Introduction

On February 18, 2021 the Mars 2020 rover Perseverance landed on the floor of Jezero crater, a ~ 45 km diameter impact crater that once contained a lake1. The crater hosts an ancient, well-preserved depositional fan/delta complex composed of sediments transported and deposited by streams >3.2-3.8 billion years ago2,3,4,5,6,7,8. Detailed investigation of the sediments and rocks in the Jezero fan can thus provide insights into the physical and chemical conditions in ancient aqueous environments on Mars, the duration and nature of past habitable environments, and their potential to sustain biological or prebiotic processes.

At the time of this writing, Perseverance has collected 11 samples from the fan for potential return to Earth, all currently stowed within the rover’s Sample and Caching System (like the sample we discuss here) or cached at the Three Forks Sample Depot. Here, we report the detailed mineralogical and chemical analysis of a conglomerate outcrop, informally named Onahu, sampled by Perseverance on Mars solar day (Sol) 882 (sample name Otis_Peak). Onahu is located on the top of Jezero’s western fan and provides the most definitive in-situ identification of Fe-phosphate minerals on Mars, observed as large (~100 s μm wide) green-blue grains (Fig. 1A).

Fig. 1: Context images for the Fe-phosphates in Onahu, and the Ouzel_Falls abrasion patch.
figure 1

A WATSON image enhanced color and with 1% contrast stretch showing the location of the green-blue Fe-phosphate minerals in the Ouzel_Falls abrasion patch on the Onahu outcrop. B HiRISE (High Resolution Imaging Science Experiment) image of Jezero crater showing the traverse path of Perseverance up to ~1000 Sols on Mars (OEB = Octavia E. Butler Landing Site). Inset (b) HiRISE image showing the location of the Onahu outcrop where PIXL analysis was conducted, and the Emerald_Lake workspace where a correlative sample was collected. C Front HazCam image of the Ouzel_Falls abrasion patch on the pebbly conglomerate outcrop Onahu. (D) WATSON image of the Ouzel_Falls abrasion patch with PIXL footprints outlined. Mars 2020 image reference numbers are provided in Table S1.

Full size image

Fe-phosphate minerals are rare on Earth and Mars, where Ca-phosphates are the dominant phosphate minerals (i.e., fluorapatite on Earth, and merrillite and chlorapatite on Mars9). Fe-phosphates have been inferred on Mars before in Gale and Gusev crater9,10,11, but, until now, have never been directly observed on the surface as we show here, using a combination of high-resolution imaging, chemical, and micro-structural analysis. Phosphate minerals are high value targets for Mars sample return because their chemistry and textural occurrence can be used to constrain the bioavailability of P for prebiotic chemical reactions and/or potential life9, past aqueous conditions on Mars9,12,13; and when igneous in origin, insights into magmatic volatile abundances14,15,16. Fe-phosphate minerals that precipitate in aqueous environments are particularly important astrobiological targets as they have been observed on Earth as products of dissimilatory Fe-reducing bacteria17,18, encrusting bacterial cells19, and growing on the hydroxyapatite components of bones, shells, and fossils that have reacted with Fe-rich fluids17,20. While biotic influences are not essential for Fe-phosphate formation on Earth or Mars, terrestrial Fe-phosphates are commonly associated with organic matter, either growing on or replacing plant detritus, or precipitating in environments where degradation of organic debris results in more reducing and P-rich conditions17.

This work summarizes the detailed textural, chemical, reflectance, and micro-structural diffraction analysis of the Onahu Fe-phosphates using the Planetary Instrument for X-ray Lithochemistry (PIXL) – a mapping X-ray fluorescence spectrometer21 – and images from the Scanning Habitable Environments with Raman and Luminescence for Organics and Chemicals (SHERLOC) instrument22, both mounted on the arm of the rover. The high-resolution PIXL and SHERLOC results are used to identify the Fe-phosphate mineral that is most likely present in Onahu, constrain the paleoenvironmental conditions that led to their formation, and assess if these conditions were habitable.

Results And Discussion

Imaging and compositional analyses of Onahu

Onahu is in the Otis_Peak formation (fm), recognized by its fluvial deposits of planar laminated and cross-stratified sandstone to granule conglomerates. Large, scattered boulders (called the ‘rubbly unit’) overlie the Otis_Peak fm, which could have been sourced from the olivine-rich units common outside Jezero crater23,24,25. The laminated, poorly sorted medium-grained sandstones of the Skrinkle_Haven member (Tenby fm) lie beneath the Otis_Peak fm23.

The Fe-phosphates described here were identified in a 50 mm circular abrasion patch informally named Ouzel_Falls abraded into the pebbly, cross-stratified conglomerate Onahu that crops out on the top of the western Jezero fan (Fig. 1B,C). The outcrop’s surface was abraded to remove surface dust and coatings and to provide a relatively smooth surface for analyses by proximity science instruments (PIXL and SHERLOC).

