Experimental timing of pyrite oxidation under various leaching conditions: consequences for rates of weathering in geological profiles

Pyrite samples were collected in the Danube–Bouchon quarry (Hautrage, Belgium), in the mining Wealden facies sediments of the Mons Basin (Fig. 1A). The Mons Basin is a geological unit almost 40 km long and 10 km wide on the NE edge of the Anglo–Paris basin. Its formation is the result of a combination of deep karstic and tectonic subsidence caused by the partial dissolution of thick levels of anhydrites and the reactivation of ante-Cretaceous faults at an extensive regional context (Dupuis and Vandycke 1989; Vandycke et al. 1991; Spagna et al. 2011). The rocks of the Mons Basin consist of Meso–Cenozoic sediments, all unconformably deposited on the Paleozoic basement that has been folded and faulted during the Variscan orogeny (Marlière 1954). In the Mons Basin, the Wealden facies sediments are the products of continental erosion ranging from middle Barremian to late Albian (Schnyder et al. 2009; Yans et al. 2010).

The Danube–Bouchon quarry (city of Hautrage) exploits the Wealden facies sediments of the Hautrage Clay Formation (HCF), which are mainly used for the manufacture of refractory materials and concrete. The HCF forms layers and lenses of a thickness of a few decimeters to several meters, with grain sizes ranging from clay to sand, hosting many plant fossils (Barral et al. 2016, 2017). The deposit is stratified with a direction from N95° E to N105° E, and a dip of 10°–25° (Yans 2007). These sediments are interpreted as a fluviatile environment (Spagna et al. 2012).

The HCF (~ 230 m thick) is divided into 9 different zones from bottom to top (Fig. 1B) (Spagna et al. 2011, 2012): (i) approximately 85 m of reddish clay facies with siderite (FeCO₃) nodules (units A to C); (ii) white, greyish, blueish, brownish or black clays intersected by lenses of sand and levels of pyritized plant debris of a total thickness of 112 m (units D to G); (iii) 35 m of sandy, sometimes conglomeratic sediments with some clayey intercalations that are rich in plant debris (units H and I). Only the upper units (E to I) are still visible in the Danube–Bouchon quarry (Yans 2007). The mineralogy of clays of HCF is mainly characterized by illite, kaolinite, chlorite and various mixed-layers minerals (Yans et al. 2010). A black clay level (Fig. 1C, D), rich in wood debris and pyrite, is observed in the upper zone of unit G (Fig. 1E). Yans (2007) suggests that the formation of pyrite in HCF is related to organic debris being epigenized into pyrite during early diagenesis.

Pyrite samples (Fig. 1E) were collected from the black clay level of the upper zone of Unit G (HCF). The collected pyrites were washed with tridistilled water directly after sampling to remove residual clays. A representative number of pyrites were analyzed by XPS to ensure that they were unoxidized/fresh before starting the experimental oxidation. The pyrite nodules show cubic aggregates on the surface, followed by a fibroradial "crown" and a microcrystalline pyrite in its center (Fig. 1E). Only the outer cubic crystals were used for the experiment here. Pyrites associated with wood were not used because of slight or even intense oxidation (= not "fresh").

After sampling and cleaning, each pyrite was numbered and exposed to various experimental conditions: (i) air (20 samples); (ii) tridistilled water (15 samples); (iii) or exposure to a continuous water drip simulating a flow (25 samples). The lab conditions were as follows: atmospheric pressure (1 bar), humidity between 45 and 55%, temperature between 19 and 23 degrees Celsius. Tridistilled water (Milli-Q) has a low conductivity (5.4 µS/cm) and pH ~ 5.8 (mostly dissolved CO₂), which is similar to rainwater pH but lacking solubilized elements.

For the drip exposure we used an ISMATEC^© Perstaltic Ecoline VC MS/CA 8/6 pump with 8 tubes of Ismaprene (0.51 mm diameter) connected to a tridistilled water tank of 50 L (ESM1). Each tube simulates a non-permanent flow (0.1 ml/min) over the pyrite. For each type of experiment (air, water, drip exposure), oxidation is analyzed at different points in time, from t = 0 to 10,000 h. To stop the oxidation process at a given time, the samples are removed from the exposure system and placed in a vacuum chamber at a pressure of 10^–7 mbar, waiting for XPS measurements. For each sample, at least one surface and one depth measurement were performed on a planar/smooth surface (cubic grains).

Chemical analyses were carried out by Activation Laboratories Ltd. (Actlabs, Canada). Major elements were identified with the help of FUS–ICP (Fusion Inductively Coupled Plasma Optical Emission Spectrometry) while the reduced iron (FeO) was determined by titration. Trace elements were analyzed by FUS–MS (Mass Spectrometry) while S was measured using the IR method (Infrared).

