Occurrence and implications of secondary olivine veinlets in lunar highland breccia Northwest Africa 11273

Lunar breccias preserve the records of geologic processes on the Moon. In this study, we report the occurrence, petrography, mineralogy, and geologic significance of the observed secondary olivine veinlets in lunar feldspathic breccia meteorite Northwest Africa (NWA) 11273. Bulk‐rock composition measurements show that this meteorite is geochemically similar to other lunar highland meteorites. In NWA 11273, five clasts are observed to host veinlets that are dominated by interconnecting olivine mineral grains. The host clasts are mainly composed of mafic minerals (i.e., pyroxene and olivine) and probably sourced from a basaltic lithology. The studied olivine veinlets (~5 to 30 μm in width) are distributed within the mafic mineral host, but do not extend into the adjacent plagioclase. Chemically, these olivine veinlets are Fe‐richer (Fo41.4–51.9), compared with other olivine grains (Fo54.3–83.1) in lithic clasts and matrix of NWA 11273. By analogy with the secondary olivine veinlets observed in meteorites from asteroid Vesta (howardite–eucrite–diogenite group samples) and lunar mare samples, our study suggests that the newly observed olivine veinlets in NWA 11273 are likely formed by secondary deposition from a lunar fluid, rather than by crystallization from a high‐temperature silicate melt. Such fluid could be sulfur‐ and phosphorous‐poor and likely had an endogenic origin on the Moon. The new occurrence of secondary olivine veinlets in breccia NWA 11273 reveals that the fluid mobility and deposition could be a previously underappreciated geological process on the Moon.


INTRODUCTION
Since the lunar magma ocean (LMO) cooled and solidified, a variety of post-LMO processes (e.g., magmatism, volcanism, and impacts) play an important role in the Moon's geological evolution (e.g., Lucey et al. 2006;St€ offler et al. 2006). Many of these processes have formed rocks that were thought by early workers to be anhydrous in nature. However, recent reexamination of these samples using new analytical techniques has shown there to be variability of "water" (H 2 O and/or OH) and other volatile elements (e.g., Saal et al. 2008;Boyce et al. 2010;Greenwood et al. 2011;Hauri et al. 2011;Hui et al. 2013;Tart ese et al. 2013; Barnes et al. 2014;Robinson et al. 2016;Potts et al. 2018). One implication of these studies is that volatilebearing mineral precipitation may have been commonplace within the lunar mantle, crust, and surface regolith.
Mineral assemblages and textures that are related to fluid processes in lunar samples can provide new insights into the behavior of fluid agents on the Moon, and also the volatile (e.g., F, Cl, H, and S) and chalcophile-siderophile element (e.g., Ni, Co, Cu, and Zn) transport in the lunar crust. However, the strong evidence in the lunar sample collection for fluid processes such as fluid-driven metasomatism is, at present, relatively limited (e.g., Neal and Taylor 1991;Isa et al. 2014;Treiman et al. 2014;Warren et al. 2014). For example, the replacement texture of olivine and pyroxene by troilite documented in Apollo 16 breccias (i.e., samples 67915 and 67016) has been suggested to represent a mineralogical record of sulfur-bearing fluids on the Moon (Haskin and Warren 1991;Colson 1992;Norman et al. 1995;Shearer et al. 2012). This texture has been used to develop a thermodynamic model of Sbearing fluid in the lunar crust (Bell et al. 2015). Mccallum and Schwartz (2001) suggested that Cl-rich fluid metasomatism could be a possible cause for the symplectic intergrowths of Mg-Al-chromite and two pyroxenes in igneous troctolite rock 76535, although an exogenous chromite-saturated melt was also proposed to account for this symplectite formation (Elardo et al. 2012). Lunar rock 79215 also records previously unrecognized processes in the lunar volatile cycle (i.e., impact-induced remobilization of P, F, and Cl volatiles and vapor-phase transport; Treiman et al. 2014).
More widely, the record of fluid mobility in crusts is also complex on other large planetary bodies. During the past few decades, secondary olivine veinlets (i.e., the veinlets formed of small olivine grains that crosscut pyroxene grains) have been discovered in a series of howardite-eucrite-diogenite (HED) meteorites that originated from asteroid 4-Vesta (Takeda et al. 1983(Takeda et al. , 1994Yamaguchi et al. 1996;Warren 2002; Barrat et al. 2011;Roszjar et al. 2011;Warren et al. 2014;Mittlefehldt 2015;Pang et al. 2017;Patzer and McSween 2018). Similar secondary olivine veinlets have also been observed in a few lunar basaltic samples crosscutting pyroxene grains: these include the basaltic regolith breccia meteorite Meteorite Hills (MET) 01210 (Arai et al. 2010) and Apollo 14 mare basalts 14072 and 14310 (Roedder and Weiblen 1972;Warren et al. 2018). No secondary olivine veinlets have been reported to date in lunar highland samples. Warren et al. (2018) suggested that the olivine veinlets in Apollo basalt 14072 are likely deposited by a water-bearing fluid. However, the formation mechanism of lunar olivine veinlets and their geological records on the Moon are still poorly understood.
Northwest Africa (NWA) 11273, with a total mass of~2.81 kg, is a newly discovered lunar feldspathic regolith breccia from Algeria in 2017 (Meteoritical Bulletin Database). The petrographic, mineralogical, and geochemical features of NWA 11273 have not been previously investigated. We have identified five olivine veinlet-bearing clasts in this meteorite. We discuss how this finding could help expand our understanding of the secondary olivine veinlets from a new geographic region on the Moon. Here, we report the detailed petrographic and mineralogical study of five olivine veinlet-bearing clasts, and also report the bulk-rock composition and possible launch regions of NWA 11273 with the goals of interpreting the origin and geological records of the secondary olivine veinlets on the Moon.

