Volcanic emission of reduced sulfur species shaped the climate of early Mars

Abstract

Sulfur and other volatiles could be transported from the martian interior to surface through magmatic processes, including mantle melting, magma differentiation, and degassing. However, these processes were not fully integrated in past sulfur cycling models because of complexity from the gas-melt interactions in chemically and dynamically evolving magmatic systems with multicomponent volatiles. Here, we incorporate these processes to simulate how sulfur, carbon, and hydrogen degas from martian melts. We find that reduced sulfur species, H₂S and S₂, are dominantly emitted through degassing at crustal to surficial pressures. These sulfur species could condense as sulfide and elemental sulfur, potentially yielding the sulfate deposits observed on the martian surface through secondary oxidation. Our models also show that evolved magmas reach graphite and sulfide saturation at crustal pressures and thus may establish sulfur and carbon reservoirs in the martian crust. The degassed H₂S and S₂ may form a hazy atmosphere with SF₆, a potent greenhouse gas, to shape the paleoclimate of Mars.

Reduced sulfur species are primarily degassed from evolving martian melts, inducing a hazy, greenhouse atmosphere on early Mars.

INTRODUCTION

Prominent observations of fluvial activity on early Mars suggest a climate condition with surface temperatures warm enough to enable the sustained presence of liquid water ~3 to 4 billion years ago (Ga) (1). Yet, the composition (2–4), duration (5), and pressure (6, 7) of the martian greenhouse gas paleoatmosphere remain largely unknown. With widespread sedimentary sulfate deposits on Mars (8–11) and abundant sulfide in martian meteorites (12–15), sulfur cycling is likely a critical process affecting the martian paleoclimate (16), which is also documented by the mass-independent sulfur isotope anomalies in martian meteorites (17). Earlier models suggested that SO₂ could induce a greenhouse effect warm enough to sustain liquid water on Mars (18–20). This notion was supported by theoretical studies on volcanic degassing of C-H-O-S volatiles (21, 22) that indicated SO₂ as a primary gas species near the surface. As SO₂ can extensively acidify water, such volcanic gases should inhibit carbonate formation on Mars. The consequential absence of carbonates, however, does not align with the growing body of carbonate observations on Mars (23–30), some of which are likely intermixed with sulfate (31). We note that previous studies may overestimate the volcanic SO₂ flux as they exclude (i) magma differentiation and (ii) saturation of sulfide and graphite. Incorporating these processes, we systematically assess the volcanic sulfur cycle and its role in shaping the climate of early Mars.

Given the recent observations of felsic through ultramafic rocks (32–34) on Mars, magma differentiation should be pervasive in the martian crust. As volatile elements (e.g., hydrogen, carbon, and sulfur) are generally incompatible in cumulate minerals, magma differentiation can concentrate them in the evolved melt and hence change the initial degassing depth and gas compositions (35). Because of the fractionation of Fe³⁺ and Fe²⁺ by cumulate minerals, magma differentiation may also substantially affect the redox condition of the evolved melt (35–38), leading to variations in gas compositions. Notably, previous works suggested that magma differentiation with degassing could vary the oxygen fugacity (ƒo₂) of martian magmas from ~0 to 4 log units below the fayalite-magnetite-quartz (FMQ) redox buffer (39, 40). Martian magmas likely reached graphite and sulfide saturation during differentiation and degassing, as evidenced by the presence of graphite (41) and abundant sulfide minerals (12–15) in igneous martian meteorites. The saturation of graphite and sulfide could also affect degassing by changing the redox budget and total contents of carbon and sulfur in the magmatic system.

Considering the aforementioned complexities, we simulated degassing of C-H-O-S volatiles from chemically and dynamically evolving magmatic systems in early Mars. Four samples were selected as representatives of primary and evolved melts to encapsulate the ƒo₂ and compositional variations during differentiation of magmas derived from two different sources (depleted versus enriched). The volatile compositions of the primary melts were estimated from martian mantle melting, whereas those of the evolved melts were adjusted according to their extents of crystallization. We conducted polybaric degassing simulations using the Magma and Gas Equilibrium Calculator (MAGEC) (35, 42) with nine volatile species (i.e., CO, CO₂, CH₄, COS, H₂, H₂O, H₂S, S₂, and SO₂) and model options suitable for martian conditions. The pressure range (5 kbars to 0.01 bars) of degassing spans the base of the crust to the surface (see Materials and Methods).

We show that differentiation of primary martian melts yields graphite and sulfide saturation in their evolved melts at crustal pressures. Graphite saturation could reduce the gaseous SO₂/H₂S ratio by about one order of magnitude, whereas sulfide saturation can increase the gaseous SO₂/H₂S ratio by ~3 to 5 orders of magnitude. The total sulfur abundances in the gas become notable when the magma ascends to near-surface depths. We find that H₂S and S₂ are the dominant sulfur species with abundances of ≳10 mol % in volcanic gases from the primary and evolved magmas. These reduced sulfur species can readily form optically thick haze through photolysis (43) and potent greenhouse gases, contributing to climate regulation on early Mars.

