Sources of iron

Iron (Fe) and sulfur (S) are the 4^th and 16^th most abundant elements in the Earth’s crust, respectively. Iron is cycled between its 2+ and 3+ redox states by various abiotic and biotic processes (Kappler et al., Reference Kappler, Bryce, Mansor, Lueder, Byrne and Swanner2021). Fe-S mineral formation requires Fe(II), which can come from multiple sources. Direct sources of Fe²⁺ originate from reduced water bodies such as groundwaters, porewaters and hydrothermal vents, as well as from the dissolution of Fe(II)-containing minerals such as siderite (FeCO₃), vivianite (Fe₃(PO₄)₂·8H₂O) and mackinawite (FeS). The reaction between dissolved Fe²⁺ and sulfide (S(-II); the sum of H₂S and HS^-) in microbial cultures typically results in the precipitation of mackinawite without further transformation to pyrite (Picard et al., Reference Picard, Gartman and Girguis2016). Therefore, recent studies on biogenic pyrite formation have shifted focus onto indirect sources of Fe(II) coming from microbial or abiotic reduction of Fe(III)-bearing minerals such as Fe(III) (oxyhydr)oxides, silicates and phosphates, which also produce intermediate sulfur species in the system.

Iron(III) (oxyhydr)oxides such as ferrihydrite (Fe(OH)₃), lepidocrocite (γ-FeOOH), goethite (α-FeOOH) and hematite (Fe₂O₃) are common constituents of soils and sediments worldwide (Table 1). In marine sediments and soils that have 3–4 wt.% Fe on average, 20–50% of the Fe exists as Fe(III) (oxyhydr)oxides that are soluble in oxalate, dithionite and/or HCl (Canfield, Reference Canfield1997; Raiswell and Canfield, Reference Raiswell and Canfield1998; Johnson et al., Reference Johnson, Beard and Weyer2020; Pasquier et al., Reference Pasquier, Fike, Révillon and Halevy2022). These minerals exist in various sizes, shapes and crystallinities, which affect their surface areas and reactivity towards sulfide (Poulton et al., Reference Poulton, Krom and Raiswell2004b). This in turn affects the supply rate of Fe(II) and the amount of residual reactive surfaces that enable interfacial reactions for pyrite formation (Peiffer et al., Reference Peiffer, Behrends, Hellige, Larese-Casanova, Wan and Pollok2015; Wan et al., Reference Wan, Schröder and Peiffer2017; Hockmann et al., Reference Hockmann, Planer-Friedrich, Johnston, Peiffer and Burton2020). For example, while ferrihydrite is highly reactive towards sulfide, its surfaces may be dissolved or coated too quickly by FeS or other minerals for pyritization to occur. By contrast, a less reactive mineral such as hematite may react too slowly with sulfide for pyritization to occur at an appreciable rate.

Table 1.

Iron-bearing minerals and their reactivity towards sulfide.

Information compiled from:

Fe(III) (oxyhydr)oxides: Raiswell and Canfield (Reference Raiswell and Canfield1998); Schwertmann and Cornell (Reference Schwertmann and Cornell2000); Poulton et al. (Reference Poulton, Krom and Raiswell2004b); Sklute et al. (Reference Sklute, Kashyap, Dyar, Holden, Tague, Wang and Jaret2018); Caraballo et al. (Reference Caraballo, Asta, Perez and Hochella2022); Jiang et al. (Reference Jiang, Liu, Roberts, Dekkers, Barrón, Torrent and Li2022).

Fe-clays: Langmuir (Reference Langmuir1997); Raiswell and Canfield (Reference Raiswell and Canfield1998); Fan et al. (Reference Fan, Wang, Fu, Li, Liu, Wang and Zhu2023).

Magnetite: Amor et al. (Reference Amor, Mathon, Monteil, Busigny and Lefevre2020); Kappler et al. (Reference Kappler, Thompson and Mansor2023).

Vivianite & ferric phosphates: Huffman et al. (Reference Huffman, Cate and Deming1960); Eynard et al. (Reference Eynard, del Campillo, Barrón and Torrent1992); Hyacinthe and Van Cappellen (Reference Hyacinthe and Van Cappellen2004); Kandori et al. (Reference Kandori, Kuwae and Ishikawa2006); Cosmidis et al. (Reference Cosmidis, Benzerara, Morin, Busigny, Lebeau, Jézéquel, Noël, Dublet and Othmane2014); Rothe et al. (Reference Rothe, Kleeberg and Hupfer2016); Schütze et al. (Reference Schütze, Gypser and Freese2020); Metz et al. (Reference Metz, Kumar, Schenkeveld and Kraemer2023).

Clays are the most abundant mineral host of Fe in nature, making up around 40–50% of total Fe in marine sediments (Raiswell and Canfield, Reference Raiswell and Canfield1998). Iron-bearing clays are thought to be poorly reactive towards sulfide, with slow pyritization possible but occurring in the timescale of hundreds to thousands of years (Raiswell and Canfield, Reference Raiswell and Canfield1996). This view is, however, slowly changing. Layered clay minerals such as illite, smectite and kaolinite contain Fe(III) in either structural, basal or edge sites with different reactivities (Fan et al., Reference Fan, Wang, Fu, Li, Liu, Wang and Zhu2023). Some of these Fe(III) are reducible by sulfide produced by sulfate-reducing microorganisms (SRM) within a timescale of days (Li et al., Reference Li, Vali, Sears, Yang, Deng and Zhang2004; Liu et al., Reference Liu, Dong, Bishop, Zhang, Wang, Xie, Wang, Huang and Eberl2012), making them potentially relevant for rapid pyrite formation (Pasquier et al., Reference Pasquier, Fike, Révillon and Halevy2022). The Fe-S minerals produced from these experiments are poorly characterized. Empirical observations have always noted close associations between diagenetic pyrite (including framboids) and clays (Canfield et al., Reference Canfield, Raiswell and Bottrell1992; Marin-Carbonne et al., Reference Marin-Carbonne, M-N, Havas, Remusat, Pasquier, Alléon, Zeyen, Bouton, Bernard, Escrig, Olivier, Vennin, Meibom, Benzerara and Thomazo2022; Sanz-Montero et al., Reference Sanz-Montero and Pablo Pérez-Soba2009; Wang et al., Reference Wang, Byrne, Perez, Thomas, Göttlicher, Höfer, Mayanna, Kontny, Kappler, Guo, Benning and Norra2020). It is possible that pore spaces with low diffusivity within clays could lead to microenvironments with high supersaturation that encourage pyrite formation.

Besides (oxyhydr)oxides and clays, Fe phosphates are also potential Fe sources for Fe sulfide formation. Recent studies have shown that biogenic pyrite could precipitate via sulfidation of ferric phosphates (Berg et al., Reference Berg, Duverger, Cordier, L-R, Guyot and Miot2020; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020). Ferric phosphates are key minerals involved in phosphorus cycling, especially in ferruginous lakes and estuarine sediments (Hyacinthe and Van Cappellen, Reference Hyacinthe and Van Cappellen2004; Cosmidis et al., Reference Cosmidis, Benzerara, Morin, Busigny, Lebeau, Jézéquel, Noël, Dublet and Othmane2014). Another phosphate mineral, the Fe(II)-containing vivianite, is also commonly formed when dissolved Fe²⁺ and phosphate are released into the solution. Such conditions are found in Fe(III) reducing zones of water columns and sediments or in microbial cultures (Rothe et al., Reference Rothe, Kleeberg and Hupfer2016; Bronner et al., Reference Bronner, Thompson, Dreher, Runge, Voggenreiter, Shuster, Wan, Joshi, Fischer, Duda, Kappler and Mansor2023). Vivianite could be sulfidized to pyrite as the mineral is buried in deeper sulfate reduction zones. Interestingly, vivianite can persist under sulfidic conditions for months in microbial cultures as observed empirically (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018; Nabeh et al., Reference Nabeh, Brokaw and Picard2022; Bronner et al., Reference Bronner, Thompson, Dreher, Runge, Voggenreiter, Shuster, Wan, Joshi, Fischer, Duda, Kappler and Mansor2023). Hence, it is unclear if vivianite is an important Fe source for pyrite formation. Nonetheless, the trends in the last few years show that the research field is shifting towards investigating alternative sources of Fe, with different interfacial chemistry, templating effects and microenvironments that promote a multitude of pathways for pyrite formation.

It is important to keep in mind that minerals with the same chemical formula and crystal structure may also exhibit different reactivities depending on their formation pathways, association with organics, trace metal contents and sizes. Natural minerals will probably have associated organics, especially when they are formed by widespread Fe-metabolizing microorganisms (Kappler et al., Reference Kappler, Bryce, Mansor, Lueder, Byrne and Swanner2021). Organics and trace metals (e.g. Ni) have been shown to retard the extent of sulfidation or pyritization (ThomasArrigo et al., Reference ThomasArrigo, Bouchet, Kaegi and Kretzschmar2020; Duverger et al., Reference Duverger, Bernard, Viennet, Miot and Busigny2021; Wang et al., Reference Wang, Wang, Wang, Kumar, Pan, Peiffer and Wang2023; Wu et al., Reference Wu, Zhang, Lanson, Yin, Cheng, Liu and He2023). Particle sizes also greatly affect pyritization rates. For example, micron-sized magnetite (mixed-valent Fe₃O₄) grains exhibit low reactivity towards sulfide with estimated half-lives in the order of 100 years (Canfield and Berner, Reference Canfield and Berner1987). By contrast, freshly precipitated nano-magnetite that are more similar to biogenic magnetite are rapidly sulfidized within days, with pyrite formation accelerated in the presence of elemental sulfur (S⁰) and polysulfides (Poulton et al., Reference Poulton, Krom and Raiswell2004b; Runge et al., Reference Runge, Mansor, Shuster, Fischer, Liu, Lunter, Kappler and Duda2023), while being retarded in the presence of organics (Runge et al., Reference Runge, Mansor, Chiu, Shuster, Fischer, Kappler and J-P2024).

Sources of sulfur

Mackinawite and greigite are Fe-S minerals in which both sulfur atoms have -2 redox states, while pyrite is an Fe polysulfide in which the sulfur atom has a -1 redox state. The availability of reduced sulfur species is tied to complex sulfur cycling driven by microbial and abiotic processes. Below, we discuss the sources of sulfide, polysulfides and S⁰ as the main sulfur-bearing species involved in sedimentary Fe-S mineral formation.

Sulfide (S(-II)) is produced in porewater as the main product of microbial sulfate reduction (MSR), the dominant mode of organic matter oxidation in marine sediments (Jørgensen et al., Reference Jørgensen, Findlay and Pellerin2019), where it is fuelled by sulfate ions present at relatively high concentrations in seawater (~28 mM) and diffusing downwards into the sediment. Sulfide speciates into two major species: H₂S and bisulfide (HS^-). The pKa of H₂S is about 7, meaning that at normal seawater pH (around 8), it represents less than 10% of the total dissolved S(-II) pool. The majority of the pool will then be in the form of HS^-. The link between MSR and pyrite formation is potentially weaker in low-sulfate lake sediments, where S(-II) may be derived from the breakdown of biomass containing reduced organic sulfur (Wei et al., Reference Wei, Yin, Kappler, Tao and Zhu2023). The breakdown of reduced organic sulfur, as well as the reduction of organic sulfite, is also thought to have been a prevalent source of reduced sulfur for pyrite formation in low-sulfate Archean oceans (Fakhraee and Katsev, Reference Fakhraee and Katsev2019). Microbial disproportionation or reduction of elemental sulfur (S⁰) and thiosulfate may also constitute a source of S(-II) for pyrite formation, as suggested by both microbial culture experiments (Canfield et al., Reference Canfield, Thamdrup and Fleischer1998) and observations from natural marine sediments (Zopfi et al., Reference Zopfi, Böttcher and Jørgensen2008).

Polysulfides are soluble ionic species consisting of relatively short chains of sulfur atoms terminated by negative charges (n in S_n^2- is typically smaller than 9, with n = 5, 6, and 4 being the more abundant forms under certain experimental conditions; Kamyshny et al., Reference Kamyshny, Goifman, Gun, Rizkov and Lev2004). At high concentrations, they exhibit a distinct yellow-green colouration with absorbances in the 250–350 nm region (e.g. Domingos et al., Reference Domingos, Runge, Dreher, T-H, Shuster, Fischer, Kappler, J-P, Xu and Mansor2023). At pH values representative of seawater and marine pore waters, polysulfides form spontaneously through the chemical reaction of S⁰ with S(-II), as depicted in the equation below:

(1) $$ {\mathrm{H}\mathrm{S}}^{{\textstyle \hbox{-} \hbox{-} }}+( n{\textstyle \hbox{--}}1)\hskip2pt {\mathrm{S}}^0={\mathrm{H}}^{+}+{{\mathrm{S}}_{\mathrm{n}}}^{2{\textstyle \hbox{--} }} $$

Polysulfides can also form through the oxidation of hydrogen sulfide by O₂, Fe(III) and manganese (oxyhydr)oxides, with kinetic reaction rates in the following order: MnO₂ > O₂ > Fe(OH)₃ > S⁰, corresponding to characteristic reaction times ranging from a few minutes to about a year at conditions relevant for sulfidic marine sediments (Chen and Morris, Reference Chen and Morris1972; Poulton et al., Reference Poulton, Fralick and Canfield2004a; Avetisyan et al., Reference Avetisyan, Buchshtav and Kamyshny2019, Reference Avetisyan, Zweig, Luther and Kamyshny2021). They can furthermore be formed biologically, as a by-product of both phototrophic and chemotrophic S-oxidative microbial metabolisms (Findlay, Reference Findlay2016). The environmental prevalence and relative abundance of polysulfide species in the environment have remained relatively elusive until the development of robust analytical methods based on derivatization almost twenty years ago (Kamyshny et al., Reference Kamyshny, Ekeltchik, Gun and Lev2006, Reference Kamyshny, Borkenstein and Ferdelman2009), and even today reliably measuring polysulfide concentrations in sediments is a difficult endeavour. However, it has been established that under many conditions, polysulfides can be expected to occur in concentrations approaching calculated equilibrium with elemental sulfur based on reaction (1) (Kamyshny and Ferdelman, Reference Kamyshny and Ferdelman2010). For this reason, we focus our discussion on factors controlling the abundance of sedimentary S⁰ in the paragraph below.