This work focuses on X-ray fluorescence (XRF) maps (3 maps collected between Sols 789–811; Fig. 1D), X-ray diffraction analysis of individual Fe-phosphate grains, and reflectance spectral images collected by PIXL, as SHERLOC Raman spectroscopy maps did not include the Fe-phosphate-rich areas. However, microscopic images collected by SHERLOC cameras (color Wide Angle Topographic Sensor for Operations and eNgineering [WATSON] images with ~15.9 µm/pixel resolution) provide complementary high-resolution context images of the Ouzel_Falls abrasion patch.

Chemistry

PIXL maps show the Ouzel_Falls abrasion patch is predominantly composed of gravel- to pebble-sized clasts of olivine (Fo61), altered mafic clasts with elevated Fe+Mg/Si (potentially basaltic polymineralic clasts that have been altered to clay minerals), and relatively Mg-rich Mg-Fe-carbonates (Mg# ~55) embedded in a fine-grained matrix. The matrix includes domains rich in Si and Mg-Fe-sulfate, and domains rich in relatively Fe-rich Fe-Mg-carbonate (Mg# ~31), intermixed with fine-grained fragments of the major clast types (Fig. 2A-C, Supplementary Figs. S4-S7, and Tables S3,S4).

Fig. 2: Compositional and textural analyzes of Ouzel_Falls by PIXL and WATSON.
figure 2

WATSON image with enhanced color and 1% contrast stretch (A) and PIXL Red-Green Blue (RGB) X-Ray-Fluorescence (XRF) maps (B, C) showing the location of the green-blue Fe-phosphate minerals, and other major phases in the Ouzel_Falls abrasion patch. RGB triangles show how color mixing between elements will appear, i.e., Fe+P = yellow in (B). The approximate footprints of the PIXL scans are shown in (A–C). D Ternary diagram illustrating compositional mixing between Ouzel_Falls silicates, carbonates, and Fe-phosphates at PIXL’s resolution with ideal linear mixing trends between the major phases and an Fe-phosphate with a P/Fe ratio of ~2/3. The high-P PMCs (PIXL Motion Counters) include those with P2O5 > 10 wt%, SiO2 < 30 wt%, and SO3 < 5 wt%. Terrestrial Fe-phosphate minerals compositions compiled from ideal stoichiometry. (E, F) PIXL XRF data illustrating how high-P PMCs are correlated with high-Fe concentrations and decorrelated with Ca. A weighted orthogonal regression through the high-P points in (E) results in a P/Fe slope ratio of 0.67 ± 0.11 (SD = Standard Deviation) (P/Fe ~2/3), b = −0.14 ±0.05. All quantitative element concentrations are corrected for surface roughness and diffraction using the method outlined in ref.38 (Supplementary Dataset 1).

Full size image

Fe-phosphates were identified in the carbonate-bearing matrix domain by their high P and Fe concentrations (up to 24.7 wt% P2O5 and 48.8 wt% FeO). These Fe-P “hot-spots” are closely correlated with the ~50–500 μm green-blue grains visible in WATSON images (Figs. 1A, 2B). Because the green-blue grains are similar to or smaller than the PIXL beam size, all the Fe-P-rich PMCs (PIXL Motion Counters = integers corresponding to each analysis point location in a PIXL map) contain elements that are present in surrounding minerals (e.g., Si and S) and are interpreted as mixtures of Fe-phosphate and surrounding matrix material. This mixing relationship is clear in Fig. 2D, which shows that the Fe-P hot-spots are consistent with mixtures of Fe-phosphate minerals and the major phases in Ouzel_Falls. Simulated models of these mixtures also produce trends similar to the empirical PIXL data presented here (see Supplementary Information and Figures S2,S3).

At PIXL’s resolution, the Fe-phosphates appear to be predominantly composed of only Fe and P based on the strong compositional and spatial correlation between the two elements (Fig. 2B,E). There are no other obvious correlations between high-P concentrations and other cations that are commonly found in phosphate minerals such as Ca, Al, Mg, Mn, and Na (Supplementary Fig. S1). No trace elements except Ni were detected in the Fe-phosphate-bearing PMCs (Supplementary Table S2).

