Apatite was analyzed by electron microprobe in 3 cumulate and 10 basaltic eucrites. Eucritic apatite is fluorine-rich with minor chlorine and hydroxyl (calculated by difference). We confirmed the hydroxyl content by measuring hydroxyl directly in apatites from three representative eucrites using secondary ionization mass spectroscopy. Overall, most eucritic apatites resemble fluorine-rich lunar mare apatites, but intriguing OH- and Cl-rich apatites suggest a role for water and/or hydrothermal fluids in the Vestan interior or on other related differentiated asteroids. Most late-stage apatite found in mesostasis has little hydroxyl or chlorine and is thought to have crystallized from a degassed magma; however, several apatites exhibit atypical compositions and/or textural characteristics. For example, the isotopically anomalous basaltic eucrite Pasamonte has apatite in the mesostasis with significant OH. Apatites in Juvinas also have significant OH and occur as veinlets crosscutting silicates. Euhedral apatites in the Moore County cumulate eucrite occur as inclusions in pyroxene and are also hydroxyl-rich (0.62 wt% OH). The OH was confirmed by SIMS analysis and this apatite clearly points to the presence of water, at least locally, in the Vestan interior. Portions of Elephant Moraine (EET) 90020 have large and abundant apatites, which may be the product of apatite accumulation in a zone of melt-rock reaction. Relatively chlorine-rich apatites occur in basaltic eucrite Graves Nunataks (GRA) 98098 (approximately 1 wt% Cl). Particularly striking is the compositional similarity between apatite in GRA 98098 and apatites in lunar KREEP, which may indicate the presence of residual magmas from an asteroid-wide magma ocean on Vesta.
Eucrites are plagioclase pigeonite meteoritic basalts, gabbros, or gabbroic cumulates that sample one of the oldest planetary crusts, approximately 4.5 Ga (Misawa et al. 2005). While most eucrites are thought to be derived from the asteroid 4 Vesta (Consolmagno and Drake 1977; Clayton 1993), some eucrites (for example, Pasamonte and Ibitira) have distinct O isotopes and may be from different parent bodies (Greenwood et al. 2005; Mittlefehldt 2005; Scott et al. 2009). Basaltic eucrites can be separated into two groups or trends based on bulk composition. Stannern trend eucrites are enriched in incompatible elements (e.g., Ti, Hf, La, etc.) compared with Nuevo Laredo trend/main group eucrites, which define a crystal fractionation trend (Fig. 1, see BVSP 1981). Much debate has centered on the petrologic conditions that caused the two groups/trends. One model involves crystallization from a magma ocean on Vesta (e.g., Righter and Drake 1997; Takeda 1997; Warren 1997; Barrat et al. 2007; McSween et al. 2011), with assimilation of crustal material to explain the Stannern trend. Another model involves varying degrees of partial melting of a mantle source and crystal fractionation (e.g., Consolmagno and Drake 1977; Stolper 1977; Mittlefehldt 1979; Mittlefehldt and Lindstrom 2003), with the Stannern trend eucrites forming by low degrees of partial melting (approximately 5%). Eucrites have also been thermally metamorphosed to varying degrees and classified into seven different metamorphic grades/types based on pyroxene chemistry and petrography: eucrites with type 1 pyroxenes are the least metamorphosed, and eucrites with type 7 pyroxenes are the most metamorphosed (Takeda and Graham 1991; Yamaguchi et al. 1996). Cumulate eucrites generally have type 6 or 7 pyroxenes (Yamaguchi et al. 1996). This thermal metamorphism could have been caused by internal heat escaping by conduction through the crust, lava flows burying existing crust, and/or burial by ejecta blankets of impactors (e.g., Yamaguchi et al. 1996).
Eucrites are characterized as being volatile-depleted (e.g., Papike 1998). For example, in a plot of K versus Th (Fig. 2), eucrites plot near the line for K/Th = 1000 (Mittlefehldt 1979; Barrat et al. 2007) and well below the lines that characterize the earth (K/Th = 2570; Newsom 1995) and Mars (K/Th = 5500; Peplowski et al. 2011), but above the line for surface lunar rocks (K/Th = 360; Peplowski et al. 2011). Thus, the eucrite parent body appears to be more volatile-depleted than the terrestrial planets, but richer in volatiles than the Moon. The causes of these differences relative to the terrestrial planets are poorly understood (Papike 1998).
The abundance and nature of volatiles are an important bulk compositional characteristic of planetary bodies. For instance, volatile species dissolved in magmas (e.g., H2O, F, and Cl) are important because they influence phase stability, magmatic densities, and melting temperatures. Fluorine and Cl are moderately volatile elements that probably condensed in fluorapatite (approximately 740K) and sodalite (approximately 950K), respectively, from the solar nebula (Lodders 2003). Consequently, F is thought to be slightly more volatile than Cl in the solar nebula. However, it is important to note that in vapor-saturated magmatic systems, Cl is concentrated in the vapor, whereas F is concentrated in the silica melt (Boudreau and McCallum 1989; Brenan 1993; Carroll and Webster 1994). In this situation, Cl is more volatile than F. Because eucrites are igneous rocks, and thus crystallized from a melt, the latter behavior of the halogens may be important in understanding the genesis of eucrites.
Eucrites contain two phosphates—merrillite and apatite. Merrillite (Ca9NaMg(PO4)7) is generally volatile-free and its presence signifies a volatile-depleted system (Patiño-Douce and Roden 2006). Apatite (M10(ZO4)6(X)2 where M = Ca, Na, Sr, Pb; Z = P, Si, As, V; and X = OH, F, Cl) is a volatile-bearing phase, and it is a ubiquitous, yet trace, mineral in terrestrial, lunar, and Martian igneous rocks (see summary in Patiño-Douce and Roden 2006 and references therein), chondrites (Jones et al. 2011), and eucrites (Delaney et al. 1984). Both phosphates generally occur as small (>10 μm) interstitial grains or in late-stage melt pockets. While the halogens and H2O do not partition equally into apatite, the composition of the X site of apatite is a function of the volatile composition of the coexisting melt (Boudreau and McCallum 1989; O'Reilly and Griffin 2000). Apatite compositions have been used to infer volatile compositions of lunar basalts and the lunar interior (e.g., Boyce et al. 2010; McCubbin et al. 2010a, 2011), but volatile partition coefficients are limited for basaltic compositions (Mathez and Webster 2005) and theoretical calculations suggest that the partitioning is complex (Patiño-Douce et al. 2011). Apatite textures can also give insight into magmatic conditions. Stubby (euhedral) apatite generally forms if a magma is fluid saturated, and acicular apatite generally forms when a melt is quenched (Wyllie et al. 1962). Euhedral apatite included in early crystallizing phases crystallized relatively early, probably due to build up of P concentrations in the melt along crystallizing pyroxene grain boundaries due to slow diffusion of P in melts (Zhang et al. 2010) and its exclusion from pyroxene, whereas apatite associated with sulfides, ilmenite, and other late-crystallizing phases in mesostasis probably crystallized relatively late. Thus, the texture of apatite can also be used to make qualitative inferences about the volatile composition of basaltic magmas during crystallization. Here, we report apatite textures and compositions in eucrites with the aim of characterizing the volatile abundances in eucritic basalts and their parent asteroid, Vesta.
