Progress in Energy and Combustion Science 32 (2006) 295–314

www.elsevier.com/locate/pecs

Decomposition and oxidation of pyrite

Guilin Hu a, Kim Dam-Johansen a,*, Stig Wedel a, Jens Peter Hansen b
a
CHEC, Department of Chemical Engineering, Technical University of Denmark, Building 229, 2800 Lyngby, Denmark
b
FL Smidth A/S, Copenhagen, Denmark
Received 23 June 2005; accepted 30 November 2005
Available online 7 February 2006

Abstract

An assessment of the mechanisms and kinetics of pyrite transformation in inert and oxygen-containing atmospheres at elevated
temperatures has been performed based on a literature review. In inert atmospheres, the transformation of pyrite proceeds through a
multi-step process in the sequence of pyrite/pyrrhotite/troilite/iron. The whole process is determined by the temperature and
the total sulfur gas pressure in the system. Under equilibrium conditions, the composition of the resulting pyrrhotite from pyrite
decomposition is determined by the temperature.
The decomposition of pyrite to pyrrhotite follows the unreacted core model and may be controlled by zero-order surface
reaction, gas film/product layer diffusion or a combination of these mechanisms, depending on the reaction conditions.
The transformation of pyrite in oxygen-containing atmospheres can take place in two different ways depending on the
conditions, such as temperature, oxygen concentration, flow and particle size. At lower temperatures (!about 800 K) and higher
oxygen concentrations, pyrite will be directly oxidized. The direct oxidation process follows the unreacted core model and may be
controlled by chemical reaction or the inward diffusion of oxygen due to the pore-blocking effect by the formation of ferric/ferrous
sulfate.
At higher temperatures (Oabout 800 K) or lower oxygen concentrations, pyrite will be transformed by a two-step process: the
thermal decomposition of the pyrite to form porous pyrrhotite as the first step, and the successive oxidation of the formed pyrrhotite
as the second step. The first step proceeds similarly to the thermal decomposition of pyrite in an inert atmosphere, but may proceed
with a much higher rate due to the extra heating effect from the oxidation of the formed sulfur gas and the reduction of eventual
diffusion resistance of the outgoing sulfur gas through the product layer and/or the gas film. The second step can proceed with
pyrrhotite in a solid or a molten state depending on the particle temperature.
Iron oxides (mainly hematite, Fe2O3 and magnetite, Fe3O4) are the main products of the oxidation of pyrite. Hematite usually
forms at lower temperatures (!1173–1273 K) and at high oxygen concentrations, whereas magnetite usually forms at higher
temperatures and/or at low oxygen concentrations.
Sulfates (mainly ferrous sulfate, FeSO4 and ferric sulfate, Fe2(SO4)3) may be formed as minor products during the oxidation of
pyrite at temperatures lower than 873–923 K. The formation of the sulfates is probably determined by the gas composition at the
reaction front. Ferrous sulfate usually forms in a gas that is rich in SO2, whereas ferric sulfate usually forms in a gas that is rich in
SO3.
q 2006 Elsevier Ltd. All rights reserved.

Keywords: Pyrite; Pyrrhotite; Oxidation; Transformation; Decomposition; Mechanism; Sulfur dioxide

* Corresponding author. Tel.: C45 45252800; fax: C45 45252845.

E-mail address: [email protected] (K. Dam-Johansen).

0360-1285/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.pecs.2005.11.004
296 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
2. Pyrite transformation in an inert atmosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
2.1. Transformation process and products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
2.2. Kinetics of pyrite decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
3. Pyrite transformation in an oxygen-containing atmosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
3.1. Transformation processes and products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
3.1.1. Transformation processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
3.1.2. The formation of iron oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
3.1.3. The formation of sulfates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
3.2. Kinetics of pyrite oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
3.2.1. Direct oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
3.2.2. The two-step process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313

Nomenclature
f fraction with fast reaction rate Tc temperature of the unreacted core
h heat transfer coefficient of the gas film Tg main gas temperature
DHr molar (solid reactant B) enthalpy change of XB conversion of solid reactant B
the reaction XB,1 conversion of the fraction with fast reaction
k thermal conductivity of the product layer rate
kr,1, kr,2 rate constants XB,2 conversion of the fraction with slow
MB mole weight of the solid reactant reaction rate
R radius of the particle t time for complete conversion
t reaction time rB density of solid reactant B

1. Introduction following is a review of the literature on the

mechanisms and kinetics of pyrite transformation at
Pyrite (FeS2) is a naturally occurring material that elevated temperatures in inert and oxygen-containing
can be found in concentrated form in nature and also as atmospheres. The transformation of pyrite in inert
impurities in coal and many other minerals. The wide atmospheres is relevant here simply because, in some
occurrence of pyrite in different minerals and coals situations, it is part of the whole transformation process
makes it one of the main sources of SO2 emission from of pyrite in oxygen-containing atmospheres.
various industrial activities, such as the metallurgical
industry, power production and cement production. 2. Pyrite transformation in an inert atmosphere
SO2 is formed when pyrite is oxidized in industrial
processes. 2.1. Transformation process and products
The transformation of pyrite in an oxygen-contain-
ing atmosphere is a complicated process, and may When heated in an inert atmosphere, pyrite (FeS2)
proceed by different mechanisms under different will decompose to release sulfur gas and form iron
conditions. Parameters such as temperature, particle sulfide of lower sulfur content—pyrrhotite FeSx (here
size, flow condition and properties of the surrounding 1%x!2). The thermodynamic properties of this
atmosphere can all affect the transformation process. process have been studied by a number of researchers
In the past decades, much research has been carried [1–9]. One of the most comprehensive and representa-
out to clarify the mechanisms and kinetics of pyrite tive investigations was performed by Toulmin et al. [5].
transformation under different conditions. The The phase diagram of the Fe–S system (Fig. 1) that they
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 297

