Framboidal pyrite formation via the oxidation of iron (II) monosulfide by hydrogen sulphide

Geochimica et Cosmochimica Acta, Vol. 64, No. 15, pp. 2665–2672, 2000 Copyright © 2000 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/00 $20.00 ⫹ .00

Pergamon

PII S0016-7037(00)00387-2

Framboidal pyrite formation via the oxidation of iron (II) monosulfide by hydrogen sulphide IAN B. BUTLER* and DAVID RICKARD Department of Earth Sciences, Cardiff University, Park Place, Cardiff CF10 3YE, Wales, UK (Received September 27, 1999; accepted in revised form March 14, 2000)

Abstract—Pyrite framboids have been synthesised as the product of the oxidation of FeS (mackinawite) by H2S in aqueous solutions at pH 6, Eh ⬎ ⫺250 mV and temperatures above 60°C in the absence of detectable O2. Under these conditions at Eh ⬍ ⫺250 mV single crystals of pyrite are formed. The reaction may proceed by dissolution of solid FeS and subsequent reaction of an aqueous FeS cluster complex with H2S. The involvement of an aqueous FeS phase facilitates transport and reaction, and allows the development of framboidal textures from dispersed fine-grained FeS reactant. The results suggest that the framboidal texture results from rapid nucleation in environments where pyrite is strongly supersaturated. At the limits of the pyrite stability field, where pyrite supersaturation is low, rapid pyrite nucleation is inhibited and single crystals form. The results demonstrate that pyrite framboids can form in the absence of molecular O2, magnetic (e.g., greigite) intermediates or biological intervention. Copyright © 2000 Elsevier Science Ltd The spectrum of models which aim to describe framboid forming processes is broad. Schneiderhohn (1923) and Love (1957) proposed that the texture was the result of pyritization of individual bacteria and bacterial colonies. Papunen (1966), Kalliokoski and Cathles (1969), Kribek (1975) and Raiswell et al. (1993) developed models based upon the pyritization of organic particles or colloids. Other workers (e.g., Rust, 1935; Berner, 1969; Sweeney and Kaplan, 1973; Wilkin and Barnes, 1997a) have favoured low temperature abiotic transformations within the Fe-S system. At present, it is the experimental work of Sweeney and Kaplan (1973) which forms the basis of models for framboidal pyrite formation (Wilkin and Barnes, 1997a), and it is their work that has implicated the magnetic thiospinel greigite (Fe3S4) as the essential prerequisite for framboidal pyrite formation. In this contribution, we provide experimental evidence to show that framboid formation can proceed without a greigite intermediate, and may result from the oxidation of aqueous iron (II) monosulphide by hydrogen sulphide. We discuss this result with reference to distributions of reactant FeS in sediments, and the mechanism of the FeS ⫹ H2S reaction.

1. INTRODUCTION

The framboidal texture of pyrite is abundant in sediments and ore deposits. Its ubiquity and remarkable physical appearance has led to the publication of a considerable literature dealing with descriptions of the texture, its laboratory synthesis, and speculating upon its mechanism of formation. However, despite the wealth of observational and experimental studies, there remains no satisfactory model that describes the process by which framboidal pyrite forms. The term framboid, from the French framboise, was first coined by Rust (1935) and pertains to the characteristic raspberry-like morphology of the texture. Essentially, a framboid is a spherical or sub-spherical structure composed of numerous microcrystals which are often equant and equidimensional (Fig. 1). Pyrite framboids commonly display highly ordered 3D arrays of close packed microcrystals (e.g., Love and Amstutz, 1966), adjacent sub-domains of microcrystals with differing packing patterns (e.g., Rickard 1970), and even a polyhedral rather than spheroidal form (Morrissey, 1972; Butler, 1994). Individual microcrystals can display a full range of crystal forms from cube to pyritohedron, although the form tends to be constant within any single framboid, and they may display rapid growth related features such as skeletal form (Butler, 1994). In sediments, framboids show a positively skewed size distribution (Love and Amstutz, 1966; Wilkin et al., 1996) with a modal size within the range 1–10 ␮m, and rare examples in excess of 50 ␮m in diameter. Microcrystals range from 0.2 to 2 ␮m, and measured ratios of microcryst to framboid diameters fall within the range 5 to 30 (Wilkin et al., 1996). Thus Wilkin et al. (1996) were able to calculate that framboids consist of between 102 and 105 discrete pyrite microcrystals, a similar, but narrower, range to that suggested by Rickard (1970) based upon theoretical packing considerations.

2. EXPERIMENTAL For this study, the methods of Rickard (1997) were utilised to allow the synthesis of framboids under conditions where oxygen is excluded. Reactions were performed both in heat sealed glass vessels (Rickard, 1997) and in sealed serum bottles fitted with rubber septa (Drobner et al., 1990). Vessels were charged with 400 mg of freshly precipitated, freeze dried, amorphous FeS in an N2 filled glove-box, and then 10 ml of HydrionTM pH 6 buffer (potassium hydrogen phosphate/sodium dihydrogen phosphate) which had been purged with N2 to reduce O2 levels to less than 0.2 ppm (Butler et al., 1994) was added. The freeze drying method was optimised to reduce air exposure of the reactant FeS to a minimum. After filtration under flowing N2, the FeS was transferred in a container to the freeze drier, and rapidly pumped down to vacuum. After drying, the freeze drier was flooded with N2 gas before the FeS was removed and transferred to a freezer and stored under N2 gas. In some reaction runs, the Eh of the solution was poised using Ti (III) citrate solution. Addition of 1 ml gives ⫺400 mV, and addition of 0.5 ml gives ⫺250 mV in 10 ml of pH 6 buffer at ambient temperatures (these values were measured using a standard Pt Eh probe). The Ti (III)