Basaltic and cumulate eucrites were selected based on four criteria.
Basaltic and cumulate eucrites were selected. We chose eucrites that have representative bulk rock Hf and Sc contents. These two elements are indicative of eucrite bulk rock composition, as the distinct Stannern and Nuevo Laredo trends for basaltic eucrites can be recognized in plots of Hf versus Sc (e.g., Mittlefehldt and Lindstrom 2003) (Fig. 1). Thus, using Hf and Sc abundances as guides to eucrite composition, Stannern and Lewis Cliff (LEW) 88010 were selected because they are representative of the Stannern trend. Main group-Nuevo Laredo trend eucrites (LEW 88009 Juvinas, Pasamonte, EET 87542, GRA 98098, and Nuevo Laredo) were selected to be representative of the range of Hf and Sc contents displayed by meteorites of this group. The cumulate eucrites ALH 85001, Serra de Magé, Moore County, EET 87548 were also selected (Fig. 1).
All eucrites selected (except Pasamonte) are monomict breccias or unbrecciated, to simplify interpretation of bulk rock and mineral compositions.
Some eucrites are falls (e.g., Stannern, Nuevo Laredo, Pasamonte, Moore County, and Serra de Magé) and thus have minimal terrestrial weathering.
Eucrites with a variety of metamorphic grades, between 2 and 6 on the Takeda and Graham (1991) and Yamaguchi et al. (1996) thermal scale, were selected to examine if a relationship exists between metamorphic grade and volatile composition of apatite.
Electron Microprobe (EMP)
We followed the analytical methods of Schrader (2009) and Patiño-Douce et al. (2011) for electron microprobe (EMP) analyses using a JEOL 8600 electron microprobe at the University of Georgia. Typical operating conditions were a 15 kV accelerating voltage and a 5 nA beam current. Grain size permitting, a 5 μm defocused electron beam was used, but most apatites were too small for a defocused beam. Thus, either a focused electron beam (approximately 1 μm) or a beam raster (approximately 1–3 μm) was used. Secondary Si and Fe fluorescence nevertheless occurred during many analyses. Thirteen elements (P, Ca, Fe, Si, Na, Mg, S, As, Sr, Y, Ce, F, and Cl) were measured in apatite, as our intent was to fully characterize the apatites and to understand ionic substitutions. Synthetic and natural standards were used for EMP work; they are as follows: P – Wilberforce apatite, Ca – Durango apatite, Fe – hematite, Si – diopside, Na – plagioclase, Mg – olivine, S – pyrite, As – InAs, Sr – SrTi, Y – garnet, Ce – CeO2, F – Wilberforce apatite, Cl – halite. After the 13-element routine was performed on select apatites from all meteorites, a shorter routine (consisting of P, Ca, Fe, Si, F, Cl; same count times as before) was used to collect additional data. This shorter analytical routine was used to limit electron beam time on the apatites (Stormer et al. 1993), so as to increase the number of analyses per day, and because Na, Mg, S, As, Sr, Y, and Ce were consistently below detection limits (Table 1).
Table 1. Representative apatite compositions measured by electron microprobe
Accurately analyzing apatite can be difficult due to halogen migration (Stormer et al. 1993; Henderson 2011; Goldoff et al. 2012) and multiple substitutions for P (e.g., Si, C, S, As) and Ca (e.g., Fe, Y, REE, Mg, Sr, Na) (Pan and Fleet 2002). To ensure accurate analyses we used the analytical routine of Schrader (2009) and Patiño-Douce et al. (2011), which measures the halogens first to minimize halogen migration (Piccoli and Candela 1994). An analytical screen described by Patiño-Douce et al. (2011) was used to reject EMP apatite analyses that are not stoichiometric. Acceptable analyses normalized to 25 oxygens (one formula unit) must have between 9.7 and 10.1 cations in the Ca site (Ca, Y, Ce, Sr, and Fe), between 5.8 and 6.1 cations in the P-site (P, Si, and As), less than 2.05 anions in the x-site and the analytical total must be between 98.5 and 100.5 (see online supplement for all acceptable analyses).
In apatite, the X site may be occupied by F and Cl, as well as H2O, OH, O, CO3, and Br (Pan and Fleet 2002); or the site can be vacant (Pan and Fleet 2002; Mason et al. 2009). The most common of these latter anions in terrestrial and lunar basalts is OH (Piccoli and Candela 2002; Boyce et al. 2010; McCubbin et al. 2010a; Greenwood et al. 2011). Thus, in this paper, we assume that the missing anion in the x-site is OH, and calculate the amount by difference from the sum of F + Cl. We check the validity of this assumption by means of SIMS analyses of selected apatite grains.
