The effect of As, Co, and Ni impurities on pyrite oxidation kinetics: An electrochemical study of synthetic pyrite

https://doi.org/10.1016/j.gca.2007.03.005Get rights and content

Abstract

Synthetic pyrite crystals doped with As, Co, or Ni, undoped pyrite, and natural arsenian pyrite from Leadville, Colorado were investigated with electrochemical techniques and solid-state measurements of semiconducting properties to determine the effect of impurity content on pyrite’s oxidation behavior. Potential step experiments, cyclic voltammetry, and AC voltammetry were performed in a standard three-electrode electrochemical cell setup. A pH 1.78 sulfuric acid solution containing 1 mM ferric iron, open to atmospheric oxygen, was chosen to approximate water affected by acid drainage. Van der Pauw/Hall effect measurements determined resistivity, carrier concentration and carrier mobility.

The anodic dissolution of pyrite and the reduction of ferric iron half-reactions are taken as proxies for natural pyrite oxidation. Pyrite containing no impurities is least reactive. Pyrite with As is more reactive than pyrite with either Ni or Co despite lower dopant concentration. As, Co, and Ni impurities introduce bulk defect states at different energy levels within the band gap. Higher reactivity of impure pyrite suggests that introduced defect levels lead to higher density of occupied surface states at the solid–solution interface and increased metallic behavior. The current density generated from potential step experiments increased with increasing As concentration. The higher reactivity of As-doped pyrite may be related to p-type conductivity and corrosion by holes. The results of this study suggest that considering the impurity content of pyrite in mining waste may lead to more accurate risk assessment of acid producing potential.

Introduction

Pyrite oxidation has been extensively studied for its role in the production of acid mine drainage, and as an important factor in ore enrichment processes. Pyrite is a semiconductor with a band gap of 0.95 eV; consequently, its oxidation properties are also of interest for their effect on its utility for solar energy conversion and electrochemical storage devices. Natural pyrite commonly contains As, Co, and Ni impurities in some combination (Palache et al., 1944, Fleischer, 1955, Abraitis et al., 2004) but there has not been a systematic study of the effect of these minor elements on its oxidation rate, because naturally occurring pyrite crystals are often heterogeneous in concentration and distribution of minor elements (Craig et al., 1998). The present study assesses electrochemical response of synthetic pyrite doped with As, Co, and Ni and natural pyrite with As impurity to determine the influence of these minor elements without the complications arising from heterogenity and the presence of more than one minor element. The advantage of electrochemical studies, compared with batch or flow-through reaction studies on crushed pyrite, is that the measured reactivity of individual, intact pyrite thick sections can be related directly to their measured composition and solid-state properties.

The principal oxidants of pyrite responsible for the abiotic production of acid mine drainage are dissolved oxygen and ferric iron (Reactions (1), (2)), with Reaction (2) dominating below pH 2 (Rimstidt and Vaughan, 2003):FeS2+7/2O2+H2OFe2++2SO42-+2H+FeS2+14Fe3++8H2O15Fe2++2SO42-+16H+Pyrite oxidation can also be modeled as three electrochemical half-reactions: the anodic dissolution of pyrite, cathodic reduction of ferric iron ions, and the cathodic reduction of oxygen given by Eqs. (4), (5), (6), respectively (Holmes and Crundwell, 2000).FeS2+8H2OFe2++2SO42-+16H++14e-Fe3++e-Fe2+O2+4H++4e-2H2OThe oxidation rate law based on these individual half-reactions (Holmes and Crundwell, 2000) agreed well with rate laws for oxidation of pyrite in solution with ferric iron, dissolved oxygen, or both, formulated by Williamson and Rimstidt (1994) on the basis of experimental results and data from the literature. Holmes and Crundwell (2000) concluded that pyrite oxidation is an electrochemical reaction and that semiconducting properties of individual samples could have an influence on oxidation rates.

