Introduction

Large-scale impact events played a vital role in the formation and evolution of the Moon, evidenced by the dense distribution of impact basins and craters on its surface. The transient high pressures and temperatures produced during an impact induce various shock effects on the minerals, such as melting, vaporization, mechanical deformation, and high-pressure phase transformation1,2. However, compared with highly shocked ordinary chondrites and Martian meteorites, lunar samples appear to contain fewer high-pressure phases, limiting our understanding of the large-scale shock effects on the Moon3,4. Reported high-pressure phases include high-pressure polymorphs of olivine (ringwoodite and wadsleyite)5,6, silica (coesite, stishovite, and seifertite)7,8,9,10,11,12, zircon (reidite)4, plagioclase (tissintite)13,14, and Ca-Al-silicate (donwilhelmsite)15. Most of these high-pressure phases are primarily observed as isolated occurrences within or near shock-induced melt veins.

The Moon’s surface is covered by a layer of fine-grained regolith. The regolith is constantly altered in both physical and compositional properties by space weathering processes—driven predominantly by micrometeoroid impacts and solar wind irradiation16,17,18,19,20,21. Notably, nanophase metallic iron (npFe0) particles, an important product of space weathering, could be responsible for the spectral properties of lunar regolith16,22,23. However, other types of space environmental effects may also play a role, including the crystallization of lunar basalt in low oxygen fugacity conditions24, the addition of meteoritic metals (Fe or FeNi)25, or larger-scale meteoroid impacts21, and the contribution of these processes is still being investigated.

Northwest Africa (NWA) 3170 is a lunar breccia containing clasts of various lithologies, including olivine gabbro, ferroan gabbro, anorthositic gabbro, and polymict breccia26,27,28. It is inferred to be paired with the NWA 773 clan, which is KREEP-like and shows distinct geochemical characteristics from the samples collected by sample return missions29. In-situ dating of baddeleyite, zircon, and apatite has primarily yielded Pb-Pb (or U-Pb) ages of approximately 3.1 Ga26, indicating that these rocks represent relatively young lunar igneous rocks. The igneous material was mixed with breccia components before being ejected from the Moon30. In this study, a detailed mineralogical investigation of NWA 3170 was conducted using a series of advanced analytical techniques, including scanning and transmission electron microscopy and Raman spectroscopy. Our results offer unequivocal evidence of abundant high-pressure polymorphs and submicroscopic metallic iron in NWA 3170. They reveal the strongest shock metamorphism in lunar samples and provide unique insights into the shock processes and effects associated with large-scale impact events.

Results

The NWA 3170 meteorite investigated in this study comprises olivine gabbro and polymict breccia lithologies (Supplementary Fig. S1). The olivine gabbro is primarily composed of olivine, pigeonite, augite, and anorthite (Supplementary Table S1). Some minerals show various shock features such as dislocations, twinning, and planar microcracks. Accessory minerals, including fayalite, silica, spinel, and phosphates, are mainly concentrated in the polymict breccia. The presence of shock melt veins in the breccia and at its boundary with the olivine gabbro indicate intense shock metamorphism (Supplementary Fig. S1). The widths of the shock melt veins are approximately 2–120 μm, and their lengths vary significantly, with the longest exceeding 10 mm (Supplementary Fig. S1). Shock-induced glasses, including maskelynite and pyroxene glass, were identified based on their Raman spectra (Supplementary Fig. S2). Additionally, we also identified abundant occurrences of high-pressure polymorphs in or at the rim of shock melt veins.

Stishovite was identified near the melt veins and pockets, as evidenced by its characteristic Raman peaks at ~232 and 756 cm−1 (Supplementary Fig. S3)3. It exhibits subhedral textures with grain sizes of up to 2 μm (Supplementary Fig. S3a). Some pyroxene fragments are enclosed in the adjacent silica glass (Supplementary Fig. S3b). Additionally, some silica grains display a tweed-like texture, indicative of transformation into seifertite (α-PbO2-type SiO2), which features diagnostic Raman peaks at 378 and 511 cm−1 (Supplementary Fig. S3e)31. The presence of shock melt residue was observed in the seifertite grain, as shown in Supplementary Fig. S3c.