SAMPLE AND ANALYTICAL METHODS
The NWA 11273 sample used in this study includes one 0.95 g meteorite slice provided by Eric Twelker, who has the main mass of NWA 11273, and one~8 g specimen provided by Hao Zhu, Beijing Planetarium. This larger 8 g specimen was cut into~1 mm thick slices by using a diamond wire saw. Seven slices were then adhered to separate 1 inch round glass slides using epoxy resin, were polished with Al 2 O 3 and diamond polishing paste, coated with carbon, and investigated using different analytical techniques. The studied five sections that contain olivine veinlet-bearing clasts are shown in Fig. S1 in supporting information.
Backscattered electron (BSE) images were collected by using an FEI Scios dual-beam focused ion beam/ scanning electron microscope (FIB/SEM) at Institute of Geochemistry, Chinese Academy of Sciences. The operating conditions were 15-20 kV accelerating voltage, 1.6 nA beam current, and 7 mm working distance. BSE montage maps (Figs. S2.1-S2.5 in supporting information) were stitched from multiple high-resolution BSE images (~5509 magnification) using the FEI's Maps software v.3.2. These images were reviewed, and then, detailed petrography and X-ray elemental mapping (Figs. S3.1-S3.5 in supporting information) of target clasts were studied using the same FEI Scios FIB/SEM and an energy dispersive detection system (EDS of EDAX), with the same current and voltage setting noted above.
One FIB slice of NWA 11273 was prepared with the same FIB/SEM (FEI Scios) at Institute of Geochemistry, Chinese Academy of Sciences. A region of interest was first coated by an electron beamdeposited and Ga ion-deposited Pt layer. Then, a slice (15 lm 9 10 lm 9 170 nm) was cut by a Ga ion operated at 30 kV with beam currents ranging from 3 to 15 nA, lifted from the sample, and mounted onto a TEM copper grid. To obtain a slice with the thickness of approximately 150 nm, the Ga ion source was operated at 30 kV with beam currents in the range of 100 pA to 1 nA. Finally, a 2-5 kV ion source and 43-48 pA beam current were used for the final processing to remove the damaged surface layer. To investigate the micromorphology of this prepared FIB slice, the scanning transmission electron microscopy (STEM) micrographs were collected using the same FEI Scios FIB/SEM operating at 30 kV accelerating voltage and 0.8 nA beam current.
Quantitative major and minor element compositions of silicates (i.e., pyroxene, olivine, and plagioclase) and chromite grains (Table S1 in supporting information) were obtained using a JXA 8230 electron microprobe analysis (EMPA) at the Guilin University of Technology. This instrument was operated at an accelerating voltage of 15 kV and electron beam current of 20 nA. A peak acquisition time of 20 s (with a background measurement time of 10 s) was used for a majority of the elements analyzed (i.e., Si, Ti, Al, Cr, Fe, Mn, Mg, and Ca), with the exception of Na and K, for which a peak acquisition time of 10 s (5 s background) was used. The measurements were calibrated with well-characterized standard samples, including olivine (Si, Mg, Fe), plagioclase (Al, Na), pyroxene (Ca), mica (K), chromite (Cr), Mn-metal (Mn), and rutile (Ti). For fine-grained (<5 lm) minerals, a focused beam (<1 lm) was used, while a defocused beam of 3-5 lm was used for other large mineral grains. The typical detection limits are approximately 0.03 wt%.
An~200 mg slice of NWA 11273 was powdered using a clean agate pestle and mortar for bulk-rock composition measurements. One powdered sample (50 mg) was dissolved in polytetrafluoroethylene-lined stainless steel bombs using a HF+HNO 3 mixture for 48 h at~190°C for bulk trace elements' measurements. The bulk-rock trace elements' content was analyzed using a PerkinElmer ELAN DRC-e inductively coupled plasma mass spectrometer (ICP-MS) at Institute of Geochemistry, Chinese Academy of Sciences, following standard procedures (Qi et al. 2000;Zeng et al. 2018a). Three standard reference materials (i.e., AGV-2, AMH-1, and GBPG-1; Table S2 in supporting information) were used for analytical quality control. Rh was used as an internal standard to monitor signal drift during counting. The uncertainties (2 RSD) for most analyzed elements are 5-10%. For bulk-rock major and minor composition measurements, the other 50 mg powdered sample was analyzed by using the inductively coupled plasma optical emission spectroscopy (ICP-OES, Agilent 720) at Institute of Geochemistry, Chinese Academy of Sciences. This sample was digested with the same procedures for ICP-MS measurement and then was analyzed following the standard procedures. The analyzed elements include Ti, Al, Fe, Mn, Mg, Ca, Na, K, P, Ni, Cu, and Ba. Four standard reference materials (i.e., GSR-3, GSD-4, GSD-6, and GSR-12; Table S3 in supporting information) were used for analytical quality control. The analytical precision is better than 5% (2 RSD) for most analyzed elements.

