An unusual clast in lunar meteorite MacAlpine Hills 88105: A unique lunar sample or projectile debris?



Lunar meteorite MacAlpine Hills (MAC) 88105 is a well-studied feldspathic regolith breccia dominated by rock and mineral fragments from the lunar highlands. Thin section MAC 88105,159 contains a small rock fragment, 400 × 350 μm in size, which is compositionally anomalous compared with other MAC 88105 lithic components. The clast is composed of olivine and plagioclase with minor pyroxene and interstitial devitrified glass component. It is magnesian, akin to samples in the lunar High Mg-Suite, and also alkali-rich, akin to samples in the lunar High Alkali Suite. It could represent a small fragment of late-stage interstitial melt from an Mg-Suite parent lithology. However, olivine and pyroxene in the clast have Fe/Mn ratios and minor element concentrations that are different from known types of lunar lithologies. As Fe/Mn ratios are notably indicative of planetary origin, the clast could either (1) have a unique lunar magmatic source, or (2) have a nonlunar origin (i.e., consist of achondritic meteorite debris that survived delivery to the lunar surface). Both hypotheses are considered and discussed.


The lunar regolith is an important boundary layer between the Moon and the surrounding space environment (Hörz et al. 1991; McKay et al. 1991; Lucey et al. 2006). At any one locality, the lunar regolith typically contains a record of diverse rock types (Korotev et al. 2003), mixed vertically and laterally by impacts, and material added to the Moon by projectiles (see Joy et al. 2012 for a summary). Interactions with the solar wind (Wieler 1998) and the galactic environment (Crawford et al. 2010) further modify the regolith.

Lunar regolith breccias (Fruland 1983) are rocks formed when the regolith was consolidated by pressure (e.g., shock, overburdening) and/or thermal sintering. They, therefore, provide a random global sampling of consolidated regolith from the Moon. These samples are not thought to have been fused by the impact cratering event that ejected them from the lunar surface into Earth-crossing orbits, because many of them have high trapped 40Ar/36Ar ratios, thought to be an indicator of sample antiquity (McKay et al. 1986; Eugster et al. 2001; Joy et al. 2011a). This suggests that they represent examples of lithified palaeoregoliths from different times in the Moon's past. Regolith breccias are, thus, time capsules: once they are consolidated into rocks, they preserve a record of ancient lunar and solar system processes. Temporally constraining this archive sheds light on different times in the Moon's past, helping to better understand the geological history of the Moon itself (McKay et al. 1986; Joy et al. 2011a), and the bombardment history of the Moon, Earth, and solar system (Joy et al. 2012).

We present here results from the serendipitous discovery of a compositionally unusual clast found in lunar meteorite MacAlpine Hills (MAC) 88105 and discuss its possible origin. MAC 88105, and its paired stone MAC 88104, are feldspathic polymict regolith breccias (Jolliff et al. 1991; Koeberl et al. 1991; Lindstrom et al. 1991; Neal et al. 1991; Warren and Kallemeyn 1991). The meteorites are composed of clasts of anorthositic igneous rocks, metaclastic granulitic clasts, impact glass and melt (Delano 1991; Taylor 1991; Cohen et al. 2005; Joy et al. 2010a), and rare mare basalt fragments (Robinson et al. 2012) consolidated by a fine-grained glassy melt matrix. The MAC 88104/05 samples have bulk compositions similar to present-day regoliths in the Outer-Feldspathic Highlands Terrane (FHT-O), including the south polar highlands area, highlands south of Tycho crater, farside far northern highlands, and feldspathic terranes surrounding Mare Australe (see fig. 11d of Joy et al. 2010a).

Sample and Methods

We were allocated a thick (100 μm) section, MAC 88105,159, by the Meteorite Working Group. The section is approximately 12 × 6 × 0.1 mm in size. We have previously analyzed the mineral chemistry of impact melt breccia clasts in the sample, and these results were published by Joy et al. (2010a). Additional mineral chemistry data are also presented here from phases in the MAC 88104,47, MAC 88105,158, and MAC 88104,48 sections. The samples were carbon coated and analyzed using the London Natural History Museum's (NHM) JEOL 5900 LV SEM fitted with an Oxford Instruments INCA energy dispersive spectrometer (EDS) X-ray microanalyzer system (20 kV, 2 nA, 1 μm beam). This technique was used to collect backscattered electron (BSE) and false-color element maps of the MAC 88105,159 section that are shown in Fig. 1. Mineral chemistry was analyzed using the NHM Cameca SX 50 electron microprobe (EMP, 20 kV, 20 nA, 1 μm beam), following the instrument setup described in full by Joy et al. (2010a). Data were checked for mineral stoichiometry and only data with analytical totals of between 98 and 102% were accepted (see supporting information).

Figure 1.

(a) Backscattered electron image and (b) false-color element maps of sample MAC 88105,159. For the false-color map, image pixels are colored to denote distribution and concentration of magnesium (green), aluminium (white), iron (red), silica (blue), titanium (pink), calcium (yellow), and potassium (cyan) (after Joy et al. 2011b). Location of the clast, which appears green as it is magnesian, is indicated with a red square inlay. Other green phases in the sample are olivine-rich clasts or single olivine mineral fragments.

We also measured olivine and pyroxene mineral chemistries using the NASA Johnson Space Center Cameca SX 100 EMP instrument using a 1 μm beam, an accelerating voltage of 20 kV, and a beam current of 40 nA following the method used by Joy et al. (2012). Long count times (200–300 s) were employed on the Mn, Ni, and Co peaks, and Co was corrected for the Fe K-β, Co K-α peak overlap. Elements were standardized to natural mineral standards and pure metals. For these high beam current settings, the detection limits were approximately 63 ppm for Mn, approximately 77 ppm for Co, and approximately 80 ppm for Ni. There is good agreement between the data acquired from the NHM and JSC Cameca SX 50 instruments (Table S1).

We measured 41 elements in a plagioclase grain, an olivine grain, and in the bulk mesostasis of the clast by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) using an Agilent 7500a quadrupole system at University College London (UCL). We operated the New Wave aperture imaged frequency quintupled Nd:YAG laser ablation system (213 nm) laser source with a pulse frequency of 20 Hz set at 75% efficiency, and with a spot size of 55 μm. Background conditions were monitored for 1 min and the sample was ablated for 30 s. Data were reduced using the GEMOC Glitter software (, 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.