RESULTS

Sample characterization

We chose two olivine-phyric shergottites, Yamato (Y-)980459 and Northwest Africa (NWA) 1068 (44), as representatives of primary melts from the depleted and enriched mantle sources. With rhyolite-MELTS (45, 46) modeling of the liquid lines of descent, we took two basaltic shergottites, Queen Alexandra Range (QUE) 94201 and Los Angeles, as the evolved melts derived from the primitive magmas of Y-980459 and NWA 1068, respectively (Fig. 1 and fig. S1). The depleted series is characterized by more reduced conditions with a smaller variation in the ƒo₂ (Y-980459: ∆FMQ = −3.7; QUE 94201: ∆FMQ = −3.0) (47) than those of the enriched series (NWA 1068: ∆FMQ = −2.5; Los Angeles: ∆FMQ = −1.3) (47, 48). Primary magma hydrogen and sulfur compositions were estimated from 10% partial melting of the mantle (49). We chose 150 parts per million by weight (ppm) H₂O and 500 ppm S for the martian mantle among the recommended ranges for H₂O (anhydrous to >500 ppm) (50–57) and sulfur (<700 to 1000 ppm) (58). Following previous assumptions of a graphite-saturated mantle (59, 60), we estimated 180 ppm C for the primary melt from the depleted source, which was also used for that from the enriched source. Applying the olivine-melt geothermometer (61) to the bulk compositions (44), we calculated the liquid temperatures of these four samples at 5 kbars to be ~1100° to 1500°C. With an assumed solidus temperature of 950°C, the liquid temperatures indicate that QUE 94201 and Los Angeles experienced ~55 and 70% crystallization. The concentrations of H-S-C in the evolved melts were thus adjusted according to the extents of crystallization (data S1), from which we estimate predegassed volatile contents for QUE 94201 (i.e., 0.33 wt % H₂O, 1.10 wt % S, and 0.04 wt % C) and Los Angeles (i.e., 0.49 wt % H₂O, 1.64 wt % S, and 0.06 wt % C).

Fig. 1. Selected meteorite compositions.

Plots (A) and (B) show Al₂O₃ and SiO₂ contents of the melt as a function of MgO. The solid lines display the rhyolite-MELTS (45, 46) modeled liquid lines of descent for the depleted (blue) and enriched (purple) magmas of Y-980459 and NWA 1068, respectively, whose bulk compositions are taken from (44). The composition of the evolved sample, Los Angeles basaltic shergottite (44), falls on the liquid line of descent of NWA 1068 to represent the enriched trend (purple triangles). Similarly, the composition of the evolved sample, QUE 94201 basaltic shergottite (44), falls on the liquid line of descent of Y-980459, to represent the depleted trend (blue circles). Compositions of other shergottites (44) that were not selected are plotted as gray circles.

Degassing pathways, saturation of graphite and sulfide, and redox variations

Results of simulations conducted with the MAGEC program show that the saturation of CO₂ and CO gases initiates degassing for the selected martian magmas in the lower to upper crust (Fig. 2A and data S2). During degassing, the primary melts (Y-980459 and NWA 1068) are undersaturated with graphite and sulfide. However, with the increase in volatile concentrations due to differentiation, the evolved melts (QUE 94201 and Los Angeles) are saturated with graphite at initial degassing depths and maintain sulfide saturation from initial degassing to ~0.3 bars.

Fig. 2. MAGEC modeling results of C-H-O-S volatiles in the gas and melt.

Plot (A) depicts the modeled gas compositions and redox variations for the selected meteorite compositions as a function of pressure. Plot (B) displays the modeled volatile compositions in the melt as a function of pressure during degassing. The pressure intervals of graphite and sulfide saturation are shown in plot (B) as gray and yellow bars, respectively.

The initial degassing pressures (P₀) of the two primary melts (Y-980459 and NWA 1068) vary from 2.0 to 1.4 kbars. This is because the higher initial ƒo₂ (FMQ-2.5) of NWA 1068 stabilizes more CO₂ (Fig. 2B), which is more soluble in the melt than CO, to initiate degassing at shallower depths. Conversely, P₀ of their evolved melts (QUE 94201 and Los Angeles) range from 1.3 to 3.2 kbars. Because these melts are saturated with graphite upon initial degassing, the total carbon budget in the predegassed melt strongly depends on the initial ƒo₂. At graphite saturation, the higher initial ƒo₂ of Los Angeles allows the melt to dissolve more CO₂ (Fig. 2B) and therefore facilitates vapor saturation at greater depths.

