You are on page 1of 10

Applied Geochemistry 41 (2014) 1–10

Contents lists available at ScienceDirect

Applied Geochemistry
journal homepage: www.elsevier.com/locate/apgeochem

Pyrrhotite dissolution in acidic media


Paul Chiritßă a,⇑, J. Donald Rimstidt b
a
University of Craiova, Department of Chemistry, Calea Bucuresßti, 107I, Craiova 200512, Romania
b
Department of Geosciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, United States

a r t i c l e i n f o a b s t r a c t

Article history: Non-oxidative dissolution rates for hexagonal pyrrhotite (Fe1xS) were measured for pH values ranging
Received 2 June 2013 from 0 to 1.3 and temperatures ranging from 25 to 70 °C. These results showed that pyrrhotite dissolution
Accepted 27 November 2013 rate increases with decreasing pH or increasing temperature.
Available online 3 December 2013
The 16 new rate data from this study were combined with 46 data taken from the literature (1 datum
Editorial handling by K.S. Savage
was discarded) to develop a rate equation for non-oxidative pyrrhotite dissolution
65;900 1
rHþ ðmol=m2 sÞ ¼ 1:58  107 e R ð Þ M1:46
T þ
H

This fit spans a range of pH values from 0 to 5 and temperatures from 20 to 90 °C. It is most reliable for
low pH (62.75), where most rate data were measured. This equation reasonably predicts rates for all pyr-
rhotite (Fe1xS) compositions regardless of the value of x, including monoclinic pyrrhotite, hexagonal pyr-
rhotite and troilite (FeS).
In addition, an equation that expresses pyrrhotite rate as a function of temperature and PO2 was devel-
oped using 35 rate data taken from the literature
30;200 1
rO2 ðmol=m2 sÞ ¼ 1:10  102 e R ð Þ P0:352
T
O2

and an equation that expresses the rate of oxidation of pyrrhotite as a function of temperature and Fe(III)
concentration was developed using 48 previously published data
33;600 1
rFeðIIIÞ ðmol=m2 sÞ ¼ 0:516e R ð Þ M0:368
T
FeðIIIÞ

The correlation coefficients for these regression models were relatively low due the narrow range of
experimental conditions, relatively few data, and possible uncontrolled variables. These equations predict
similar rates for all three reactions near pH 2 where many of the experiments were carried out leading to
concerns that some of the experiments may have measured combined rates. For natural conditions, the
rates of the H+ and Fe(III) reactions are predicted to be relatively fast and similar at pH < 3. The O2 reaction
is dominant at higher pH > 3.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction 2002; Karpenko and Norris, 2002). At the present time other
sources for sulfuric acid and copperas are more economic so that
Pyrrhotite (Fe1xS) is the second most abundant sulfide mineral pyrrhotite is considered gangue and is disposed in mine wastes
in the Earth’s crust (Belzile et al., 2004; Cai et al., 2005). It is found and in mill tailings where it typically oxidizes to produce acid mine
disseminated in many igneous and metamorphic rocks and is drainage (AMD). Very fine-grained pyrrhotite can cause problems
highly concentrated in certain ore deposits, such as massive sul- inside mines and mills because pyrrhotite oxidation is strongly
fides like Ducktown, Tennessee, USA, where it is the dominant sul- exothermic. When the oxidation process is very fast, build up of
fide phase (Laurence, 1965; Craig and Vokes, 1993). heat can lead to spontaneous fires. This process also releases large
Pyrrhotite has been a raw material for the production of sulfuric amounts of noxious SO2 into the air.
acid (Lin, 1997; Quinn, 1993) and at times it has been used to man- Exposing pyrrhotite to the atmosphere and water leads to a
ufacture copperas (also known as green vitriol), which is synthetic variety of reactions that eventually convert it to iron oxyhydrox-
melanterite (FeSO47H2O) (Hammarstrom et al., 2005; Johnsson, ides and sulfuric acid. This process is complicated (Pratt and Nes-
bitt, 1997; Mycroft et al., 1995; Belzile et al., 2004), consisting of
⇑ Corresponding author. Tel.: +40 788849387. parallel and consecutive elementary reactions, with multiple reac-
E-mail addresses: paulxchirita@gmail.com (P. Chiritßă), jdr02@vt.edu
tion steps. Full comprehension of the nature of the overall process
(J.D. Rimstidt). requires a detailed knowledge of each rate. Because pyrrhotite

0883-2927/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.apgeochem.2013.11.013
2 P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10

Nomenclature

A preexponential factor in rate equation, mol/m2 s n reaction order with respect to the concentration of H+
Ageo specific surface area calculated from grain diameter, m2/g R gas constant, 8.314 J/mol K
b initial rate (i.e., rate at t = 0), mol/L s q density of mineral, g/cm3
De effective grain diameter, m ri rate for reaction involving species i, mol/m2 s
Dmax maximum grain diameter, m T temperature, K
Dmin minimum grain diameter, m t time, s
Ea activation energy, J/mol
k rate constant, s1
Mi concentration of species i, mol/L