Based on the lack of correlation with other major and trace elements, the composition of the Fe-phosphates is estimated here by determining the P/Fe ratio of the PMCs that contain the most Fe-phosphate (PMCs with P2O5 > 10 wt%, SiO2 <30 wt%, SO3 < 5 wt%). The slope of a weighted orthogonal regression through these points (a weighted linear regression that accounts for errors in both x and y variables carried out using the scipy.odr package in Python26,27) indicates the Fe-phosphates have P/Fe molar ratios of approximately 0.67 ± 0.11 (~2/3; Fig. 2E). The non-zero intercept from this regression (0.14±0.05) is consistent with mixing with surrounding Fe-bearing phases. According to our literature review and the RRUFF database of International Mineralogical Association approved minerals28, 19 known terrestrial Fe-phosphate minerals have P/Fe ratios in this range (see Supplementary Table S5). It should be noted however that the P/Fe ratio determined here is a first order estimate based on discretionary SiO2, P2O5, and SO3 thresholds (Figure S1), and there is likely more unquantified uncertainty as the Fe-phosphates are mixed with multiple end-members which is difficult to model with a linear correlation trend.

Minor phases identified by PIXL in the Ouzel_Falls abrasion include sulfates and spinels. The bulk composition of Ouzel_Falls contains significant sulfur (~6.6 wt% SO3; Table S2), most of which can be ascribed to Mg-Fe-sulfate minerals. Areas enriched in Mg and S were found across the PIXL scans, and typically associated with the Si-rich matrix (Supplementary Fig. S6). Ca-sulfate is uncommon in Ouzel_Falls and occurs in discrete monomineralic areas. Small grains of Cr-Ti spinel (likely Ti-bearing chromite) are rare and randomly distributed (Supplementary Fig. S8).

Reflectance

To identify which Fe-phosphate mineral with P/Fe ~2/3 is most likely present in Ouzel_Falls we examined their reflectance spectra in more detail using PIXL’s Micro Context Camera (MCC). The Fe-phosphates can be readily delineated using Green/Near Infrared (G/NIR) band ratio images, standing out as regions with the highest G/NIR in the abrasion patch (Fig. 3A-C).

Fig. 3: PIXL MCC (Micro Context Camera) reflectance analysis of the Onahu Fe-phosphates.
figure 3

A WATSON image of the Fe-phosphates within the Ouzel_Falls abrasion patch with no enhancement so colors are more comparable to PIXL MCC. B PIXL MCC false color RGB image, C PIXL MCC Green/Near Infrared (G/NIR) image. Fe-phosphate grains are circled in each image, with 4 representative grains selected for reflectance spectra comparison. D Binary plot showing the variation in UV/NIR and G/NIR in the Ouzel_Falls abrasion patch. The Ouzel_Falls Fe-phosphate grains plot in between terrestrial ferrous and ferric phosphates and surrounding matrix indicating they are likely mixtures at PIXL’s MCC resolution. Compilation of reflectance spectra for terrestrial phosphate minerals (sourced from the RELAB database29 [samples: vivianite specimen JB-JLB-G65, strunzite specimen JB-JLB-B99-B, strengite specimen JB-JLB-742], University of Winnipeg for metavivianite [collected using methods from75], and30 for synthetic vivianite* subjected to 1 minute of oxidation vs. 23 hours of oxidation.

Full size image

Four of the grains with the highest P concentrations were selected for further analysis and comparison with terrestrial Fe-phosphate reflectance spectra mainly compiled by the RELAB database29. We focus on reflectance ratios baselined with the NIR band, as this band is dominant in discerning the grains from the matrix. The available bands of PIXL’s MCC are; ultraviolet [UV, 385 nm], green [G; 530 nm], blue [B, 450 nm], and near-infrared [NIR, 735 nm]. As such, detailed comparisons between PIXL spectral profiles which contain only 4 bands, and high resolution RELAB measurements are limited. Moreover, there is additional uncertainty due to particle size effects on the spectral profiles, as RELAB measurements are based on powdered material.

The Fe-phosphate grains in Ouzel_Falls have distinct UV/NIR and G/NIR (Fig. 3D). The average reflectance ratios of Grains 1-3 are relatively similar, varying between 0.91-0.94 G/NIR and 0.60-0.71 UV/NIR, while Grain 4’s averages are much lower with 0.79 G/NIR, and 0.46 UV/NIR, revealing an overall increase in NIR relative to other wavelengths in this grain and possibly a different Fe-phosphate phase. Furthermore, many of the individual pixels within Grains 1–3 have similar reflectance ratios to the more ferrous terrestrial Fe-phosphates (vivianite and metavivianite), while the individual pixels from the core of Grain 4 are more similar to terrestrial ferric phosphates (strunzite and strengite).

Only the cores of each grain were used for comparison to terrestrial Fe-phosphates in Fig. 3D in order to minimize any mixing with surrounding Ouzel_Falls matrix. In order to assess if any zoning is present, additional analysis of the cores and rims of each grain were carried out (Supplementary Materials and Figures S9-S16, Table S6). The G/NIR ratio of Grains 1–3 were noted to decrease from grain core to rim, either indicative of mixing between the green-blue phosphates with surrounding reddish matrix, and/or possible zoning between green cores and redder rims within the phosphates themselves. Grain 4 is again unique, with very similar G/NIR ratios from core to rim, and a very slight increase in G/NIR from core to rim – opposite to the trend observed in Grains 1–3. The relatively low G/NIR signature from the core of Grain 4 could be due to the presence of a redder Fe-phosphate phase in the core vs rim, or insufficient abrasion as this grain is slightly recessed relative to the other grains within the abrasion patch.

