Sulfites and sulfates formed by weathering of early martian carbonates in a sulfur dioxide-bearing atmosphere Authors: V. F. Chevrier, C. Gil-Lozano, E. Dehouck, S. T. Altheide Published: 2024 ABSTRACT Despite ample evidence that extensive water–rock interactions occurred under a CO2-dominated atmosphere on early Mars, carbonate minerals are relatively rare at the surface. One possibility to explain this scarcity is that carbonates were initially abundant, but were later destroyed when atmospheric conditions changed, particularly as a result of volcanism releasing large volumes of sulfur dioxide SO2 into the atmosphere. However, despite some early theoretical and experimental results, no study has investigated the stability of the most common carbonates (Ca, Mg and Fe) in the presence of abundant SO2 gas. Here we present the results of experiments demonstrating that carbonates are systematically unstable when exposed to 0.8 bar of SO2 in moderately oxidizing (SO2 + H2O) or strongly oxidizing (SO2 + H2O + H2O2) environments. In both environments, the reaction end products are systematically sulfates, except for calcium carbonate, which predominantly transforms into calcium sulfite (hannebachite) in moderately oxidizing conditions. Based on these results, carbonates formed early in martian history would have been rapidly decomposed and replaced by sulfates (and sulfites if calcium was abundant enough) in the presence of abundant SO2 released by major volcanic events such as those associated with the build-up of the Tharsis rise. Introduction The surface of Mars, although largely dominated by basaltic materials, exhibits a wide variety of secondary phases, resulting from various processes of alteration under different conditions at different times1. It is generally assumed that early Mars was covered by a thick CO2atmosphere, which appears necessary to maintain warm conditions through greenhouse effect and allow liquid water for long periods of time2,3. Such CO2-dominated atmosphere should have eventually led to thick and ubiquitous layers of carbonates. However, carbonate deposits are limited to only a few small spots on the surface of Mars4–8. Thus, several hypotheses have been put forth to explain this apparent discrepancy. In the first hypothesis, Mars never had a thick atmosphere of CO2 and possibly remained cold or was warmed by other greenhouse gases9–11. Even in the presence of a thick CO2 atmosphere, the precipitation of carbonates could have been limited in anoxic environments, either by very high supersaturation thresholds impossible to achieve at high water-to-rock ratios12, by the mineral composition of the host rock13, or because of the lower solubility of sulfites (SO32-) compared to carbonates14,15. Finally, another hypothesis is carbonates that initially precipitated under a thick CO2 atmosphere or through other processes like hydrothermalism16, could have been destroyed later on by acidic conditions resulting from intense volcanism. In this paper, we investigate the stability of early martian carbonates against a later episode of SO2-rich volcanism. In our previous work on the weathering of the early martian surface, we studied the effect of a thick CO2 atmosphere in contact with iron phases like magnetite and pyrrhotite17,18, and on mixtures of sulfides and silicates19. These experiments were carried out in moderately (H2O) or strongly (H2O + H2O2) oxidizing environments. Previous experiments with SO2 have only focused on the relative abundance of SO2 compared to CO2 and under oxygen-free conditions14. These experiments showed that even low concentrations of SO2 hinder the precipitation of calcium carbonate in favor of calcium sulfite (hannebachite). Here, we present the results of experiments in which we evaluated the stability of various carbonates (calcium-, iron- and magnesium-bearing) relevant to early Mars in the presence of a SO2-rich atmosphere in moderately (H2O) or strongly oxidizing (H2O + H2O2) conditions (see methods section). Results All the phases altered in the SO2 + H2O atmosphere showed significant transformation after two years, as well as different levels of “wetness” or liquid water saturation. The original