Semiconducting electrical properties such as resistivity, charge carrier concentration, and mobility are related to the type of impurity in pyrite (Bither et al., 1968, Chandler and Bene, 1973, Li et al., 1974, Zhao et al., 1993, Eyert et al., 1998, Lehner et al., 2006). We have studied the relationships among these electrical properties, impurity concentrations, and stoichiometry of synthetic pyrite doped with As, Co, and Ni and with negligible impurity (Lehner et al., 2006). From Hall effect measurements we found that resistivity, carrier concentration, and mobility vary predictably for pyrite with each type of impurity. For example, Co-doped pyrite has very low resistivity resulting from high carrier concentration and mobility. Electrons (n-type) or holes (p-type) may be charge carriers. All our synthetic pyrites are n-type except for As-doped pyrite which tends to be p-type.

The volume of literature on pyrite and its oxidation behavior is vast and spans well over a century of work. Pyrite has long been recognized as a semiconductor. The electromagnetic properties of sulfide ores were studied as early as the 19th century by Fox, 1830, Skey, 1871 who measured the rest potentials of pyrite and other sulfides. Sveshnikov and Dobychin, 1956, Sato, 1960 discovered that potential gradients between sulfide phases can lead to electrochemical dissolution of the phase with lower potential. As the field of electrochemistry has matured there have been many electrochemical studies of pyrite oxidation. Some have focused on the correlation of pyrite semiconducting properties with oxidation behavior. Springer (1970) reported no difference in anodic dissolution between n- and p-type pyrite and concluded that the semiconducting properties of natural pyrite have no effect on oxidation behavior. Biegler and Swift (1979) also found no systematic difference between n- and p-type natural pyrite in anodic dissolution experiments in acid solutions.

A number of other studies, however, do suggest that variable electrical properties arising from impurities and defects in pyrite may affect its oxidation rate. Crundwell (1988) asserts that the solid-state electrical properties of sulfide minerals must be taken into account in order to understand the anodic dissolution behavior. Becker et al., 2001, Rosso and Becker, 2003 reported the “proximity effect” whereby electrons from one crystal site are transferred to remote sites where a reaction is taking place. This may imply that variations in resistivity such as we have observed (Lehner et al., 2006) could play a role in charge transfer kinetics. Dependence of oxidation rates on photo-excitation indicates that charge carrier type and concentration may be significant. For example, Schoonen et al. (2000) found a strong temperature dependence for pyrite oxidation by dissolved oxygen, and an accelerated rate when pyrite is exposed to visible light. Jaegermann and Tributsch (1983) attributed pyrite’s photocurrent to photo-generated holes.

There is conflicting evidence on the role of holes in facilitating pyrite oxidation. Crundwell (Crundwell, 1988) suggested that pyrite resists dissolution by holes due to the non-bonding nature of the valence band orbitals. However, Wei and Osseo-Asare (1996) studied electrochemical dissolution of synthetic particulate n-type pyrite and concluded from the photo-response that anodic dissolution occurs by transfer of holes from the valence band to the surface, while cathodic dissolution involves transfer of electrons from the conduction band. They report in a subsequent study (Wei and Osseo-Asare, 1997) that the mechanism of anodic dissolution in acidic media involves both chemical and electrochemical processes requiring the presence of water. By comparing synthetic n- and p-type pyrite, our experiments provide new evidence for the possible role of holes.