In low-temperature environments, S⁰ is formed by oxidation of more reduced sulfur species. This oxidation process can occur abiotically in the presence of oxygen or oxidized Fe or Mn phases (e.g. Rickard and Luther, Reference Rickard and Luther2007) or it can be mediated by S-oxidizing bacteria and archaea (Dahl et al., Reference Dahl, Schulte, Stockdreher, Hong, Grimm, Sander, Kim, Kim and Shin2008). Since measured rates of prokaryotic S-oxidation are several orders of magnitudes faster than rates of chemical sulfide oxidation to S⁰ by molecular oxygen (Luther et al., Reference Luther, Findlay, MacDonald, Owings, Hanson, Beinart and Girguis2011), it is often assumed that S⁰ formed in low-temperature environments is mostly formed as a result of microbial activity. A diversity of phototrophs and chemotrophs are able to biomineralize S⁰ in the form of intra- or extra-cellular S⁰ globules (Dahl and Prange, Reference Dahl, Prange and Shively2006; Cron et al., Reference Cron, Henri, Chan, Macalady and Cosmidis2019; Marnocha et al., Reference Marnocha, Sabanayagam, Modla, Powell, Henri, Steele, Hanson, Webb and Chan2019) or extracellular S⁰ filaments (Wirsen et al., Reference Wirsen, Sievert, Cavanaugh, Molyneaux, Ahmad, Taylor, DeLong and Taylor2002; Sievert et al., Reference Sievert, Wieringa, Wirsen and Taylor2007). Sulfur rarely accumulates in sedimentary environments, due to its chemical and biological instability. Chemically, S⁰ is only stable in a very narrow range of Eh-pH conditions, and not at all above neutral pH values (Rickard and Luther, Reference Rickard and Luther2007). It has actually been found to be thermodynamically unstable in a range of natural sulfidic waters (Helz, Reference Helz2014). Biologically, S⁰ is used as a source of energy for a diverse range of S-oxidizers, S-reducers and microorganisms that perform S⁰ disproportionation (Dahl, Reference Dahl and Lens2020a). Some microbes, such as the thermoacidophile Acidianus, are particularly efficient at recycling S⁰ as they can grow from all three reactions (Amenabar and Boyd, Reference Amenabar and Boyd2018). Abundances of S⁰ in sediments typically range from 11 μmol/kg to 60 mmol/kg (see compilations in Ye and Jing (Reference Ye and Jing2022), their Table S-1 and Zopfi et al. (Reference Zopfi, Ferdelman and Fossing2004).

Not all forms of S⁰ are created equal, as a number of factors can influence the chemical and biological reactivity of S⁰ in the environment. Obviously, S⁰ biominerals stored intracellularly (Dahl, Reference Dahl and Jendrossek2020b) are unavailable for extracellular chemical reactions and consumption by other microorganisms. Some microbes that produce S⁰ biominerals as an extracellular energy storage resource have also evolved strategies to increase their environmental stability and/or avoid piracy by other S⁰-consuming cells (Cosmidis and Benzerara, Reference Cosmidis and Benzerara2022). As an example, some S-oxidizers encapsulate their extracellular S⁰ globules in an organic membrane, allowing S⁰ to be formed and persist in the extracellular medium in a thermodynamically unstable state (Cron et al., Reference Cron, Henri, Chan, Macalady and Cosmidis2019, Reference Cron, Macalady and Cosmidis2021; Marnocha et al., Reference Marnocha, Sabanayagam, Modla, Powell, Henri, Steele, Hanson, Webb and Chan2019). Sulfur minerals produced chemically by oxidation of sulfide in the presence of organics (a process called S⁰ organomineralization) can also exist outside of their thermodynamic stability domain (Lau et al., Reference Lau, Cosmidis, Grasby, Trivedi, Spear and Templeton2017; Cosmidis et al., Reference Cosmidis, Nims, Diercks and Templeton2019). The size of S⁰ particles affects their chemical reactivity (Steudel, Reference Steudel and Steudel2003) and bioavailability (Franz et al., Reference Franz, Lichtenberg, Hormes, Modrow, Dahl and Prange2007), with smaller particles being overall more unstable. In sediments, S⁰ is thought to exist mainly as colloidal sols, which are more reactive than crystalline S⁰ (Zopfi et al., Reference Zopfi, Ferdelman and Fossing2004). Due to its reactivity, colloidal or nanoparticulate S⁰ is likely to be the main source of polysulfides in sulfidic sediments (Kleinjan et al., Reference Kleinjan, De Keizer and Janssen2005; Mol et al., Reference Mol, Pruim, de Korte, Meuwissen, van der Weijden, Klok, Keesman and Buisman2022), but abundances of S⁰ particles in the micron or sub-micron size range are rarely reported (Findlay et al., Reference Findlay, Gartman, Macdonald, Hanson, Shaw and Luther2014). Such considerations on S⁰ reactivity should be taken into account in experimental studies investigating the role of this mineral in Fe-S mineral formation.

Mackinawite and greigite formation at low temperature

Mackinawite (FeS) and greigite (Fe₃S₄) are described as metastable Fe-S minerals with respect to pyrite (cubic FeS₂). For that reason, they are generally assumed to be present in modern sedimentary environments but absent from ancient rocks and sediments (Rickard, Reference Rickard2012b). Recent modelling studies have nonetheless found that greigite could be much more stable than originally expected (Subramani et al., Reference Subramani, Lilova, Abramchuk, Leinenweber and Navrotsky2020; Shumway et al., Reference Shumway, Wilson, Lilova, Subramani, Navrotsky and Woodfield2022; Son et al., Reference Son, Hyun, Charlet and Kwon2022). In natural environments, metastable Fe-S minerals are assumed to be the main constituent of acid-volatile sulfides (AVS), which are the solid and aqueous phases that produce sulfide after treatment of samples with HCl. The contribution of mackinawite and greigite to the AVS fraction is likely to vary depending on the environment considered and it is possible that the AVS fraction does not capture all of the mackinawite and greigite content of an environment, as these two minerals might not completely dissolve in HCl (Rickard and Morse, Reference Rickard and Morse2005). While direct identification of mackinawite and greigite is most easily done using X-ray diffraction (XRD) (Berner, Reference Berner1962; Evans Jr et al., Reference Evans, Milton, Chao, Adler, Mead, Ingram and Berner1964; Skinner et al., Reference Skinner, Erd and Grimaldi1964; Lennie, Reference Lennie1995), combined studies including AVS analysis and XRD in sedimentary environments are not common. Biogenic precipitates of mackinawite and greigite are certainly present in anoxic sediments, and understanding their physical properties is of crucial importance because they could react differently from abiotic precipitates to analytical procedures, such as leaching procedures for AVS quantification.

Abiotic mackinawite has a tetragonal layer structure with cell parameters a=b=3.6735 Å and c=5.0329 Å (Lennie, Reference Lennie1995). Greigite has a cubic unit cell, with a=9.876 Å (Skinner et al., Reference Skinner, Erd and Grimaldi1964). Mackinawite is stoichiometric FeS and precipitates rapidly from the reaction between aqueous Fe²⁺ and dissolved sulfide (H₂S/HS^-) (Rickard, Reference Rickard1995, Reference Rickard2024; Rickard et al., Reference Rickard, Griffith, Oldroyd, Butler, Lopez-Capel, Manning and Apperley2006). The chemical formula for the greigite formula is averaged to Fe₃S₄, although its exact composition has not been determined (Rickard, Reference Rickard2012b). Greigite forms through the solid-state transformation of mackinawite, which is driven by the oxidation of Fe atoms and their rearrangement (Rickard, Reference Rickard2012b, Lennie et al., Reference Lennie, Redfern, Champness, Stoddart, Schofield and Vaughan1997). As greigite does not precipitate directly from the solution, its characterization is difficult because residual mackinawite is always present (Rickard, Reference Rickard2012b). Owing to its magnetic properties, greigite can be detected in the sedimentary record but could be difficult to differentiate from magnetite (Fe₃O₄) (Roberts et al., Reference Roberts, Chang, Rowan, C-S and Florindo2011). In experimental studies and microbial cultures, a strong neodymium magnet can be used to check for its presence.

In low-temperature environments, the main source of sulfide for mackinawite and greigite formation is microbial sulfate reduction (MSR) (Rickard et al., Reference Rickard, Mussmann and Steadman2017). As discussed in the ‘Sources of sulfur’ section above, other microbial processes can provide sulfide for metastable Fe-S mineral formation, but their significance varies depending on the environment (e.g. Jørgensen et al., Reference Jørgensen, Findlay and Pellerin2019; Wu et al., Reference Wu, Liu, Fang, Yang, Chen, He and Wang2021). Mesophilic sulfate-reducing bacteria (SRB) have been used in most experimental studies to decipher the role of microorganisms in the formation of extracellular Fe-S minerals at low temperatures, e.g. through biologically induced mineralization (Pósfai and Dunin-Borkowski, Reference Pósfai and Dunin-Borkowski2006; Picard et al., Reference Picard, Gartman and Girguis2016; Park and Faivre, Reference Park and Faivre2022). The few sulfate-reducing archaea (SRA) available in cultures are (hyper)thermophilic and have not been used for Fe-S mineral formation investigations. Experimental studies consist of precipitating biogenic Fe-S minerals by adding a source of Fe to cultures of SRB and (ideally) comparing them to abiotic Fe-S minerals precipitated by adding sulfide to the culture medium containing the same source of Fe. The following strains have been used in a range of temperatures between room temperature and 35°C: Desulfovibrio capillatus (Ikogou et al., Reference Ikogou, Ona-Nguema, Juillot, Le Pape, Menguy, Richeux, Guigner, Noël, Brest, Baptiste and Morin2017), Desulfovibrio desulfuricans (Rickard, Reference Rickard1969b; Neal et al., Reference Neal, Techkarnjanaruk, Dohnalkova, McCready, Peyton and Geesey2001; Li et al., Reference Li, Vali, Yang, Phelps and Zhang2006; Stanley and Southam, Reference Stanley and Southam2018; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020), Desulfovibrio hydrothermalis (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018; Nabeh et al., Reference Nabeh, Brokaw and Picard2022), Desulfovibrio magneticus (Nabeh et al., Reference Nabeh, Brokaw and Picard2022), Desulfovibrio vulgaris (Zhou et al., Reference Zhou, Vannela, Hayes and Rittmann2014; Mansor et al., Reference Mansor and Fantle2019), Desulfovibrio spp. (Li et al., Reference Li, Vali, Sears, Yang, Deng and Zhang2004), Desulfosporosinus orientis (Stanley and Southam, Reference Stanley and Southam2018), Desulfotomaculum sp. (Fortin et al., Reference Fortin, Southam and Beveridge1994). Additionally, uncharacterized enrichments of SRB from various environments have also been used in experimental studies (Herbert et al., Reference Herbert, Benner, Pratt and Blowes1998; Donald and Southam, Reference Donald and Southam1999; Gramp et al., Reference Gramp, Bigham, Jones and Tuovinen2010). The redox state of the initial source of Fe in cultures of SRB appears to determine the mineralogy of the final Fe-S mineral products (Rickard, Reference Rickard1969b; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020). In experimental studies using Fe²⁺ as a unique source of Fe, mackinawite precipitates first and transforms into greigite over time (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). In experimental studies using Fe(III), the reduction of the latter by sulfide produces Fe(II) and intermediate sulfur species, which appear necessary for the formation of pyrite (see the section ‘Biogenic pyrite formation at low temperatures’ below).

Owing to the semi-conducting properties of Fe-S minerals, the role of mackinawite in improving the efficiency of extracellular electron transfer in microbial cultures has been investigated (Nakamura et al., Reference Nakamura, Okamoto, Tajima, Newton, Kai, Takashima and Hashimoto2010; Jiang et al., Reference Jiang, Hu, Lieber, Jackan, Biffinger, Fitzgerald, Ringeisen and Lieber2014; Kondo et al., Reference Kondo, Okamoto, Hashimoto and Nakamura2015; Zhu et al., Reference Zhu, Huang, Ni, Tang, Zhu, Long-er and Zou2022). These studies have explored the production of biogenic Fe-S minerals by Fe(III)-reducing bacteria, e.g. Shewanella or Geobacter, which also have the ability to reduce intermediate sulfur species. In those experimental systems, microbial reduction of thiosulfate or elemental sulfur produces sulfide, while microbial and/or chemical reduction of Fe(III) by sulfide is the source of Fe(II) that promotes the precipitation of Fe-S minerals. Although many studies of interest also considered biogenic mackinawite in the context of bioremediation studies (e.g. Sharma et al., Reference Sharma, Yan, Feng, Xu, Pan, Kong and Li2024; Yang et al., Reference Yang, Chen, Sumona, Gupta, Sun, Hu and Zhan2017), we restrict this section to studies that have attempted to characterize and quantify the differences between abiotic and biogenic mackinawite and greigite experimentally.

Microbial influence on the physical characteristics of mackinawite

The presence of microorganisms in experimental systems does not prevent the formation of mackinawite. However, the availability of cell surfaces as templates for mineral nucleation and growth can impact its physical properties, such as the size of crystallite domains, crystallinity and propensity to aggregate, and can lead to cell encrustation (Picard et al., Reference Picard, Gartman and Girguis2016, Reference Picard, Gartman, Clarke and Girguis2018, Reference Picard, Gartman and Girguis2021; Mansor et al., Reference Mansor, Berti, Hochella, Murayama and Xu2019; Nabeh et al., Reference Nabeh, Brokaw and Picard2022). Templating occurs when Fe²⁺ first interacts with negatively charged bacterial cell surfaces before precipitating with sulfide (Beveridge, Reference Beveridge1989); i.e. when SRBs are grown in a culture medium that contains millimolar concentrations of Fe²⁺ (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). Physical and chemical characteristics are important when considering reactivity and transformation of solid phases in natural environments.

X-ray diffractograms of ‘templated’ minerals display fine peaks with high intensity, while those of ‘non-templated’ minerals display broad peaks with low intensity (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018; Mansor et al., Reference Mansor, Berti, Hochella, Murayama and Xu2019; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020, Reference Duverger, Bernard, Viennet, Miot and Busigny2021, Nabeh et al., Reference Nabeh, Brokaw and Picard2022). For a quantitative approach, XRD data can be fitted to estimate the average size of the crystalline domains using the Scherrer equation (Wolthers et al., Reference Wolthers, Van der Gaast and Rickard2003; Unruh and Forbes, Reference Unruh, Forbes, Kenney, Veeramani and Alessi2019). Abiotic mackinawite precipitated in water is nanocrystalline and has an average particle size of 7.4 nm (a/b axis) x 2.9 nm (c axis) (Wolthers et al., Reference Wolthers, Van der Gaast and Rickard2003; Ohfuji and Rickard, Reference Ohfuji and Rickard2006). While there is variability among the few studies that reported crystallite size data, biogenic mackinawite grows along both the a/b axis and the c axis more than the abiotic controls (Zhou et al., Reference Zhou, Vannela, Hayes and Rittmann2014; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018; Mansor et al., Reference Mansor, Berti, Hochella, Murayama and Xu2019). Specifically, crystallite domains of biogenic mackinawite precipitated with SRB grown with Fe²⁺, and of mackinawite precipitated with dead SRB incubated with Fe²⁺ then with sulfide, are on average significantly larger than those of abiotic mackinawite precipitated in water and the SRB medium (with or without simple or complex organic molecules), and those of biogenic mackinawite precipitated in SRB cultures to which Fe²⁺ has been added after growth and sulfide production (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018, Reference Picard, Gartman and Girguis2021). When SRBs are grown with Fe(III)-citrate, biogenic mackinawite displays smaller crystallite domains than when Fe²⁺ is directly available (Ikogou et al., Reference Ikogou, Ona-Nguema, Juillot, Le Pape, Menguy, Richeux, Guigner, Noël, Brest, Baptiste and Morin2017). In all conditions in which crystallite domains are small, templating cannot occur for the two following reasons: (1) salts and organic molecules in the medium do not provide scaffolds for mineral nucleation and growth and (2) Fe²⁺ does not have the opportunity to bind to cell surfaces if sulfide is already present in the environment. A number of other studies have investigated the formation of biogenic Fe-S sulfide minerals using Fe(III) minerals as starting Fe source (e.g. Fe(III) (oxyhydr)oxides, Fe(III)-containing clays, Fe(III) phosphate minerals); however, they did not provide information about the size of crystallite domains (Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020).

