The Possible Aqueous Origins of Manganese Alteration Minerals in the Máaz Formation of Jezero Crater

5.1. Guillaumes

Guillaumes has the highest Mn concentrations in all of the crater floor scans, up to 9.7 wt% MnO (Fig. 3A), which is an order of magnitude greater than MnO concentrations measured in Martian igneous meteorites (Udry et al., 2020). The six high-manganese PMCs form a single contiguous region (Fig. 3B), with clear correlations with respect to CaO and SO₃ (Fig. 3C,D), which indicate a mineral rich in Mn, S, and Ca. There are eight known minerals that contain these elements (Supplementary Table S1). Three of these minerals are silicates (“Na-metaschoepite,” Roeblingite, and “M2006-18-SiO: CaClFFeHMnNaZr”), and one of them is an (Al,Fe)-sulfate that contains Mg (svyazhinite). For the PMCs of interest, Mn is anti-correlated with Si (Supplementary Fig. S1), which makes silicate minerals unlikely. Similarly, Mn is anti-correlated with Al and Fe with a slight positive correlation between Mg and Mn (Supplementary Fig. S1), which makes svyazhinite unlikely. The remaining four minerals are despujolsite (Ca₃Mn⁴⁺(SO₄)₂(OH)₆ · 3H₂O), jouravskite (Ca₃Mn⁴⁺(SO₄)(CO₃)(OH)₆ · 12H₂O), saccoite (Ca₂Mn₂³⁺F(OH)₈ · 0.5(SO₄)), and an unnamed possible Mn(IV) analog of sturmanite (Ca₆Mn₂⁴⁺(SO₄)₂[B(OH)₄](OH)₁₀O₂ · nH₂O).

OPEN IN VIEWER

FIG. 3.

(A) Guillaumes PIXL map (overlaid on merged MCC/WATSON image) with MnO wt% colormap overlaid, highlighting the several PMCs close to 9 wt% at the bottom of the map. (B) Guillaumes PIXL map with 6 PMCs potentially identified as despujolsite shown in purple. (C) Binary plot of MnO and SO₃ for the entire Guillaumes scan, with the high MnO points that come out above the main cluster in a clear linear relationship shown in purple. These points are also spatially correlated, as seen in panel B. The expected molar ratios for possible Ca Mn sulfate phases are plotted, demonstrating the close match of the data points to the expected slope for despujolsite of 1:2. (D) Binary plot of MnO and CaO for the Guillaumes scan, with the high MnO points that come out above the main cluster in a clear linear relationship shown in purple. The expected molar ratios for possible Ca Mn sulfate phases are plotted, demonstrating the close match of the data points to the expected slope for despujolsite of 1:3 (as well as jouravskite and the sturmanite analog, though panel C demonstrates that jouravskite and the sturmanite analog are poor matches for the SO₃ data). MCC, micro-context camera; PMCs, PIXL Motor Controls.

To determine the most likely candidate from these minerals, we investigate the molar ratios of MnO to SO₃ and CaO (Tice et al., 2022; Jones et al., 2025; Liu et al., 2022; Hurowitz et al., 2021; Schmidt et al., 2025). The MnO to SO₃ ratio for jouravskite and the sturmanite analog is 1:1, despujolsite is 1:2, and saccoite is 4:1. The high-Mn PMCs in Guillaumes have a clear MnO:SO₃ ratio of 1:2 (Fig. 3C), which suggests that despujolsite is the most likely candidate. Similarly, the MnO:CaO ratio of saccoite is 1:2, while despujolsite, jouravskite, and the sturmanite analog have a 1:3 MnO:CaO ratio. The high-Mn PMCs in Guillaumes have a clear MnO:CaO ratio close to 1:3 (Fig. 3D). A linear regression through the high-Mn PMCs gives a molar MnO:SO₃ ratio of 0.51 and a molar MnO:CaO ratio of 0.39, which are consistent with despujolsite’s stoichiometry (molar Mn:S ratio of 0.5 and molar Mn:Ca ratio of 0.33) (Fig. 3C,D). The near-zero intercepts of these regressions (−0.024 mmol g⁻¹ for MnO:CaO; 0.017 mmol g⁻¹ for MnO:SO₃) indicate that all the MnO, CaO, and SO₃ within these high-Mn PMCs are contained within the likely despujolsite [i.e., no mineral stripping (Tice et al., 2022) is required].