Analyses were calibrated with NIST 612 external standard measurements (a synthetic doped glass; Pearce et al. 1997) using the same instrument setup. Calcium (42Ca) was used as an internal standard for the plagioclase and mesostasis analyses, using CaO abundance in clasts determined by EMP analysis (Table 1). For the olivine grain, we assessed the suitability of which element is an appropriate internal standard. We considered using 42Ca (85262 ppm in NIST 612; Pearce et al. 1997), 47Ti (48.11 ppm in NIST 612; Pearce et al. 1997), and 55Mn (38.43 ppm in NIST 612; Pearce et al. 1997) as these three elements are measured with both the EMP and the LA-ICP-MS. NIST 612 (8.4 wt% Ca) is not a good matrix match for lunar olivine (typically <0.2 wt% Ca) and using it with 42Ca for internal normalization results in low trace element concentrations (Table S2). The concentration of 55Mn in NIST 612 is also not a good match for lunar olivine (which typically has >700 ppm Mn: e.g., Papike et al. 1998; Shearer and Papike 2005; Schnare et al. 2008; Fagan et al. 2013) and using it for internal normalization results in high trace element concentrations (Table S2; note also unrealistically high Ca abundances of approximately 7 wt%). 47Ti has the most similar concentrations in the NIST 612 standard to lunar olivine (typically 40–400 ppm Ti, occasionally up to 1000 ppm Ti) and so we selected it as the element best suited to act as an internal LA-ICP-MS standard when using NIST 612 as the external standard for olivine analysis.

Table 1. Major (EMPA) and trace element (LA-ICP-MS) composition of an olivine grain, a plagioclase grain, and a mixed mesostasis area (pyroxene and K-rich glass and Ti-phases) in the clast in MAC 88105,159. Also listed is the bulk composition of the clast estimated (1) by a raster-beam EDS analysis (see Joy et al. [2010a] for details) and (2) by modal reconstruction using proportions 30% plagioclase, 28 olivine, and 30% mesostasis (Fig. 1).
PhaseOlivine in MAC 88105 clastPlagioclase in MAC 88105 clastMesostasis in MAC 88105 clastClast bulk composition
NameNHM 6NHM 2Normalized raster-beam EDS analysis(1) Normalized raster-beam EDS analysis(2) Modal recombination
An# 81.966.274.676.7
Ca426965 ± 373119677 ± 379068640 ± 2172 72147
Sc451.17 ± 0.111.15 ± 0.118.59 ± 0.81 3.37
Ti47270 ± 9533 ± 306958 ± 401 2375
V515.27 ± 0.554.70 ± 0.4514.73 ± 1.45 7.83
Cr53171 ± 2246 ± 6554 ± 75 233
Mn55175 ± 1175 ± 4355 ± 20 187
Co591.93 ± 0.190.45 ± 0.052.60 ± 0.26 1.51
Ni6013.72 ± 1.371.92 ± 0.2322.19 ± 2.41 11.29
Cu630.49 ± 0.041.20 ± 0.112.40 ± 0.20 1.35
Zn660.89 ± 0.170.63 ± 0.142.92 ± 0.6 1.38
Ga690.58 ± 0.0616.54 ± 1.4631.75 ± 2.87 16.51
Ge720.21 ± 0.020.14 ± 0.050.94 ± 0.09 0.40
Rb850.36 ± 0.041.04 ± 0.1223.61 ± 2.59 7.58
Sr888.64 ± 0.46243 ± 9135.89 ± 4.97 143.9
Y895.52 ± 0.673.56 ± 0.4479.13 ± 10.02 26.66
Zr9016.52 ± 2.046 ± 0.75619 ± 78 192
Nb931.05 ± 0.060.74 ± 0.0545.58 ± 2.12 14.212
Mo980.006 ± 0.002b.d.0.006 ± 0.003 <0.006
Cs1334.61 ± 0.5555.27 ± 6.213.4 ± 0.39 25.3
Ba1377.09 ± 0.56234 ± 16467 ± 33 239
La1391.23 ± 0.115.70 ± 0.4826.29 ± 2.23 10.563
Ce1403.30 ± 0.2513.68 ± 0.9480.41 ± 5.61 30.620
Pr1410.53 ± 0.031.50 ± 0.079.89 ± 0.41 3.726
Nd1462.48 ± 0.185.59 ± 0.3943.46 ± 2.90 15.996
Sm1470.73 ± 0.061.27 ± 0.1212.31 ± 0.95 4.408
Eu1530.049 ± 0.0042.31 ± 0.121.48 ± 0.07 1.414
Gd1570.87 ± 0.090.89 ± 0.1112.97 ± 1.42 4.487
Tb1590.14 ± 0.010.14 ± 0.022.29 ± 0.24 0.784
Dy1630.95 ± 0.110.80 ± 0.1015.31 ± 1.79 5.171
Ho1650.20 ± 0.020.13 ± 0.023.17 ± 0.36 1.055
Er1660.58 ± 0.070.33 ± 0.059.31 ± 1.07 3.080
Tm1690.08 ± 0.010.04 ± 0.011.35 ± 0.16 0.441
Yb1720.53 ± 0.060.38 ± 0.059.31 ± 0.96 3.088
Lu1750.08 ± 0.010.04 ± 0.011.21 ± 0.14 0.400
Hf1780.39 ± 0.040.18 ± 0.0314.04 ± 1.59 4.375
Ta1810.037 ± 0.0040.021 ± 0.0041.79 ± 0.15 0.554
W1820.007 ± 0.0020.10 ± 0.020.05 ± 0.01 0.057
Pb2080.09 ± 0.020.31 ± 0.062.58 ± 0.52 0.923
Th2320.21 ± 0.020.17 ± 0.027.88 ± 0.77 2.481
U2380.04 ± 0.010.16 ± 0.022.22 ± 0.30 0.743

When using 42Ca as the internal standard, repeatability of the NIST 612 external standard measurements has a total relative standard deviation range of between 0.7 and 7% for all elements analyzed and was on average 3.5%. Accuracy was assessed by comparing our repeat NIST 612 measurements with the Pearce et al. (1997) NIST 612 values, where the percentage relative difference had a range of between 0.58 and 8.56% and an average of 2.6%. When using 47Ti as the internal standard, repeatability of the NIST 612 standard measurements has a total relative standard deviation range of between 1.5 and 7% for all elements analyzed and was on average 3.6%. Accuracy was assessed by comparing our repeat NIST 612 measurements to the Pearce et al. (1997) NIST 612 values, where the percent error relative difference had a range of between 0.15 and 16% and on average 10.3%. Reported errors (Table 1) are one sigma as calculated by the Glitter software.