High-resolution studies of biogenic mackinawite indicated that it becomes crystalline very rapidly. Selected-area electron diffraction (SAED) patterns showed a shift from a polycrystalline material after one week to a single-crystal pattern after one month in mackinawite produced in cultures of Desulfovibrio desulfuricans grown with Fe²⁺. As early as one week of incubation, and in longer experiments, the 5 Å d-spacing of mackinawite can be clearly observed using high-resolution transmission electron microscopy (HR-TEM) (Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020). It is usually assumed in abiotic studies that mackinawite is transient and short-lived in anoxic sedimentary environments. The crystallinity of biogenic mackinawite will probably play an important role in its stability and further transformations and should be considered.

Aggregation of biogenic mackinawite particles in cultures of SRB grown with Fe²⁺ is visible when observing cultures by eye. Minerals observed in unshaken cultures of SRB grown with Fe(II) form sticky clumps, while abiotic mackinawite sediments homogeneously at the bottom of serum vials. When biogenic minerals are resuspended in solutions, they also appear less opaque than abiotic minerals (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). Observations using scanning and/or transmission electron microscopy (SEM and/or TEM) reveal that biogenic mackinawite has a flaky texture and can reach mm-range sizes, much larger than aggregates of abiotic mackinawite (Herbert et al., Reference Herbert, Benner, Pratt and Blowes1998; Gramp et al., Reference Gramp, Bigham, Jones and Tuovinen2010; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020). Biogenic particles precipitated in cultures of SRB grown with Fe²⁺ form larger aggregates than abiotic particles precipitated in the culture medium, as determined using dynamic light scattering (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018).

Cell encrustation by Fe-S minerals has been reported in cultures of SRB grown with Fe²⁺. This observation is consistent with the ability of cells to serve as templates, e.g. when Fe²⁺ is available to bind to the cell surface before interacting with sulfide. Cells from cultures of SRB grown with and without Fe(II) have a similar smooth aspect when imaged with SEM (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). To detect the presence of Fe-S crusts at the surface of cells, it is best to prepare thin sections of resin-embedded pellets that contain cells and minerals and to image them using TEM (Shuster et al., Reference Shuster, Reith, Southam, Kenney, Veeramani and Alessi2019; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). Fe-S minerals have been directly observed at the surface of both gram-negative and gram-positive SRB using TEM imaging of thin sections prepared after short incubations (Fortin et al., Reference Fortin, Southam and Beveridge1994; Donald and Southam, Reference Donald and Southam1999; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018; Stanley and Southam, Reference Stanley and Southam2018). Although no direct mineral characterization of the crusts has been performed in these studies using SAED, encrusted cells were imaged after short periods of time (one week or less), when mackinawite is the only mineral phase detected by powder X-ray diffraction (XRD) (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). It is unknown if cell encrustation in Fe-S minerals is a common situation for microbial cells in natural environments. There have been only two reports of microorganisms encrusted with Fe-S minerals in mine tailing sediments, where SRB are ubiquitous (Ferris et al., Reference Ferris, Fyfe and Beveridge1987; Fortin and Beveridge, Reference Fortin and Beveridge1997). The impact of Fe-S mineral encrustation on microbial metabolic activity is also unknown. The assimilation of carbon substrates by bacteria encrusted in Fe(III) oxyhydroxides appears to be inhibited (Miot et al., Reference Miot, Remusat, Duprat, Gonzalez, Pont and Poinsot2015). Duverger et al. (Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020) suggested that sulfate reduction in cultures of Desulfovibrio desulfuricans grown with Fe²⁺ is hindered by cell encrustation in Fe-S minerals. It is possible that the high concentration of soluble Fe²⁺ (20 mM) used in their study, rather than encrustation, inhibited metabolic activity. Indeed, other studies that described cell crusts around cells used lower concentrations of Fe²⁺ (e.g. 3.5–4.0 mM) and did not report inhibition of sulfate reduction (Fortin et al., Reference Fortin, Southam and Beveridge1994; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). Although Fe(II) is required by SRB for growth, high concentrations of soluble Fe²⁺ might trigger stress or other metabolic responses. SRB released seven unidentified organic molecules in response to 4 mM Fe²⁺ in the growth medium (Picard et al., Reference Picard, Gartman, Cosmidis, Obst, Vidoudez, Clarke and Girguis2019). It is possible that (1) these extracellular compounds are inhibitory at higher concentrations, and/or that (2) high concentrations of Fe²⁺ are directly inhibitory for SRB. As Fe-S minerals are semi-conducting materials, some studies have evaluated the potential role that biogenic Fe-S minerals could play in enhancing extracellular electron transfer (EET) between microorganisms and solid phases. Direct contact between SRB and Fe-S minerals might be beneficial, and electrochemical studies indicated an increase in electron transfer from cells to electrodes in the presence of biogenic Fe-S minerals (Deng et al., Reference Deng, Dohmae, Kaksonen and Okamoto2020). This observation has also been reported in studies in which Shewanella strains produced Fe-S minerals when grown with thiosulfate and a source of Fe (Nakamura et al., Reference Nakamura, Okamoto, Tajima, Newton, Kai, Takashima and Hashimoto2010; Jiang et al., Reference Jiang, Hu, Lieber, Jackan, Biffinger, Fitzgerald, Ringeisen and Lieber2014; Kondo et al., Reference Kondo, Okamoto, Hashimoto and Nakamura2015). The facilitated electron transfer through biogenic Fe-S minerals could be of use in energy-depleted environments. As discussed later in this review, Fe-S minerals can store significant amounts of organic carbon that could be potentially accessed as energy sources by SRB and other microorganisms (Picard et al., Reference Picard, Gartman, Cosmidis, Obst, Vidoudez, Clarke and Girguis2019; Nabeh et al., Reference Nabeh, Brokaw and Picard2022).

Microbial influence on the transformation of mackinawite to greigite under anoxic conditions

In abiotic experiments, greigite formation from mackinawite requires an oxidant, which can be traces of oxygen at the surface of mackinawite (Benning et al., Reference Benning, Wilkin and Barnes2000), aldehydes (with the intention of sterilizing the experimental system) (Rickard et al., Reference Rickard, Butler and Oldroyd2001), or polysulfides (Benning et al., Reference Benning, Wilkin and Barnes2000). High temperature accelerates the abiotic transformation of mackinawite to greigite (Lennie, Reference Lennie1995). A recent modelling study indicated that greigite formation is favourable in anoxic, alkaline and low-temperature environments where Fe is enriched and sulfide limited (Turney et al., Reference Turney, Weiss, Muxworthy and Fraser2023). There is experimental evidence that SRM could be driving and/or accelerating the transformation of mackinawite to greigite under strictly anoxic conditions. As noted above, greigite can form in the absence of microorganisms. However, in experimental studies under strict anoxic conditions, greigite forms in cultures of SRM grown with Fe²⁺ at their optimal pH and temperature after several months of incubation, while it does not form in abiotic experiments (Rickard, Reference Rickard1969b; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018, Reference Picard, Gartman and Girguis2021; Mansor et al., Reference Mansor, Berti, Hochella, Murayama and Xu2019; Nabeh et al., Reference Nabeh, Brokaw and Picard2022). In old cultures, greigite is stable and does not transform further, and mackinawite is still detectable after several years of incubation, suggesting that the full transformation of mackinawite to greigite in these experimental conditions is slow (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018; Nabeh et al., Reference Nabeh, Brokaw and Picard2022; Picard, unpublished). In the study by Mansor et al. (Reference Mansor, Berti, Hochella, Murayama and Xu2019), greigite could be detected by SAED in the TEM after six months of incubation but not by XRD, suggesting that transformation yield can vary with experimental conditions and microbial strains. In abiotic studies that maintain strict anoxic conditions, mackinawite remains stable and does not transform into greigite (Benning et al., Reference Benning, Wilkin and Barnes2000; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018, Reference Picard, Gartman and Girguis2021).

Biogenic greigite produced in cultures of SRB displays an average crystalline domain size of 19.2 nm, which is in the size range of biogenic mackinawite precipitated in cultures of SRB grown with Fe²⁺, supporting the hypothesis of solid-state transformation from biogenic mackinawite (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). The ‘small’ mackinawite produced in cultures to which Fe²⁺ has been added after sulfide production does not transform to greigite, nor does abiotic mackinawite precipitated with organic molecules or mixtures of complex organics (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018, Reference Picard, Gartman and Girguis2021). Interestingly, the ‘large’ mackinawite that is produced at the surface of dead cells also does not transform into greigite (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). This indicates that under strict anoxic conditions, crystalline mackinawite and metabolically active cells are required to form greigite. The oxidant necessary to oxidize Fe(II) in mackinawite could originate from H⁺ produced by residual activity of SRB in old cultures (Mansor et al., Reference Mansor, Berti, Hochella, Murayama and Xu2019). However, it is unknown why the transformation only takes place after several months, when greigite precipitation is actually favourable in anoxic and alkaline conditions at low temperatures (Turney et al., Reference Turney, Weiss, Muxworthy and Fraser2023). The experiments cited above have been performed in anoxic conditions, around neutral pH and at temperatures between RT and 35°C (Rickard, Reference Rickard1969b; Zhou et al., Reference Zhou, Vannela, Hayes and Rittmann2014; Picard et al., Reference Picard, Gartman, Clarke and Girguis2018, Reference Picard, Gartman and Girguis2021; Mansor et al., Reference Mansor, Berti, Hochella, Murayama and Xu2019).

There are still many unknowns into what controls greigite formation in low-temperature environments. Changes in one or several parameters in culture media can promote or suppress the formation of greigite. For example, varying the amount of sulfide produced in cultures of Desulfovibrio hydrothermalis, and of Desulfovibrio magneticus RS-1 grown with Fe(II) resulted in different patterns of greigite production (Nabeh et al., Reference Nabeh, Brokaw and Picard2022). A recent report of the formation of greigite in cultures of Geobacter sulfurreducens grown with Fe(III) and S(0) suggests that other conditions than the ones described above might exist to promote greigite formation in the presence of microorganisms (Bronner et al., Reference Bronner, Thompson, Dreher, Runge, Voggenreiter, Shuster, Wan, Joshi, Fischer, Duda, Kappler and Mansor2023). The encrustation of cells by greigite has not been reported. As noted in the previous section, encrustation has been reported in relatively short-term cultures when mackinawite is the only phase detected by XRD. It is thus unknown how the interactions between SRM and Fe-S minerals evolve and if the transformation of mackinawite to greigite affects the localization of minerals on microbial cells.

While its significance in terms of amounts precipitated is unknown, biogenic greigite produced intracellularly by magnetotactic bacteria (MTB) through biologically controlled mineralization (as opposed to biologically induced mineralization described above) has attracted attention for their potential to participate in the magnetic properties of sediments and sedimentary rocks and to produce biosignatures (Amor et al., Reference Amor, Mathon, Monteil, Busigny and Lefevre2020). Two morphological types of intracellular greigite producers have been described: magnetotactic multicellular prokaryotes (MMPs) and large rod-shaped bacteria. Intracellular greigite producers are found in reducing and sulfidic environments and use sulfate as an electron acceptor (Farina et al., Reference Farina, Esquivel and Lins de Barros1990; Mann et al., Reference Mann, Sparks, Frankel, Bazylinski and Jannasch1990; Lefèvre et al., Reference Lefèvre, Menguy, Abreu, Lins, Pósfai, Prozorov, Pignol, Frankel and Bazylinski2011; Descamps et al., Reference Descamps, Monteil, Menguy, Ginet, Pignol, Bazylinski and Lefèvre2017). Intracellular greigite formation is favoured at low Fe and high sulfide concentrations (Descamps et al., Reference Descamps, Monteil, Menguy, Ginet, Pignol, Bazylinski and Lefèvre2017). Similar to extracellular greigite, intracellular crystals in MTB form from the solid-state transformation of mackinawite (Pósfai et al., Reference Pósfai, Buseck, Bazylinski and Franke1998). Mackinawite can be detected at the end of magnetosome chains during short periods of time before transforming to greigite in a matter of days to weeks, suggesting that the intracellular formation of greigite is faster than that of extracellular greigite (Pósfai et al., Reference Pósfai, Buseck, Bazylinski and Franke1998).

Biogenic pyrite formation at low temperatures

Pyrite is the most widespread iron sulfide on the Earth’s surface and is often ascribed a biological origin, as microbial sulfate reduction (MSR) is the main pathway to produce sulfide precursors from the particularly stable molecule sulfate. Chemically speaking, pyrite is one of two iron disulfide (FeS₂) polymorphs, along with marcasite. Pyrite’s ubiquitous presence in the geological record is due to its incredible thermodynamic stability in comparison to other reduced sulfur minerals (Schoonen, Reference Schoonen2004). It is thus tempting to assume that its formation results from the simple precipitation of its constituent cation (Fe²⁺) and anion (S₂^2-), as is the case for most non-silicate authigenic sedimentary minerals (e.g. halides, sulfates, carbonates, phosphates). It was therefore assumed that pyrite is relatively easy to obtain in laboratory experiments. Indeed, a long list of recipes for pyrite formation can be found in the literature, as compiled in Rickard (2012).

However, the conditions used to precipitate pyrite in the laboratory are often far removed from those found in natural sediments, particularly those supporting microbial life. Indeed, many experiments have been carried out at so-called ‘low temperatures’ of <100°C, which are lower than hydrothermal temperatures but not so low in a biological sense. In fact, temperatures above 45°C and 80°C are in the ecological niche ranges of thermophiles and hyperthermophiles, respectively, which are not responsible for most MSR globally. Similarly, experiments carried out at pH levels far from neutral raise questions about the parallels that can be drawn with the natural environment. All these abiotic syntheses have been crucial to understanding the mechanisms of pyrite formation and the different environmental factors at play. However, they do not allow us to correctly decipher the role played by microorganisms in pyrite formation.

