A Noachian hydrosphere component in the 4.1-Ga Martian meteorite Teghaza 001 Authors: Lee Saper, Yang Liu, Yunbin Guan, Chi Ma, Elizabeth Ann Bell, Carl B. Agee Published: 2026 ABSTRACT The recent discovery of a 4.1-billion-year-old (Ga) Martian gabbroic diorite enables an assessment of water reservoirs on early Mars. Measurements of H2O and D/H in igneous Ca-phosphates record mixing between D-poor magmatic water and a D-rich component that retains an imprint of the ancient Martian hydrosphere. The D/H ratio of magmatic water is similar in Martian magmas with different ages and mantle sources, indicating that early differentiation processes did not fractionate H isotopes in the primordial Martian mantle. The D-rich component was assimilated by migrating magmas that interacted with aqueously altered basaltic crust. Our minimum bound on the D/H ratio of the Martian hydrosphere at 4.1 Ga (~2× the D/H ratio of Earth seawater) supports models of rapid H loss to space from Mars’ juvenile atmosphere. INTRODUCTION During the Noachian period [ ~3.7 to 4.1 billion years ago (Ga)], Mars was characterized by a semiarid climate that transiently supported liquid water at the surface and potentially habitable environments (1, 2). The confluence of high impact flux, extensive magmatism and volcanic outgassing, and aqueous alteration of basalts promoted the exchange of H between different reservoirs: the mantle, crust, surface, and atmosphere. Progressive sequestration of H in the crust (3) and escape of H to space (4) resulted in the secular drying of the Martian surface such that, by the Amazonian period (<3 Ga), the hydrosphere had evolved to resemble its currently dry and largely horizontally stratified state. Until materials are returned directly from Mars, meteorites are the only samples that can be dated precisely and measured for H isotopes (H and D) to trace the distribution and evolution of water among the different reservoirs over time. Although approximately half of the Martian surface has a Noachian exposure age (5), nearly all meteorites in our collection are geologically young (Fig. 1), limiting our understanding of water during Mars’ early and critical Noachian period. The previously studied Noachian-aged meteorite, Allan Hills (ALH) 84001, is a nearly monomineralic orthopyroxenite that is a unique rock type among all other Martian samples and likely derives from the same mantle source as the common shergottite meteorite class (6). The regolith breccia NWA 7034 and its pairs contain fragments of Noachian-aged material; however, they are entrained in a complex physical mixture that has been subjected to protracted metamorphic reworking and impact processing since at least ~1.6 Ga (7). H isotope compositions of Martian materials versus time. H isotope compositions [δD (‰, V-SMOW)] of apatites in Martian meteorites compared to Mars Science Laboratory D/H measurements of mudstones (51) and atmosphere (50) in Gale Crater. Symbols indicate apatite D/H data reported in this study: Teghaza (Teg) 001 (purple diamonds), Dhofar (Dho) 019 (orange right triangles), and Northwest Africa (NWA) 13134 (blue squares). Literature apatite H isotope data were compiled previously in (13), and ages represent apatite 207Pb/206Pb ages, except for the depleted shergottites, Nakhla, and Chassigny, which represent bulk Sm-Nd ages. The high δD box shown for ALH 84001 represents the high D/H apatite datum from (27). The depositional ages of mudstones in Gale Crater (3.5 to 2.9 Ga) are inferred from impact crater densities (33); however, it is unclear whether the measured D/H ratios record postdepositional events related to burial diagenesis and lithification up to the time of scarp retreat at ~1.5 Ga. Note that the shergottites can be classified into enriched (blue vertical bar), intermediate, and depleted (orange vertical bar) groups by their degree of enrichment in trace elements and in radiogenic isotopes. The enriched and depleted shergottites are interpreted to form either by variable degrees of assimilation of incompatible-element enriched crustal materials into ascending magmas (20) or by melting of distinct and chemically isolated enriched and depleted mantle sources that formed in the early Noachian (30) and variable mixing of these partial melts. δD=[(DH)sample(DH)V−SMOW−1]×1000‰, where V-SMOW is Vienna Standard Mean Ocean Water with D/H = 0.00015576 ± 0.000005 (National Institute of Standards and Technology). Here, we present H isotope compositions and 207Pb/206Pb ages measured in Ca-phosphate crystals in a newly recognized Noachian-aged (≥4.1 Ga) Martian meteorite, Teghaza (Teg) 001 (8). Teg 001 is a plagioclase-rich gabbroic diorite with evidence for alteration and metasomatism on Mars. Its mineralogy and bulk composition indicates that it represents previously unsampled crust that was derived from a distinct mantle reservoir (8), providing a glimpse into conditions on ancient Mars. We use