Oxygen-isotope compositions were analyzed in situ with the University of Hawai‘i Cameca ims-1280 ion microprobe using a technique similar to that described by Makide et al. (2009) and Joy et al. (2012). A 400 pA focused Cs+ primary ion beam was rastered over a 7 × 7 μm2 area for 100 s to remove carbon coating and any surface contaminants. Then, the raster was reduced to 5 × 5 μm2 and data were collected. The secondary ion mass spectrometer was operated at −10 keV with a 50 eV energy window. Three oxygen isotopes were collected using multicollection mode. 16O was measured on a Faraday cup, while 17O and 18O were measured with electron multipliers. The mass resolving power for 16O and 18O was approximately 2000, and that for 17O was approximately 6000, sufficient to separate interfering 16OH. A normal-incidence electron flood gun was used for charge compensation.

Oxygen-isotope analyses are reported in standard δ notation where δ18O has been calculated as: δ 18O = (([18Osample/16Osample]/[18Oref/16Oref]) − 1) × 1000, and similarly for δ17O using 17O/16O ratio. Δ17O (deviation from the terrestrial fractionation line) is calculated as δ 17O − 0.52 × δ18O.

Terrestrial standards (San Carlos olivine and Miyakejima anorthite) were used to set up the instrument and check reproducibility of our measurement protocol. To minimize any possible differences in instrumental effects associated with different sample mounts, we analyzed lunar plagioclase grains in the host MAC 88105,159 rock as an internal standard. The weighted mean of Δ17O on lunar plagioclase measurements was assumed to be Δ17O = 0, and data for the clast are reported relative to the lunar plagioclase. To verify the positions of the sputtered region, the phases studied for oxygen isotopes were imaged in secondary and backscattered electrons using the University of Hawai‘i JEOL 5900LV scanning electron microscope after ion probe measurements.


The MAC 88105,159 section is composed of a feldspathic regolith breccia with impact melt breccia clasts, anorthositic clasts, and rare basalt and granitic lithologies (Fig. 1, see also Joy et al. 2010a). We identified a magnesian lithic clast (Fig. 2) as being compositionally distinct (Mg-rich and K-rich) in a false-color element map of the sample (Fig. 1b). The clast is 400 × 350 μm in size and is transected by a 50–80 μm wide fracture that also cross-cuts the surrounding matrix (Fig. 2). The clast has a hypocrystalline texture (crystals within a glassy mesostasis groundmass) and is fine grained. It is composed of blocky subhedral olivine and plagioclase crystals trapping elongate xenomorphic pyroxenes and a late-stage glassy mesostasis (Fig. 2). There are no particles of Fe-metal present that would be indicative of an impact melt origin. Modal abundances of minerals by mode were determined using analysis of BSE and element map images (following the methods outlined in Snape et al. 2011) and these phase proportions (Fig. 2c) indicate that the clast is an olivine-gabbro. However, given the small size of the clast, this may not be representative of the parent lithology from where it was sourced. Mineral trace elements measured in the clast are plotted in Fig. 3, and major and minor element data are plotted in Figs. 4-8.

Figure 2.

Close-up images of clast in MAC 88105,159. (a) Backscattered electron image of clast. Red circles denote collection locations and size of SIMS oxygen measurements. Blue circles denote location and size of LA-ICP-MS pits for trace element analysis. (b) False-color element map of the clast (see Fig. 1b for color details). Minerals phases are denoted where Ol = olivine, pyx = pyroxene, plag = plagioclase, and ms = mesostasis. (c) Mineral distribution within the clast where blue = plagioclase, red = olivine, green = pyroxene, yellow = mesostasis glass, pink = Ti-rich phase, and white =holes or host meteorite MAC 88105,159.

Figure 3.

REE concentrations in the clast (Table 1). (a) CI chondrite-normalized REE values of plagioclase and olivine mineral grains and bulk area mesostasis (glass + ilmenite + pyroxene). REE abundances of CI chondrite were from Anders and Grevesse (1989). Also shown are the modeled (modal recombination) bulk clast composition (Table 1) and the composition of high-K KREEP (Warren 1989) for comparison. (b) Clast plagioclase REE value compared with those from the lunar ferroan anorthosite (FAN) suite (Papike et al. 1997; Floss et al. 1998), the Mg-Suite (HMS: medium gray box; data from Papike et al. 1996; Shervais and McGee 1998) and the High Alkali Suite (HAS: dark gray box; data from Shervais and McGee 1999). Error bars shown are 2 sigma.

Figure 4.

Pyroxene compositions measured in the clast plotted onto a pyroxene quadrilateral. Data are compared with pyroxene in clasts and mineral fragments in the host MAC 88105,159 meteorite and also in MAC 88104,47, MAC 88105,158, and MAC 88104,48.

Figure 5.

Minor elements in pyroxene in the clast compared with pyroxene in lunar meteorites and Apollo samples. Note that for a given Mg#, the clast pyroxene have higher concentrations of Na, Ti, and Al compared with most other lunar materials. Meteoritic data sources are as follows: MAC 88104/05, Dar al Gani 400, Meteorite Hills 01210, and Pecora Escarpment 02007: Joy et al. (2010a); North West Africa 4472: Joy et al. (2011c); Miller Range 07006: Joy et al. (2010b); Robinson et al. (2012); La Paz 02205 and pairs: Joy et al. (2006); and Miller Range 05035: Joy et al. (2008). Data for Apollo samples include feldspathic lithologies, Mg-Suite, KREEP, and mare basalts (Takeda et al. 1975; Papike et al. 1991, 1996, 1998; Shervais and McGee 2012,1999; Jolliff et al. 1999; Schnare et al. 2008; Taylor et al. 2012).