The relationship between pyrite’s band energy structure and intrinsic surface states, which are energy levels available to electrons within the band gap at the semiconductor surface, may also influence oxidation behavior. Bronold et al. (1994) conclude that Fe surface states are located within the band gap due to loss of degeneracy from the surface coordination. They report that if the concentration of surface states is high enough it can lead to the Fermi level being pinned near the valence band edge. Fermi level pinning describes a situation in which the average energy level of electrons at the surface is controlled by surface states independent of applied potential (Bard et al., 1980). Under these conditions a pyrite electrode behaves as a metal. Ennaoui et al. (1986) report Fermi level pinning caused by surface states, and indicate that the oxidation reaction is mediated by these intra-band gap surface states. Fan and Bard (1991) also postulate a localized surface state within the band gap based upon results from scanning tunneling microscopy and tunneling spectroscopy on pyrite surfaces. Density functional theory calculations by Hung et al., 2002a, Hung et al., 2002b indicate that the band gap is reduced and in some cases eliminated at intrinsic surface states on pyrite. Due to Fe coordination numbers of 4 and 5, compared with 6 in the bulk crystal, the Fe t2g electrons at the top of the valence band lose their degeneracy and assume energy levels above one another, approaching the conduction band. At some sites (for example, on {1 1 1} planes) an electron from the highest energy electron pair will be able to occupy an antibonding orbital in the conduction band; the loss of spin parity results in paramagnetic regions. These regions are more attractive to molecular oxygen when exposed to air or water, and so may explain why pyrite oxidation has been observed to proceed from step and kink sites.

Several studies, noting that the ideal pyrite {1 0 0} surface would contain dangling Fe2+ bonds as surface states, suggest that pyrite oxidation is initiated as Fe2+ is oxidized to Fe3+ and instantly reduced by electrons from the lattice, ultimately resulting in the oxidation of S22- anions (Rosso et al., 1999, Rimstidt and Vaughan, 2003). However, the surface of pyrite never breaks ideally along the {1 0 0} plane (Nesbitt et al., 1998). In conchoidal fractured pyrite where the surface may contain terraces, kinks, and perhaps fracture planes such as the {1 1 0}, {2 1 0}, and {1 1 1}, about 40% of the S–S bonds are broken. The surface S oxidizes the dangling Fe2+ resulting in about 40% of the Fe surface states being Fe3+ (Nesbitt et al., 1998). Schaufuss et al. (1998) used synchrotron sourced X-ray photoelectron spectroscopy to study the surface sulfur states and sulfur oxidation products of freshly fractured natural pyrite in air. They report sulfur surface species in order of decreasing reactivity: S2−, S22- with surface coordination, and S22- with bulk coordination. The S2− species form immediately from S at fractured S–S bonds consistent with Nesbitt et al. (1998) reduction of S by Fe2+. Schaufuss et al. (1998) proposed a mechanism whereby oxidation proceeds from the resulting Fe3+ dangling bonds. From these studies it appears that intrinsic surface states are a major factor in the initiation and mechanism of pyrite oxidation, but the relation between impurity atoms and surface states has not been explored.

In the present study, by comparing single-dopant and undoped synthetic pyrite that display different solid-state electrical properties, we address the relation between electrical properties and oxidation behavior without the complexity introduced by heterogeneous natural pyrite samples containing multiple minor elements. We employed cyclic voltammetry (CV), steady state voltammetry, and alternating current (AC) voltammetry to compare the electrochemically driven rates of pyrite oxidation for synthetic pyrite with different impurity compositions. We also investigated natural pyrite from Leadville, Colorado with As and Pb as the main impurities. We chose a system for the cyclic and AC voltammetry experiments below pH 2 that would approximate water affected by acid drainage. No attempt was made to eliminate oxygen.

Previous studies suggest that Fe3+ is the dominant pyrite oxidant and that Reaction (2) is the dominant redox reaction in this range (Rimstidt and Vaughan, 2003). We therefore focused on the current generated by cyclic and AC voltammetry corresponding to the reduction of ferric iron (Reaction (4)) but also on the forced anodic dissolution (Reaction (3)). The release of pyrite-derived ions into solution (Reaction (3)) allows electrochemically driven reaction rates to be assessed by comparing the magnitude of the current generated by different pyrite electrode materials at the applied potentials where Reactions (3), (4) occur. Current density, which is related to the stoichiometry of the reaction (the number of electrons transferred), is measured in A cm−2 or C s−1 cm−2. Coulombs (C) are related to moles reacted through Faraday’s constant. This study was designed to assess the behavior of differently doped pyrite crystals under identical electrochemical and solution conditions, rather than to develop explicit rate laws.