Because this paper focuses on the volatiles, careful attention was given to analyzing F and Cl. Some writers have suggested that repeated exposure to an electron beam permanently affects fluorapatite composition (Stormer et al. 1993; Henderson 2011). McCubbin et al. (2011) described in detail an apparent time-dependent increase in fluorine X-ray intensities during EMP analyses. These writers argued that F count rates increase during an analysis, and that count rates will stay elevated if the same spot is analyzed again. This suggests that F might migrate up the crystal lattice toward the surface of the apatite crystal. Using fluorapatite (i.e., Durango and Wilberforce) as a standard could therefore be problematic. If standards are permanently damaged by exposure to the electron beam and as a result have an elevated F concentration at the surface of the crystal, one would underestimate the F content in an unknown. This underestimation of F content in an unknown would cause an overestimation of OH calculated by difference. To ensure that our F analyses are accurate, a second Durango apatite grain was analyzed as an unknown during each analytical session. To address possible time-dependent increase in F X-ray intensities, F was measured first with a relatively low beam current (5 nA) for 10 s (to minimize F migration) and an analytical screen was used to reject apatite analyses with an excess of Cl + F relative to what can be accommodated in the anion site. With these protocols in place, we tried to minimize the potential errors in F analysis of apatite unknowns. Secondary ionization mass spectroscopy (SIMS) was used as a final check to ensure accurate OH, F, and Cl abundances.
Rare-Earth Elements (REEs) in Apatite
Rare earth elements were measured with a 25 nA focused beam (1 μm). Only P, Ca, Ce, and Y (as a proxy for the heavy REE) were measured because the goal was to determine if the REEs were present in significant amounts in apatite from eucrites. Count times for Ce and Y were increased to five minutes on the peak and background positions while keeping 10 s count times for P and Ca. To assess quality of the data, P and Ca concentrations were compared with previous P and Ca measurements performed with the longer analytical routine described above.
Precision and Accuracy by Electron Microprobe (EMP)
The apatite standards, Durango and Wilberforce, were analyzed as unknowns multiple times to examine reproducibility of the EMP data. For Durango apatite, measured F, Cl, P2O5, and CaO contents varied by 5.6%, 14%, 1.3%, and 1.0% (1σ). The analytical precision for OH was calculated using standard error propagation; OH varied by 15%. We assessed accuracy by comparing the compositions of Durango and Wilberforce apatite that we measured with the published values of Jarosewich et al. (1980) and the values established by the C. M. Taylor Company (Taylor standard block, information from Taylor multi-element standard documentation, by C. M. Taylor, C. M. Taylor Company), respectively (Table 1). Our analyses closely match the published values. Small grain sizes may have negatively impacted the accuracy and precision of some analyses, but our unknowns should have a similar precision as the standards.
Secondary Ionization Mass Spectrometry (SIMS)
To assess the accuracy of calculated OH values, SIMS was used to measure OH directly in samples that contained apatites large enough to be analyzed by SIMS. We used the Cameca IMS 1280 ion probe at the Woods Hole Oceanographic Institution (WHOI) for SIMS analyses. Analytical methods were based on those of Hauri et al. (2002) and Shaw et al. (2010); deviations from that method are as follows: Before SIMS analyses began, the secondary beam was centered in the field aperture using the image plate. Then the signal for 31P− was maximized with additional centering of the beam using the electron multiplier. Finally, masses 12C− and 16O1H− were checked on the electron multiplier for a homogenous signal, ensuring that the beam was not on a crack or contamination. A −10 kV sample voltage and a 190 pA primary current were used. Mass resolving power (M/ΔM) was approximately 6800, enough to differentiate between 17O− and 16O1H−. A Cs+ primary beam was rastered over a 30, 20, and 10 μm2 area and a field aperture was used to collect ions from the center of the secondary ion beam, approximately 15, 10, and 6 μm2, respectively, to minimize crater edge effects. Secondary negative ions 12C−, 16O1H−, 18O−, 19F−, 31P−, 32S−, 35Cl−, 19F16O−, and 37Cl− were collected for ten cycles; 18O− and 31P− were used as reference masses. Calibration curves were used to determine elemental concentrations and were calculated using the inversion method of Boyce et al. (2012).
Detection Limit of Hydroxyl in Apatite by Secondary Ionization Mass Spectroscopy
To obtain a low detection limit of OH, samples must be free of any contamination containing H. All samples were repolished and ultrasonically cleaned in ultrapure water to remove any remnant carbon coat or polishing residua. Once the thin sections and the standard mount were cleaned and dried, they were put into an oven (70°) for 24 h, then placed under vacuum (approximately 5 × 10−7 torr) for approximately 3 days. Following gold coating, they were placed under high vacuum (approximately 5 × 10−9 torr) for >8 h. Because the calibration curve for OH does not intersect 0, a contaminant must be present. This nonzero intercept was also seen by Boyce et al. (2010) who attributed this observation to the presence of an unknown hydrocarbon contaminant. Plagioclase and pyroxene (assumed to be anhydrous) were used to determine instrumental “blank” values. The detection limit for OH is defined by the maximum apparent OH intensities for plagioclase and pyroxene (assumed to be anhydrous) adjacent to apatites. Hydroxyl concentrations in eucritic apatite were always above the instrumental “blank” values for plagioclase and pyroxene. Matrix effects of these “blank” measurements are insignificant compared with the two orders of magnitude difference in H content of the “blank” and the eucritic apatite.
Confocal Raman Spectroscopy
Silica veins were found during EMP analyses of GRA 98089. To determine if the veins had a similar mineralogy as veins found in other eucrites (e.g., Serra de Magé), confocal Raman spectroscopy was used. Vein mineralogy can help determine whether the vein crystallized from a melt or fluid, and vein mineralogy can give constraints on temperature of crystallization. For example, if tridymite is found, then the melt/fluid must have crystallized above approximately 1000 ° C. Raman scattering was excited using a 633 nm wavelength laser and measurements were performed with a LabRam HR spectrometer equipped with a Peltier-cooled CCD detector at WHOI. The unknown silica polymorph was analyzed by focusing the laser slightly below the surface with a approximately 1 μm wide spot, under an Olympus confocal microscope. The confocal hole was set to 200 μm to constrain the analyzed volume in the vertical direction, and a slit width of 100 μm was used to obtain sufficient spectral resolution. Characteristic peaks (128 and 206 cm−1 for quartz and 401 cm−1 for tridymite) were used to identify the SiO2 polymorph (Frezzotti et al. 2012).
Scanning Electron Microscopy (SEM)
To help identify some phases, a Zeiss 1450EP variable pressure scanning electron microscope at the University of Georgia was used. Qualitative phase compositions were measured using an Oxford INCA energy dispersive detector with a 20 kV operating voltage.
