Introduction

The Moon is increasingly considered as a strategic natural resource, supporting not only sustained human exploration of the Moon itself, but also enabling deeper ventures into the wider Solar System1,2. Although the Moon is significantly smaller than Earth, conducting an arbitrary search for resources is impractical unless substantial resource potential can be unambiguously identified1. In a new era of lunar exploration, it is important to develop and further test reliable strategies that can predict concentrations and distributions of valuable mineral deposits.

On Earth, magmatic stratiform ore deposits are critical sources of base metals, such as Ni, and some of the most high-value platinum group metals (PGM) that include the platinum group elements (PGE: Os, Ir, Ru, Rh, Pt, Pd, and Au). Economically exploitable terrestrial stratiform magmatic deposits have PGE concentrations enriched up to one thousand times (~μg/g, or 1/1,000,000) the average abundance of these elements in Earth’s mantle (~ng/g, or 1/1,000,000,000), inheriting fractionated Pt/Ir, Pd/Ir and Ru/Ir ratios during the enrichment process. The largest mafic intrusion on Earth, the ~1,000,000 km3 Rustenberg Layered Suite of the 2.05 Ga Bushveld Igneous Complex, hosts 90% of the world’s PGM resource3,4. The PGM enrichment occurred due to a high silicate melt to sulfide melt ratio, enabling the PGE to be effectively scavenged due to their high sulfide/silicate partition coefficients5,6,7. For the Moon, the feasibility of similar stratiform ore mineralization has been considered8, with a high probability for such deposits due to the five to ten times higher particle settling velocity in lunar magmas compared to Earth.

The abundance of the PGE in the Moon, as for other differentiated bodies, is primarily controlled by metal-silicate equilibration and subsequent post-core formation late accretion processes, partitioning most of the PGE to metallic cores9,10,11,12. The Moon, despite having similar crust and mantle partitioning behavior to Earth, consistently shows ~40 times lower absolute PGE abundances13,14,15. Lunar impact breccias (LIB) samples, on the other hand, show relatively high PGE abundances (ng/g) and, in several cases, supra-chondritic Pt/Ir, Pd/Ir and Ru/Ir ratios, with supra-chondritic 187Os/188Os16,17,18,19,20,21,22. The elevated and fractionated PGE abundances of LIB are similar to those of unmineralized terrestrial layered mafic-ultramafic intrusion magmas (refs. 7,23 and references therein).

Uncertainties exist in understanding the resource potential on the Moon. Critically, most lunar samples studied for PGE abundances are from the Apollo missions and originated from the Procellarum KREEP Terrane, which covers only ~16% of the lunar surface24. Lunar meteorites originating away from the PKT regions are important as they offer a means for understanding if processes were global, or local to the PKT. We analyzed four lunar meteorites: Asuka (A)-881757, Yamato (Y)-86032, Y 981031 (will be referred to as Y981031), and Y 983885 (will be referred to as Y983885) for major, minor and trace elements, including PGE + Re abundances and 187Re–187Os systematics to expand our understanding of global lunar PGE distributions (Fig. 1, Supplementary Figs. 1, 2, and Supplementary Tables S1–S3).

Fig. 1: The PGE content of A−881757, Y981031, Y983885, and Y−86032 compared with published data on lunar rocks.
figure 1

The PGE composition of A-881757 (this study) and MIL 0503514 is compared to Apollo mare basalts14 and LAP basalt meteorites14. The YAMM basalts (A-881757 and MIL 05035) lie well within the previously analyzed endogenous mare basalt samples. Y981031, Y983885, and Y-86032 are compared with previously analyzed impact melt rocks16,17,18,19,20,21,22. CI chondrite normalization from ref. 75.