The concentration of MnO + CaO + SO₃ in these PMCs averages 19.27 ± 0.22 wt%; this suggests that despujolsite constitutes approximately 20% of these PMCs if all the MnO, CaO, and SO₃ are contained in despujolsite, which is likely given that the previous analysis indicates a lack of mixing with other minerals. As PIXL cannot measure elements lighter than Na (atomic number Z < 11), PIQUANT assumes that all measured elements are present as oxides for its quantification by default (Heirwegh et al., 2022). Therefore, if all the elements are present as oxides and there are no additional elements lighter than Na, PIQUANT returns an ideal total quantification of 100% (Heirwegh et al., 2021; Heirwegh, 2023). Because of this, PIQUANT does not enforce geo-chemical summing to 100% (Allwood et al., 2020; Heirwegh et al., 2022). The difference between the ideal total quantification and the measured total quantification is referred to as the “mass deficit” and can be due to low Z elements in hydroxyl groups, water molecules, or carbonates, for example (Heirwegh et al., 2021; Heirwegh, 2023). For the high-Mn PMCs, the total quantification averages 90.96 ± 0.73 wt%, giving a mass deficit of about 9 wt%. This mass deficit may be spread across other minerals in this region and not entirely be due to low Z elements in despujolsite. Assuming the despujolsite makes up 20% of the PMCs, as discussed previously, the expected mass deficit due to the low Z elements in despujolsite [(OH)₆ and 3H₂O] is approximately 5 wt%, which is within the 9 wt% mass deficit for these PMCs allowing for a 4 wt% mass deficit contained in other minerals (for example Tosca et al., 2025; Supplementary Table S1).

This potential despujolsite identification cannot be confirmed by SHERLOC because the Raman spectra were not collocated with the high-Mn grain (Fig. 2,3). However, the SHERLOC spectra of other regions in the Guillaumes abrasion reveal the presence of hydrated Mg sulfate and anhydrous Ca sulfate (Siljeström et al., 2024). Interestingly, SHERLOC was unable to identify a specific mineral species for the sulfate spectra in Guillaumes, since it has features from many standards such as gypsum, epsomite, natural kieserite, and synthetic magnesium sulfate, suggesting mixed calcium and magnesium cations. While not collocated with the potential despujolsite grain, the mix of sulfate phases observed by SHERLOC in the same abrasion, some of which are hydrated and show signs of mixed cations, suggests the likely presence of hydrated Ca–Mn sulfates.

The magnetite grain identified in this abrasion patch (Tosca et al., 2025; Fig. 4A) overlaps with a high MnO region, with MnO concentrations up to 4 wt% at the center that decrease outward (Fig. 4B). The highest MnO values are in four PMCs contiguous with the magnetite grain, coinciding with a small group of sulfate-rich PMCs (SO₃ > 2 wt %; Fig. 4B,C). Although bulk Alfalfa contains < 1 wt % SO₃, a few PMCs reach up to 17 wt % SO₃. These points show a strong correlation between SO₃ and both MnO and CaO (Fig. 4D,E). CaO rises linearly with SO₃ (R² = 0.96, slope = 0.69 ± 0.09, intercept = 0.1 ± 0.1; Fig. 4D), while MnO displays a weaker correlation (R² = 0.3, slope = 0.02 ± 0.02, intercept = 0.54 ± 0.03; Fig. 4E). These trends indicate that the sulfur resides primarily in a Ca-dominant sulfate with very minor Mn, and that the rest of the Mn may be hosted by a sulfur-free Mn oxide (magnetite). FeO and MgO both negatively correlate with SO₃ (Supplementary Fig. S4); this suggests the absence of Fe and Mg in the sulfate lattice. FeO also shows little variation between the bulk and the high MnO PMCs along the MnO/SO₃ trend (Fig. 4E), consistent with Fe precipitation occurring independently of Mn-Ca formation. The data are consistent with a Ca-dominant sulfate phase such as gypsum (CaSO₄·2H₂O) or anhydrite (CaSO₄), with minor Mn that could reflect limited Mn substitution into the sulfate structure or contributions from co-occurring Mn-bearing phases. The Ca:Mn ratio in the four highest MnO PMCs, 36:1, does not match the stoichiometry of any possible Ca-Mn sulfate phases (Supplementary Table S1). If the Mn in these PMCs is incorporated into the sulfate lattice, the composition would correspond to gypsum or anhydrite with ∼2.7 wt% Mn substitution for Ca. The PIXL elemental data alone do not uniquely distinguish between gypsum and anhydrite, and resolving their hydration state is beyond the scope of this study.