the apatite data to assess the distribution and evolution of H in the ancient Martian mantle, crust, and atmosphere, showing that early differentiation processes did not fractionate H isotopes among Martian mantle reservoirs and that by 4.1 Ga, the juvenile Martian hydrosphere had become highly fractionated due rapid H loss to space. RESULTS Apatite morphologies, crystal orientations, and elemental compositions We focus our study on apatite [Ca10(PO4)6(Cl,F,OH)2] because it contains structural H and is the dominant reservoir for volatile elements in Martian meteorites. Figure 2 shows three apatite occurrences in Teg 001: large polycrystalline aggregates (Fig. 2A), elongate vein shapes (Fig. 2C), and lath-shaped apatite single crystals (Fig. 2E). Polycrystalline aggregates form 0.5 mm–by–0.8 mm masses composed of <10- to 50-μm polycrystals of dominantly Cl-rich apatite with <10% merrillite, Ca9Na(Mg,Fe)(PO4)7 (Fig. 2B). Individual polycrystals are homogeneous, have <1° of internal misorientation, typically share straight edges, and commonly intersect at triple junctions. Phosphate aggregates are intergrown with low-Ca pyroxene and are surrounded by shocked plagioclase (maskelynite), silica groundmass, and evolved melt (mesostasis). Vein-shaped apatites (Fig. 2C) extend as undulating crystals up to ~1 mm long with widths of <100 μm; are interspersed with minor merrillite; and share grain boundaries with pyroxene, maskelynite, ilmenite, and mesostasis. Vein-shaped apatites also display low degrees of internal misorientation across the length of the crystal (<2°; Fig. 2D). Lath apatites are smaller, up to ~40 μm by 400 μm, are associated with mesostasis (Fig. 2E), and have uniform orientations indicating single crystals (Fig. 2F). Electron images of three apatite morphologies observed in Teg 001. The grayscale panels are backscattered electron images; the colored panels are apatite inverse pole figure Z (IPF-Z) orientation maps from electron backscatter diffraction (EBSD) data. The wedge-shaped inset shows the IPF-Z color code. (A and B) A polycrystalline phosphate composed of randomly oriented chlorapatite (Cl-ap) and merrillite (mer) domains, intergrown with pigeonite (pgt) and maskelynite (mask). (C and D) A vein-shaped apatite between maskelynite and late-stage melt [mesostasis (meso)]. (E and F) A lath-shaped apatite surrounded by mesostasis and in contact with ilmenite (ilm). The circular gaps in EBSD map data (indicated by white arrows) are secondary ion mass spectrometry (SIMS) pits. Scale bars, 100 μm. Lower-magnification images of these apatites and their surroundings are shown in figs. S9, S12, S13, and S18. Apatites in Teg 001 are Cl-rich (Table 1) and overlap in composition with apatites measured in geologically young (~0.2 to 0.5 Ga) shergottites (fig. S1). Lath apatites are Cl-rich compared to vein-shaped apatites. Apatites in polycrystalline aggregates span the Cl/F range of the other two morphologies. The proximity and overlapping elemental compositions of the three apatite occurrences reflect similar formation conditions over a limited range of Cl and F activities. Table 1.Compositions and ages of Teg 001 apatites.Sample IDApatite typeCl (wt %)F (wt %)H2O (ppm)2 SEδD (‰)2 SE207Pb/206Pb age (Ga)2 SEA5_ap1Aggregate2.890.717567274105730A5_ap2Aggregate2.950.737428219911234.2460.156A5_ap3aAggregate2.780.756245115654404.1090.054A5_ap3bAggregate4.0610.154A5_ap5Aggregate2.920.819423629110431NW-A5_ap1aAggregate2.890.798050359762294.1500.066NW-A5_ap1cAggregate4.0100.106NW-A5_ap2Aggregate2.920.817515388975304.2470.142AE_ap2Aggregate3.490.5573194367142AE_ap3bAggregate3.210.5866324636423AE_ap5Aggregate2.940.776688144AE_ap6Aggregate88937AE_ap9*Aggregate2.16 0.66744559108740NE_AE_Ap1*Aggregate1.920.73820741112622A3_ap2Vein3.440.446256219764594.0650.066A3_ap4Lath4.220.447906545497 304.1690.058*Cl and F determined by SIMS; all other Cl and F concentrations determined by EPMA. H2O contents, δD values, and 207Pb/206Pb ages determined by SIMS. Compositions and ages of Teg 001 apatites. Cl and F determined by SIMS; all other Cl and F concentrations determined by EPMA. H2O contents, δD values, and 207Pb/206Pb ages determined by SIMS. SIMS apatite 207Pb/206Pb ages We analyzed U, Pb, and Th isotopes in apatite grains in three polycrystalline aggregates, in one vein-shaped apatite, and in one lath apatite. The 207Pb/206Pb ages of the apatite grains all overlap at the 95% confidence level, with individual ages ranging from 4.109 ± 0.054 Ga (2 SE) to 4.247 ± 0.141 Ga (2 SE) (Table 1). Collectively, the average 207Pb/206Pb age of Teg 001 apatites is 4.132 ± 0.061 Ga (2 SE, n = 8); the weighted average is 4.124 ± 0.028 Ga [2 SE, mean square weighted deviation (MSWD) = 2.47]. This age overlaps with the weighted 207Pb/206Pb age of zircons measured in the same sample (4.111 ± 0.013 Ga, 2 SE, MSWD = 1.55) (fig. S2) (8). The 207Pb/206Pb ages also overlap with the in situ 87Rb/86Sr isochron