Figure 6.

Minor elements in olivine in the clast compared with olivine in lunar meteorites (see Fig. 5 caption for literature sources) and Apollo samples including feldspathic lithologies, Mg-Suite, KREEP, and mare basalts (see Fig. 5 caption for literature sources). Note that for a given Mg#, the clast olivine have equivalent Ca and Ti higher concentrations of Cr and lower FeO/MnO ratios.

Figure 7.

Average Mg# of olivine and pyroxene versus plagioclase 100 × Ca/[Ca+Na] (average 79) in the clast. Error bars show range of compositions in the clast. Note that the clast plagioclase data have been recalculated from that presented in the text, which was reported for 100 × Ca/[Ca+Na+K]. The data are compared with possibly pristine nonmare rocks listed by Warren (1993). The outer ferroan anorthosite suite (FAS) field was taken from Warren (1993). The inner ferroan anorthosite suite field and High Mg-Suite (HMS) fields outline those rocks that have high confidence of pristinity (i.e., those with pristinity values of >8: Warren 1993). The approximate boundary (dashed line) between Mg-Suite and High Alkali Suite rocks was taken from Wieczorek et al. (2006).

Figure 8.

Mn versus total Fe atoms per formula unit in (a) and (b) olivine, and (c) and (d) pyroxene in the clast. Data in (a) and (c) are compared with mafic phases in other lunar meteorites and numerous Earth rocks (taken from the PetDB database including basalts, peridotites, lherzolite, troctolites, gabbros, gabbronorite, harzbergites, etc. where reported Fe data are converted from wt% FeOtotal). Also shown are average planetary trend lines where the Moon lines are linear fits (olivine: Mn = [0.0114 × Fe]−0.0003; pyroxene: Mn = [0.0116 × Fe]−0.0038) to lunar meteorite pyroxene and olivine data as reported in Fig. 5 caption; the Earth line is taken from a linear fit (olivine: Mn = [0.0194 × Fe]−0.0015; pyroxene: Mn = [0.0309 × Fe]−0.0028) to data compiled in the PetDB database from numerous terrestrial rocks; planetary trend lines for SNC meteorites (Mars) and HED meteorites (Vesta) are from Papike et al. (2009), and ordinary chondrites (OC) and CO-type carbonaceous chondrites are from Berlin et al. (2011). In (b) and (d), data from olivine and pyroxene phases in lunar meteorite MAC 88104/05 (Joy et al. [2010a] and this study) and Apollo Mg-Suite are plotted for comparison (Papike et al. 1998; Shervais and McGee 1998).

Mineral Chemistry Results

The clast has approximately 30% (by area) zoned forsteritic olivine grains (Fo83–93; Table S1). Olivine has Ni at concentrations of <160 ppm (by EMP analysis, often less than detection limits of approximately 80 ppm; Table S1). Concentrations of other minor elements in olivine are plotted in Fig. 6 compared with a wide range of olivine in lunar meteorites and Apollo samples. Olivine grains in the clast have higher Cr concentrations (0.07–0.28 wt% Cr2O3; Fig. 6) and marginally higher CaO (0.15–0.44 wt%) and Ti (approximately 170–1000 ppm) than lunar samples with similar Mg-rich olivine (i.e., those from the Mg-Suite and KREEP basalts; Papike et al. 1998; Shervais and McGee 1998; Taylor et al. 2012).

Approximately 42% (by area) of the clast is zoned blocky plagioclase (An72–82, where An# = atomic Ca/[Ca+Na+K]; Mg# = 59–81, where Mg# = atomic 100 × Mg/[Mg+Fe]; Table S3). Plagioclase grains have a positive Eu-anomaly (Eu/Eu* = 6.6 where Eu/Eu* is calculated as Eucn/√[Smcn × Gdcn] and where cn are the chondrite-normalized values using the CI concentrations reported by Anders and Grevesse 1989) with trivalent REE at ×2cn to ×24cn (Fig. 3).

Elongate pyroxene crystals, which contribute to approximately 17% by area of the clast, are associated with olivine and plagioclase grain boundaries and cross-cut the mesostasis. These have augite compositions (Fig. 4: En45–55 Fs8–12 Wo36–45; Mg# = 79–86; Table S4). Minor element concentrations in pyroxene are plotted in Fig. 5 and show that the clast has notably higher Al, Na, Ti, and marginally higher Cr concentrations (2.5–5 wt% Al2O3, 0.19–0.32 wt% Na2O, 1.96–2.97 wt% TiO2: Fig. 5) compared with similar Mg-rich lunar pyroxene (i.e., those from the Mg-Suite and KREEP basalts: Papike et al. 1998; Shervais and McGee 1998; Taylor et al. 2012).

Fe/Mn ratios in olivine (46 ± 10; quoted error is two standard deviations) and pyroxene (23 ± 5) in the clast are significantly lower than Fe/Mn ratios in olivine (95 ± 15) and pyroxene (57 ± 13) in the host MAC 88105,159 meteorite (Figs. 8 and 9). They are also dissimilar in terms of Fe/Mn ratio to olivine and pyroxene in other lunar meteorites and Apollo samples (Figs. 6d, 8, and 9).

Figure 9.

Range of typical plagioclase composition (where An# = Ca/[Ca+Na+K]) versus olivine and pyroxene atomic Mn/Fe ratios for different planetary bodies (fields for meteorite groups, Earth and Apollo basalts taken from Papike et al. 2003). Also shown are the total range of plagioclase, pyroxene, and olivine compositions reported in lunar meteorites (references listed in Fig. 5 caption). Average composition of the MAC 88105,159 clast is plotted in red where the bars denote the range in composition plagioclase and mafic phases.