Full size image

Lunar impact melt sheets have high precious metal potential

Mantle-derived magmatic deposits on Earth enriched in the PGE are formed due to the magma’s ability to interact with crustal wall rocks25,26. To form an economically viable terrestrial PGE deposit, the magma must be of sufficient volume and must become saturated in sulfide to effectively scavenge and concentrate the PGE. Impact melt pools formed on the Moon, especially in the case of basin-forming impacts generated large volumes of impact melt27, comparable to the magma volumes of the world’s largest source of the PGE, the Bushveld Igneous Complex (~1 × 106 km3; ref. 28). On Earth there also exists an example of an impact melt sheet where exactly such a process has occurred. The 1.85 Gyr (billion years) old impact melt sheet of Sudbury, Northern Ontario, Canada29 is the source of significant base metal and PGM resources30. The meteoroid impact at Sudbury flash-melted a mixture of mafic Archean and early Proterozoic crust, forming a large impact melt sheet, where sulfide saturation under superheated conditions scavenged the crustal PGE7,31,32,33.

In the case of the Moon, following the scaling laws of Cintala and Grieve34, formation of basins such as Orientale (~930 km diameter), generated ~1.0 × 106 km3 of impact melt27. This is two orders of magnitude larger than the Sudbury Impact melt pool (~3.1 × 104 km3; ref. 35), demonstrating that impacts on the Moon could generate sufficient melt to facilitate PGE scavenging.

Another important factor to assist PGE scavenging is sulfide saturation. Lunar magmas are sufficiently undersaturated with respect to sulfur36,37, and therefore, they must be driven to sulfide saturation by crustal contamination from the S-bearing lithologies. Among lunar rocks, mare basalts have the highest S concentrations (0.1–0.2 wt%; ref. 11) and within mare basalts, high-Ti basalts are richer in S than low-Ti basalts37,38. We estimated the possibilities of sulfide saturation in the mixtures of representative lunar basalts and highland rocks at low pressure using the O’Neill39 parameterization, and our calculations suggest that without anomalously high impactor addition (>5%), it is difficult to saturate the impact melt sheets with sulfur (Methods; Fig. 2a and Supplementary Table S4).

Fig. 2: Plot showing the effect of sulfide saturation and sulfide to silicate ratio, controlling PGE enrichment.
figure 2

a Comparison of calculated sulfide solubility (expressed as sulfur content at sulfide saturation; SCSS) to measured sulfur contents of lunar basalts, martian shergottites and a combination of mare basalts and highland rocks with variable (1–10%) impactor addition. The SCSS calculations were performed using parameterization from O’Neill39 applicable to 1 bar pressure (Supplementary Table S4). All sulfur from the impactor (CI chondrite) was assumed to be evaporated from the impact event, and the calculation only accounts for the S content of the target lithology. The results suggest low probabilities for melt in large impact melt sheets to undergo sulfide saturation based on lunar S alone. Significant impactor retention (>5%) would be needed to supply enough Ni to reduce the sulfide saturation threshold and enable sulfur saturation. b A plot showing the level of precious metal (Ir, Ru, Pt, and Pd) enrichment assuming variable ratios of silicate to sulfide (mass of silicate/mass of sulfide). The model assumes that the melt to be sulfide saturated and show the possible enrichment as a result of terrestrial-style PGE enrichment process. The pristine average crust with low PGE abundance is incapable of enriching the crust to the level observed in PGE deposits, and only with increasing impactor contribution such values can be achieved.

Full size image

A more likely scenario for melt attaining sulfide saturation is through contamination by preexisting sulfide-saturated regions of the Moon. Notably, even for planets like Mars having a sulfide-saturated mantle, similar processes have been proposed, indicating that sulfide saturation can occur through interaction with sulfide-rich reservoirs rather than solely through intrinsic magmatic evolution40. It has been shown experimentally that at the end of lunar magma ocean crystallization (\(\ge\)90%), the residual melt reached sulfide saturation11. Therefore, sulfide-saturated lithologies should occur at shallow depths (~10 km) close to the lunar surface41. Experimental studies also suggest that sulfur concentration at sulfide saturation (SCSS) decreases with increasing Ni in silicate melts11,37,39. Therefore, the addition of Ni-rich impactor material to already sulfide-saturated shallow LMO lithologies37 provides a reasonable scenario for the precipitation of sulfide in lunar impact melt sheets. All these lines of reasoning collectively suggest that impact melt sheets formed by basin-size impacts on the Moon had a high probability of providing conditions favorable for forming PGE resources.