In summary, based on the reflectance ratios of individual pixels within each grain (Fig. 3D), the Fe-phosphates reflectance spectra in Ouzel_Falls likely represent mixtures of relatively ferrous phosphates and more oxidized phosphates with higher NIR reflectance, and/or mixtures with the high NIR matrix that surrounds each grain at Ouzel_Falls. However, the significantly higher NIR reflectance in Grain 4 in comparison to Grains 1–3 could indicate it is a separate, more oxidized phase (Fig. 3D). The decrease in G/NIR and UV/NIR as oxidation increases in Fe-phosphates has also been observed in experimentally oxidized synthetic vivianite30.

Bright regions in the G/NIR images are also associated with vitreous olivine grains, and as such, estimating the volume of Fe-phosphates across the abrasion patch using MCC imaging is ambiguous. If the PIXL scan areas are assumed to be representative of the entire rock, Onahu contains ~1 vol% Fe-phosphates (78 PMCs > 10 wt% P2O5 / 6495 PMCs total).

Crystal Structure

As terrestrial reflectance data from Fe-phosphates are limited, further analysis is needed to narrow down the most likely Fe-phosphate phase(s) at Ouzel_Falls. Here we use X-ray diffraction peaks detected by PIXL to assess the Fe-phosphates’ crystallinity and compare their respective diffraction patterns to terrestrial Fe-phosphates. X-ray diffraction peaks in the Ouzel_Falls Fe-P hotspots indicate that the Fe-phosphates are predominantly crystalline, i.e., with grain sizes larger than >30–50 μm in the region with diffraction peaks (Fig. 4, and Supplementary Information). Diffraction peaks were also detected from the olivine and carbonate. The clays and high-Si matrix regions either do not diffract or diffract poorly, indicating they are either amorphous or micro-crystalline with grain sizes <30–50 μm31.

Fig. 4: Diffraction analysis results.
figure 4

A PIXL X-ray Fluorescence (XRF) map of the Ouzel_Falls abrasion patch overlain with the results of the sum diffraction analysis for each high-P region (dashed ovals) shown together with the relevant p value for that region. Grains 1-4 were also used for reflectance analysis and are colored in the same scheme as Fig. 3. One region labeled “(Ferro)Beraunite” is likely a combination of ferroberaunite and beraunite. B Desaturated PIXL XRF map of the Ouzel_Falls abrasion patch overlain with the results of the diffraction analysis per PMC. Circles over the analyzed PMCs are colored according to the identified mineral with the cross-correlation value represented as transparency (min = 0.679, max = 0.979, mean = 0.897). PMCs where the diffraction analysis could not uniquely identify a single Fe-phosphate mineral with high confidence are labeled “Not determined” either as a result of no diffraction detected (i.e., are amorphous to PIXL31) or the results of the diffraction analysis were ambiguous and thus no reliable mineral identification could be made. C–F Examples of spectra from each of PIXLs detectors are shown with the measured diffraction (identified with black arrows where there is a peak in one spectrum and not the other) that is compared to a look up table for all possible diffraction profiles for each together with the cross-correlation between the Look Up Table (LUT) and the measured diffraction (see supplementary material).

Full size image

From the 19 Fe-phosphate minerals with P/Fe ratios ~2/3 (Supplementary Tables S5, S7), 10 were selected for diffraction pattern comparison with the Ouzel_Falls Fe-phosphates in order to identify the most likely phase(s). The 10 phosphates were selected either based on their similarity to the Ouzel_Falls phosphates in color and reflectance, their prevalence on Earth and Mars, and/or as compositional end-members for completeness, and include:

  • Vivianite [Fe2+3(PO4)2•8H2O] — often blue/green in color with similar reflectance band ratios to some of the Ouzel_Falls Fe-phosphates, the most common Fe-phosphate in sedimentary environments on Earth17.

  • Metavivianite [Fe2+Fe3+2(PO4)2(OH)2•6H2O]— often dark blue/green to black in color with similar reflectance band ratios to some of the Ouzel_Falls Fe-phosphates, and an oxidation product of vivianite30.

  • Beraunite [Fe3+6(PO4)4O(OH)4•6H2O] — often red to red-brown, a ferric end-member for comparison.

  • Ferroberaunite [Fe2+Fe3+5(PO4)4(OH)5•6H2O] — often green to brown or gray, and a more reduced form of beraunite.