*Author to whom correspondence should be addressed (ButlerIB@Cardiff. ac.uk). 2665

2666

I. B. Butler and D. Rickard citrate solution was prepared according to the method of Zehnder and Wuhrmann (1976) and the maximum citrate concentration in the reaction vessels at the start of a reaction run was 0.02 M. Reaction vessels were attached to a gas-transfer manifold and repeatedly flushed with analytical grade N2 which had been passed through a Supelco high capacity gas purifier to remove trace O2. Addition of H2S gas, generated by acid decomposition of Na2S 䡠 9H2O, was performed at this stage. Reaction vessels were heated submerged in a water bath and allowed to proceed unagitated at temperatures of 60 –95°C for periods of 5 to 45 d. Reactions at 100 –140° were performed in a reaction oven. After reaction and extraction, solid reaction products were analysed using Debye-Scherrer XRD (CoK␣ radiation) and imaged using a Cambridge Instruments Stereoscan S360 analytical SEM. In situ analyses of reaction products were performed using the EDAXTM system. Extracted gas from sealed glass reaction vessels was analysed using a PerkinElmer GC with a PorapakQTM column and a thermal conductivity detector (TCD). 3. RESULTS

Fig. 1. SEM photomicrographs of framboidal pyrite from the Chattanooga Shale, Upper Devonian, USA. A) A framboid composed of octahedral microcrystals which have developed an interpenetrant habit. The microcrystals share no particular orientation, and the framboid appears to be disordered. B) A subspheroidal framboid composed of numerous cubo-octahedral microcrystals. Individual microcrystals are packed into a regular 3-D array within the framboid, and share a common orientation. The range of framboid morphology, and the diversity of environments in which they form indicate a robust formation mechanism.

The results of the experiments are summarised in Table 1. In all experiments, pyrite (and trace mackinawite) was the only product identifiable by XRD, and EDAXTM analysis of the reaction products returned Fe:S ratios consistent with FeS2 stoichiometry, based on a pyrite standard. There was no evidence for stoichiometries other than FeS2, and reaction products showed no magnetic response. Details of the textures of reactants and products are shown in Figure 2. The amorphous FeS reactant (Fig. 2a) is an extremely fine grained material, the individual particles of which cannot be resolved on the SEM, suggesting a particle diameter of less than 30 nm. Reaction with H2S generates three distinct product morphologies; euhedral pyrite (Fig. 2b), protoframboidal clusters (Fig. 2c), and framboidal pyrite (Fig. 2d–f). The texture of the product is apparently a function of the Eh of the initial reaction system. Reactions poised initially at ⫺400 mV produced euhedral pyrite as small (⬍0.5 ␮m), discrete octahedra. Reactions poised at ⫺250 mV produced protoframboidal clusters of the type described by Rickard (1997). The protoframboids are small sub-

Table 1. Reaction conditions and results. Moles H2S

Reaction temp °C

Reaction time (days)

Eh poise mV

Product

Texture Framboids Protoframboids N/A Protoframboids Euhedra Euhedra Euhedra Euhedra Framboids Framboids Framboids ⫹ Euhedra Framboids ⫹ Euhedra

Run

Reps

Mass FeS mg.

ac 1c 2a,c 3c 4 5 6 7 8 9 10

1 7 2 6 4 4 2 6 4 4 2

400 400 400 400 100 100 100 100 100 100 400

0.004 0.004 0.004 0.004 0.001 0.001 0.004 0.004 0.004 0.004 0.004

95 60 80 60 140 140 60 60 80 80 100

14 21 N/A 14 ⫹ 45b 14 14 15 14 7 14 5

none ⫺250 ⫺250 ⫺250 ⫺400 ⫺400 ⫺400 ⫺400 none none none

Pyrite (mk)d Pyrite (mk) Oxides Pyrite (mk) Pyrite (mk) Pyrite (mk) Pyrite (mk) Pyrite (mk) Pyrite (mk) Pyrite (mk) Pyrite (mk)

11

2

400

0.004

100

5

none

Pyrite (mk)

a

The reaction vessels leaked and the sulphide reactant was exposed to air. Oxidation was rapid and complete. No textural difference was observed between sulphides reacted for 14 days or 45 days. All textural transformations are therefore the result of the pyrite forming reaction. c Method of Drobner et al., (1990) used. All other runs use method of Rickard, 1997. d Mk refers to trace mackinawite revealed by XRD. b

Framboidal pyrite formation

2667

Fig. 2. SEM photomicrographs of experimental reactants and products. A) Amorphous FeS reactant. The reactant consists of clumps of fine FeS particles which cannot be resolved, suggesting a particle size of less than 30 nm. B) Reactions poised at ⬃⫺400 mV produce euhedral pyrite, here showing an octahedral habit. C) Reactions poised at ⬃⫺250 mV produce protoframboidal pyrite. The protoframboids are small and composed of poorly formed cuboidal pyrite microcrystals. D) Framboidal pyrite forms where no Eh poise is applied to the system. The surface morphology of each microcrystal is characteristic of these experimentally produced framboids. E) Framboidal products may be up to 5 ␮m in diameter and are composed of well formed cubic microcrystals. F) Circular surface features (indicated by arrows) occur on all experimentally produced framboid microcrystals and may represent screw dislocations formed during crystal growth.