oxidation reactions produce hydrogen ions, non-oxidative dissolu- the oxidation and bio-oxidation of Fe2+ produce large amounts of
tion reactions, which consume these hydrogen ions, are potentially Fe3+. The FeS2 produced by the oxidative leaching of pyrrhotite,
important in the overall pyrrhotite weathering process. Pyrrhotites called ‘‘Zwischenproduckt’’ by Ramdohr (1980), is recognized by
(po) are iron deficient and their non-oxidative dissolution can pro- its ‘‘birds eye’’ texture under the ore microscope. Pratt et al.
duce zero valence sulfur that appears in the reaction products as (1994a) and Mycroft et al. (1995) explain the initial stage of this
disulfide, polysulfide, or elemental sulfur (Jones et al., 1992; Pratt process. It is not clear how much of the Zwischenproduckt ob-
et al., 1994a; Mycroft et al., 1995; Thomas et al., 1998; Mikhlin served in nature is the result of Fe3+ released by non-oxidative dis-
et al., 2000,2003; Chirita et al., 2008; Chirita, 2009; Harries et al., solution and how much is produced by oxidation reactions.
2013). Non-oxidative dissolution reactions (1) and (2) cannot occur
unless there is a source of hydrogen ions and hydrogen ions are
Fe1x SðpoÞ þ 2ð1  xÞHþ ¼ ð1  xÞFe2þ þ ð1  xÞH2 S þ xS0 ð1Þ only produced by oxidation reactions. Pyrrhotite persists for geo-
H2S released by non-oxidative dissolution appears to play an impor- logic time spans in the subsurface as long as it is isolated from oxy-
tant role in the self-heating and spontaneous combustion of pyr- gen. However, when it is exhumed and exposed to air a complex
rothite. Somot and Finch (2010) detected H2S in self-heating series of oxidation reactions occur (Gunsinger et al., 2006). The
sulfide masses by its reaction with copper sulfate (so that black overall process produces goethite and 1 mol of sulfuric acid for
amorphous copper sulfides are formed) or metallic copper (so that every mole of pyrrhotite oxidized.
covellite and some digenite and chalcocite are formed). Fe1x S þ ðð9  3xÞ=4ÞO2 þ ðð3  xÞ=2ÞH2 O
Spectroscopy studies (Pratt et al., 1994a; Mycroft et al., 1995;
Mikhlin and Tomashevic, 2005; Skinner et al., 2004) suggest that ¼ ð1  xÞFeOOH þ 2Hþ þ SO2
4 ð5Þ
the charge deficiency due to the missing Fe2+ in the pyrrhotite Under the acid conditions found in acid mine drainage, this
structure is accommodated by the presence of Fe3+. That means reaction requires the participation of Acidithiobacillus ferrooxidans
that the pyrrhotite chemical formula could be written as and related microbes that oxidize the Fe2+ to Fe3+ because the abi-
FeIIð13xÞ FeIII
ð2xÞ S and the dissolution reaction in acidic solution based otic Fe2+ oxidation rate is too slow to be important (Williamson
on this stoichiometry might release both Fe2+ and Fe3+ into et al., 2006; Singer and Stumm, 1970). When these microbes are
solution. not active because of low water activity, low oxygen concentra-
FeIIð13xÞ FeIII þ 2þ
þ ð2xÞFe3þ tions, or high temperature, pyrrhotite oxidizes to a mixture of fer-
ð2xÞ SðpoÞ þ 2H ¼ H2 S þ ð1  3xÞFe ð2Þ
rous sulfate and sulfuric acid.
This reaction does not explain the accumulation of zero and
intermediate valence sulfur associated with the sulfur rich layer. Fe1x S þ ðð4  xÞ=2ÞO2 þ xH2 O ¼ ð1  xÞFe2þ þ 2xHþ þ SO2
4 ð6Þ
However, a subsequent reaction between Fe3+ and H2S (Asai
This reaction seems to be important under confined conditions with
et al., 1990; Ebrahimi et al., 2003; Gholami et al., 2009) could pro-
limited water and oxygen availability. Under these conditions pyr-
duce zero valence sulfur.
rothite can convert to melanterite (FeSO47H2O) (Jerz and Rimstidt,
xH2 S þ ð2xÞFe3þ ¼ ð2xÞFe2þ þ 2xHþ þ xS0 ð3Þ 2003). Regardless of whether reaction (5) or (6) predominates, the
rate of hydrogen ion diffusion ðDHþ ¼ 9:3  109 m2 =sÞ (Li and Greg-
Eqs. (2) and (3) add to give reaction (1) which could produce ory, 1974) into granular pyrrhotite masses is about four times faster
zero valence sulfur. Alternatively, oxidation of the pyrrhotite sur- than dissolved O2 ðDO2 ¼ 2:2  109 m2 =sÞ (Ferrell and Himmelblau,
face by Fe3+ could produce a sulfur rich layer (SRL) composed from 1967). This means that some combination of reactions (1)–(4)are
disulfide, polysulfide and elemental sulfur (Pratt et al., 1994a; My- potentially important in the interior of the pyrrhotite granular
croft et al., 1995; Chirita and Schlegel, 2012). The SRL can be fur- masses where they are likely to be the first stage of the pyrrhotite
ther covered by an outermost layer containing ferric weathering process. The relative importance of each of these reac-
oxyhydroxides (Pratt et al., 1994b). tions will depend upon its relative rate, which in turn depends upon
This scenario is further complicated by an oxidative leaching solution composition, especially pH, and temperature.
reaction in which Fe3+ reacts with pyrrhotite to form marcasite Pyrrhotite dissolution involves multiple reactions and pro-
(mc) (Fleet, 1978). cesses, which vary in their contribution depending upon the reac-
tion conditions. Deriving rate equations for the key reactions is a
2Fe1x S þ 2ð1  2xÞFe3þ ¼ FeS2 ðmcÞ þ 3ð1  2xÞFe2þ ð4Þ
reasonable first step toward achieving a quantitative understand-
3+ 2
In this reaction the Fe extracts electrons from S in the pyr- ing of pyrrhotite reactivity. This paper focuses on developing a rate
rhotite and the loss of those electrons converts the sulfide anions equation for reaction (1) as a function of pH and temperature.
to S2
2 . Charge balance is maintained by releasing Fe
2+
into the solu- Some rate data are already available in the literature (Bugajinski
tion. This reaction is important under oxidizing conditions where and Gamsjager, 1982; Gleisner, 2005; Janzen et al., 2000; Thomas
P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10 3

et al., 2000, 2001). These were supplemented with additional mea-


surements and these new data were combined with the literature
data to develop a rate equation that describes pyrrhotite non-oxi-
dative dissolution rates as a function of pH and temperature. This
reaction was chosen for study because of its relative simplicity
compared to the oxidation reactions and because the available data
appear to be reasonably reliable. In order to evaluate the impor-
tance of the non-oxidative reaction, rate equations for the oxida-
tion of pyrrhotite by Fe(III) and by O2 were also developed from
published rates. Side by side comparison of these three rate equa-
tions is a first step toward untangling the complex and competing
processes that occur during pyrrhotite weathering.

2. Experimental procedure

2.1. Materials Fig. 1. Schematic representation of the experimental setup.

A sample of natural pyrrhotite, described by Chirita et al.