Apatite Volatile Contents Measured by SIMS and EMP
SIMS analyses confirm the accuracy of volatile contents measured (F and Cl) and calculated (OH) by EMP analyses (Tables 1 and 2). For example, apatite in Moore County has 82% fluorapatite component measured by EMP and 18% hydroxyapatite component calculated by difference, compared to 81.7% fluorapatite component and 18.1% hydroxyapatite component measured by SIMS. Apatite in GRA 98098 has an 11% chlorapatite component measured by EMP and an 11% hydroxyapatite component calculated by difference compared to an 11.5% chlorapatite and a 12.6% hydroxyapatite component measured by SIMS. On the basis of these results, we infer that OH contents calculated by difference on other apatites are also reliable.
Table 2. SIMS analyses of apatites
Textural Analysis of Apatite in Eucrites
Apatite displays a wide range of crystal habits and sizes in the eucrites that we studied. Apatite ranges from euhedral to subhedral grains in Moore County to subhedral grains in most eucrites (ALH 85001, ALH 87548, Serra de Magé, Bates Nunataks [BTN] 00300, LEW 88009, EET 87542, LEW 88010, EET 90020, GRA 98098, Nuevo Laredo, Pasamonte, Stannern) to anhedral veinlets in Juvinas. A summary of meteorite textures is given in Table 3, and selected textures are discussed below.
Table 3. Petrography of eucrites
Plagioclase length (mm)
Pyroxene width (mm)
Apatite length (μm)
Apatite crystal habit
na = not applicable. Mineral abbreviations are: trd = tridymite, il = ilmenite, sp = spinel, tr = troilite, Fe-Ni = Fe-Ni metal, ap = apatite, mer = merrillite, zrn = zircon, si = silica, ol = Fe-rich olivine, crs = cristobalite, Fe = Fe metal, aug = augite, trd = tridymite.
Moore County is an unbrecciated cumulate eucrite with a Pb-Pb age of 4.484 Ga interpreted to be a crystallization age (Tera et al. 1997). It contains granular to subophitic plagioclase (0.5–1.5 mm long; An90–An93); subhedral to euhedral orthopyroxene with augite exsolution lamellae (0.2–6 mm wide); and minor amounts of tridymite, ilmenite, chromite, troilite, Fe-Ni metal, and apatite (Hess and Henderson 1949; Mayne et al. 2009). Apatite occurs as euhedral to subhedral grains (15–40 μm) included in pyroxene (Fig. 3b) and as acicular grains (100 μm long) included in interstitial tridymite grains (Hess and Henderson 1949).
Serra de Magé
Serra de Magé has a 4.399 Ga Pb-Pb age interpreted by Tera et al. (1997) to be a crystallization age. The meteorite is an unbrecciated cumulate eucrite with weakly oriented plagioclase grains (An94–96); orthopyroxene with augite lamellae; pigeonite; and minor phases including chromite, silica, Fe-Ni metal, troilite, ilmenite, apatite, and zircon (Prinz et al. 1977; Mayne et al. 2009). Apatite occurs as anhedral grains (approximately 30 μm long) in mesostasis with silica, troilite, and pyroxene (Figs. 3c and 3d). The mesostasis is irregularly shaped and approximately 10–1000 μm in maximum dimension. Serra de Magé was probably altered by aqueous fluids while on its parent body, which produced quartz veining (Treiman et al. 2004). A phase we interpret to be an Fe oxide or hydroxide with approximately 77 wt% Fe2O3 occurs in the mesostasis and as veinlets (Figs. 3c and 3d). This phase may be a replacement of troilite because troilite is often rimmed by the Fe oxide/hydroxide, veinlets of Fe oxide/hydroxide often terminate at troilite rimed by the Fe oxide/hydroxide, and material with a composition intermediate between troilite and the Fe oxide/hydroxide occurs with a similar shape as the troilite grains. This intermediate composition material has less S than troilite, but it has the same amount of Fe as the Fe hydroxide. The origin of this phase is a puzzle, but without proof of an extraterrestrial origin, we suggest that it is terrestrial alteration even though Serra de Magé is a fall.
EET 90020 is an unbrecciated basaltic eucrite with type 5 pyroxenes, indicating that this meteorite underwent significant thermal alteration (Takeda and Graham 1991; Yamaguchi et al. 2001). This eucrite contains two different lithologies, a coarse-grained granulitic lithology and a fine-grained subophitic-textured lithology (Yamaguchi et al. 2001, 2009; Mittlefehldt and Lindstrom 2003). EET 90020 also has narrow voids or vugs that are concentrated in the fine-grained lithology, but also occur in the coarse-grained lithology (Bogard and Garrison 1997; Yamaguchi et al. 2001). According to Yamaguchi et al. (2001), the vugs provide evidence that EET 90020 was a breccia prior to a metamorphic annealing event. The fine-grained lithology has a subophitic texture with plagioclase (250–800 μm long; An88–92) and anhedral pigeonite (type 5) with augite exsolution lamellae (400–700 μm wide) (Yamaguchi et al. 2001). The coarse-grained lithology has a granular texture with laths of plagioclase (150–400 μm long; An89–91) and anhedral to granular pigeonite with augite exsolution lamellae (600–900 μm wide). A 2 × 1 mm area within the coarse-grained lithology is finer grained, and consists of granular to stubby plagioclase, anhedral pyroxene, interstitial tridymite, and abundant apatite (30 μm–1 mm long; Fig. 4a). The minor phases are similar in the two lithologies, and include ilmenite, spinel, and laths of tridymite associated with each other; troilite; Fe-Ni metal; Fe-rich olivine (sometimes associated with oxides); and Ca-phosphates (apatite and merrillite) (Yamaguchi et al. 2001), although the coarse-grained lithology has significantly less minor ilmenite, spinel, troilite and metal than the fine-grained lithology. As Yamaguchi et al. (2001), we found only merrillite in the fine-grained lithology and only apatite in the coarse-grained lithology, where it is associated with fine-grained stubby plagioclase. The crystallization age of EET 90020 is unknown, but it underwent a strong reheating event at 4.49 Ga (Yamaguchi et al. 2001).