age of 4.238 ± 0.208 Ga (2 SE) measured in silicates (8). Teg 001 experienced at least one alteration or metasomatic event on Mars, evidenced by Na enrichments in zircon and plagioclase rims and severe alteration of pyroxene cores (8). If the overlapping U-Pb, Pb-Pb, and Rb-Sr ages correspond to a secondary event that uniformly reset all chronometers, then the crystallization age of Teg 001 is ≥4.1 Ga. This resetting event would have occurred at sufficiently high temperature to produce concordant U-Pb ages in zircons, and, thus, the apatite Pb-Pb ages also likely reflect resetting by the same 4.1-Ga event. The apatite ages thus record either a low-degree metamorphic event occurring at 4.1 Ga in the Martian crust or processes (including magma-crust interactions and metasosmatism) that occurred during the initial cooling of the Teg 001 magma body. SIMS apatite H2O contents and D/H ratios We measured H concentrations and isotope abundances in Teg 001 apatites (Table 1) adjacent to the 207Pb/206Pb age measurements. In addition to data on Teg 001, we analyzed H2O and D/H in apatites from the enriched gabbroic shergottite Northwest Africa (NWA) 13134 (9) and the depleted basaltic shergottite Dhofar (Dho) 019 (10) (see the Supplementary Materials). Teg 001 apatite H2O contents range from 5731 to 9423 parts per million (ppm) by weight, similar to other Martian apatites (Fig. 3 and fig. S3). D/H ratios in Teg 001 apatites are enriched by a factor of ~1.3 to 2.0 relative to terrestrial seawater [apatite δD ranges from +364 ± 23 per mil (‰) to +1126 ± 22‰, 2 SE]. Figure 3 shows that H2O and δD in Teg 001 apatites have a positively sloped relationship, similar to the trend observed in the enriched shergottites NWA 13134 and Grove Mountains (GRV) 020090 (11). SIMS measurements of H2O (ppm) and δD (‰, V-SMOW) of apatites in Martian meteorites. Data from this study: Teg 001 (purple diamonds), NWA 13134 (blue squares, enriched shergottite), and Dho 019 (orange right triangles, depleted shergottite). Data for GRV 020090 (enriched shergottite) are from (11). The δD measured in the modern atmosphere (50) is shown as a gray rectangle on the right y axis. The red circle is the inferred value for the primitive Martian mantle (20, 52). The purple and green curves show two-component mixing calculations for Teg 001 and for the shergottites, respectively, where the low-δD end-member is defined by the primitive mantle value (δD = +275‰). The high-H2O, high-δD component for the shergottites was defined by fluids in equilibrium with the modern atmosphere (50), and the calculation has no free parameters. For Teg 001, the high-δD component was a free parameter resulting in best-fit values of δD = 1967 ± 60‰, which reflects the H isotope composition of the Noachian hydrosphere at 4.1 Ga. The purple rectangle shows the range of best-fit δD values consistent with the apatite data (see the Supplementary Materials). Error bars are 2 SE and are sometimes smaller than the symbols. DISCUSSION Systematics of H2O and D/H in Martian apatite Apatite in Martian meteorites display substantial intrasample variability in H contents and isotopes, characterized by either positively sloped (Fig. 3) or negatively sloped (fig. S3) correlations between H2O and δD (12, 13). Several processes can potentially fractionate H isotopes in apatite leading to the negatively sloped relationship shown in fig. S3, including kinetic fractionation by diffusion, H2 degassing (14), mixing processes on Mars (13), low-temperature alteration (15), and terrestrial contamination (16). Deconvolving these processes is challenging, and mixing alone requires at least three distinct mantle D/H end-members to explain these data (13) (see the Supplementary Materials). We note that secondary ion mass spectrometry (SIMS) data from ~4.0-Ga (17) apatites in ALH 84001 fall on this negatively sloped trend. Figure 3 shows that 4.1-Ga apatite in Teg 001 and 0.2- to 0.5-Ga apatite in some shergottites display a positively sloped relationship between H2O and δD. This relationship is inconsistent with kinetic fractionation and terrestrial contamination (see the Supplementary Materials). The positive correlation is not due to shock implantation of H into apatite because of the high H2O contents in apatite relative to the modern Martian atmosphere, the high δD of the modern atmosphere compared to high-D/H Teg 001 apatites (Fig. 3), and the lack of the high-pressure high-temperature polymorph tuite in Raman data of the three apatite morphologies in Teg 001 (fig. S4). Instead, the positive relationships in Fig. 3 can be described by a two-component mixture of a low-H2O low-D/H end-member and a high-H2O high-D/H end-member. Following the interpretation for GRV 020090 apatite (11, 12) and for similar positive correlations measured in quenched impact melts (16, 18, 19), these end-members represent a D-poor mantle reservoir and a fractionated D-rich reservoir that exchanged