sample of hydromagnesite appeared completely liquefied in the SO2 + H2O atmosphere, while the siderite sample showed abundant liquid and the calcite sample was mostly dry (Fig. 1). By comparison, the samples altered in the SO2 + H2O + H2O2 atmosphere appeared completely dry and showed low abundances of secondary products, although more oxidized. Table 1 shows the secondary products identified in the samples.Fig. 1Photographs of the three carbonate samples (initials: A, D, G) tested in our experiments after two years of weathering in 0.8 bar SO2 atmospheres under moderately (B, E, H: H2O alone) or strongly oxidizing (C, F, I: H2O + H2O2) conditions. A. Initial calcite (CaCO3). B. Calcite in SO2 + H2O. C. Calcite in SO2 + H2O + H2O2. D. Initial siderite (FeCO3). E. Siderite in SO2 + H2O. F. Siderite in SO2 + H2O + H2O2. G. Initial hydromagnesite (Mg5(CO3)4(OH)2.4H2O). H. Hydromagnesite in SO2 + H2O. I. Hydromagnesite in SO2 + H2O + H2O2.Table 1Summary of secondary phases identified in our experiments, after two years of exposure to SO2 + H2O or SO2 + H2O + H2O2.Initial mineralsSO2 + H2OSO2 + H2O + H2O2CalciteHannebachiteGypsumSideriteLepidocrociteGoethiteMelanteriteRozeniteButleriteGypsumHexahydriteNesquehoniteFe (oxy)hydroxidesHydromagnesiteKieseritePentahydriteHexahydriteMg-sulfate Photographs of the three carbonate samples (initials: A, D, G) tested in our experiments after two years of weathering in 0.8 bar SO2 atmospheres under moderately (B, E, H: H2O alone) or strongly oxidizing (C, F, I: H2O + H2O2) conditions. A. Initial calcite (CaCO3). B. Calcite in SO2 + H2O. C. Calcite in SO2 + H2O + H2O2. D. Initial siderite (FeCO3). E. Siderite in SO2 + H2O. F. Siderite in SO2 + H2O + H2O2. G. Initial hydromagnesite (Mg5(CO3)4(OH)2.4H2O). H. Hydromagnesite in SO2 + H2O. I. Hydromagnesite in SO2 + H2O + H2O2. Table 1Summary of secondary phases identified in our experiments, after two years of exposure to SO2 + H2O or SO2 + H2O + H2O2.Initial mineralsSO2 + H2OSO2 + H2O + H2O2CalciteHannebachiteGypsumSideriteLepidocrociteGoethiteMelanteriteRozeniteButleriteGypsumHexahydriteNesquehoniteFe (oxy)hydroxidesHydromagnesiteKieseritePentahydriteHexahydriteMg-sulfate Summary of secondary phases identified in our experiments, after two years of exposure to SO2 + H2O or SO2 + H2O + H2O2. Lepidocrocite Goethite Melanterite Rozenite Butlerite Gypsum Hexahydrite Nesquehonite Kieserite Pentahydrite Hexahydrite Mineralogically, the calcite sample (Figs. 1A, 2A) altered in the SO2 + H2O atmosphere was composed almost exclusively of hannebachite (Figs. 1B, 2B), the Ca-sulfite of formula CaSO3.1/2H2O. Hannebachite accounts for ~ 97% of the XRD signal, along with minor gypsum ~ 3%; The initial calcite was not detectable anymore in the sample. In contrast, the sample altered in SO2 + H2O + H2O2 (Figs. 1C, 2C) shows significant amounts of gypsum, which represents about 40% of the sample’s XRD signal, with the remaining 60% being the initial calcite.Fig. 2X-Ray diffraction spectra for calcite, siderite, and hydromagnesite samples exposed to a SO2 atmosphere in the presence of H2O or H2O + H2O2. A. Initial Ca-carbonate (calcite). B. Ca-carbonate sample altered in SO2 + H2O. C. Ca-carbonate sample altered in SO2 + H2O + H2O2. D. Initial Fe-carbonate (siderite). E. Solid residue of Fe-carbonate sample altered in SO2 + H2O. F. Dried liquid brine from Fe-carbonate sample altered in SO2 + H2O. G. Initial Mg-carbonate (hydromagnesite). H. Dried liquid brine after Mg-carbonate alteration in SO2 + H2O. X-Ray diffraction spectra for calcite, siderite, and hydromagnesite samples exposed to a SO2 atmosphere in the presence of H2O or H2O + H2O2. A. Initial Ca-carbonate (calcite). B. Ca-carbonate sample altered in SO2 + H2O. C. Ca-carbonate sample altered in SO2 + H2O + H2O2. D. Initial Fe-carbonate (siderite). E. Solid residue of Fe-carbonate sample altered in SO2 + H2O. F. Dried liquid brine from Fe-carbonate sample altered in SO2 + H2O. G. Initial Mg-carbonate (hydromagnesite). H. Dried liquid brine after Mg-carbonate alteration in SO2 + H2O. The NIR spectrum of the calcite sample altered in SO2 + H2O shows absorption bands at 1.47, 1.83, 1.98, 2.13, 2.47 and 2.57 µm, which