GRA 98098 is an unbrecciated granoblastic basaltic eucrite, which consists of mm-sized plagioclase (An83–93) and orthopyroxene with 1–5 μm thick lamellae of augite (Antarctic Meteorite Newsletter 22-2; Mayne et al. 2009); it has distinct white veins that Mayne et al. (2009) described as tabular silica-rich veins, and we found similar veins composed of a subequal mixture of euhedral plagioclase (100–500 μm long) and granoblastic tridymite (approximately 50 μm in the long direction) (tridymite determined by Raman spectroscopy) with minor pyroxene (Figs. 4d and 4e). Minor phases include chromite, ilmenite, troilite, apatite, Fe-rich olivine, and merrillite (Antarctic Meteorite Newsletter 22-2; Mittlefehldt and Lee 2001). Apatite occurs as subhedral to euhedral interstitial grains (20–300 μm long) that are in close proximity, but not within the tridymite/plagioclase veins (Figs. 4c and 4d). Mittlefehldt and Lee (2001) found phosphates in the tridymite/plagioclase veins, but they did not specify if the phosphate was apatite or merrillite; we found apatite adjacent to, but not within the tridymite/plagioclase. Bogard and Garrison (2003) reported a 4.49 ± 0.02 Ga Ar-Ar age for this meteorite, which they attributed to an approximately 4.48 Ga impact event that ejected GRA 98098 and other unbrecciated eucrites from Vesta.
Juvinas is a monomict basaltic eucrite with lithic clasts composed of ophitic plagioclase (An92), type 5 pigeonite (200 μm–2 mm) with augite lamellae and minor phases including silica, ilmenite, chromite, merrillite, and apatite (Harlow and Klimentidis 1980; Takeda and Graham 1991). Plagioclase has angular fractures, which often contain minor phases, including apatite. Apatite occurs as anhedral crystals filling the fractures in plagioclase and as veins along grain boundaries and crosscutting plagioclase and pyroxene (Fig. 4b). The veins often include minor pyroxene and troilite; thus, veins in Juvinas consist of apatite ± pyroxene and troilite. Juvinas has a crystallization age of 4.539 Ga (Manhes et al. 1984).
Pasamonte is a polymict basaltic eucrite with fine- to medium-grained lithic fragments set in a granulated eucritic matrix. The lithic fragments consist of type 2 pigeonite (0.1–0.5 mm wide), plagioclase (0.1–1 mm long; An80–95), and mesostasis (BVSP 1981; Takeda and Graham 1991). Mesostasis consists of tridymite, ilmenite, Fe-Ni metal, troilite, merrillite, apatite, zircon, olivine, baddeleyite, Ca-rich pyroxene, and plagioclase (Miyamoto et al. 1985). Apatite occurs as 1–30 μm subhedral to anhedral grains in the mesostasis. Apatite and merrillite were not found associated with one another. Pasamonte has a crystallization age of 4.565 Ga (Trinquier et al. 2008). Scott et al. (2009) argued that Pasamonte originated in a different parent body, because it has a Δ17O value that is anomalous for eucrites. Interestingly, Pasamonte displays secondary (hydrous?) Fe-enrichment along cracks and fractures (Barrat et al. 2011), but no Fe-enrichment was noted in the vicinity of apatite.
Representative apatite analyses are presented in Table 1, and atomic proportions of F-Cl-OH are plotted in Fig. 5. Most of the tabulated data represent a single analysis of an apatite grain, although some apatites (e.g., apatites in GRA 98098, EET 90020 and cumulate eucrites) were large enough to perform multiple analyses on them. No correlation was seen between apatite composition and whether the eucrite belonged to the Stannern trend or main group-Nuevo Laredo trend; there was also no correlation between apatite composition and metamorphic grade.
Apatites from all eucrites exhibit a range of OH and F, while Cl remains nearly constant. For the EMP data, considerable uncertainty exists in the calculation of OH by difference, derived from uncertainties in the F analyses (see Fig. 5). However, SIMS data suggest that average OH contents calculated by difference in EMP analyses are accurate (Fig. 5). Apatites from cumulate eucrites generally plot on or near the F-OH join and are essentially Cl-free although apatite from ALH 85001 has up to 0.2 wt% Cl. These apatites range from pure fluorapatite (with minimal Fe and Si) to 80% fluorapatite – 20% hydroxyapatite. Most apatites in basaltic eucrites are also F-rich and typically Cl-poor, and thus similar to the apatites from the cumulate eucrites as evidenced by a similar range in composition from pure fluorapatite to 80% fluorapatite – 20% hydroxyapatite. Apatites from Juvinas, Pasamonte, and Moore County are richest in hydroxyapatite component, 0.15–0.21 XOH, which corresponds to 0.46–0.78 wt% OH. Apatite from GRA 98098 has a remarkable composition with up to 1.0 wt% Cl and 15% chlorapatite component. Except for apatite from GRA 98098, all apatites analyzed from basaltic eucrites have 0.33 wt% Cl or less.
Substitutions in apatite are common and many of them are coupled and/or can involve vacancies (Pan and Fleet 2002). For example, divalent anions (e.g., O2−, CO32−, or S2−) can substitute in the X site coupled with REE3+ in the M site (Pan and Fleet 2002). Minor substituents, like Na, Mg, S, As, Sr, Y, and the REE (Pan and Fleet 2002) are, however, consistently below or near the detection limit in the eucrite apatites. Apatites from eucrites that are representative of the range of volatile compositions that we observed were chosen for an extended analytical routine designed to detect Ce and Y (as a proxy for HREE). These apatites were (see also Fig. 5): GRA 98098 (high Cl content), Juvinas (high OH content), and EET 90020 (low OH, Cl contents). Apatites from all three of these eucrites have similar and low Ce and Y concentrations (Table 1). Cerium is below detection limit (approximately 0.1 wt%) and Y is approximately 0.1 wt%. The low REE abundances that we determined are consistent with relatively low REE in apatite from low-Mg clasts in howardites (Barrat et al. 2012). However, measurable Fe (0–3.26 wt%, mean – 0.65 wt%) and Si (0–2.34 wt%, mean – 0.26 wt%) occur in most apatite grains. Most of the silica, and the high amounts of Fe measured, are probably due to secondary fluorescence of Si and Fe from neighboring silicates but small amounts of Fe (<1 wt.%) may substitute as Fe+2 for Ca2+ (Pan and Fleet 2002).
If Vesta accreted with chondritic abundances of volatiles, that signature is not present in eucrites (McSween et al. 2010). The volatiles could have been lost during evolution of an asteroid-wide magma ocean (e.g., Righter and Drake 1997), or the eucrite parent body could have originally been volatile-poor. Either evolutionary path would result in the volatile depletion seen in eucrites. The volatile-bearing nature of apatite gives us an opportunity to at least infer the volatile content and composition of the eucrite parent magmas.