  • Strengite [Fe3+PO4•2H2O] — occurs in a variety of colors, compositionally unlikely (i.e., molar P/Fe ratio of 1:1), but reflectance is relatively similar to some of the Ouzel_Falls Fe-phosphates, and previously inferred to occur on Mars in the Paso Robles soils, Gusev Crater11.

  • Rockbridgeite [(Fe2+0.5Fe3+0.5)2Fe3+3(PO4)3(OH)5] — often blue/green in color, and an Fe-rich end-member.

  • Sarcopside [Fe2+3(PO4)2] — occurs in a variety of colors, and an anhydrous end-member for comparison.

  • Ferrolaueite [Fe2+Fe3+2(PO4)2(OH)2•8H2O] — typically brown to orange-brown, but its Mn-rich end-member laueite [Mn2+Fe3+2(PO4)2(OH)2•8H2O] was recently inferred to be present on Mars as an oxidation product of Mn-rich vivianite at the Groken Site, Gale Crater10. Laueite was also included in the diffraction analysis for completeness.

  • Ferrostrunzite and Ferristrunzite [Fe3+ Fe2+2(PO4)2(OH)2•6H2O and Fe3+Fe2+2(PO4)2(OH)3•5H2O] — typically yellow to light-brown, but their Mn-rich end-member strunzite [Mn2+Fe3+2(PO4)2(OH)2•6H2O] was also recently inferred to be on Mars with laueite as an oxidation product of Mn-rich vivianite at the Groken Site, Gale Crater10. Ferristrunzite has also been inferred in P-rich Paso Robles soil, Gusev Crater11. Strunzite was also included in the diffraction analysis for completeness and due to its similar reflectance band ratios to some of the Ouzel_Falls Fe-phosphates.

Out of the 12 Fe-phosphates listed above (also including laueite and strunzite), the intensity and position of the detected X-ray diffraction peaks are most consistent with metavivianite, beraunite, ferroberaunite, and/or ferrolaueite (Fig. 4). The results from diffraction analysis are consistent with reflectance spectra for each grain as those with the highest relative NIR reflectance (i.e., Grain 4 Fig. 4D) are identified as Fe3+-rich beraunite, while grains with relatively low NIR reflectance (and higher G/NIR ratios) are identified as mixed valence metavivianite and ferrolaueite (Grains 1–3, Fig. 4E,F).

Fe-Phosphate Identification

Chemical, reflectance, and diffraction analysis of the Fe-phosphates in Ouzel_Falls by PIXL indicate that the secondary phosphate minerals metavivianite, (ferro)beraunite, and/or ferrolaueite are likely present. This is the first time these rare Fe-phosphate minerals have been identified on Mars.

These minerals (as well as others of the 19 Fe-phosphate minerals with P/Fe molar ratios ~2/3) can also be differentiated based on their Fe-oxidation state and H2O contents. While PIXL is unable to directly measure these components, other instruments on the Perseverance Rover are sensitive to hydration and oxidation (SuperCam and SHERLOC). Unfortunately, these instruments did not target the Fe-phosphates in Ouzel_Falls.

Metavivianite, beraunite, and ferrolaueite can all form on Earth in association with the Fe-phosphate mineral vivianite30,32,33. Vivianite occurs as green-blue crystals and is the most common Fe-phosphate formed in sedimentary environments on Earth17. It is also found as a secondary phase in a variety of hydrothermal settings17. Metavivianite (triclinic) forms from vivianite (monoclinic) as Fe3+ increases, resulting in a degradation of crystal symmetry and darkening in color34. If Fe-oxidation continues, eventually the orange-brown amorphous pseudo-mineral santabarbaraite (Fe3+3(PO4)2(OH)3•5H2O) forms after metavivianite20,30. Reflectance spectra for santabarbaraite were not available for comparison at this time, however, we hypothesize that mixtures of dark green-blue to black metavivianite and orange-brown santabarbaraite and/or other Fe3+-phosphates could also produce the high NIR reflectance observed in the Ouzel_Falls phosphates (Fig. 3D).

Beraunite and ferrolaueite are much less common on Earth than vivianite, but both can occur associated with vivianite as Fe-oxidation products like metavivianite; or conversely be reduced to form vivianite32,33,35,36. Ferroberaunite is a relatively new mineral37, and slightly reduced form of beraunite that has been often misidentified as green beraunite in similar formation environments.

As all of these phases can occur together, it is possible that the Fe-phosphate assemblage at Ouzel_Falls comprises a mixture of (ferro)beraunite, metavivianite, ferrolaueite, santabarbaraite, vivianite, and/or other Fe-phosphate minerals that we are not able to differentiate at PIXL’s resolution, or where terrestrial analog data needed for comparison are limited. For example, no reflectance spectra for beraunite, ferroberaunite, or ferrolaueite were available for comparison at the time of this writing.