matches the spectrum of hannebachite (Fig. 3A,B). The NIR spectrum of the calcite sample altered in SO2 + H2O + H2O2 displays absorption bands at 1.44, 1.75, 1.94, 2.21 µm, as well as a reflectance drop longward of ~ 2.4 µm, which closely matches a reference spectrum of pure gypsum (Fig. 3A,B). The small band located at 2.34 µm is likely due to remaining calcite.Fig. 3Near-infrared reflectance spectra for calcite, siderite, and hydromagnesite exposed to SO2 atmosphere in the presence of H2O or H2O + H2O2. All the scale bars correspond to 20% reflectance A. Initial Ca-carbonate (calcite, black, labelled “initial”), compared to solid residue after exposure to SO2 + H2O atmosphere (blue, labelled “H2O”) and solid residue after exposure to SO2 + H2O + H2O2 atmosphere (green, labelled “H2O2”). B. Reference spectra of calcite (RELAB database #LACB12B); hannebachite (RELAB database #BIR1LH058) and gypsum (USGS database #HS333.3B). C. Initial Fe-carbonate (siderite, black, labelled “initial”), compared to solid residue after exposure to SO2 + H2O atmosphere (blue, labelled “H2O-solid”), dried liquid from sample exposed to SO2 + H2O atmosphere (blue, labelled “H2O-brine”) and solid residue after exposure to SO2 + H2O + H2O2 atmosphere (green, labelled “H2O2”). D. Reference spectra of siderite (RELAB database #F1CC07B); melanterite (RELAB database #LASF44A); rozenite (RELAB database #BKR1JB626B); butlerite (USGS database #GDS25); gypsum (USGS database #HS333.3B); hexahydrite54; nesquehonite19. E. Initial Mg-carbonate (hydromagnesite, black, labelled “initial”) compared to dried liquid from sample exposed to SO2 + H2O atmosphere (blue, labelled “H2O-brine”) and solid sample exposed to SO2 + H2O + H2O2 atmosphere (green, labelled “H2O2”). F. Reference spectra of hydromagnesite (RELAB database #CACB35); starkeyite, pentahydrite, hexahydrite and epsomite54. Near-infrared reflectance spectra for calcite, siderite, and hydromagnesite exposed to SO2 atmosphere in the presence of H2O or H2O + H2O2. All the scale bars correspond to 20% reflectance A. Initial Ca-carbonate (calcite, black, labelled “initial”), compared to solid residue after exposure to SO2 + H2O atmosphere (blue, labelled “H2O”) and solid residue after exposure to SO2 + H2O + H2O2 atmosphere (green, labelled “H2O2”). B. Reference spectra of calcite (RELAB database #LACB12B); hannebachite (RELAB database #BIR1LH058) and gypsum (USGS database #HS333.3B). C. Initial Fe-carbonate (siderite, black, labelled “initial”), compared to solid residue after exposure to SO2 + H2O atmosphere (blue, labelled “H2O-solid”), dried liquid from sample exposed to SO2 + H2O atmosphere (blue, labelled “H2O-brine”) and solid residue after exposure to SO2 + H2O + H2O2 atmosphere (green, labelled “H2O2”). D. Reference spectra of siderite (RELAB database #F1CC07B); melanterite (RELAB database #LASF44A); rozenite (RELAB database #BKR1JB626B); butlerite (USGS database #GDS25); gypsum (USGS database #HS333.3B); hexahydrite54; nesquehonite19. E. Initial Mg-carbonate (hydromagnesite, black, labelled “initial”) compared to dried liquid from sample exposed to SO2 + H2O atmosphere (blue, labelled “H2O-brine”) and solid sample exposed to SO2 + H2O + H2O2 atmosphere (green, labelled “H2O2”). F. Reference spectra of hydromagnesite (RELAB database #CACB35); starkeyite, pentahydrite, hexahydrite and epsomite54. The siderite sample (Figs. 1D, 2D) that was exposed to the SO2 + H2O atmosphere was separated into two fractions: solid residue and liquid phase (brine, Fig. 1E). XRD of the solid residue shows siderite and the iron (oxy)hydroxides lepidocrocite and goethite (Fig. 2E). XRD spectra of the sample crystallized from the brine shows a complex mixture that cannot be resolved with a single phase (Fig. 2F). The XRD pattern shows peaks of various ferrous (melanterite FeIISO4·7H2O and rozenite FeIISO4·4H2O) and ferric (butlerite FeIII(SO4)(OH)·2H2O) sulfate phases. Moreover, we also identify calcium sulfate (gypsum), magnesium sulfate (hexahydrite MgSO4·6H2O), and magnesium carbonate (nesquehonite MgCO3·3H2O) in the sample, likely resulting from the presence of Ca and Mg in the initial siderite sample, either as substitutions for Fe and/or