Introduction

Based on the current theories of the Moon formation, solidification of the lunar magma ocean (LMO) produced in the aftermath of the Moon-forming giant impact resulted in a chemically stratified lunar mantle consisting of mafic silicate cumulates (primarily olivine and orthopyroxene) and ilmenite-bearing Fe,Ti-rich cumulates, a low-density anorthositic lunar crust, and a uniform residuum interlayer that is rich in K, rare earth elements (REE), and P, known as urKREEP between the lunar crust and mantle1,2,3,4. This primordial differentiation was possibly followed immediately by density-driven overturn and partial melting of the cumulate mantle leading to secondary magmatism on the Moon such as the lunar magnesian-suite (Mg-suite) rocks (e.g.,5,6). The Mg-suite rocks, including dunites, troctolites, norites, and gabbronorites, represent a rare subset of lunar highland samples with endogenous igneous origin and compositionally distinct from the anorthositic rocks1,4,7,8. Most Mg-suite rocks appear as plutons intruded into the primary anorthositic lunar crust8, while some of them may have erupted as lavas onto the Moon’s surface9.

The role of KREEP in the petrogenesis of Mg-suite rocks remains a long-standing enigma. Early studies highlighted that many Mg-suite rocks from the Apollo landing sites exhibit elevated trace element signatures, despite their high Mg# characteristic10,11. Consequently, a hybridized cumulate Mg-suite source containing a residual KREEP layer has been suggested1,7,12,13. However, the recent discovery of KREEP-depleted Mg-suite-related clasts in samples returned from other missions and meteorites has sparked debates on whether the petrogenesis of the Mg-suite rocks has to be KREEP-related6,9,14,15,16,17,18,19,20. Fractional crystallization of low-degree partial melts from KREEP-free mantle sources has been proposed as a mechanism to replicate the mineralogy of KREEP-depleted Apollo-15 Fe-norites, suggesting a potential new magmatic evolution trend on the Moon independent of KREEP21. Nonetheless, further research regarding detailed major- and trace-element analysis and modeling is essential to demonstrate the geological setting of this KREEP-free trend and its genetic connection to Mg-suite-related clasts in meteorites5,22.

The spatial distribution and formation timing of Mg-suite rocks on the lunar surface also remain uncertain. Most Mg-suite samples collected by the sample-return missions are breccias, complicating efforts to determine their precise provenance. These rocks are hypothesized to be linked to nearside-specific magmatism associated with the Procellarum KREEP Terrane (PKT)4. However, remote sensing observations and geodynamic modeling suggest that Mg-suite rocks may be distributed globally across the Moon6,8,23,24. Furthermore, the radiometric ages of Mg-suite-related rocks collected at Apollo sites cluster around 4.33–4.36 Ga, hinting at a nearly contemporaneous formation25,26. If Mg-suite rocks are indeed globally distributed, the 4.33–4.36 Ga interval may represent a global-scale magmatic event, necessitating further geochemical and chronological evidence from Mg-suite lithologies in meteorites. However, the linkage of meteoritic Mg-suite clasts to their Apollo bulk-rock counterparts in terms of petrogenesis is complicated by the lack of age constraints for meteoritic samples7. Additionally, Mg-suite samples identified in meteorites are typically small (e.g., 1–10 mm) clasts, and whole-rock compositions derived from these clasts may not accurately reflect those of their parent rocks, making it difficult to constrain their exact origins.

Officially approved on 19th March 2023, Arguin 002 is the first whole-rock sample in the lunar meteorite collection that has been compellingly classified as a norite (see Meteorite Bulletin Database). Arguin 002 is enveloped by a prominent, fresh fusion crust (Fig. 1), suggesting a recent landing and minimal terrestrial alteration. This research presents the detailed petrology, mineralogy, geochemistry, and U-Pb chronology of Arguin 002. Our results reveal that this meteorite is a chemically pristine but heavily-shocked, KREEP-depleted Mg-suite rock with a plutonic origin, formed contemporaneously with Apollo Mg-suite rocks but in geologically distinct locations. These findings can advance our understanding of the petrogenesis and spatial distribution of Mg-suite rocks beyond the PKT and shed light on the processes that triggered lunar Mg-suite magmatism.

Fig. 1: Macro and microstructure of the norite meteorite Arguin 002.
figure 1

a Mosaic backscattered electron (BSE) image of Arguin 002 with enhanced contrast, taken with TIMA. b Image depicting the mineral distribution in Arguin 002, taken with TESCAN Integrated Mineral Analyzer (TIMA). c Hand specimen photo of Arguin 002.

Full size image

Results

Petrology and mineralogy

Arguin 002 is primarily composed of orthopyroxene (47.2%), plagioclase (36.0%), and impact melt glass (9.8%). Lesser amounts of clinopyroxene (4.6%) and silica (2.1%) are also present (Fig. 1). Trace phases in the meteorite include merrillite, apatite, chromite, troilite, FeNi metal, ilmenite, and zirconium-bearing phases (Fig. 1).