Ringwoodite and ahrensite, the Mg- and Fe-end-members of γ-(Mg, Fe)2SiO4, were identified within and near the melt veins (Fig. 1a), with a strong characteristic Raman peak around 842 cm−1 (Fig. 1c)32,33. Olivine with an Fa content of 28 (100×Fe/(Fe+Mg), atomic ratio) in contact with the melt vein was completely transformed into ringwoodite, exhibiting a significantly different contrast (Fig. 1a). The identified ahrensite grains had higher Fa values (Fa > 97, composition of Ahr-1 and Ahr-2 presented in Supplementary Table S2) and featured three occurrences: (1) completely transformed into ahrensite adjacent to the melt vein (Fig. 1b); (2) as ahrensite lamella in the fayalite (Fig. 1d); and (3) as sub-µm granular ahrensite clusters in the melt-vein matrix (Fig. 1e). The Raman peak at 668 cm−1 (R216 and R574 in Fig. 1c) may correspond to the bridgmanite that was vitrified under decompression, similar to those observed in some Martian and chondrite meteorites34,35. Additionally, in the first occurrence, large amounts of bright nanoparticles coexisting with polycrystalline ahrensite were observed in the magnified scanning electron microscopy (SEM) image (Fig. 2a). Subsequently, we prepared an ultrathin foil (Foil 1) for further structural analysis. The HAADF image and its corresponding EDS maps revealed that these nanoparticles were rich in Fe but poor in O (Fig. 2c–e). Quantitative TEM-EDS analysis of one nanoparticle showed an Fe content of 74 wt% with small amounts of Si and O, which may be influenced by the surrounding ahrensite (composition of Foil 1 is presented in Supplementary Table S3). The HRTEM image and corresponding fast Fourier transform (FFT) analyses confirmed that these nanoparticles were metallic iron (Fig. 2g).

Fig. 1: Occurrences of ringwoodite, ahrensite and tuite.
Fig. 1: Occurrences of ringwoodite, ahrensite and tuite.
Full size image

a BSE image of ringwoodite in contact with the longest shock-induced melt vein. b BSE image of ahrensite and tuite near the melt vein. c Raman spectra of various minerals. d Lamellar ahrensite near the melt vein. e BSE image of ahrensite + vitrified bridgmanite in the melt vein. Rwt ringwoodite, Ol olivine, Ahr ahrensite, Fa fayalite, Msk maskelynite, Brg G bridgmanite glass. Yellow markers (e.g., E35) indicate the positions of EPMA measurements. White lines indicate the boundaries of the melt veins.

Fig. 2: Microstructures of polycrystalline ahrensite and submicroscopic metallic iron.
Fig. 2: Microstructures of polycrystalline ahrensite and submicroscopic metallic iron.
Full size image

a Enlarged BSE image of ahrensite near the melt vein. The blue line indicates the location from which Foil 1 was extracted. Abundant submicroscopic metallic iron can be observed at the transition between ahrensite and fayalite. b HAADF image showing ahrensite, defect-rich fayalite, and the transition zone between fayalite and ahrensite. c HAADF image of the transition zone showing metallic iron coexisting with ahrensite. d, e HAADF and corresponding EDS maps showing that the nanoparticles are Fe-rich and O-poor. f HRTEM image of ahrensite and its corresponding FFT pattern. g HRTEM image of metallic iron and its corresponding FFT pattern. Ahr ahrensite, Fa fayalite. White lines indicate the boundaries of the melt veins.