Pyrite formation in the presence of microbes has more recently been reported in environmental studies of sediments (Quevedo et al., Reference Quevedo, Jiménez-Millán, Cifuentes, Gálvez, Castellanos-Rozo and Jiménez-Espinosa2021; Tribovillard et al., Reference Tribovillard, Bout-Roumazeilles, Delattre, Ventalon and Bensadok2022), thermal springs (Tsyrenova et al., Reference Tsyrenova, Barkhutova, Buryukhaev, Lazareva, Bryanskaya and Zamana2018) and deep geological repositories (Boylan et al., Reference Boylan, Perez-Mon, Guillard, Burzan, Loreggian, Maisch, Kappler, Byrne and Bernier-Latmani2019), as well as in numerous recent experimental studies such as in metal corrosion (Etim et al., Reference Etim, Wei, Dong, Xu, Chen, Wei, Su and Ke2018; Jia et al., Reference Jia, Tan, Jin, Blackwood, Xu and Gu2018, Reference Jia, Wang, Jin, Unsal, Yang, Yang, Xu and Gu2019; Rasheed et al., Reference Rasheed, Jabbar, Rasool, Pandey, Sliem, Helal, Samara, Abdullah and Mahmoud2019), arsenic bioremediation in contaminated soils (Pi et al., Reference Pi, Wang, Xie, Ma, Liu, Su, Zhu and Wang2017; Saunders et al., Reference Saunders, M-K, Dhakal, Ghandehari, Wilson, Billor and Uddin2018; Lee et al., Reference Lee, Saunders, Wilson, Levitt, Ghandehari, Dhakal, Redwine, Marks, Billor, Miller, Han and Wang2019), various microbial enrichment cultures (Gao et al., Reference Gao, Jiang, Guo, Zeng, Fan, Zhang, Reinfelder, Huang, Lu and Dang2019, Reference Gao, Zheng, Deng and Jiang2021; Berg et al., Reference Berg, Duverger, Cordier, L-R, Guyot and Miot2020; Allen et al., Reference Allen, Wegener, Sublett, Bodnar, Feng, Wendt and White2021; Ikkert et al., Reference Ikkert, Ivanov, Ukhova, Zuysman, Glukhova, Avakyan and Karnachuk2021; Karnachuk et al., Reference Karnachuk, Ikkert, Avakyan, Knyazev, Volochaev, Zyusman, Panov, Kadnikov, Mardanov and Ravin2021; Wei et al., Reference Wei, Yin, Kappler, Tao and Zhu2023; Yang et al., Reference Yang, Cheng and Zhou2023) and (co-)cultures (Thiel et al., Reference Thiel, Byrne, Kappler, Schink and Pester2019; Zhou et al., Reference Zhou, Gao, Xie, Wang, Yue, Wei, Yang, Li and Chen2019; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020; Sun et al., Reference Sun, Feng, Zheng, Kong, Wu, Zhang, Zhou and Liu2023; Ke et al., Reference Ke, Deng, Zhang, Ren, Liu, He, Wu, Dang and Guo2024). It should be pointed out that the role of microorganisms in the formation of biogenic pyrite was not the main objective of some of these studies and, therefore, although anoxic conditions were applied during the microbial culture, they were not maintained throughout the experiments, particularly during the analysis, which may result in the transformation of metastable iron sulfide phases into more stable pyrite that may not have been present initially. Moreover, the identification of pyrite in some of these studies is not infallible: SEM images rely solely on morphology, semi-quantitative SEM-EDS analyses may be subject to large errors due to surface effects, chemical extractions are not very selective to pyrite (e.g. chromium-reducible sulfur can also contain zero-valent sulfur species if not initially separated with methanol and HNO₃ can also target iron silicates) and there are still poorly indexed X-ray diffractograms in the literature that over-interpret the presence of pyrite, as they are made on multi-phase systems (containing elemental sulfur or vivianite, for example) that generate numerous peaks, many of which coincide with those of pyrite. Despite the fact that it is widely accepted that microorganisms play a key role in the production of pyrite, it is interesting to note that most attempts to produce pyrite in the presence of microorganisms have failed (Picard et al., Reference Picard, Gartman and Girguis2016, Reference Picard, Gartman, Clarke and Girguis2018; Ikogou et al., Reference Ikogou, Ona-Nguema, Juillot, Le Pape, Menguy, Richeux, Guigner, Noël, Brest, Baptiste and Morin2017; Zhou et al., Reference Zhou, Liu and Dong2017; Stanley and Southam, Reference Stanley and Southam2018; Mansor et al., Reference Mansor, Berti, Hochella, Murayama and Xu2019; Zhang et al., Reference Zhang, Zhu, Liao, Dang and Guo2021; Nabeh et al., Reference Nabeh, Brokaw and Picard2022) with only a few known exceptions (Rickard, Reference Rickard1969b; Ivarson and Hallberg, Reference Ivarson and Hallberg1976; Donald and Southam, Reference Donald and Southam1999; Thiel et al., Reference Thiel, Byrne, Kappler, Schink and Pester2019; Zhou et al., Reference Zhou, Gao, Xie, Wang, Yue, Wei, Yang, Li and Chen2019; Berg et al., Reference Berg, Duverger, Cordier, L-R, Guyot and Miot2020; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020).

For a quarter of a century, there has been a consensus on two reaction pathways for pyrite formation: the polysulfide pathway (Rickard, Reference Rickard1969a, Reference Rickard1975) and the H₂S pathway (Rickard, Reference Rickard1997; Rickard and Luther, Reference Rickard and Luther1997). Others have fallen into disuse because they can be interpreted as variants of the two main pathways (Table 2). Recently a third reaction pathway has been proposed under the name ferric-hydroxide-surface pathway (FHS; Wan et al., Reference Wan, Schröder and Peiffer2017).

Table 2.

Example of proposed reactions for pyrite formation.

The role of polysulfides in pyrite formation has long been known. Apart from the synthesis of pyrite by metallurgical processes requiring very high temperatures, two almost concurrent works attesting to the formation of pyrite from polysulfides at relatively low temperatures go as far back as the mid-19th century. During an expedition to Iceland following the 1845–1846 Hekla eruption, the formation of pyrite was observed in fumarole systems (Bunsen, Reference Bunsen1847). Based on the contemporaneous chemical knowledge about alkaline polysulfide dissolution of ferrous sulfide, which then reprecipitates, it was postulated that hot hydrogen sulfide vapours emitted by fumaroles transform the iron minerals in weathered basalt into pyrite through the transient formation of iron sulfide and polysulfides. Several years later, in pioneering hydrothermal synthesis experiments, several metal sulfide minerals including pyrite were successfully formed from a mixture of iron salts and ‘persulfide’ (obsolete word for polysulfide S_n^2-) in a sealed tube heated to 165–180°C (Senarmont, Reference Senarmont1851). Later investigations on hydrothermal synthesis of iron sulfide showed that pyrite could be formed by heating various mixtures of H₂S and iron salts, elemental sulfur and iron sulfide or dissolved polysulfide and ferrous salts, reproducing acidic to alkaline environments (Allen et al., Reference Allen, Crenshaw, Johnston and Larsen1912). The authors generalized these three different methods by “the action of sulphur on ferrous sulphide”, laying the foundations for a common pathway to pyrite formation regardless of the reagents used. The idea that elemental sulfur is directly involved in the formation of sedimentary pyrite subsequently gained ground, not least because of its common presence in the sediments (GW Harmsen, Reference Harmsen1954; Kaplan et al., Reference Kaplan, Emery and Rittenbebg1963; Berner, Reference Berner1970; Sweeney and Kaplan, Reference Sweeney and Kaplan1973).

Because the solid-state reaction between elemental sulfur and iron sulfide at low temperatures is mechanically impossible, it is now generally accepted that pyrite formation resulting from the addition of a sulfur compound to iron monosulfide occurs via the dissolved polysulfide reagent. This reaction is known as the polysulfide pathway (Rickard, Reference Rickard1975; Luther, Reference Luther1991). The reaction of HS^- with elemental sulfur (Reaction 1, Table 2) is a ubiquitous source of dissolved polysulfide which helps explain how pyrite can easily be formed anywhere from modern sediments to early experiments with elemental sulfur and iron sulfide. Isotopic measurements have revealed that pyrite formed by the polysulfide pathway inherits the polysulfide isotopic signature, meaning that the sulfur atom from the initial FeS is replaced by two sulfur atoms of the polysulfide via a cyclic reaction (Butler et al., Reference Butler, Böttcher, Rickard and Oldroyd2004).

The polysulfide pathway is probably of relevance in transitional redox environments with limited molecular oxygen content (Rickard, Reference Rickard1997). In addition to oxygen, other oxidants could promote this pathway in the environment. For instance, H₂S released by sulfate-reducing microorganisms (SRM) can be oxidized to polysulfide during abiotic ferric iron reduction (Wei and Osseo-Asare, Reference Wei and Osseo-Asare1997; Morin et al., Reference Morin, Noël, Menguy, Brest, Baptiste, Tharaud, Ona-Nguema, Ikogou, Viollier and Juillot2017; Baya et al., Reference Baya, Le Pape, Baptiste, Menguy, Delbes, Morand, Rouelle, Aubry, Ona-Nguema, Noël, Juillot and Morin2022). It is interesting to note that under strictly anoxic conditions, the few studies reporting pyrite formation in microbial cultures of SRM have mostly been carried out in the presence of ferric iron: goethite (FeO(OH), Rickard, Reference Rickard1969a), jarosite (KFe₃(SO₄)₂(OH)₆, Ivarson and Hallberg, Reference Ivarson and Hallberg1976) or ferric phosphate (FePO₄.2H₂O, Berg et al., Reference Berg, Duverger, Cordier, L-R, Guyot and Miot2020; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020). In these various works, amorphous black precipitates were initially observed and pyrite was only detected after longer incubation times. It would therefore seem that the H₂S produced by microbial sulfate reduction (MSR) reacted with the ferric iron minerals, producing not only ferrous iron and subsequently iron sulfide but also intermediate sulfur compounds paving the way for pyrite formation via the polysulfide pathway. A similar mechanism is probably at play in sulfur-disproportionating bacterial cultures that grow only with the addition of Fe(III) (oxyhydr)oxides as a hydrogen sulfide scavenger (Thamdrup et al., Reference Thamdrup, Finster, Hansen and Bak1993; Canfield et al., Reference Canfield, Thamdrup and Fleischer1998; Finster et al., Reference Finster, Liesack and Thamdrup1998). Polysulfides can also be directly produced by microorganisms and eventually released into the environment via polyvalent metabolisms such as incomplete sulfide oxidation (Berg et al., Reference Berg, Schwedt, A-C, Kuypers and Milucka2014; Findlay, Reference Findlay2016). The rate law for pyrite formation via the polysulfide pathway was derived by Rickard (Reference Rickard1975), with a correction later published due to incorrect unit conversion in the original publication (Wan et al., Reference Wan, Schröder and Peiffer2017).

The thermodynamic feasibility of pyrite formation via the H₂S pathway, even at low-temperature, biological conditions, was known long before being experimentally demonstrated (Berner, Reference Berner1970; Rickard, Reference Rickard1997; Rickard and Luther, Reference Rickard and Luther1997). The first observations of pyrite formation coupled with dihydrogen production came from hydrothermal experiments (Wikjord et al., Reference Wikjord, Rummery and Doern1976, Reference Wikjord, Rummery, Doern and Owen1980; Taylor et al., Reference Taylor, Rummery and Owen1979). In highly reducing environments where stronger oxidants are absent, H₂S can oxidize FeS to pyrite (Rickard, Reference Rickard1997; Rickard and Luther, Reference Rickard and Luther1997). This direct reaction leads to the production of H₂, which can then reduce carbon dioxide to organic molecules in the presence of a mineral or enzyme catalyst. In fact, the formation of pyrite via the H₂S pathway helped fuel the iron-sulfur world hypothesis for the origin of life (Wächtershäuser, Reference Wächtershäuser1988, Reference Wächtershäuser1992; Drobner et al., Reference Drobner, Huber, Wächtershäuser, Rose and Stetter1990). Although another mechanism, known as the iron-loss pathway, was proposed to explain the production of dihydrogen by the oxidation of two FeS molecules to pyrite and dissolved Fe²⁺ (Schoonen and Barnes, Reference Schoonen and Barnes1991; Wilkin and Barnes, Reference Wilkin and Barnes1996), it is now considered to be derived from the H₂S pathway (Butler et al., Reference Butler, Böttcher, Rickard and Oldroyd2004). There is evidence from the sulfur isotopic composition of pyrite formed via the H₂S route that it is a product of the equimolar mixing of the FeS and H₂S pools (Butler et al., Reference Butler, Böttcher, Rickard and Oldroyd2004).

Although HS^- is simply the deprotonated form of H₂S, differences in electron orbital energy levels make HS^- an incompatible electron acceptor in the H₂S pathway. For this reason, the H₂S pathway was initially suggested to be dominant in neutral to acidic environments, while the polysulfide pathway is dominant under alkaline conditions (Rickard and Luther, Reference Rickard and Luther1997; Thiel et al., Reference Thiel, Byrne, Kappler, Schink and Pester2019). The favourability of the reaction is also influenced by the microbial scavenging of H₂. For example, hydrogenotrophic methanogens growing together with sulfate-reducing bacteria probably maintain this reaction in pyrite-precipitating enrichment cultures (Thiel et al., Reference Thiel, Byrne, Kappler, Schink and Pester2019). The H₂S pathway is assumed to be more widespread in strictly anoxic subsurface sediments (Rickard, Reference Rickard1997) and could have been the dominant pyrite formation mechanism in the ferruginous Archean Ocean before the Great Oxygenation Event (Lyons et al., Reference Lyons, Reinhard and Planavsky2014).

Nonetheless, studies have shown that mackinawite and H₂S can remain incredibly stable over months in strict anoxic conditions even during hydrothermal processing (Benning et al., Reference Benning, Wilkin and Barnes2000; Cahill et al., Reference Cahill, Benning, Barnes and Parise2000; Swanner et al., Reference Swanner, Webb and Kappler2019), in contrast to the fast rates of the H₂S pathway initially reported in Rickard (Reference Rickard1997). These studies demonstrated that slightly oxidized iron sulfide precursors actually drove pyrite formation. Anaerobic cultivation techniques and sensing technologies used to measure oxygen have evolved a long way since early experiments (Wikjord et al., Reference Wikjord, Rummery and Doern1976; Taylor et al., Reference Taylor, Rummery and Owen1979; Drobner et al., Reference Drobner, Huber, Wächtershäuser, Rose and Stetter1990; Rickard, Reference Rickard1997; Rickard and Luther, Reference Rickard and Luther1997), so one cannot exclude the possible influence of very low oxygen concentrations in previous studies. In these cases, the surface oxidation of iron sulfide precursors may have been sufficient to drive pyrite formation under not strictly anoxic conditions, perhaps via a combination of different pathways.