As pressure decreases during degassing, the primary melts show little variation in ƒo₂ but become gradually reduced at pressures of ≲1 bar (Fig. 2A). Their decreases of ƒo₂ near surface are due to the amplified degassing of oxidized sulfur and hydrogen species (e.g., SO₂ and H₂O; Fig. 2A). Graphite saturation in the evolved melts at their initial degassing depths appears to decrease the ƒo₂ by ~0.3 to 1 log unit following the graphite–CO–CO₂ redox buffer (Fig. 2A). Upon cessation of graphite saturation, the evolved melts become increasingly oxidized by ~0.5 to 2 log units resulting from removal of S²⁻ by sulfide saturation.

To better characterize the reducing capacity of the magmatic gases, we calculated the index of reducing power (R_p) for the magmatic gases following the definition from a previous work (35), which is determined by the ratio of all reduced to oxidized species. The R_p values of magmatic gases (Fig. 3A) generally display an inverse trend with ƒo₂ (Fig. 2A). Because of the relatively more reduced conditions, the gases emitted by the primary melts generally show greater R_p values than those from the evolved melts (Fig. 3A). Notably, at near-surface pressure (~0.1 to 1 bar), the R_p values of all magmatic gases vary within a small range (~−0.2 to +0.5; Fig. 3A), comparable to those of terrestrial submarine gases (−0.02 ± 0.2) (35).

Fig. 3. Reducing capacity and volatile speciation of magmatic gases.

Plot (A) shows the reducing power (R_p) of the magmatic gases during degassing as a function of pressure for the four tested meteorite compositions. The gray boxes denote the log(R_p) values of terrestrial submarine and subaerial gases (34). Plots (B to F) display the molar ratios of CO₂/CO, H₂O/H₂, SO₂/H₂S, S₂/H₂S, and COS/CO₂ in the magmatic gases.

C-H-O-S volatiles in magmatic gases

The magmatic gases are mainly composed of CO, CO₂, H₂, H₂O, H₂S, S₂, and SO₂ with minor COS and trace CH₄ (Fig. 2A). Among the degassed species, CO and CO₂ are the most abundant species degassed at crustal pressures of ≳2 to 20 bars due to their relatively lower solubilities. At more reduced conditions, the high-pressure magmatic gases from the depleted series (Y-980459 and QUE 94201) show much lower CO₂/CO molar ratios (≈0.1 to 0.8) than those (≈0.9 to 7) from the enriched series (NWA 1068 and Los Angeles) (Fig. 3B). Following extensive degassing of CO and CO₂ at high pressures, H₂O becomes a dominant species in the gas (≳30 to 40 mol %) near the martian surface (~1 bar) because of its lower solubility at shallower pressures (Fig. 2A). At near-surface depths, the lower ƒo₂ conditions of the depleted series permit considerable H₂ degassing with a H₂O/H₂ molar ratio of ≈1 to 4, whereas the higher ƒo₂ conditions of the enriched series produce more H₂O-rich gases with a H₂O/H₂ molar ratio of ≈2 to 15 (Fig. 3C).

Our simulations show that H₂S and S₂ are the predominant sulfur species in the magmatic gases at all tested pressures (Fig. 3, D and E). Although the abundances of SO₂ and H₂S are equivalent at pressures of ≲0.1 bars for enriched martian magmas, SO₂ never appears to be more abundant than H₂S and is at maximum only ~17% of the total gaseous sulfur. For all tested compositions, COS in the gas follows CO₂ with nearly constant ratios for individual samples (Fig. 3F), but its concentration decreases from ~1 mol % at the initial degassing pressure to ~0.001 mol % at near-surface pressures (Fig. 2A). The degassing of CH₄ is positively correlated with the bulk carbon contents of the melts and the decreasing ƒo₂. Given the lower carbon contents, the primary melts permit negligible degassing of CH₄ with concentrations (≲0.0001 mol %) less than those from the evolved melts by ~3 to 4 orders of magnitude. There is a slight increase in CH₄ degassing from the evolved melts associated with graphite saturation and the corresponding decrease in ƒo₂ (Fig. 2A). However, with the increasing ƒo₂ spurred by sulfide saturation, CH₄ degassing from the evolved melts also becomes negligible at pressures of ≲10 bars.

C-H-O-S volatiles dissolved in martian melts

Martian melts from our simulations contain S²⁻, H₂O, CO₂, CO, and H₂ with minor S⁶⁺ and trace CH₄. Our results indicate that, across the tested range of degassing pressures (0.01 bars to 5 kbars), reduced sulfur (S²⁻) is consistently the most abundant species (~0.2 to 0.5 wt %) in both primary and evolved martian magmas (Fig. 2B). The dissolved S²⁻ contents in the primary melts, which are sulfide undersaturated, decrease from 0.5 to ~0.2 wt % as the degassing pressure decreases to 0.01 bars. However, S²⁻ concentrations in the evolved melts remain at ~0.5 wt % at pressures of ≳0.3 bars due to sulfide saturation and then decrease slightly to ~0.3 wt % at 0.01 bars by degassing. At crustal pressures, H₂O is highly soluble in martian magmas. Although the predegassed melt concentrations of H₂O can range from ~0.15 to 0.5 wt % depending on differentiation, heightened H₂O degassing at near-surface pressures causes all melts to contain ~40 to 70 ppm H₂O at ~0.01 bars. The reduced conditions of the depleted series cause the degassed melt to have a lower dissolved CO₂/CO ratio by weight (≈0.01 to 2) in comparison to that (≈0.1 to 15) of the oxidized, enriched series (Fig. 2B). The dissolved CO₂ contents of all melts decrease by ~4 to 6 orders of magnitude from 5 kbars to surface pressures as CO₂ is degassed.