Petrologic Texture
NWA 11273 is a polymict feldspathic regolith breccia meteorite that is composed of a wide range of feldspathic clasts, mineral fragments, and glass components. A BSE image of the representative region of NWA 11273 is shown in Fig. 1. In this study, five lithic clasts (i.e., clasts V1-V5; Table 1   It should be noted that we did not observe the olivine veinlets occurring in other mineral phases such as plagioclase. Clast V1 is a large (~1500 lm in size; Fig. 2a) mafic clast, mainly composed of pyroxene (~85%, by area), plagioclase (~10%), and minor (<5%) chromite and troilite grains. Pyroxene in this clast consists mainly of low-Ca pyroxene with exsolution lamellae of clinopyroxene (<3 lm in width; Fig. 2c). Fine-grained (<10 lm) chromite grains are scattered within this clast. Two olivine veinlets were observed cutting across pyroxene grains with the width of 25 lm (Fig. 2b). Smaller (<20 lm) grains of low-Ca pyroxene and chromite are detected in close association with the olivine veinlet (Fig. 2c).
Clast V2 is another olivine veinlet-bearing maficrich clast that mainly consists of~75% pyroxene with 25% plagioclase (Fig. 2d). The olivine veinlets (~25 lm in width) cut across the pyroxene crystal and did not cut across the plagioclase grains in these clasts (Fig. 2d). The host low-Ca pyroxene contains thin (<3 lm in width) lamellae of clinopyroxene (Fig. 2e). Most olivine grains in this veinlets show~120°triple junction grain boundaries (Fig. 2e). In addition to the mafic clasts V1 and V2, olivine veinlets were also observed in three pyroxene fragments (i.e., clasts V3-V5; Fig. 3) in NWA 11273. Clast V3 is about 250 lm in size and is cut by irregular olivine veinlets with a width of 3-10 lm (Fig. 3a). This pyroxene fragment contains thin (<2 lm) exsolution lamellae of clinopyroxene. A few chromite grains are observed in the olivine veinlets in clast V3 (Fig. 3b). One low-Ca pyroxene fragment (i.e., clast V4) has an irregular clast boundary (~200 lm in size) and does not contain the clinopyroxene exsolution lamellae. Several olivine veinlets (~10 to~35 lm in width) are found within and surrounding this fragment (Fig. 3c). The small (<15 lm) grains of pyroxene and chromite are presented within the olivine veinlets of Clast V4 (Fig. 3d). Clast V5 is a large (~250 9 400 lm in size; Fig. 3e) pyroxene containing several irregular olivine and plagioclase veinlets (~10 lm in width) that cut across this clast. This host low-Ca pyroxene contains thin (<2 lm) exsolution lamellae of clinopyroxene and also a few small plagioclase, troilite, and merrillite grains (Fig. 3f). A few chromite grains are observed in both olivine veinlets and the host pyroxene crystal (Fig. 3f).

STEM Observations
The boundaries between secondary olivine veinlets and their host pyroxene could help us to understand the formation mechanism of olivine veinlets. To observe the micromorphology between secondary olivine veinlets and their host pyroxene, an FIB section of 15 9 10 lm was extracted from clast V5 (see Fig. 3f). The STEM image shows that the olivine veinlets have distinct boundaries with their host pyroxene grains (Fig. 4). No replacement texture of pyroxene by olivine was observed at submicron scale. The pyroxene grains show serrated grain boundary, resulting in the wedge-shaped hole with a length of~1 lm (Fig. 4b). Such microstructures more likely represent the brittle fracture of pyroxene (e.g., see also similar textures in rocks from the terrestrial Lockne impact crater, Sweden; fig. 3 of Agarwal et al. 2017).

Mineral Composition
The mineral compositions for lithic clasts V1-V5 in NWA 11273 are provided in Tables S1 and S2. Both secondary olivine veinlets and their host pyroxene in clasts V1-V5 have Fe-Mn (atomic) trends similar to the Fe-Mn trends for other olivine and pyroxene in the rest of the NWA 11273 meteorite and Apollo returned samples (Fig. 5).

BULK-ROCK METEORITE COMPOSITION
The bulk-rock major, minor, and trace element composition of NWA 11273 is listed in Table 3. This meteorite has bulk Al 2 O 3 and FeO contents of 25.1 and 4.35 wt%, respectively (Table 3), similar to the Apollo 16 soils (Lucey et al. 2006) and lunar feldspathic meteorites (see comparison in Fig. 8a). The CInormalized (Anders and Grevesse 1989) rare earth elements (REE) pattern of NWA 11273 shows lower REE abundance (5 9 CI La, 3 9 CI Lu; Fig. 8b) compared with Apollo 16 typical mature soils and the average lunar feldspathic highland crust (Fig. 8b). Compared with the Apollo 16 regolith samples that generally have Sm >2 ppm and Th >1 ppm (e.g.,    Table 3).