OPEN IN VIEWER

FIG. 4.

(A) PIXL Alfalfa map (overlaid on merged MCC and WATSON image) with magnetite grain in orange. (B) Color map of MnO concentration in the Alfalfa scan, showing the highest concentrations of MnO in the center of the sulfate/magnetite grain. (C) Color map of SO₃ concentration in the Alfalfa scan, showing high concentrations of SO₃ overlapping with the magnetite grain in A. (D) Binary plot of SO₃ and CaO, with a linear regression fit to the 4 highest MnO PMCs, showing a linear relationship between CaO:SO₃ with a slope of around 0.7. This implies a primary Ca sulfate. (E) Binary plot of MnO and SO_3, with points colored by their FeO concentration. The 4 highest MnO PMCs are again fit with a linear regression, showing a linear relationship implying Mn substitution in the Ca sulfate. The consistency in FeO concentrations between bulk and the Ca-Mn sulfate assemblage confirms that Fe is not in the sulfate phase. (F) PIXL Alfalfa map with magnetite in orange, Ca-Mn sulfate assemblage in green, plagioclase feldspar in maroon, pyroxene in purple, and Fe silicate in blue.

Additionally, two of the most sulfate-rich PMCs in this scan have much lower MnO content, even though one of these PMCs is contiguous with the Ca–Mn sulfate assemblage identified above. These points could represent a Mn-poor Ca sulfate or a late-stage gypsum that formed after Mn was removed from the alteration fluid. Mn heterogeneity is expected since Mn²⁺ solubility (and therefore its availability to enter gypsum) drops sharply as soon as pH rises above ∼6 or as soon as Mn is oxidized to Mn³⁺/Mn⁴⁺. A fluid wave that has already oxidized/lost Mn can still carry Ca²⁺ and SO₄^2- and precipitate pure gypsum.

Many PMCs above 1.5 wt% MnO are scattered around the scan (Fig. 4B). These MnO-enriched PMCs are found in Fe silicate material at the top of the scan, which can be olivine or serpentine, as well as pyroxene at the bottom of the scan (Fig. 4F), in the form of augite. This is consistent with a Mn source from the host igneous rock that was later altered. The highest-MnO region is bordered by Fe silicate material (Fig. 4F), with the highest Mn concentration in the center of the sulfate/magnetite grain (Fig. 4B). The concentrations at the center of the magnetite/sulfate grain are ∼4 wt%, while concentrations fall to ∼2 wt% away from the sulfate grain, and then to ∼1 wt% at the border of the magnetite grain (Supplementary Fig. S5A). The uncertainty on the PMCs with > 2 wt% MnO has a 1 σ of 0.32–0.59 wt% (Supplementary Table S2). The measured gradient we discuss (from 4 wt% at the center to 1 wt% at the rim) spans 3 wt%, which is 5–9 σ above uncertainty. There is also a possibility of Mn leaching preferentially with respect to Fe from the host silicate into the sulfate (Supplementary Fig. S5B), which can happen under mildly acidic conditions where Mn²⁺ remains soluble and mobile over a wide pH–Eh range, enabling transport and local sequestration into secondary phases, while Fe is rapidly immobilized during oxidation (Khosravi et al., 2020). These observations are compatible with, but not diagnostic of, olivine alteration instigating secondary magnetite formation, and later alteration of the magnetite to form Mn sulfate.