The precious metal-devoid pristine lunar crust

The lunar mantle has been suggested to be heterogeneous at a range of scales based on major and trace-elemental abundances, mineralogy and isotopic compositions in returned mare basalts and unbrecciated mare basalt meteorites42,43,44,45,46. Despite having large differences in lithophile isotope systematics, there are no clear differences in the PGE + Re abundance of Apollo 12, 15, or 17 mare basalts that can be attributed to mantle source heterogeneity18. Therefore, similar PGE + Re abundance in varied Apollo mare basalts sources are considered to suggest PGE + Re homogeneity in the lunar mantle. However, Apollo basalts only characterize the mantle source beneath the PKT region.

To further understand the PGE distribution in non-PKT regions, we studied meteorite A-881757. Lunar meteorite A-881757 is one of the two unbrecciated gabbros (the other being MIL 05035) from the YAMM meteorite clan47 and is composed of coarse-grained Fe-rich basalts with pyroxene and plagioclase as primary constituents (Supplementary Fig. 1 and Supplementary Text S1; refs. 46,48). The low PGE + Re abundance (~pg/g; Supplementary Text S2 and Supplementary Table S3) of A-881757 and MIL 0503514 indicates that YAMM basalts are pristine and free from impactor contamination. The 187Os/188Os of A-881757 (0.1247 ± 0.0003; Supplementary Table S3 and Supplementary Fig. 3a) is slightly sub-chondritic with fractionated PGE + Re patterns (less fractionated than MIL 05035), similar to previously analyzed low-Mg mare basalt meteorites. The observed differences between MIL 05035 and A-881757 can be effectively modeled as the result of ~0.05% metal and sulfide fractionation using a starting composition of A-881757 (Supplementary Text S3, Supplementary Table S4, and Supplementary Fig. 3b). Overall, the PGE + Re abundance of lunar meteorites and Apollo mare basalts reflects post-core formation accretion during LMO crystallization. Moreover, the observed homogeneity in PGE abundance also implies a spatially homogenous lunar mantle.

Further, we estimated if lunar basaltic or highland lithologies alone possess sufficient PGE to reach the level of enrichment as seen in the terrestrial PGE-enriched deposits. Assuming a lunar pristine crustal average PGE+Re from ref. 13, we modeled the terrestrial enrichment scenario in the impact melt pool, also accounting for subsequent metal segregation. The economically viable terrestrial PGE-enriched deposits show an average of 1–5 g/t of PGE (ref. 7 and references therein). Our numerical modeling suggests that without any impactor contribution, the average crustal PGE abundance (both in highland and mare regions), even with an unusually high silicate to sulfide melt ratio (~10,000:1), can only scavenge ~0.005 g/t PGE; a value too low to host any PGE resources (Methods; Fig. 2b).

Crustal PGE enrichment preserved in lunar impact breccias

In contrast to the low PGE primary crustal sample, the returned samples from the Moon also contain a substantial amount (~95 kg from the Apollo 16 mission alone) of impact breccias. Among impact breccias, several varieties have been identified, such as regolith breccia, fragmental breccia, crystalline/glassy Impact Melt Breccia (IMB) and granulitic breccia49. Although the PGE abundance in the impact breccias ranges from 0.1 to 0.01 times the abundance in CI chondrite16,17,18,19,20,21,22,50,51,52,53,54, the IMB consistently exhibit high PGE abundance with relatively flat chondritic relative patterns when compared to regolith/fragmental breccias, which have relatively low PGE abundance with fractionated PGE/Ir signatures (Fig. 1). Any signs of PGE enrichment in the lunar crust are likely to be preserved in the lithologies from large impact melt basins.