Minor and trace C-H-O-S constituents (H₂, S⁶⁺, and CH₄) are all in abundances of ≲10 ppm in martian magmas (Fig. 2B). Because of their lower abundance of hydrogen before degassing, the primary melts contain less dissolved H₂ than the evolved melts by a factor of 10. The concentration of H₂ in the primary melts declines at ~1 bar in alignment with the magnified degassing of oxidized sulfur and hydrogen species at shallow depths, resulting in an overall decrease in the H₂ concentration by ~3 orders of magnitude. Within evolved melts, the dissolved H₂ concentration decreases at ~10 bars by ~5 orders of magnitude because of their increase in oxidation state associated with sulfide saturation and graphite undersaturation. Like H₂, the concentration of S⁶⁺ is greater in the evolved melts than the primary melts; however, the maximum S⁶⁺ melt concentration is merely ~1 ppm. Its abundance in the melt is highly reflective of the ƒo₂ conditions, causing the S⁶⁺ concentration in evolved melts to fluctuate substantially more than in primary melts. The least abundant component of the melt is CH₄ at a maximum concentration of ~0.1 ppm.

DISCUSSION

Sulfur degassing and speciation

The most notable observation from our simulations is the more dominant presence of reduced sulfur species (S₂ and H₂S) over oxidized sulfur species (SO₂) in the gases from martian magmas at crustal and surface pressures. To highlight the change of sulfur speciation during degassing, we plotted the ratios of three major sulfur species in the magmatic gas as a function of pressure (Fig. 3, D and E). The magmatic gases from Y-980459 show a lower SO₂/H₂S ratio than that of the NWA 1068 melt due to its comparatively reduced condition. As pressure decreases, the SO₂/H₂S molar ratio of magmatic gases from these two primary melts increases by a factor of ~5 to 10 and reaches up to ~0.1 to 1.

The systematic variation of gaseous SO₂/H₂S ratios is likely controlled by the degassed major volatile species (hydrogen and carbon) and redox condition. At crustal pressures (≳2 to 20 bars), CO₂ and CO are the major species in the gas and likely dominate the sulfur species through

$${\mathrm{H}}_2{\mathrm{S}}_{\left(\mathrm{g}\right)}+{\text{3CO}}_{2\left(\mathrm{g}\right)}={\text{SO}}_{2\left(\mathrm{g}\right)}+{\text{3CO}}_{\left(\mathrm{g}\right)}+{\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{g}\right)}$$ (1)

At near-surface pressures (≲1 bar), H₂ and H₂O could regulate sulfur speciation by the equilibrium reaction

$${\mathrm{H}}_2{\mathrm{S}}_{\left(\mathrm{g}\right)}+{\text{2H}}_2{\mathrm{O}}_{\left(\mathrm{g}\right)}={\text{SO}}_{2\left(\mathrm{g}\right)}+{\text{3H}}_{2\left(\mathrm{g}\right)}$$ (2)

Note that a lower pressure facilitates both reactions toward the right-hand side and thus increases the SO₂/H₂S ratio. During degassing, the variation in SO₂/H₂S ratios of the magmatic gases from the evolved melts is also strongly associated with the change in ƒo₂. In particular, the SO₂/H₂S ratios of the gases from the evolved melts decrease by ~1 to 2 orders of magnitude at graphite saturation and then increase by ~3 to 4 orders of magnitude (Fig. 3D) along with sulfide saturation (Fig. 2F). However, the maximum SO₂/H₂S molar ratios only reach ~0.2 to 1.3 at 0.01 bars (Fig. 3D).

The S₂/H₂S molar ratios in the gas follow similar trends as SO₂/H₂S ratios during degassing and can reach up to ≈4 to 10 at ~0.01 bars (Fig. 3E). Notably, the relative abundances of S₂ and H₂S in magmatic gases from Y-980459, QUE 94201, and Los Angeles show a crossover (S₂/H₂S = 1) at pressures near the martian surface (~1 bar). With a relatively more oxidized condition, the magmatic gases from NWA 1068 show higher S₂/H₂S ratios than those from Y-980459. Yet, the magmatic gases from both primary melts maintain greater S₂/H₂S ratios than those from their evolved melts although the primary magmas release slightly less S₂ (~7 to 18 mol %) than evolved magmas (~11 to 30 mol %) at martian surface pressures. This is because H₂S degassing is likely governed by the equilibrium reaction

$${{\mathrm{S}}^{2-}}_{\left(\text{melt}\right)}+{{\text{2OH}}^{-}}_{\left(\text{melt}\right)}={\mathrm{H}}_2{\mathrm{S}}_{\left(\mathrm{g}\right)}+{{\text{2O}}^{2-}}_{\left(\text{melt}\right)}$$ (3)

With lower H₂O contents, the primary magmas emit less H₂S (~0.001 to 10 mol %) than the evolved melts (~0.1 to 20 mol %) at crustal pressures of >1 bar.