Tuite, a high-pressure polymorph of Ca-phosphate with the structure of γ-Ca3(PO4)2, was discovered near the shock melt vein (Fig. 1b), exhibiting a strong Raman peak at 973 cm−1 (R214 in Fig. 1c)36. Additionally, xieite, a high-pressure polymorph of chromite (CaTi2O4-type structure), was identified in direct contact with the melt vein (Supplementary Fig. S4). Xieite shows significantly different Raman spectral characteristics from chromite (Supplementary Fig. S4b). The corresponding EBSD pattern matched well with the xieite structure (an orthorhombic phase characterized by cell parameters of a = 2.88 Å, b = 9.52 Å, c = 9.75 Å)37, with a mean angular deviation (MAD) of 0.77 (Supplementary Fig. S4c). Xieite grains exhibit a preferred orientation (Supplementary Fig. S4e) and have a similar composition to the adjacent chromite (Supplementary Table S2). Some lamellae observed slightly away from the shock vein were likely chenmingite that were formed at lower temperatures (Supplementary Fig. S4a)38.

Exsolution lamellae comprising high- and low-Ca components are prevalent in pyroxenes. However, some low-Ca lamellae near the melt vein appeared brighter (Fig. 3a), displaying a strong Raman peak at 916 cm−1 (R284 in Fig. 3e) that was clearly distinct from those peaks at 671 and 1002 cm−1 for the adjacent pyroxene (R286 in Fig. 3e). The Raman peak at 916 cm−1 can be attributed to a garnet-like mineral39, whereas that at 667 cm−1 corresponds to bridgmanite glass34,35. The foil (Foil 2) was extracted for TEM analysis. Both the HRTEM image and SAED pattern were consistent with a low-Ca majorite (garnet) structure (Fig. 3d). The presence of weak {h0l} reflections (where h, l = odd) suggested that the space group featured tetragonal symmetry (I41/a) rather than cubic symmetry (\({\rm{Ia}}\bar{3}{\rm{d}}\))35,40. However, high-Ca lamellae still exhibited a monoclinic pyroxene structure, as demonstrated from the identical SAED patterns at different areas (Fig. 3f).

Fig. 3: Microstructure of low-Ca majorite.
Fig. 3: Microstructure of low-Ca majorite.
Full size image

a BSE image of majorite and pyroxene near the melt vein. The white circles indicate the location of Raman analysis. Yellow marker E6 indicates the position of EPMA measurement. The blue line represents the location from which Foil 2 was extracted. b TEM image of Foil 2 showing the lamellar texture of majorite. The white dashed lines indicate the boundary between pyroxene and majorite. c STEM-EDS map showing the low-Ca and high-Ca pyroxene lamella. d HRTEM image and SAED pattern confirming the transformation of low-Ca pyroxene to majorite (garnet). e Raman spectra of pyroxene and majorite. f TEM image of lamellar high-Ca pyroxene and the SAED patterns of different regions. Maj majorite, Px pyroxene.

Majorite-pyrope solid solution, a major rock-forming mineral in the upper mantle and transition zone of the Earth41, was observed as granular crystals within the melt vein, with diameters ranging from a few hundred nanometers to several micrometers (Fig. 4a). They had a characteristic Raman peak at around 918 cm−1 (R571 in Fig. 4b), and their structure can also be confirmed by SAED patterns along the [110] and [111] zone axes (Fig. 4f, h). Some grains exhibited distinct compositional zoning in the BSE (Fig. 4a) and HAADF images of extracted ultra-foil (Supplementary Fig. S5) owing to the fast cooling rate35,42. Additionally, fine-grained Fe-rich particles were observed both within majorite-pyrope (Fig. 4d–e) and the matrix (Fig. 4h–j). These particles comprised subgrains, including metallic iron (α-Fe) and iron-sulfide (Fig. 4e, and Supplementary Figs. S5 and S6). Semi-quantitative TEM-EDS analysis of these nanoparticles showed an iron concentration of up to 85 wt% and a small amount of adjacent material (compositions of Foil 3 and Foil 4 are presented in Supplementary Table S3). The metallic iron and iron-sulfide assemblage in Foil-3 extracted from the center of the melt vein contains approximately 2 wt% Ni, whereas the metallic iron in Foil-4 extracted near the edge of the melt vein contains no detectable Ni. The structure of metallic iron can be confirmed in high-resolution TEM images (Fig. 4g) and CBED patterns (Fig. 4j), with measured d-spacings of approximately 2.05 and 1.18 Å (insets of Fig. 4g, j), which are consistent with the (110) and (\(\bar{1}21\)) planes of the metallic iron (α-Fe) phase, respectively.