REMOTE SENSING DATA AND POTENTIAL LAUNCH REGIONS
Remote sensing missions provide key insights to the geochemical diversity of the Moon. For example, the Lunar Prospector mission gamma ray spectrometer (LP-GRS) has measured the abundance of rock-forming and radioactive incompatible elements (e.g., O, Si, Fe, Ti, Mg, Al, Th, and K) within tens of centimeters of the lunar surface (Binder 1998). In addition, bulk-rock composition of lunar regolith breccia meteorites also extend our understanding of the geochemistry of lunar surface (i.e., few centimeters to meters) materials, where the meteorites ejected from (Lorenzetti et al. 2005;Joy et al. 2010;Calzada-Diaz et al. 2015). Measurements of Apollo samples and lunar meteorites shown that bulk-rock FeO and Th contents have been used to help identify (distinguish) different lithologies on the Moon (e.g., fig. 17 of Korotev 2005). Specifically, lunar highland feldspathic-rich materials are characterized by relative low abundance of FeO and low ITEs (e.g., Th), while mafic-rich basaltic lithologies exhibit high content of FeO and high ITEs (e.g., Th). Therefore, by comparing bulk-rock composition of regolith breccia meteorites with LP-GRS geochemical data (i.e., FeO and Th), the meteorite's likely launch location(s) can be constrained.
In this work, we used the bulk FeO (4.35 AE 1 wt%) and Th (0.17 AE 1 ppm) contents of NWA 11273 as input parameters to locate regoliths with similar compositions following the method reported in Joy et al. (2010) and Zeng et al. (2018a). The resulting map ( Fig. 9) shows that NWA 11273 is compositionally similar to the regolith in the anorthositic-Feldspathic Highland Terrane (FHT-An), and also to regolith around the FHT-An in the outer-Feldspathic Highland Terrane (FHT-O).

Lunar Origin of the Olivine Veinlet-Bearing Clasts V1-V5
Although exogenic mineral and lithic components (e.g., chondrite and iron meteorites) could be preserved in polymict lunar breccias (e.g., Day et al. 2006;Joy et al. 2012Joy et al. , 2016, the mineralogy of clasts V1-V5 suggest that these clasts have a lunar origin. Evidence includes that (1)  NWA 11273 is a hot desert meteorite that could be contaminated by weathering processes on the Earth (Lee and Bland 2004;Saunier et al. 2010;Velbel 2014). Previous studies demonstrated that the terrestrial weathering can modify the mineralogical and geochemical information of desert meteorites; the weathering products commonly include calcite (e.g., Fig. 2b), barite (e.g., Fig. 3c), Fe-oxides, and clay, but do not include olivine grains (Lee and Bland 2004;Al-Kathiri et al. 2005;Saunier et al. 2010;Hyde et al. 2014;Velbel 2014). This suggests that the olivine veinlets in clasts V1-V5 are not likely terrestrial weathering products. In addition, the olivine veinlets in NWA 11273 were observed to be cut through by multiple fractures (probably induced by shock events); see Figs. 2c, 2e, 3b, and 3d), suggesting that these veinlets were formed before this meteorite was ejected off the Moon and/or fell to the Earth. Host Rock of the Olivine Veinlet-Bearing Clasts V1-V5 Mineralogically, clasts V1-V5 are mafic rich predominately composed of pyroxene with minor plagioclase and accessory chromite grains (Figs. 2 and  3). This suggests that clasts V1-V5 are not like plagioclase-rich lunar highland crustal rocks (e.g., the anorthosite and feldspathic impact-melt clasts; Cahill et al. 2004;Yamaguchi et al. 2010;Wittmann et al. 2014), but are more similar to basaltic lithologies found in other lunar breccia meteorites (e.g., Snape et al. 2011;Mercer et al. 2013;Joy et al. 2014). This conclusion is further supported by the mineral composition of these clasts (Fig. 10): clasts V1-V5 have pyroxene compositions (i.e., Fe# and Ti) similar to pyroxene from Apollo low-titanium (Low-Ti) and very low-titanium (VLT) basalts (Fig. 10). We note that the variation of Fe# is relatively small within individual lithic clasts, b.d. = below detection limit; Ol = olivine; Px = pyroxene; Px* = pyroxene grains within the secondary olivine; Pl* = plagioclase grains within the secondary olivine; Pl = plagioclase, Chr = chromite, Chr* = chromite grains within the secondary olivine, Spl = Usp = ulvospinel, Hc = hercynite. a Considering the thin (<2 lm) exsolution lamellae of clinopyroxene, the reported clinopyroxene composition could be collected from a mixture of orthopyroxene and clinopyroxene.
while Ti# shows relatively large variation. Such a chemical trend could be caused by different diffusion rates of these elements (i.e., the Fe and Mg have faster diffusion rates than that for Ti and Cr; Arai et al. 1996;Zhang and Hsu 2009;Fagan et al. 2014). This difference makes Fe# more easily homogenized than Ti# in pyroxene when the subsequent annealing processes occur (Arai et al. 1996). Such geochemical character (i.e., Fe# and Ti# trend) suggests that the host rock of clasts V1-V5 underwent postmagmatic reequilibration. This interpretation is strongly supported by the exsolution texture observed in these clasts (i.e., clasts V1, V2, V3, and V5; Figs. 2 and 3). The olivine veinlet-bearing clasts V1-V5 show a variety of petrographic texture and mineralogy. For example, (1) clast V1 distinctively contains oval-shaped high-Ca pyroxene, (2) clast V2 distinctively contains rounded plagioclase grains, and (3) clast V4 distinctively contains no exsolution lamellae at microscale. These features indicate that clasts V1-V5 might represent at least three different parent rocks and clearly experienced different formation histories as well.