(2002), was used in this study. Electron microprobe analysis the reactor contained holes for a condenser, a gas (nitrogen) inlet,
(MBT, accelerating voltage 25 kV and beam current 20.13 nA) of a thermometer and a central shaft driven by a motor. The reactor
the sample gave 47.03 at.% Fe and 52.80 at.% S, respectively. The was plugged into a flask containing KOH, which acted as a H2S trap,
composition (Fe0.89S) is consistent with that of hexagonal pyrrho- through a condenser and a safety flask, half filled with oxygen-free
tite. X-ray powder diffraction (XRD) analysis of pyrrhotite sample water. Fig. 1 shows the experimental setup. Before adding the pyr-
with Co radiation generated at 40 kV and 25 mA confirmed the rhotite, the reactor was filled with HCl solution and purged for an
presence of hexagonal pyrrhotite. No Ni, Cu, or Pb was detected hour with high purity nitrogen in order to remove oxygen and CO2.
by microprobe analysis. The Co concentration is 0.04 at.%; the Zn To prevent secondary precipitation reactions produced by H2S(aq),
concentration is 0.01 at.%; and the As concentration is 0.12 at.%. the reaction system was continuously purged with high purity
The pyrrhotite was crushed in an agate mortar and separated nitrogen. No attempt was made to determine the amount of
into 80–150 lm, 50–80 lm, and 25–50 lm size fractions by dry evolved H2S because some of it reacted with the dissolved Sn(II)
screening. The effective grain diameter (De, m) for each fraction (Liteanu, 1971; Alexeyev, 1969) released by Sn(0) dissolution. Tin
was calculated from the relationship given by Tester et al. (1994). residues were always found in the reactor at the end of experi-
ments, indicating that reducing conditions were preserved during
Dmax  Dmin dissolution.
De ¼   ð7Þ
ln DDmax Periodically, a 2 mL solution sample was withdrawn from the
min
reactor, filtered with 0.22 lm pore size membrane, and analyzed
The effective diameter and the density of pyrrhotite (q = 4.88 g/ for total dissolved iron using Atomic Absorption Spectroscopy
cm3 calculated from the molar volume and molecular weight given (AAS). The detection limit was 5 lmol/L. Iron was used as the reac-
by Robie and Hemingway (1995)) was used to calculate the geo- tion progress variable to calculate the dissolution rate.
metric surface area for each size fraction (Kimball et al., 2010).
  2.3. Determination of the initial rate
6  106 m2
Ageo ¼ ð8Þ
De q g Initial dissolution rates were determined by the method de-
Although reaction rates are often expressed in terms of BET sur- scribed in Rimstidt and Newcomb (1993). For each experiment,
face area, geometric surface area is equally valid and is more con- the concentration of iron released at each sample time was fitted
venient to measure and to use to model field conditions as to a second order polynomial.
explained in Rimstidt et al. (2012). Each pyrrhotite fraction was
MFe ¼ a þ bt þ ct 2 ð9Þ
ultrasonically cleaned in ethanol to remove the fine particles
adhering to the larger particles (Descostes et al., 2004). The frac- The derivative of this equation is
tions were stored in an evacuated desiccator until use.
dMFe
Distilled oxygen-free water and analytical grade purity HCl ¼ b þ 2ct ð10Þ
were used for the experiments. Oxygen-free water was prepared dt
by bubbling ultra high purity nitrogen through water for at least The coefficient b is the rate at t = 0, i.e. the initial rate (mol/L s). The
2 h. Metallic tin (99.5%) (Sn(0)) or SnCl22H2O (Sn(II)) was added b coefficients were divided by the surface area of the grains (m2/g)
to most of the experiments in order to consume any traces of oxy- and multiplied by the volume of solution to find the dissolution rate
gen, which were not removed by nitrogen bubbling along with in mol/m2 s.
Fe3+, which might participate in reactions (3) or (6).
2.4. Deriving the rate equation
2.2. Dissolution experiments
In order to develop a rate equation for non-oxidative pyrrhotite
The dissolution experiments were performed in a sealed 0.5 L dissolution it is desirable to use a large number of data. In order to
glass reactor under an atmosphere of ultra high purity nitrogen. obtain these data, we scrutinized the literature to gather 47 addi-
For each experiment metallic tin (0.20–0.25 g) and 0.25 g of pyr- tional dissolution rates from five previously published papers
rhotite were added to 250 mL oxygen-free, acidic solutions (0.05 (Table 1). These data were combined with the 16 rate data
to 1 mol/L HCl), which had been thermally equilibrated to desired obtained in this study to develop a rate equation. The data set
temperature (25–70 °C). Each experiment lasted 4 h. The cover of spanned the pH range from 0 to 5 and temperatures from 20 to
4 P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10

Table 1
Summary of the literature sources for the pyrrhotite dissolution rate data.

Identifier reference Experimental Run material Reactor Rate determination Activation energy, Number of
conditionsa type method kJ/mol data
Bu 82 (Bugajinski and 40–90 °C Monoclinic pyrrhotite BR IRM 59 22
Gamsjager, 1982)
pH 1
Ja00 (Janzen et al., 2000) 25 °C Monoclinic and hexagonal BR IRM – 12
pyrrhotite
pH 2.75
Gl05 (Gleisner, 2005) 25 °C Pyrrhotite from unoxidized CE RDFCE – 3
tailings
pH 4 and 5
Th00 (Thomas et al., 2000) 50 °C Synthetic hexagonal BR RPT – 1
pyrrhotite
pH 1
Th01 (Thomas et al., 2001) 50 °C Polished natural hexagonal BR RPT – 9
pyrrhotite
pH 1

BR = batch reactor; CE = column experiments. IRM = initial rate method; RDFCE = rates derived from column experiments; RPT = rate at particular time.
a
pH and temperature for each experiment can be found in Table S1.

Table 2
Summary of the literature sources for the pyrrhotite oxidative dissolution rate data.