We studied three compositionally distinct regolith breccias, Y981031 (anorthosite-bearing basaltic regolith breccia), Y983885 (feldspathic regolith breccia), and Y-86032 (feldspathic breccia with impact melts) that are likely sourced from three distinct regions on the Moon (Supplementary Fig. 4 and Supplementary Text S4). The PGE concentrations in the studied regolith breccias range from ~0.01 to 0.1 times CI chondrite values (Fig. 1 and Supplementary Table S3). These concentrations are ten to a thousand times higher than those observed in A-881757, mare basalts, and other pristine lunar lithologies (Fig. 1; refs. 13,14,42). Similar to the majority of previously studied impact melt breccias in Apollo samples17,21,50,51,52,53,54, the 187Os/188Os ratio and PGE/Ir (Ru/Ir, Pt/Ir, and Pd/Ir) of sample Y-86032 point toward an ordinary chondrite type impactor composition (Fig. 3 and Supplementary Figs. 5, 6). Only Os/Ir ratios show a slightly higher value when compared to ordinary chondrite, similar to the fractionated Os/Ir of ordinary chondrite partially melted bodies known as brachinites (Supplementary Figs. 5, 6). While the PGE/Ir ratios for Y-86032 fall within the known chondritic range, samples Y983885 and Y981031 do not consistently fall into any of the known chondrite or iron meteorite fields (Supplementary Figs. 5–9 and Supplementary Text S5). Therefore, we further explore the possibility of other processes to explain the observed PGE fractionation. The canonical exposition for the supra-chondritic 187Os/188Os and high Pt/Ir, Pd/Ir and Ru/Ir are that these variations could reflect: (i) a mixture of chondritic and differentiated impactors17,21, (ii) a signature of unsampled inner solar system bodies whose composition differ from the modern-day population of chondrites16,18,19,21,54, (iii) PGE fractionations (both small scale and/or large scale) in impact melt sheets or the lunar crust9,20,22.

Fig. 3: Variation in PGE/Ir ratios with respect to Pt/Ir for the studied rocks, terrestrial PGE deposits, previously analyzed impact breccia rocks, lunar anorthositic regolith breccia (ARB), and impact melt coats (IMC).
figure 3

a Pd/Ir versus Pt/Ir, b Ru/Ir versus Pt/Ir, and c Os/Ir versus Pt/Ir. Only Y981031 and some samples from ref. 22, show PGE fractionation similar to the terrestrial PGE deposits, indicating the distinct nature of these samples.

Full size image

Evidence suggesting that PGE fractionations were inherited from multiple impactors comes from a positive trend of 187Os/188Os versus Ru/Ir, Pt/Ir, and Pd/Ir17,19,20,21,55. However, our samples, particularly Y981031, deviate from these trends, showing distinct or opposite patterns (Supplementary Fig. 5). The ratios in these samples also do not align with the magmatic differentiation trends of IIA, IIIA, and IVA iron meteorites, suggesting that their PGE/Ir ratios do not correspond to any known iron impactor compositions (Supplementary Fig. 8 and Supplementary Text S5). It is also possible that the observed PGE/Ir reflects signatures of previously unknown impactors, distinct from modern chondrites and achondrites16. The varying PGE compositions in LIB and regolith breccia meteorites16,19,21,22,54,55 suggest early impactors with different PGE signatures than present-day chondrites. Sample Y983885, with an age of ~4.0 Ga56, could support this idea, but for the younger sample Y981031 (~3.5 Ga; ref. 57), the likelihood of PGE fractionation from an unknown impactor is low.

Another potential mechanism for fractionating the PGE in impact melt rocks is the gravitational removal of solid metal from an unfractionated metallic melt20. In-situ analysis of PGE-bearing phases in meteorites indicates that fractionation can occur at hand sample scales12,20,22. While impact breccias often retain PGE from the impactor, siderophile elements like S, P, Fe, and Ni may have both lunar and meteoritic origins. To investigate if the Y981031 PGE + Re composition reflects the melt or is influenced by large-scale fractionation, we modeled PGE + Re partitioning during Fe-Ni-S crystallization (Supplementary Text S6 and Supplementary Figs. 6, 9). The model, using a mix of impactor and host lithologies, shows that the modeled PGE ratios do not match the observed ratios in the samples (Supplementary Figs. 6, 9). Although PGE abundance increases with metal segregation in the model, the Pd/Ir ratio increases alongside other PGE/Ir ratios, leading to trends that differ from sample Y981031. Consequently, metal/sulfide segregation in the impact melt pool, with impactor compositions (chondritic, differentiated, or a mix), is unlikely to explain the elevated Pt/Ir and Pd/Ir ratios.