Origin of Secondary Olivine Veinlets in NWA 11273: Fluid Deposition?
The secondary olivine veinlets that cut across pyroxene grains have been previously observed in several of HED meteorite samples (Takeda et al. 1983,   Fe-rich olivine veinlets (Takeda et al. 1983(Takeda et al. , 1994Barrat et al. 2011;Roszjar et al. 2011;Mittlefehldt et al. 2013;Warren et al. 2014;Patzer and McSween 2018). The origin of these secondary olivine veinlets is still under debate; multiple formation mechanisms have been proposed to account for veinlets in HED meteorites and lunar mare basalts. These are outlined below. Pang et al. (2017) proposed that the shock-induced localized melting of forsteritic olivine (Fa 44.9-45.3 ) accounted for the formation of Mg-rich olivine veinlets with polycrystalline aggregates and highly irregular veinlet boundaries in eucrite NWA 1109 (Fig. 11a) (Korotev et al. 2003), and Apollo 16 typical mature soils (green circle; Lucey et al. 2006). The highlands and mare regions are taken from Korotev et al. (2003). (b) The CI-normalized (Anders and Grevesse 1989) REE pattern of NWA 11273 (Table 3) compared with the lunar feldspathic meteorites, average feldspathic crust (Korotev et al. 2003), and average Apollo 16 soils (Lucey et al. 2006 Fig. 9. Compositional constraints of likely launch regions of NWA 11273 using the Lunar Prospector 0.5°/pixel FeO (wt%) and Th (ppm) data sets (Lawrence et al. 2002(Lawrence et al. , 2003. The cyan areas are compositional similar regoliths to the composition of NWA 11273 (i.e., Fe = 4.35 AE 1 wt%, Th = 0.17 AE 1 ppm; Table 3). Locations of the Apollo and Luna sample return missions are noted, along with the Chang'e-3/Chang'e-4 lander and rover landing site. The map highlights with dashed outlines the main geochemical terranes defined by Jolliff et al. (2000).  Pang et al. 2017) contain FeNi metals with high Ni/Co ratios of 5-11, suggestive of meteoritic origin. In contrast, no FeNi metal was detected in the studied veinlet-bearing clasts V1-V5. These suggest that the impact-induced localized melting of phenocrystic olivine does not explain the formation of olivine veinlets in clasts V1-V5 of NWA 11273.

Shock-Induced Localized Melting Model
Textural relationships demonstrate that the studied olivine veinlets were not likely a typical primary crystallization product of clasts V1-V5's parental magma, but they are more likely to represent secondary filling of pyroxene grains that have previously been fractured by impact shock brittle deformation. One possible source of the secondary olivine veinlets in clasts V1-V5 is that they were crystallized from injection of a shock-induced silicate melt at high temperature (~1300°C; M edard and Grove 2007). If this scenario is correct, such silicate melt could percolate through clast V1-V5's parent rock along previously existing cracks and grain boundaries, then locally forming the secondary olivine veinlets and other accessory minerals (i.e., plagioclase and chromite). However, the textures of secondary olivine veinlets in clasts V1-V5 are not consistent with this scenario. (1) Shock-induced melt veins in meteorites commonly cut through various mineral phases (e.g., Walton and Shaw 2009;Miyahara et al. 2013). In contrast, the studied olivine veinlets in NWA 11273 are only distributed in pyroxene crystals, but do not extend into the adjacent plagioclase (e.g., clast V2; Fig. 2d). Also, olivine veinlets in NWA 11273 were observed enclosed within host crystals (e.g., Figs. 2b and 3c), which are very texturally different from shock-induced melt veins. (2) A number of mineral phases (e.g., silicate grain, FeNi metal, troilite, and glass) are commonly observed in a shock-induced melt veins (e.g., Tomioka and Miyahara 2017;Zeng et al. 2018b). Such mineralogy is largely different in the mineral assemblages (i.e., olivine, plagioclase, and chromite) of olivine veinlets in NWA 11273. (3) If a shock-induced melt percolated cracks with clast V1-V5's parent rock, the temperature would have been too high (~1300°C) to form the wedge-shaped holes between olivine veinlets and surrounding pyroxene grains in clasts V1-V5 (Figs. 2-4). This is because some remelting of the host pyroxene would have likely occurred.