spherical clusters of poorly formed cubic particles. Reactions carried out in the absence of an Eh poise (i.e. at Eh ⬎ ⫺250 mV) produced framboidal pyrite, and more rarely protoframboids and euhedral pyrite. Where framboids were formed at ⬍100°C, they were found to be the sole reaction product,

although runs 10 and 11 produced framboids at 100°C in association with euhedral pyrite. The framboids are spherical clusters with a narrow size distribution (diameter 2–5 ␮m) and consist of well formed cubic pyrite microcrystals, which are commonly interpen-

2668

I. B. Butler and D. Rickard

etrant. The microcrystals display circular features on their crystal faces (Fig. 2d,f). Similar structures were recognised on synthetic pyrite by Rickard (1997), and may represent screw dislocations developed on the microcrystal surface during growth. 4. DISCUSSION

Sweeney and Kaplan (1973) proposed that framboidal pyrite formation proceeds via the solid phase reaction of FeS and elemental sulphur: 3FeS(mk) ⫹ S°(s) 3 Fe3S4(gr)

(1)

Fe3S4(gr) ⫹ 2S°(s) 3 3FeS2(py)

(2)

is greigite, S°(s) is

where FeS(mk) is mackinawite, Fe3S4(gr) elemental sulphur and FeS2(py) is pyrite. Their work established that micro-organisms were not necessary for pyrite framboid formation and that the texture could be developed abiotically. Additionally, their results are the source of the implication of the magnetic thiospinel greigite (Fe3S4) as the essential precursor for framboid formation. It is important to note that the reaction as written in (1) and (2) is a solid phase process, and while it proceeds readily at elevated temperature, it appears to be relatively slow at ambient temperatures. Schoonen and Barnes (1991b) noted that the zerovalent sulphur in reactions (1) and (2) may be considered to represent any zero-valent sulphur species. The most reasonable 2⫺ candidate for such a reactant is polysulphide, SnS(aq) , but it should be noted that experimental studies using polysulphides (e.g., Rickard 1975; Luther, 1991) have always produced euhedral rather than framboidal pyrite. Wilkin and Barnes (1997a) have redefined the role of greigite as a precursor phase, attributing the framboidal form to the aggregation of greigite microcrystals under magnetic forces and their subsequent pyritization. They propose a general reaction scheme whereby greigite is formed via: ⫺ ⫹ 3FeS(mk) ⫹ HS(aq) ⫹ 1⁄2O2(aq) ⫹ H(aq) 3 Fe3S4(gr) ⫹ H2O(1)

(3)

⫹ 2⫹ 4FeS(mk) ⫹ 1⁄2O2(aq) ⫹ 2H(aq) 3 Fe3S4(gr) ⫹ Fe(aq) ⫹ H2O(l)

(4)

Pyritization of the aggregated greigite framboid may then proceed via the addition of sulphur (Eqn. 2) or through loss of Fe (Furukawa and Barnes, 1995): ⫹ 2⫹ Fe3S4(gr) ⫹ 2H(aq) 3 2FeS2(py) ⫹ Fe(s) ⫹ H2(g)

(5)

It is worth noting at this point that Wilkin and Barnes (1996) found that framboid formation was not an inevitable consequence of greigite formation, and they speculated that, in their experiments, the formation of framboidal pyrite must be primarily related to the rate of greigite formation. Clearly, in open systems reactions 3 and 4 may proceed at low O2 concentrations. However, if we consider reactions 3 and 4 within the context of our closed-system reaction vessels, and those used by Rickard (1997) (volume ⫽ 120 ml), then to transform 400 mg (4.5 mmol) of iron (II) monosulphide to greigite requires 0.56 to 0.75 millimoles of molecular oxygen, that is, an atmosphere containing between 12–15% O2. Measurement of the reactant solution using a Jenway 3200 dissolved oxygen electrode revealed no detectable O2 (i.e. ⬍ 0.01 ␮g 䡠 g⫺1), and GC analysis of the gas phase again