H with the Martian hydrosphere, either directly or by interacting with fluids that were at some point in isotopic equilibrium with the atmosphere. Mantle-derived low-D/H water in Martian meteorites The lowest δD values measured in Teg 001 apatites (δD = +364 ± 23‰ and +367 ± 142‰) are similar to the value measured in an olivine-hosted melt inclusion in Yamato 980459 [+275 ± 20‰; (20)], interpreted to represent the primordial δD of the depleted shergottite mantle source. Low δD mantle values are also inferred from SIMS measurements of undegassed pyroxene cores in nakhlites (1.3 Ga) (21) and of quenched melt inclusions in Chassigny (1.3 Ga) (18). Considering that Teg 001, nakhlites and Chassigny, and the shergottites sample discrete mantle reservoirs with variable H2O contents (22), the similar inferred δD compositions of their mantle sources support the presence of a common primordial H reservoir that is widespread in the Martian mantle (19). This suggests that early differentiation processes did not efficiently fractionate H isotopes among distinct mantle reservoirs and that the value of +275‰ represents a maximum estimate of the silicate residue of the magma ocean (23). The value is high compared to Earth seawater (δD = 0‰), possibly as a consequence of magma ocean degassing processes and delivery of D-rich cometary material to early Mars (24). Higher δD values measured in minerals and glasses record incorporation of D-rich materials into mantle-derived magmas during migration through the crust (20) or, possibly, incorporated into the mantle sources themselves by recycling and redistribution of H (e.g., burial and metasomatism) (13). A Noachian hydrosphere component in Teg 001 apatite For Teg 001, we calculated a range of plausible δD values for the end-member representing the Noachian hydrosphere by fixing the magmatic value to +275‰, propagating uncertainties in measured D/H ratios, and minimizing the misfit between the model and apatite data (Materials and Methods). The resulting best-fit high-H2O end-member has δD = +1967 ± 60‰. Slightly higher values are permitted if the magmatic value is also allowed to vary, e.g., if the magmatic value was Earth like with δD = 0‰, then the best-fit δD for the high-H2O end-member is +2500‰. Our calculated δD value overlaps with values measured in 3.9-Ga secondary carbonates in ALH 84001 [avg. δD = +1051‰, up to +2029‰; (18, 25)], providing strong support that our estimate is a signature of the δD of the Noachian atmosphere at 4.1 Ga. These D-rich ancient minerals indicate that Mars lost a substantial proportion of its juvenile atmosphere to space within a few hundred million years after the magma ocean solidified (26) and that, during its earliest history, there was extensive exchange of H between the atmosphere and the crust. The measurements provide evidence for an active and vertically integrated Noachian hydrosphere (3). We note that our modeling suggests that the high δD value (+3000‰) measured in one apatite in ALH 84001 (27) might reflect an additional component or a secondary process that modified the original D/H ratio (see the Supplementary Materials). The shergottite data in Fig. 3 can be explained by a mixture of D-poor magmatic H and a D-rich component defined by the modern atmosphere [δD = +4950 ± 1080‰; (6)]. This is consistent with D/H measurements in shergottite impact glasses (16, 28) and indicates that the Martian atmosphere has had a roughly constant D/H ratio over at least the last ~0.5 Ga due to low H escape rates (15, 29). Our comparison of apatites of different ages shows a secular increase in the D/H ratio of the Martian hydrosphere due a combination of efficient Jeans’ escape of H (4), progressive alteration of basaltic crust, and subsequent burial during the earliest phases of Mars’ evolution (3). Nature and timing of assimilation of D-rich fluids by Martian magmas Much of the debate about the D/H ratio of the Martian mantle has focused on shergottites because they are the most abundant Martian meteorite class (29) and sample discrete Nd and Sr isotopic mantle reservoirs that separated ~40 million years after Mars formed (22, 30). On the basis of D/H ratios measured in melt inclusions in primitive olivines, Usui et al. (19, 20) inferred that the mantle sources of both enriched and depleted shergottites have a uniform D/H ratio and that enriched shergottite magmas had preferentially assimilated D-rich and incompatible-element rich crustal materials during melting and ascent. In contrast, Barnes et al. (13) used averages of D/H measurements in several mineral and glass phases to argue that enriched and depleted shergottites sample distinct D-rich and D-poor mantle reservoirs, respectively, and that both magma types mixed with an intermediate crustal reservoir with a range in δD of +1000 to +2700‰. However, some apatite δD values in depleted shergottites exceed +2700‰ (e.g., Dho 019; Fig. 3) and cannot be