Remelting of Olivine Mesostasis and Incongruent Melting of Pyroxene Models
To account for the origin of Fe-rich olivine veinlets in HED meteorites, two possible mechanisms have been proposed; i.e., (1) shock-induced remelting of late-stage phases such as Fe-rich olivine mesostasis ( Fig. 11b; Takeda et al. 1983Takeda et al. , 1994 and (2) incongruent melting of primary pyroxene in the magma chamber ( Fig. 11c; Roszjar et al. 2011). The further studies of HED meteorites have ruled out these two models (e.g., see the discussion in Pang et al. 2017). Likewise, applying this framework to lunar sample NWA 11273, these two models are difficult to account for the observed textural and mineralogical features in this meteorite as (1) we do not observe regions of mesostasis in any of the clasts V1-V5, the clasts themselves do not show excessive shock metamorphic textures (e.g., glassy veins), and the olivine veinlets are not seen in other mineral phases such as plagioclase (Figs. 2 and 3). All these textures imply that the shock-induced remelting model is not likely the formation mechanism of olivine veinlets in NWA 11273. (2) The incongruent melting of pyroxene model (Roszjar et al. 2011) is not consistent with the sharp boundaries between olivine veinlets and their host pyroxene grains in clasts V1-V5 (Figs. 2-4); if incongruent melting occurs, then a reaction texture  (1978) and Arai et al. (1996). (Color figure can be viewed at wileyonlinelibrary.com.) would be expected at the boundaries of olivine veinlets and host pyroxene grains.

Fluid Deposition Model
Previous studies suggest that the fluid deposition model (Fig. 11d) is currently the most accepted model to explain the secondary Fe-rich olivine veinlets and associated Ca-plagioclase within veinlets in a series of HED meteorites Warren et al. 2014;Mittlefehldt 2015;Pang et al. 2017). Similar hydrothermally deposited fayalitic olivine (e.g., Rasmussen et al. 1998) and hydrothermally deposited anorthitic plagioclase (e.g., Python et al. 2007;Mora et al. 2009) are common on the Earth. Also, secondary fayalitic olivine that are related to fluid-mediated metasomatism have been described in many carbonaceous chondrite meteorites (e.g., Krot et al. 2000Krot et al. , 2002Krot et al. , 2004Krot et al. , 2007Brearley and Krot 2013;Li et al. 2017). Recently, a fluid deposition model has also been used to explain the occurrence of olivine veinlets in lunar basalt 14072 (Warren et al. 2018). The reported olivine veinlets in lunar sample 14072 are similar to the olivine veinlets within NWA 11273 (1) olivine veins are widespread within pyroxene, but absent in plagioclase, (2) olivine veins have similar chemical composition (Fig. 6), and (3) both of them are associated with chromite grains. Accordingly, we have considered this process as the most likely mechanism for explaining the formation of secondary olivine veinlets in NWA 11273. The fluid deposition process can account for texture and mineralogy (e.g., irregular sharp boundaries) of the olivine veinletbearing clasts V1-V5.
In most cases, the fluid-driven metasomatism would occur along cracks or grain boundaries, resulting in dissolution or replacement of primary minerals (e.g., Krot et al. 2000Krot et al. , 2004Shearer et al. 2012;Li et al. 2017;Warren et al. 2018;Zhang et al. 2018). However, the olivine veinlets in clasts V1-V5 are straight with sharp boundaries with their host pyroxene grains (Figs. 2-4). This suggests that olivine veinlets in NWA 11273 might not be a reaction product from the in situ replacement of their host pyroxene by fluids. These secondary olivine veinlets and associated plagioclase could have formed by direct deposition from the fluid percolating through the pyroxene fractures. This interpretation is further supported by the chemical composition difference between the secondary olivine veinlets and other olivine grains in NWA 11273 (Fig. 7). According to the fluid deposition model, the formation of secondary olivine veinlets with accessory anorthitic plagioclase can be interpreted by the general reaction: fluid ? olivine + Ca-plagioclase. Specifically, some type of aqueous fluid agent (i.e., water bearing) could transport the olivine/Ca-plagioclase cations (e.g., Fe 2+ , Mg 2+ , Si 4+ , Ca 2+ , and Al 3+ ) and then direct deposited the secondary minerals by the following reactions (Brearley and Krot 2013 The absence of reaction boundaries in clasts V1-V5 suggests potentially relatively low temperature (<550 K) Fe-rich olivine mesostasis (a) Localized melting of olivine model (Pang et al. 2017) (b) Remelting of olivine mesostasis model (Takeda et al. 1983(Takeda et al. , 1994 (d) Fluid deposition model (e.g., Barrat et al. 2011;Warren et al. 2014) (c) Incongruent melting of pyroxene model (Roszjar et al. 2011  for the formation of secondary olivine and Caplagioclase from aqueous solutions (Krot et al. 2004;Brearley and Krot 2013). In addition, the presence of wedge-shaped holes possibly suggests that the fluid contains a certain amount of volatile (e.g., H 2 as proposed in reaction 2). In most circumstances, Cr 3+ in hydrous fluid has low solubility and high immobility (Roeder and Reynolds 1991;Klein-Bendavid et al. 2011). This suggests that the small chromite grains sometimes found within the secondary olivine veinlets were more likely incorporated from their host pyroxene than formed in situ. In addition, pyroxene grains seen within secondary olivine veinlets have the same chemical composition as the host pyroxene ( Table 2), suggesting that they could be the broken pieces of the host that have become mobilized by the fluid flow. This interpretation is supported by the irregular shape of pyroxene and chromite grains in olivine veinlets (Figs. 2 and 3). These irregular pyroxene and chromite grains are likely related to the brittle fracture of pyroxene, which was implicated by the observed wedge-shaped holes (Fig. 4a) and the irregular grain boundary (Figs. 2 and 3) of host pyroxene.
Consequently, the fluid deposition seems at present the most plausible candidate for explaining the secondary olivine veinlets in NWA 11273, although this model cannot readily account for certain textural observations such as the absence of olivine veinlets in other host mineral phases (e.g., plagioclase). Our studies and previous works have shown that olivine veinlets are commonly distributed in pyroxene crystals, but do not extend into the adjacent plagioclase (e.g., fig. 4 of Barrat et al. 2011;fig. 4 of Pang et al. 2017;figs. 1 and 2 of Warren et al. 2018). Pang et al. (2017) suggested that the lack of secondary olivine veins in plagioclase of HED meteorites is probably due to the different shock behavior of pyroxene and plagioclase, that is, the fractures could be widely developed in pyroxene, but are rare in plagioclase (e.g., fig. 4 of Chen and El Goresy 2000; fig. 2b of Zhang et al. 2011;fig. 2 of Ray et al. 2017; fig. 5e of Zeng et al. 2018b). This interpretation is consistent with our observations in clasts V1 and V2 that contain irregular crystalline pyroxene with fractures/cracks terminating at the boundaries of plagioclase grains (e.g., Fig. 2e).