Orthopyroxene and plagioclase in Arguin 002 are coarse-grained, both with typical sizes ranging from 1–5 mm. Some orthopyroxene grains may reach over 10 mm, as inferred from the near-complete parallelism of exsolved lamellae in these grains across the imaged area (Fig. 1). Clinopyroxene occurs exclusively as exsolution patches or lamellae within orthopyroxene, typically measuring 5–10 μm in width (Fig. 1). Most plagioclase grains have been transformed into maskelynite, and some pyroxene grains exhibit mosaic extinction under cross-polarized light, indicative of a high degree of impact metamorphism (Supplementary Fig. 1). The compositions of these primary minerals are homogeneous, with a range of Wo3.2–4.7En62.0–65.3Fs30.8–34.0 for orthopyroxene, Wo39.5–42.0En42.0–44.0Fs14.0–17.4 for clinopyroxene, and An92.3–93.1Ab6.4–7.2Or0.4–0.6 for plagioclase. When plotted on a Mg# versus An# diagram (Fig. 2a), the mineral compositions of Arguin 002 fall into a gap between the typical Apollo Mg-suite lithologies and FANs, but overlapping with the compositions reported for “atypical” Fe-noritic clasts found in Apollo-15 regolith breccia17 and Mg-suite-related clasts found in brecciated meteorites (Fig. 2a). Together, these data illustrate a steeper Mg-suite differentiation trend than is classically portrayed in Fig. 2a.

Fig. 2: Major and trace element diagrams for the bulk rock and minerals of Arguin 002.
figure 2

a Mg# of orthopyroxene vs. An# in plagioclase for Arguin 002, Apollo norites/alkali norites/gabbronorites, atypical Apollo-15 Fe-norites, Mg-suite-related clasts in meteorites, and the compositional ranges of typical Apollo Mg- and alkali-suite, and ferroan anorthosites (FANs). Results from thermodynamic calculations for partial melts of Mg-suite analog sources with varying amounts of KREEP are represented by colored curves (Elardo2023:31; Prissel2020:21). b The bulk-rock rare-earth element (REE) pattern of Arguin 002, compared with those of typical Apollo norites/alkali norites/gabbronorites and atypical Apollo-15 Fe-norites. c–e The REE patterns of silicate minerals (clinopyroxene, orthopyroxene, plagioclase) in Arguin 002, compared with the compositional ranges of Apollo Mg-suite, FANs, and Mg-suite-related clasts in meteorites. The typical Apollo data in (a, b) and (c–e) are from4 and12 respectively. The Apollo-15 Fe-norite data are from17. Data of Mg-suite-related clasts in meteorites are from15,18,20,27. The compositional ranges (colored shades) in (a) are from Ref. 31.

Full size image

Impact melt glass is ubiquitously dispersed along inter-crystal boundaries and intra-crystal fractures (Fig. 1), with a composition resembling a mixture of plagioclase and pyroxene. Large, continuous impact melt glass can reach widths up to 1 mm, occasionally featuring elliptical bubbles (Fig. 1). The CaO/Al2O3 and MgO/Al2O3 ratios of the glass (Supplementary Fig. 2) support its impact origin.

Trace and accessory minerals are typically smaller in grain size and are commonly found in interstitial areas or hosted within the primary minerals (Fig. 1). Silica appears as anhedral crystals with a typical size of 200–500 μm, often associated with plagioclase and impact melt. Apatite and merrillite are often embedded in orthopyroxene or plagioclase, or associated with impact melt (Supplementary Fig. 3a, b). While typically sized between 5–20 μm, some merrillite grains are larger, reaching up to 30 × 200 μm in plagioclase. EPMA data confirm apatite as fluorapatite (F0.76–0.86Cl0.08–0.14OH0.06–0.10), and merrillite containing trace amounts of Cl ( ~ 100 ppm). The zirconium-bearing phases in the sample include zircon, baddeleyite and zirconolite. Zircon grains are 30–50 μm in size, euhedral to subhedral, and usually coexist with plagioclase and silica (Supplementary Fig. 3c). Baddeleyite and zirconolite grains are relatively small, typically less than 5 μm (Supplementary Fig. 3d). FeNi metal and troilite grains are typically 2–10 μm in size, with Ni/Co ratios (0.96–2.85 for FeNi metal and 0.31–2.08 for troilite) consistent with those of indigenous metal/sulfides in pristine lunar rocks (Supplementary Fig. 4).

The bulk Na2O and K2O contents of the sample are below the alkali content threshold typical of alkali-suite rocks ( > 0.1 wt% K2O and 0.3 wt% Na2O,3), confirming its classification as a Mg-suite rock. The bulk Mg# of the sample (65.4–68.6) indicates that it is too Mg-rich to be in equilibrium with its constituent pyroxene (Mg# = 65.5–69.0). This discrepancy suggests that the bulk composition does not represent the parental magma, consistent with the cumulate nature of the sample. Utilizing the partition coefficient \({K}_{Fe-Mg}^{opx-liq}\) = 0.28, the Mg# of the parental melt calculated from bulk pyroxene ranges from 34.7 to 38.4. This value indicates that the parental magma of Arguin 002 is more evolved than most Apollo norites (parental magma Mg# = 36–60;10) and noritic clasts in meteorites (parental magma Mg# = 43–55;18,27). In contrast, the evolution extent of its parental magma is comparable to some of the evolved Fe-noritic clasts reported in the Apollo-15 regolith breccia17.

Geochemistry

The bulk-rock trace-element contents of Arguin 002 measured in the three parallel analyses show moderate variation, with an enriched signature of La  ~ 13.6–25.0 × CI to Lu  ~ 15.7–20.2 × CI, accompanied by a flat to slight negative Eu anomaly. The chemical differences between the analyses are likely due to minor variation of phosphate abundances in the samples (Supplementary Fig. 5a). Despite this variation, the average rare-earth element (REE) content of Arguin 002 is similar to, but slightly depleted in light REEs compared to typical Apollo norites (Fig. 2b). Additionally, the REE content of Arguin 002 is on the same order of magnitude with some atypical Apollo-15 Fe-norites, which show enrichments of  ~ 10–30 × CI (Fig. 2b).