Fe-Phosphates Origin and Paleoenvironmental Conditions

In order to constrain the paleoenvironmental conditions that led to the formation of the Fe-phosphates in Ouzel_Falls, we first need to consider if they were incorporated into the Onahu conglomerate as detrital phases, or if they formed in place.

In either case, the Fe-phosphates described here are the first to be identified and sampled by Perseverance, indicating a unique formation environment (Ca-phosphates are common in Jezero crater9,38,39). Additionally, as metavivianite, ferrolaueite, (ferro)beraunite, and/or santabarbaraite are all known to form after oxidation of vivianite on Earth, it is assumed that they first formed as vivianite in a more reducing environment before being exposed to oxidizing conditions either within the Onahu conglomerate itself, or within the Jezero paleolake watershed.

Authigenic origin

The Fe-phosphates occur only in the Fe-Mg-carbonate-rich matrix, suggesting that they might have precipitated in-place (as vivianite), in the pore space of the coarse-grained conglomerate at the same time as the carbonate, or filled the remaining pore space after carbonate precipitation.

Experimental studies have shown that vivianite readily precipitates from aqueous solution in the presence of Fe2+ and PO43– 40,41. Because the speciation of aqueous phosphate is strongly dependent on pH, and more aqueous phosphate is converted to PO43- at higher pH values, vivianite often forms in natural systems at circumneutral pH (e.g., pH 6–917), and where Fe2+ and dissolved phosphate are present40,41. In anoxic lakes on Earth, the P/Fe2+ ratio needed for authigenic formation is relatively high, typically ~1:6 or more, and >1:30 in laboratory experiments42. These conditions may be met in inorganic systems and in biologically influenced systems. For instance, thermodynamic calculations based on new solubility measurements have shown that vivianite may form through the evaporation of dilute surface waters on Mars under anoxic conditions that support ferrous iron43. On Earth, vivianite tends to form where dissimilatory microbial reduction of Fe(III)-bearing phases in sedimentary environments increases pH, and releases Fe2+ and dissolved phosphate that is commonly bound to Fe(III)-oxide minerals17. Fe-rich carbonate, which occurs with the Fe-phosphates in Onahu, is also known to precipitate with vivianite on Earth in anoxic zones through the same process when organic carbon oxidation is coupled with reduction of Fe(III)-oxides17,44. Still, on an anoxic, less Fe(III)-rich45,46, and relatively P-rich early Mars47, similar associations of microbes and iron would not be needed to produce the reducing, Fe2+, and P-rich conditions where vivianite (and carbonate) can precipitate.

Authigenic vivianite formation has been previously inferred on Mars at the Groken Site, Gale Crater (although it was not directly observed) where it may have precipitated in P-Mn-rich nodules10,48. The Groken P-Mn-rich nodules are thought to have originally precipitated as Mn-rich vivianite within sediment pore space from a locally P-rich, reduced groundwater10. Over time, the vivianite was likely oxidized to laueite and/or strunzite, and eventually to the main components of the nodules today – hematite and Mn-hydroxide10.

The presence of metavivianite, ferrolaueite, (ferro)beraunite, and/or santabarbaraite in Ouzel_Falls suggests that the original vivianite crystals must have also undergone oxidation in-place if authigenic. The formation of separate phosphate-carbonate-rich, and Si-sulfate-rich domains in the conglomerate matrix (Fig. 2C) indicates that diagenesis in Onahu occurred under evolving paleoenvironmental conditions and potentially different redox states. It is hypothesized here that the vivianite originally precipitated in a reducing Fe2+-rich fluid associated with the formation of the carbonate-rich matrix (based on their co-occurrence), and was then eventually oxidized either from alteration by infiltrating oxidizing Si-S-rich fluids that formed the Si- and sulfate-rich (oxidized S) matrix in the remaining pore space and/or from exposure to the oxidizing martian atmosphere (Fig. 5). Based on the persistence of carbonate, it is assumed that if a Si- and sulfate-rich fluid infiltrated into the Fe-phosphates-carbonate matrix, the fluid was not strongly acidic, and/or the water-rock interaction was short lived. Regardless, vivianite is also relatively insoluble in low pH environments (in comparison to Ca-phosphates)9.

Fig. 5: Cartoon illustrating the potential paleoenvironmental history of the Onahu outcrop and Fe-phosphates if they formed authigenically.
figure 5

Multiple episodes of fluid-rock interactions or evolving fluid compositions are needed to explain the distinct occurrences of carbonate-rich matrix and Si,S-rich matrix. Detrital Fe-phosphates could also have been incorporated into the conglomerate before fluid infiltration.

Full size image

The Ouzel_Falls abrasion patch is also the most P-enriched PIXL target analyzed within the western fan (Supplementary Fig. S17; Table S2), particularly in comparison to other carbonate-rich siliciclastic rocks, indicating that, if authigenic, the Fe-phosphates likely formed from a localized P-rich fluid.