The remaining 11% of the clast is composed of a K-rich, partially devitrified, interstitial glassy mesostasis with a bulk alkali-calcic dacite composition (Table 1). Also present in the clast are small (<10 μm) accessory (approximately 0.5%) Ti-rich phases (Fig. 2). Attempts to determine their nature were hampered by their small phase size, resulting in mixed analyses with surrounding minerals. The element maps of the clast reveal that two grains are Cr-bearing, suggesting that at least two of the grains are probably Cr-bearing spinel. All the other grains are only Ti-bearing. In two cases, the ratio of TiO2/FeO measured in mixed EMPA analyses is more similar to ilmenite than Ti-rich spinel, suggesting, thus, that some of these grains are small ilmenite crystals, but this cannot be confirmed with the existing data. A mixed area of mesostasis with some pyroxene and Ti-rich phases (see Fig. 2a where the left-hand blue circle is located) has bulk trace elements with a negative Eu-anomaly (Eu/Eu* = 0.35) and trivalent REE of ×51cn to ×133cn (Fig. 3a; Table 1).

Clast Reconstructed Bulk Composition

The bulk composition of the clast is listed in Table 1. Major element composition was estimated by two independent approaches: (1) normalized raster-beam EDS analysis, where EDS X-ray spectra were collected from each digitized pixel of a selected region (polygon) of the clast. The accumulated X-ray counts were added together and inbuilt system matrix corrections performed on the total counts to derive element atomic abundances (see method of Joy et al. [2010a] for full details), and (2) modal recombination of the plagioclase (41% by area), olivine (30% by area), and bulk mesostasis region (29% by area) compositions as listed in Table 1. The bulk trace element composition was estimated using the same modal recombination approach using the phase proportions listed above, and the mineral compositions listed in Table 1.

The modeled bulk clast composition supports the observations from mineral chemistry that the clast is both magnesian and rich in alkali (volatile) and incompatible elements (Table 1). In terms of bulk SiO2 and alkalis, it is classified as a basalt. It has essentially no Eu-anomaly (Eu/Eu* = 0.97, Fig. 3a), and compared with bulk-rock MAC 88104/05 (Joy et al. 2010a), it has high trivalent REE abundances (bulk MAC 88104/05: approximately ×6cn to ×12cn; bulk clast: ×17cn to ×51cn). It has K/Th ratios (approximately 3000) that are an order of magnitude higher than the bulk lunar regolith observed from remote sensing measurements (the average lunar surface has a K/Th ratio of approximately 360: Peplowski et al. 2011; to approximately 810 in the northern farside highlands: Gillis et al. 2004), but which are similar to some rare plutonic High Alkali Suite hand specimen samples (as reported in the electronic index of Wieczorek et al. 2006).

Oxygen-Isotope Results

To investigate whether or not the clast originated in the Earth–Moon system or elsewhere, we performed in situ ion microprobe oxygen-isotope analysis of plagioclase, olivine, and pyroxene grains in the clast and compared these data with oxygen isotopes measured in the host MAC 88105,159 lunar material. The weighted mean of the host meteorite MAC 88105,159 lunar plagioclase grains (14 data points) was assumed to lie on the terrestrial fractionation line (TFL) (Δ17O = 0.00 ± 0.15, 2σ standard error: Fig. 10; Table 2). The clast oxygen-isotope data (8 data points) gave a weighted mean of Δ17O = 0.12 ± 0.20 (2σ standard error; Fig. 10). These results show that, in terms of Δ17O, the clast is statistically indistinguishable from the TFL ((0.12 ± 0.20) − (0.00 ± 0.15) = (0.12 ± 0.25)). Its weighted mean composition is also statistically indistinguishable from the average of SNC (Shergottite-Nakhla-Chassigny) martian meteorites (Fig. 10). However, the weighted mean of the clast is isotopically distinct from the average composition of HED (howardite-eucrite-diogenite) meteorites (Fig. 10) that are thought to have originated from the asteroid Vesta.

Table 2. Results of in situ oxygen-isotope studies. Data were collected from plagioclase in host meteorite MAC 88105,159 (top) and minerals phases in the clast (bottom). The weighted mean Δ17O of the lunar plagioclase was normalized to the TFL (Δ17O = 0) and then the lunar data set and clast data set were normalized by the same amount (right-hand columns). Weighted mean host MAC 88105,159 data and clast data errors are 2σ standard error (standard deviation of data divided by the square root of number of measurements).
Lunar meteorite phaseAnalysis orderΔ17O 2σΔ17O normalized to TFL
MAC 88105,159 Lunar plagioclase11.47 ± 0.580.37 ± 0.58
21.50 ± 0.500.40 ± 0.50
30.87 ± 0.55−0.22 ± 0.55
41.06 ± 0.56−0.04 ± 0.56
60.81 ± 0.73−0.28 ± 0.73
80.76 ± 0.59−0.34 ± 0.59
101.24 ± 0.610.14 ± 0.61
110.59 ± 0.64−0.51 ± 0.64
120.57 ± 0.60−0.52 ± 0.60
131.29 ± 0.460.19 ± 0.46
140.93 ± 0.59−0.17 ± 0.59
171.04 ± 0.57−0.06 ± 0.57
201.39 ± 0.550.29 ± 0.55
241.44 ± 0.660.34 ± 0.66
Weighted mean and 2σ standard error 1.10 ± 0.150.00 ± 0.15
Clast phaseAnalysis orderΔ17O 2σΔ17O normalized to lunar portion
MAC 88105,159 Clast—olivine71.38 ± 0.580.28 ± 0.60
91.27 ± 0.600.17 ± 0.62
MAC 88105,159 Clast—plagioclase151.26 ± 0.530.16 ± 0.55
MAC 88105,159 Clast—olivine161.41 ± 0.560.31 ± 0.58
MAC 88105,159 Clast—pyroxene181.11 ± 0.490.02 ± 0.51
MAC 88105,159 Clast—olivine191.48 ± 0.490.38 ± 0.51
211.05 ± 0.60−0.05 ± 0.62
220.80 ± 0.52−0.30 ± 0.54
Weighted mean and 2σ standard error 1.22 ± 0.190.12 ± 0.20
Figure 10.

