Olivine and igneous pyrite samples were collected from the island of Lanzarote, Spain. The samples were crushed and dry sieved to a particle size of less than 20 microns to increase the surface area of the grains. X-ray diffraction patterns showed that the olivine -rich sample (Fo₈₅) also contains enstatite and diopside, whereas no other mineral was identified in the pyrite sample (Supplementary Figure S1). Powder samples were cleaned using a stepwise procedure consisting of sonication followed by centrifugation. First, the samples were sonicated in ethanol to remove organic contaminants, followed by hydrochloric acid (HCl, 0.1N) to dissolve surface oxides. Finally, the samples were washed twice in deoxygenated deionized water (Milli-Q, N₂-purged) to ensure the removal of residual impurities. The FeS₂-Fo₈₅ pairing sample was prepared by weight (60 wt% pyrite and 40 wt% olivine). Despite recent advances in the field of Martian simulants preparations (Ramkissoon et al., 2019), a simplified mixture of olivine and a high amount of pyrite was intentionally used to clearly identify surface alteration within the time scale of laboratory experiments. The selection of olivine as an accessory mineral was based on its rapid weathering behaviour and the development of an SiO₂-rich alteration layer during the initial stages of alteration (Delvigne et al., 1979). The sample was then exposed to anoxic water (Milli-Q, N₂-purged) at a fluid to rock ratio in mass units of ⁓10 g H₂O/1 g mineral mixture (olivine + pyrite) under CO₂ atmosphere (pCO₂ = 0.5 bar) to simulate a thicker and anoxic early Martian atmosphere (Supplementary Figure S2A). Visible light was included to accelerate the dissolution of pyrite (Schoonen et al., 2000). The duration of aqueous weathering (42 days) was based on previous studies that observed surface mineral alteration within this time lapse (Dehouck et al., 2014; Gil-Lozano et al., 2017). The sample was then dried for 5 days under the same CO₂ atmospheric conditions using silica gel as a desiccant (room temperature) to ensure removal of excess water that was not chemically or physically adsorbed.

The exposure of the dried weathered sample to UV-like Martian surface conditions was performed inside the Planetary Atmospheres Surfaces Chamber (PASC) at the Centro de Astrobiologia (CAB) (Mateo-Martí et al., 2006) (Supplementary Figure S2B). The simulated Martian atmosphere is composed of pCO₂ = 7 mbar. To achieve this controlled atmosphere, the chamber was first pumped down to a vacuum conditions of about 1 × 10⁻⁶ mbar (high vacuum) to reduce contamination from O₂ or other gases inside the chamber. CO₂ gas (Alphagaz bottle: CO₂ ⁓99.7 %v/v, H₂O ≤ 200 ppm, O₂ < 25 ppm) was then introduced to raise the pressure from 1 × 10⁻⁶ mbar–7 mbar. A constant temperature of 295 K was used to accelerate the alteration process. The FeS₂-Fo₈₅ weathered sample was irradiated with a UV deuterium lamp for a total of 72 h. This duration was chosen based on previous work confirming that clean single-crystal pyrite exhibits measurable photoreactivity within a few hours under similar conditions (Mateo-Marti et al., 2019). The UV source emits a continuous spectrum in the 200–400 nm range, which is representative of the UV radiation reaching the Martian surface (Patel et al., 2004). The source has a dose emission of 2.3 W/m², corresponding to F = 2.3 × 10¹⁴ (6 eV photons) cm^-2 s^-1 (Caro et al., 2006). This emission is about 10 times lower than the UV flux measured by the REMS instrument onboard the Curiosity Rover (Gómez-Elvira et al., 2014).

Powder X-ray Diffraction (PXRD) analyses were performed to determine the presence of accessory minerals in the initial samples. No new mineral phases were identified by XRD, as the alteration of the sample was not sufficient to be detected by this bulk technique. PXRD was performed using a Seifert 3003 TT with Cu Kα radiation (λ = 1.542 Å). The X-ray generator was set to an acceleration voltage of 40 kV and a filament current of 40 mA. The samples were scanned between 5° and 60° (2θ) with a step size of 0.05° (2θ) and a count time of 2 s per step. Qualitative analysis of the XRD patterns was performed with QualX2 software using the PDF-2 database (ICDD), while Rietveld analyses were performed with the Fullprof software (Roisnel and Rodriguez-Carvajal, 2001). A scanning electron microscope (SEM), JEOL JSM-5600 LV, equipped with an energy dispersive X-ray spectroscopy (EDX) INCA detector (20 kV) was used to characterize the morphology and chemical composition of the samples after exposure to the simulated Martian conditions.