Since FeS formation can be directly linked to the sulfidation of Fe(III) (hydr)oxides, it eventually became clear that surface-mediated reactions could play a role in pyrite formation. Nevertheless, it required a combination of bulk chemical and modern micro- to nano-scale imaging methods to finally elucidate the steps following the electron transfer reaction between sulfide and Fe(III) (hydr)oxides. In experiments mixing dissolved sulfide and lepidocrocite, TEM, XRD and Mössbauer were used to observe a very rapid (<2 h) surface reaction with the formation of a mackinawite rim on lepidocrocite, which eventually dissolved at the expense of pyrite precipitation delocalized from the lepidocrocite surface (Hellige et al., Reference Hellige, Pollok, Larese-Casanova, Behrends and Peiffer2012). Further investigations with iron minerals of different crystallinity (ferrihydrite < lepidocrocite < goethite) revealed that sulfidation proceeds at different rates and that the formation of FeS (hours) and pyrite (weeks) is decoupled in time (Peiffer et al., Reference Peiffer, Behrends, Hellige, Larese-Casanova, Wan and Pollok2015). This led to the proposal of a third pathway for pyrite formation catalysed by Fe(III) minerals. The so-called ferric hydroxide surface pathway (FHS) begins with the formation of surface-bound, non-sulfur-associated Fe(II) (>Fe(II)OH₂⁺) formed through surface complexation reactions (Table 2). These iron-hydroxy groups react with sulfide radicals to form surface-bound Fe(II)S₂⁻ species and induce pyrite nucleation in conditions below supersaturation by creating a new equilibrium with the aqueous phase. The FHS pathway is predicted to be prominent in aquatic systems with abundant Fe(III) minerals, i.e. with terrestrial influence, (Wan et al., Reference Wan, Schröder and Peiffer2017) because the reaction only proceeds if the precursor species >Fe^IIS₂^– incompletely covers the ferric hydroxide surface. Other compounds, like organic matter, have been found to mask ferric hydroxide reactive sites, slowing the formation of pyrite via the FHS pathway from 120 days (Wan et al., Reference Wan, Schröder and Peiffer2017) to 12 months (ThomasArrigo et al., Reference ThomasArrigo, Bouchet, Kaegi and Kretzschmar2020). Interestingly, the addition of dissolved (complexed) Fe³⁺ has commonly produced pyrite in abiotic experiments (Wei and Osseo-Asare, Reference Wei and Osseo-Asare1997; Morin et al., Reference Morin, Noël, Menguy, Brest, Baptiste, Tharaud, Ona-Nguema, Ikogou, Viollier and Juillot2017; Baya et al., Reference Baya, Le Pape, Baptiste, Menguy, Delbes, Morand, Rouelle, Aubry, Ona-Nguema, Noël, Juillot and Morin2022) but only metastable iron sulfide phases in microbial experiments (Bertel et al., Reference Bertel, Peck, Quick and Senko2012; Ikogou et al., Reference Ikogou, Ona-Nguema, Juillot, Le Pape, Menguy, Richeux, Guigner, Noël, Brest, Baptiste and Morin2017). The few successes in obtaining biogenic pyrite in laboratory experiments have used solid Fe(III) minerals (Rickard, Reference Rickard1969b; Ivarson and Hallberg, Reference Ivarson and Hallberg1976; Berg et al., Reference Berg, Duverger, Cordier, L-R, Guyot and Miot2020; Duverger et al., Reference Duverger, Berg, Busigny, Guyot, Bernard and Miot2020). Together, these results suggest that further studies on sulfidization pathways on common iron mineral phases should include organic matter and microorganisms in order to better represent environmental conditions.

The black smokers as Fe- and S-rich systems

The most extreme hyperthermophilic microorganisms have been isolated from hydrothermal vents (Huber et al., Reference Huber, Huber, Jones, Lauerer, Neuner, Segerer, Stetter and Degens1991; Blöchl et al., Reference Blöchl, Rachel, Burggraf, Hafenbradl, Jannasch and Stetter1997; Takai et al., Reference Takai, Komatsu and Horikoshi2001), which represent the most biologically active sites in the deep ocean found along mid-ocean ridges and discovered in 1977 (Corliss et al., Reference Corliss, Dymond, Gordon, Edmond, von Herzen, Ballard, Green, Williams, Bainbridge, Crane and Andel1979; Hannington et al., Reference Hannington, Tivey, Larocque, Petersen and Rona1995). These environments are characterized by unique physical and chemical properties such as high hydrostatic pressures, high temperatures and often highly dissolved metal contents. Among those, black smokers are iron- and sulfur-rich anaerobic systems (Holden et al., Reference Holden, Breier, Rogers, Schulte and Toner2012) that form chimneys up to 45 metres tall (Hannington et al., Reference Hannington, Tivey, Larocque, Petersen and Rona1995). Cold seawater infiltrates down through the oceanic crust, is heated in contact with the magma chamber, becomes less dense, and then rises to the seafloor, dissolving metals and sulfides from the surrounding basaltic rocks (Humphris and Mccollom, Reference Humphris and Mccollom1998; Fouquet et al., Reference Fouquet, Cambon, Etoubleau, Charlou, Ondréas, Barriga, Cherkashov, Semkova, Poroshina, Bohn and Donval2010). Sulfur species are abundant and present in several redox states in hydrothermal environments (from –2 to +6), both in inorganic and organic forms. Inorganic sulfur appears in various forms such as metal sulfides (chalcopyrite CuFeS₂, pyrite FeS₂, sphalerite ZnS) (Tivey and Delaney, Reference Tivey and Delaney1986; Peng and Zhou, Reference Peng and Zhou2005), as well as nanoparticles of elemental sulfur (S⁰), hydrogen sulfide (H₂S), hydrosulfide (HS^-), hydrogen polysulfides (HS_n^-) and polysulfides (S_n^2-) (Schwarzenbach and Fischer, Reference Schwarzenbach and Fischer1960; Rickard and Luther, Reference Rickard and Luther2007; Gartman et al., Reference Gartman, Yücel, Madison, Chu, Ma, Janzen, Becker, Beinart, Girguis and Luther2011). These are predominantly found in plumes and in the reducing parts of hydrothermal chimneys (Findlay et al., Reference Findlay, Gartman, Macdonald, Hanson, Shaw and Luther2014; Findlay, Reference Findlay2016). The presence of both oxidized and reduced sulfur compounds in the hydrothermal ecosystem supports the development of metabolically diverse microorganisms (Orcutt et al., Reference Orcutt, Sylvan, Knab and Edwards2011; Dick, Reference Dick2019). Sulfides and hydrogen sulfide can also result from MSR and/or reduction of elemental sulfur. Consequently, the sulfide content of black smokers originates from a mixture of abiotically produced sulfide through high-temperature chemical reactions and biotically produced sulfide via microbial activities, complicating the interpretation of sulfur-based biosignatures in such ecosystems.

Iron is found at very high concentrations, up to 25 mM (Holden and Adams, Reference Holden and Adams2003; Tivey, Reference Tivey2007; Toner et al., Reference Toner, Rouxel, Santelli, Bach and Edwards2016). In the anaerobic, reducing and high-temperature hydrothermal fluid, iron is mainly present as the soluble ferrous form Fe²⁺, while in the surrounding oxygenated seawater, iron is rapidly oxidized into the ferric form Fe(III) and precipitates as iron (oxyhydr)oxide minerals (Rickard and Luther, Reference Rickard and Luther2007; Scholten et al., Reference Scholten, Schmidt, Lecumberri-Sanchez, Newville, Lanzirotti, Sirbescu and Steele-MacInnis2019). In the fluid (>250°C), Fe²⁺ reacts with sulfide to form inorganic massive iron sulfide deposits, beginning with FeS mackinawite/ pyrrhotite, which is thermodynamically unstable, and then rapidly evolving into FeS₂ pyrite (Rickard and Luther, Reference Rickard and Luther2007), the major sulfide component of the interior of black smokers (Fouquet et al., Reference Fouquet, Schultz, Herrington and Nesbitt1997). Whereas Fe₃S₄ greigite is considered an intermediate phase in the process of pyrite formation (Hunger and Benning, Reference Hunger and Benning2007), no occurrence of greigite has been reported in hydrothermal chimneys. Pyritization is thus the main process in black smokers but the mechanism of pyrite formation is still being debated, especially for pyrite forming at relatively low temperatures (<150°C) in the external parts of chimneys, which may involve the living compartment (Juniper and Martineu, Reference Juniper and Martineu1995; McCollom, Reference McCollom2007).

To elucidate the process of microbial synthesis of pyrite, several experiments have aimed to mimic environmental conditions in laboratory experiments using mesophilic microorganisms to produce pyrite (see section ‘Biogenic pyrite formation at low temperature’). But pyrite formation in direct connection to the activity of hyperthermophilic microorganisms has rarely been reported in the literature (Stetter et al., Reference Stetter, König and Stackebrandt1983; Gorlas et al., Reference Gorlas, Jacquemot, Guigner, Gill, Forterre and Guyot2018, Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022; Truong et al., Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023).

Pyrite formation by hyperthermophiles

More than two decades ago, Stetter and colleagues published the first observation of pyrite formation resulting from a biogenic process at a high temperature (Stetter et al., Reference Stetter, König and Stackebrandt1983). Pyrite was formed during the growth of the sulfur-reducing archaeon Pyrodictium occultum in a medium containing a final concentration of 10 μM FeSO₄, in a fermenter at 95°C. This biogenic pyrite formation could be explained by an interaction between the metabolic end-product H₂S, formed from sulfur reduction, with dissolved Fe²⁺ according to the H₂S pathway (Table 2).

Unfortunately, there have been no further studies to confirm or refute this following these initial observations. Recent experiments were specifically designed to investigate the formation of Fe-S minerals by hyperthermophilic Archaea belonging to the Thermococcales order (Gorlas et al., Reference Gorlas, Jacquemot, Guigner, Gill, Forterre and Guyot2018), which are predominant inhabitants of the hottest parts of hydrothermal environments (Takai et al., Reference Takai, Komatsu and Horikoshi2001). Under laboratory conditions, the sulfur-reducers Thermococcales are able to generate significant amounts of pyrite within a few hours and could be an important contributor to pyrite formation in their ecosystem (Gorlas et al., Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022). Cells grown in a modified rich medium supplemented with S⁰ (1 g/L or 31 mM) were incubated with a solution of ferrous sulfate (5 mM of FeSO₄), under strict anoxia at 85°C (Gorlas et al., Reference Gorlas, Jacquemot, Guigner, Gill, Forterre and Guyot2018, Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022). The amount of pyrite production increases as the mineralization duration lengthens (Truong et al., Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023). Pyrite nanocrystals are consistently found to have a close association with the cells and with sulfur vesicles produced by Thermococcales during their growth (Gorlas et al., Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022). This leads to the production of pyrite spherules with ultra-smooth surfaces having diameters ranging between 200 nm to 1 μm (with a marked abundance maximum around 1 μm) made up of an assemblage of very numerous domains ranging between 5 and 15 nm (Truong et al., Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023). This assembly of small domains explains the ultra-smooth appearance of pyrite spherules. Interestingly, the presence of pyrite was detected only when Thermococcales were cultivated in a growth medium initially containing zero-valent sulfur S⁰ (Gorlas et al., Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022). In these growth conditions, Thermococcales cells internalized high sulfur concentrations leading to the production of numerous sulfur vesicles probably derived from polysulfides. This process has been interpreted as a polysulfide detoxification mechanism (Gorlas et al., Reference Gorlas, Marguet, Gill, Geslin, Guigner, Guyot and Forterre2015). Gorlas et al. (Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022) proposed that in the presence of abundant Fe²⁺, those sulfur vesicles allowing reactive sulfur to be exposed at cell surfaces could act as precursors for pyrite formation. Conversely, under growth conditions in which Thermococcales do not produce such sulfur vesicles but sulfide instead, no formation of pyrite occurs, suggesting that pyrite is produced through Thermococcales preferentially via the polysulfide pathway rather than the H₂S pathway (Gorlas et al., Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022).

The initial occurrence of the FeS nano-mackinawite, observed in short-term experiments (i.e. after 5 hours of mineralization) (Truong et al., Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023), could be attributed to the interaction between Fe²⁺ and the H₂S produced during Thermococcales growth. Thus Truong et al. (Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023) proposed that the formation of pyrite particles induced by the presence of Thermococcales and their sulfur vesicles occurs due to a redox comproportionation of S⁰ (from elemental sulfur) and sulfide (S-II) (from FeS) to yield S(-I) in pyrite particles. This suggests a high degree of metabolic adaptability by Thermococcales. This capability allows Thermococcales to adjust their metabolic activities in response to variations in sulfur and iron availability, which may fluctuate in hydrothermal vent environments.

Greigite formation by Thermococcales

Thermococcales have also been recognized for producing cuboidal extracellular nanocrystals of Fe₃S₄ greigite within a few days, with sizes ranging from 40 to 60 nm, regardless of the presence or absence of sulfur vesicles (Gorlas et al., Reference Gorlas, Jacquemot, Guigner, Gill, Forterre and Guyot2018, Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022; Truong et al., Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023). Greigite formation occurs in close proximity to cells and vesicles that yield pyrite spherules. While conventional models of greigite formation from FeS typically involve an excess of sulfur or an iron loss pathway (Wilkin and Barnes, Reference Wilkin and Barnes1996), Gorlas et al. (Reference Gorlas, Jacquemot, Guigner, Gill, Forterre and Guyot2018) proposed that those greigites related to the biological activity of Thermococcales were formed by sulfurization of amorphous Fe(III)-bearing phosphates loaded onto cellular debris. Production of greigite was also observed during the growth of the hyperthermophile methanogen Methanocaldoccus jannaschii at 80°C when hematite was added to the growth medium (Igarashi et al., Reference Igarashi, Yamamura and Kuwabara2016). Following the reduction of sulfur by the methanogen, hematite was reduced to form amorphous FeS, subsequently reacting with residual Fe(III) into greigite-like nanoflakes. The presence of Fe(III), either because it is added or because it is intrinsically produced, therefore seems to be a sufficient condition to form greigites in these strictly anoxic environments. In the case of Thermococcales, the mechanism by which Fe(III) is generated has yet to be elucidated (Kish et al., Reference Kish, Miot, Lombard, Guigner, Bernard, Zirah and Guyot2016). Finally, it is likely that over long periods of time, some of the greigites formed by these processes subsequently evolve into pyrite, but it is clear that in the case of Thermococcales, greigite formation is not just an intermediate towards pyrite but constitutes a pathway on its own.

The efficiency with which Thermococcales produce pyrite and greigite in laboratory settings suggests that they are able to replicate this biomineralization process in their native environment. Consequently, the hyperthermophiles Thermococcales may play a significant role in the formation of ‘low temperature’ pyrite in their ecosystem. Furthermore, this biologically induced mineralization mechanism of Fe-S minerals by Thermococcales could be a key component of their survival strategy to thrive in highly mineralized, high-temperature environments (Gorlas et al., Reference Gorlas, Mariotte, Morey, Truong, Bernard, J‐M, Oberto, Baudin, Landrot, Baya, Le Pape, Morin, Forterre and Guyot2022).