XPS measurements were performed using a K-alpha spectrometer (Thermofisher), equipped with a monochromatic Al Kα source (1,486.6 eV) with a spot size of 300 µm. The Thermo Avantage software was used for data processing. Survey and high-resolution spectra were calibrated with C_1s with an energy step of 1.0 and 0.1 eV, and a pass energy of 200 eV and 20 eV respectively. Spectra recorded during the depth profiles were measured using the so-called snapshot mode. This mode allows to acquire faster profiles with sufficient energy resolution to separate sulfate and sulfide peaks. The sputtering was performed using Ar⁺ ions with an energy of 2 keV and a raster size of 1.5 × 1.5 mm². A flood gun with combined electron and low energy ions was also used during analysis to prevent surface charging. The depth profile analysis consists of several sputtering steps ranging from 5 to 30 s, depending on the total profile time to be reached. After each sputtering step, measurements of S 2p, Fe 2p_3/2, O 1s, Si_2p and C_1s were performed using a snapshot acquisition. For S 2p and Fe 2p3/2 spectra, the different peaks and binding energies associated with pyrite and oxidation products are listed in Table 1. Ar⁺ sputtering can slightly affect the structure of the pyrite to form some Fe–S-type surface species resulting in a small flat peak at 708.8 eV (Karthe et al. 1993), which is, however, negligible compared to the induced oxidation processes and which can be easily discerned from the oxidized species formed (sulfates and Fe-oxides).

The sputtering rate was evaluated by measuring the depth of the crater that was left after the sputtering process on polished pyrite, using a DektakXT stylus profilometer by Bruker^© (Fig. 2A). The average sputtering rate was 0.6 nm.s⁻¹ and is reproducible. The applied XPS measurements provide exact information on the elemental composition of samples with an excellent depth resolution (< 10 nm). Despite the oxidation process being relatively slow, this method allows to determine the evolution of the oxidation depth already within a single year in the laboratory, while the natural process takes place over thousands or even millions of years. Moreover, the data on chemical bonds allow to distinguish pyrite (Fe-sulfides) from Fe-sulfates or even Fe-oxides, even if they occur only in small quantities and in depth. However, the shortcoming of this method is its limited sensitivity (~ 0.1 at. %). Therefore, XPS cannot determine the evolution of trace elements, which were monitored by FUS–MS and hardly impacted the experiments.

For each depth profile, the thickness of the oxidation layer was derived from several characteristics: the Fe/S ratio, the oxygen concentration of the depth profile, and the presence of sulfates or oxides (mostly iron). The evolution of S, Fe and O concentrations of a typical XPS depth profile is shown in Fig. 2B. The oxygen content is never zero due to the shadowing effect and roughness of the sample, but it remains stable when no more oxidation happens (which is also observed for fresh pyrite). The limit of the oxidized layer is also characterized by a Fe/S ratio of 0.5 (Fig. 2B, etching time 1125 s). It should be noted that a Si signal can reveal the presence of a potential clay-filled (micro)fracture, which would affect the result. Once the limits of the oxidation layer are defined, the associated sputtering time is converted to oxidation depth by using the sputtering rate. Error bars are mainly caused by slight variabilities in depth profile measurements, but also by the natural character of pyrite causing variation in experimental time periods. Thus, for a given time period, the error bar represents both the variability of the oxidation depth between samples, and the measurement error of the crater depth with DektakXT, which is, however, negligible compared to the natural variability of the sample.

Natural specimens of pyrite do not have ideal chemistry (FeS₂) and contain small quantities of minor and trace elements (Co, Ni, As, …), even if in most specimen/samples the ratio of molar proportions of Fe and S is close to 1:2 (Abraitis et al. 2004). For example, Doyle and Mirza (1996) determined the S/Fe ratio via electron probe microanalysis with values ranging from 1.983 to 2.015, which corresponds to less than 1% deviation from the ideal formula without considering the influence of trace elements. Oberthur et al. (1997) report average compositions of pyrite from the Great Dyke (Zimbabwe) as 45.44 wt.% Fe and 53.25 wt.% S (molar ratio S/Fe ~ 2.041) with Co and Ni as minor elements of less than 2 wt.%. However, such minor deviations from ideal stoichiometry could have important consequences regarding the electrical properties of the mineral (Abraitis et al. 2004). In pyrite, lattice substitutions can be either stoichiometric or non-stoichiometric. Stoichiometric substitutions mainly result from Co²⁺ and Ni²⁺ replacing Fe²⁺, and Se or Te replacing S (Abraitis et al. 2004). The most common non-stoichiometric substitutions in pyrite include the incorporation of As by substituting Fleischer (1955) and Vaughan and Craig (1978) found the following maximum contents for minor elements in pyrite: As (5 wt.%), Pb (5,000), V (1,000), Sb (700), Ti (600), Sn (400), Tl (340), Se (300), Ag (200), Au (200), Ga (100), Bi (100), Ge (50), Mo (30), Cd (10), In (10) and Hg (0.4) (all concentrations given in wt.ppm except As). Several authors (Wells and Mullens 1973; Huston et al. 1995; Oberthur et al. 1997; Large et al. 2009; Sykora et al. 2018; Steadman et al. 2021) determined minor and trace elements in pyrite (Table 3).

