Framboidal pyrite has been used as a paleo-redox proxy and a biomarker in ancient sediments, but the interpretation of pyrite framboids can be controversial, especially where later overgrowths have obscured primary textures. Here we show how nano-scale chemical mapping of organic carbon and nitrogen (CNorg) can detect relict framboids within Precambrian pyrite grains and determine their formation mechanism. Pyrite grains associated with an Ediacaran fossil Lagerstätte from Newfoundland (ca. 560 Ma) hold significance for our understanding of taphonomy and redox history of the earliest macrofossil assemblages. They show distinct chemical zoning with respect to CNorg. Relict framboids are revealed as spheroidal zones within larger pyrite grains, whereby pure pyrite microcrystals are enclosed by a mesh-like matrix of pyrite possessing elevated CNorg, replicating observations from framboids growing within modern biofilms. Subsequent pyrite overgrowths also incorporated CNorg from biofilms, with concentric CNorg zoning showing that the availability of CNorg progressively decreased during later pyrite growth. Multiple framboids are commonly cemented together by these overgrowths to form larger grains, with relict framboids only detectable in CNorg maps. In situ sulfur isotope data (δ34S = ∼−24‰ to −15‰) show that the source of sulfur for the pyrite was also biologically mediated, most likely via a sulfate-reducing microbial metabolism within the biofilms. Relict framboids have significantly smaller diameters than the pyrite grains that enclose them, suggesting that the use of framboid diameters to infer water column paleo-redox conditions should be approached with caution. This work shows that pyrite framboids have formed within organic biofilms for at least 560 m.y., and provides a novel methodology that could readily be extended to search for such biomarkers in older rocks and potentially on other planets.


Framboidal pyrite is a common component of the geological record, frequently being the most abundant pyrite texture in ancient sediments, but its formation mechanism has long been debated (Papunen, 1966; Ohfuji and Rickard, 2005; Ohfuji et al., 2005; Rickard, 2012). Pyrite framboids are defined as microscopic spheroidal to subspheroidal clusters of equidimensional and equimorphic pyrite microcrystals (Ohfuji and Rickard, 2005). A single framboid may contain up to 106 approximately cubic or octahedral pyrite microcrystals, and may be 1–250 μm across (Ohfuji and Rickard, 2005), although they are most commonly 10–20 μm in diameter (Wilkin et al., 1996; Wang et al., 2012). Pyrite framboids have been used as a proxy for local redox conditions in paleo-environmental reconstructions, with their size distributions used to discriminate between formation within euxinic water columns and formation in sediments below oxygenated water columns (Wilkin et al., 1996; Wang et al., 2012). They have also been suggested as potential biomarkers in very ancient sediments or on other planets (Popa et al., 2004; MacLean et al., 2008). Hence, there is a pressing need for robust ways to identify framboids, accurately measure their size distributions, and determine their biogenicity throughout the geological record.

Early studies noted a frequent association of organic matter with pyrite framboids, leading to the suggestion that their characteristic texture was directly controlled by biology, with some studies speculating that framboids were pyritized microfossils (e.g., Love, 1957). However, the discovery of framboids in high-temperature volcanic and hydrothermal settings (Love and Amstutz, 1969), plus the experimental synthesis of pyrite framboids in the laboratory without the presence of organic material (Sweeney and Kaplan, 1973), indicated that biology was not a prerequisite for framboid formation. Substantial debate followed about the extent to which biology contributes to framboid formation (e.g., Ohfuji and Rickard, 2005; Kohn et al., 1998). Regarding modern low-temperature sedimentary environments, much of this debate was resolved by the work of Large et al. (2001) and MacLean et al. (2008). These authors used high-spatial-resolution cryogenic scanning electron microscopy (SEM) (Large et al., 2001), plus focused ion beam SEM and X-ray spectroscopy (MacLean et al., 2008), to demonstrate the presence of biofilms coating both the outer surface of complete pyrite framboids and the surfaces of individual microcrystals within a framboid. Partially formed “proto-framboids” were found to be embedded in particularly large quantities of biofilm and possessed microcrystals with anhedral crystal faces, suggesting that biofilms provide an organic template (constrained growth space) for the growth and aggregation of pyrite microcrystals (MacLean et al., 2008). Furthermore, the polysaccharide-dominated surfaces of biofilms have a strong affinity for Fe2+ ions, providing ideal nucleation sites for iron sulfides, and may also play a role in stabilizing the framboids during sediment compaction or disturbance (Large et al., 2001).