Surface analyses were performed using Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS). Infrared spectra of the mineral samples were analyzed using the diffuse reflectance technique (DRIFT) with a Nicolet FTIR spectrometer. Spectra were collected using a DTGS-KBr detector with 2 cm^-1 resolution in the mid-infrared region (from 4,000 to 400 cm^-1) with an XT-KBr beamsplitter. Each spectrum is an average of three independent measurements (64 scans, 2 cm^-1 resolution). XPS analyses of the samples were performed in an ultrahigh vacuum chamber equipped with a hemispherical electron analyzer (Phoibos 150 MCD) and an Al Kα X-ray source (1,486.7 eV) with an aperture of 7 × 20 mm. The base pressure in the ultra-high vacuum chamber was 1 × 10⁻⁸ mbar and the experiment was performed at room temperature. A 30 eV pass energy was used for the acquisition of the overview sample, while a 20 eV pass energy was used for the analysis of the following core-level spectra: O 1s, C 1s, N 1s, Fe 2p, S 2p and Si 2p. The XPS spectra regions were fitted and deconvoluted using the CasaXPS software (Fairley et al., 2021).

The geochemical simulations were performed with the PHREEQCI 3.7.3 speciation-reaction code (Parkhurst and Appelo, 2013) using the Lawrence Livermore National Laboratory (LLNL) thermodynamic database. This database was modified to include thermodynamic data for a variety of smectite and zeolite minerals from the Thermoddem database, the kinetic rate constants for mineral dissolution and precipitation (Zhang et al., 2019), and a kinetic expression for the formation of H₂O₂ during pyrite dissolution (see Supplementary Appendix 1 in the Supplementary Material for script example).

The goal of the models is to analyze the secondary mineral formation pathways resulting from the weathering of a pyrite-rich and a pyrite-free basalt substrate under the same environmental conditions. In particular, the geochemical scenario simulates the water-rock reaction of a stagnant surface water in equilibrium with an anoxic CO₂-rich atmosphere (Supplementary Figure S3). After 150 years, this system simultaneously loses pCO₂ and cools, representing a transitional warm event in a generally cold and wet early Martian environment (Bishop et al., 2018; Fairén, 2010; Feldman et al., 2024).

The compositions of the pyrite-rich and pyrite-free basalt substrates used in the models are described in Table 1. An olivine-rich basalt composition was intentionally used to have a quick source of iron when modeling the pyrite-free basalt scenario. In each run, a pyrite-rich or pyrite-free basalt substrate was reacted with an initial solution of pure water equilibrated with a CO₂ atmosphere (pCO₂ = 0.5 bar, T = 20°C). The models consider the same set of potential secondary minerals (Supplementary Table S1) that could be formed and redissolved during the simulation time (up to 300 years). The list of secondary minerals includes phyllosilicates, sulfates, oxides, carbonates, and zeolites that are representative of the mineral alteration observed on the Martian surface. In the models, therefore, the amount of mineral reacting with the solution is controlled by the dissolution rate, which depends on the surface area (A₀/V surface area of mineral per unit water mass) of the primary minerals (Fairén et al., 2017). Kinetic rates were modeled using the kinetic script library developed by Zhang et al. (2019) based on the transition state theory rate law (Lasaga, 1998; Palandri and Kharaka, 2004). In particular, the kinetic rate law of pyrite is catalyzed by the presence of H₂O₂, Fe³⁺, and O₂ (McKibben and Barnes, 1986). In the simulations, the primary oxidant is the H₂O₂ formed on the reactive surface of pyrite, simulating an autocatalytic process. The formation of H₂O₂ at the pyrite surface was modeled using a kinetic expression that relates the number of defects undergoing oxidation to the surface area of pyrite. This model predicts that H₂O₂ is continuously introduced into the system as long as pyrite dissolution proceeds (see Supplementary Appendix 1 in the Supplementary Material).