6.1. Overview of alteration pathways in the crater floor

Previous analyses of crater floor rocks have identified multiple stages of aqueous alteration that affected both the olivine-rich Séítah formation and the overlying basaltic Máaz formation. Observations made with PIXL suggest a three-phase aqueous history corresponding to low-temperature olivine carbonation in the Séítah formation, followed by sulfate precipitation from vadose zone brines and perchlorate infiltration from oxidizing fluids under low water–rock ratios that extended through the crater floor (Tice et al., 2022). The SHERLOC instrument found similar evidence for distinct aqueous episodes in the crater, breaking up the sequence into an initial episode of carbonation in Séítah and a later briny episode during sulfate and perchlorate formation (Scheller et al., 2022). Because the SHERLOC detection of perchlorate accompanied the sulfate detection in the same scan, Scheller et al. concluded that they formed at the same time during a single briny episode, rather than as separate events as inferred from the PIXL-based sequence (Tice et al., 2022), which puts perchlorate infiltration as a later stage. The Mn-bearing minerals identified in the Máaz formation expand on this sequence; this indicates that fluid chemistry and redox state evolved substantially through time.

6.2. Formation of Mn sulfate phases in the Roubion member

6.2.5. Evidence for serpentinization and Mn mobility

The presence of serpentine group minerals in Roubion implies serpentinization as a possible alteration pathway for Mn leaching from primary olivine, followed by later aqueous alteration leaching Mn from Fe silicates to form Mn sulfate. This pathway is consistent with the presence of an altered grain of olivine, bordered by serpentine, within < 2 mm of the Mn sulfate (Supplementary Fig. S3). The Mn/Fe gradient in this olivine grain (Supplementary Fig. S3) shows surface depletion and core enrichment in Mn, consistent with Mn leaching and reprecipitation during successive alteration. Although Fe silicates do not directly rim the sulfate region (Supplementary Fig. S2C and D), Mn-bearing fluids could have precipitated sulfate nearby. Therefore, it is possible that the Mn sulfate could be a later alteration product of serpentine/olivine in Guillaumes. This is in contrast to the allocthonous origin story for Ca and Mg in the sulfate minerals in Guillaumes (Tice et al., 2022) but is consistent with the redistribution of Mn during serpentinization at the Alfalfa abrasion site as described by Tosca et al. (2025), who observed enrichment of Mn in Fe-rich secondary silicates and magnetite (their Figs. 2–5), implying Mn mobilization from primary olivine.

OPEN IN VIEWER

FIG. 5.

Plot of crater floor scans with high Mn occurrence, demonstrating the different Mn/Fe ratios between the Máaz (‘o’ marker) and Séítah (‘x’ marker) formations.

6.3. Serpentinization and Mn mobility in the Ch’al member

At the Alfalfa abrasion target in the Ch’al member, multiple alteration episodes appear recorded in the mineral textures. Manganese was likely transferred from primary olivine to secondary serpentine and magnetite during an early stage of serpentinization, followed by a later sulfate overprint from oxidizing near-surface fluids. These two alteration stages are further evaluated below based on mineral textures, compositional zoning and terrestrial analogs.