In ancient environments, however, where significant pyrite recrystallization may have taken place and framboid-containing rocks may have experienced both low-temperature and high-temperature conditions (cf. Scott et al., 2009), it is more difficult to securely identify pyrite framboids and to demonstrate a biological formation mechanism. Some ancient framboids still retain their characteristic morphology when viewed under reflected light or SEM, but many others, such as those studied here, may be “hidden” within larger grains. Chemical etching may hint at hidden framboids (Rickard and Zweifel, 1975), and δ34S data may indicate whether the sulfur incorporated into framboids has a biogenic source (Kohn et al., 1998), but these data do not reveal whether framboid growth occurred within a biological matrix. Furthermore, the small size of framboids means that conventional bulk isotopic and elemental analyses lack the spatial resolution required to provide meaningful data. Here we combine in situ secondary ion mass spectrometry (SIMS) and transmission electron microscopy (TEM) to provide a new way to detect and measure relict framboids within ancient pyrite grains, and evaluate the contribution of biology to their nucleation and growth mechanisms.


Ion mapping was performed on portions of standard geological thin sections using a CAMECA NanoSIMS 50, with instrument parameters optimized as described by Wacey et al. (2011). TEM wafers were extracted from geological thin sections using a FEI xT Nova NanoLab 200 focused ion beam SEM, and TEM data were obtained using a FEI Titan G2 80–200 TEM/STEM with ChemiSTEM Technology, plus a JEOL 2100 LaB6 TEM. Sulfur isotope data were obtained using a CAMECA NanoSIMS 50 and a CAMECA IMS 1280, following protocols described by McLoughlin et al. (2012) and Farquhar et al. (2013), respectively. For detailed methods, see the GSA Data Repository1.


Pyrite Chemistry and Nano-Texture

Turbiditic siltstones of the ca. 560 Ma Fermeuse Formation at Back Cove, Bonavista Peninsula, Newfoundland (Canada), contain clusters and laminae of small (<50 μm) pyrite grains (Fig. DR1 in the Data Repository). Pyrite morphology ranges from rounded through to subhedral and euhedral cubes. Secondary electron images and NanoSIMS sulfur (34S) ion maps show no indication of framboidal morphologies; indeed they indicate a rather homogenous textural and chemical composition for all pyrite grains (Figs. 1 and 2C), with no significant differences between small rounded grains and larger euhedral grains (compare the 34S images of a small rounded grain at the top of Fig. 1 with the large grain in Fig. 2). In contrast, carbon (12C) and nitrogen (26CN) NanoSIMS ion maps reveal distinct nano-scale chemical zoning within and between these grains. Many pyrite grains possess an inner spheroidal zone (ISZ; Fig. 1, dashed circle in enlarged 26CN map) comprising numerous microcrystals of pure pyrite (Fig. 1, black cuboids in enlarged 26CN map) set within a mesh-like matrix of C-rich and N-rich pyrite (Fig. 1, pink areas within dashed circle in 26CN map). The ISZ is usually surrounded by an outer zone (OZ) of C-rich and N-rich pyrite (Fig. 1, outer blue and pink zones). Co-occurrence of C and N, combined with the absence of ions such as Ca or O, indicates that the C and N signals come from organic material preserved within the pyrite grains (from this point on referred to as CNorg).

TEM images of ultrathin (∼100 nm) cross sections through well-preserved grains demonstrate the nano-texture of the ISZ and the OZ, and also show that the grains have a distinct thin (<500 nm) outer rim (Figs. 2D and 2E). In the ISZ, pyrite microcrystals are closely packed, as expected for framboidal pyrite (cf. Ohfuji and Rickard, 2005), and each of these pyrite microcrystals has a thin (∼50–100 nm) nano-porous rim (Figs. 2D and 2E, arrows). The OZ also has a nano-porous texture, and these pores tend to be slightly larger than those in the ISZ (Fig. 2D). The nano-porous rims of the ISZ together with the nano-porous OZ correspond precisely to the areas of CNorg enrichment seen in NanoSIMS ion maps. Unfortunately, the relatively poor detection limits for N, together with the extreme thinness of the TEM sample, preclude accurate mapping of CNorg in the TEM. Only relatively large clumps of CNorg at some crystal boundaries can be visualized in the TEM (Fig. DR2). We suggest that the nano-pores house the remaining CNorg seen in NanoSIMS maps, but it is also possible that CNorg is held within the pyrite crystal lattice itself in these areas. Arsenic and nickel occur in trace amounts restricted to the nano-porous rims of the ISZ (and to a lesser amount in the OZ), correlating with the CNorg enrichment seen in the NanoSIMS maps (Fig. DR2). At the boundary between the ISZ and the OZ, the ISZ microcrystals are sharply terminated against the OZ (Fig. 2D, white line) and are not equimorphic with microcrystals toward the center of the ISZ. This suggests either recrystallization of outer portions of the ISZ to form the OZ or a fairly rapid change in local chemistry, such as reduction of free iron concentration, that changed pyrite growth morphology. The thin outer rim records minor late-stage oxidation of pyrite to iron oxide (Fig. DR3).