We studied the photocatalysis of a mixture of FeS₂-Fo₈₅ microparticles after exposure to present-day Martian surface conditions (pCO₂ ∼ 7 mbar, UV (200–400 nm) flux ⁓ 2.3 W/m²). The initial sample was previously altered in water equilibrated with a CO₂-rich atmosphere (pCO₂ = 0.5 bar, W/R ⁓ 10 wt%) and then evaporated under the same atmosphere. This process simulates a warmer period of early Mars and aims to evaluate the effect of a partially oxidized surface layer on the photocatalytic properties of the disulfide-silicate mixture.

Figure 1 shows the DRIFT spectra, where we can distinguish surface alteration by-products characterized by the presence of several bands in the region of 2,300–750 cm^-1, related to the asymmetric and symmetric stretching vibrations of Si-O (from olivine and amorphous silica) and S-O (from sulfoxyanions), as well as their corresponding bending vibrations in the region below 750 cm^-1 (Lane, 2007; Ellerbrock et al., 2022). The specific band assignments in these regions are complex due to the strong band overlap caused by the coexistence of different chemical groups (e.g., polysulfides, sulfite, sulfate, silicate, amorphous silica) in different chemical environments (e.g., bridging, outer, inner sphere) (Peak et al., 1999). Overall, after exposure to current Martian surface conditions, the sample showed a decrease in disulfide (S-S) stretching at 435 cm^-1 and an increase in stretching (3,800–2,800 cm^-1) and bending (1700–1,600 cm^-1) modes of H₂O. In addition, the bands at 1,145 and 1,102 cm^-1 showed a significant increase. We tentatively attribute the band at 1,145 cm^-1 to an increase in sulfate, likely in a chemical environment similar to that of gypsum, with reduced distortion of the v₃ band, resulting in the predominance of a single band (Lane, 2007). Additionally, the band at 1,100 cm^-1 is assigned to an enhancement of the amorphous silica coating (Ellerbrock et al., 2022). The increase in hydration of this sample also supports the increase in hygroscopic by-products such as silica and sulfate. Interestingly, we also observed changes in the v₃ stretching region of carbonates (i.e., 1,275–1,590 cm^-1) (Bishop et al., 2021). The initial weathered sample showed a band at 1,425 cm^-1, assigned to calcite. However, after exposure to Mars’ surface conditions, this band split into three bands, indicating a symmetry reduction of the carbonate compound (Bishop et al., 2021).

Using XPS, we further analyzed the evolution of the surface alteration layer after exposure to current Mars-like surface conditions. Quantification of the XPS survey spectra showed a significant increase in the oxygen contribution after exposure to current Martian surface conditions (Supplementary Figure S4), which is consistent with the increase in silica, sulfate, and the increase of the hydration level found in the DRIFT spectra. The most notable changes related to pyrite alteration occurred in the Fe 2p_3/2 and S 2p_3/2 orbitals, where surface species shifted towards higher values of binding energies. This is associated with an increase in the oxidation state of the corresponding species (Schaufuß et al., 1998). Considering that high-spin Fe(II) and Fe(III) compounds give rise to multiplet splitting and satellite features (Grosvenor et al., 2004), the fitting of Fe 2p in chemical environments of different iron species (such as sulfates, oxides and oxyhydroxides) is challenging due to peak overlap. Nevertheless, the shape of the Fe 2p spectra showed significant differences before and after exposure to Martian surface conditions (Figure 2). The initial weathered sample showed a well resolved peak assigned to the Fe-S contribution of pyrite (Fe 2p_3/2 ⁓ 707.6 eV) and a broad tail towards higher binding energies, indicating the presence of Fe(II) and possibly some Fe(III) compounds. However, after simulating Martian surface conditions, the Fe-S peak showed a significant decrease while the asymmetric tail increased at higher binding energies, consistent with an increase in Fe(III) compounds. This is further supported by the appearance of Fe(III) satellite peaks (Grosvenor et al., 2004). Figure 3A illustrates the fitting of the S 2p spectra. We identified three spin-orbit components, assigned to S₂ ^2-(S 2p_3/2 162.1 ± 0.1eV), S_n ² (S 2p_3/2 164.2 ± 0.1eV), SO₃ (S 2p_3/2 166.7 ± 0.1eV) and SO₄ ^2- (S 2p_3/2 169.5 ± 0.1eV), respectively. Following exposure to Martian surface conditions, the sample exhibited a notable increase in sulfate contributions, accompanied by the emergence of a new peak at the highest binding energy (172.5 eV). This peak can be attributed to the emergence of a ferric hydroxysulfate compound (Todd et al., 2003), possibly due to hydroxylation at pyrite defect sites (Xian et al., 2019), along with an increase in the Si 2s plasmon, which is also observed in the Si 2p plasmon. Overall, the Fe 2p and S 2p spectra showed a decrease in iron disulfide species in favor of iron sulfate and iron oxide/oxyhydroxide formation.