revealed no detectable O2 (i.e., ⬍0.01%). If conversion of greigite to pyrite was to proceed by sulphur addition (rather than iron loss as suggested by Furukawa and Barnes, 1995), then further molecular oxygen would be required. In fact, the presence of such O2 levels in sealed iron sulphide reaction systems results in rapid transformation to red Fe oxyhydroxides (Table 1). Work by Wilkin and Barnes, 1996 and Butler et al., (1999) shows that some oxidation of the FeS surface is required for pyrite formation. In fact it appears that O2 acts as a reaction initiator (Rickard, 1999) for the FeS H2S reaction which has been shown to kinetically inhibited if there is no oxidation of the FeS reactant (Wilkin and Barnes, 1996; Butler et al., 1999). Minor surficial oxidation is apparently related to the freezedrying process (Benning et al., in press) used to prepare the FeS reactant. The role of the oxidation purely as a reaction initiator is consistent with the results of Taylor et al. (1979) and Rickard (1997) who found both stoichiometric and non-stoichiometric amounts of H2 produced, consistent with the reaction of FeS and H2S to produce pyrite and hydrogen. We may further discount the role of greigite as a reaction intermediate or product in our systems because the reaction products show no magnetic response. Greigite is strongly ferrimagnetic, and even trace amounts which are undetectable using XRD methods cause the iron sulphide product to react strongly to a hand magnet. The synthetic production of apparently magnetic pyrite was first commented on by Schoonen and Barnes (1991b). Subsequently Butler (1994) and Butler and Schoonen (unpublished data) were able to demonstrate the persistence of magnetic response and greigitic Mo¨ssbauer spectra in pyrite formed from greigite, even after extended reaction durations at temperatures of 95°C. Moreover, Rickard (1997) examined experimental products formed after a few hours and showed that at no stage of the reaction between FeS and H2S is greigite formed as a reaction intermediate. There are no data to implicate greigite in the formation of the pyrite, framboidal or otherwise, produced in our experimental runs. Since many of our experiments resulted in the formation of framboids we conclude that greigite is not an essential prerequisite for the formation of framboids, and we contend that framboidal pyrite may form directly via the oxidation of amorphous FeS by aqueous H2S. The experiments of Wilkin and Barnes (1996) produced framboidal pyrite, apparently by the magnetic accretion of rapidly formed greigite particles. Since greigite is not present in our experimentation, it seems possible that framboidal textures may develop via different pathways. It is certainly true that the magnetic aggregation model of Wilkin and Barnes (1997a) is a potential process that is applicable to the formation of the framboidal texture in less reduced environments where oxidised S compounds or molecular oxygen are present at low concentrations. The process we propose herein has applicability to framboidal pyrite both in strictly anoxic environments where S is present as S(-II), as well as less reduced environments, where reduced S is still by far the dominant S species. The reaction of FeS and H2S to form FeS2 and H2 was investigated experimentally by Taylor et al. (1979) at temperatures of between 100 and 160°C, and found that the process generated quantities of hydrogen consistent with the stoichiometry:

Framboidal pyrite formation FeS ⫹ H2S 3 FeS2 ⫹ H2

(6)

Drobner et al. (1990) repeated the process at lower temperatures, but found hydrogen gas levels one hundred times lower than expected. Rickard (1997) and Rickard and Luther (1997) determined the kinetics and mechanism of the process. They found that the oxidation of iron (II) monosulphide by hydrogen sulphide proceeds via an FeS 3 SH2 intermediate which allows the formation of an S-S bond, the breaking of H-S bonds and the conversion of Fe(II) from high to low spin. It is likely that the FeS2 and H2 products interact, leading to occlusion of H2 on the pyrite surface. This feature might explain the nonstoichiometric quantities of H2 analysed by Drobner et al. (1990) and Rickard (1997), and this argument was used by Rickard (1997) to explain the variable amounts of H2 measured from the reaction. This interpretation remains controversial, however Guevremont et al. (1998) observe that the reaction of surface hydrogen to form gaseous hydrogen on the FeS2 surface is not a facile process, and expect that the fate of hydrogen is to incorporate into the pyrite lattice during growth. The experimentally determined lowest unoccupied molecular orbital (LUMO) for H2S is ⫺1.1 eV (Radzig and Smirnov, 1985), making it an excellent electron acceptor. Indeed the LUMO for H2S compares favourably with the LUMO for O2, which is ⫺0.47 eV. By contrast, HS⫺ cannot be an electron acceptor for pyrite formation because the calculated LUMO is ⫹8.015 eV (Rickard and Luther 1997), and so pyrite forming reactions utilising HS⫺ as a sulphur source require an additional electron acceptor. Thus, since pK1 for H2S at 298 K is 6.98 (e.g., Suleimenov and Seward, 1997), the reaction is favoured at acid to weakly alkaline pH, and is expected to proceed under conditions typical of sediments and hydrothermal fluids. Pyrite formation occurs via this pathway without the requirement for oxidised, reactive, sulphur species (cf. Schoonen and Barnes, 1991b; Wilkin and Barnes, 1996) which are present in nature at low concentrations. Thus, pyrite formation in strictly anoxic environments (e.g., Boesen and Postma, 1998) and framboidal pyrite formation in many hydrothermal systems (e.g., Reedman et al., 1985) can be accounted for. The mechanism of reaction proposed by Rickard and Luther (1997) is: FeS(mk) 3 FeS(aq) (⬍100°C fast. ⬎100°C slow)

(7)

FeS(aq) ⫹ H2S(aq) 3 {FeS3 SH2} (fast)

(8)

{FeS3 SH2} 3 [FeS2.H2(occluded)] (fast)

(9)

[FeS2.H2(occluded)] 3 FeS2(py) ⫹ H2(g) (⬍100°C slow. ⬎100°C fast) (10)

where FeS(aq) is an electroactive, dissolved species and represents cluster complexes of quantum-sized particles of FeS. This mechanism is important to framboid formation since it involves a dissolved phase. That is, the pyrite microcrysts do not retain any textural memory of the precursor mackinawite, as might be the case through a solid state reaction. Indeed, Rickard and Luther (1997) were able to demonstrate clearly that the intervention of a dissolved phase enables transport of the reaction components away from the mackinawite source. This is further confirmed in this experimentation, where pyrite microcrysts grow at distances of at least 2–5 ␮m (i.e., one framboid diameter) from the reactant mackinawite (e.g., Fig. 2d,e).