Most eucritic apatites have igneous textures, and probably crystallized at high (magmatic) temperatures similar to those at which terrestrial apatites crystallize in basaltic magmas (Piccoli and Candela 2002). The volatile composition of the parental magma is the key aspect that we are interested in. Partitioning data are needed to determine magmatic volatile contents (e.g., Mathez and Webster 2005; Filiberto and Treiman 2009; McCubbin et al. 2012), but comprehensive data do not exist on apatite-merrillite-melt assemblages, at the pressures, temperatures, fO2 (one log unit below the iron-wüstite oxygen buffer; Stolper 1977), and bulk compositions characteristic of eucritic magmas. To complicate matters further, most apatites studied are late-stage, which means that they crystallized from residual melts of unknown compositions. Other apatites, e.g., Moore County, appear to have crystallized early, but we cannot constrain the timing of this crystallization. Absolute estimates of the volatile content of the parent eucrite melt may not be possible.
In addition to uncertainties with regard to partitioning, possible volatile loss from eucrite magmas must be taken into account. Because H2O, F, and Cl are all incompatible in basaltic melts (Workman et al. 2006), the ratios between these elements should remain nearly constant during early fractionation of olivine, pyroxene, and plagioclase, unless a degassing event occurs. Basaltic eucrites were either lava flows or shallow sills, so most eucrites (Nuevo Laredo, Stannern, LEW 88009, EET 87542, BTN 00300, LEW 88010) are probably degassed. Given the low overburden pressure in Vesta (0.25–0.3 kbar at approximately 40 km depth; calculated from ρ * g * z, where ρ is density, g is the acceleration due to gravity, and z is depth), even the cumulate eucrites, Serra de Magé and ALH 85001, may have crystallized from degassed magmas, as a magma at 40 km depth can hold approximately 1.5 wt% H2O (calculated from Dixon et al. 1995). Thus, much of the magmatic Cl and H2O could have been lost to the evolved gas (Ustunisik et al. 2011).
ALH 85001, Serra De Magé, BTN 00300, LEW 88009, EET 87542, LEW 88010, EET 90020, Nuevo Laredo, and Stannern
Apatites in basaltic eucrites BTN 00300, LEW 88009, EET 87542, LEW 88010, EET 90020, Nuevo Laredo, and Stannern as well as the cumulate eucrites ALH 85001 and Serra de Magé generally occur as small (>1–40 μm) subhedral to euhedral interstitial grains. These apatites have magmatic textures, are associated with late crystallizing phases, and they are F-rich (Figs. 5 and 6). This suggests that these apatites are primary magmatic phases and that they crystallized either from a melt that had degassed or from a melt that was initially depleted in Cl and H-species. As it is not possible to discriminate between the alternative explanations, these meteorites can tell us little about the volatile reservoir(s) on Vesta.
Moore County, Pasamonte, and Juvinas
Moore County contains euhedral apatite included in pyroxene, suggesting that this apatite crystallized early. The apatite is relatively OH-rich and could have sampled a relatively undegassed magma. Figure 6 shows the average OH, F, and Cl composition (atomic proportions) of apatite from Moore County analyzed by SIMS and EMP, with the inferred melt composition for a fluid-undersaturated melt in equilibrium with this apatite using ratios of the partition coefficients of Mathez and Webster (2005) and McCubbin et al. (2010a). This inferred melt may have had more H2O than F and more F than Cl (Fig. 6), but this conclusion is subject to large uncertainties, as discussed below.
Apatite in Moore County, Juvinas, and Pasamonte is richer in OH than apatite from any of the other eucrites that we studied. We are certain that OH contents in Moore County apatites are real, and not an artifact arising from recalculation of electron probe analyses, because they were determined by SIMS (Tables 1 and 2). The importance of this conclusion is that it unambiguously shows that at least some regions in Vesta (or in the Pasamonte parent body) still contained H2O when basaltic melts were crystallizing. Boyce et al. (2010) and McCubbin et al. (2010a) estimated apatite-melt H2O partition coefficients, DOH, from the results of Mathez and Webster's (2005) experiments at pressures of ≈2 kbar. If we applied the values of DOH (≈0.4) suggested by Boyce et al. (2010) to the relatively early crystallizing apatite in Moore County, we would conclude that this phase crystallized from a melt containing approximately 1.5 wt% H2O. Using the partition coefficients calculated by McCubbin and coworkers (0.05–0.3) yields melt H2O contents higher than this. Crystallization of a basalt with 1.5 wt% H2O requires a minimum pressure of ≈250–300 bar (the pressure at which basaltic melt becomes saturated with this amount of H2O, e.g., Dixon et al. 1995), which corresponds to a minimum crystallization depth of ≈30–40 km in Vesta. Depth would have to have been greater than this if the melt was not vapor-saturated, if the water content was higher than 1.5 wt%, or if the melt was saturated in a vapor containing significant concentrations of other volatile species (for instance, in the COHS system). A minimum apatite crystallization depth of 21–40 km is about the thickness of the Vestian basaltic crust (Ruzicka et al. 1997). Although the apatite inclusions could have crystallized at depth with relatively high water contents, we believe that it is unlikely that Moore County crystallized from a basaltic melt containing ≈1.5 wt% H2O, and that melt H2O content was probably less than this. We suggest that the effects of pressure and composition on apatite-melt partition coefficients for volatile species are not well understood, and that use of partition coefficients calibrated for a single pressure and melt composition to estimate water and halogen contents of melts that crystallized at unknown conditions is problematic. The apatite in Moore County undoubtedly points to the presence of H2O in the Vestan interior, which is an important finding, but we do not believe that it is possible at present to transform this qualitative conclusion into a quantitative statement.