Time order analysis of Δ17O oxygen-isotope analysis of phases in the lunar portion of host meteorite MAC 88105,159 (gray symbols) where the weighted mean (gray line and square gray symbol) has been normalized to the TFL. Δ17O. The 2σ standard errors of these lunar measurements (±0.15) are shown as error bars on gray square symbol. Analyses of phases in the clast are shown as red symbols where the weighted mean (normalized to the weighted mean of the lunar portion corrected to the TFL) is Δ17O = 0.12 shown as the solid red line and red square symbol. The 2σ uncertainty (±0.25) levels for the mean of the clast (including the standard error on the means for both the clast and lunar measurements) are shown as error bars on the red square symbol. These errors are appropriate to compare the Δ17O of the clast with those of average SNC (Δ17O 0.29: average of data compiled by Mittlefehldt et al. 2008), HED (Δ17O-0.22: average of data compiled by Mittlefehldt et al. 2008) and angrite meteorites (Δ17O-0.07: Rumble et al. 2008).


Petrological History

The mineral chemistry and bulk clast chemistry show that the clast is magnesian, but also rich in alkali (volatile) and incompatible-trace elements. Although we have to bear in mind that the clast itself is very small, and may not be representative of its parent lithology (Warren 2012), these are unusual characteristics of a rock sourced from a primitive melt. It implies the presence of mixing of an evolved melt component in the clast's parent melt source region or later assimilation of an evolved melt component.


The clast has mineral chemistry characteristics that differentiate it from known lunar lithologies (Figs. 5, 6, 8, and 9). In particular, the olivine and pyroxene crystals have Fe/Mn ratios that are unique compared with previously sampled lunar rocks types. The ratio of Fe/Mn in mafic minerals and bulk samples is indicative of planetary reservoirs and subsequent geological evolution of planetary bodies (i.e., volatile loss, metal segregation during core removal, oxygen fugacity, and melt fractionation: Drake et al. 1989; Papike 1998; Karner et al. 2003, 2006; Papike et al. 2003; Goodrich and Delaney 2000; Gross and Treiman 2010; Gross et al. 2011). The possible planetary sources of the clast are discussed below.

Sampling a Unique Region of Lunar Crust?

Although the olivine and pyroxene mineral compositions are not lunar-like (Figs. 5, 6, 8, and 9), other characteristics may be consistent with the clast being derived from lunar rocks. Its magnesian nature is similar to rocks from the lunar Mg-Suite, although plagioclase is alkali-rich compared with plagioclase in Mg-Suite rocks (Fig. 7). Conversely, although the clast's aluminous and alkali-rich nature is more similar to samples from the High Alkali Suite, olivine and pyroxene in the clast are too magnesian (Fig. 7). Plagioclase trace element concentrations (Fig. 3) are similar to rocks from both the High Alkali Suite and Mg-Suite. If the rock is lunar, then it shares characteristics of both these magmatic suites, although it differs from both. It is plausible that the rock represents a Mg-Suite cumulate that was infiltrated by late-stage–evolved K-rich fluids (akin to High Alkali Suite or KREEP basalt melts) to account for the alkali-rich plagioclase and trapped mesostasis Na-K–rich glass.

If the clast originated from the Moon, then an explanation is required for the nonlunar Fe/Mn ratios, in both the early formed olivine (Figs. 8a and 8b) and the later crystallized pyroxene (Figs. 8c and 8d). The following mechanisms could account for difference in Fe/Mn ratios between the clast and lunar rocks:

  1. Oxygen fugacity effects. Low Fe/Mn ratio in the clast could result from a source region with higher ƒO2 than typical lunar melts. However, no ferric mineral phases are present in the clast that would support this model. Alternatively, the low ratio could imply that the clast has experienced reduction to remove Fe from olivine and pyroxene that could have decreased both minerals’ Fe/Mg and Fe/Mn ratios, and increased Ni concentrations in the clast compared with lunar rocks. However, no metallic Fe is observed in the clast, so if reduction occurred, the resulting metal products and siderophile elements were effectively removed from the rock before it crystallized.
  2. Crystallization or fractionation effects. A decrease in Fe/Mn ratio in mafic phases could indicate that olivine was removed (fractionated) from the sample's source region, as Mn is somewhat incompatible in olivine and Fe is compatible (Humayun et al. 2004; Qin and Humayun 2008); this process could decrease the system's bulk Fe and increase the bulk Fe/Mg and decrease the Fe/Mn ratio. However, as the clast has a bulk-rock Mg# of 84–89 (Table 1), precipitating olivine in equilibrium should be Fo94–96 (calculated using equation 3 of Joy et al. 2008 and references therein). As these calculated values are similar to the most primitive olivine composition measured in the clast (Fo93), this indicates that little or no olivine was removed from the parent system and so this is not likely to be the cause of the Fe/Mn variation. Fe/Mn ratios could potentially also be lowered if Fe-Ti or Fe-Cr-Al oxides precipitated as an early phase removing Fe from the melt (Karner et al. 2003; Gross et al. 2011); however, both olivine and pyroxene in the clast are generally Ti-rich, Cr-rich, and Al-rich compared with lunar phases with similar Mg# (Figs. 5 and 6), suggesting that early oxide removal has not been extensive.
  3. Unique lunar crustal or mantle mineralogy. The lunar mantle and crust is heterogeneous, with regions that contain differing amounts of volatile elements (e.g., Hauri et al. 2011; McCubbin et al. 2011; Tartèse et al. 2013). The clast could, therefore, have been sourced from a region with higher concentrations of volatile elements. Manganese is a moderately volatile element and generally has low concentrations in lunar materials, presumably because it was volatilized and depleted during the Moon's formation by giant impact (e.g., Hartmann and Davis 1975; Papike et al. 2003; O'Neill and Palme 2008). However, in principle, as yet undiscovered relatively volatile-rich regions may exist in the lunar crust or mantle from which this clast might have been derived. An origin in such a region might also explain the relatively alkali-rich nature of the plagioclase grains within the clast.

In summary, although there are possibly mechanisms to account for the clasts's non–lunar-like Fe/Mn ratios in olivine and pyroxene, such models would also have to account for the clast's different mineral chemistry compared with known lunar rock types (Table 3). Indeed, the clast appears sufficiently compositionally unique compared with known lunar rocks that it may not be lunar at all.