6.4. Late-stage oxidizing alteration

SHERLOC identified Na-perchlorate in the bright-white void-filling material in Guillaumes, commonly associated with hydrated Ca and Mg sulfate (Farley et al., 2022; Corpolongo et al., 2023; Scheller et al., 2022). These white grains, seen by the SHERLOC scan (purple box, Fig. 2B), correspond to the white, void-filling, Cl-rich material in the adjacent PIXL scan (cyan box, Fig. 2B; Fig. 5). Elemental abundances show a Cl:Na molar ratio of ∼1.5:1 (Fig. 6B,C), consistent with Na-perchlorate (NaClO₄) due to oxygen’s influence on Cl fluorescence shifting the trend toward a Cl:Na ratio of ∼1.5 in laboratory studies (Gellert et al., 2022; Tice et al., 2022). However, we cannot rule out the presence of multiple chlorine phases, including halite or other oxychlorines, such as chlorate (Mitra, 2025). For simplicity, we refer to these phases collectively as “perchlorate” in the discussion that follows.

OPEN IN VIEWER

FIG. 6.

(A) Merged MCC and WATSON image of the Guillaumes abrasion patch with overlaid PIXL footprint. Grains with Cl > 5 wt% are shown in dark yellow. (B) The high Cl PMCs plotted in dark yellow on a ternary with FeT, Cl, and Na in mmol/g, showing Cl:Na molar ratio of 3:2. (C) The high Cl PMCs plotted in dark yellow on a ternary with SiO₂, Cl, and Na in mmol/g, showing Cl:Na molar ratio of 3:2. In panels B and C, white points represent PMCs with low Cl.

Given the identification of perchlorate by SHERLOC and SuperCam, and the widespread detection of perchlorates across Mars (Mitra, 2025; Clark and Kounaves, 2016), it is likely the Na-Cl phase in the bright white material in Guillaumes is Na-perchlorate. The occurrence of sulfates at the boundaries of the same white grains (Corpolongo et al., 2023) and minor sulfate peaks within SHERLOC perchlorate spectra indicate that these phases are closely associated, if not contemporaneous. Raman spectra from Bellegarde (in the Rochette member of Máaz [see Table 1]) also indicate the presence of sodium perchlorate (Wiens et al., 2022), which further supports that perchlorate formation occurred broadly across the Máaz formation.

The perchlorates in Guillaumes likely represent a late-stage phase of near-surface oxidation and brine evaporation. These salts may have precipitated from later fluids, distinct from those responsible for earlier sulfate deposition, as proposed for Séítah (Scheller et al., 2021). Although despujolsite and Na-perchlorate occur in close proximity within Guillaumes, their relationship is perhaps sequential rather than contemporaneous, given the possible hydrothermal origin of the despujolsite.

The Mn/Fe ratios across Jezero’s crater floor show greater Mn enrichment in Máaz (Guillaumes and Alfalfa abrasions) than in Séítah (Dourbes and Quartier abrasions) (Fig. 5). This enrichment could reflect differences in host mineralogy, fluid chemistry, redox/pH conditions, or the extent of hydrothermal alteration. In Séítah, alteration occurred in at least two stages, broken up into an early meteoric or shallow aquifer episode followed by brine infiltration at low water-to-rock ratios (Tice et al., 2022; Scheller et al., 2022). In contrast, Máaz records a more complex alteration history. Its carbonation did not form as many intergranular carbonates as in Séítah, but instead produced Fe silicates consistent with serpentinization, suggesting that the fluids interacting with the two formations were distinct. Both units, however, record late-stage sulfate deposition that likely occurred under similar oxidizing conditions. The Mn-rich sulfates in Máaz may have formed from high-temperature, low water-to-rock fluids or hypersaline brines similar to those inferred for Séítah.

Guillaumes is in a lower stratigraphic unit (Roubion), and Alfalfa is in the highest (Ch’al) (Fig. 1B), yet the extensive secondary mineralization seen in the bottom of the Máaz formation is not present in the top of the formation. Therefore, Máaz likely experienced multiple phases of aqueous alteration, with an early episode of serpentinization and a later acidic hydrothermal or alkaline lacustrine alteration. If serpentinization occurred in both members simultaneously, it likely preceded sulfate formation, since serpentinizing conditions could have altered Mn-bearing precursors. Serpentinization of ferroan olivine was likely the earliest alteration, producing pore space in the altered olivine for later sulfate deposition. The absence of serpentine minerals in terrestrial despujolsite analogs suggests serpentinization is not a prerequisite for despujolsite formation, which implies different serpentinization histories in Guillaumes and Alfalfa.