Fractionation of the PGE has been observed in pristine lunar lithologies (Fig. 1; refs. 13,14). Most mare basalts and highland rocks exhibit elevated PGE/Ir ratios, which can be attributed to fractional crystallization and the partitioning behavior of PGE in lunar magmas (Fig. 1). Day et al.9 demonstrated that mixing trends between lunar crustal compositions, such as FAN and magnesian suite rocks, and a CI chondrite composition suggest that lunar crustal materials could contribute to the compositions of impact breccias. However, to explain the relative mixing required for lunar impact melt rocks, prior scavenging of PGE in Fe-Ni metal and sulfides is necessary. This can occur through the combination of lunar sulfur (S) and phosphorus (P) to form metal-sulfide assemblages in some impact breccias under sulfide-saturated conditions, similar to those observed in terrestrial PGE deposits (Fig. 4a). The fractionated PGE signature observed in Y981031 could therefore reflect crustal signatures that were amplified by the enrichment process on the Moon (Fig. 3).

Fig. 4: A model showing PGE scavenging on the lunar surface, resulting from the terrestrial like PGE enrichment process, can explain the observed fractionation of PGE/Ir ratios in impact breccias.
figure 4

a Schematic showing a possible scenario in order where basin-forming impact on the lunar crust taps into shallow S-rich layers providing enough volume of melt and sulfur needed for sulfide saturation which further leads to scavenging of PGE by metal + sulfides. Then, later impacts on the already PGE-enriched lunar surface, and subsequent minor (<1%) metal segregation gives rise to the observed fractionation in impact breccias. The schematic also show that fractionated PGE signatures are possibly recorded in PGE-enriched sulfide phases which are likely incorporated within some lunar impact breccia rocks. b Modeling results showing that the terrestrial-style PGE-enriched lunar crust and its subsequent fractionation in an impact melt pool can produce the observed PGE/Ir ratio in Y981031. The model assumes fractionation in the Fe-Ni-S system, and <1% fractionation matches the observed composition (Methods; Supplementary Text S8 and Supplementary Table S7). PGE platinum group elements, IC impactor composition, CMB crystalline matrix breccias, FB feldspathic breccias, IM impact melt rocks, PB polymict breccias .

Full size image

Lunar impact melt sheets as PGE resources

PGE fractionation is observed in both impacts contaminated as well as lunar pristine lithologies (Fig. 1). The observed fractionation in the lunar pristine lithologies such as mare basalts and ferroan anorthosites are, however, difficult to transfer to lunar impact breccias. This is mostly because the impactor has high PGE content (~ng/g) and almost no PGE fractionation, overprinting the preexisting fractionated PGE signature (~pg/g) of the lunar crustal lithologies (Fig. 4a). Sample Y981031, should it indeed record signatures of PGE enrichment, must have its PGE composition explained by some combination of preexisting PGE-enriched lithology and a possible impactor. Therefore, it is essential to know the preimpact lithology and its PGE composition (Supplementary Text S7).

With observed similarities in composition and source location for A-881757 and Y981031 (Supplementary Fig. 4 and Supplementary Text S7), we assume A-881757 (YAMM basalts) is the best representative of the preimpact basaltic lithology for Y981031. The assumed preimpact PGE concentration of sample Y981031 (80% basaltic lithologies + 20% highland lithologies) can be taken as 0.80 ×YAMM-PGE + 0.20× lunar pristine crustal average PGE from Day et al.13. Using this preimpact lithology, we model terrestrial-style PGE enrichment scenario in the impact melt pool for sample Y981031, also accounting for subsequent metal segregation (Fig. 4, Supplementary Table S5, and Supplementary Text S8). With ~1–10% impactor contribution, the PGE concentration reaches values akin to terrestrial economic PGE-enriched deposits with reasonable sulfide to silicate ratios (1:100 to 1:1000; Fig. 2b). A <1% removal of metal successfully mimics the observed PGE/Ir ratios with elevated Pt/Ir and Pd/Ir in sample Y981031 (Fig. 4b).