2669

Voltammetric studies of lake water columns (Buffle et al., 1988; Davison et al., 1988), marine waters and hydrothermal fluids (Theberge and Luther, 1997), and estuarine sediment porewaters (Rickard et al., 1999) provide direct evidence for the presence of electroactive FeS°(aq) complexes in natural systems. Buffle et al. (1988) found that FeS°(aq) forms at IAP ⬎ 0.25 Ksp FeSam, and Theberge and Luther (1997) found that KfeS°(aq) ⫽ Ksp FeSam. Thus, FeS°(aq) may be present in pyrite-forming systems where FeSam is undersaturated or only just saturated. A study of sediment cores from the Loughor Estuary, S.Wales, by Rickard et al. (1999) suggests that the dissolution of FeSam provides the FeS°(aq) reactant for sedimentary pyrite formation. The thiospinel, greigite, has been reported by a number of workers from sedimentary environments and from anoxic water columns (e.g., Murumoto et al., 1991). Usually, greigite has been reported from near-shore or shoreline systems such as saltmarshes (e.g., Cutter and Velinsky, 1988), and from freshwater sediments (e.g., Hilton et al., 1986; Hilton, 1990) where the environment shows a seasonal oxic/anoxic cyclicity. In these environments, results might suggest that greigite may be an intermediate in the formation of pyrite (e.g., Cutter and Velinsky, 1988). We note that if greigite dissolves, its role in pyrite formation is indistinguishable from that of any other reactant iron source in sediments. However, Morse and Cornwell (1987) examined the nature and distribution of solid iron sulphides at 14 sites typifying anoxic marine environments, and their data emphasise the possible importance of a soluble FeS°(aq) phase in pyrite formation. They found that identifiable iron sulphide phases in marine sediments are almost exclusively pyrite. In the Morse and Cornwell (1987) study, mackinawite was never present as an observable mineral phase, and greigite was restricted to a single occurrence within vascular plant tissue, and was never observed within magnetic separates. Similarly, magnetic separates recovered from black anoxic sediments of the Humber Estuary, UK, contain no observable greigite, and sulphur is below EDAXTM detection limits (S. Lee, pers. comm., 1998). Despite the lack of observable iron monosulphide phases, Morse and Cornwell (1987) found that acid volatile sulphides (AVS) were present in the sediments studied at concentrations of up to 100 ␮mol 䡠 g⫺1. They concluded that AVS in sediments is present exclusively as extremely fine-grained coatings on mineral grain surfaces. In natural sediments, as in our laboratory experiments, it seems likely that it is the solution transport of iron (II) monosulphide as FeS°(aq) that is essential for the development and growth of observable pyrite textures from dispersed, fine grained monosulphide reactants. The oxidation of aqueous iron (II) monosulphide by H2S is by far the most rapid of the pyrite forming reactions hitherto identified (Rickard, 1997). The rate of pyrite formation by the dissolution of solid amorphous FeS and reaction with H2S between 25 and 125°C is described by the equation (Rickard, 1997): dFeS2/dt ⫽ k(FeS) (cH2S(aq))

(11)

and the second order rate constant, k, is 1.03 ⫻ 10⫺4 L mol⫺1 s⫺1 at 25°C. The reaction is considerably faster than the polysulphide pathway (Rickard, 1975; Luther, 1991) for which the

2670

I. B. Butler and D. Rickard

second order rate constant is typically of the order of 10⫺20 L mol⫺1 s⫺1 at 25°C. Kinetic data have not been determined for other pyrite forming reactions, but those proceeding via solid phase reactions are expected to be slower than the polysulphide pathway at ambient temperatures. From the rate data, Rickard (1997) calculated that pyrite formation, within the typical range of sedimentary FeS and H2S concentrations, falls within the range 3 ⫻10⫺10 to 3 ⫻ 10⫺5 FeS2 g⫺1 of sediment per annum at pH ⫽ 8 and between 1.5 ⫻ 10⫺9 and 1.5 ⫻ 10⫺4 mol FeS2 g⫺1 of sediment per annum at pH ⫽ 7. In most anoxic marine environments where there is a ready supply of organic matter and reactive iron minerals, rates will tend towards the upper limit suggested by Rickard (1997). In some cases, such as Framvaren Fjord in S. Norway, H2S concentrations exceed 7– 8 mM at pH ⫽ 7 (Skei, 1988), and rates of pyrite formation may be an order of magnitude greater than estimated by Rickard (1997). Wilkin et al. (1996a) found total pyrite concentrations of between 40 and 280 ␮mol 䡠 g⫺1 in the top 1–5 centimetres of modern anoxic sediment columns, and they noted that framboidal pyrite constituted between 60 and 95% of the pyrite present. These concentrations could be formed within a timeframe of 1–10 y at pH ⫽ 8, based on experimental rate data. These rates fit well with typical sedimentation rates for the environments considered, and are consistent with framboid formation within the top few centimetres of the anoxic sediment column. The reaction mechanism and rate data are also consistent with observations of pyrite framboid formation within anoxic water columns (e.g., Skei, 1988; Wilkin and Barnes, 1997b). Rates of pyrite formation via other processes are expected to be subordinate due to the generally low concentration of comparatively oxidised reactants such as greigite, elemental sulphur or polysulphide in natural environments. Our experimental data suggest a redox dependence of framboid formation in the experimental systems. The redox state is the initial Eh of the reaction system poised using Ti (III) citrate, and varied between ⫺250 and ⫺400 mV. This equates to an equilibrium fO2 of between 10⫺73 and 10⫺86 atm at 298 K, indicating that free molecular oxygen is not involved in the reaction. The predominance of single crystals at low Eh and of framboids at high Eh is likely a function of the relative rates of crystal growth and nucleation. The oxidation of FeS(mk) by H2S to form pyrite is energetically favourable (⌬G°r ⫽ ⫺34.14 kJ 䡠 mol⫺1 at 25°C). However, Schoonen and Barnes (1991a) showed that pyrite formation was inhibited by a nucleation barrier and suggested that active FeS(mk) surfaces could act to overcome this barrier. The role of nucleation inhibition in pyrite formation probably explains the contrasting experimental results published for pyrite formation. However, it is expected that pyrite itself acts as a site for further pyrite nucleation. Thus, the formation of a first pyrite nucleus may result in a cascade effect as further pyrite nuclei form on the pre-formed pyrite. This process has been suggested as the reason for the rapid formation of clusters of pyrite microcrysts in mutual contact which characterise the framboid texture (Butler 1994). Framboids, with limited microcryst size and multiple (up to 105) pyrite nuclei are evidence for rapid nucleation and limited crystal growth, which implies that reactants are transported rapidly to the reaction site with respect to the reaction rate. The role of FeS°(aq) may be especially crucial in this regard. Single crystals, growing to larger sizes than individual framboid mi-