explained solely by mixing with this proposed intermediate reservoir. Likewise, δD values lower than +1000‰ measured in enriched shergottite apatites (e.g., GRV 020090) (Fig. 3) (11) would require an additional mechanism or mixing component. To explain high δD values in some depleted shergottites, Barnes et al. (13) suggested redistribution of water from the D-rich enriched mantle to the depleted mantle, but it is unclear how such exchange occurs. Furthermore, given the early separation of the enriched and depleted reservoirs, the mixing model in (13) requires that the enriched mantle acquired its high δD (~+4000‰) by ~4.5 Ga; however, the process that drove this extreme fractionation remains unexplained. We find that the overlapping ranges in magmatic D/H ratios of the enriched and depleted shergottites (Figs. 1 and 3) imply that trace element and radiogenic isotope enrichments are decoupled from H isotopes in Martian magmas. This is supported by our measurements of Teg 001, which is highly enriched in incompatible elements relative to shergottites and samples a distinct mantle reservoir (8) but has low magmatic D/H ratios. The decoupling suggests that, while enriched Martian magmas may have obtained their incompatible-element rich signature directly from their sources and/or due to crustal assimilation during ascent, a separate process influencing nearly all Martian magmas resulted in assimilation of a D-rich component derived from the continuously fractionating hydrosphere. The apatite data indicate that some magmas were further subjected to secondary H loss (degassing) and redistribution and/or require exchange with additional D-rich component(s) that have low water activities (fig. S3) (13, 19), e.g., continuous low-temperature exchange with fractionated thin films. We propose that assimilation of D-rich crustal fluids occurred during protracted cooling of magma bodies in the shallow altered basaltic crust or during metasomatic episodes in the shallow crust that postdated the initial cooling of the magma body (Fig. 4). This is consistent with O and Cl isotope evidence for assimilation of hydrous fluids derived from the crust or surface in other Martian meteorites (31, 32). It is noteworthy that pyroxenes in both Teg 001 and NWA 13134 have coarse (≤2 μm) exsolution lamellae (8, 9), indicative of slow cooling, which may have promoted magma-fluid interactions in the crust. Cartoon depicting conditions on early Mars and interactions between ascending magmas and crustal fluids. Potential fluid sources Teg 001 apatites provide a signature of crustal water during the rapidly evolving pre-Noachian and Noachian periods, when Mars had an active hydrosphere that promoted exchange between atmospheric and interior reservoirs (3, 23, 24, 26). High δD values in shergottites indicate that the crust continued to evolve such that fluids with D/H ratios approaching the modern atmosphere value were assimilated by ascending magmas. Although the age and significance of H isotopes measured in Gale Crater mudstones is unclear, if the high measured δD values (>3000‰; Fig. 1) correspond to the time of deposition [3.5 to 2.9 Ga; (33)], then it suggests that the D/H ratio of the Martian atmosphere had evolved to nearly its modern value by ~3 Ga. During episodes of volcanism, ascending magmas stalled in the crust and exchanged hydrogen with crustal fluids or, possibly, tapped meteoric waters directly by interacting with shallow hydrothermal systems induced by impacts or by magmatism (Fig. 4) (34, 35). Potential sources of fluids include buried aquifers of meteoric ice or water, dehydration of altered and metamorphosed basaltic crust (36), and fluids evolved from other degassing magma bodies (37). Exposed Noachian crust is altered and clay-rich, with orbital detections of serpentine, prehnite, and chlorite in eroded Noachian terrains (38), confirmed by in situ measurements of basaltic alteration products by the Perseverance rover (39). Interspersed within the altered igneous crust are buried impact craters and sedimentary basins (40, 41), filled with wind-blown and water-laden sediments and evaporites that could host buried aquifers present as ice or as groundwater (Fig. 4). Hydrated minerals and pore fluids present in altered basalts and sediments would be dominated by the D/H ratio of fluids in equilibrium with the Martian atmosphere (42) [clay minerals would have slightly lower in D/H than the coexisting fluid; (3)]. Subsequent volcanism would progressively bury these altered basalts and associated volcaniclastic sediments forming the thick Noachian crust. A major fraction of this crust formed early and before the creation of most of the major basin-forming impacts and volcanic provinces (43). Zircon, apatite, and pyroxene microstructures in Teg 001 indicate some degree of high-pressure, low-temperature metamorphism, and deformation after emplacement, possibly due to impact reworking. Subsequent