The apatite textures in Juvinas are the most remarkable in any of the eucrites that we studied. Although some apatite in Juvinas occurs as anhedral grains in mesostasis, much of the apatite coats angular cracks in plagioclase and pyroxene associated with minor troilite (Fig. 4b). We have described a similar apatite texture in mantle xenoliths from Nunivak Island, Alaska (Patiño-Douce et al. 2011). The apatite ± pyroxene and troilite veins are approximately 10–30 μm wide and up to several millimeters in length. This texture suggests that these phases precipitated from a fluid, and the relatively H2O-rich apatite composition suggests that the fluid may have had significant H2O (Brenan 1993). If this fluid was at a temperature higher than the stability field of hydrous phases other than apatite, then veins composed of anhydrous minerals would form as suggested previously (McCollom and Shock 1998; Python et al. 2007; Barrat et al. 2011). Such fluids may have been present locally in the near-surface environment on Vesta.
Apatite from the isotopically anomalous eucrite Pasamonte is also OH-rich and similar in composition to apatites from Moore County and Juvinas. Texturally, the apatite occurs in the mesostasis and consequently is texturally similar to the F-rich apatites (ALH 85001 etc.) discussed above. However, the OH-rich nature of the apatite suggests that it crystallized from a magma richer in H2O than eucrites with similar apatite textures but OH-poor compositions. Pasamonte is therefore distinct from most other eucrites, in that it appears to have crystallized from a relatively H2O-rich parent magma, in addition to having a distinct oxygen isotope composition. These two characteristics may be linked: the parent body for the Pasamonte meteorite may have been inherently more volatile-rich than Vesta.
EET 90020 is as an unbrecciated eucrite originally thought to belong to the Nuevo Laredo trend/main group (Yamaguchi et al. 2001; Mittlefehldt and Lindstrom 2003; Scott et al. 2009). Recently, Yamaguchi et al. 2009) suggested that EET 90020 is an example of a “residual eucrite”—a eucrite that underwent partial melting, from which at least some of the melt escaped. The loss of a partial melt explains depletion in LREEs and other incompatible trace elements that characterize such “residual eucrites” (Yamaguchi et al. 2009). EET 90020 is also unusual because of the distribution of phosphates between the two lithologies: only REE-rich merrillite occurs in the fine-grained granulitic lithology, and only REE-poor apatite occurs in the coarse-grained lithology (Yamaguchi et al. 2001). Complicated models have been suggested to explain this complex petrographic relationship, including the residual model described above and subsolidus redistribution of REE (Mittlefehldt and Lindstrom 2003).
Yamaguchi et al. (2001) report modal abundance of apatite in the coarse-grained lithology to be 0.1 (vol%), but the apatite is clearly heterogeneously distributed and a significant concentration of apatite is located close to the fusion crust, which Yamaguchi et al. (2001) notes coats the interiors of some vugs (Fig. 4a). The apatite-rich region in the coarse-grained lithology is characterized by fine-grained equigranular plagioclase with interstitial pyroxene (Fig. 4a) and forms a thin zone, as if precipitated from either a melt or a fluid that flowed through a fracture. This could be the melt suggested by Yamaguchi et al. (2001, 2009) to have formed during heating at 4.49 Ga.
Sarafian and Marschall (2013) suggested an alternative hypothesis in which a melt-rock reaction took place in EET 90020, during which apatite accumulated along a melt-rock reaction surface. This model is favored because the high modal abundance of apatite (approximately 18% by volume) suggests accumulation of apatite rather than crystallization of a melt, because only approximately 0.5 wt% P2O5 is soluble in a typical eucritic melt (calculated from Tollori et al. ). Because apatite (rather than volatile-free merrillite) accumulated, volatiles, specifically F and H2O, must have been important in controlling the nature of the phosphate in EET 90020 (e.g., Patiño-Douce et al. 2011).
GRA 98098 is a metamorphosed basaltic eucrite. Key aspects of this meteorite are the presence of crosscutting tridymite-plagioclase (An83–93) veins (Mittlefehldt and Lee 2001; Scott et al. 2009) and relatively large, Cl-rich apatites (Figs. 4C and 6). The meteorite is slightly more sodic and Cr-rich than most eucrites, and contains approximately 3–5 times more REEs than a typical eucrite (Mittlefehldt and Lee 2001). The veins in GRA 98098 have been described as due to an infiltrating melt, but contrary to expectations, the vein material has low REE abundances (Mittlefehldt and Galindo 2002). Apatite in GRA 98098 is the most Cl-rich apatite that we studied, and its crystals are relatively large compared with most apatites in the other eucrites. We propose two possible explanations for the relatively Cl-rich composition of the apatites in GRA 98098.
The first possibility is that GRA 98098 was infiltrated by a fluid, similar to the fluid that Treiman et al. (2004) inferred to have crystallized quartz and sealed cracks in Serra de Magé. The composition of the apatite suggests that the fluid is likely to have been Cl-rich and T is likely to have been greater than 1000 ° C, as tridymite crystallized rather than quartz. This relatively Cl-rich fluid could have exsolved from a melt lower in the crust of Vesta, and could have metasomatized a portion of the overlying crust. It could also have contributed to the elevated bulk rock Na, because Na is soluble in fluids (e.g., Armellini and Tester 1993). Crack-seal quartz veins occur in Serra de Magé, but are thinner and less extensive than tridymite/plagioclase veins in GRA 98098 (Treiman et al. 2004), suggesting that the fluid infiltration in GRA 98098 was more pervasive and possibly hotter than that in Serra de Magé. Unlike typical terrestrial veins where hydrous phases are common, no hydrous phases occur in GRA 98098 or in Serra de Magé. Hence, the SiO2-rich veins either formed at high temperatures where hydrous minerals are not stable (McCollom and Shock 1998; Python et al. 2007; Barrat et al. 2011), or the fluid was water-poor. In either case, the veins would be composed of anhydrous minerals.
The second possibility is that GRA 98098 was infiltrated by a residual melt on Vesta — such melts have been suggested to be parental to silica-rich veins in a number of eucrites including GRA 98098 (Mittlefehldt and Galindo 2002). The apatite composition indicates that the putative melt was significantly more Cl-rich than other eucrite magmas that precipitated apatite (Fig. 6). If this melt was saturated with merrillite prior to its infiltration to GRA 98098, this could account for the low abundance of REEs in the vein material (Mittlefehldt and Galindo 2002), because merrillite concentrates the REE and can account for approximately 90% of the REEs in eucrites (Hsu and Crozaz 1996). In any case, the Cl-rich nature of the apatites in GRA 98098 suggests that a Cl-rich mobile phase was present on Vesta.