Background

In contrast to metastable iron sulfides, pyrite is prevalent in sedimentary rocks, with its presence dating back to approximately 3.8 billion years ago (Ga) (Smith et al., Reference Smith, Evensen, York and Moorbath2005). The ubiquity of pyrite in sedimentary rocks, as shown in Fig. 1, has led to the use of pyrite abundance, isotopic characteristics, and trace metal concentrations to reconstruct past environmental conditions, aiding in the understanding of historical atmospheric oxygen levels, as well as global sulfur and iron geochemical cycles over Earth’s geological history. For instance, the recognition of detrital pyrite grains apparently eroded by flowing water serves as a robust indicator of low oxygen (O₂) levels before approximately 2.4 Ga (Johnson et al., Reference Johnson, Gerpheide, Lamb and Fischer2014). This timing aligns with the loss of sulfur mass-independent fractionation (S-MIF, see box), which is commonly attributed to the shift from an anoxic atmosphere (pO₂ < 10^-15 PAL) to an oxic atmosphere (Farquhar et al., Reference Farquhar, Bao and Thiemens2000).

Furthermore, the primary mechanism for the long-term storage of reduced sulfur species on geological timescales is the formation and subsequent burial of pyrite. This process, in conjunction with the burial of organic carbon, contributes significantly to maintaining the oxidized surface conditions of Earth (Canfield and Farquhar, Reference Canfield and Farquhar2009). Consequently, methods have been developed to gauge past oxidation states and water chemistry by examining the iron (-sulfur) mineralogy in sediment and sedimentary rocks (Raiswell and Canfield, Reference Raiswell and Canfield2012; Raiswell et al., Reference Raiswell, Hardisty, Lyons, Canfield, Owens, Planavsky, Poulton and Reinhard2018). These methods are grounded in the differential reactivity of iron phases towards sulfide, and thus, rely on pyrite precipitation. The presently utilized framework (Poulton and Canfield, Reference Poulton and Canfield2005), known as the ‘Fe speciation’ proxy, involves quantifying the ratio of highly reactive iron to total iron (Fe_HR/Fe_T) and the ratio of pyrite to highly reactive iron (Fe_PYR/Fe_HR). The application of Fe speciation data, whether derived from stratigraphic variations or collected across various locations over time or space, has significantly contributed to our understanding of the evolution of ocean redox conditions (Poulton, Reference Poulton2021 and references therein). It has revealed that during the Archean and Phanerozoic eras, oceans were predominantly ferruginous (characterized by anoxic conditions and high Fe²⁺ levels) and oxygenated, respectively. In the Proterozoic era, rising oxygen levels facilitated the development of euxinic conditions at mid-depth, particularly in restricted basins while the deep ocean maintained predominantly ferruginous conditions (Poulton et al., Reference Poulton, Fralick and Canfield2004a; Planavsky et al., Reference Planavsky, McGoldrick, Scott, Li, Reinhard, Kelly, Chu, Bekker, Love and Lyons2011; Ostrander et al., Reference Ostrander, Nielsen, Owens, Kendall, Gordon, Romaniello and Anbar2019). It is important to note that recent concerns have been raised regarding the effectiveness of the extraction protocol in retrieving the targeted mineral phases (Hepburn et al., Reference Hepburn, Butler, Boyce and Schröder2020; Slotznick et al., Reference Slotznick, Sperling, Tosca, Miller, Clayton, van Helmond, Slomp and Swanson-Hysell2020) and the influence of early diagenesis (Eroglu et al., Reference Eroglu, Scholz, Salvatteci, Siebert, Schneider and Frank2021; Hutchings and Turchyn, Reference Hutchings and Turchyn2021; Pasquier et al., Reference Pasquier, Fike, Révillon and Halevy2022). These concerns question the capacity of the ‘Fe speciation’ approach to accurately constrain the chemistry and oxidation state of water columns, both in the present and in the past (Pasquier et al., Reference Pasquier, Fike, Révillon and Halevy2022).

Texture and composition

Pyrite grains display a large range of sizes, spanning from centimetres to nanometres, and a variety of textures, e.g. euhedral/anhedral, nodule and framboidal are among the most common (see Fig. 1). Framboids, in particular, are characterized as spherical to subspherical clusters comprised of numerous microcrystals of pyrite, predominantly found in sedimentary environments (Wang and Morse, Reference Wang and Morse1996; Wilkin and Barnes, Reference Wilkin and Barnes1997; Rickard, Reference Rickard2019). Irrespective of its shape, during its crystallization, pyrite has the capability to integrate various trace elements (TE), with chalcophile and siderophile elements being the most frequently encountered. These TE can be integrated into pyrite through two distinct mechanisms: either by substituting for Fe or S within the pyrite structure, or by existing as inclusions of distinct mineral phases or amorphous masses enclosed within the pyrite matrix (Gregory, Reference Gregory2020 and references therein). Of particular interest in enhancing our understanding of past environmental conditions is the observation that the TE content in pyrite is directly linked to the TE content of the water from which it precipitates (Gregory et al., Reference Gregory, Meffre and Large2014). Consequently, TE within pyrite has been utilized to decipher the formation of ore deposits (Kusebauch et al., Reference Kusebauch, Gleeson and Oelze2019) and to unravel the evolution of ocean and atmospheric chemistry (e.g. Large et al., Reference Large, Halpin, Danyushevsky, Maslennikov, Bull, Long, Gregory, Lounejeva, Lyons, Sack, McGoldrick and Calver2014; Gregory et al., Reference Gregory, Lyons, Large, Jiang, Stepanov, Diamond, Figueroa and Olin2017). Nevertheless, recent investigations have revealed that pyrite’s TE concentration varies with depth within the sediment, ultimately reaching a plateau in composition (Gregory et al., Reference Gregory, Kovarik, Taylor, Perea, Owens, Atienza and Lyons2022a). This suggests that the relationship between the water column and pyrite’s TE concentration is more intricate than previously assumed. Hence, TE contents in pyrite not only reflect the composition of the water column but also the composition of pore water, maybe allowing for tracking the release of trace elements from organic matter and/or iron (oxyhydr)oxides during diagenesis, particularly when examined at a finer spatial resolution (Tribovillard et al., Reference Tribovillard, Algeo, Lyons and Riboulleau2006; Gregory et al., Reference Gregory, Lyons, Large and Stepanov2022b; Atienza et al., Reference Atienza, Gregory, Taylor, Swing, Perea, Owens and Lyons2023). Biogenic pyrite experiments in the presence of TE have not been performed and it is unclear how biological activities can affect the incorporation of TE into pyrite. A recent study has also highlighted the importance of pyrite growth via particle attachment under certain conditions, which might impact the morphology and distribution of trace metals and isotopes within pyrite (Domingos et al., Reference Domingos, Runge, Dreher, T-H, Shuster, Fischer, Kappler, J-P, Xu and Mansor2023).

Sulfur isotopes

Sedimentary pyrite retains a distinct isotopic signature that reflects a combination of microbial metabolic processes and physical transport and mineralization. A broad taxonomic spectrum of microbes can change the oxidation state of sulfur to gain the energy required for cellular function and growth. Three main metabolic pathways are particularly important for the sulfur isotopic composition of pyrite: microbial sulfate reduction (MSR), sulfur disproportionation, and sulfide oxidation (Fike et al., Reference Fike, Bradley and Rose2015; Jørgensen et al., Reference Jørgensen, Findlay and Pellerin2019). From each of them arises a metabolic-specific microbial fractionation of sulfur isotopes which discriminates heavy isotopes, reported afterwards as ³⁴ε-³³λ (see isotope notation box).

Over the past several decades, laboratory experiments using pure cultures of SRM have yielded variable apparent isotopic fractionation, from 65‰ to ≈5‰ (Detmers et al., Reference Detmers, Brüchert, Habicht and Kuever2001; Johnston et al., Reference Johnston, Farquhar, Wing, Kaufman, Canfield and Habicht2005b; Hoek et al., Reference Hoek, Reysenbach, Habicht and Canfield2006; Johnston et al., Reference Johnston, Farquhar and Canfield2007; Sim et al., Reference Sim, Bosak and Ono2011a; Sim et al., Reference Sim, Ono, Donovan, Templer and Bosak2011b; Sim et al., Reference Sim, Ono and Bosak2012; Leavitt et al., Reference Leavitt, Halevy, Bradley and Johnston2013; Deusner et al., Reference Deusner, Holler, Arnold, Bernasconi, Formolo and Brunner2014; Pellerin et al., Reference Pellerin, Anderson-Trocmé, Whyte, Zane, Wall and Wing2015; Bradley et al., Reference Bradley, Leavitt, Schmidt, Knoll, Girguis and Johnston2016; Smith et al., Reference Smith, Fike, Johnston and Bradley2020). However, it was not until we understood the enzymatic processes that drive MSR that the debate over the extent of microbial fractionation could be settled (Goldhaber and Kaplan, Reference Goldhaber and Kaplan1980; Ohmoto et al., Reference Ohmoto, Kaiser and Geer1990; Rudnicki et al., Reference Rudnicki, Elderfield and Spiro2001; Wortmann et al., Reference Wortmann, Bernasconi and Böttcher2001; Brunner et al., Reference Brunner, Bernasconi, Kleikemper and Schroth2005). Recent work shows that the microbial reduction of sulfate to sulfide is catalysed by a reaction network of four enzymatic steps, all of which are reversible (Johnston et al., Reference Johnston, Farquhar, Wing, Kaufman, Canfield and Habicht2005b; Johnston et al., Reference Johnston, Farquhar and Canfield2007; Sim et al., Reference Sim, Bosak and Ono2011a; Sim et al., Reference Sim, Ono, Donovan, Templer and Bosak2011b; Sim et al., Reference Sim, Ono and Bosak2012; Leavitt et al., Reference Leavitt, Halevy, Bradley and Johnston2013, Reference Leavitt, Waldbauer, Venceslau, Sub, Zhang, Boidi, Plummer, Diaz, Pereira and Bradley2024; Bradley et al., Reference Bradley, Leavitt, Schmidt, Knoll, Girguis and Johnston2016; Smith et al., Reference Smith, Fike, Johnston and Bradley2020). The emerging picture from laboratory cultures, bio-isotopic models, and modern environments is that a large ³⁴ε_DSR predominates in natural environments, ranging from 66 to 78‰ (Wing and Halevy, Reference Wing and Halevy2014; Halevy et al., Reference Halevy, Fike, Pasquier, Bryant, Wenk, Turchyn, Johnston and Claypool2023). This substantial fractionation, which closely resembles the thermodynamic equilibrium between sulfate and sulfide (i.e. ≈70‰ at 20°C; Eldridge et al., Reference Eldridge, Guo and Farquhar2016), is attributed to the inherently low cell-specific sulfate reduction rates (csSRR) in marine sediments. In other words, all steps of the enzymatic pathway are fully reversible (i.e. equilibrium between reactant and product; Wing and Halevy, Reference Wing and Halevy2014). The observed inverse relationship between the ³⁴ε_DSR and the csSRR is responsible for the overall deviation from thermodynamic equilibrium observed in laboratory experiments (i.e. lower apparent fractionation). This relationship between ³⁴ε_DSR and csSRR can be extended to ³³S and, consequently, one can expect sulfide produced from MSR to be characterized by large, near-equilibrium ³⁴ε_DSR and ³³λ_DSR values. Culture experiments with green sulfur bacteria that oxidize sulfide showed smaller microbial S-isotope fractionations and small but nonzero ³³λ deviations from equilibrium (Zerkle et al., Reference Zerkle, Farquhar, Johnston, Cox and Canfield2009), whereas larger microbial S-isotope fractionations and ³³λ deviations from equilibrium were observed in pure cultures of sulfur disproportionators ( Johnston et al., Reference Johnston, Wing, Farquhar, Wing, Kaufman, Canfield and Habicht2005a; Fig. 5).

Figure 5.

Comparison of S-isotope measurements in culture experiments to assess isotopic microbial fractionation, with the S-isotopic composition of pyrite preserved in natural samples over the geological record (measured by bulk and microscale techniques). Data are from: MSR (Detmers et al., Reference Detmers, Brüchert, Habicht and Kuever2001; Johnston et al., Reference Johnston, Farquhar, Wing, Kaufman, Canfield and Habicht2005b; Hoek et al., Reference Hoek, Reysenbach, Habicht and Canfield2006; Johnston et al., Reference Johnston, Farquhar and Canfield2007; Sim et al., Reference Sim, Bosak and Ono2011a; Reference Sim, Ono, Donovan, Templer and Bosak2011b; Sim et al., Reference Sim, Ono and Bosak2012; Leavitt et al., Reference Leavitt, Halevy, Bradley and Johnston2013, Reference Leavitt, Waldbauer, Venceslau, Sub, Zhang, Boidi, Plummer, Diaz, Pereira and Bradley2024; Deusner et al., Reference Deusner, Holler, Arnold, Bernasconi, Formolo and Brunner2014; Pellerin et al., Reference Pellerin, Anderson-Trocmé, Whyte, Zane, Wall and Wing2015; Bradley et al., Reference Bradley, Leavitt, Schmidt, Knoll, Girguis and Johnston2016; Smith et al., Reference Smith, Fike, Johnston and Bradley2020); S-oxidizers (Zerkle et al., Reference Zerkle, Farquhar, Johnston, Cox and Canfield2009); disproportionators (Johnston et al., Reference Johnston, Wing, Farquhar, Wing, Kaufman, Canfield and Habicht2005a); bulk and microscale pyrite (Halevy et al., Reference Halevy, Fike, Pasquier, Bryant, Wenk, Turchyn, Johnston and Claypool2023). Modelled DSR refers to bio-isotopic model outputs analysed under a wide range of environmental parameters (i.e. temperature, sulfate, organic matter and Fe availabilities) expected to reflect modern marine conditions.

Once generated, a portion of this ³⁴S-depleted sulfide can react with dissolved ferrous iron (Fe²⁺) or iron-bearing minerals (Fe(III)) to form FeS. The formation of FeS and its subsequent transformation into pyrite involves a relatively small isotopic fractionation, typically not exceeding a few ‰ (>5‰; Fry et al., Reference Fry, Cox, Gest and Hayes1986; Böttcher et al., Reference Böttcher, Smock and Cypionka1998). Consequently, pyrite captures the isotopic composition of the products of microbial sulfur metabolisms (H₂S, HS^-), including small deviations from MDF, and thus may serve as a good recorder of past microbial activity. Interestingly, bulk pyrite S-isotope data do not meet the experimental expectations above. Several non-unique explanations that involve mixing between different sulfur pools or combinations of metabolic effects have been invoked to explain the apparent mismatch, clearly demonstrating that our current methodology is not suitable for uniquely distinguishing the various microbial pathways involved in pyrite formation (see Johnston, Reference Johnston2011 for a recent review).