Our studied pyrite samples show an S/Fe ratio of 2.018, which is very close to the ideal stoichiometry. The low percentage of SiO₂, Al₂O₃ and other oxides (Table 2) may be associated with very low clay contents in the microfractures/structures of the pyrite. Moreover, the main minor elements (Table 3) are present in very low concentrations, indicating a lack of inclusions of other sulfide minerals. Thus, the stability, and hence, oxidation rate of the pyrite samples studied here is only slightly or even unaffected by trace elements.

Pyrite oxidation rates have already been studied using modeling based on specific surface area. Several authors (e.g., McKibben and Barnes 1986; Pesic et al. 1989; Nicholson et al. 1990; Moses and Herman 1991; Williamson and Rimstidt 1994; Holmes and Crundwell 2000; Kameia and Ohmotob 2000; Jerz and Rimstidt 2004; Huminicki and Rimstidt 2009; Wang et al. 2019; Verron et al. 2019) defined laws for pyrite oxidation under different experimental setups (Table 4). Most of these models use the specific area (m².g⁻¹) of the pyrite and therefore the surface reaction rate (mol.m⁻².s⁻¹) calculated by dividing the amount of change in reactants or products by the total surface area of pyrite. These equations generally apply only under precise experimental conditions and are very difficult to use in geological applications that require extrapolation over millions of years.

Wang et al. (2019) propose a rate model that includes the rate of pyrite destruction and the rate of unreacted core shrinking. Their model assumes that a diffusion barrier around fresh pyrite forms as a thin layer predominantly by precipitation of oxides, which decreases the diffusion of oxygen to the pyrite surface over time. Some other authors (Garrels and Thompson 1960; Huminicki and Rimstidt 2009; Wang et al. 2022) also mention the presence of a thin layer of ferrous sulfate, iron oxides and sulfuric acid on the pyrite surface that delays oxygen transport. However, Manaka (2023) considers that the oxy-hydroxides layer do not passivate the oxidation of pyrite. In carbonated systems, the precipitation of gypsum or anhydrite (sulfates) with calcium and sulfuric acid could have two opposed effects: (i) the same barrier effect as mentioned above; (ii) the consumption of sulfates could favor pyrite oxidation (Verron et al. 2019) according to the Le Châtelier principle. Verron et al. (2019) mention that the addition of clay phases or other detrital silicates has no impact on pyrite oxidation rate.

Our study based on new XPS analyses provides pyrite oxidation rates that can be extrapolated (linear regression) considering that the barrier effect of sulfates (and oxides) on the surface is constant and therefore, oxidation is also constant over time. This regression/assumption leads to an average oxidation rate of about 0.10 ± 0.02 m.Ma⁻¹ under air exposure. The same approach applied for the data from the drip exposure experiment results in an oxidation rate of 4.3 ± 0.6 m.Ma⁻¹. However, the assumption is only valid with regards to the oxides’ barrier since sulfates are almost entirely leached. These two oxidation rates are different. In the case of air exposure, only the ambient humidity plays a role in the weathering of pyrite. Furthermore, the mineral phases that form on the surface (mainly sulfates) are not leached which probably increases the sulfate barrier effect mentioned by (Butler and Rickard 2000) compared to the drip exposure experiment. Underwater pyrite is almost unaffected by oxidation with values remaining below 1 cm.Ma⁻¹.

Although the presence of clays slightly buffered pH values in the experiment with tridistilled water during the oxidation of pyrite (Fig. 4), the effect was very limited and vanished over time. Nevertheless, pyrite from the Danube–Bouchon quarry is embedded in clay (Hautrage Clay Formation), which preserved them from the oxidizing conditions on the surface. Furthermore, it has already been demonstrated that chlorite (Brown 1967) or even carbonates (Descostes et al. 2002) present in the solution/rock can neutralize acidic fluids, which will play a significant part in the regulation of the pH and the subsequent oxidation process. Thus, the evolution of the pH will have an impact on the oxidation of pyrite (McKibben and Barnes 1986; Williamson and Rimstidt 1994; Jerz and Rimstidt 2004; Wang et al. 2019; Verron et al. 2019).