Variable mantle volatile compositions

We note that our simulations were on the basis of fixed hydrogen and sulfur compositions for a graphite-saturated mantle (H₂O = 150 ppm; S = 500 ppm). It is plausible that the martian mantle contains volatile compositions different from those originally modeled here. In particular, the mantle H₂O content of Mars is contested, with estimates of the shergottite mantle source ranging from nearly anhydrous (~15 ppm) to >500 ppm H₂O (50–57), whereas estimates for sulfur in the martian mantle range from 360 to 1000 ppm (58, 62). To test whether this would substantially affect the simulation results, we rerun the MAGEC degassing model by considering additional scenarios of variable H₂O (30 and 300 ppm) and sulfur contents (300 and 1000 ppm) in the martian mantle (figs. S2 to S5 and data S3 to S6).

When considering the large difference in mantle H₂O, one might expect there to be proportional differences in the concentrations of H₂O in the melts, which may cause more notable variations to the gaseous SO₂/H₂S and S₂/H₂S ratios (cf. Eq. 3). This anticipated trend is evident at crustal depths where H₂O is highly soluble in the melt. As the mantle H₂O content increased by a factor of 10, we found that the SO₂/H₂S and S₂/H₂S ratios of the magmatic gases decreased by two orders of magnitude at crustal pressures, which is consistent with the stoichiometric coefficient of dissolved H₂O in Eq. 3 (Fig. 4). Yet, the expected trend does not appear near the surface where H₂O is extensively degassed. In this case, the concentrations of H₂O in the melts from the H₂O-rich mantle are slightly greater than those derived from the H₂O-poor mantle by a factor of ~1.5 to 2. Such an increase in dissolved H₂O in the melts at near-surface pressure decreases the gaseous SO₂/H₂S and S₂/H₂S ratios of the enriched series by a factor of ~3 (cf. Eq. 3). However, for the depleted series, the SO₂/H₂S and S₂/H₂S ratios are substantially affected by their more reduced conditions that can lower H₂O content in the melt.

Fig. 4. Variations in SO₂/H₂S and S₂/H₂S ratios of magmatic gases due to variable mantle H₂O compositions.

The plot depicts the molar ratios of SO₂/H₂S (circles) and S₂/H₂S (diamonds) in magmatic gases at crustal (1 kbar; unshaded) and atmospheric (0.1 bars; shaded) pressures for three choices of mantle H₂O contents (30, 150, and 300 ppm). Magmas that are graphite and/or sulfide saturated at the specific pressure condition are marked by an “x” and/or a cross, respectively.

In the case of variable mantle sulfur contents, the increase in mantle sulfur generally leads to the increase in S₂ in the magmatic gases from the primary and evolved melts at sulfide-undersaturated conditions. However, the gaseous SO₂ and H₂S from the primary and evolved melts show distinct trends, as manifested by the SO₂/H₂S ratios in Fig. 5. With the increase in mantle sulfur, the SO₂/H₂S ratios of the magmatic gases from evolved melts increase by a factor of ~2 to 6, whereas those from primary melts decrease by a factor of ~2 to 3. The differing trends of the gaseous SO₂/H₂S ratios for the evolved versus the primary melts can be explained by the disproportionation of S₂ exothermically into SO₂ (i.e., ½ S₂ + O₂ = SO₂) and endothermically into H₂S (i.e., ½ S₂ + H₂O = H₂S + ½ O₂). As the mantle sulfur content increases, the cooler temperatures of the evolved melts favor increased SO₂ emission (i.e., raising the gaseous SO₂/H₂S ratios) whereas the warmer temperatures of the primary melts prefer degassing more H₂S (i.e., decreasing the gaseous SO₂/H₂S ratios). When the melts reach sulfide saturation (e.g., Los Angeles and QUE 94201 at 1 kbar; Fig. 5), the concentrations of SO₂, S₂, and H₂S in their magmatic gases do not appear to be affected by the variation of mantle sulfur contents. This is because sulfide saturation buffers the activity of S²⁻ in the melt and hence controls the degassing of different sulfur species (e.g., H₂S in Eq. 3) to be independent of variable total sulfur budgets in the magmatic system. We would like to emphasize that H₂S and S₂ remain the predominant sulfur species in the magmatic gases from both primary and evolved melts at the crustal and surface pressures.