The Ca-Mn sulfate assemblage in Ch’al and the despujolsite in Guillaumes could reflect distinct but overlapping alteration stages. Both require acidic, high-temperature environments, but Mn²⁺ stabilizes in the reducing to mildly oxidizing waters that formed the Ca-Mn sulfate assemblage in Alfalfa, while Mn⁴⁺ in despujolsite demands stronger oxidation in Guillaumes. Additionally, the lower degree of secondary mineralization in Alfalfa suggests that the conditions that formed despujolsite in Guillaumes could be separate from the conditions that formed the Ca-Mn sulfate assemblage in Alfalfa. However, the acidic oxidizing conditions required to form the Ca-Mn sulfate assemblage in Alfalfa are similar to the acidic hydrothermal environment that could form despujolsite in Guillaumes.

The final stage of alteration in Máaz was the emplacement of Cl-bearing phases. Oxychlorine salts, identified in both Guillaumes and Alfalfa, likely represent near-surface oxidation or evaporitic concentration that postdated the sulfate-forming episodes.

Stratigraphic observations suggest a complex timing of these events. Roubion shows substantial weathering, while Ch’al does not appear to underlie the delta, which implies a time gap between their emplacements that possibly places lake activity before Ch’al emplacement (Horgan et al., 2023; Crumpler et al., 2023). Additionally, there is evidence that the Máaz formation experienced tilting after emplacement, potentially due to the uplift of Séítah (Horgan et al., 2023). Stratigraphically coherent serpentinization in Roubion and Ch’al, coupled with their shared tilt, favors alteration prior to uplift, although localized, post-uplift serpentinization restricted to Máaz due to localized fluid flow or lithologic controls cannot be ruled out. These observations support two potential alteration scenarios:

7. Conclusion

We report the first tentative identification of despujolsite, Ca₃Mn⁴⁺(SO₄)₂(OH)₆ · 3H₂O, on Mars, found in the Guillaumes abrasion patch in the Máaz basalts, on the floor of Jezero crater. The mineral phase was deduced by elemental chemistry from the PIXL instrument on the Perseverance rover. We also report Mn-bearing Ca sulfate-rich PMCs in the Alfalfa abrasion patch, also in the Máaz basalts, consistent with either minor Mn incorporation into gypsum or anhydrite or mixing with adjacent Mn-rich phases. The possible presence of these minerals, combined with manganese-enriched magnetite and Cl-bearing phases, suggests a multi-stage geochemical history of alteration in the Máaz formation. The Jezero crater floor records a dynamic environment where serpentinized ultramafic rocks have been overprinted by near-surface oxidative, acidic conditions, generating sulfate minerals, and possibly evaporitic or hydrothermal fluids introducing Cl-phases.

Although the Guillaumes sample core was lost, there may be occurrences of despujolsite in other returned samples from Máaz, which can enable critical insights into the formation, environment, and habitability of Jezero. Detailed sample return analyses such as radiometric dating to understand depositional history, carbon isotope analysis to search for biosignatures, and mineral identification to distinguish between formation hypotheses presented in this study could refine the emplacement history of the Máaz formation and constrain the age of the paleolake. Thus, returning these samples could refine our knowledge of the aqueous history of Mars and bring us one step closer to the search for life in our solar system.

Authors’ Contributions

K.P.S.: Conceptualization, formal analysis, investigation, data curation, visualization, writing—original draft, writing—review and editing. B.C.C.: Investigation, formal analysis, writing—review and editing. M.W.M.J.: Formal analysis, visualization, writing—review and editing. D.C.C.: Conceptualization, supervision, writing—review and editing. W.T.E.: Methodology, supervision, writing—review and editing. Y.L.: Conceptualization, writing—review and editing.