pulses of volcanism migrated through the layered crust, stalling and crystallizing during ascent. While residing in the altered crust, magma bodies with relatively low water activities assimilated crustal fluids by driving small-scale hydrothermal systems in the altered crust or by direct assimilation of meteoric fluids. The probability of impact-driven magmatism and hydrothermalism was higher in the Noachian, which could also drive magma-fluid interactions. These represent candidate scenarios for supplying D-rich fluids that variably interacted with Martian magmas (31, 32, 44). Such interactions may have stabilized Cl-rich apatites, which crystallized and recorded the progressive assimilation of D-rich crustal fluids during late stages of magma cooling (Fig. 3). The consistent relationships between apatite H2O and δD values in both ancient and young Martian apatite suggest that the Martian mantle contains a widespread primordial component that is D-rich relative to Earth and has remained unfractionated since the solidification of the magma ocean. Most Martian magmas appear to have assimilated D-rich fluids that were at some point in isotopic equilibrium with Martian atmosphere. The D-rich crustal reservoir is laterally extensive and is dominated by materials that were processed during Mars first billion years, when there was widespread surface alteration, efficient crust building and burial of altered basalts and sediments, and extensive redistribution of H between the atmosphere and crust. MATERIALS AND METHODS All data are from polished thick sections of Teg 001, NWA 13134, and Dho 019. The sections were cleaned with isopropanol and initially coated with a ~20-nm carbon coat for scanning electron microscopy (SEM) and electron probe microanalysis (EPMA). A thicker carbon coat was applied before SIMS H isotope measurements. For Teg 001, the thick carbon coat was removed using 0.25-μm diamond suspension, and the section was then cleaned in isopropanol, dried, and recoated with gold before the U-Pb SIMS session. After each SIMS session, the sections were imaged using SEM to document SIMS pit locations and to identify spots that intersected cracks or overlapping phases. The gold coat was removed from Teg 001 using a KCN solution, and the phosphates measured by SIMS were analyzed using Raman spectroscopy. Last, the uncoated section was vibrapolished in colloidal silica for 1 hour before electron backscatter diffraction (EBSD) analysis. SEM and EBSD A Hitachi SU-3500 SEM at JPL was used for initial backscatter electron imaging and phase identification to locate apatites in all three sections. For Teg 001, additional imaging and EBSD analysis used a ZEISS 1550VP field-emission SEM at Caltech. EBSD maps were collected using the AZtecHKL EBSD system in four regions of interest: one large map covering a region ~1.6 mm–by–2.3 mm with several large polycrystalline apatite aggregates (subset in Fig. 2A), a magnified 250 μm–by–250 μm area of a polycrystalline apatite (fig. S5), and maps of the vein (Fig. 2C) and lath (Fig. 2E) apatites. EBSD analyses were performed in variable pressure mode (25 Pa) at 20 kV and 6 nA using a focused beam with the sample tilted at a 70° angle. The step size was 0.75 μm in the large map and 1.5 to 2 μm for the remaining maps with an acquisition speed of ~13.5 Hz. EBSD data were analyzed using HKL Channel5 software Tango and Mambo. Electron probe microanalysis We collected wavelength-dispersive x-ray spectroscopy measurements of apatite and merrillite using a JEOL JXA-iHP200F field-emission EPMA at Caltech. Beam conditions were 15 kV, 10 nA, and a 5-μm spot size. Peak count times were 20 s for all elements. For F, the backgrounds were measured off-peak for 10 s each; for all other elements, we used the mean atomic number method (45). Raw x-ray counts were reduced using the CITZAF procedure (46). We used a set of synthetic and natural standards for element quantification. Diffracting crystals and standards: F (LDE1, phologopite), Cl (PETL, sodalite), Ca (PETJ, Durango apatite), P (PETJ, Durango apatite), Fe (LiFL, fayalite), Mg (TAPL, forsterite), Na (TAPL, Amelia albite) Mn (LiFL, tephroite), and S (PETL, Durango apatite). SIMS H2O and D/H H abundances and isotopes were quantified using a Cameca 7f-GEO at Caltech and followed procedures detailed in (16, 28). The carbon-coated thick sections were placed in a high-vacuum sample storage chamber for a minimum of 3 to 5 days before analysis to reduce hydrogen backgrounds. Apatites in NWA 13134 were measured in April 2024, Dho 019 in November 2024, and Teg 001 in November 2024 and February 2026. The primary ion beam was Cs+. Using a mass resolution power (MRP) of 5000, we measured negative ions of masses 12 (C), 17.003 (OH), 18 (O), 19 (F), 28 (Si) (30 Si was used in the February 2026 session), 31 (P), 32 (S), and 35 (Cl). The beam current was held at 4 to 5 nA, and each spot was presputtered for 120 s over