Fig. 4: Microstructure of majorite-pyrope.
Fig. 4: Microstructure of majorite-pyrope.
Full size image

a BSE image of majorite-pyrope (Maj-pyr) in the shock melt vein. The blue lines represent the locations where Foil 3 and Foil 4 were extracted. b Raman spectrum of majorite-pyrope (R571). c HAADF image showing the microstructure of Foil 3. d HAADF image of majorite-pyrope showing some bright nanoparticles within it. e The nanoparticles are composed of subgrains, including metallic iron and sulfide. f HRTEM image of majorite-pyrope and its SAED pattern along the [110] zone axis. g HRTEM image of metallic iron and sulfide. The inset is the FFT pattern of metallic iron. h HAADF image showing the microstructure of Foil 4. The inset shows the SAED pattern of majorite-pyrope along the [111] zone axis. i EDS map showing the metallic iron nanoparticles and (Fe, Ti)-oxides in the shock melt vein. j TEM image of metallic iron and majorite-pyrope. The inset shows the CBED pattern of metallic iron along the [\(1\bar{1}3\)] zone axis.

Discussion and implications

High-pressure phase transformation

Various high-pressure minerals were observed in the shock veins of NWA 3170, including stishovite, seifertite, ringwoodite, ahrensite, tuite, xieite, majorite, majorite-pyrope, and vitrified bridgmanite, to the best of our knowledge, six of which are reported for the first time in lunar samples (Table 1). This suggests that substantial amounts of high-pressure minerals can also be produced and well-preserved in lunar samples.

Table 1 High-pressure phases reported in lunar samples
Full size table

It has been extensively demonstrated that high-pressure minerals form either by solid-state transformation of low-pressure polymorphs or crystallization from shock melts3,43,44. In this study, some of these high-pressure phases, such as ringwoodite and xieite, are definitively formed through a solid-solid phase transformation, as inferred from their occurrences and similar composition to the host minerals (Supplementary Table S2). Tuite has a relatively high FeO content, and no apatite or Cl- or F-bearing minerals were found nearby. Therefore, it is very likely formed through the solid-state phase transformation of merrillite rather than by the decomposition of apatite31,45,46. Majorities formed by either solid-state transformation or melt crystallization have been documented in many ordinary chondrites and Martian meteorites3,47. However, in this study, we presumed that the low-Ca majorite identified here should have been formed by the solid-state transformation of low-Ca pyroxene, as evidenced by the parallel exsolution lamellar texture (Fig. 3b). Such occurrence of majorite has never been reported in previous studies.

Ahrensite, with both polycrystalline and lamellar textures, was observed in the host fayalite grains (Fig. 1b–d), indicating a solid-solid phase transition48. However, ahrensite clusters in the melt vein (Fig. 1e) likely crystallized from the shock melt49. Similarly, the majorite-pyrope should have also crystallized from the shock melt, as evidenced by its granular morphology and internal sulfides (Fig. 4). Moreover, we observed some mineral fragments in the silica glass beside stishovite and melt residues in the seifertite grain (Supplementary Fig. S3c), indicating that some stishovite and seifertite crystallized from the shock melt. However, the formation mechanism of most high-pressure phases of silica is difficult to determine. Apart from the high-pressure phases, we only observed some silica glass. Considering that most silica grains in lunar meteorites and Apollo samples are cristobalite, we cannot rule out the possibility that some high-pressure phases of silica may be formed by solid-state transitions, as reported in other lunar samples8,9.