as accessory minerals (calcite, ankerite; Fig. 2D). Finally, some peaks remain unidentified in the dried brine sample, particularly a high intensity peak at 2θ (CoKalpha)≈19.5º (~5.26 Å) and possible unassigned peaks at high 2θ values above 35º, suggesting the presence of “exotic” phases (possibly some Fe or Mg sulfites, which are not present in the reference XRD databases). Due to the lack of identification of all the crystalline phases present in the mixture, we could not perform any quantitative analysis on the abundance of phases. In complete contrast, the sample exposed to the SO2 + H2O + H2O2 atmosphere showed superficial discoloration (Fig. 1F) but the secondary phases responsible for it were formed in trace amounts and thus could not be identified with a bulk XRD analysis. The brownish color, however, indicates that this secondary product is likely some iron (oxy)hydroxide (possibly amorphous). The NIR spectra of the solid altered siderite samples are almost identical between both atmospheres, strongly resembling the initial siderite (Fig. 3C), except for the slightly deeper 1.9-µm band in the H2O2 sample, which suggests the formation of minor hydrated phases. In contrast, the spectrum of the crystallized brine extracted from the SO2 + H2O atmosphere is significantly different from the spectrum of the initial siderite and exhibits absorption bands near 1.4, 1.9 and 2.4 µm (Fig. 3C), indicating the presence of hydrated sulfates (Fig. 3D). The detection of additional absorption features at 1.73 and 2.24 µm suggests the presence of gypsum (which may have formed from minor amounts of Ca in the starting material), but other sulfate or sulfite species likely contribute to the overall spectrum too (Fig. 3D). The hydromagnesite sample (Fig. 1G) altered in the SO2 + H2O atmosphere was completely liquified (Fig. 1H). XRD analysis of the solid residue after drying the liquid sample shows that it is entirely composed of Mg-sulfates (Fig. 2H) with various hydration states, mostly pentahydrite (MgSO4·5H2O), kieserite (MgSO4·H2O), and hexahydrite (MgSO4·6H2O), in order of decreasing abundance. The NIR spectrum is consistent with this interpretation, although the best match given the exact position of the observed bands (1.50 and 1.95 µm) is epsomite (MgSO4.7H2O, Fig. 3E,F). Interestingly, we found no evidence of magnesium sulfite in the resulting salt mixture, either by XRD or NIR spectroscopy. Finally, the hydromagnesite sample exposed to the SO2 + H2O + H2O2 atmosphere was visually unchanged (Fig. 1I) but clearly exhibits a Mg-sulfate signature in the NIR data (Fig. 3E,F). The overall spectrum appears similar to the SO2 + H2O counterpart, suggesting similar highly hydrated Mg-sulfates, most likely pentahydrate or hexahydrite. Discussion There are two interesting effects that our study addresses. The first effect is the water activity in the initial liquid at the bottom of the desiccator in equilibrium with the atmosphere. The presence of 30% H2O2 in the liquid reduces the water activity of the solution (for a 30 wt% H2O2 solution, the water activity is around 0.78), and therefore the relative humidity in equilibrium with the liquid phase. This explains why the samples in SO2 + H2O + H2O2 exhibit a lower abundance of secondary phases but a generallya higher oxidation state. In contrast, the samples in SO2 + H2O appear saturated with liquid water (since the relative humidity in the desiccator is close to 100%), but less oxidized. The low relative humidity in the H2O2 atmosphere prevents water from condensing at the surface of the samples (besides adsorption), and therefore the reactions were akin to heterogeneous solid – gas. Conversely, in the SO2 + H2O atmosphere, the relative humidity remains much higher, and water condensed on the surface of the samples, which did not evaporate due to the formation of hygroscopic salts and brines on the surface of the samples. Reaction with the carbonates created a thin liquid film on their surface. The solubility of the resulting salts (sulfates or sulfites) in turn lowered the water activity of the liquid films and increased their hygroscopic behavior allowing them to absorb more water vapor over