Fig. 5. Variations in SO₂/H₂S and S₂/H₂S ratios of magmatic gases due to variable mantle sulfur compositions.

The plot shows the molar ratios of SO₂/H₂S (circles) and S₂/H₂S (diamonds) in magmatic gases at crustal (1 kbar; unshaded) and atmospheric (0.1 bars; shaded) pressures for three choices of mantle sulfur contents (300, 500, and 1000 ppm). Magmas that are graphite and/or sulfide saturated at the specific pressure condition are marked by an “x” and/or a cross, respectively.

Implications for the early martian climate

The atmosphere and climate of early Mars remain unknown, but abundant observations of sulfides in martian meteorites (12–15) and widespread detections of sulfates at the martian surface (8–11) suggest that sulfur cycling could have played a key role in shaping the climate of Mars ~3 to 4 Ga. Through a comprehensive assessment of C-H-O-S degassing in evolving martian magmas, we show that reduced sulfur, as H₂S and S₂, were the dominant sulfur species emitted from early martian magmatism as opposed to oxidized sulfur as SO₂. This notion appears to be consistent across a wide range of mantle volatile compositions. Lowering the pressure can increase SO₂ in magmatic gases dominated by hydrogen and/or carbon volatiles; however, the low redox budget of the martian melts before degassing limits the relative proportion of oxidized versus reduced sulfur species.

With the incorporation of magmatic differentiation, our models indicate evolved martian magmas, distinct from their primary counterparts, obtain sulfide saturation at pressures spanning the martian crust through surface because of their increased dissolved S²⁻ content. The observations of sulfides as rounded droplets and ellipsoidal blebs in shergottites, as well as correlation between sulfides in shergottites with differentiation proxies (e.g., bulk Mg number and Ni content in sulfide) (15, 63) support our modeled results of igneous sulfide formation in evolved magmas. The precipitated sulfide in the martian crust could potentially serve as a sulfur reservoir to extend the duration of sulfur cycling if reworked by later magmatism. Degassed sulfur on Mars ~3 to 4 Ga would likely be reduced, in the form of H₂S at crustal pressures and S₂ at pressures relevant to the martian surface. At the surface (~0.1 to 1 bar), martian magmatic gases can contain ≳11 mol % H₂S and ≳30 mol % S₂.

We note that the emitted H₂S and S₂ vapors could readily precipitate into sulfide and elemental sulfur, respectively, on the martian surface. The Curiosity rover recently observed elemental sulfur on Mars at Gediz Vallis, Gale Crater (64), supporting our simulation results. To reconcile reduced sulfur degassing with observations of oxidized sulfates at the martian surface, we suggest that the precipitated sulfide and elemental sulfur from magmatic gases likely undergo oxidative hydrologic weathering. A previous study proposes that hydrologic weathering could oxidize sulfide (FeS) and elemental sulfur (S⁰) alongside weathering of the basaltic martian crust to produce the (Ca, Mg, Fe)SO₄ sedimentary sulfate deposits (65). Taken collectively, we posit that sulfur cycling on early Mars was characterized by multistage processes (Fig. 6): (i) the release of reduced sulfur species from magmatic degassing, (ii) formation of sulfide and elemental sulfur at the martian surface, and (iii) subsequent conversion into sulfate through a secondary oxidative weathering mechanism.

Fig. 6. Schematic illustration of sulfur cycling from an evolving crustal magmatic system on early Mars.

Primary magmas in the martian crust are undersaturated with graphite and sulfide, but their differentiation can yield graphite and sulfide saturation in their evolved magmas. Martian magmas degas variable C-H-O-S volatile species, among which the dominant ones are S₂ and H₂S. These reduced sulfur species (S₂ and H₂S) can condense to produce elemental sulfur and sulfide, which then undergo hydrologic weathering alongside the basaltic crust for conversion into sulfate deposits widely observed on the martian surface. In addition, the degassed S₂ and H₂S could also modulate the climate of early Mars by forming a sulfur-rich hazy atmosphere with the potent greenhouse gas, SF₆. The latter can be readily produced by reacting the reduced sulfur species with F₂, which is also likely emitted from martian magmas.

Our simulation results are also consistent with the observations of igneous graphite in martian meteorites (41). The presence of graphite at mid- to upper-crustal pressures in only evolved magmas introduces the possibility of a secondary graphitic crust on Mars—separate from magmas produced from the graphite-saturated mantle (59, 60)—generated through differentiation of magmas. Furthermore, the pressures at which evolved melts are no longer graphite saturated (~1 kbar $\approx$ 10 km) approximately align with the first of the major seismic discontinuities on Mars (8 to 13 km) (66) and would contribute to a less dense martian crust as observed from the InSight mission (66, 67). Graphite in the martian crust could provide an important reservoir of carbon. Lithospheric foundering, if occurred (68), could cycle this crustal carbon reservoir back to the mantle. This carbon cycling could potentially extend the duration of a CO₂-dominated greenhouse gas atmosphere for a habitable ancient Mars.