Shock condition constraints

The unique assemblage of high-pressure phases and shock features in NWA 3170 can be used to estimate the pressure and temperature (P–T) conditions of its shock metamorphism (Supplementary Fig. S7). Previous studies have suggested that pyroxene glasses form under shock pressures of 26–35 GPa49. The plagioclase in the olivine gabbro has an anorthite content (An = Ca/[Ca+ Na]) of approximately 92, and the corresponding transition pressure of plagioclase to maskelynite is ~25 GPa50. Under the high-temperature conditions of the shock melt vein, the transition pressure should be lower than ~25 GPa. The presence of xieite near the melt vein requires the shock pressure and temperature to be >18 GPa and >1400 °C, respectively37,51, whereas studies on α-cristobalite indicate that seifertite appears at pressure as low as ∼11 GPa under nonhydrostatic conditions9,52. Moreover, based on the majorite barometer53, the required pressure for majorite is estimated to be 21.87 \(\pm\) 0.55 GPa. Bridgmanite forms at pressures of 23–25 GPa49,54, whereas the majorite-pyrope garnet and ringwoodite assemblage in the Allende meteorite requires approximately 20–24 GPa and ~2000 °C (Supplementary Fig. S7)55. Therefore, we infer that NWA 3170 experienced shock pressures of 22–24 GPa and a temperature of approximately 2000 °C. The heterogeneity of high-pressure mineral assemblage indicates that the P-T conditions of the experienced shock metamorphism were heterogeneously distributed. This is consistent with the results of other studies on high-pressure phases in meteorites31,56,57,58 and the quasichaotic nature of shock propagation in heterogeneous and/or porous materials59.

The crystallization assemblages in the shock melt vein depend on the quench time and duration of the high shock pressure59. The quench time for the shock vein, namely, the lag time between the crystallization of the edge and center of the vein, is estimated to be approximately 15 ms; this is based on the thermal conduction equation, namely, t = L2/α, where L is the width of the shock vein (calculated using the maximum width of melt vein of ~120 μm) and α is the thermal diffusion coefficient that is set to 10−6 m2 s−1 60. The rapid cooling rate is likely attributed to the presence of the shock vein being primarily at the interface between the breccia and the olivine gabbro, where the olivine gabbro acts as a heat sink. This is consistent with the results of numerical simulation of impact-induced compaction in the poorly consolidated mix of chondrules and high-porosity matrix; that is, the matrix cools rapidly after the shock owing to the proximity to non-porous “cold” chondrules61. If the quench time is considerably longer than the duration of shock pressure, crystallization will occur during pressure release, and a difference in mineralogy will be observed between the vein edge and the vein center59,62. As the entire shock vein contains high-pressure minerals that crystallized from the melt and shows no apparent back transformation to low-pressure polymorphs, the duration of shock pressure was at least 15 ms.

Formation mechanism of metallic iron

In this study, two distinct occurrences of metallic iron were identified. The first comprised tiny metallic iron nanoparticles embedded within ahrensite (Fig. 2). The disproportionation reaction (3Fe2+ = 2Fe3+ + Fe0) could theoretically occur under extremely high temperature and pressure (i.e., approximately 20 GPa)63,64. However, this hypothesis faces two inconsistencies. First, ahrensite decomposes into wuestite and stishovite at pressures above 17 GPa65. Second, the metallic iron nanoparticles show preferential distribution in the ahrensite-fayalite transition zone, slightly away from the melt vein (Fig. 2a, b). This spatial distribution reveals that the formation pressure and temperature conditions are close to the threshold of high-pressure phase transition, but lower than those required for the disproportionation reaction. Instead, the observed Raman peak of bridgmanite glass (R216 in Fig. 1c) within this region supports an alternative decomposition pathway: fayalite/ahrensite decomposed into bridgmanite + metallic iron. The products were different from those in the previously reported dissociated olivine (vitrified bridgmanite + magnesiowüstite) in Martian meteorites and chondrites33,66,67. Temperature and oxygen fugacity may be important controlling factors for this discrepancy, but further investigation is required.