time. Therefore, in the SO2 + H2O environment, the relative humidity above the liquid remains higher than the water activity of the films formed in contact with the carbonates. In the SO2 + H2O + H2O2 atmosphere, the situation is opposite because the relative humidity above the liquid mixture is lower than the water activity of the brine films. For example, in the SO2 + H2O atmosphere, the Mg-carbonate essentially leads to the formation of Mg-sulfates. At our experimental temperature, the water activity of the saturated MgSO4brine is ~ 0.85–0.9020 which is lower than the relative humidity above the pure water. Therefore, the brine acts as a sink for water vapor, allowing the film to grow and the carbonate to be progressively dissolved. Yet this same water activity is higher than the water activity of the H2O + H2O2 solution (0.78), which then prevents any liquid from being stable in contact with the carbonate. Thus, these samples remain completely dry. However, direct oxidation by H2O2 vapor still occurs. The second effect is due to the change of redox conditions between a strongly oxidizing atmosphere (SO2 + H2O + H2O2) and a moderately oxidizing atmosphere (SO2 + H2O). The most interesting result regarding the effect of redox conditions is the transformation of calcite into the calcium sulfite phase hannebachite in the SO2 + H2O atmosphere, while gypsum forms in the more oxidizing atmosphere. Indeed, the simple reaction of SO2 and water forms H2SO3 (sulfurous acid which does not exist in solution and decomposes into H+ and HSO3-), while the formation of sulfate first requires the oxidation of SO2 into SO3, followed by reaction with water to form H2SO4. Thus, our results confirm previous experimental observations by other groups that, in the presence of SO2, hannebachite is more stable than calcite14 but also demonstrate that calcite can be directly replaced by hannebachite through pseudomorphosis, due to the lower solubility of the sulfite compared to the carbonate. This explains why the sample remained dry because the low solubility of the compounds (Ca-sulfite and Ca-carbonate) did not affect the water activity in contact with the grains, and therefore liquid brine films could not grow through deliquescence. The resulting hannebachite appears quite resistant to further oxidation since we observed only minor amounts of gypsum even after two years in the same atmosphere. Therefore, our results show that carbonates destabilize without the need for highly acidic conditions. Even though sulfurous acid (H2SO3) is a weak acid compared to sulfuric acid, the complete transformation of calcite into hannebachite is likely due to a mild acidic attack on the carbonate combined with the lower solubility of hannebachite compared to calcite. Interestingly, this reaction occurs much faster than the oxidation of sulfite into sulfate. The alteration of siderite in the SO2 + H2O atmosphere also shows interesting results with the presence of both ferric and ferrous sulfates, and possibly sulfites, suggesting that significant oxidation of Fe2+ into Fe3+ and sulfite into sulfate occurred in the liquid part of the sample. However, siderite in the SO2 + H2O + H2O2 atmosphere did not transform into sulfate, contrary to what we observe with the calcite sample. This is likely due to the difference in solubility of the resulting products. Gypsum is quite insoluble and will immediately precipitate even in a low humidity atmosphere. In comparison, ferric (or ferrous) sulfates are highly soluble21–23, while iron (oxy)hydroxides are extremely insoluble and will immediately precipitate out of solution24. This will act as a Fe3+ trap, thus hindering the precipitation of ferric sulfates. The presence of Fe (oxy)hydroxides in the peroxide-bearing environment also indicates a higher pH resulting from the very limited amounts of reactant in contact with the carbonate. In other words, the “liquid” films in contact with siderite do not develop enough due to the low water activity imposed by the H2O + H2O2 solution and are instantly buffered by the carbonate. Previous studies have similarly observed that jarosite formation from acidic alteration of ferrihydrite is only possible if the