To investigate Reaction (4), we use AC voltammetry, a technique that is very sensitive to reaction rates due to the small amplitude and sinusoidal frequency of the applied potential. The amplitude of the AC current near the equilibrium potential of the ferric/ferrous iron couple with which the pyrite electrode responds to alternating voltage is directly proportional to the standard heterogeneous rate constant and frequency (Breyer and Bauer, 1963, Smith, 1966, Bard and Faulkner, 2001). The standard heterogeneous rate constant corresponds to the exchange current which is the current of the forward and backward redox reactions when they are equal, i.e., at equilibrium. It is a measure of the kinetic facility of a reaction.

We report here on four proxies for the rate of pyrite oxidation that can be related to the impurity concentrations determined by LA-ICPMS and the bulk electrical properties determined by Hall effect-four-probe method for the same pyrite thick section: (i) the cyclic voltammetry current peak generated from the reduction of ferric iron and (ii) the amplitude of the AC voltammetry peak from the reduction of ferric iron, both corresponding to Reaction (4); (iii) the current generated from the anodic dissolution of pyrite at 1.1 V versus the Ag/AgCl 3 M KCl reference electrode, corresponding to Reaction (3); and (iv) the standard heterogeneous rate constant, ko, calculated from the phase shift between the current and the applied potential vectors from the reduction of ferric iron in response to AC voltammetry. We use a pH 1.78 H2SO4 solution with 1 mM ferric iron added as FeCl3. In addition to these four proxies, evidence of higher oxidation rate for pyrite with As impurity over the other types is presented in the form of the anodic branch of Tafel slopes generated from potential step experiments in pH 2 H2SO4 solutions between 0.65 and 0.1 V versus the Ag/AgCl 3 M KCl reference electrode.

In AC voltammetry a sinusoidal voltage is applied to the electrode during a constant DC scan rate. The frequency of the sinusoidal voltage is much higher than the DC scan rate and the amplitude of voltage is small relative to the applied DC voltage. Both the alternating current and alternating voltage are expressed as phasors. They are related to each other by the impedance, a vector quantity, which has a resistive and a capacitive component. It is expressed as a complex number with the resistive component being the real part and the capacitive reactance the imaginary part:Z(ω)=ZRe-jZIm,where ZRe is resistance, j=-1, and ZIm is the capacitative reactance (1ωC), where ω is angular frequency and C is the capacitance (Bard and Faulkner, 2001).

The phase angle (ϕ) between the applied voltage and the resulting current is equal to the phase angle between the impedance vector Z(ω) and the resistive (ZRe) component. As the AC signal is applied successively through the potential region most favorable to a particular reaction the maximum value for the cotangent of the phase angle occurs near the equilibrium potential for the reaction. Cot (ϕ) at this potential is inversely proportional to the standard heterogeneous rate constant:ko=(2DOβDRα)1/2ω1/2[(cotϕ)max-1]αβ-α+αββ,where DO and DR are the diffusion coefficients for the oxidized and reduced species respectively; α is the transfer coefficient, β = 1  α; and ω is the angular frequency (from Eq. 10.5.26, Bard and Faulkner, 2001). This relationship holds for irreversible and quasi-reversible reactions (Bard and Faulkner, 2001). The phase angle at the potential of maximum cot (ϕ) is independent of the bulk solution concentration of the oxidant, the electrode surface area, the amplitude of the alternating potential, and the frequency of the AC signal. Thus, the ko values calculated from the phase angles are evidence that the experimental results are not due only to surface area irregularities or to varying solution conditions.