The second type of metallic iron is present in the shock vein and sometimes coexists with sulfides (Fig. 4). This is similar to the occurrence in the impact melt rocks and can be attributed to the liquid immiscibility between the sulfides and silicate melt during impact68,69. The crystallized product is metallic iron, rather than magnesiowüstite, typically found in chondritic meteorites43,70, probably because of the high vacuum and low oxygen fugacity on the lunar surface.

Both types of submicroscopic metallic iron are produced directly during the shock metamorphism, and are significantly different from those formed by space weathering (e.g., solar wind irradiation and micrometeoroid impacts) on the surface of regolith grains17,21,71. Metallic iron formed in this way may be ubiquitous on the lunar surface owing to the massive, large-scale impact events on the Moon. In particular, the early intense asteroid bombardment on the Moon may have generated substantial amounts of metallic iron. These metallic iron nanoparticles could aggregate and grow during subsequent multi-scale impact events, thereby considerably altering the spectral characteristics and magnetic properties of lunar samples.

Shock history of NWA 3170 and implications for high-pressure phases on the Moon

The unique high-pressure phase assemblage in NWA 3170 suggests that it experienced distinct impact histories compared with other lunar samples. The continuous shock vein in contact with olivine gabbro cuts across the gabbroic clast and breccia matrix, and therefore indicates that the shock vein formed after brecciation, namely, the latest impact event when the meteorite was ejected from the Moon. This inference is consistent with the chronological data of the NWA 773 clan meteorites to which NWA 3170 belongs. The Ar-Ar age of NWA 773 is ~2.9 Ga72, which is significantly younger than the Pb-Pb and U-Pb ages (~3.1 Ga) of baddeleyite and zircon26,73, suggesting that they were excavated some time after ~2.9 Ga (Fig. 5)74. After this collision, the ejecta of anorthositic crust, mare basalt, and gabbroic intrusive bodies were consolidated and brecciated to form the precursor lithology of NWA 317075. According to the total cosmic-ray exposure age, this meteorite was subsequently ejected from the Moon at <160 Ma72,76. Our finding suggests that abundant high-pressure minerals were formed during this shock event.

Fig. 5: Schematics of the impact history of meteorite NWA 3170.
Fig. 5: Schematics of the impact history of meteorite NWA 3170.
Full size image

a The sample was brecciated after the first impact. b The meteorite was ejected from the Moon after a large second impact, after ref. 75.

We can estimate the approximate size of the second asteroid impactor on the Moon using the shock pressure (~24 GPa) and duration (~15 ms) based on the Rankine-Hugoniot relationship8,74,77. The size was estimated to be 40 m, which would have formed a crater with a diameter of 274–607 m (calculation details are provided in Supplementary Text S1). The estimated crater size is consistent with the calculated result of 400–800 m, derived from the shock conditions of a paired meteorite NWA 272774. By comparing the geochemical composition and geochronological data with the remote-sensing datasets, the source region of the NWA 773 clan is most likely a fresh crater (~300 m in diameter) on the edge of the Le Verrier D crater in Mare Imbrium, within the Procellarum KREEP Terrane74.