During the time when large impactors strike the surface of a planetary body, it is understood that as much as ~99% of the impactor evaporates58. Such a high-energy environment creates large impact melt pools, which may facilitate mineralization and subsequent metal-sulfide segregation. The Y981031 composition is best matched to originating from either Mare Crisium (>700 km diameter; ref. 59) or Mare Fecunditatis (>840 km diameter; ref. 60). Using the scaling laws of ref. 34, the volume of impact melt generated from these basin formations would be ~1.36 × 106 km3 and 2.74 × 106 km3, respectively. With such a large transient cavity diameter, the Crisium or Fecunditatis forming impactors might have excavated up to ~70–80 km depth into the Moon, tapping the shallow sulfide-rich layer41. The large impact melt pool with sulfide-rich melt might have provided favorable conditions for PGE scavenging by sulfides. Later impacts on the already PGE-enriched lunar surface, and subsequent minor (<1%) metal segregation, gave rise to the observed fractionation in impact breccias. The fractionated Pd/Ir and Pt/Ir ratios observed in sample Y981031 are therefore, indigenous to the Moon and likely show PGE enrichment similar to that observed in terrestrial PGE settings. Based on settling, we expect any HSE-enriched deposits to be close to the edges of the impact melt sheet against the cavity walls, akin to massive sulfide side-wall deposits in terrestrial layered intrusions.

The meteorite Y981031 is among a few samples in the lunar sample suites that shows potential evidence for terrestrial-style PGE enrichment processes acting in the Moon (Figs. 3, 4). However, our sample size is minor compared to the extent and size of such deposits. The limited sampling may reflect the fact that PGE-enriched deposits are likely to be spatially and volumetrically limited, as they are on Earth, with random sampling as might occur from surface landing missions or meteorites being close to zero. Knowledge from studying terrestrial PGE deposits suggests that only a few among many such lunar deposits would be a potential economic PGE reserve.

The exploitation of extraterrestrial resources, especially the PGE, has been discussed for several decades (e.g., refs. 1,61,62,63,64), with commercial interest largely focused on the near-Earth asteroids. Recent numerical modeling discussing delivery and retention of PGE-rich asteroids has suggested that the Moon might preserve PGE more efficiently (potentially by several orders of magnitude) than near-Earth asteroids65. With NASA’s Moon to Mars program and multiple upcoming lunar missions prioritizing ISRU, systematic assessment of PGE enrichment on the Moon is becoming both timely and feasible. A major challenge in evaluating this potential is the limited understanding of the geological occurrence of PGE in lunar crustal settings. The distinct PGE fractionation signatures in lunar meteorite Y981031 shown here provide the first evidence directly linking impact melt processes to the potential formation of economical PGE concentrations on the Moon. Although no specific exploitable PGE deposits have yet been identified, our geochemical investigation suggests that impact-related lithologies will be the plausible reservoirs for enrichment. Future exploration strategies should therefore combine sample analyses with remote sensing approaches sensitive to PGE-bearing minerals, such as sulfides. Prospected instruments operating in the far-infrared (FIR) spectrum, such as MIRORES66, offer promising avenues for reconnaissance.

Relatively sulfide-rich high-Ti mare basalts regions in the Procellarum KREEP Terrane might appear the most viable targets for reconnaissance. Our calculation for sulfide saturation however, indicates that anorthositic targets, which may have covered much of the Early Moon, were more likely to reach sulfide saturation than the mare basalts (Fig. 2). Exposed impact melt facies within basins such as Imbrium, Crisium, and Orientale are especially promising, though their accessibility is often complicated by subsequent mare basalt infilling, thick regolith deposits, and overlying ejecta blankets. Our results for lunar meteorite Y981031 point to Mare Crisium and Mare Fecunditatis as the most likely source regions (Supplementary Fig. 4). While the impact melt facies of Mare Fecunditatis remain largely unexplored, Yerkes crater67 in Mare Crisium offers a potential window into Crisium’s impact melt facies. Despite the Orientale basin being less ideal for a lunar base due to limited sunlight (lunar libration), communication challenges, and rugged terrain, it emerges as an exceptionally favorable target for robotic missions. This is primarily because it has relatively minimal mare infilling compared to other nearside basins. Local impact melt exposures in regions such as the Maunder crater68 represent important locales where both geological processes of PGE enrichment and their potential resource relevance can be directly evaluated.