Given the relatively low bulk concentrations of Sm (1.59–3.18 ppm) and Th (0.41–0.86 ppm), Arguin 002 is located near the conjunction line of feldspathic highland terrain (FHT) and maria on the bulk Sc-Sm and FeO-Th plots, distinguishing it from both known lunar meteorites and samples from return missions (Fig. 3). Furthermore, the bulk Ni/Co ratio (0.79–0.83) of the sample is among the lowest in lunar meteorites (Supplementary Fig. 6), indicating minimal meteoritic contamination and supporting the pristine nature of the sample.

Fig. 3: Representations of bulk compositional spaces for lunar samples.
figure 3

a Sm vs Sc diagram; b FeO vs Th diagram. The bulk composition of Arguin 002 is highlighted using orange pentagrams. Compositional spaces (FHT, PKT, maria) defined by returned lunar samples and the compositional data points of 278 lunar meteorites71 and Apollo-15 Fe-norites17 are plotted for comparison. A number-density plot technique is employed, where brighter colors indicate higher concentrations of data points, reflecting a greater number of meteorites in those regions. The diagrams are adapted from71,72.

Full size image

The REE concentrations in the major minerals of Arguin 002 generally fall between those observed in the major minerals of Apollo Mg-suite samples and those in Mg-suite-related clasts found in lunar meteorites (Fig. 2c–e). Both clinopyroxene and orthopyroxene show pronounced negative Eu anomalies (\(Eu/E{u}^{* }=E{u}_{N}/\sqrt{S{m}_{N}\times G{d}_{N}}\) = 0.01–0.02, subscript “N” refers to chondrite-normalized concentrations), suggesting substantial plagioclase fractionation of their parental magma (Fig. 2c, d). The impact melt glass has a slightly LREE-enriched pattern (La  ~ 11–33 × CI, Lu  ~ 9–18 × CI) and varying Eu anomalies (Eu/Eu* = 0.5–3.0), consistent with mixtures of pyroxene and plagioclase in differing proportions (Supplementary Data 1). Apatite and merrillite are the primary REE hosts, with La contents of 350–870 × CI for apatite and 46,000–47,000 × CI for merrillite, and Lu contents of 69–220 × CI for apatite and 4500–4800 × CI for merrillite. Both phases exhibit deep negative Eu anomalies (Eu/Eu* = 0.02–0.08 for apatite; Eu/Eu* = 0.003–0.006 for merrillite) (Supplementary Data 1). The uniformity of REE patterns and concentrations among grains of the same mineral phase suggests chemical equilibrium and inheritance from the parental melt, arguing against post-crystallization trace-element addition via metasomatism10.

Chronology

U-Pb isotopic analyses were performed on four spots from two apatite grains, three spots from one zircon grain, and four spots from three merrillite grains. Those grains were chosen due to their sizes large enough for SIMS analyses with spot sizes of 4 μm. All the analyzed spots have U-Pb isotopic compositions (corrected for minor terrestrial common Pb based on 204Pb) lying on or close to the Tera-Wasserburg concordia (Fig. 4a), confirming closed U-Pb systems with minimal Pb loss in the analyzed domains. The analyses yielded weighted mean 207Pb/206Pb dates of 4343  ± 17 Ma for apatite, 4335  ± 14 Ma for zircon, and 4280  ± 22 Ma for merrillite (Fig. 4b; uncertainties reported here and below are 95% confidence intervals calculated using Isoplot 3.75,28). The 207Pb/206Pb dates of apatite and zircon agree with each other within error, and together define a weighted mean date of 4341.5  ± 9.3 Ma, while the 207Pb/206Pb date of merrillite is resolvably younger than that of zircon and apatite by 62  ± 24 Ma.

Fig. 4: SIMS U-Pb dating results of Arguin 002.
figure 4

a Tera-Wasserburg plot of zircon and phosphate analyses. b Weighted mean 207Pb/206Pb ages for zircon and phosphates.

Full size image

Discussion

Evidence for a KREEP-free Mg-suite source region

Lunar Mg-suite rocks sampled during Apollo missions are typically enriched in trace elements, suggesting interactions with KREEPy reservoirs in their petrogenesis8,12. These rocks are retrieved from the lunar nearside and believed to have a genetic association with urKREEP. Potential mechanisms for the incorporation of the KREEP signature include melting of a hybridized lunar mantle or assimilation of urKREEP near the base of lunar crust during the magma ascent5,8. Prior studies29,30 have also proposed that the cumulate lithologies of the Mg-suite may result from fractional crystallization of KREEP-like magmas. More recent thermodynamic modeling supports that typical Apollo Mg-suite rocks likely originated from sources with 10–25 wt% KREEP components31.

In contrast, our analysis suggests that Arguin 002 was derived from a source region with considerably less KREEP content than typical Apollo Mg-suite rocks, based on the following observations: (1) The lower Mg# in both bulk rock and orthopyroxene of Arguin 002 indicates an evolved magma origin (Fig. 2a). Despite this evolved nature, the REE abundances in both the bulk rock (Fig. 2b) and mineral phases (Fig. 2c–e) are similar to or lower than those of typical Apollo norites, suggesting that its parental magma contained lower incompatible element concentrations (i.e., KREEP components). (2) The Sm and Th concentrations in Arguin 002 differ from those in typical Apollo samples (Fig. 3), pointing to a source region likely located far from the PKT, with minimal interaction with KREEPy materials32,33.