a 10 μm–by–10 μm area. For acquisition, the raster size was reduced to 2 μm by 2 μm, and data were collected in peak-jumping mode from masses 12 to 35 for 15 cycles. A field aperture was used to only collect secondary ion signals from the 8- to 10-μm crater center. Electron gating of 80 to 90% was used to further reduce any possible edge contributions. A suite of apatite standards [compositions reported in (22)] was used to quantify H2O (fig. S6) and other element concentrations: Ap003 (Durango), Ap004 (Atlas Mountains), Ap005 (Crystal Lode), and Ap018 (Lake Baikal). We measured San Carlos olivine to subtract background H on the standard indium mount. For the Martian sections, we measured nearby maskelynite and pyroxene to monitor and subtract background contributions to OH counts, which were always small (~100 ppm) relative to OH measured in apatite (~4000 to 9000 ppm) (data S1). A similar set up was used to analyze H and D, but the MRP was 1200 and the beam current was slightly lower at 2 to 3 nA. Using 1H− ion images, we relocated spots measured using the protocol described in the above paragraph and reanalyzed the same spots for H isotopes; the ion images were also used to identify and avoid cracks in the analytical volume. Presputtering was set to 30 s in these cases. For a few analyses, the D/H measurement was made on a clean area of apatite adjacent to the corresponding OH pit, with a 120-s presputter before acquisition. Images of all SIMS spots and detailed SIMS data are included in the Supplementary Materials. Ion counts were collected for 30, 50, or 100 cycles. Increasing the number of cycles reduced the standard error slightly, but D/H ratios were largely unchanged. Each analysis was inspected for severe D/H ratio drift or abrupt step changes in counts, and the SIMS pits were inspected under SEM for potential beam overlap with cracks and with other phases such as merrillite. Analyses of maskelynite were used to determine background D/H ratios; D counts in maskelynite were low, and so additional measurements up to 600 cycles were made to verify the background D/H values (−100 to −16‰). The maskelynite analyses contain an unknown but likely major contribution from terrestrial H given their low intrinsic H contents. We used standard Ap004 [δDaccepted = −45 ± 5‰; (22)] to monitor instrumental mass fractionation using the following equationsα=(D/H)measured/(D/H)accepted(D/H)IMF−corrected=(D/H)measured×(1/α) For the April 2024 session (NWA 13134), we obtained α = 0.573868 ± 0.005459 (2 SD). For the November 2024 session (Teg 001, Dho 001, and NWA 13134), we obtained α = 0.833558 ± 0.013327 (2 SD). For the February 2024 session (Teg 001), we obtained α = 0.951776 ± 0.032214 (2 SD). For Teg 001 and for NWA 13134, replicate measurements of nearby areas apatite in different analytical sessions produced internally consistent D/H values (data S1). SIMS U-Pb isotopes Apatite U-Pb dating was performed on the UCLA CAMECA ims1290HR using a Hyperion-II oxygen ion source. Samples were gold coated before measurements. We used an either 7- or 10-nA O− primary beam focused to a ~10- to 15-μm spot, with a 10-μm raster to preclean the surface during presputtering to limit common Pb contamination and a 2-μm raster during analysis to limit the depth of the pit and the associated charging. Oxygen flooding to ~1 × 10−5 to 2 × 10−5 torr and MRP of ~5000 were used to separate interference species, primarily molecular interferences on the Pb isotopes. Species measured included 40Ca231P16O4, 204Pb, 206Pb, 207Pb, 208Pb, 232Th, 238U, 232Th16O, 238U16O, 232Th16O2, and 238U16O2, all as singly charged ions. Following (47), an age calibration was established using the 238U16O2/238U16O versus 206Pb/238U16O Relative Sensitivity Factor of the primary age standard [BR5, 2040 Ma; (47)]. The higher abundance of the U oxide ions over atomic U+ made this calibration higher precision than in calibrations using UOx/U (47). The accuracy of the calibration line was checked against secondary age standard AFG2 [478 Ma; (47)]. Common Pb contamination used a 204Pb correction. For the standards, common Pb compositions recommended in (47) were used for corrections, while, for the unknowns, we assumed that common Pb derived from laboratory contamination and used the composition of southern California environmental Pb as measured in San Diego Bay (48). Unknown apatite U concentrations were estimated using the observed 238U16O2/40Ca231P16O4 of BR5 and AFG2, for which Kennedy et al. (47) report U concentrations. In the Teg 001 apatites, 232Th16O2/40Ca231P16O4 ratios were substantially lower than in the standards, such that a linear calibration curve led to negative calculated Th concentrations. Lower Th concentration standards are required for quantification. Raman spectroscopy We used a Renishaw inVia Raman spectrometer at Caltech to identify phosphate mineralogy in Teg 001. The system was calibrated using a Si standard