Eucrite Apatites Compared to Lunar Apatites
The volatile composition of lunar apatite has been extensively studied (e.g., Patiño-Douce and Roden 2006; Greenwood et al. 2008, 2011; McCubbin et al. 2010a, 2011). Lunar apatite is predominately F-rich, but subtle differences exist among apatites from various lunar rock types. Apatite from lunar mare basalts is generally more OH-rich and Cl-poor compared with apatites in KREEP-bearing rocks from the lunar highlands (Fig. 5). The F-rich apatite from lunar mare basalt has a volatile composition similar to apatite in most eucrites studied here, which means that the two bodies could have had a similar volatile content, or each of the basalts degassed to varying degrees, ending with apatites with a similar volatile content.
The unique characteristics of GRA 98098 suggest that it may represent a late-stage melt. Late-stage melts are often enriched in Si and incompatible elements (e.g., REEs, Na, P, Cl, H). GRA 98098 is also intriguingly similar to KREEP, in that it has elevated incompatible elements (Na and REEs), and large relatively Cl-rich apatites. Most KREEP is restricted to early magmatic rocks on the Moon (4.17–4.42 Ga), and its formation predates most mare basalts (Palme 1977; Nyquist and Shih 1992; Edmunson et al. 2009; Nemchin et al. 2009; Arai et al. 2010). However, some KREEP and mare basalts crystallized contemporaneously (Taylor et al. 1983). Likewise, GRA 98098 was metamorphosed and possibly ejected from Vesta by approximately 4.48 Ga (Bogard and Garrison 2003), which implies GRA 98098 and its veins crystallized at a similar time as most eucrites. As stated above, KREEP basalts and GRA 98098 have relatively elevated incompatible element abundances and similar apatite compositions, but differences exist. KREEP has approximately 10× the amount of REE in typical lunar mare basalts (Rhodes and Hubbard 1973; Warren and Wasson 1979), whereas GRA 98098 has 3–5× the amount of REEs in typical eucrites (e.g., Consolmagno and Drake 1977; Warren et al. 2009). Compared with terrestrial and lunar basalts, eucrites are depleted in REEs (Consolmagno and Drake 1977; Warren and Wasson 1979). A residual melt on Vesta might not be as rich in the REE as KREEP, because REEs are not as abundant in eucrites. Textures and mineralogy of these two lithologies are different as well; KREEP often has more late-stage minerals, like baddeleyite and zircon, whereas zircon or baddeleyite has not been described in GRA 98098. Also, GRA 98098 is texturally more equigranular, whereas KREEP is generally more ophitic.
The textural differences can be explained by the pervasive thermal equilibration on Vesta (e.g., Yamaguchi et al. 2009) and the mineralogy differences can be explained by the inherent low abundance of incompatible elements on Vesta. Finally, apatite in GRA 98098, with about 0.36% H2O, is more OH-rich than apatite in KREEP, which typically contains less than 0.1 wt% H2O, and rarely up to 0.2 wt% (Boyce et al. 2010; Barnes et al. 2012; Robinson et al. 2012). If these two lithologies formed in a similar manner, then the difference in OH contents of apatites suggests that the GRA 98098 parental melt had a higher H2O/F ratio than the KREEP parent magmas, which is consistent with the greater volatile inventory of Vesta deduced from the higher K/Th of eucrite meteorites compared with the lunar surface (Fig. 1). We suggest that GRA 98098 and/or its veins may have formed in a way similar to that of lunar KREEP.
Apatite from 10 basaltic and 3 cumulate eucrites are F-rich with minor OH and/or Cl, as shown by electron microprobe analyses and confirmed on a subset of the apatites by SIMS. No resolvable compositional difference exists between the volatile contents of apatites from the two eucrite trends (Stannern versus Nuevo Laredo), nor is there any apparent correlation between apatite composition and degree of metamorphism. Apatite in most of the eucrites that we analyzed (ALH 85001, Serra De Magé, BTN 00300, LEW 88009, EET 87542, LEW 88010, EET 90020, Nuevo Laredo, and Stannern) is near pure fluorapatite and probably crystallized from degassed magmas. A portion of EET 90020 is apatite-rich, and the apatite enrichment possibly reflects apatite accumulation from a melt in which fluorine and water were abundant enough to suppress merrillite precipitation. The volatile composition of many of the apatites measured here closely resembles that of apatites in lunar mare basalts as well as in many terrestrial igneous apatites, but not enough data are available to quantify the volatile content of eucritic basalts at this time.
Apatites in Moore County, Pasamonte, Juvinas, and GRA 98098 provide the clearest evidence for an enhanced role of H2O and/or Cl in Vesta. Apatite included in pyroxene in Moore County is relatively OH-rich and probably equilibrated with a relatively undegassed magma. The apatite in Moore County undoubtedly points to the presence of H2O in the vestian interior, which is an important and unexpected finding, but we do not believe that it is possible at present to transform this qualitative conclusion into a quantitative statement. Apatite from Juvinas is also OH-rich and occurs as crack fillings associated with troilite and pyroxene. This textural habit strongly suggests the presence of high temperature fluids in Vesta, as suggested by Barrat et al. (2011). Although Pasamonte may be from a distinct parent asteroid, its relatively OH-rich apatite further supports the presence of H2O in the interiors of at least some differentiated asteroids. Finally, apatite in GRA 98098 is relatively Cl-rich. The apatite in GRA 98098 is compositionally similar to some in lunar KREEP, and thus could have a similar origin to lunar KREEP; if that is the case, then further examples of Cl-rich apatites in relatively REE-rich eucrite hosts are likely to be discovered.
H. Marschall is thanked for his advice on diffusion. Samples were provided by the American Museum of Natural History and the Meteorite Working Group; we thank J. Boesenberg and K. Righter for assistance with the allocation of samples. We also thank C. Fleisher, N. Shimizu, B. Monteleone, and F. Klein for assistance with analytical work. D. Crowe and E. Tursack are thanked for valuable input during the writing of this manuscript. J.-A. Barrat, F. McCubbin, and A. Treiman are warmly acknowledged for their constructive reviews, and D. Mittlefehldt is thanked for his editorial handling. J. Gray, C. Schrader, and J. Boyce are thanked for providing valuable standards. This project was supported by a grant-in-aid from Sigma Xi and the University of Georgia Disability services center and the University of Georgia Geology department. The authors acknowledge that no conflict of interest exists.