While the textural occurrence of Fe-phosphates within the Fe-rich carbonate-bearing matrix, the presence of two discrete matrix domains, and the overall P-enrichment in Ouzel_Falls support a model where vivianite precipitated and oxidized in-place, we should also consider if the Fe-phosphates may have precipitated from a similar process in sediments upstream within the Jezero paleolake watershed and were concentrated in the carbonate-rich matrix in Onahu through sorting.

Detrital origin

Terrestrial Fe-phosphates are not common detrital phases, and to our knowledge, have never been identified. Vivianite and metavivianite are soft (Mohs hardness of 1.5-2, and 2.5, respectively17,49) and unlikely to persist during extended fluvial or eolian transport. Beraunite, ferroberaunite, and ferrolaueite are slightly harder, with Mohs hardness of 3–432,37,50.

A source of detrital Fe-phosphates in Onahu could be upstream sediments where vivianite formed in a similar process as described above. However, vivianite and its oxidation products are also known to occur on Earth as secondary minerals in altered P-rich pegmatites36,51,52 or in metal-rich ore bodies (with either igneous or sedimentary protoliths) and skarn deposits50,53. At this time, there is no evidence of pegmatites, metal-rich ore bodies, and/or skarn deposits in the Jezero watershed (or elsewhere on Mars), but the discovery of highly evolved Fe-rich basalts on the Jezero crater floor38,54 indicates that more extreme differentiated lithologies could exist in the watershed that we are not able to identify with remote analyses alone.

The Jezero watershed is also known to contain significantly altered rock based on the presence of smectite, carbonate, and incision of the altered basement and mottled terrain units by fluvial valleys5. Aqueous alteration models demonstrate it is possible for Fe-phosphates to form during weathering of Ca-phosphate-bearing ultramafic rocks55; as such, they could have originally formed in the Jezero watershed from this process.

Regardless of the inferred origin of the Fe-phosphates, based on our current knowledge of the Jezero watershed, and comparisons to terrestrial environments, the Onahu Fe-phosphates most likely are not igneous or hydrothermal in origin and formed in a P- and Fe-rich aqueous environment at relatively low temperatures either as precipitates in sediments or as alteration products of P-rich rocks. If detrital, oxidation of vivianite could have occurred before or after transport and incorporation into Onahu.

Implications for habitability and Mars sample return

The identification of secondary Fe-phosphate minerals (metavivianite, ferrolaueite, (ferro)beraunite, and/or santabarbaraite) after vivianite in Ouzel_Falls provides evidence that many of the materials and energy source(s) needed for habitable conditions may have existed in the Jezero fan top, and/or within the Jezero paleolake watershed. Vivianite and its oxidation products have been inferred on Mars before9,10,11, but this is the first time the Fe-phosphates noted here have been identified using correlative high-resolution chemical, structural, reflectance, and imaging analyses.

Based on comparisons to natural systems on Earth, the vivianite in Ouzel_Falls most likely precipitated in conditions favorable to potential martian life within a low temperature, anoxic, reducing aqueous medium at approximately neutral pH17, with high concentrations of bio-limiting elements including carbon and phosphorus, and Fe-redox gradients that could provide an energy source. For example, dissimilatory reduction of ferric ions by bacteria when phosphorus is available, to the ferrous form in vivianite, can enable access to metabolic energy56,57. These conditions are most likely to have occurred when Mars was warmer and wetter, around the time of the Jezero fan formation ( > 3.2 Ga2,3,4,5,6,7,8).

Although vivianite is thought to have been important in the biogeochemical P cycle on early Earth43,58,59, it is largely absent from the ancient geologic record, possibly due to oxidation/dissolution driven by diagenesis and metamorphism58. In contrast, the results presented here, combined with the well-preserved nature of the western fan8 suggest the Onahu outcrop experienced minimal metamorphic alteration following the formation of the fan ca. 3.2 Ga, thus extending the preservational window of vivianite. However, while the available evidence supports a potentially ancient origin for these well-preserved Fe-phosphates, we cannot rule out their formation in another more recent event.

Many of the conditions needed for vivianite formation are also suitable for the preservation of chemical biosignatures, if present60. Terrestrial Fe-phosphates have been observed encrusting bacteria, and growing on/replacing organic detritus and the hydroxyapatite components of bones and fossils17. Furthermore, authigenic carbonate minerals like those that occur in the Ouzel_Falls matrix are particularly conducive to the preservation of chemical and morphological biosignatures, from organic molecules to microbial fossils61,62.