Identifier Experimental Run material Oxidantb Reaction progress Reactor Rate Activation Number
Reference conditionsa variables type determination energy, kJ/mol of data
method
Ja00 (Janzen 25 °C Monoclinic and O2 and Iron and sulfate BR IRM 47.7–62.5c 47
et al., 2000) hexagonal pyrrhotite Fe3+
d
pH 2.50 and 2.75 22.8–63.0
Be04 (Belzile 22–35 °C Museum grade O2 and Iron and sulfate Unreported Unreported – 5
et al., 2004) pyrrhotite Fe3+
pH 2–4
Ca05 (Cai et al., 10–40 °C Monoclinic and O2 and Iron and sulfate BR IRM – 12
2005) hexagonal pyrrhotite Fe3+
pH 2
Ch08 (Chirita 25–45 °C Monoclinic pyrrhotite O2 Iron and H+ BR IRM 41.6 4
et al., 2008)
pH 2.75–3.45
Ro12 (Romano, 4–35 °C Monoclinic pyrrhotite O2 Iron BR IRM 40.26 15
2012)
pH 1.97–3.9

BR = batch reactor; CE = column experiments; IRM = initial rate method.


a
pH and temperature for each experiment can be found in Table S3.
b
Oxidant for each experiment can be found in Table S3.
c
Based on iron release in the presence of dissolved oxygen.
d
Based on iron release in the presence of Fe3+.

90 °C. All rate data are found in Table S1 in the Supplementary reported as 1 standard error. Data for the rate of pyrrhotite oxida-
material. tion by Fe(III) and O2 were also collected from the literature
These data were fit to a generalized rate equation that relates (Table 2) and fit to linearized rate equations with the form of Eq.
the rate to hydrogen ion molarity, which is hereafter expressed (12) but with MHþ replaced either by MFe(III) or by P O2 .
in terms of pH (assuming that the activity coefficient for H+ is near
one), and temperature.
3. Results
n
r ¼ kMHþ ¼ ðAeEa =RT ÞM nHþ ð11Þ
3.1. Experimental results
This equation was transformed to a linear form by taking the
logarithm of both sides. Table 3 shows the experimental conditions and the rates deter-
Ea 1 mined from each pyrrhotite dissolution experiment. The concen-
log r ¼ log A   npH ð12Þ tration of total iron in solution (MFetotal) versus time data for
2:303R T
these experiments are given in the Supplementary material
Log r was fitted as a function of pH and 1/T using the multiple linear (Table S2). Representative plots of MFetotal as a function of time
module in JMP version 9.0.0 statistical software from SAS Institute are shown in Fig. 2.
Inc., Cary, NC. T-tests were used to determine the significance of The effect of two different reducing agents on pyrrhotite disso-
the fitted parameters and residual plots were examined to identify lution was investigated at 40 °C, pH 1, 700 rpm and particles diam-
outliers in the data set. The uncertainty in the fitted parameters is eter in range 50–80 lm. The first experiment (#R02) was
P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10 5

Table 3
Experimental conditions and measured rates. Sn(0) was present in all experiments except R01, which contained no reducing agent, and R02, which was performed with a
0.01 mol/L Sn(II) solution.

Experiment # Grain size, lm Ageo, m2/g Stirring, rpm pH T, °C Rate, mol/m2 s


T01 50–80 0.0205 700 1.00 25 6.23  106
T02 50–80 0.0205 700 1.00 40 8.30  106
T03 50–80 0.0205 700 1.00 50 9.76  106
T04 50–80 0.0205 700 1.00 55 1.21  105
T05 50–80 0.0205 700 1.00 70 1.71  105
F01 80–150 0.0118 700 1.00 40 9.12  106
F02 25–50 0.0363 700 1.00 40 7.37  106
P01 50–80 0.0205 700 1.30 40 5.24  106
P02 50–80 0.0205 700 0.30 40 2.60  105
P03 50–80 0.0205 700 0.10 40 4.00  105
P04 50–80 0.0205 700 0.00 40 5.60  105
S01 50–80 0.0205 400 1.00 40 7.47  106
S02 50–80 0.0205 500 1.00 40 8.30  106
S03 50–80 0.0205 800 1.00 40 8.98  106
R01 50–80 0.0205 700 1.00 40 4.24  106
R02 50–80 0.0205 700 1.00 40 6.85  106

The effect of the temperature on the rate of pyrrhotite dissolu-


tion was determined using several experiments conducted at tem-
peratures from 25 to 70 °C (experiments #T01-T05). As expected
the pyrrhotite dissolution rate increases when reaction tempera-
ture increases from 25 °C to 70 °C.
The initial pH was varied over the range of 0–1.3. This had a
very significant effect on the dissolution rate so that the dissolu-
tion rate increases from 5.24  106 mol/m2 s to 5.60  105 mol/
m2 s as the hydrogen ion concentration increases from 0.05 to
1 mol/L.

3.2. Rate equation for the H+ reaction

Sixteen rate measurements from this study were combined


Fig. 2. Graphs illustrating concentration of total iron (MFetotal) versus time for some with 47 rate data taken from the literature (see Table 1 for the data
typical experiments. sources) to create a set of 63 independent measurements. These
data spanned pH values from 0 to 5 and temperatures from 20 to
90 °C. An initial regression model found that one datum fell 5.7
performed with a 0.01 mol/L Sn(II) solution, the second (#T02) was standard deviations away from the best-fit line. This point was re-
conducted in the presence of metallic tin. A control experiment moved from the data set and the remaining 62 data were regressed
without the reducing agents (#R01) was carried out under similar to produce:
reaction conditions. When the reducing agents were added to the
reactor pyrrhotite dissolution rates were slightly higher, log rHþ ¼ 7:20ð0:59Þ  1:46ð0:04ÞpH
6.85  106 mol/m2 s for #R02 and 8.30  106 mol/m2 s for 3443ð194Þ
 ðR2
#T02, compared to the control experiment’s (#R01) rate of T
4.24  106 mol/m2 s. These results might reflect a minor rate ef- ¼ 0:98Þ ð13Þ
fect caused by the reduction of ferric iron on pyrrhotite surface
and in its lattice (Hsieh et al., 2002). The numbers in parentheses are one standard error of the
In order to investigate the effect of stirring speed on the rate parameter estimate. An example of the method to forward propa-
of pyrrhotite dissolution a series of experiments (#S01-S04 and gate these errors to find the error in log r is given in the Supple-
#T02) was carried out in the presence of Sn(0) at 40 °C, pH 1, mental materials for this paper. Fig. 3 shows the distribution of
and stirring rates ranging from 400 to 800 rpm. The rate of pyr- data and the goodness of the fit. The reaction order with respect
rhotite dissolution was essentially unaffected by the stirring to the concentration of H+ is 1.46 and the activation energy is
speeds greater than 400 rpm. This finding suggests that the 65.9 kJ/mol. Because diffusion coefficients have activation energy
mass transfer in solution is not the rate-determining step. A values near 15 kJ/mol, this high activation energy is further confir-
stirring speed of 700 rpm was chosen for the rest of the mation that the reaction is not limited by ion transport through
experiments. solution.
To study the effect of particle size on pyrrhotite dissolution,
three particle size fractions (25–50, 50–80 and 80–150 lm) were 3.3. Rate equations for the oxidation reactions
exposed to acidic solutions, in the presence of Sn(0), at 40 °C and
pH 1. It was found that the rate of iron release from pyrrhotite in- To highlight the role of non-oxidative dissolution of pyrrhotite
creased as the particle size decreased. According to Mudunkotuwa in the oxidative process we collected rate data from literature
et al. (2012), when the diameter of the particles decreases surface sources listed in Table 2. These data were used to develop rate
area and the number of kinks, edges and defect site density in- equations for oxidative dissolution of pyrrhotite by reaction with
crease creating hot spots of dissolution. O2 or Fe(III). The rate data are listed in Table 3S and details about
6 P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10

