ICP-MS TiO2 values); b) Sc/Sm versus Mg#; c) Eu/Sm versus Sc (ppm); d) Cr (ppm) versus Mg#; e) Sc versus Sm [n.b. Sc is highly compatible in the pyroxene crystal structure and can be a used a good assessment of how much mafic material has been incorporated into the melt (Jolliff and Haskin 1995). Sm is used as a measure of incompatible trace elements in lunar impact melts (Jolliff and Haskin 1995), where high ITE signatures are typically acquired from KREEP assimilation]. Impact melt clasts compositions in (e) are compared with the rocks identified by McKinley et al. (1984) and Korotev (1994) as group 1 (A16-G1) Apollo 16 "LKFM" varieties and KREEPy basalts; group 2 (A16-G2) VHA basalts, impact-melts splashes and bombs; group 3 (A16-G3) Anorthositic basalts, FAN rocks; group 4 (A16-G4) Feldspathic, fragment laden, melt breccia, intergranular breccias and feldspathic microporphyritic melt breccias.]; f) Sm versus Thcn/Smcn (where the magnesian granulitic breccias and Apollo 16 ferroan noritic anorthosite fields are taken from figure 16a of Korotev et al. [2003]). Bulk compositions of Apollo and meteorite samples are taken from a variety of references including those listed in Fig. 5 and in BVSP (1981); Taylor et al. (1991); Papike et al. (1998), the "mare basalt database" that is part of the electronic appendix of Jolliff et al. (2006) and the online supporting material table A3.1 to A3.12 of Wieczorek et al. (2006). FAN symbol represents sample 15415 "genesis rock" (Ryder 1985) and 60025 (Haskin et al. 1973). Other sample composition references are as described in Fig. 5. Cold desert feldspathic lunar meteorites typically have bulk CaO/Al2O3 ratios of 0.6). Six clasts in DaG 400 (Feldspathic 1-3, 5, 7-8: Table S1) appear have CaO/Al2O3 ratios of <