Some single pyrite grains possess several ISZs separated and surrounded by OZ pyrite (Figs. 2A and 2B; Fig. DR4). In these cases, the grains commonly exhibit a rather euhedral shape that disguises their original formation mechanism (Fig. 2). In other examples, the OZ is chemically heterogenous in CNorg, with clear concentric zoning marked by outward-decreasing CNorg contents (Fig. DR5). Occasionally the CNorg chemical microstructure of the ISZ is completely lost and only a spheroid of pure pyrite is seen (Fig. DR6).

We interpret the ISZ in these grains as relict primary pyrite framboids and the OZ as secondary pyrite. For those pyrite grains that possess several ISZs, these are interpreted as multiple, closely spaced relict primary framboids that were cemented together by secondary pyrite overgrowths. The distribution of CNorg in most ISZs closely resembles the pattern of organic enrichment seen in modern framboidal pyrite that nucleated and grew within biofilms (Large et al., 2001). The nano-texture of the ISZ is also nearly identical to that seen in modern biologically mediated framboids, where every pyrite microcrystal in the framboid is enclosed by a thin biofilm (compare Figs. 2D and 2E with MacLean et al. [2008], their figure 2c). Hence, we conclude that our 560 Ma framboids likewise nucleated and grew within biofilms, or within some other organic material such as extracellular polymeric substances. Wispy carbonaceous laminae observed in thin sections close to the framboid-bearing layer (Fig. DR1a) likely represent the remains of biofilms. Biofilms provide a number of favorable conditions for framboid formation: (1) the organic framework provides a pre-existing confined growth space that can control crystal size and morphology (MacLean et al., 2008); (2) biofilms replicate conditions of high Fe and S supersaturation (required for successful experimental abiotic precipitation of pyrite framboids; Ohfuji and Rickard, 2005) so that nucleation rate is significantly greater than crystal growth rate; (3) biofilms contain large amounts of polysaccharides that have a high Fe2+ binding capacity, enhancing crystal nucleation (Flemming, 1995; likewise, Ni and As could also be preferentially bound by biofilms, consistent with their enrichment in CNorg zones); (4) cell walls within biofilms provide further preferred nucleation sites for metal sulfides (Ferris et al., 1987); and (5) metal-reducing and sulfate-reducing bacteria in biofilms can provide a local source of reactive iron and sulfide (Rickard, 2012). In modern framboids, a layer of biofilm also tends to enclose the entire framboid (Large et al., 2001; MacLean et al., 2008). While this could equate to the OZs observed in our pyrite grains, our OZs are much thicker than modern biofilm coatings. This, together with our nano-textural observations above, suggests that the OZ is instead a zone of secondary overgrowth and/or recrystallization.

Insights from Sulfur Isotope Geochemistry

Sulfur isotope data (Table DR1; Fig. 3) were obtained in situ from polished geological thin sections using Cameca NanoSIMS 50 and Cameca IMS 1280 ion probes. Pyrite grains show a uniformly light δ34S signal (−15.2‰ to −24.3‰; mean = −21.5‰; n = 33). There is no significant difference in data obtained using NanoSIMS (mean = −21.7‰) and IMS 1280 (mean = −21.4‰).