To understand the utility of the phase angle (ϕ) it is useful to consider metal electrodes. The DC voltage controls the ratio of oxidized to reduced species near the surface of the electrode. Because the AC signal is small compared to the applied DC potential, this ratio near the electrode surface has the same effect as if it were the bulk ratio in solution. Changing the DC potential is therefore equivalent to conducting multiple experiments using solutions with different redox conditions. For a metal electrode, the total measured capacitance of the system is due to the capacity of the solution/electrode interface known as the Helmholtz capacitance or double layer capacitance, in parallel with the so-called pseudocapacitance of the charged redox species reacting with the surface. The pseudocapacitance arises from the faradaic process (the redox reaction) and is a function of the redox condition, the diffusion characteristics of the redox couple and the frequency of the AC signal. At a given redox condition (fixed DC potential) and frequency, the pseudocapacity should be the same for different metal electrodes; however, the resistance component of the impedance will vary depending on the charge transfer resistance for the different electrode material.

For a completely reversible redox reaction, by definition there is no charge transfer resistance and so ϕ = 45°. In this case both the capacitive reactance and the resistance vector are defined by the diffusion characteristics of the redox couple and they are equal. As reactions become less reversible the charge transfer resistance increases the resistance vector, ϕ decreases, and cot (ϕ) increases. Therefore ko, the rate constant for the equilibrium exchange current, is proportional to ϕ in a semi-reversible to irreversible system. The measured values have only to be corrected for the solution resistance and Helmholtz capacitance (Bard and Faulkner, 1980).

Pyrite, being a semiconductor, has at least two additional sources of capacitance that function in parallel to each other and in series with the Helmholtz capacitance (Fig. 1, dashed region). One is the space charge region which is charged relative to the bulk pyrite due to surface states and reaction with species in solution. The result is a capacity due to either depletion or an excess of charge carriers. The capacitance of the space charge region is much smaller than the Helmholtz capacitance and therefore its capacitive reactance is much larger. The Helmholtz capacitance is considered a minor source of error in semiconductor surface characterization whereas the space charge capacitance is significant and changes as a function of applied potential (Morrison, 1980). The other source of additional capacitance is the surface states themselves: the available energy levels resulting from surface defects, adsorbed species, and impurity atoms. Unlike the space charge capacitance and the Helmholtz capacitance, which are similar to a parallel plate capacitor in that they can reversibly gain or lose charge, the surface state capacitance is dependent on the mobility of electrons between surface states and the bulk or species in solution. If the surface state capacitance is large, as may be the case if there are a large number of defect and impurity states, then the surface state capacitance should dominate the space charge capacitance and the electrode could behave as a metal. To obtain meaningful phase angles, measured values have to be corrected for all sources of non-faradaic capacitance and resistance. Fig. 1 is an equivalent circuit showing some of the possible circuit elements affecting the semiconductor/solution interface. We can infer a relationship between surface states and impurity concentration by measuring the non-faradaic capacitance and resistance and applying the correction.

Section snippets

Samples

The pyrite used in this study is from two sources. Most samples were synthesized in our laboratory where pyrite crystals are grown with chemical vapor transport (CVT) in sealed evacuated quartz tubes in a temperature gradient of 700–550 °C over 16 cm using FeBr3 as a transport agent (Lehner et al., 2006). The synthesized pyrite was doped with either As, Co, Ni or nothing at all. Pyrite from the Black Cloud Mine (Leadville, CO) that was found to contain primarily As and Pb impurities was also

Composition and crystallinity

Dopant concentrations are reported in Table 1. Samples with impurity concentrations less than 50 ppm were classified as undoped (but are labeled in Table 1 with their intended dopant). Fig. 3 shows the results of the synchrotron X-ray diffraction analyses and a scanning electron micrograph of a typical Ni-doped crystal. The crystal morphology and sharp diffraction peaks that match the planar spacings of pyrite are evidence of the material’s identity as pyrite.

AC and cyclic voltammetry

Pyrite with As impurity generated

Discussion

As-, Co-, and Ni-doped pyrite produce more current from the electrochemical reactions than undoped pyrite; As-doped pyrite is the most reactive. Two proposed explanations are discussed below:

  • 1.

    As-doped pyrite has higher relative reactivity because it has p-type conductivity and is susceptible to corrosion by holes. Evidence comes from the response to anodic polarization.

  • 2.

    Bulk defect states arising from impurities in pyrite increase occupied surface state density, resulting in increased reactivity.