Fig. 3. (a) Comparison of log r values predicted by Eq. (13) with the measured values (62 data) (R2 = 0.984). (b) Residual plot showing the difference between the predicted
and measured values of log r versus log r (predicted). (c) Leverage plot showing the effect of pH on log r. (d) Leverage plot showing the effect of 1/T on log r. The dashed
horizontal line in each graph is the average of the dependent variable and the dashed lines surrounding the solid regression lines shows the confidence interval for the fit. If
the horizontal line lies entirely within the confidence interval, the regression variable does not significantly affect the dependent variable. The legend refers to the data
sources listed in Table 1.

Fig. 4. (a) Comparison of log r values predicted by Eq. (14) with the measured values (35 data) (R2 = 0.207). (b) Residual plot showing the difference between the predicted
and measured values of log r versus log r (predicted). Leverage plot showing the effect of 1/T on log r. (d) Leverage plot showing the effect of log P O2 on log r. (c) The dashed
horizontal line in each graph is the average of the dependent variable and the dashed lines surrounding the solid regression lines shows the confidence interval for the fit. If
the horizontal line lies entirely within the confidence interval, the regression variable does not significantly affect the dependent variable. The legend refers to the literature
sources listed in Table 2.
P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10 7

Fig. 5. (a) Comparison of log r values predicted by Eq. (15) with the measured values (48 data) (R2 = 0.390). (b) Residual plot showing the difference between the predicted
and measured values of log r versus log r (predicted). (c) Leverage plot showing the effect of log MFe(III) on log r. (d) Leverage plot showing the effect of 1/T on log r. The dashed
horizontal line in each graph is the average of the dependent variable and the dashed lines surrounding the solid regression lines shows the confidence interval for the fit. If
the horizontal line lies entirely within the confidence interval, the regression variable does not significantly affect the dependent variable. The legend refers to the literature
sources listed in Table 2.

the literature sources and experimental conditions are given in Ta- reactions discussed in the Introduction must be occurring, the sim-
ble 2. Multiple linear regression of these data showed that pH has plicity of the rate equation suggests that one of them dominates the
no significant effect on the rates for either reaction so the rates overall dissolution process giving us hope that the reaction process
were fit as functions of temperature and oxidant concentration is relatively simple. The scatter of the rates around the regression
or pressure. For the 35 data for O2 oxidation the resulting rate model is much larger than would be expected by forward propaga-
equation is: tion of measurement errors but it is typical of mineral dissolution
rate data (see Rimstidt et al. (2012) for a discussion of this problem).
log r O2 ¼  1:96ð2:6Þ þ 0:352ð0:162Þ log P O2
Fortunately, the relatively large number of rate data spanning a
1575ð675Þ wide range of temperatures and pH offsets this problem allowing
 ðR2 ¼ 0:21Þ ð14Þ
T us to develop an acceptable rate equation.
The reaction order for O2 is 0.35 and the activation energy is An interesting comparison can be made between pyrrhotite and
30.1 kJ/mol. For the 48 data for Fe(III) oxidation the rate equation troilite non-oxidative dissolution at 50 °C (Fig. 6). In order to deter-
is: mine rate of troilite dissolution, the data from Chirita and

log r FeðIIIÞ ¼  0:278ð1:7Þ þ 0:368ð0:089Þ log M FeðIIIÞ


1753ð494Þ
 ðR2 ¼ 0:39Þ ð15Þ
T
The reaction order for Fe(III) is 0.37 and the activation energy is
33.6 kJ/mol. The numbers in parentheses are one standard error of
the parameter estimate. Figs. 4 and 5 show the distribution of data
and the goodness of the fit.

4. Discussion and conclusions

Combining rate data from other laboratories with the new data
from the experiments reported in this paper allowed us to develop a
rate equation that not only spans a wider range of conditions but is
also more reliable than those previously obtained. The R2 value of
0.98 shows that 98% of the variance of the rates is explained by
the hydrogen ion concentration and temperature. This means that
even though the experiments were carried out using different reac-
Fig. 6. Comparison of the non-oxidative dissolution rates of troilite (filled circles)
tor schemes, extents of reaction, solution chemistries and pyrrho- from Chirita and Descostes (2006) with the rates for pyrrhotite predicted by Eq. (13)
tite samples those experimental differences appear to have no (solid line). The dashed lines represent 1 standard error for the value predicted by
significant effect on the rates. Although we know that all of the Eq. (13).
8 P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10

Table 4
Initial rates of non-oxidative dissolution of troilite at 50 °C based
on concentration versus time data taken from Chirita and
Descostes (2006).