Perseverance can provide immense amounts of information on martian paleoenvironments through the identification of rare minerals like the Fe-phosphates described here when high-resolution proximity science instruments are deployed. Still, textural, compositional, and isotopic analyses of the returned sample (Otis_Peak) will be essential to constrain the ancient environmental conditions, chemical reactions, or even microbial activity that produced these unusual minerals.

Methods

PIXL Analyses

The PIXL instrument produces maps of elemental abundances on martian surface materials using X-ray Fluorescence (XRF). PIXL’s focused X-ray beam interacts with the surface in a ~ 120 µm diameter spot size for fluoresced X-rays of 8 keV, and larger diameters for lower energies21,31,63. PIXL rasters its focused X-ray beam across a predefined area of interest, and so produces analyses of multiple spots that are correlated with beam positions; from which the PIXLISE software package64 produces maps of elemental abundances (see21,31,38,39 for more details on PIXL’s operation). PIXL maps can include thousands of individual X-ray fluorescence spectra, each with unique positions recorded as PIXL Motion Counters (PMCs - integers tracking each time PIXL moves on to a new software activity that correspond to locations of each PIXL analysis in a scanned area).

PIXL obtained three XRF raster scans on the Ouzel_Falls abrasion patch (Sols 789, 790, and 811) at step sizes of 0.125 mm, for a total scan area of 97 mm2 (Fig. 1D). Energy-dispersive XRF spectra were acquired for each point for 10 s each (6495 PMCs total), with an additional 30 s of dwell time when PIXL’s adaptive sampling algorithm was triggered on rare phases of interest (such as phosphate minerals) in order to improve signal-to-noise ratios at these locations (123 PMCS, e.g.,65). PIXL’s micro context camera (MCC) also acquired images of the abrasion patch illuminated by ultraviolet (UV, 385 nm), blue (B, 450 nm), green (G, 530 nm) and near-infrared (735 nm) light emitting diodes (LEDs) with a spatial resolution of ~50 µm/pixel. The preflight calibration of the LED-MCC system allows multispectral analysis of these four wavelengths66,67. The XRF and MCC data were then analyzed using PIXLISE, allowing for detailed examination of elemental compositions and X-ray spectral characteristics correlated to context images. PIXLISE fully incorporates instrument calibration of optics, geometry, XRF quantification as well as the in-situ conditions of topographic effects and positional drift caused by diurnal thermal environment, enabling accurate correlation between XRF spectra and context images to within 50 microns68.

Mineral mixing modeling

PIXL’s analytical spot size varies between elements because lighter elements fluoresce at lower energies63. The beam diameter varies inversely with energy, meaning higher energy X-rays are focused to a smaller diameter. High-energy X-rays excite more high-energy fluorescence, so the relationship between PIXL’s analytical spot size and X-ray energy results in a larger analytical volume for lighter atoms and a correspondingly inflated measured distribution. While these effects are well-characterized and accounted for in a homogenous sample, they become more complex at grain boundaries. Mixing relationships between neighboring mineral grains appear nonlinear, complicating efforts to determine the “unmixed” end-member compositions of mineral phases that are smaller than, or close to, PIXL’s spot size (including the Fe-phosphates of interest here). Thus, to constrain the composition of Fe-phosphates, we modeled compositional mixing between the main minerals in Ouzel_Falls and various Fe-phosphate endmembers, taking the energy-dependent variation in spot size for different elements into account. See Supplementary Fig. S3 and Supplementary Information.

Crystallographic mapping and diffraction phase matching

PIXL has two detectors, and so can separate anisotropic X-ray diffraction peaks occurring from crystalline materials from isotropic X-ray fluorescence peaks when energies of PIXL’s incident radiation happen to be in a diffracting condition with d-spacings and orientations of crystal lattices in the target materials69. Diffraction, as well as surface roughness, can result in misestimation of elemental concentrations, as such, all XRF compositions reported here have been corrected for these effects using the method outlined in ref. 38.

When detected, X-ray diffraction can also be used to assess crystallinity, grain size, and grain texture of a target31,38,39 and to distinguish among minerals with similar elemental compositions70. Here, diffraction patterns from the Fe-P-rich areas of the Ouzel_Falls abrasion patch were compared to modeled diffraction patterns for several candidate Fe-phosphate minerals to identify the most likely phases (see Supplementary Information).

Sampling

The Perseverance rover attempted twice to collect a sample from the Onahu outcrop near and on the Ouzel_Falls abrasion patch, but the rock was too friable to acquire an intact core. A final sampling attempt was carried out ~30 meters NNW of the Onahu outcrop in a workspace informally named Emerald_Lake, in the same stratigraphic layer as Onahu. The Otis_Peak sample core was successfully acquired from this location.

As PIXL was not used to analyze the Emerald_Lake workspace, data from the SuperCam, WATSON, and CacheCam instruments on Perseverance are used here to confirm its similarity to Onahu (see Supplementary Information, Table S8 and Figures S18,S19).