Methods

Sample petrography and determination of metal compositions

Polished thick sections of A-881757(88a & 88b), Y981031(,100), Y983885(,74) and Y-86032(,137) and bulk rock fragments of A-881757(,89), Y981031(,90), Y983885(,72), and Y-86032(,136) were provided by NIPR. The polished thick sections were mapped using BSE imaging and later used to determine the association and composition of metals and sulfides grains. Metal and associated sulfide grains were analyzed for major and minor element compositions using JEOL JXA 8530F field emission electron probe micro analyzers (FE-EPMA) housed at both the National Institute of Polar Research, Japan and the Physical Research Laboratory, Ahmedabad, India. The analytical methods used at both locations were identical with identical reproducibility. Polished sections were coated with either carbon or gold prior to analysis, and water usage was minimized to avoid oxidation of metals. Analyses were performed with wavelength dispersive spectrometers, using an accelerating voltage of 15 kV, beam current of 30 nA, beam diameter <1 µm for native metals and applying ZAF correction. The point selection was carefully made by avoiding beam overlap between metal grains and surrounding silicates and oxides. An overlap correction for emission lines of Co (Kα)-Fe (Kβ) and Cu (Kα)-Ni (Kβ) was also applied. The standardization was performed using standard pure metals (Fe, Ni, Co, Cu, Cr, and Si), olivine (Mg), sphalerite (Zn and S), and P on Fe3P and only totals exceeding 97.5 wt% were considered acceptable. The peak count and background count times for all the elements were kept at 30 and 15 s, respectively, except for Si where it was adjusted to 60 and 30 s, respectively. The results are presented in Supplementary Table S1.

Bulk rock major, minor, and trace-element abundance determinations

Homogeneous powders of the bulk samples were prepared using a dedicated agate mortar and pestle that was thoroughly cleaned between uses. The bulk chemical composition was obtained on these powdered rock fragments provided by the NIPR. A Thermo Scientific iCAP-Q inductively coupled plasma mass spectrometer (ICP-MS) housed at the Scripps Isotope Geochemistry Laboratory (SIGL), Scripps Institution of Oceanography, was used for the bulk major and trace element analysis. A ~50 mg aliquot of homogenized sample powder was digested in Teflon-distilled concentrated HF (4 mL) and HNO3 (1 mL) for >72 h on a hotplate at 140 °C, along with total procedural blanks and terrestrial basalt and andesite standards (AGV-2, BHVO-2). Samples were sequentially dried and taken up in concentrated HNO3 to destroy fluorides and then diluted to a factor of 50,000 for major-element determination and 5000 for trace-element determination, followed by doping with 1 ppb indium to monitor instrumental drift during analysis. Major-element abundances were obtained in low-resolution mode. For major elements, Si was derived by difference, with reproducibility of other elements measured on the BHVO-2 reference material being better than 3%, except Na2O (6.9%). The results are presented in the Supplementary Table S2.

PGE + Re abundance and Re-Os isotope determinations

Osmium isotope and PGE + Re abundance measurements were performed at the Scripps Isotope Geochemistry Laboratory (SIGL). The homogenized powder was used to measure the bulk PGE + Re abundance and 187Os/188Os of the samples. Homogenized sample powders and associated total procedural blanks were prepared with appropriate quantities of isotopically enriched multi-elemental spike (99Ru, 106Pd, 185Re, 190Os, 191Ir, and 194Pt) in single-use pre-cleaned borosilicate Carius tubes (see ref. 14 for details). Samples were digested by adding a 1:2 mixture of Teflon-distilled HCl and Teflon-distilled purged HNO3 in the Carius tube. After sealing, the Carius tubes were placed in a convective oven for >72 h at a temperature of 250 ± 1 °C. Osmium was triply extracted using CCl4 and then back-extracted into quadruple Teflon-distilled HBr69. Double micro-distillation was performed to further purify the extracted osmium70. The remaining PGE + Re were purified from the residual solutions using anion exchange separation techniques9.