Arguin 002, atypical Apollo-15 Fe-noritic clasts, and KREEP-free Mg-suite-related clasts from regolith meteorites define a steeper Mg-suite differentiation trend in orthopyroxene Mg# vs. plagioclase An# space, diverging from the canonical trend of typical Apollo Mg-suite rocks (Fig. 2a). Thermodynamic modeling of major-element evolution21,31 suggests that this trend can arise from a primary magma formed without KREEP addition (Fig. 2a). Prissel & Gross21 propose that KREEP-free Mg-suite lithologies form through high-degree ( ~ 50–87% for troctolites and  > 87% for norites) fractional crystallization of a primary magma generated by low-degree ( ~ 1–2 wt%) batch melting of deep (2.1 GPa, equivalent to  ~ 420 km in depth) primitive LMO cumulates. Given the similarities of Mg# and An# between Arguin 002 and Apollo-15 Fe-norites, the major-element compositional evolution of minerals in Arguin 002 could have formed through a similar process. Does the trace-element compositional evolution of minerals in Arguin 002 also support this scenario?

To explore the genetic relationships between KREEP-free Mg-suite lithologies and Arguin 002, we modeled trace-element partitioning during mantle melting and fractional crystallization. Results show that REE contents and patterns of KREEP-free Mg-suite-related clasts in lunar meteorites can be reproduced by  ~ 80–95% fractional crystallization of olivine and plagioclase from a primary melt generated by 1% partial melting of a KREEP-free mantle source (Fig. 5). A little higher degree of fractionation ( ~ 95–97%) of this primary melt would yield an evolved magma with REE contents and patterns consistent with the parental magma of Arguin 002 (Fig. 5). By contrast, the REE abundances in Apollo Mg-suite parental magmas are too high to replicate through similar processes, indicating that their source regions likely contain notable amounts of KREEP (Fig. 5). These findings support the hypothesis that KREEP-free norites like Arguin 002 originated from a mantle region similar to that of KREEP-free Mg-suite-related clasts in meteorites but experienced higher degrees of fractional crystallization, resulting in the observed REE enrichment and deeper Eu anomalies.

Fig. 5: Trace-element partitioning behaviors of a KREEP-free lunar primary mantle source (LPUM) during partial melting and fractional crystallization.
figure 5

The figure displays modeled REE contents of the source region, 1% partial melts from the source, and evolved melts (dashed black curves) at varying degrees of fractionation. These are compared with calculated parental magmas of Mg-suite-related clasts from NWA 761127, NWA 1040115, NWA 1098618, and NWA 1178820, illustrated with colored curves. The REE range for Apollo Mg-suite parental melts is sourced from31. Numbers adjacent to the dashed black curves indicate the fractionation degree for each evolved melt. Error bars represent 1σ uncertainties.

Full size image

In summary, our study suggests that Arguin 002 originated from a distinct KREEP-free Mg-suite source region, likely part of the primitive lunar lower mantle, composed primarily of KREEP-free olivine cumulates from LMO solidification1,21. KREEP-free Mg-suite magmatism would initiate through low-degree ( < 2%) partial melting of this primitive source, followed by high-degree ( > 90%) fractional crystallization of olivine, plagioclase, and orthopyroxene, forming KREEP-free lithologies such as troctolites and norites. The limited partial melting of the KREEP-free mantle source may explain the lower abundance of Mg-suite materials in the lunar highlands compared to the PKT region, which is consistent with recent remote-sensing findings6. Future radiogenic isotope analyses (e.g., Rb-Sr, Sm-Nd, U-Pb) will be crucial to confirm the KREEP-free nature of Arguin 002 and its source region.

The petrogenesis of KREEP-free Mg-suite rocks

The petrogenesis of lunar Mg-suite rocks remains a subject of debate. Current models for their origin can be grouped into three categories (5 and references therein): (1) products of LMO crystallization, (2) solidification of impact melts, and (3) crystallization of partial melts derived from mantle regions at varying depths and KREEP contents, followed by emplacement into the deep crust or onto the lunar surface.

Model (1) is unlikely for Arguin 002, as its evolved whole-rock chemistry (Mg# = 65.4–68.8) and silica-bearing composition do not align with the Mg-rich (Mg#  > 80) mantle lithologies hypothesized to result from LMO crystallization2,7,34,35. Additionally, Arguin 002 is unlikely to represent a major component of the lunar primordial lower crust, as Mg-rich noritic materials are not expected to directly derive from the LMO36.

Furthermore, while some thermodynamic models support norite formation via fractional crystallization of impact melt from a mixture of mantle dunite (77–80 wt%) and crustal anorthosite (13–20 wt%)37,38, this mechanism appears inapplicable to Arguin 002 for several reasons. First, the presence of coarse-grained pyroxene and plagioclase strongly indicates that Arguin 002 is a pristine plutonic rock rather than an impact melt product. Second, impact melts from such mixtures are typically plagioclase-saturated and favor Mg-Al spinel crystallization39. However, Arguin 002 contains chromite rather than Mg-Al spinel, indicating it could not have crystallized from these melts. Third, clinopyroxene thermobarometric calculations place the formation conditions of Arguin 002 at 0.6–1.0 kbar and 1129–1144 °C, corresponding to depths of  ~ 12–20 km within the lunar crust (Supplementary Fig. 7). If Arguin 002 formed from a thick impact melt sheet at such depths, zirconium-bearing phases in the sample would exhibit microstructures indicative of reversion from a cubic-ZrO2 phase, which forms at extremely high temperatures ( > 2100 °C) characteristic of thick impact melt sheets40. However, no such microstructures were identified in Arguin 002 using Raman spectrometer. Fourth, the low Ni/Co ratios observed in both FeNi metal and the whole rock suggest minimal assimilation of impactor materials, further challenging the impact-related origin hypothesis.