transformation into goethite is inhibited25. The precipitation of excess Fe3+ into (oxy)hydroxides in the residual solid phase of the SO2 + H2O experiment (Fig. 2E) would also explain the relative abundance of Ca- and Mg-sulfates in the crystallized brine. The interpretations above about the effect of oxidation and water activity are also confirmed by the results we obtained for hydromagnesite. In the SO2 + H2O + H2O2 atmosphere, no complete transformation was observed because the dry atmosphere prevented liquid from condensing on, and reacting with, the sample. However, in the SO2 + H2O atmosphere, the sample was completely liquefied and fully oxidized. The complete absence of sulfite in the hydromagnesite sample altered in SO2 + H2O contrasts with the presence of hannebachite in the calcite sample. The simplest explanation for this observation is that the oxidation of magnesium sulfite proceeds at a much faster rate in the liquid phase compared to heterogeneous oxidation of the solid (and very insoluble) hannebachite. This result is akin to the abundant oxidation products that we observed on the siderite sample in the SO2 + H2O atmosphere. Our experiments were done with a high partial pressure of SO2 (0.8 bar). Such SO2 abundance is unlikely to have occurred on Mars, even in periods of high volcanism. Typical concentrations on early Mars are usually estimated to range from 10 to 10,000 ppm26. Moreover, the partial pressure of CO2 could have been much higher, possibly as high as several bar3. Therefore, to test if our experimental results would still be valid in such conditions, we constructed a phase diagram for calcium phases as a function of the partial pressures of CO2 and SO2. For most atmospheric compositions in SO2 and CO2, hannebachite is the thermodynamically stable phase (Fig. 4A). Even at 1 bar of CO2, hannebachite becomes stable over calcite at SO2 concentrations above ~1 ppm. We also made a similar plot including gypsum (Fig. 4B) as a function of the CO2 partial pressure and concentration of sulfate in the solution. Calcite is only stable at high CO2 partial pressures, above ~12 bar, while gypsum is only stable at high sulfate concentrations (0.1 mol/L). Therefore, hannebachite appears as the most stable phase compared to other calcium secondary minerals. However, this situation is only valid in reducing or weakly oxidizing conditions. Our experiments show that hannebachite becomes unstable with respect to gypsum in strongly oxidizing conditions.Fig. 4Thermodynamic phase diagrams for calcium phases (sulfite, sulfate and carbonate) in aqueous solution. A. Partial pressure of CO2 versus partial pressure of SO2. The concentration of Ca2+ is 0.01 mol/L, the temperature is 25 °C, the total pressure 1 bar, and the pH is 7. B. Partial pressure of CO2 versus concentration of sulfate SO42- in solution at a SO2 partial pressure of 10-6 bar (1 ppm), same conditions as figure A except the pH set at 6. We used the Geochemist’s Workbench software package and the Thermo_Minteq database, which contains solubility constants for hannebachite (CaSO3.1/2H2O). Thermodynamic phase diagrams for calcium phases (sulfite, sulfate and carbonate) in aqueous solution. A. Partial pressure of CO2 versus partial pressure of SO2. The concentration of Ca2+ is 0.01 mol/L, the temperature is 25 °C, the total pressure 1 bar, and the pH is 7. B. Partial pressure of CO2 versus concentration of sulfate SO42- in solution at a SO2 partial pressure of 10-6 bar (1 ppm), same conditions as figure A except the pH set at 6. We used the Geochemist’s Workbench software package and the Thermo_Minteq database, which contains solubility constants for hannebachite (CaSO3.1/2H2O). Other processes might limit the conversion of carbonates into sulfites / sulfates. The formation of passivation layers, particularly on solid carbonate deposits (contrary to our loose powders), can reduce dissolution rates depending on the porosity of the coating. Previous studies show that gypsum precipitation limits H⁺ diffusion to the calcite surface, slowing its dissolution27,28. On Mars, spectral signatures of carbonates associated with sulfates have been identified in weathering profiles29. The