There are two possible reasons for the scarcity of high-pressure minerals in lunar meteorites. On the one hand, most lunar meteorites are launched at velocities below the escape velocity of the Earth-moon system (2.78 km s−1)78,79. When lunar ejecta are launched at higher velocities, they tend to enter independent heliocentric orbits, which substantially reduces the probability of reaching Earth78. This suggests that lunar meteorites generally experience less intense shock metamorphism compared with Martian meteorites. On the other hand, the lunar surface comprises a layer of unconsolidated regolith covering the underlying bedrock. Shock recovery experiments reveal that temperature increases are greater in a porous materials than non-porous ones59. For instance, an impact on a highly porous lunar regolith with an average peak pressure of ~10 GPa is capable of achieving a shock temperature of 1473 K4. However, high temperatures are not conducive to the preservation of most high-pressure minerals, such as causing the back transformation of wadsleyite to olivine80. This could also be a key reason for the scarcity of high-pressure minerals in lunar samples. Hence, an impact with appropriate energy combined with a unique cooling history is are critical factor for the formation and preservation of high-pressure minerals in lunar samples. In addition, it has been noted that lunar meteorites generally show a higher level of shock than returned samples81. One explanation for this phenomenon is location bias, wherein lunar meteorites represent a random global sampling, whereas returned samples originate from limited regions on the lunar surface81. Alternatively, it could be attributed to differences in sample collection, as lunar meteorites were ejected from the Moon by impacts rather than being collected directly. We emphasize the need for further studies of high-pressure polymorphs in lunar meteorites (e.g., NWA 773 clan) using advanced microanalysis techniques. This can reveal new shock effects on the Moon and enhance our understanding of the behavior of materials under extreme conditions.

Materials and methods

SEM

Petrographic observations of NWA 3170 were conducted via backscattered electron (BSE) imaging, using the Thermo Fisher Apreo S and Nova 450 field emission scanning electron microscopes (SEMs) equipped with energy dispersive X-ray spectroscopy (EDS) detectors at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS). A large area BSE image was acquired using the MAPS software. The instruments were operated at accelerating voltages of 10–20 kV with beam currents of 1.6–6.4 nA. Electron back-scattered diffraction (EBSD) data were collected with an Oxford Symmetry EBSD detector and processed with Aztec Crystal to determine the mineral phases.

EPMA

Quantitative chemical compositions of the minerals were determined using a JEOL 8100 electron probe microanalyzer (EPMA) with wavelength dispersive spectrometers (WDS) at IGGCAS. It was operated at an accelerating voltage of 15 kV, a probe current of 20 nA, and spot size of approximately 1–2 μm spot size. The following natural and synthetic mineral standards were used: albite for Na, Si, and Al; diopside for Mg and Ca; fluorite for F; apatite for P; tugtupite for Cl; hematite for Fe; bustamite for Mn; sanidine for K; rutile for Ti; and Cr2O3 for Cr. All data were processed using the ZAF correction procedure.

Raman

Raman spectra of the minerals were acquired using a WITec alpha 300 R confocal Raman system. An optical microscope with 100× objective lens (Numerical Aperture: 0.9) was used to focus the 532-nm excitation laser beam onto the sample. A laser power of approximately 2 mW was used to identify the high-pressure minerals.

FIB and TEM

The ultrathin foils (~100 nm thick) were prepared using a Zeiss Auriga Compact focused ion beam (FIB) microscope equipped with an Omniprobe AutoProbe 200 micromanipulator. Ion beam milling and final polishing were performed at 5–30 kV and various beam currents (20 pA to 4 nA). Prior to FIB cutting, Pt was deposited to protect the sample surface from FIB Ga+ damage. The microstructures of the foils were examined via transmission electron microscopy (TEM) using JEOL JEM-2100 and Thermo Fisher Talos 200X instruments. The crystallographic structures were analyzed through selected area electron diffraction (SAED), convergent-beam electron diffraction (CBED), high-resolution TEM (HRTEM) imaging, and high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM). EDS point analyses and STEM-EDS mapping were used to determine the chemical compositions of the nanoscale phases. Semi-quantitative composition analysis was conducted using the Oxford Aztec and Thermo Scientific Velox software.

Estimation of impactor size and crater diameter

The size of the impactor and crater can be estimated from the parameters of launch impact based on the Rankine–Hugoniot’s relations77,82. Detailed procedures are provided in Supplementary Text S1.