The genesis of Arguin 002 is more consistent with endogenous magmatic processes occurring after the formation of the primordial lunar crust, similar to the model (3). The similarities in mineral compositions (Fig. 2a) and bulk-rock trace-element contents (Fig. 2b) between Arguin 002 and Apollo-15 Fe-norites possibly suggest a similar genesis through fractional crystallization of melts derived from magnesian LMO cumulates, with minimal KREEP involvement21. The bulk major- and trace-element differences (e.g., Eu anomaly, Sm, Sc, Th, and FeO contents) between Arguin 002 and Apollo-15 Fe-norites (Fig. 2b, 3) can be attributed to a higher modal abundance of plagioclase (50–80%) in Apollo-15 Fe-norites (Supplementary Fig. 5b), likely due to the earlier fractionation and accumulation of plagioclase relative to orthopyroxene, as well as potential sampling bias.

Given that Arguin 002 and related KREEP-free Mg-suite rocks most likely originated from deep-seated emplacement within the lunar crust, it is probable that this rock was transported to the lunar surface only during the formation of large impact basins, which possess the capability to excavate rocks from several kilometers within the crust (e.g.,41). Thus, we propose that the formation process of KREEP-free Mg-suite rocks involves the following stages: (1) low-degree ( < 2%) partial melting of KREEP-free primitive magnesian cumulates in the lunar mantle to form the primary magma (Fig. 6a), possibly triggered by the mantle overturn; (2) diapiric ascent of the magma through the overlying anorthositic crust, forming Mg-suite cumulates at depths of  ~ 12–20 km (Fig. 6b) following high-degree ( > 80%) fractional crystallization of Mg-rich minerals and plagioclase; (3) transportation of the cumulates to the lunar surface by large basin-forming impact events (Fig. 6c); (4) subsequent smaller impacts ejecting the exposed troctolitic/noritic materials to Earth as meteorites (Fig. 6d). This implies that lunar Mg-suite magmatism is not necessarily driven by the radiogenic heating of KREEP components, but rather, can be caused by alternative endogenous processes.

Fig. 6: A schematic diagram illustrating the endogenous genesis of the lunar meteorite Arguin 002 and related KREEP-free Mg-suite rocks.
figure 6

a A KREEP-free primary melt forms via partial melting of the deep lunar mantle, potentially triggered by mantle overturn. b The partial melt undergoes fractional crystallization and intrudes into the lunar crust, where it slowly cools to form noritic cumulates. c A major impact event transports these noritic materials to the lunar surface, producing ejecta. d A subsequent impact event propels KREEP-free Mg-suite rocks from the lunar surface to Earth.

Full size image

The timescale and global distribution of Mg-suite magmatism

Radiometric dating of Arguin 002 provides essential constraints on the formation timeline of Mg-suite rocks. U-Pb dating of apatite and zircon in this study yields a consistent age of 4341.5  ± 9.3 Ma. We interpret this as the primary crystallization age of Arguin 002 for two reasons: (1) the well-defined Raman peaks and uniform contrast in cathodoluminescence images of apatite suggest minimal effects from post-crystallization shock processes (Supplementary Fig. 8); (2) the analyzed zircon grain does not exhibit granular neoblastic recrystallization structure, indicating that it effectively preserved its original 207Pb/206Pb age during impact events (e.g.,42,43). Conversely, the younger 207Pb/206Pb age of merrillite (4279.6  ± 22.1 Ma) likely reflects impact-induced resetting. The presence of residual apatite within merrillite grains suggests that merrillite may have formed from the transformation of pre-existing apatite (Supplementary Fig. 3a, b) due to impact-induced shock compression44.

High-precision isochron ages for Apollo Mg-suite rocks cluster between 4.33 and 4.36 Ga (e.g.,7,25,45), consistent with the 207Pb/206Pb age of Arguin 002 determined from zircon and apatite (Fig. 7). This synchronization suggests that Arguin 002 may have occurred concurrently with that of Apollo Mg-suite rocks. Through the comparison of bulk-rock data with lunar surface remote sensing spectral data, we have identified four potential source sites for Mg-suite rocks similar in composition and mineral modal abundances to Arguin 002, with the northeastern part of the South-Pole Aitken (SPA) basin being the most likely (Supplementary Fig. 9). This launch region satisfies both dynamic37 and age constraints22, suggesting that the excavation of noritic lithology from the deep lunar crust ( ~ 12–20 km) likely occurred contemporaneously with the formation of the SPA basin (Supplementary Text 2.2). Even if the  ~ 4.34 Ga age recorded by apatite and zircon reflects a complete resetting of the U-Pb systematics during a large impact event, this age remains consistent with the estimated formation time of the SPA basin within error22, again supporting an SPA-basin origin for Arguin 002 (Supplementary Text 2.2). Thus, the synchronization of ages and the widespread distribution of Mg-suite rocks, including Arguin 002 and Apollo samples, indicate that a global event likely triggered Mg-suite magmatism between 4.33 and 4.36 Ga.