Table 3. Summary of compositional and mineralogical similarities and differences between the clast in MAC 88105,159 and other planetary bodies.
ObservationInterpretationMoonEarthMars (SNC)Vesta (HED)MercuryVenusBasaltic Achondrite
Oxygen isotopesClast weighted mean Δ17O indistinguishable (2 sigma) from TFL, distinct from HED samplesYesYesYesNoUnknownUnknownYes
Bulk K and Th valuesWithin range of lunar HAS rocks, higher than HMS rocks. Within range of terrestrial Archean rocks. Higher than terrestrial oceanic crust. Higher than HED and SNC samples. Higher than that recorded Mercury surface regoliths and Venus crust.Uncommon, but like some HAS samplesYes for Archean rocks, No for oceanic basaltic crustNoNoNoNoMaybe
Magnesian mafic mineralsMore magnesian than HED and SNC samples studied to date. Similar to lunar Mg-Suite and terrestrial basalts/Archean rocksYes like lunar Mg-Suite rocksYesNo NoUnknownUnknownYes
Olivine FeO/MnO ratiosNonlunar olivine FeO/MnO ratiosNoOl: Overlap but not same trendOl: YesNoUnknownUnknownYes
Pyroxene FeO/MnO ratiosNonlunar pyroxene FeO/MnO ratiosNoPyx: Overlap but not same trendPyx: NoPyx: NoUnknownUnknownYes
Plagioclase An#81, Eu × ~40cnHigher than equivalent An# plag. in terrestrial basalts and SNCs, but similar to lunar HAS Suite and HEDs (Karner et al. 2004)Yes—within range of lunar HAS Suite and phases in late-stage fractionates of some mare basaltsYesNo, too calcicYesUnknownUnknownMaybe
Sodic augitesNot-HED affiliated or martian; unusual for the Moon, but some overlap with terrestrial pyroxenesUnusualYesNoNoUnknownUnknownMaybe
Olivine low-Ni (<150 ppm)Suggest nonterrestrial origin (Archean and modern day oceanic basalt olivines typically >600 ppm Ni)YesNoYesYesUnknownUnknownMaybe

A Nonlunar Origin?

As the Fe/Mn ratios of the olivine and pyroxene are not lunar-like (Figs. 8 and 9), it is plausible that the clast may have been sourced from a different parent body and survived delivery to the Moon as impact debris. Meteoritic debris have previously been identified on the Moon as rare samples found in the lunar regolith (see Joy et al. 2012 for a summary).

Compositional and isotopic constraints for a meteoritic origin, and potential parent bodies, are listed in Table 3. Olivine grains in the clast have nonlunar Fe/Mn ratios that are more similar to trends in martian meteorites, terrestrial samples, and some chondrules in ordinary chondrites (Fig. 8). However, we do not consider that the clast is a chondrule relic as there (1) are no Fe-metal, sulfide, Al-rich spinel, or nepheline grains present, indicative of plagioclase-rich chondrules; (2) the clast bulk MgO/Al2O3 ratio (approximately 0.8) is lower than bulk chondrules (typically >>1.5: McSween 1977), and clast bulk MgO/TiO2 is typically lower (<0.21) than in chondrules (typically >>45: McSween 1977); and (3) plagioclase grains in the clast are a lot blockier than found in plagioclase-rich chondrules (Krot et al. 2002). The clast pyroxenes have Fe/Mn ratios that are distinct from most basaltic achondrite groups, although are within the spread of terrestrial pyroxene data (Fig. 8).

Additional constraints are provided by the oxygen-isotope data. Minerals in the clast have oxygen-isotope ratios (Table 2) that are (1) statistically indistinguishable from the Terrestrial Fractionation Line (i.e., the clast could be a terrestrial or a lunar sample), (2) are statistically (2σ error) different from the bulk HED meteorite trend, and (3) are within 2σ error of the bulk SNC meteorite and angrite meteorite oxygen-isotope trends (Fig. 10).

Although the oxygen isotopic composition is consistent with the TFL, and the plagioclase An# values overlap with terrestrial values (Fig. 9), we provisionally discount a terrestrial origin for the clast. This is because the olivine Ni contents are lower (Clast = <150 ppm Ni: Table S1) than high-Fo (Fo>80) olivine in terrestrial mafic rocks (typically >500 ppm Ni; Karner et al. 2003; PetDB database and Archean samples (>600 ppm Ni; Barnes et al. 1983; Karner et al. 2003; Mondal et al. 2006; Pettigrew and Hattori 2006; Cheng and Kusky 2007). In addition, although the Fe/Mn ratios in the clast's mafic phases overlap with examples from terrestrial samples, they do not exactly follow the terrestrial Fe/Mn ratio trend (Fig. 8).

It is notable that the Fe/Mn ratios in olivine fall very close to the martian olivine trend (Figs. 8a and 8b), and that the oxygen-isotope values do not rule out a martian origin (Fig. 10). However, evidence from other mineral chemistry data appears to discount a martian source, as the olivine and pyroxene mineral compositions are atypically magnesian, and the plagioclase too Ca-rich (anorthitic) compared with known martian meteorites (i.e., Papike et al. 2003, 2009; Karner et al. 2003, 2004, 2006; Sarbadhikari et al. 2009; Fig. 9). Moreover, unlike the olivine, the Fe/Mn ratios in pyroxene do not follow the martian trend (Karner et al. 2003, 2006; see Figs. 8 and 9). However, we have to recognize that our current set of martian meteorites are derived from a very few locations on Mars, and it would be unwise to assume that we have anything approaching a complete picture of the range of the composition of martian igneous rocks that we could use for such a comparison.

As we have no recognizable meteorite samples from Venus or Mercury to compare with, it is difficult to assess if these planets could have been the source of the clast. However, the clast bulk K (7300–8400 ppm; Table 1) and Th (approximately 2.5 ppm) in the clast is notably higher than that recorded in any Mercury surface regoliths by the Messenger mission gamma-ray spectrometer (GRS) instrument (1150 ± 220 ppm K; 0.22 ± 0.06 ppm Th; see Peplowski et al. 2011). It is also higher than that recorded by the GRS instrument on board the Venera landers (3000–4500 ppm K; 0.7–2 ppm Th; see fig. 2 of Peplowski et al. 2011 and references therein). This suggests that the clast is unlike typical rocks in Mercury's or Venus's upper crust, although clearly the full diversity of these crustal rocks is presumably greater than deduced from relatively low spatial resolution orbital remote sensing of Mercury and three in situ measurements made on Venus.