According to our data on the evolution of pH and oxidation depth for pyrite in water (Fig. 6A), the first rapid oxidation induces a simultaneous drop in pH and is mainly due to the oxygen dissolved in the water, which is consumed to oxidize the pyrite. Once this amount of dissolved oxygen has been depleted, only trace amounts of oxygen are introduced into the system via the exchange between the atmosphere and the closed plastic pot containing tridistilled water and pyrite. The diffusion coefficient of oxygen in water is also relatively low. As a result, the evolution of pyrite oxidation and pH is constant but much slower (Fig. 6A), which is confirmed by a stability field diagram (Fig. 6B) that assumes that pyrite can be stable at low pH with an Eh close to zero (Butler and Rickard 2000).

Our experimental data can be related to different zones of weathering profiles (Fig. 7). Our drip exposure experiment can be associated with the vadose zone, where rainfall and runoff oxidize and leach pyrite in laterite and saprolite parts of the weathering profile. Thus, the average oxidation rate should range between the two values we determined via drip (4.3 ± 0.6 m.Ma⁻¹) and air exposure (0.10 ± 0.02 m.Ma⁻¹), since most regions of the world experience both drier and wetter seasons. Dry periods promote the formation of sulfates while wet periods would favor leaching and the development of oxidized levels. Because sulfides are stable below the water table level under reducing conditions, our observations on pyrite in water can also be extrapolated to weathering profiles, referring to the cementation zone (secondary sulfides) (Guilbert and Park 1986; Brookins 1988; Robb 2005).

Regarding pyrite oxidation rates in geological conditions, our results are similar to those of Lapakko and Antonson (2006) (Table 5). The latter authors show that pyrite oxidation rates, normalized for exposed surface area, range from 4 × 10⁻¹⁰ to 18 × 10⁻¹⁰ mol m⁻² s⁻¹. If we convert their data, considering the molecular weight of pyrite (119.975 g mol⁻¹) and its density (5 g.cm⁻³), we obtain an oxidation rate of 1.36 m.Ma⁻¹. This oxidation rate lies between the two values (air and drip exposure) observed experimentally in this study.

Experimentally obtained weathering rates are probably lower than actual weathering rates in the geological field. As White and Brantley (2003) have shown with a study on silicate minerals, the weathering rate depends on (i) intrinsic characteristics of the mineral/fluids, and (ii) extrinsic characteristics related to the weathering environment. Intrinsic effects include increased physical surface area due to surface roughness, concurrent decrease in reactive surface area due to reduced compositional and structural heterogeneities, and physical occlusion by secondary minerals (see discussion above). Extrinsic factors might have a significant effect but are more difficult to reproduce in the laboratory. In contrast to natural/geological weathering, which involves much lower fluid/mineral interactions over much longer time periods, experimental rates are measured with high fluid/mineral ratios reacting over short periods of time. Sun et al. (2019) simulated an average silicate (granite) weathering rate of 7.0 t km⁻² y⁻¹, which corresponds to 2.6 m Ma⁻¹ (calculated with a granite density of 2.7). Other studies estimated weathering rates using different dating methods. Dequincey et al. (1999) estimated the weathering rate of Goyoum laterite (Cameroum) via radioactive disequilibria (²³⁴U/²³⁷U), based on the hypothesis that the measured disequilibria represent geochemical fractionations that would be characteristic for the formation and evolution of laterite profiles. The resulting weathering rate of 6–8 m Ma⁻¹ is higher but still comparable to our experimental data considering the humid tropical climate of the area in Cameroun (promoting the leaching of sulfates) in addition to the fact that experimental data slightly underestimated geological values. de Oliveira Carmo and Vasconcelos (2006) studied grains of cryptomelane in the Cachoeira Mine weathering profile, Minas Gerais (SE Brazil), and estimated the weathering front propagation rate, measured by ⁴⁰Ar/³⁹Ar ages, to be 8.9 ± 1.1 m Ma⁻¹, which corresponds to a saprolitization rate of 24.9 ± 3.1 tkm⁻² y⁻¹. Vasconcelos and Conroy (2003) conducted a similar study with grains of supergene jarosite, alunite, and cryptomelane from weathering profiles in the Dugald River area (Australia) using ⁴⁰Ar/³⁹Ar dating and estimated a weathering rate of 3.8 m.Ma⁻¹. These values correspond to our data, as other weathering rates (Table 5).

All of these weathering rates are therefore close to the experimental ones. However, deposits in natural systems do not predominantly consist of pure pyrite (or more generally sulfides). Mineral inclusions in pyrite such as galena or sphalerite (up to 0.9 wt.% of Pb and Zn) can occur and are generally associated with arsenian pyrite, which is strongly correlated with high contents of Au (Table 3) (Large et al. 2009). These sulfides are embedded in a host rock with variable chemical–mineralogical–petrographical compositions, porosities, fractures, and catalysts such as bacteria. These geological parameters (Table 6), as well as climate change over time (Boni and Large 2003), will influence the weathering rate. The role of bacteria (thiosulfates) is not observed in our experiments but could play a role in the pyrite oxidation process in natural systems.