Fig. 7: Schematic diagram summarizing important age anchors for lunar formation, LMO differentiation, post-LMO magmatism, and formation of South-Pole Aitken (SPA) basin.
figure 7

The chronological data from22,25,26,73,74 (and references therein) are shown for comparison.

Full size image

Lunar dynamics models suggest that gravitational instability-induced mantle overturn of LMO cumulates may have triggered widespread partial melting of the KREEP-free lunar lower mantle, which matches both the estimated abundances of Mg-suite rocks and their global distribution as detected in remote sensing data6,46,47. Seismological evidence further supports this global mantle overturn occurring after the crystallization of the LMO48. In this context, Mg-suite magmatism could have been rapidly initiated due to gravitational restructuring of the LMO cumulates, potentially driven by the formation of dense, low-viscosity ilmenite-bearing cumulates or even a silicate-driven mantle overturn prior to complete LMO solidification6,49,50. This scenario may explain the age overlap between Apollo Mg-suite rocks and Arguin 002, with the age upper bound of lunar mantle overturn potentially constrained by the formation of the primordial lunar crust ( ~ 4.36–4.37 Ga,25,51). Notably, this age coincides with the formation of the oldest Mg-suite rock (Apollo-17 sample 72215) at  ~ 4.36 Ga25, suggesting a short timeframe for the onset of global Mg-suite magmatism in response to mantle overturn. The age disparity between the oldest Mg-suite rock and Arguin 002 further implies that Mg-suite magmatism was a prolonged event, lasting at least 20–30 Myr.

The Chang’e-6 mission, conducted by the China National Space Administration, has successfully sampled lunar surface materials within the SPA basin (41.623°S, 153.917°W). Recent Sr-Nd-Pb isotope analysis of the Chang’e-6 low-Ti basalts reveal compelling evidence for their derivation from a KREEP-free mantle source52. Noritic materials have also been identified in the Chang’e-6 regolith breccias53, consistent with the widespread noritic lithologies observed near the landing site and across the SPA basin54,55,56. Given the possible origin of Arguin 002 near the Chang’e-6 landing site (Supplementary Text 2.2), it is plausible that Arguin 002 shares a genetic correlation with the noritic materials in the region. The increasing number of lunar samples sourced from KREEP-free mantle regions, as exemplified by the Chang’e-6 samples and meteorites15,20,52, presents an opportunity for deeper insights into Mg-suite magmatism. Future analyses of these samples and their potential genetic links to Arguin 002 could provide critical insights into the timing and spatial distribution of Mg-suite magmatic events, the evolution of the lunar highland crust, and the formation processes of the SPA basin.

Conclusion

Arguin 002 is the first unbrecciated, chemically pristine, fresh whole-rock meteorite definitively identified as an Mg-suite norite. Detailed petrological, mineralogical, geochemical, and chronological analyses in this study yield the following conclusions:

  1. (1)

    The mineral assemblages of Arguin 002 likely formed through high-degree ( > 90 %) fractional crystallization of a primary magma generated by low-degree ( < 2 %) partial melting of the KREEP-free lower lunar mantle. This finding suggests that lunar Mg-suite magmatism is not necessarily driven by or associated with KREEP components.

  2. (2)

    The KREEP-free Mg-suite rocks, exemplified by Arguin 002, likely have an endogenous plutonic origin. Their parental magma may have intruded into the lunar crust at depths of  ~ 12–20 km before being excavated by a large basin-forming impact, such as that created the SPA basin.

  3. (3)

    The crystallization age of Arguin 002 (4341.5  ± 9.3 Ma), aligns with the ages of Apollo Mg-suite and related rocks, supporting the hypothesis that Mg-suite magmatism was a global, nearly synchronous event driven by mantle overturn. The lunar Mg-suite magmatism persisted for at least 20–30 Myr.

Methods

Structural and chemical characterizations of minerals

The mineral modal abundances of Arguin 002 were analyzed using the TESCAN Integrated Mineral Analyzer (TIMA) system. The microstructure of the sample was examined using an electron microprobe equipped with a Cathodoluminescence (CL) imaging detector. Raman spectrometry was employed to identify the spectral characteristics of the minerals within the sample. Additionally, the major and trace element compositions of the mineral phases were determined using an Electron Microprobe Analyzer (EPMA) and Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS), respectively. For detailed descriptions and accuracies of these methods, readers are referred to Supplementary Text 1.

Mineral trace-element analysis

Trace-element concentrations of minerals in the sample were analyzed using LA-ICP-MS at the Wuhan SampleSolution Analytical Technology Co., Ltd., Wuhan, China. Details of the operating conditions for laser ablation and ICP-MS instruments, as well as data reduction procedures can be found in57. A GeolasPro laser ablation system, comprised of a COMPexPro 102 ArF excimer laser (193 nm wavelength; 200 mJ max energy) and a MicroLas optical setup, was employed for laser sampling. Atoms were ionized and ion-signal intensities were registered using an Agilent 7900 ICP-MS instrument equipped with a secondary electron multiplier. Helium served as the carrier gas, and argon, utilized as the supplementary gas, was introduced via a T-connector prior to ICP entry. The laser operational settings were adjusted to 44 μm spot size, 80 mJ energy, and 5 Hz frequency. A range of reference materials (BCR-2G, BHVO-2G, BIR-1G, and GSE-1G) were employed to calibrate trace element concentrations58. Analysis of each spot included a background acquisition of approximately 20–30 s followed by 50 s of data acquisition of the sample. ICPMSDataCal, an Excel-based software, was used off-line for selection and integration of background and analyzed signals, time-drift correction, and quantitative calibration58. Results for the reference materials, detailed in Supplementary Data 2, are consistent with the GeoReM database, showing relative errors of  < 10% (mostly  < 5%) for key elements (e.g., Sc, Th, and REEs) analyzed in this study.