Compared with asteroid material sampled at the present day by near Earth objects (NEOs), mafic mineral compositions in the clast are dissimilar to lithologies sampled by aubrite (Brearley and Jones 1998), winonaite (Benedix et al. 2005), acopulcoite, lodronite (McCoy et al. 1996, 1997), mesosiderite (Nehru et al. 1980), and ureilite (Downes et al. 2008) achondritic meteorites. Plagioclase grains in the clast are not as calcic, and Fe/Mn ratios in olivine and pyroxene are lower than in angrite meteorites (Fig. 9) (Papike et al. 2003). Fe/Mn ratios and the augite-rich Na-bearing nature of pyroxene are also dissimilar to those in HED pyroxene (Figs. 8c and 8d, and Fig. 9; see also McSween et al. 2012; Beck et al. 2012). HED meteorites have already been shown to not fit well with the clast's oxygen-isotope composition (Fig. 10).

The clast could, therefore, have originated from a different, so-far unsampled, achondritic parent body with differentiates that were melted from a primitive reserve (to account for magnesian mafic phases) and included a fractionated residual liquid component (to account for the Na-K–rich mesostasis). Rare granitic igneous fragments and glasses (some magnesian) have been reported in a number of meteorites that are presumed to have originated within differentiated crusts by magmatism or impact processes on small asteroidal parent bodies (e.g., Bonin 2012). It is, therefore, possible that this clast could represent a lithology from a differentiated asteroid parent body that is poorly represented, or not represented, in meteorites being delivered to Earth at the present day.

Whatever its source, if the clast is exogenous to the Moon, the timing of its delivery to the lunar surface could help to shed new light on the sources of projectiles being delivered to the Moon at different points in lunar history. Constraining the age of lunar regoliths is complicated as they contain many different rock types that may have undergone several formation and space-exposure episodes. In regolith samples that have undergone exposure to the space environment, the bulk-rock ratio of “trapped” (parentless) 40Ar to solar wind–implanted 36Ar has been shown to be indicative of the last time the regolith system was closed from surface exposure (i.e., it was turned from a soil into a rock). This isotope ratio can then be calibrated to a temporal antiquity age record using the argon isotope record of Apollo samples of known age (Eugster et al. 2001; Joy et al. 2011a). The trapped 40Ar/36Ar of MAC 88015 was measured by Eugster et al. (1991) to be 5.7. Applying this ratio to the age calibration of Joy et al. (2011a) implies that MAC 88105 was closed from lunar surface exposure at approximately 2.82 Ga. Therefore, any meteorite components in the MAC 88105 parent regolith would have to have been delivered to the lunar surface before this time. Eugster et al. (1991) report that the parent regolith was immature and had a surface residence time of about 650 Ma prior to brecciation: this implies that the clast possibly was delivered between approximately 3.47 Ga and approximately 2.82 Ga during Late Imbrian epoch to early Eratosthenian period.

Highly siderophile-element signatures for impact melts (e.g., Puchtel et al. 2008; Fischer-Gödde and Becker 2012; Galenas et al. 2012), and discoveries of projectiles in ancient breccias (Joy et al. 2012), imply that chondritic asteroids were common sources of impactors during the basin-forming epoch (>3.7 Ga). Delivery of achondritic material to the lunar surface during an interval of approximately 3.47 Ga and approximately 2.82 Ga is consistent with a variety of impactors (chondritic, achondritic, iron) found in younger Apollo 16 regolith breccias and Apollo landing site soils (see Joy et al. 2012 for a summary), and reflects a possible diversification of impactor sources in post–basin-forming epoch (<3.7 Ga) projectile populations.


We have discovered a compositionally unusual clast within lunar meteorite MAC 88105,159. The clast is composed of forsteritic olivine, bytownitic plagioclase, augitic pyroxene, and a mesostasis of devitrified K-rich glass with an alkali-calcic dacite composition. In terms of olivine and pyroxene mineral Mg#, it is similar to Mg-Suite samples; however, in terms of An# plagioclase are sodic and more akin to samples of the high alkali suite. This indicates that the rock may represent a new type of lunar lithology that experienced an unusual petrological origin combining a primitive mafic melt with a late-stage alkali-element (ITE-rich) component. However, despite these similarities to some known lunar rock types, pyroxene and olivine in the clast have Fe/Mn ratios that are notably different from any known indigenous lunar samples (Figs. 8 and 9). As Fe/Mn ratios are key indicators of planetary heritage, this evidence suggests that the clast may not have originated from the Moon, and instead may represent material from another differentiated parent body. We suggest that these Fe/Mn ratios and other unique compositional characteristics point toward derivation from an achondritic basaltic meteorite that was derived from a parent body that was more oxidized and more volatile-rich than the Moon.

Although in this study we have not been able to definitively identify the parent body from which this clast is derived, the plausible discovery of an achondritic meteorite implanted in the lunar regolith prior to approximately 2.82 Ga adds to the diverse suit of meteoritic material already known to be sampled in regolith breccias and Apollo soils (see Joy et al. 2012 for a summary). This further underlines the importance of the lunar regolith as an archive of impact debris derived from other bodies in the solar system, including possible samples of the early Earth of astrobiological significance that may not be preserved anywhere else (e.g., Armstrong et al. 2002; Crawford et al. 2008). Identifying such materials, both within the existing lunar sample collection and in samples collected by future lunar missions, will be an important aspect of lunar science in the coming decades.


Thanks to Anne Peslier at JSC, John Spratt and Anton Kearsley at the NHM, and Andy Beard at Birkbeck for analytical assistance. We thank the referees Drs. Tomoko Arai, Duck Mittlefehldt, James Karner, and Juliane Gross, and AE Dr. Cyrena Goodrich for constructive comments. This research was facilitated by Leverhulme Trust grants F/07 112/P to IAC and 2011-569 to KHJ. We acknowledge a NASA Lunar Science Institute contract NNA09DB33A to David A. Kring PI, which supported the JSC EMP analyses and, grant NNX08AG58G to Gary R. Huss.

Editorial Handling

Dr. Cyrena Goodrich