M Hþ , mol/L Rate, mol/m2 s


0.2 2.95  105
0.1 2.02  105
0.06 7.60  106
0.04 6.20  106

Descostes (2006) were analyzed using the initial rate method. The
rates were derived from the slope of the regression lines for data
collected at the beginning of the experiments (0–1300 s). These
rates are given in Table 4. Fig. 6 shows that at low pH the non-oxi-
dative dissolution rates of pyrrhotite and troilite are essentially the
same. This similarity suggests that Eq. (13) is applicable to all pyr-
rhotite compositions regardless of the composition of pyrrhotite, Fig. 7. Rates of the H+, O2, and Fe(III) reactions predicted by the rate Eqs. (13)–(15)
including monoclinic pyrrhotite, hexagonal pyrrhotite and troilite. for the experimental conditions (25 °C) where the rate data were derived (shaded
areas and H+ line). The shaded rectangles represent ranges of pH and oxidant
This hypothesis is also supported by the small effect of reducing
concentrations for the rate data used to develop Eqs. (14) and (15). The overlap of
agents of Fe3+ on pyrrhotite dissolution rate in acidic media. the rates for these experimental conditions suggests that the rates determined by
Although reducing agents may change the composition of pyrrho- some of the experiments are a composite of two reactions.
tite surface by reducing the Fe3+ present the dissolution rates are
not significantly affected. narrow pH range of 2.0–2.75 (all experimental conditions lie
The regression models for the O2 and Fe(III) oxidation data are within the small rectangle) and the overlap between this range
somewhat disappointing because the correlation coefficients are and the H+ rates strongly suggest that the rates measured near
low, which means that the regression variables explain only 21% pH 2 represent a combination of rates from the H+ and Fe(III) reac-
and 39% of the variance of the rates respectively. This means that tions. Finally, the measured rates for O2 and Fe(III) reactions over-
the random and systematic errors in the data are quite large com- lap and this suggests that in O2 rich atmospheres the measured
pared to the effect of O2 or Fe(III) concentrations on the observed Fe(III) oxidation rates may contain a significant contribution from
rate. None the less, these equations represent the current best esti- O2 oxidation. These possible overlapping rates are a point of con-
mates for the effect of O2 and Fe(III) on pyrrhotite dissolution rates. cern and may help to explain the low correlation coefficients for
This lack of fit may be the result of one or more uncontrolled vari- the O2 and Fe(III) rate equations.
ables such as reactor design, extent of reaction, solution chemistry, These rate equations can also help us to interpret geochemical
pyrrhotite impurities or simply caused by the narrow range of processes. Although there are large uncertainties associated with
experimental conditions and relatively small numbers of data. the O2 and Fe(III) rate equations, they are the most reliable model-
Several previous studies have suggested that the pyrrhotite ing tools currently available and the pattern of rates versus pH
dissolution rates are limited by diffusion of reactants through a shown in Fig. 8 seem to provide a reasonable conceptual model.
SRL (Mycroft et al., 1995; Thomas et al., 1998, 2001, 2003; Mikhlin This pattern is similar to that for pyrite (Williamson et al., 2006),
et al., 2003; Chirita et al., 2008; Chirita, 2009). However, our results which is a useful comparison. In acid mine drainage settings Fe(III)
do not support that idea because the calculated activation energy concentrations tend to be controlled by the solubility of a ferric
values for the Fe(III) and O2 reactions are near 30 kJ/mol, which oxyhydroxide phase. Fig. 8 shows the Fe(III) rates recast in terms
is significantly larger than the 10 to 20 kJ/mol activation energy of Fe(III) concentrations for solutions in equilibrium with goethite
values expected for diffusion processes. More rate data for a wider and ferrihydrite. This figure shows that for pH < 2, Fe(III) oxidation
range of temperature and oxidant concentrations are needed to re- rates and H+ dissolution rates overlap. If the process of self-heating
solve this matter. Even thought the O2 and Fe(III) oxidation rate
equations could be better, the fitted parameters are the most prob-
able values based on existing data. As a result we can make at least
order-of-magnitude comparisons of reaction rates in order to gain
some sense of the relative importance of each of the reactions.
Fig. 7 compares the rates of pyrrhotite dissolution for each of
the three reactions based on the predictions of the rate Eqs.
(13)–(15). This figure allows some tentative observations regard-
ing the rate experiments. The H+ experiments spanned the pH
range from 0 to 5 and were carried out under anoxic conditions.
These results appear to be quite reliable and unaffected by O2
and Fe(III) oxidation reactions. Even though some Fe(III) was re-
leased by the anoxic dissolution process, the good correlation be-
tween the experiments, which trapped that Fe(III) using Sn
reported in this paper, and the others, which did not, suggests that
the Fe(III) did not significantly affect the observed rates. The O2
rate experiments were performed over a pH range of 1.97–3.5
(all experimental conditions lie within the large rectangle). The
rates at the low pH end of this range are similar to the H+ rates Fig. 8. Rates of the H+, O2, and Fe(III) reactions predicted by the rate equations for
and there is a reasonable possibility that those rates are a compos- field conditions (25 °C) where the Fe(III) concentrations tend to be controlled by
ite of both the H+ and O2 reactions. The Fe(III) experiments span a ferric oxyhydroxide phases and the P O2 ¼ 0:21.
P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10 9