Bulk-rock major- and trace-element analysis

To ensure representative sampling, three chips from different parts of the whole rock, each weighing  ~ 1 g, were crushed to particles smaller than 70 μm using an agate mortar. Then, 40 mg of powder was randomly sampled from each of the three  ~ 1 g portions of powder for analysis. To ensure consistent test results between testing agencies, one powder sample was analyzed at Nanjing FocuMS Technology Co. Ltd, and the other two at China University of Geosciences Beijing. In each testing agency, the crushed sample was digested in a high-pressure PTFE bomb with 0.5 ml of 60 wt% HNO3 and 1.0 ml of 40 wt% HF. The bomb was then placed in an oven at 195 °C for 3 days to ensure complete digestion. After cooling and drying, the sample was mixed with 1 ml of Rh internal standard solution and re-dissolved in 5 ml of 15 wt% HNO3 in the oven at 150 °C overnight. Aliquots of the dissolved sample were prepared for analysis: one (2000 × dilution) for trace elements, analyzed using an Agilent Technologies 7700x quadrupole ICP-MS (Tokyo, Japan); and another (500 × dilution) for major elements (excluding silicon), analyzed using an Agilent Technologies 5110 ICP-OES (Penang, Malaysia). The bulk-rock SiO2 content was calculated by subtracting the combined weights of major and trace element oxides from 100 wt%. USGS geochemical reference materials (BHVO-2 and AGV-2 at Nanjing FocuMS, and AGV-2, GSR-1, GSR-3, and BHVO-2 at China University of Geosciences Beijing) served as quality control benchmarks. The reference material analysis results, detailed in Supplementary Data 2, agree with published data in the GeoReM database, with relative errors of  < 10%.

SIMS U-Pb dating

U-Pb isotopic compositions of zircon and phosphates (apatite and merrillite) in Arguin 002 were analyzed using a CAMECA IMS-1280HR ion microprobe at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS). Prior to analysis, the sample mounts were coated with a 20-nm-thick gold layer. A Gaussian illumination mode was used to focus a primary beam of 16O– to a diameter of 4 μm. A description of the instruments used for analyses and the analytical procedure can be found in59 and60 for zircon and in61 for phosphates, and only a brief summary is provided here.

The first session focused on the U-Pb isotope measurement of zircon. A single electron multiplier was used in ion-counting mode to measure secondary-ion beam intensities by a peak jumping sequence, with each measurement consisting of seven cycles. Measured U-Pb ratios were corrected for U oxides using the power-law relationship between 206Pb/238U versus 238U16O2/238U of reference zircon Plešovice. A second standard, GBW04705 (Qinghu), was also analyzed with zircon grains as an unknown. Long-term uncertainty for the 206Pb/238U measurements of the standard zircons was propagated to the unknowns (1 relative standard deviation (RSD) = 1.5%).

The second session focused on the U-Pb isotope measurement of phosphates. A multi-collector mode with five electron multipliers was used. A hybrid dynamic multicollector technique with five detectors was used to determine the ion beam intensities of 204Pb+, 206Pb+, 207Pb+, 238U+, 232Th16O+, 238U16O+, and a matrix reference peak of 40Ca231P16O3+ at a mass resolution of 7000 (defined at 50% height). The 206Pb+ peak was used as a reference peak for centering the secondary ion beam, energy, and mass calibration. NIST610 glass was used to calibrate the relative yield of different electron multipliers and evaluate the external reproducibility of Pb isotopic compositions. Pb/U ratios were calibrated with a power law relationship between 206Pb/238U and 238U16O/238U relative to an apatite standard of NW-1.

After measurement, data reduction was carried out using the Isoplot 3.0 program28. An average present-day crustal Pb isotopic composition was used for terrestrial common Pb correction62.

Mantle melting and fractional crystallization modeling

It has been demonstrated that the major elemental compositions of primary melts for KREEP-poor/free Mg-suite troctolites can be produced by low-degree ( ~ 1–2%) batch melting at  ~ 2.1 GPa within the cumulate lunar mantle, formed through  ~ 50% equilibrium crystallization of an LMO with a “lunar primitive upper mantle” (LPUM) initial composition21. This study investigates whether the REE patterns and contents of Arguin 002 and KREEP-free Mg-suite-related clasts in lunar meteorites could also arise from this process.

To test this hypothesis, we applied a geochemical model to calculate the REE concentrations in the mantle source, the primary melts, and the parental magmas of KREEP-free Mg-suite rocks, accounting for both partial melting and fractional crystallization processes. The model comprises four steps, as described below in detail.