These spatially resolved δ34S data inform on the source of the sulfur for the pyrite. Taking δ34S = +25‰ as a mean estimate for the isotopic composition of seawater sulfate at 560 Ma (Fike et al., 2006), then the pyrite grains show maximum fractionations from Ediacaran seawater sulfate (Δδ34S) of almost 50‰ (Δδ34S = 40.2‰–49.3‰). This clearly shows biological processing of sulfur prior to incorporation into the pyrite. Such large Δδ34S fractionations may occur in two ways: (1) during microbial sulfate reduction (MSR) under conditions of limited electron donor supply and/or poor reactivity of organic material (Leavitt et al., 2013); and (2) during oxidative sulfur cycling where fractionations associated with MSR are supplemented by those occurring during disproportionation of oxidized sulfide (Canfield and Thamdrup, 1994). We favor the former mechanism here due to the inferred deep-sea setting (Hofmann et al., 2008), leading to low Corg delivery and relatively low total organic carbon contents of these sediments (mostly <0.1 wt%; Canfield et al., 2007). Although microbial biomorphs have not yet been observed in the pyrite grains, it is likely that sulfate-reducing bacteria were part of the living biofilm in which the framboids nucleated and grew, producing a localized source of 34S-depleted H2S for incorporation into the pyrite microcrystals.

δ34S does not change significantly with distance from the center of a pyrite grain (Fig. 3). This supports our earlier hypothesis that pyrite overgrowths could have resulted from recrystallization of primary framboidal zones (hence inheriting their isotopic signal). This mechanism is also consistent with the homogenous enrichment of CNorg in the overgrowths of many grains; here, CNorg that was present in biofilms coating framboid microcrystals could have been redistributed and incorporated into nano-pores or the crystal lattice of pyrite as it recrystallized. For those grains with concentrically zoned and outward-diminishing CNorg contents (e.g., Fig. DR5), we infer that recrystallization was followed by continued early diagenetic pyrite growth in pore waters that were CNorg poor but were still open to a supply of iron and seawater sulfate (hence, no Rayleigh-type 34S enrichment in the δ34S data).

Implications for the Use of Framboid Diameters as a Paleo-Redox Proxy

Framboid diameters have been used to determine whether pyrite precipitation took place within a euxinic water column or within sediments below an oxic or dysoxic water column (e.g., Wilkin et al., 1996; Wang et al., 2012). In modern environments, framboids precipitated within a euxinic water column over a short time span tend to be <10 μm in diameter, whereas those precipitated in sediment pore waters have larger and more variable diameters (Wilkin et al., 1996). This protocol has been extended into the geological record and used to infer the oxidation state of some ancient water columns (Wang et al., 2012). However, when we compare the diameters of the relict framboids (defined as the ISZ in the NanoSIMS CNorg maps) with the total diameter of the enclosing grains, we find that the relict framboids make up only 15% to 71% of total grain diameter (n = 28). Moreover, several enclosing grains retain a spheroidal, pseudo-framboidal shape despite clearly not being primary framboids. These data suggest that simple transects using light or electron microscopy are insufficient to determine the portion of a framboid that is primary in nature, and hence extreme caution should be exercised if attempting to use framboid diameters as paleo-redox proxies.


Pyrite grains from 560 Ma sediments exhibit distinctive distributions of CNorg that highlight zones of primary framboid growth and subsequent pyrite overgrowths. The pattern of CNorg enrichment correlates with pyrite nano-textures showing that organic material is retained at grain boundaries between pyrite microcrystals and in thin nano-porous rims coating each microcrystal. Organic material is also found preserved in pyrite overgrowth zones but commonly decreases outwards with progressive overgrowth. The CNorg distribution permits accurate measurements of relict framboid versus overgrowth diameters, which is essential if framboids are to be used as paleo-redox proxies. Our data show that 560 Ma framboids nucleated and grew within organic biofilms, extending geological evidence for this growth mechanism back into the Precambrian. These biofilms also contained sulfate-reducing bacteria, and perhaps also metal-reducing bacteria, that provided a local source of reduced S and Fe for the pyrite. Our work provides a straightforward protocol for finding hidden framboids in larger euhedral crystals, and for determining the biogenicity of ancient framboids, enhancing their potential as biomarkers of the early Earth and on other planets.

We acknowledge the Australian Microscopy & Microanalysis Research Facility at The University of Western Australia and The University of New South Wales. These facilities are funded by the universities and State and Commonwealth governments. Wacey is funded by the Australian Research Council, via a grant to the Centre of Excellence for Core to Crust Fluid Systems.

1GSA Data Repository item 2015025, supplementary Figures DR1–DR6, Table DR1 (sulfur isotope data), and detailed methods, is available online at www.geosociety.org/pubs/ft2015.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.