that leads to spontaneous combustion involves pore waters that Bugajinski, J., Gamsjager, H., 1982. The kinetics of dissolution of monoclinic
pyrrhotite in aqueous acid solutions. Monat. Chem. 113, 1087–1092.
have this very low pH, rapid oxidation and H2S generation could
Cai, M.F., Dang, Z., Chen, Y.W., Belzile, N., 2005. The passivation of pyrrhotite by
co-occur and this would explain the observed H2S in self-heating surface coating. Chemosphere 61, 659–667.
ores. Acid mine drainage solutions seldom attain pH < 2.5 because Chirita, P., 2009. Iron monosulfide (FeS) oxidation by dissolved oxygen:
of the sulfate-bisulfate buffer. Because of this, neither the H+ nor characteristics of the product layer. Surf. Interface Anal. 41, 405–411.
Chirita, P., Descostes, M., 2006. Anoxic dissolution of troilite in acidic media. J.
the Fe(III) reaction is likely to play a significant role in acid mine Colloid Interface Sci. 294, 376–384.
drainage production from pyrrhotite rich rocks. This contrasts with Chirita, P., Schlegel, M., 2012. Reaction of FeS with Fe(III)-bearing acidic solutions.
pyrite rich rocks where the Fe(III) oxidation reaction becomes sig- Chem. Geol. 334, 131–138.
Chirita, P., Preda, M., Rusu, O., 2002. Natural pyrrhotite dissolution in aqueous
nificant near pH 4 leading to runaway acid production at lower pH. solution. South. Braz. J. Chem. 10, 11–17.
The pyrrhotite oxidation process is dominated by the O2 reaction at Chirita, P., Descostes, M., Schlegel, M.L., 2008. Oxidation of FeS by oxygen-bearing
pH > 3 and runaway acid production caused by increasing Fe(III) acidic solutions. J. Colloid Interface Sci. 321, 84–95.
Craig, J.R., Vokes, F.M., 1993. The metamorphism of pyrite and pyritic ores: an
concentration linked to declining pH can barely begin before the overview. Mineral. Mag. 57, 3–18.
pH decline is halted by the sulfate–bisulfate buffer. An important Descostes, M., Vitorge, P., Beaucaire, C., 2004. Pyrite dissolution in acidic media.
but unanswered question is whether the sulfur rich layer that Geochim. Cosmochim. Acta 68, 4559–4569.
Ebrahimi, S., Kleerebezem, R., van Loosdrecht, M.C.M., Heijnen, J.J., 2003. Kinetics of
has been found in laboratory experiments plays a role in acid mine the reactive absorption of hydrogen sulfide into aqueous ferric sulfate solutions.
drainage settings. Elemental sulfur is seldom reported by field Chem. Eng. Sci. 58, 417–427.
studies of pyrrhotite oxidation. This observation can be explained Ferrell, R.T., Himmelblau, D.M., 1967. Diffusion coefficients of nitrogen and oxygen
in water. J. Chem. Eng. Data 12, 111–115.
if we consider that sulfur rich layers are related to H+ reaction,
Fleet, M.E., 1978. The pyrrhotite–marcasite transformation. Can. Mineral. 16, 31–35.
which favors the release of iron relative to sulfur. Since H+ reaction Gholami, Z., Torabi Angaji, M., Gholami, F., Razavi Alavi, S.A., 2009. Reactive
is not important at pH > 3 (field conditions), it appears unlikely absorption of hydrogen sulfide in aqueous ferric sulfate solution. Int. J. Chem.
that layers of elemental sulfur that are thick enough to cause to dif- Biol. Eng. 2, 88–90.
Gleisner, M., 2005. Quantification of mineral weathering rates in sulfidic mine
fusion-limited reactions are present. Furthermore if significant sul- tailings under water-saturated conditions. Biogeochemistry. Stockholm
fur-rich layers did form they would likely reduce the oxidation rate University.
to the point that pyrrhotite rich rocks would produce little or no Gunsinger, M.R., Ptacek, C.J., Blowes, D.W., Jambor, J.L., 2006. Evaluation of long
term sulfide oxidation processes within pyrrhotite-rich tailings, Lynn Lake,
acid. Also, it is possible that significant sulfur rich layers do not Manitoba. J. Contam. Hydrol. 83, 149–170.
form because microbes consume the sulfur rich layer as quickly Hammarstrom, J.M., Seal II, R.R., Meier, A.L., Kornfeld, J.M., 2005. Secondary
as it forms. The absence of significant sulfur rich layers for field sulfate minerals associated with acid drainage in the eastern US:
recycling of metals and acidity in surficial environments. Chem. Geol.
conditions means that the rates for experiments with little or no 251, 407–431.
sulfur coating (likely the experiments at pH > 3), reported in Harries, D., Pollok, K., Langenhorst, F., 2013. Oxidative dissolution of 4C- and NC-
the literature or acquired in the future studies, could be used to de- pyrrhotite: intrinsic reactivity differences, pH dependence, and the effect of
anisotropy. Geochim. Cosmochim. Acta 102, 23–44.
velop better rate equations of pyrrhotite oxidation, applicable to Hsieh, Y.P., Chung, S.W., Tsau, Y.J., Sue, C.T., 2002. Analysis of sulfides in the
natural settings. presence of ferric minerals by diffusion methods. Chem. Geol. 182, 195–
We hope that this paper will prompt the readers to take a 201.
Janzen, M.P., Nicholson, R.V., Scharer, J.M., 2000. Pyrrhotite reaction kinetics:
thoughtful approach toward the design and interpretation of rate
reaction rates for oxidation by oxygen, ferric iron, and for nonoxidative
experiments. Rate experiments are often carried out over a very dissolution. Geochim. Cosmochim. Acta 64, 1511–1522.
limited set of conditions and the resulting rate data are seldom Jerz, J.K., Rimstidt, J.D., 2003. Efflorescent iron sulfate minerals: paragenesis, relative
compared to rates reported by other sources to identify reliable stability and environmental impact. Am. Mineral. 88, 1919–1932.
Johnsson, J., 2002. South Strafford’s Elizabeth copper mine: the Tyson Years, 1880–
patterns and to remediate inconsistencies. This can lead to over 1902. Vermont History 70, 130–152.
confidence in narrowly constrained rates and to a tendency to over Jones, C.F., LeCount, S., Smart, R.St.C., White, T.J., 1992. Compositional and structural
interpret them. Compilation and comparison of rate data from alteration of pyrrhotite surfaces in solution: XPS and XRD studies. Appl. Surf.
Sci. 55, 65–85.
multiple sources gives a fresh and broad view of a reaction pro- Karpenko, V., Norris, J.A., 2002. Vitrol in the history of chemistry. Chem. Listy 96,
cesses. The widespread use of computers and the availability of 997–1005.
statistical software allow for easy correlation of the data to pro- Kimball, B.E., Rimstidt, J.D., Brantley, S.L., 2010. Chalcopyrite dissolution rate laws.
Appl. Geochem. 25, 972–983.
duce rate equations that recast the experimental data into a form Laurence, R.A., 1965. Field Trip No. 2 Ducktown, Tennessee. Joint ACA-MSA Meeting,
that can be used to model geochemical processes. Gatlinburg, TN. 18–37.
Li, Y.H., Gregory, S., 1974. Diffusion of ions in sea water and in deep-sea sediments.
Geochim. Cosmochim. Acta 38, 703–714.
Acknowledgements Lin, Z., 1997. Mineralogical and chemical characterization of wastes from the
sulfuric acid industry in Fauln, Sweden. Environ. Geol. 30, 152–162.
This work was supported by a grant of the Romanian National Liteanu, C., 1971. Chimie Analitica Cantitativa. Volumetria. Editura Didactica si
Pedagogica, Bucuresti.
Authority for Scientific Research, CNDI–UEFISCDI, Project Number
Mikhlin, Y., Tomashevic, Y., 2005. Pristine and reacted surfaces of pyrrhotite and
51/2012. The authors thank Wayne Nesbitt for his helpful sugges- arsenopyrite as studied by X-ray absorption near-edge structure spectroscopy.
tions. The comments of K. Savage and two anonymous reviewers Phys. Chem. Miner. 32, 19–27.
Mikhlin, Y., Varnek, V., Asanov, I., Tomashevich, Y., Okotrub, A., Livshits, A., Selyutin,
led to significant improvement of the manuscript.
G., Pashkov, G., 2000. Reactivity of pyrrhotite (Fe9S10) surfaces: spectroscopic
studies. Phys. Chem. Chem. Phys. 2, 4393–4398.
Appendix A. Supplementary material Mikhlin, Yu.L., Kuklinskiy, A.V., Pavlenko, N.I., Varnek, V.A., Asanov, I.P., Okotrub,
A.V., Selyutin, G.E., Solovyev, L.A., 2003. Spectroscopic and XRD studies of the air
degradation of acid-reacted pyrrhotite. Geochim. Cosmochim. Acta 66, 4057–
Supplementary data associated with this article can be found, in 4067.
the online version, at http://dx.doi.org/10.1016/j.apgeochem. Mudunkotuwa, I.A., Rupasinghe, T., Wu, C.M., Grassian, V.H., 2012. Dissolution of
ZnO nanoparticles at circumneutral pH: a study of size effects in the presence
2013.11.013. and absence of citric acid. Langmuir 28, 396–403.
Mycroft, J.R., Nesbitt, H.W., Pratt, A.R., 1995. X-ray photoelectron and Auger electron
References spectroscopy of air-oxidized pyrrhotite: distribution of oxidized species with
depth. Geochim. Cosmochim. Acta 59, 721–733.
Pratt, A.R., Nesbitt, H.W., 1997. Pyrrhotite leaching in acid mixtures of HCl and
Alexeyev, V., 1969. Quantitative Analyses. Mir, Moscow.
H2SO4. Am. J. Sci. 297, 807–820.
Asai, S., Konishi, Y., Yabu, T., 1990. Kinetics of absorption of hydrogen sulfide into
Pratt, A.R., Muir, I.J., Nesbitt, H.W., 1994a. X-ray photoelectron and Auger electron
aqueous ferric sulfate solutions. AIChE J. 36, 1331–1338.
spectroscopic studies of pyrrhotite and mechanism of air oxidation. Geochim.
Belzile, N., Chen, Y.W., Chai, M.F., Li, Y., 2004. A review on pyrrhotite oxidation. J.
Cosmochim. Acta 58, 827–841.
Geochem. Explor. 84, 65–76.
10 P. Chiritßă, J.D. Rimstidt / Applied Geochemistry 41 (2014) 1–10