Similarly, we observed a significant change in the Si 2p spectra (Figure 3B). The initial weathered sample showed the major peak at 102.5 ± 0.1 eV, assigned to silicate. After exposure to Martian surface conditions, this contribution decreased, while the peak at 105.1 ± 0.1 eV, assigned to hydroxylated silica, increased (Kwak et al., 2011). The O 1s spectra of the sample after exposure to current Martian surface conditions also shifted to higher binding energies, which is related to the increased OH and H₂O species on the surface of the sample (Supplementary Figure S5). Unfortunately, the coexistence of various compounds within the narrow binding energy range of the O 1s orbital makes accurate deconvolution challenging. These findings are further corroborated by EDX analysis, which revealed an oxygen enrichment in olivine particles after exposure to current Martian surface conditions (Supplementary Figure S6).

The geochemical simulations represent ponded water in equilibrium with a CO₂ atmosphere undergoing progressive cooling and pCO₂ reduction (Fairén et al., 2009; Fairén et al., 2011) (Supplementary Figure S3). The boundary conditions, including the primary mineral assemblage and the list of secondary minerals that could form, are the same in both simulations except for the presence of pyrite in one of them (Table 1; Supplementary Table S2). Therefore, the model results allowed us to specifically evaluate the effect of H₂O₂ production from pyrite dissolution on the secondary minerals formed.

As expected, the pH evolution is strongly dependent on pCO₂ variations in the atmosphere, regulated by the CO₂-bicarbonate-carbonate buffering system (Figure 4). In both simulations, the pH of the initial solution is acidic (pH₀ ⁓ 4) but rises rapidly due to chemical weathering of silicate until it reaches a pH close to 6.4, which allows carbonate precipitation and buffers the pH. However, in the pyrite-rich substrate, the pH buffering value is slightly lower, around 6.2, and when pCO₂ is reduced (in the cooling phase), the increase of the pH is less pronounced due to the acidic dissolution of pyrite.

It is noteworthy that in the simulation with the pyrite-rich substrate, the formation of siderite (iron carbonate) is inhibited, despite the iron supply by the pyrite dissolution. This is due to the oxidizing conditions that favor instead the formation of hematite (Fe(III) iron oxide) (Baron et al., 2019), which is kept stable during the whole simulation (warm and cold periods), while nontronite (Fe(III)-rich clay) is only formed in the cooling period. In the pyrite-free scenario, however, hematite (followed by nontronite) is only formed in the cooling period, when basalt dissolution slows down and Eh gradually increases. Interestingly, this increase in Eh causes the destabilization of siderite, which begins to dissolve at the end of the simulation. Consequently, Fe²⁺ is released into the solution, lowering the redox conditions again and favoring the formation of magnetite (Fe(II)Fe(III) iron oxide).

Since Ca-plagioclase (anorthite) is a major component of the substrate and dissolves quickly under these geochemical conditions, kaolinite and montmorillonite are the most abundant clay minerals formed in both simulations.

In the pyrite-rich substrate, we also observe the formation of sulfates due to the higher oxidation of the iron sulfide. The first sulfate to precipitate is alunite, a common weathering product of pyrite dissolution. It begins to form in year 29 and remains stable until year 257, when it is dissolved by the increase in pH that favors the formation of zeolite (Ca-heulandite). Gypsum, which is more soluble, begins to form in year 115 and remains stable since then. Interestingly, at the end of the simulation gypsum experienced a rapid increase coinciding with the start of calcite dissolution, which releases Ca²⁺ into the solution.

These results revealed the strong competition between minerals for the available cations throughout the whole simulation time (Fairén et al., 2017). Overall, the main difference between both simulations shows that dissolution of the pyrite-rich basalt substrate induces higher redox conditions favoring the formation of ferric-bearing minerals and sulfates even in a CO₂ rich anoxic atmosphere.