  1. (1)

    Modeling of mantle source composition: We modeled the bulk REE composition of the mantle source for KREEP-free Mg-suite rocks, formed from the solidification of an LPUM LMO with an initial REE composition of  ~ 2.7 × CI chondrites63. Following the crystallization protocol in1, we used equilibrium crystallization equations to calculate the REE contents of LMO cumulates and residual liquid:

    $${C}_{L}=\frac{{C}_{LMO}}{F+D(1-F)},{C}_{S}=D{C}_{L}$$
    (1)

    where D is the total partition coefficient of REE between mantle cumulates and residual liquid, F is the crystallization degree, and CLMO, CL, and CS are the REE contents of the initial LMO, residual liquid, and fractionated solids, respectively. REE partition coefficients were sourced from63. After 50% equilibrium crystallization of the LMO, with  ~ 3% trapped instantaneous residual liquid (TIRL)1, the resulting mantle source comprised primarily olivine, with La and Lu contents of 0.17 × and 0.34 × CI chondrites, respectively.

  2. (2)

    Modeling of primary melt composition: We calculated the REE concentrations of a low-degree ( ~ 1–2%) partial melt from this mantle source by batch melting, assuming each mineral phase melts in proportion to its modal abundance, as outlined in64,65. The batch melting equation is similar to Eqn. (1), with the only differences being that CLMO is replaced by CL, and F represents the batch melting degree. The batch melting produced a primary melt with relatively flat REE contents from La (16.0 × CI chondrites) to Lu (7.8 × CI chondrites) (1% partial melting), or from La (8.2 × CI chondrites) to Lu (6.4 × CI chondrites) (2% partial melting).

  3. (3)

    Modeling of parental magma compositions: Since Mg-suite rocks are mostly cumulate rocks and their bulk compositions do not reflect parental magma compositions, we reconstructed the REE contents of their parental magmas using REE concentrations in silicate minerals (plagioclase, clinopyroxene, and orthopyroxene) and partition coefficients between these minerals and melts. For consistency, REE partition coefficients were again sourced from63. Besides Arguin 002, we determined the parental magmas of KREEP-free Mg-suite-related clasts from NWA 761127, NWA 1040115, NWA 1098618, and NWA 1178820 using the same approach.

  4. (4)

    Modeling of fractional crystallization of primary magma: To predict the REE contents of KREEP-free Mg-suite parental magmas, we modeled the fractional crystallization of the primary magma using the following equation:

    $${C}_{LR}={C}_{L}{(1-F)}^{D-1}$$
    (2)

    where CLR is the REE content of the residual melt after fractional crystallization, and F is the degree of mineral fractionation. The mineral fractionation scheme follows the results in21 and is summarized here. Before the fractionation degree reaching 87 wt%, the magma first fractionates olivine then co-crystallizes olivine and plagioclase, forming troctolitic lithology. At fractionation degree  > 87 wt%, orthopyroxene begins to crystallize and thus forms noritic lithology. For Arguin 002, the formation of noritic lithology with an observed pyroxene Mg# range of 65.5–69.0 corresponds to a fractionation degree of  > 95 wt%21. If the fractional crystallization modeling in such a range of fractionation degree can reproduce the REE contents and patterns of the parental magma of Arguin 002 calculated in the step (3) within error range, we can infer that the lithology of Arguin 002 likely formed through the aforementioned processes, suggesting a possible genetic relationship with KREEP-free Mg-suite troctolites, as expected by21.

The MATLAB codes for implementing the four-step modeling process are provided in this study. For detailed instructions on the implementation process, please refer to the Code Availability Statement section.

Launch site identification

The procedure for identifying the possible launch sites of Arguin 002 included two steps. First, the bulk-rock chemical composition measured for Arguin 002 was compared with a global dataset of lunar surface chemical compositions. This dataset consist of high-resolution ( ~ 59 m/pixel) maps of key elemental oxides (i.e., Al2O3, CaO, FeO, MgO, and TiO2) spanning a latitude range between  ± 65°66 and a global Th map67. Possible launch sites were preliminarily selected based on matching between the spectral and meteorite elemental contents considering their uncertainties. For the meteorite elemental data, the highest and lowest values measured in the three parallel bulk-rock ICP-MS measurements were used as the error range. For the remote sensing data, estimated errors of  ± 2 wt% for MgO and CaO,  ± 1 wt% for Al2O3 and FeO,  ± 0.5 wt% for TiO2, 2 ppm for Th were applied to reflect the prediction accuracy of individual oxide content on the lunar surface. Based on these selections, in the second step, a global dataset of lunar surface mineral abundances was used to further select source sites that match the modal mineral abundances observed in the meteorite. The global mineral abundance datasets were created from reflectance data obtained by the Mineral Mapper of the JAXA SELENE/Kaguya mission and corrected for topography, with a spatial resolution of  ~ 59 m/pixel and covering latitudes between  ± 50°68. To ensure comparability with the remote sensing datasets, modal abundances of major phases, including clinopyroxene, orthopyroxene, plagioclase, and impact melt in the meteorite sample were first normalized to a sum of 100%. Because the impact melt (modal abundance 10.1%) in the meteorite has a chemical composition representing a mixture of pyroxene and plagioclase with unknown proportions, they add significant uncertainties to the mineral abundances in the pre-shock rock. For this reason, a  ± 10.1% total uncertainty was assigned to the modal abundances of pyroxene and plagioclase when making the comparison, resulting in the following matching thresholds: 38.3–58.4% for orthopyroxene, 0–14.8% for clinopyroxene, and 26.8–47.0% for plagioclase. Given the absence of olivine in the meteorite sample, only the sites with an olivine content  < 8% were considered as potential source sites for Arguin 002, based on the fitting error of the global mineral abundance dataset69. The implementation of these two steps was carried out using the “Map Algebra” toolbox in ArcGIS 10.8 software.