Pratt, A.R., Muir, I.J., Nesbitt, H.W., 1994b. Generation of acids from mine waste: Somot, S., Finch, J.A., 2010. Possible role of hydrogen sulfide gas in self-heating of
oxidative leaching of pyrrhotite in dilute H2SO4 solutions at pH 3.0. Geochim. phrrhotite-rich materials. Miner. Eng. 23, 104–110.
Cosmochim. Acta 58, 5147–5159. Tester, J.W., Worley, W.G., Robinson, B.A., Grigsby, C.O., Feerer, J.L., 1994. Correlating
Quinn, M.L., 1993. Industry and environment in the Appalachian copper basin, quartz dissolution kinetics in pure water from 25 to 625 °C. Geochim.
1890–1930. Technol. Cult. 34, 575–612. Cosmochim. Acta 58, 2407–2420.
Ramdohr, P., 1980. The Ore Minerals and their Intergrowths. Pergamon Press, Thomas, J.E., Jones, C.F., Skinner, W.M., Smart, R.St.C., 1998. The role of surface sulfur
Oxford. species in the inhibition of pyrrhotite dissolution in acid conditions. Geochim.
Rimstidt, J.D., Newcomb, W.D., 1993. Measurement and analysis of rate data: the Cosmochim. Acta 62, 1555–1565.
rate of reaction of ferric iron with pyrite. Geochim. Cosmochim. Acta 57, 1919– Thomas, J.E., Smart, R.S.C., Skinner, W.M., 2000. Kinetic factors for
1934. oxidative and non-oxidative dissolution of iron sulfides. Miner. Eng.
Rimstidt, J.D., Brantley, S.L., Olsen, A.A., 2012. Systematic review of forsterite 13, 1149–1159.
dissolution data. Geochim. Cosmochim. Acta 99, 159–178. Thomas, J.E., Skinner, W.M., Smart, R.S.C., 2001. A mechanism to explain sudden
Robie, R.A., Hemingway, B.S., 1995. Thermodynamic Properties of Minerals and changes in rates and products for pyrrhotite dissolution in acid solution.
Related Substances at 298.15 K and 1 bar (105 Pascals) Pressure and at Higher Geochim. Cosmochim. Acta 65, 1–12.
Temperatures. U.S. Geological Survey, Washington, DC. Thomas, J.E., Skinner, W.M., Smart, R.S.C., 2003. A comparison of the dissolution
Romano, G.Y., 2012. Kinetics of Pyrrhotite Oxidation in Seawater: Implications for behavior of troilite with other iron(II) sulfides; implications of structure.
Mining Seafloor Hotsprings. Master Thesis. University of California Riverside. Geochim. Cosmochim. Acta 67, 831–843.
Singer, P.C., Stumm, W., 1970. Acidic mine drainage: the rate determining step. Williamson, M.A., Kirby, C.S., Rimstidt, J.D., 2006. Iron dynamics in acid mine
Science 167, 1121–1123. drainage. In: 7th International Conference on Acid Rock Drainage (ICARD).
Skinner, W.M., Nesbitt, H.W., Pratt, A.R., 2004. XPS identification of bulk hole defects American Society of Mining and Reclamation (AMSR), Lexington, KY, pp. 2411–
and iterant Fe 3d electrons in natural trolite (FeS). Geochim. Cosmochim. Acta 2423.
68, 2259–2263.

You might also like