Magmatically degassed H₂S and S₂ on early Mars may have induced a hazy atmosphere. Studies on exoplanet atmospheres indicate that even trace concentrations of H₂S can be critical in organosulfur aerosol haze formation, especially in CO₂-rich atmospheres as assumed on early Mars (43, 69, 70). The inclusion of even low concentrations of H₂S in a CO₂-rich atmosphere appears to produce diverse and complex sulfur species, such as CH₃SH, C₂H₄S, CS₂, and S₈, and generate a thicker haze in comparison to H₂S-free atmospheres (43, 70). The organosulfur compounds could produce additional COS (43), which also is a minor constituent in our modeled magmatic gases. COS is of particular interest with regard to habitability because of its role as a reactant in the production of diverse sulfur molecules (71, 72) and as a catalyst in the formation of peptides from amino acids (73). In addition, degassed S₂ could potentially be readily converted to S₈ and contribute to haze formation in the martian atmosphere. However, there is no unanimous conclusion as to the effect of haze on the early martian climate. Scattering caused by S₈ aerosols could induce a cooling effect of up to ~20°C (74). Yet, it is possible that a fractal aggregate haze of organic particles could have a strong ultraviolet shielding effect, protecting less stable reduced species from photolysis and possibly allowing for the formation of a dense, reduced greenhouse atmosphere (75). Although not modeled here, halogens (chlorine and fluorine) are among the degassed volatile species on early Mars (14). Condensed S₈ could react with degassed F₂ to produce SF₆, which is a highly potent greenhouse gas >22,000 times more powerful than CO₂ (76). Regardless of its specific climatic effect, sulfur degassing likely played a critical role in modulating the climate condition of early Mars and needs to be considered in future paleoclimate models.

Last, our findings may also have important implications for understanding the habitability potential of martian hydrothermal systems. Results of our modeling indicate that the magmatic gases in the martian surface environment share similar reducing capacities as those from the hydrothermal systems in terrestrial submarine environments (Fig. 3A). Noting that magmatic fluids can provide not only essential metabolic sources (77) but also crucial redox interfaces (78) for the diverse microbial communities in submarine hydrothermal systems, we speculate that magmatic gases in martian hydrological settings may also play a critical role in nurturing habitable environments for potential life on early Mars.

MATERIALS AND METHODS

Sample selection and determination of their extents of crystallization

Given the importance of redox conditions on degassed and saturated sulfur speciation, it is necessary to highlight the systematic variation in ƒo₂ of martian magmatic sources. The most abundant and varied class of martian meteorites are shergottites, which can be subclassified as basaltic, olivine-phyric, augite-rich, or poikilitic and categorized as from a depleted or enriched mantle source (79, 80). Shergottite crystallization ages range from ~0.35 to 2.4 Ga (81, 82) with source ages that span the Pre-Noachian through the Hesperian (~4.2 to 3.3 Ga) (75). Shergottite variance in both composition and age indicates that shergottites formed from a long-lived magmatic source and throughout multiple magmatic systems (80), making shergottite conditions ideal for modeling ancient martian magmatic systems. There is a correlation between shergottite ƒo₂ and the nature of the magmatic source, as quantified by the meteoritic La/Yb ratio (40, 80). Accordingly, it has been inferred that shergottites are formed through the mixing of two distinct sources, one enriched and oxidized and the other depleted and reduced (80).

To model degassing as a function of crystallization, we consider the compositions and redox conditions of four shergottites from both the depleted and enriched sources: Y-980459, QUE 94201, NWA 1068, and Los Angeles. These samples were selected on the basis of their petrogenetic relations through rhyolite-MELTS modeling (Fig. 1 and fig. S1). We modeled the liquid lines of descent for both the Y-980459 and NWA 1068 with rhyolite-MELTS Version 1.2.0 (45, 46) (data S1). The initial conditions were set to temperatures at 1540°C and pressures at 1 kbar (data S1). The ƒo₂ conditions were set at FMQ-4 and FMQ-3 for Y-980459 and NWA 1068, respectively. We simulated the liquid lines of descent for these meteorite compositions with the “Fractionate Solids” mode in rhyolite-MELTS (45, 46). Our results indicate that the enriched Los Angeles basaltic shergottite falls on the liquid line of descent for NWA 1068, whereas the depleted QUE 94201 basaltic shergottite plots onto the liquid line of descent for Y-980459 (Fig. 1). The rhyolite-MELTS (45, 46) modeling of these meteorites suggests that the composition of the Los Angeles basaltic shergottite characterizes an evolved magmatic composition of the enriched magma source for NWA 1068, and similarly, QUE 94201 represents an evolved melt composition of the depleted magma source of Y-980459. The selected meteorites not only capture a range of major oxide compositions but also exemplify increasing ƒo₂ with crystallization.