Osmium isotope measurements were done using a Triton TIMS in negative mode, and the other PGE + Re abundances were determined using a Cetac Aridus II desolvating nebulizer coupled to an iCAPQc ICP-MS. Precision on the 187Os/188Os measurement was determined using a 35 pg UMCP Johnson-Matthey standard load and was better than ±0.2% (2 SD; 0.11368 ± 27; n = 9). Osmium measurements were also corrected for oxide interferences, instrumental mass fractionation (assuming the exponential law) using 192Os/188Os = 3.0821, spike addition, and the TAB (total analytical blank). The PGE + Re abundance data were corrected offline for mass fractionation using multiple measurements of a standard solution throughout the run. The external reproducibility on the PGE + Re analyses was better than 0.25% on 0.5 ppb solutions. Total procedural blanks analyzed run with the samples in two analytical campaigns and from the same batch of Carius tubes had relatively consistent 187Os/188Os = 0.140 ± 0.013, and (in pg) Re = 4.9 ± 2.0, Pd = 20.5 ± 5.2, Pt = 3.9 ± 2.0, Ru = 70 ± 10, Ir = 2.5 ± 1.0, and Os = 1.7 ± 0.8 (n = 3). The uncertainties on concentration measurements were calculated by propagating from external reproducibility on standard measurements, together with associated blank percentages. The results are presented in Supplementary Table S3.

Numerical modeling of factors influencing PGE enrichment in lunar impact scenarios

Sulfide saturation for various lunar lithologies

Utilizing the O’Neill39 parameterization, we estimated the sulfide saturation potential across a combination of lunar lithologies, including representative lunar mare basalts (both high-Ti and low-Ti) and highland rocks (FAN), under low-pressure conditions. This parameterization, based on the only available low-pressure experimental data, has been previously applied to lunar basalts71. We modeled various combinations (50/50 and 80/20 mixtures) of average representative low-Ti basalts72, high-Ti basalts (Lunar Sourcebook), representative FAN (60025; Lunar Compendium), and 0–10% CI73 to reflect the variability in target lithologies. To validate our calculations, we also determined the sulfur content at sulfide saturation (SCSS) for martian shergottites, which are well-documented as sulfide undersaturated under low-pressure conditions74. The results are presented in Fig. 2a and Supplementary Table S4.

R-factor enrichment and PGE resource potential

In terrestrial settings, significant enrichment of the PGE + Re is possible by formation of metal-sulfide assemblages in impact melt rocks, such as at Sudbury, Ontario29. Under these conditions, an “R-factor” type process occurs where the high partition coefficients of PGE + Re into sulfide versus silicate, combined with the high mass ratio of silicate melt to sulfide melt available, lead to significant enrichment from sources with otherwise low PGE + Re abundances57. The basin-size impact melting, provided with high volume of sulfide-saturated impact melt can result in R-factor type (terrestrial-style PGE) enrichment.

R-factor type enrichment is governed by the equation:

$${({{\mathrm{PGE}}}+{{\mathrm{Re}}})}_{{{\mathrm{sulfide}}}\,{{\mathrm{melt}}}}=\frac{{({{\mathrm{PGE}}}+{{\mathrm{Re}}})}_{{{\mathrm{silicate\,melt}}}}\times {D}_{{{\mathrm{silicate}}}}^{{{\mathrm{Sulfide}}}}\times (R+1)}{R+\,{D}_{{{\mathrm{silicate}}}}^{{{\mathrm{Sulfide}}}}}$$
(1)

where R is the ratio of silicate to sulfide liquid and \({D}_{{{\mathrm{silicate}}}}^{{{\mathrm{Sulfide}}}}\) is the partition coefficient of sulfide and silicate melt (taken from ref. 28). The values used in the model are provided in Supplementary Tables S6, S7.

We perform calculations by logarithmically varying R values from 1 to 100,000, spanning the range from near-equilibrium sulfide segregation to highly silicate-dominated systems. Initial PGE concentrations for each scenario were derived from four mixing models of average lunar crust13 with CI chondrite impactor contributions of 0%, 0.1, 1, and 10% (Supplementary Tables S6, S7). For each scenario and element, concentrations were calculated across the full R range to determine enrichment trajectories (Fig. 2b). All calculations were carried out in Python using vectorized routines in numpy, and results are presented in Fig. 2b.