persistence of these carbonates may be influenced by factors such as the diffusion of S-rich fluids30 or the alkalinity of the host silicates. Indeed, early martian carbonates, like those found at Jezero Crater31,32 or in the Noachian orthopyroxenite ALH8400133,34 are often associated with mafic silicates with high natural alkalinity. However, our results show that in humid environments (SO₂ + H₂O), even with the formation of hannebachite as a potential passivation layer, complete conversion of calcite occurs within two years. Over geological timescales, this process likely affected carbonate layers in a similar way. Furthermore, any remaining carbonates would likely be obscured by secondary products, making detection by remote sensing instruments difficult. Implications for Mars’ carbonates Based on our results, the conspicuous absence of large Noachian carbonate deposits on the martian surface could be explained by the release of SO2 during important or global volcanic episodes such as the formation of the Tharsis dome, which would have subsequently and rapidly erased these carbonate deposits, even in reducing conditions. Although the presence of massive carbonate deposits would typically be the result of a warm early Mars and a dense CO2atmosphere35, recent efforts in climate modelling have rather suggested that Mars remained a cold planet even early in its history36. In this scenario, warming related to SO2would have been likely more episodic, and the result of multiple events, in an overall cold martian environment37–39. This scenario would better explain the preservation of carbonate deposits in locations such as Nili Fossae40, Gale Crater41,42and Jezero crater31,32,43. The high solubility of SO2 in liquid water would have prevented significant transport to regions located far from any active volcano or vent and resulted in a more heterogeneous alteration of carbonates. On a global scale, carbonates are typically associated with phyllosilicates44. However, several studies have noticed local associations between phyllosilicates and sulfates45–47. Those could be remnants of former carbonate and phyllosilicate parageneses that were further weathered by SO2-enriched solutions30. Phyllosilicates would have survived because they are significantly more resistant to acidic alteration compared to carbonates48. The continuous or sporadic alteration of carbonates, depending on the volumes affected, would have resulted in the release of significant CO2 back into the atmosphere. Therefore, our experiments do not provide a direct answer for the “missing carbonate” problem. However, our results show that even in a low pressure CO2 atmosphere, carbonates formed by more “local” processes such as hydrothermalism16,49, could be destabilized by episodic SO2 release. Alternatively, carbonates buried deeper in the subsurface50,51 could remain unaffected because highly reactive SO2-rich solutions would likely get neutralized by the layers of regolith above. Even if the overall conditions on early Mars remained cold and dry36, the formation of concentrated brines from carbonates, with lower freezing temperatures, would have allowed alteration to proceed at temperatures below 0 °C10. However, our results also show that such a dry environment would have resulted in a more limited or much slower alteration process. In our experiments in H2O2 (dry environment), alteration can proceed without significant involvement of liquid water, although at a much slower rate than in a humid environment. Thus, the conditions on early Mars might have alternated between our “humid” model in the SO2 + H2O atmosphere and our “dry” model in the SO2 + H2O + H2O2 atmosphere. Conclusions We presented new experimental results on the alteration of carbonates relevant to the martian surface in the presence of SO2 and a moderately (H2O) or strongly oxidizing (H2O + H2O2) atmosphere. The main conclusion of this study is that carbonates are systematically unstable in the presence of SO2 in moderately or strongly oxidizing atmospheres. They rapidly transform into sulfur-containing species and oxides (for iron carbonate). Our results show that this alteration process occurs rapidly in wet, moderately