Fig. 3. Eh-pH stability field diagram for the system Fe-S-H20 at 25°C, generated using HSC Chemistry for WindowsTM v4.0. Once formed, FeS2 is expected to be a persistent phase within any part of Eh-pH space, however, it is also true that FeS2 can only form within its own stability field. Increasing total Fe and S concentrations from micromolal to millimolal concentrations and to levels approximating our experimental conditions (10⫺7 m to 10⫺1 m) results in an increase in the extent of the FeS2 stability field (grey). The FeS2 field extends sub-parallel to the pH axis and broadens parallel to the Eh axis. The extension parallel to the Eh axis is asymmetric, with the upper boundary of the field remaining approximately fixed, but the lower boundary extending below the limit of H2O stability. At low total Fe and S, pyrite is undersaturated or slightly supersaturated at very low Eh, but saturation becomes greater as total Fe and S increase. Due to the asymmetric growth of the pyrite stability field, lines of equal supersaturation are compressed together close to the upper boundary of the field. Thus, systems at comparatively oxidised Eh may show greater supersaturation with respect to FeS2 than do systems with the same total Fe and S at highly reducing Eh. Within our experimental systems, reactions poised at ⫺400 mV are considerably less supersaturated with respect to FeS2 than reactions carried out in the absence of a redox poise.

crocrysts represent a lower rate of nucleation, but a greater rate of crystal growth. The implication of the Eh dependence of pyrite framboid formation in our experimentation is that redox state has an influence of pyrite nucleation rates (Fig. 3). A possibility is that at higher Eh, the reaction solution is firmly within the thermodynamic stability field of pyrite, and thus pyrite is able to nucleate rapidly from considerable supersaturation onto suitable surfaces. At lower Eh pyrite is not as heavily supersaturated (see Fig. 3), and nucleation rates are slower. Our experimental design does not permit significant changes in pH in the reaction system, but we predict that framboidal pyrite formation will be particularly favoured under conditions of lower pH and comparatively high Eh. This is consistent with the involvement of H2S (as against HS⫺) as a reactant. The apparent redox dependence for framboid formation is again consistent with observations of framboid formation at or close to the redoxcline in sediments. We propose, however, that this is a function of relative nucleation and crystal growth rates in that environment rather than a supply of suitable electron acceptors other than H2S in these environments. The concept that the generation of framboidal versus euhedral forms is dependant upon relative nucleation versus growth rates is not novel in itself and has been previously proposed by Sweeney and Kaplan (1973) and Wilkin and Barnes (1997a). However, in this published work, the relative

Framboidal pyrite formation

rates of nucleation and growth are dependant upon a reaction which forms pyrite using a variety of electron acceptors other than H2S or oxidised S phases (e.g., O2, SnS2⫺, S°). The model proposed here suggests that the pH and redox potential of the local environment control saturation and thus the relative nucleation versus growth rate for pyrite formed by the FeS ⫹ H2S reaction. In this experimentation, the form of the framboidal microcrysts is almost exclusively cubic. Octahedral forms are commonly observed in single crystals (Fig. 2b), but only rarely in framboids. Murowchick and Barnes (1987) related the cubic and octahedral crystal forms of pyrite to the degree of supersaturation of the solution with respect to pyrite. In their scheme, cubic faces developed at lower supersaturations. In single crystal products, pyrite faces terminating octahedra are comparatively abundant, and this is consistent with the idea of cube faces representing the final stages of supersaturation in solution (IAP ⬎ Ksp). The scarcity of octahedral forms in the framboidal reaction products suggests that throughout the reaction the IAP was never much greater than the Ksp(pyrite). This in turn is consistent with enhanced pyrite nucleation rates during framboid formation. In detail, the cubic pyrite microcrystals of which the framboids are composed, are poorly formed. The faces show irregular or circular features about 30nm in diameter (Fig. 2d–f). Such structures are the product of screw dislocation crystal growth which is characteristic of fast growth. Similar features have been observed in natural framboidal pyrite from fossil wood of the Eocene London Clay and from Kuroko style mineralization at Nukundamu, Vanua Atu, Fiji (Butler, 1994).