Formation History of the Olivine Veinlet-Bearing Clasts in NWA 11273
NWA 11273 has bulk-rock composition similar to lunar highland feldspathic meteorites (Fig. 8), suggesting that the studied secondary olivine veinlets hosted in this meteorite are most likely to be formed in lunar highland area. These olivine veinlets are restricted to individual lithic clasts and do not cut across the matrix surrounding the clasts (Figs. 2 and 3). This implies that these olivine veinlets were formed before the whole sample breccia-forming event.
The petrological and mineralogical features of clasts V1-V5 suggest that the formation of secondary olivine veinlets in NWA 11273 includes a variety of complex processes (see Fig. 12). (1) The basaltic host rocks of clasts V1-V5 were crystallized from an Fe-rich basaltic magma; (2) the basaltic host rocks experienced postmagmatic re-equilibration (e.g., the exsolution lamellae of clinopyroxene; Figs. 2 and 3) and also were shocked by an impact(s) to create a brittle fracture network; (3) fluids percolated through the host rock along these preexisting fractures, forming irregular secondary olivine veinlets with accessory Ca-plagioclase grains; (4) many of the secondary olivine grains in clasts V1-V5 show 120°triple junction grain boundaries (e.g., Figs. 2e and 3b), suggesting that a later thermal metamorphic event of~700-1000°C occurred (temperatures estimated from the two-pyroxene thermometer of Lindsley and Andersen 1983;Figs. 6b-d). (5) Finally, the olivine veinlet-bearing host rock was impact ejected from their bedrock source, mixed with other lunar components (e.g., mineral fragments and highlands lithologies), and then consolidated to form a regolith breccia, which gave us meteorite NWA 11273.

Implications of the Secondary Olivine Veinlets in NWA 11273
In addition to the secondary olivine veinlets in highlands breccia NWA 11273, similar secondary olivine veinlets have also been previously observed in a few basaltic lunar samples (i.e., basaltic breccia meteorite MET 01210; Arai et al. 2010, and Apollo 14 mare basalt 14072; Warren et al. 2018). These findings imply that the formation mechanism for these olivine veinlets could be an underappreciated geological process within the lunar crust.
In HED meteorites, metasomatic system P-and S-bearing mineral phases (i.e., merrillite and troilite) have been detected associated with the secondary fayalitic olivine veinlets (e.g., Barrat et al. [2011] and references within; Pang et al. 2017). However, in this study, the secondary olivine veinlets in NWA 11273 were observed associated with only a few accessory mineral grains of chromite, pyroxene, and plagioclase (Figs. 2 and 3). The P-rich phases (e.g., phosphate) and S-rich minerals (e.g., troilite) were not detected in the olivine veinlet in clasts V1-V5 (Figs. 2 and 3; Table 1, see also Figs. S3.1-S3.5). This may indicate that the concentrations of P and S (or F and Cl, which are significant for the formation of the mineral apatite; McCubbin et al. 2015) in the fluid that formed the secondary olivine in NWA 11273 (at least in clasts V1-V5) was relatively low.
In eucrite NWA 5738, Warren et al. (2014) observed a set of ferroan olivine and Ca-plagioclase curving microveins that are likely formed from aqueousrich fluids (~700°C). These authors indicated that the fluids could have come from an impactor that mobilized rock-forming elements (i.e., Fe, Mg, Si) from 4-Vesta target rocks, before re-precipitation occurred as veinlets . For the source of fluid related to secondary olivine veinlets in NWA 11273, it is possible that this fluid originated from an impactor (hydrated asteroid or comet); however, we favor a lunar origin for the fluids for the secondary olivine in NWA 11273. This is supported by the mineral composition (e.g., Fe/Mn ratios for mafic minerals and plagioclase composition) of the secondary olivine veinlets in NWA 11273, which are similar to the lunar samples (Figs. 5 and 7).
Metasomatic fluids have evidently played a complex role in different lunar magmatic systems. Lunar highland rocks show multiple lines of mineralogical and textural evidence for the metasomatism (e.g., Bell et al. 2015). For example, the observed sulfide replacement textures in Mg-Suite lithologies of lunar samples 67915 and 67016 are indicative of the presence of S-bearing fluid in lunar crust Bell et al. 2015). Dickinson et al. (1989) suggested that the source region for Apollo 14 aluminous mare basalts may have assimilated volatile-rich (Ge-rich) materials that had been metasomatized within the lunar mantle. In addition, during the thermal degassing of lunar regolith induced by the heat of emplaced basalts or nearby impact events, H-Cl-bearing volatile phases could have been released from the surface regolith to interact with the surface materials, modifying the dD and d 37 Cl signatures in lunar volcanic/magmatic samples (e.g., Greenwood et al. 2011;Tart ese et al. 2014;Pernet-Fisher 2016;Treiman et al. 2016;Potts et al. 2018). Taken together, these findings indicate that lunar metasomatic fluids were commonly enriched in volatiles (e.g., P, S, Ge, and Cl; McCubbin et al. 2015; Barnes et al. 2016aBarnes et al. , 2016bPotts et al. 2018), REE (Neal and Taylor 1991), and/or water/hydrogen (e.g., Barnes et al. 2014Barnes et al. , 2016cMcCubbin et al. 2015;Robinson et al. 2016). However, the fluid responsible for the studied secondary olivine veinlets in NWA 11273 is significantly different from previous studies, as it is likely volatile element (e.g., S, P, or Cl) depleted.
In this study and previous investigations (e.g., Arai et al. 2010;Warren et al. 2018), only small amounts of microscale (<~30 lm in width) secondary olivine veinlets have been observed. The low abundances of veins identified in lunar samples indicate that fluid flow was rare (and small scale) on the Moon. If a large amounts of a fluid moved (flowed) through a rock, then larger portion of olivine veinlet-bearing clasts and larger scale (e.g., hundreds of microns to several millimeters in size) fluid deposited products would be expected. The existing sample collection gives a tantalizing hint of this poorly understood process. The upcoming Chang'e-5 sample return mission to the Mons Rumker and intermediate Ti mare basalt lava fields (Qian et al. 2018) provides a new opportunity to access new types of lavas and an evolved silicic volcanic dome. Future studies should be carried out to seek and characterize evidence of the past fluid mobility and deposition within these returned volcanic rocks in order better relate the Moon's magmatic history to different volatile reservoirs and fluid mobility processes.