oxidizing conditions. Considering that early martian conditions were likely reducing or at least not strongly oxidizing11,52, this scenario is quite plausible. In our experiments, the only sulfite we observed was calcium-bearing hannebachite and only in the moderately oxidizing environment. We have not observed iron or magnesium sulfites, as the end-products are systematically sulfates. Carbonate minerals detected by rover missions so far are mainly magnesium and iron bearing16,32, but if calcium-rich carbonates formed elsewhere on Mars, the identification of calcium sulfites would be an excellent marker of moderately acidic and moderately oxidizing conditions on early Mars, compared to later highly acidic and highly oxidizing conditions that dominated during sulfate formation. Interestingly, the accumulation of SO2 in the Venusian atmosphere could also have destabilized carbonates there53, although the absence of water would have likely been a limiting factor. Methods We selected as initial samples three different carbonates likely to have been present on early Mars: calcite (CaCO3), siderite (FeCO3), and hydromagnesite (Mg5(OH)2(CO3)4·4H2O). Siderite is a natural sample from St. Pons Mine, Barcelonnette, Alpes de Haute Provence, France, and contains some impurities such as ankerite, calcite and quartz. The other two carbonates are synthetic, purchased from Mallinckrodt chemicals (catalogue #4071–12, lot C23585) for calcite and J. T. Baker (sold as magnesium carbonate n-hydrate, catalogue #2437–01, lot C44H05) for hydromagnesite. For the experimental set-up, three grams of finely powdered mineral (sieved to ≤ 60 µm) were put in individual petri dishes. Since the duration of the experiments (2 years) is relatively short compared to the intermittent periods of SO₂ volcanism on Mars, which could have lasted from decades to centuries 37, we used a small grain size for the starting minerals to increase their surface area and promote a higher dissolution rate. The petri dishes were placed in a desiccator, the bottom of which was filled with 1L of either pure DI water or DI water containing 33% hydrogen peroxide (H2O2); hydrogen peroxide was chosen because of its high oxidation potential, higher than the redox couple H2O / O2, and the fact that its decomposition products are oxygen and water, so they would not introduce more ions in the system and complicate the results. The samples were not in direct contact with the liquid. Both desiccators (1 per atmosphere with all three samples) were then equilibrated with gaseous SO2 at an initial pressure of 0.8 bar and the temperature was maintained in the range 15–20 °C. Such conditions were used to increase the rates of weathering, even if it is unlikely that Mars had 0.8 bar of SO2 at the global scale; however, such high pressure could model local conditions in the vicinity of large volcanoes or fumarolic activity (the high solubility of SO2 in liquid water likely prevented it from diffusing very far from the source). Pressure was controlled every 2–3 months and SO2 was reinjected if necessary. The desiccators were opened after two years of reaction and the petri dishes were weighted. Powders in the H2O + H2O2 desiccator remained dry during the entire experiment while powders in the H2O desiccator became progressively wetter, and some were entirely liquid by the end of the two-year long experiment. In this case, the solid residue was separated from the liquid by centrifugation, followed by recrystallization of a fraction of the liquid phase at 50 °C. All solid products were analyzed using X-ray diffraction (XRD) and near-infrared reflectance spectroscopy (NIR). Powder X-ray Difracction (PXRD) was performed using a Terra portable X-ray diffraction/XRF instrument (CheMin prototype) with Co Kα radiation (λ = 1.789 Å). XRD patterns were acquired from 5 to 55° (2 theta) with a step size of 0.5°. NIR spectra were acquired using a Thermo Nicolet 6700 Fourier Transform spectrometer, equipped with a tungsten source, a CaF2 beamsplitter and a TECD detector, allowing a spectral range from 0.8 to 2.6 µm. We systematically acquired 400 spectra per sample that were averaged to improve the signal to noise ratio.