to an accuracy of 0.1 cm−1. We used a green 514-nm laser at 10% power and 50× optical magnification with a 3000 lines/mm grating. Data were collected from 55 to 1500 cm−1 over three accumulations with a 10-s exposure time. Spectra are compared to library Raman data (RRUFF) in fig. S4. All three apatite morphologies have Raman shift peaks overlapping with chlorapatite. Analyses of merrillite show two diagnostic peaks near 957 and 973 cm−1, distinct from the single peak near 955.6 cm−1 measured in chlorapatite, and consistent with library spectra of merrillite. We found no evidence for the high-pressure high-temperature polymorph tuite. Mixing calculations We constructed a simple two-component linear mixing calculation to model the positive relationship between H2O and δD shown in Fig. 3. The low-H2O low-D/H magmatic component (A) is an apatite-saturated melt and the high-H2O high-D/H component (B) is an assimilant, either a hydrous melt or aqueous fluid in isotopic equilibrium with the Martian atmosphere. We apply a constant partition coefficient of H2O between apatite and melt of 0.2 (49) and assume no D/H fractionation accompanies apatite crystallization. We then simply mixed the two components and calculated the D/H ratio of the melt as a function of mixing proportions and the corresponding apatite H2O concentrations and D/H ratios. In 1/H2O versus D/H space, the mixing hyperbola reduces to a line (e.g., fig. S7) such that D/H ratios in the melt can be calculated as a function of the mixing proportions of components A and B(D/H)calc,melt=[XH2OA×XH2OB]×[(D/H)B−(D/H)A]XH2Omelt×(XH2OA−XH2OB)+XH2OA×(D/H)A−XH2OB×(D/H)BXH2OA−XH2OB(1)where (D/H)calc,melt is the calculated D/H ratio in the melt at each mixing proportion, given the mass fraction of H2O (XH2O) and D/H ratio in components A and B. For these calculations, we assume that component B is pure H2O and that it dissolves as molecular H2O in the melt, which further simplifies the mixing equation. As a result, our calculations represent minimum mixing proportions (<3% by mass) required to explain the high-D/H Teg 001 apatites and maximum calculated (D/H)B values in our inversion (see below for sensitivity tests). For the shergottite mixing calculation, there were no free parameters: XH2OA = 5,000 ppm, (D/H)A = 1.98 × 10−4, and (D/H)B = 7.55 × 10−4 to 1.09 × 10−3, which is plus and minus two SDs about the average value measured in the modern atmosphere by the Sample Analysis at Mars instrument on the Mars Science Laboratory (50). For Teg 001, the magmatic component was defined as XH2OA = 25,000 ppm and (D/H)A = 1.98 × 10−4, corresponding to δD = +275‰. The XH2OA value represents an evolved melt in equilibrium with apatite containing 5000-ppm H2O, applying the partition coefficient reported in (49) for Martian Cl-rich apatite. We then set up a linear least squares routine to calculate (D/H)B or the D/H ratio of the assimilant. To do this, we employed a Monte Carlo routine using the measured H2O contents and D/H ratios in Teg 001 apatites and resampling each analysis by its analytical uncertainty, treating H2O and D/H errors as independent (Table 1). For 1,000,000 synthetic datasets, (D/H)B was then varied to minimize the misfit between the calculated mixing hyperbola and the apatite H2O and D/H data. The result of this exercise resulted in a best-fit (D/H)B = 4.62 × 10−4 with an upper bound of 4.71 × 10−4 and lower bound of 4.53 × 10−4, corresponding to δD = +1967‰ (+58‰/−55‰). The relatively narrow confidence interval of ±60‰ reflects the precision of the SIMS measurements. We also performed sensitivity tests on our fixed or assumed parameters (fig. S8). We varied XH2OB, or the H2O content of the assimilant, from 50,000 ppm (e.g., a hydrous melt in equilibrium with chlorapatite containing 10,000-ppm H2O) to pure H2O and calculated the best-fit (D/H)B. The result shows a positive relationship that has a minimum best-fit δD = +1142‰ and asymptotes to a value of ~+1967‰ (fig. S8, bottom panel). This gives us confidence that our estimates derived from the Monte Carlo simulation represent maximum bounds on the D/H of the Noachian atmosphere consistent with our apatite data and that lower values are permitted if the assimilant had lower H2O activity. Slightly higher values for (D/H)B are permitted if (D/H)A is lower than the assumed value of +275‰. For example, if (D/H)A = 1.56 × 10−4 (δD = 0), then the maximum best-fit (D/H)B = +2487‰. Producing even higher values of (D/H)B that are consistent with the δD of Teg 001 apatites would require an extremely D-depleted source with negative δD, which is unsupported by measurements of D/H in Martian materials and by models of the early solar nebula and accretion of Mars. Apatite values that fall below the trends for both the shergottites and Teg 001 could be explained by crystallization from melt that continued to evolve and concentrate H2O before, or without, assimilating component B.