Evidence from electrochemical etching of the electrodes

Most pyrite electrodes must be electrochemically etched in order to react quasi-reversibly with ferric iron in solution. Without this etching the electrodes did not respond to the ferric iron in the pH 1.78 H2SO4 solution with 1 mM ferric iron when the potential was scanned from 0.7 to 0.2 V with CV or AC voltammetry. The etching consists of the anodic dissolution from the CV sweep to 1.1 V, after which the ferric iron reduction peak occurs at approximately 0.44 V in the return cathodic scan. If

Conclusions

In sulfuric acid solution at pH 1.78 with added ferric iron in the presence of oxygen, pyrite with impurities generates more current from the electrochemical half-reactions: the reduction of ferric iron (Reaction (4)) and the anodic dissolution of pyrite (Reaction (3)). Pyrite containing As impurity is more reactive than pyrite containing little or no impurities. Pyrite with As is also more reactive than pyrite with either Ni or Co in the same solution. Pyrite containing the lowest impurities

Acknowledgments

This manuscript benefitted from the comments of three anonymous reviewers. We thank Leonard Feldman (Vanderbilt University Physics Department) for use of the Hall effect equipment. We are grateful to John Ayers and Peixin He of CH Instruments, Inc. for helpful discussion, and to Daniela Stefan for her assistance with pyrite synthesis and for providing the Leadville, CO thick sections. We appreciate assistance from Alan Wiseman (Vanderbilt University Dept. of Earth and Environmental Sciences)

References (49)

  • J.D. Rimstidt et al.

    Pyrite oxidation: a state-of-the-art assessment of the reaction mechanism

    Geochimica et Cosmochimica Acta

    (2003)
  • K.M. Rosso et al.

    Proximity effects on semiconducting mineral surfaces II: distance dependence of indirect interactions

    Geochimica et Cosmochimica Acta

    (2003)
  • Y. Tomm et al.

    Growth mechanism and electronic properties of doped pyrite (FeS2) crystals

    Journal of Crystal Growth

    (1995)
  • M.A. Williamson et al.

    The kinetics and electrochemical rate-determining step of aqueous pyrite oxidation

    Geochimica et Cosmochimica Acta

    (1994)
  • A.J. Bard et al.

    Electrochemical Methods: Fundamentals and Applications

    (1980)
  • A.J. Bard et al.

    Electrochemical Methods: Fundamentals and Applications

    (2001)
  • A.J. Bard et al.

    The concept of fermi level pinning at semiconductor/liquid junctions. Consequences for energy conversion efficiency and selection of useful solution redox couples in solar devices

    Journal of the American Chemical Society

    (1980)
  • T.A. Bither et al.

    Transition metal pyrite dichalcogenides, high-pressure synthesis and correlation of properties

    Inorganic Chemistry

    (1968)
  • B. Breyer et al.

    Alternating Current Polarography and Tensammetry

    (1963)
  • K. Buker et al.

    Influence of electrochemical activation and surface orientation on the photoresponse of single crystalline pyrite/electrolyte and pyrite/metal junctions

    Berichte Der Bunsen-Gesellschaft-Physical Chemistry Chemical Physics

    (1994)
  • K. Buker et al.

    Photoelectrochemical investigations of complex formation phenomena on oriented n-pyrite (FeS2) crystal surfaces

    Berichte der Bunsengesellschaft für Physikalische Chemie

    (1996)
  • R.N. Chandler et al.

    EPR study of the solid solutions NixFe1–xS2, CoxFe1–xS2, and CoxNiyFe1–x–yS2

    Physical Review B

    (1973)
  • J.R. Craig et al.

    Pyrite: physical and chemical textures

    Mineralium Deposita

    (1998)
  • S.E. Creager et al.

    A new way of using AC Voltammetry to study redox kinetics in electroactive monolayers

    Analytical Chemistry

    (1998)
  • Cited by (76)

    View all citing articles on Scopus
    View full text