The uncertainty in the interpretation of bulk δ³⁴S_PYR data can be resolved by studying the isotopic composition of individual pyrite grains, which collectively contribute to the bulk signal (Fike et al., Reference Fike, Bradley and Rose2015; Marin-Carbonne et al., Reference Marin-Carbonne, M-N, Havas, Remusat, Pasquier, Alléon, Zeyen, Bouton, Bernard, Escrig, Olivier, Vennin, Meibom, Benzerara and Thomazo2022; Bryant et al., Reference Bryant, Houghton, Jones, Pasquier, Halevy and Fike2023). As illustrated in Fike et al. (Reference Fike, Bradley and Rose2015), when MSR occurs in a transport-limited system, the concentration of residual sulfate in porewater decreases, as do δ³⁴S_SO4 and Δ³³S_SO4, due to Rayleigh distillation. This results in a parallel increase in the instantaneous δ³⁴S_H2S-Δ³³S_H2S that can be preserved in the accumulated pyrite pool. The magnitude of this enrichment depends on the fraction of sulfate consumed, the connectivity of the porewater sulfate pool with the overlying seawater column, and the fractionation associated with the redox transformation itself. Most of these factors depend on the local sedimentation regime, including the sedimentation rate, the iron and organic carbon loading and reactivity (Pasquier et al., Reference Pasquier, Sansjofre, Rabineau, Revillon, Houghton and Fike2017; Liu et al., Reference Liu, Antler, Pellerin, Izon, Dohrmann, Findlay, Røy, Ono, Turchyn, Kasten and Jørgensen2021; Pasquier et al., Reference Pasquier, Bryant, Fike and Halevy2021a; Pasquier et al., Reference Pasquier, Fike and Halevy2021b; Houghton et al., Reference Houghton, Scarponi, Capraro and Fike2022; Bryant et al., Reference Bryant, Houghton, Jones, Pasquier, Halevy and Fike2023; Halevy et al., Reference Halevy, Fike, Pasquier, Bryant, Wenk, Turchyn, Johnston and Claypool2023). As pyrite grains grow within the sediment, they continuously sample the evolving isotopic composition of microbially-produced sulfide throughout the sample’s burial history, and when pyrite grains grow quickly, over short durations as for framboid morphology, it is expected that a time series of increasing δ³⁴S_PYR values will be recorded within the population of pyrite grains in each sample. In such instances, the minimum δ³⁴S_PYR can be used to assess the microbial fractionation specific to the depositional environment (Marin-Carbonne et al., Reference Marin-Carbonne, M-N, Havas, Remusat, Pasquier, Alléon, Zeyen, Bouton, Bernard, Escrig, Olivier, Vennin, Meibom, Benzerara and Thomazo2022; Bryant et al., Reference Bryant, Houghton, Jones, Pasquier, Halevy and Fike2023) whereas the overall distribution of δ³⁴S is more likely to reflect the diagenetic evolution over the history of the sediment burial (Halevy et al., Reference Halevy, Fike, Pasquier, Bryant, Wenk, Turchyn, Johnston and Claypool2023).

Assuming that MSR was established approximately 3.5 billion years ago (Shen and Buick, Reference Shen and Buick2004; Mateos et al., Reference Mateos, Chappell, Klos, LE, Boden, Stüeken and Anderson2023), the sulfur isotopic compositions of pyrite (denoted as δ³⁴S_PYR) are commonly utilized in both contemporary and ancient contexts to gain insights into the processes and fluxes within the global sulfur cycle. Within the framework of a global steady-state S cycle, the lowest δ³⁴S values reflect the maximum microbial fractionation whereas the departure from those low microbial values toward higher values indicates the evolution of MSR during the burial history of the sediment, irrespectively of the local and/or global oxygenation (Fig. 5). This pattern is typically interpreted as indicative of rising oxygen levels over time. This increase in oxygen may have resulted in more elevated oceanic sulfate levels and/or a reduction of the pyrite burial flux in response to more efficient organic matter oxidation (i.e. more aerobic respiration; Habicht et al., Reference Habicht, Gade, Thamdrup, Berg and Canfield2002; Wu et al., Reference Wu, Farquhar, Strauss, Kim and Canfield2010; Leavitt et al., Reference Leavitt, Halevy, Bradley and Johnston2013).

To delve further into this, Archean sedimentary rocks exhibit relatively low apparent fractionation between sulfate and pyrite (Δ_PYR) values, which are often interpreted as a consequence of the limited sulfate reservoir in the ancient oceans (Habicht et al., Reference Habicht, Gade, Thamdrup, Berg and Canfield2002), with sulfate estimates ranging from hundreds to tens of micromoles per litre (μM) (Crowe et al., Reference Crowe, Paris, Katsev, Jones, S-T, Zerkle, Nomosatryo, Fowle, Adkins, Sessions, Farquhar and Canfield2014). While the rates and isotopic implications of organic sulfur breakdown remain uncertain, both modelling and observations suggest that organic sulfur probably played a role in extremely low-sulfate systems (Fakhraee and Katsev, Reference Fakhraee and Katsev2019). This has potential implications for understanding the pathways and isotopic compositions involved in early Earth pyrite formation. The transition from the Archean to the Early Proterozoic era is marked by the gradual oxidation of the Earth’s surface environment. This led to an increase in seawater sulfate content, resulting in the disappearance of sulfur mass-independent fractionation (S-MIF) signals (see Box) and an overall rise in Δ_PYR values (Fig. 5). Throughout much of the Proterozoic era, δ³⁴S_PYR values exhibit significant variability, which is commonly attributed to fluctuations in the pyrite burial rate and shifts in ocean redox conditions, such as changes in the extent of ocean anoxia (Emmings et al., Reference Emmings, Poulton, Walsh, Leeming, Ross and Peters2022). After the oxygenation of the ocean atmosphere, the range of δ³⁴S_PYR values remained relatively constant at around -50‰ (Fig. 5), and the observed decreases during the Paleozoic era are probably associated with marine sulfate isotopic secular evolution (Fig. 5; e.g. Owens et al., Reference Owens, Gill, Jenkyns, Bates, Severmann, Kuypers, Woodfine and Lyons2013; Gomes et al., Reference Gomes, Hurtgen and Sageman2016; Raven et al., Reference Raven, Fike, Gomes, Webb, Bradley and McClelland2018). An essential aspect of the Phanerozoic era is the emergence of episodic intervals of ocean anoxia, known as Oceanic Anoxic Events (OAEs), driven by large-scale disturbances in the carbon cycle. High-resolution analysis of pyrite and seawater sulfate sulfur isotopic compositions and their utilization in isotopic box models have unveiled brief intervals of severe anoxia which led to significant drawdown of the seawater sulfate during those brief episodes of Earth’s history (100s kyr to My, Song et al., Reference Song, Wignall, Chu, Tong, Sun, Song, He and Tian2014; Bauer et al., Reference Bauer, Bottini, Katsev, Jellinek, Francois, Erba and Crowe2022).

More recently, an alternative reading of the first-order pattern preserved in the S-isotope geologic record has emerged due to a growing body of evidence highlighting the importance of local sedimentary processes in shaping the preserved isotopic composition of pyrite (Fike et al., Reference Fike, Bradley and Rose2015; Halevy et al., Reference Halevy, Fike, Pasquier, Bryant, Wenk, Turchyn, Johnston and Claypool2023). These findings are based on the study of stratigraphic δ³⁴S_PYR variations within 100 kyr Pleistocene glacial-interglacial cycles, which have been linked to sedimentary parameters such as sedimentation rates and the availability and reactivity of iron and/or organic carbon, rather than changes in microbial fractionation (Bryant et al., Reference Bryant, Houghton, Jones, Pasquier, Halevy and Fike2023). These findings challenge the conventional view that sedimentary archives merely passively record the global sulfur cycle, especially given the long residence time of sulfate in the ocean (13 Myr, Kah et al., Reference Kah, Lyons and Frank2004). The emerging understanding suggests that, in addition to the expansion of the marine sulfate reservoir, sedimentary parameters through their influence on the accessibility of sulfate within the sediment play a crucial role in shaping the wide spectrum of Δ_PYR preserved in sedimentary rock records, rather than large-scale temporal shifts in the global sulfur cycle.

Iron isotopes

The formation of pyrite necessitates the presence of Fe(II), which in natural environments mostly originates from either dissimilatory iron reduction (DIR) or from abiotic sulfidization of Fe(III) (oxyhydr)oxides. When Fe(III) reducers are grown in pure cultures, they produce Fe²⁺ that exhibits a ⁵⁶Fe-depletion of approximately 2.9±0.9‰ compared to the initial Fe(III) substrates (Crosby et al., Reference Crosby, Johnson, Roden and Beard2005, Reference Crosby, Roden, Johnson and Beard2007). This microbial fractionation remains consistent across different Fe(III) substrates and bacterial strains, and it closely matches the thermodynamic equilibrium fractionation between Fe(III) and Fe(II) (Welch et al., Reference Welch, Beard, Johnson and Braterman2003) (Fig. 6). In contrast, laboratory experiments involving the abiotic sulfidization of Fe(III) (oxyhydr)oxides release Fe²⁺ with a ⁵⁶Fe-depletion of approximately 0.8±0.3‰ relative to the Fe(III) minerals (McAnena et al., Reference McAnena, Severmann, Guilbaud and Poulton2024). Interestingly, this abiotic fractionation appears to be independent of the dissolution rate of the Fe(III) (oxyhydr)oxides, which itself seems to be influenced by the mineralogy of Fe(III) or the S(-II):Fe(III) ratio.

Figure 6.

Compilation of Fe-isotopes measurements during the reduction of Fe(III) minerals, from abiotic processes involved during mineral precipitation and from pyrite preserved in modern environments over the geological record (measured by bulk and microscale techniques). Data are from: dissimilatory iron reduction DIR (Crosby et al., Reference Crosby, Johnson, Roden and Beard2005, Reference Crosby, Roden, Johnson and Beard2007); sulfidization (McAnena et al., Reference McAnena, Severmann, Guilbaud and Poulton2024); bulk and microscale pyrite (Dupeyron et al., Reference Dupeyron, Decraene, Marin-Carbonne and Busigny2023).

Furthermore, apart from the isotopic fractionation associated with the reduction step, the conversion of Fe²⁺ to FeS species also leads to a noticeable isotopic fractionation. Laboratory experiments have shown that this fractionation can vary between –0.8±0.2‰ (Butler et al., Reference Butler, Archer, Vance, Oldroyd and Rickard2005) and an apparent equilibrium FeS-Fe²⁺ isotope fractionation of +0.4±0.2‰ (Wu et al., Reference Wu, Druschel, Findlay, Beard and Johnson2012). Near-equilibrium between Fe²⁺ to FeS in natural environments is consistent with the observed isotopic exchange between FeS and aqueous sulfide (Butler et al., Reference Butler, Böttcher, Rickard and Oldroyd2004). Under laboratory conditions, the formation of pyrite from FeS results in pyrite that is ⁵⁶Fe-depleted by 0.5 to 2.2‰ compared to the FeS pool (Guilbaud et al., Reference Guilbaud, Butler and Ellam2011; Mansor and Fantle, Reference Mansor and Fantle2019). The overall Fe isotope fractionation between pyrite and Fe²⁺ (or Fe(III) (oxyhydr)oxides) depends on the relative importance of these reactions and the rate-dependent expression of kinetic and equilibrium isotope effects associated with these processes.

A limited number of studies (n=9) collectively provide valuable insights into δ⁵⁶Fe values associated with pyrite found in modern marine sedimentary environments, thereby enhancing our understanding of the influences of water column redox processes (e.g. Busigny et al., Reference Busigny, Planavsky, Jézéquel, Crowe, Louvat, Moureau, Viollier and Lyons2014; Rolison et al., Reference Rolison, Stirling, Middag, Gault-Ringold, George and Rijkenberg2018), offering a contemporary perspective on the benthic iron shuttle (Severmann et al., Reference Severmann, Lyons, Anbar, McManus and Gordon2008; Scholz et al., Reference Scholz, Severmann, McManus, Noffke, Lomnitz and Hensen2014), and examining how early diagenesis can impact δ⁵⁶Fe_PYR in marine sediments (as observed in Fehr et al., Reference Fehr, Andersson, Hålenius and Mörth2008, Reference Fehr, Andersson, Hålenius, Gustafsson and Morth2010; Lin et al., Reference Lin, Sun, Lu, Strauss, Xu, Gong, Teichert, Lu, Lu, Sun and Peckmann2017, Reference Lin, Sun, Lu, Strauss, Xu, Chen, Lu and Peckmann2018). However, it should be noted that the majority of bulk δ⁵⁶Fe_PYR data fail to align with the combined expectations derived from experimental and theoretical considerations mentioned earlier. One can expect that with recent advancements in in situ δ⁵⁶Fe_PYR analyses, they can be used to remove some of the uncertainty in interpreting bulk δ⁵⁶Fe_PYR data, particularly when conducted across a range of well-defined modern depositional settings.

By taking into account some of the effects mentioned above, researchers have employed the iron isotopic composition of pyrite, denoted as δ⁵⁶Fe_PYR, to explore how the biogeochemical iron (Fe) cycle in Earth’s oceans and the processes underlying pyrite formation may have evolved over geological history. In contrast to S-isotopes, the δ⁵⁶Fe_PYR record is fragmented, with a primary focus on significant shifts in Earth’s redox history, particularly associated with two major events of atmospheric oxygenation – the Great Oxidation Event (GOE) and the Neoproterozoic Oxidation Event (NOE). Consequently, it remains challenging to construct a comprehensive long-term narrative of the Fe cycle’s evolution.

Various environmental factors, whether in conjunction with the co-evolution of microbial life or not, have been suggested as potential explanations for the bulk negative isotopic variations that occurred prior to the GOE around 2.4 billion years ago. For instance, Rouxel et al., (Reference Rouxel, Bekker and Edwards2005) leveraged δ⁵⁶Fe_PYR values to glean insights into the oxygenation status of ancient oceans. Lighter δ⁵⁶Fe_PYR values were associated with partial oxidation of a vast Fe²⁺ oceanic reservoir indicative of more reducing conditions, while heavier values were indicative of more oxidizing conditions leading up to the GOE. Furthermore, the transition towards more positive δ⁵⁶Fe_PYR values has been suggested to attest to the onset of pyrite weathering (Heard et al., Reference Heard, Dauphas, Guilbaud, Rouxel, Butler, Nie and Bekker2020). This would have significantly increased the availability of sulfate in the ocean, thereby affecting the interplay between kinetic and equilibrium processes during pyrite precipitation (Mansor and Fantle, Reference Mansor and Fantle2019; Heard et al., Reference Heard, Dauphas, Guilbaud, Rouxel, Butler, Nie and Bekker2020). Alternatively, the prevalence of low δ⁵⁶Fe_PYR values during the Neoarchean era (2.8–2.5 billion years ago) has been used to underscore the early emergence of DIR and the inference of a microbially driven Fe cycle (Johnson et al., Reference Johnson, Beard and Roden2008). These low values have also been interpreted as reflecting strong kinetic effects during the abiotic formation of pyrite from FeS precursors (Guilbaud et al., Reference Guilbaud, Butler and Ellam2011). It’s important to highlight that a recent collection of microscale δ⁵⁶Fe_PYR values across the GOE presents distinctive patterns when compared with results obtained through conventional bulk analyses (Dupeyron et al., Reference Dupeyron, Decraene, Marin-Carbonne and Busigny2023). The discrepancy in the evolution of δ⁵⁶Fe_PYR between bulk and in situ measurements may be linked to a sampling bias, because the majority of traditional bulk investigations have been performed on millimetre-scale pyrite grains, often extracted from black shale matrices.