5. CONCLUSIONS

Framboidal pyrite can be synthesised via the reaction of amorphous iron (II) monosulphide and aqueous hydrogen sulphide. Crucially, the experimental synthesis of framboids proceeds without the involvement of a greigite intermediate. We conclude that, contrary to previous conceptual models, greigite is not an essential prerequisite for framboid formation, and consequently framboid formation probably does not always proceed via the magnetic aggregation and subsequent transformation of greigite microcrystals (Wilkin and Barnes, 1997a). The reaction is apparently dependant upon Eh, with framboidal pyrite formed at an Eh of ⬎⫺250 mV at pH 6. We suggest that this may be a supersaturation effect related to position within the pyrite stability field. The appearance of framboids close to the initiation of bacterial sulphate reduction in anoxic sediments and water columns may be unrelated to the presence of electron acceptors other than H2S as has been previously supposed. The size and form of the pyrite products of our reactions (both euhedral and framboidal) point to a nucleation vs crystal growth rate control over the nature of the pyrite texture. Framboidal pyrite appears to formed under conditions favourable for rapid nucleation, but growth is possibly substrate-limited. Single crystals indicate slow nucleation and normal crystal growth without substrate limitation. This demonstration that framboidal pyrite forms directly via the oxidation of FeS(aq) by H2S without a greigite intermediate

2671

has considerable implications for physical models of framboid formation. Acknowledgments—This work was supported by NERC grants GR9/ 603 and GR3/7476 to DR and a NERC framework studentship award GT4/90/GS/127 to IB. Many thanks to Anthony Oldroyd for experimental expertise and to Peter Fisher for his assistance with the SEM. The authors would like to thank J. Murowchick and two anonymous reviewers for their constructive criticism of the original manuscript. REFERENCES Berner R. A. (1967) Thermodynamic stability of sedimentary iron sulphides. Am. J. Sci. 265, 773–785. Berner R. A. (1969) The synthesis of framboidal pyrite. Econ. Geol. 64, 383–384. Boesen C. and Postma D. (1988) Pyrite formation in anoxic sediments of the Baltic. Am. J. Sci. 288, 275– 603. Buffle J., De Vitre R., Perret D., and Leppard G. G. (1988) Combining field measurements for speciation in non-perturbable water samples. In Metal Speciation: Theory, Analysis and Application (eds. J. R. Kramer and H. E. Allen), pp 99 –124. Lewis Publishers Inc. Benning L. G., Wilkin R. T., and Barnes H. L. (in press) Reaction pathways in the Fe-S system below 100°C. Chem. Geol. Butler I. B. (1994) Framboid formation. PhD thesis, Univ. Wales. Butler I. B., Schoonen M. A. A., and Rickard D. (1994) Removal of dissolved oxygen from water: A comparison of four common techniques. Talanta 41, 211–215. Butler I. B., Grimes S. T., and Rickard D., (1999) Iron sulphide oxidation in an anoxic chemostated reaction system. Proc. 9th V.M. Goldschmidt Conference LPI Houston. p. 45. Cutter G. A. and Velinsky D. J. (1988) Temporal variations of sedimentary sulphur in a Delaware salt marsh. Mar. Chem. 23, 311–327. Davison W., Buffle J., and De Vitre R. (1988) Direct polarographic determination of O2, Fe(II), Mn(II), S(-II) and related species in anoxic waters. Pure Appl. Chem. 60, 1535–1548. Drobner E., Hubner H., Wa¨chterhauser G., Rose D., and Setter K. O. (1990) Pyrite formation linked with hydrogen evolution under anaerobic conditions. Nature 346, 742–744. Furukawa Y. and Barnes H. L. (1995) Reactions forming pyrite from precipitated amorphous ferrous sulphide. In Geochemical Transformations of Sedimentary Sulphur (eds. M. A. Vairavamurthy and M. A. A. Schoonen), pp 194 –205, Am. Chem. Soc. Guevremont J. M., Strongin D. R., and Schoonen M. A. A. (1998) Thermal chemistry of H2S and H2O on the (100) plane of pyrite: Unique reactivity of defect sites. Am. Mineral. 83, 1246 –1255. Hilton J. (1990) Greigite and the magnetic properties of sediments. Limnol. Oceanogr. 35, 509 –520. Hilton J., Lishman J. P., and Chapman J. S. (1986) Magnetic and chemical characterisation of a diagenetic magnetic mineral formed in sediments of productive lakes. Chem. Geol. 56, 325–333. Kalliokoski J. and Cathles L. (1969) Morphology, mode of formation and diagenetic changes in framboids. Bull. Geol. Soc. Finland 41, 125–133. Kribek B. (1975) The origin of framboids as a surface effect on sulphur grains. Mineralium Deposita 10, 389 –396. Love L. G. (1957) Micro-organisms and the presence of syngenetic pyrite. Q. J. Geol. Soc. Lond. 113, 429 – 440. Love L. G. and Amstutz G. C. (1966) Review of microscopic pyrite from the Devonian Chatanooga shale and Rammelsberg Banderz. Fortschr. Miner. 43, 273–309. Morrissey C. J. (1972) A quasi-framboidal form of syn-sedimentary pyrite. Inst. Min. Met. Trans. 81, B55–B56. Morse J. W. and Cornwell J. C. (1987) Analysis and distribution of iron sulphide minerals in recent anoxic marine sediments. Mar. Chem. 22, 55– 69. Murowchick J. B. and Barnes H. L. (1987) Effects of temperature and degree of supersaturation on pyrite morphology. Am. Min. 72, 1241– 1250. Murumoto J. A., Honjo S., Fry B., Hay B. J., Howarth R. W., and Cisne J. L. (1991) Sulphur, iron and organic carbon fluxes in the Black Sea:

2672

I. B. Butler and D. Rickard

Sulphur isotopic evidence for the origin of sulphur fluxes. Deep Sea Res. 38, S1151–S1187. Papunen H. (1966) Framboidal textures of the pyritic layer found in a peat bog in S.E. Finland. Bull. Comm. Geol. Finlande 222, 117–125. Pearson R. G. (1976) Symmetry Rules for Chemical Reactions. Wiley. Radzig A. A. and Smirnov B. M. (1985) Reference data on atoms, molecules and ions. Vol. 31, pp 375– 438. Springer-Verlag. Raiswell R., Whaler K., Dean S., Coleman M. L., and Briggs D. E. G. (1993) A simple three-dimensional model of diffusion with precipitation applied to localised pyrite formation in framboids, fossils and detrital iron minerals. Mar. Geol. 113, 89 –100. Reedman A. J., Colman T. B., Campbell D. D. G., and Howells M. F. (1985) Volcanogenic mineralisation related to the Snowdon Group (Ordovician), Gwynedd, North Wales. J. Geol. Soc. Lond. 142, 875– 888. Rickard D. (1970) The origin of framboids. Lithos 3, 269 –293. Rickard D. (1997) Kinetics of pyrite formation by the H2S oxidation of iron (II) monosulphide in aqueous solutions between 25°C and 125°C: The rate equation. Geochim. Cosmochim. Acta 61, 115–134. Rickard D. and Luther, III, G. W. (1997) Kinetics of pyrite formation by the H2S oxidation of iron (II) monosulphide in aqueous solutions between 25°C and 125°C: The mechanism. Geochim. Cosmochim. Acta 61, 135–147. Rickard D., Oldroyd A., and Cramp A. (1999) Voltammetric evidence for soluble FeS complexes in anoxic estuarine muds. Estuaries 22, 693–701. Rickard D. (1999) A pyrite grand unified theory. Proc. 9th V.M. Goldschmidt Conference LPI Houston. p246. Rust G. W. (1935) Colloidal primary copper ores at Cornwall Mines, Missouri. J. Geol. 43, 398 – 426. Schneiderhohn H. (1923) Chalkographische untersuchung des Mansfelder Kupferscheifers. N. Jb. Miner. Geol. Pala¨ont. 47, 1–38. Schoonen M. A. A. and Barnes H. L. (1991a) Reactions forming pyrite

and marcasite from solution: I. Nucleation of FeS2 below 100°C. Geochim. Cosmochim. Acta 55, 1495–1504. Schoonen M. A. A. and Barnes H. L. (1991b) Reactions forming pyrite and marcasite from solution. II. via FeS precursors below 100°C. Geochim. Cosmochim. Acta 60, 115–134. Skei J. M. (1988) Formation of framboidal pyrite in the waters of a permanently anoxic fjord, Framvaren, S. Norway. Marine Chemistry 23, 345–352. Suleimenov O. M. and Seward T. M. (1997) A spectrophotometric study of hydrogen sulphide ionisation in aqueous solutions to 350°C. Geochim. Cosmochim. Acta 61, 5187–5198. Sweeney R. E. and Kaplan I. R. (1973) Pyrite framboid formation: Laboratory synthesis and marine sediments. Econ. Geol. 68, 618 – 634. Taylor P., Rummery T. E., and Owen D. G. (1979) Reactions of iron monosulphide solids with aqueous hydrogen sulphide up to 160°C. J. Inorg. Nucl. Chem. 41, 1683–1687. Theberge S. and Luther, III, G. W. (1997) Determination of the electrochemical properties of a soluble aqueous FeS species present in sulphidic solution. Aqua. Geochem. 3, 191–211. Wilkin R. T., Barnes H. L., and Brantley S. L. (1996) The size distribution of framboidal pyrite in modern sediments: An indicator of redox conditions. Geochim. Cosmochim. Acta 60, 3897–3912. Wilkin R. T. and Barnes H. L. (1996) Pyrite formation by reactions of iron monosulphides with dissolved inorganic and organic sulphur species. Geochim. Cosmochim. Acta 60, 41267– 4179. Wilkin R. T. and Barnes H. L. (1997a). Formation processes of framboidal pyrite. Geochim. Cosmochim. Acta 61, 323–339. Wilkin R. T. and Barnes H. L. (1997b). Pyrite formation in an anoxic estuarine basin. Am. J. Sci. 297, 620 – 650. Zehnder A. J. B. and Wuhrmann K. (1976) Titanium (III) citrate as a non-toxic oxidation-reduction system for the culture of obligate anaerobes. Science 192, 1165–1166