CONCLUSIONS
NWA 11273 is a feldspathic breccia, compositionally (e.g., Al 2 O 3 = 25.1 wt%, Sm = 0.49 ppm, and Th = 0.17 ppm) similar to other lunar highland meteorites. In this sample, we have recognized five lithic pyroxene clasts that host secondary olivine veinlets (i.e., clasts V1-V5). These olivine veinlets cross cut their host pyroxene grains. They exhibit similar mineralogy to each other and are mainly composed of olivine with a few accessory mineral grains of chromite, pyroxene, and plagioclase. Compared with other olivine grains in lithic clasts and matrix of NWA 11273 (Fo 54.3-83.1 ), the olivine veinlets in clasts V1-V5 are relatively fayalitic  ). The mineralogy of clasts V1-V5 suggests that their host pyroxene represents basaltic lithologies, similar to low-Ti or VLT mare basalts.
By analogy with the secondary olivine veinlets observed in HED meteorite, the fluid deposition processes seem the most plausible mechanism for interpreting the formation of secondary olivine veinlets in NWA 11273. Petrological and mineralogical features of clasts V1-V5 reveal that formation history of secondary olivine veinlets in NWA 11273 includes a variety of complex processes; i.e., (1) the basaltic host rocks formed in lunar crust, (2) postmagmatic re-equilibration and fracturing by impact events, (3) secondary veinlets were formed by deposition from metasomatic fluid, and (4) NWA 11273 was consolidated with the olivine veinletbearing clasts (Fig. 12).
The observed secondary olivine veinlets in NWA 11273 may be evidence of fluid deposition at or close to the lunar surface. On the basis of the currently available data, we propose that such fluid was likely S,P-poor and was from an endogenic origin on the Moon. To better understand the formation mechanism of secondary olivine veinlets on the Moon, more olivine veinlet-bearing lunar clasts are need to be examined and comparative studies of olivine veinlets between lunar samples (such as those that will be collected by the upcoming Chang'e-5 mission) and HED meteorites are also necessary. In addition, the systematic studies of in situ oxygen isotope and trace elements, in the future, would allow for greater revealing the geological records of such secondary olivine veinlets in lunar samples.
Acknowledgments-We thank Hao Zhu (Beijing Planetarium) for donating lunar meteorite NWA 11273. We thank Chris Herd for the editorial handling and two anonymous reviewers for their constructive comments on an earlier version of this manuscript. We also thank Dr. Fagan and an anonymous reviewer for their fruitful comments, as well as Dr. Righter for editorial handling. X.Z. thank YiZhi Liu and Yan Zhong for their help during the EMPA measurements at the Guilin University of Technology. This work was financially supported by the National Natural Science Foundation of China (nos 41473067, 41490630, 41572037, and 41941003)   Editorial Handling-Dr. Kevin Righter

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article:. Figs. S3.1-3.5. The BSE image and X-ray elemental mapping results for clasts V1-V5 .  Table S1. The mineral composition of the olivine veinlet-bearing clasts V1-V5 in NWA 11273.