Sequestration of organic carbon by Fe-S minerals

Associations between Fe-S minerals and organic carbon have been hardly investigated, even though they might provide a mechanism for the long-term preservation of organic matter in rocks and sediments. Anoxic zones in the global ocean are expanding as a result of warming climate (Keeling et al., Reference Keeling, Körtzinger and Gruber2010; Levin and Bris, Reference Levin and Bris2015; Keil, Reference Keil2017; Breitburg et al., Reference Breitburg, Levin, Oschlies, Grégoire, Chavez, Conley, Garçon, Gilbert, Gutiérrez, Isensee, Jacinto, Limburg, Montes, Naqvi, Pitcher, Rabalais, Roman, Rose, Seibel, Telszewski, Yasuhara and Zhang2018; Ruvalcaba Baroni et al., Reference Ruvalcaba Baroni, Palastanga and Slomp2020). Therefore, understanding interactions between Fe-S minerals and organic carbon might be of importance to predict how the biogeochemical cycles of Fe, S and C might evolve in marine environments. The importance of Fe(III) (oxyhydr)oxides for the preservation of organic carbon has been well explored and recognized because of their abundance in oxygenated surface environments (Lalonde et al., Reference Lalonde, Mucci, Ouellet and Gélinas2012; Longman et al., Reference Longman, Faust, Bryce, Homoky and März2022; Moore et al., Reference Moore, Curti, Woulds, Bradley, Babakhani, Mills, Homoky, Xiao, Bray, Fisher, Kazemian, Kaulich, Dale and Peacock2023). Recent experimental work highlighted the role of microbial cell surfaces in the growth and nucleation of Fe-S minerals, therefore suggesting that some type of association must form between microbial organic carbon and Fe-S minerals (Picard et al., Reference Picard, Gartman, Clarke and Girguis2018). It has been suggested that Fe-S minerals might play an important role in the preservation of organic carbon in anoxic marine sediments (Barber et al., Reference Barber, Brandes, Leri, Lalonde, Balind, Wirick, Wang and Gélinas2017).

In laboratory experiments, Fe-S minerals can bind organic matter from various sources: live and dead microbial cells (Herbert et al., Reference Herbert, Benner, Pratt and Blowes1998; Picard et al., Reference Picard, Gartman, Cosmidis, Obst, Vidoudez, Clarke and Girguis2019, Reference Picard, Gartman and Girguis2021; Nabeh et al., Reference Nabeh, Brokaw and Picard2022; Truong et al., Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023), simple organic molecules (i.e. sugars, amino acids) and complex organic mixtures used in microbiological media (i.e. tryptone and yeast extract) (Picard et al., Reference Picard, Gartman and Girguis2021; Nabeh et al., Reference Nabeh, Brokaw and Picard2022), water-soluble extracts from aged compost soil, microalgae biomass and corn leaf (Tétrault and Gélinas, Reference Tétrault and Gélinas2022). The capacity of mackinawite to sequester organic carbon is at least comparable to that of ferrihydrite (Wang et al., Reference Wang, Zhang, Han, Sun, Jin, Yang, Yang, Hao, Liu and Xing2019; Ma et al., Reference Ma, Zhu, Yang, Li, Li, Liu and Li2022). Light elements (e.g. C, O) can be detected in Fe-S minerals by energy dispersive X-ray spectroscopy (EDS) in the scanning electron microscope (SEM) (Picard et al., Reference Picard, Gartman, Cosmidis, Obst, Vidoudez, Clarke and Girguis2019). To characterize the speciation and redox state of light elements, the use of spectroscopic methods, such as X-ray photoelectron spectroscopy (XPS) and near-edge X-ray absorption fine structure (NEXAFS) spectroscopy, is required. These methods identified C and N in biogenic Fe-S minerals originating from organic molecules, and O originating from organic functional groups rather than from oxidation of Fe-S minerals (Herbert et al., Reference Herbert, Benner, Pratt and Blowes1998; Picard et al., Reference Picard, Gartman, Cosmidis, Obst, Vidoudez, Clarke and Girguis2019, Reference Picard, Gartman and Girguis2021). The use of scanning transmission X-ray microscopy (STXM), coupled with NEXAFS spectroscopy, determined that organic carbon is homogenously distributed on biogenic Fe-S mineral aggregates precipitated with SRB, on biogenic pyrite and Fe-S minerals produced in cultures of sulfur-reducing archaea, and on abiotic minerals precipitated with simple and complex organic mixtures (Picard et al., Reference Picard, Gartman, Cosmidis, Obst, Vidoudez, Clarke and Girguis2019, Reference Picard, Gartman and Girguis2021; Truong et al., Reference Truong, Bernard, Le Pape, Morin, Baya, Merrot, Gorlas and Guyot2023). Some organic carbon also occurs as ‘hot spots’ in mineral aggregates precipitated with SRB (whether Fe²⁺ was in the growth medium or added after growth), interpreted as the preservation of intact cells and contents (Picard et al., Reference Picard, Gartman, Cosmidis, Obst, Vidoudez, Clarke and Girguis2019, Reference Picard, Gartman and Girguis2021).

In long-term laboratory experiments, Fe-S minerals could sequester significant amounts of organic carbon and nitrogen, as quantified by elemental analysis of solid phases (Nabeh et al., Reference Nabeh, Brokaw and Picard2022). The sequestered amounts decreased as a function of time but stabilized rapidly within a few months (Nabeh et al., Reference Nabeh, Brokaw and Picard2022). Semi-quantitative analysis from STXM/NEXAFS data also (Picard et al., Reference Picard, Gartman and Girguis2021). The highest stabilized levels of C and N were 13.5 w/w% and 3.3 w/w%, respectively. In a marine medium, the organic carbon removal capacity of Fe-S minerals was higher when microbial biomass was present (42–51% organic carbon removed from medium) than when organic mixtures that did not contain cells (e.g. amino acid mixtures, tryptone, yeast extract) were used (1.2–5.2% organic carbon removed from medium). In freshwater medium, the removal capacity of microbial biomass was even higher than in marine medium (67–137%; values above 100% linked to uncertainties in elemental analysis of biomass) (Nabeh et al., Reference Nabeh, Brokaw and Picard2022).

The nature of associations between Fe-S minerals and organic carbon has yet to be thoroughly investigated. Quantification of organic carbon in the studies reported above was done after several washes with anoxic ultrapure water, therefore removing loosely bound and water-soluble organics. Spectroscopic studies indicated that glucose and mannose bind less to Fe-S minerals than protein-rich organic mixtures (Picard et al., Reference Picard, Gartman and Girguis2021). Synthetic mackinawite studied with Fourier transform infrared spectroscopy (FTIR) revealed that it mostly associates with polysaccharides when adsorbed to extracts from natural organic matter (corn leaf, marine algae and soil) (Tetrault and Gelinas Reference Tétrault and Gélinas2022). The only study on natural pyrite (n = 5 framboids) determined fairly high organic carbon contents of 2.8–6.5 wt.% (Tribovillard et al., Reference Tribovillard, Bout-Roumazeilles, Delattre, Ventalon and Bensadok2022). Overall, the studies indicated a high association between Fe-S minerals and organic carbon, which needs to be investigated further in the context of climate and long-term organic carbon preservation (Fig. 7).

Figure 7.

Summary of the relevance of biogenic Fe-S minerals in various research fields and current societal issues.

Pyrite oxidation in the context of acidic drainages, metal recovery and nitrate removal

Pyrite oxidation in nature has been a subject of intense interest due to its key role in generating acid mine drainages (AMD) (Fig. 7). Decades of ore and metal mining have left pyrite-containing deposits exposed to the air and susceptible to oxidation mediated by aerobic Fe(II) and sulfur-oxidizing microorganisms, generating Fe³⁺ and sulfuric acid. Further oxidation of pyrite by Fe³⁺ generates a feedback loop that amplifies the oxidative reaction, resulting in AMD with low pH (< 5) and elevated toxic metals (e.g. Fe, Ni, Co, Cu, Zn) that negatively affect ecosystems worldwide (Baker and Banfield, Reference Baker and Banfield2003).

Despite its negative associations, research on acidic drainages presents some exciting opportunities to understand life in extreme environments. It is being increasingly recognized that some acidic drainages are natural and are better termed acid rock drainages (ARD) to distinguish them from anthropogenically-generated AMD. The most famous example is perhaps the Río Tinto, a 92 km long river that drains the Iberian Pyrite Belt in southwest Spain. Biogeochemical conditions in such acidic environments are similar to those that could have been found on ancient Mars as well as on early Earth (Amils and Fernández-Remolar, Reference Amils, Fernández-Remolar, Seckbah and Stan-Lotter2021). In particular, it was proposed that widespread ARDs occurred on the early Earth ~2.4 billion years ago, due to the GOE that accelerated terrestrial pyrite weathering (Konhauser et al., Reference Konhauser, Lalonde, Planavsky, Pecoits, Lyons, Mojzsis, Rouxel, Barley, Rosìere, Fralick, Kump and Bekker2011). Thus, modern AMD and ARD act as analogues to provide valuable insights into ancient life and biogeochemical cycles that may have operated in the past and other habitable worlds.

Further downstream, the toxicity of acidic drainages is mitigated by several processes including hydrological dilution, pH buffering by bedrock (e.g. limestone) and sulfate reduction by SRM. Enhancing the activity of microbial sulfate reduction is of particular interest in bioremediation and resource recovery, as sulfate reduction increases pH and generates sulfide that reacts readily with metals, immobilizing them as various metal sulfides (e.g. FeS, CuS, ZnS, NiS). The precious metals can then be recovered and used for other applications (Fig. 7). Selective metal recovery has been demonstrated in some cases using precise pH control in bioreactors, although the efficiency highly depends on the specific composition of the feed solution (reviewed in Johnson and Sánchez-Andrea, Reference Johnson and Sánchez-Andrea2019).

Besides aerobic pyrite oxidation, the environmental relevance of pyrite oxidation under anoxic conditions is being increasingly recognized. In fact, most of the pyrite oxidation that contributes to the aforementioned Río Tinto system is now thought to come primarily from subsurface pyrite oxidation coupled with nitrate reduction by anaerobic microorganisms (Amils et al., Reference Amils, Escudero, Oggerin, Puente Sánchez, Arce Rodríguez, Fernández Remolar, Rodríguez, García Villadangos, Sanz, Briones, Sánchez, Gómez, Leandro, Moreno-Paz, Prieto-Ballesteros, Molina, Tornos, Sánchez-Andrea, Timmis, Pieper and Parro2023). This process is also one of the major controls for the fate of agriculturally-sourced nitrate, a pollutant of ground- and drinking water in various parts of the world. It has been shown that pyrite of different sizes and shapes are associated with different degrees of pyrite oxidation and nitrate removal, and whether the final product is dissolved ammonium, harmless N₂ gas or the greenhouse gas N₂O (Bosch et al., Reference Bosch, Lee, Jordan, K-W and Meckenstock2011; Yan et al., Reference Yan, Kappler, Muehe, Knorr, Horn, Poser, Lohmayer and Peiffer2019; Mansor and Xu, Reference Mansor and Xu2020; Pang and Wang, Reference Pang and Wang2020; Jakus et al., Reference Jakus, Mellage, Carmen, Maisch, Byrne, Mueller, Grathwohl and Kappler2021; Kappler et al., Reference Kappler, Bryce, Mansor, Lueder, Byrne and Swanner2021). Subsurface pyrite oxidation is also an important factor to consider in engineering, as this process can lead to ground collapse and significant loss of economic and public life (Czerewko and Cripps, Reference Czerewko and Cripps2023).

While this section is focused primarily on pyrite oxidation, it is important to recognize that redox heterogeneity is expected in space and time (Peiffer et al., Reference Peiffer, Kappler, Haderlein, Schmidt, Byrne, Kleindienst, Vogt, Richnow, Obst, Angenent, Bryce, McCammon and Planer-Friedrich2021). For example, sediment cores in Río Tinto are black within a few centimetres of depth, which can be attributed to sulfate reduction at depth (Sánchez-Andrea et al., Reference Sánchez-Andrea, Knittel, Amann, Amils and Sanz2012) and the formation of new metal sulfides that probably include pyrite. In the subsurface, hydrological fluctuations over time can shift redox conditions from being conducive to pyrite oxidation to being conducive to pyrite formation via the activity of sulfate/sulfur/iron-reducing microorganisms. It is likely that new pyrite is continuously being formed and redissolved in the environment. Thus, investigating biogenic pyrite formation will be key to understanding the reactivity of these dynamic phases.

Biogenic pyrite as future photovoltaics and semiconductors

As the world’s population continues to increase, there is a need for environmentally sustainable solutions to meet the future demands in energy. The optimization of energy gain from solar cells and output efficiencies of various electronics are considered high priorities in this regard. Silicon-based materials have traditionally dominated the field of photovoltaics and semiconductors. In recent years, it has been recognized that pyrite has the potential to replace silicon-based materials at a lower cost and with higher efficiency (Fig. 7). Pyrite has a suitable band gap (energy gap between two electron states) around 0.95 eV, which is neither too small (i.e. as a conductor) or too large (i.e. insulator) for it to function as a semiconductor. It further has a light absorption coefficient that is about two orders of magnitude higher than silicon. The relative abundance and non-toxicity of pyrite components (e.g. Fe, S) also make it more attractive compared to other explored materials (Wadia et al., Reference Wadia, Alivisatos and Kammen2009).

For pyrite to be feasible as an energy material of the future, a method to obtain high-quality pyrite in sufficient quantities needs to be developed. Thin films of pyrite have initially been synthesized through various methods such as hydro/solvothermal, hot injection and vapour deposition. Modifications of these procedures can lead to different sizes, shapes and trace metal contents, which allow for tunable band gap energy and electrical current production. These methods often require high temperatures and toxic solvents and tend to produce impurities such as the high-temperature phase marcasite (orthorhombic FeS₂) and pyrrhotite (Fe₇S₈) (Zaka et al., Reference Zaka, Alhassan and Nayfeh2022). High-energy milling of pyrite ore can be used to obtain nanoparticles prior to thin film preparation, but there are concerns that this will exacerbate environmental problems associated with acid mine drainages. Furthermore, robust supplies of high-quality pyrite ore cannot be guaranteed as there are huge variations in the semiconductive properties of pyrite even within a deposit (Wang et al., Reference Wang, Shen, Du, Xu, Zhang and Liu2021).

Could pyrite in the future be grown as a biomanufactured material? (Cosmidis, Reference Cosmidis2023). This approach takes its inspiration from nature, which has produced pyrite sustainably at a rate of ~200,000 tons per day for the last millions of years (Rickard, Reference Rickard2015). A mechanistic understanding of biogenic pyrite formation will be essential to allow for tunable bio-synthesis. The gap between laboratory experiments to large-scale production is vast. However, achieving this will be a worthwhile target.