Using the experimental data for martian melt compositions from (83, 84), we tested the liquidus calculations of rhyolite-MELTS and found that rhyolite-MELTS overestimates the liquidus temperature of martian magmas by ~50° to 125°C (table S1). The olivine-melt geothermometer from (61), however, can predict the experimental liquidus temperatures to within ±35°C. Applying this geothermometer, we calculated the liquid temperatures of the selected samples to be ~1100° to 1500°C at 5 kbars. The fraction of the residual liquid (F) was estimated with the following equation

$$F=\frac{T_{\text{melt}}-T_{\text{sol}}}{T_{\text{liq}}-T_{\text{sol}}}$$ (4)

where T_sol is the approximate temperature of the solidus (950°C), T_melt is the temperature of the meteorite obtained by the geothermometer, and T_liq is the temperature of the liquidus (i.e., the temperature received from the geothermometer for Y-980459 or NWA 1068; data S1). Using Eq. 4, we determined that QUE 94201 was produced through ~55% crystallization (F = 0.45) and Los Angeles formed from ~70% crystallization (F = 0.30).

Magmatic volatile compositions

The abundance of volatiles in the martian mantle and magmas remains highly uncertain. Considering the assumption of a graphite-saturated mantle (60, 85), we used an existing model to determine magmatic carbon content at graphite saturation that is calibrated to reduced martian basalts (59). The only magma assumed to be generated from the graphite-saturated martian mantle is that equivalent to Y-980459 (86). Although representative of a primitive, enriched magma, NWA 1068 is not considered to be mantle derived (86). When applying the same model to NWA 1068, the relatively oxidized conditions of NWA 1068 yield unrealistically high carbon contents for the differentiated Los Angeles composition. Therefore, the carbon content determined for Y-980459 was applied to both the depleted and enriched magmas. Hydrogen and sulfur volatiles were considered highly incompatible, so the primary magma volatile compositions were calculated with the assumption of 10% partial melting (49), whereas evolved compositions were further concentrated in incompatible volatile elements via their calculated extents of crystallization (data S1). There are currently no rigorous constraints on the H₂O composition of the martian mantle and predictions for the shergottite mantle source range from nearly anhydrous to >500 ppm H₂O (50–57). An intermediate value of 150 ppm H₂O was selected as the baseline for our calculations, but we investigated a range of mantle H₂O from 30 to 300 ppm. As with graphite, the martian mantle was assumed to be sulfide saturated with primary magmas sufficiently consuming all sulfur in the mantle (i.e., below sulfide saturation) and is estimated to contain <700 to 1000 ppm S (58). To maintain this assumption of sulfide-undersaturated primary magmas, we use a mantle with 500 ppm S as the reference point but the test mantle sulfur content ranging from 300 to 1000 ppm (58, 62).

Magma and gas equilibrium calculation model

The MAGEC program is a thermodynamic modeling software that highlights gas-melt interactions and is applicable to both magmatic degassing and crystallization (35, 42). The redox/chemical equations and mass balance relationships are concurrently solved at variable P-T conditions and an inputted ƒo₂, buffered through redox equations as the system progresses (35). Currently, MAGEC is the only available petrological tool that can simulate C-H-O-S degassing for melts with variable compositions at the saturation of graphite and sulfide. Depending on the selected solubility models and graphite/sulfide saturation models, the modeling results in the earlier version of MAGEC can have intrinsic uncertainties of up to ~20% (35). Error propagations were not included in the MAGEC code due to the expensive computational cost. To minimize the intrinsic uncertainties, we used the updated MAGEC with recently calibrated models for Fe- and S-redox equilibria (42) and graphite-saturation in mafic melts (59). As demonstrated previously (35, 42), the validity of MAGEC has been benchmarked against existing solubility experiments, mid-ocean ridge basalts, and alkaline melt inclusions from Canary Islands.

To simulate degassing and crystallization processes, we used the major element compositions of the four meteorites with different extents of crystallization and modeled polybaric degassing along the melt adiabat for the calculated initial liquidus temperatures (data S1 to S6). According to the seismic constraints from InSight (66), the average thickness (~50 km) of the martian crust indicates a pressure of ~5 kbars at the base of the crust. Estimates for the martian paleoatmosphere generally range from ~0.1 to 1 bar with a possible lower limit at ~0.01 bars (4). Thus, we considered a degassing pressure range from 5 kbars to 0.01 bars. The output from the MAGEC software contains volatile compositions of the melt (H₂, H₂O, CO, CO₂, CH₄, S²⁻, and S⁶⁺) and the gas (H₂, H₂O, CO, CO₂, CH₄, H₂, SO₂, S₂, COS, and O₂); the ƒo₂ after degassing; the activities of C⁰, S²⁻, and S⁶⁺; and the extent of degassing.

Supplementary Materials

The PDF file includes:

Figs. S1 to S5

Table S1

Legends for data S1 to S6

Other Supplementary Material for this manuscript includes the following:

Data S1 to S6