Fig. 1. Phase diagram of the Fe–S system (modified from Toulmin and Barton [5]). In the figure: fS2 , the fugacity of sulfur gas; S(l), liquid sulfur;
S(v), sulfur vapor; PY, pyrite; PO, pyrrhotite; Fe, iron; N, fraction of FeS in pyrrhotite when pyrrhotite is expressed as (FeS)N(S2)1KN. The line
marked with PY/PO is the equilibrium line between pyrite and pyrrhotite. The line marked with PO/Fe is the equilibrium line between troilite and
elemental iron. The lines between the equilibrium lines PY/PO and PO/Fe represent isopleths of pyrrhotite of different N values. b/g is the transition
point of the crystal structure of pyrrhotite from the b form to the g form. The dashed line illustrates the transformation process of pyrite at a fixed
temperature.

presented gives a good illustration of the transformation content. In a practical decomposition process, there will
process of pyrite at elevated temperatures and shows be a gradient of the partial pressure of sulfur gas across
the equilibrium relationship between pyrite, pyrrhotite the formed pyrrhotite layer due to the outgoing
and other related decomposition products. In the diffusion of the formed sulfur gas. The existence of
diagram, pyrrhotite is expressed as (FeS)N(S2)1KN this gradient means that the partial pressure of sulfur
because pyrrhotite was supposed to be a solid solution gas in the pyrrhotite layer will deviate from the
of sulfur in troilite (FeS). Troilite is FeSx with xZ1, and equilibrium state, and thus cause further decomposition
is also called pyrrhotite by some authors. x and N are of the primary pyrrhotite before pyrite is completely
related by the equation xZ(2KN)/N. decomposed. Fegley and Lodders [11] and Lambert et al.
The diagram shows that at a fixed temperature pyrite [10] identified two different pyrrhotite phases in
will start to decompose to form pyrrhotite and sulfur partially decomposed pyrite particles. A reasonable
gas when the sulfur gas pressure is lowered to the explanation for their observations is the further
corresponding equilibrium pressure (at the point A
where the dashed line crosses the line marked PY/PO).
This process can be represented by the following
overall reaction:
FeS2 ðsÞ4 FeSx ðsÞ C ð1K0:5xÞS2 ðgÞ (1)
Under equilibrium conditions, sulfur content in the
formed pyrrhotite (x in FeSx) is determined by the
temperature. Fig. 2, created by using the empirical
equation obtained by Lambert et al. [10], shows the
correlation between x and the decomposition tempera-
ture. The correlation is valid in the temperature interval
from 523 to 1016 K.
The primary pyrrhotite formed from the decompo-
sition of pyrite will continue to release sulfur gas if the Fig. 2. Dependence of x (moles of S per mole of Fe in pyrrhotite FeSx)
sulfur gas pressure in the system is lowered further. on the decomposition temperature of pyrite (created by using the
This process will result in pyrrhotite of lower sulfur equation shown in the figure obtained by Lambert et al. [10]).
298 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

decomposition of the primary pyrrhotite due to the

deviation of the partial pressure of sulfur gas from the
equilibrium pressure.
If this process continues, the stoichiometric iron
sulfide FeS will be formed (at the point B where the
dashed line crosses the line marked PO/Fe in Fig. 1).
This process can be represented by the following
overall reaction:
FeSx ðsÞ4 FeSðsÞ C 0:5ðxK1ÞS2 ðgÞ (2)
If the sulfur gas pressure is lowered further, FeS will
decompose to form elemental iron (the line marked by
PO/Fe in Fig. 1). This can be represented by the Fig. 3. Molar fraction of allotropic species of sulfur gas over pyrite
following overall reaction: under equilibrium conditions (the diagram is calculated by using data
from Gurvich et al. [19] and Hong and Fegley [9]).
FeSðsÞ4 FeðsÞ C 0:5S2 ðgÞ (3)
The thermal decomposition of pyrite should be a
reversible process in light of the fact that pyrite/
shows the molar fractions of the different allotropic
pyrrhotite can be readily synthesized by the sulphida-
species in the temperature interval that is relevant to
tion reaction between elemental iron and sulfur gas
pyrite decomposition under equilibrium conditions.
under fairly mild conditions [12–16].
The figure is constructed by using the equilibrium
Pyrite has an incongruent melting point of 1016 K,
constants compiled by Gurvich et al. [19] and the
with a total equilibrium sulfur gas pressure of around
equation presented by Hong et al. [9] (as shown below)
1 MPa [2]. The incongruent melting point of pyrite
for the calculation of the total equilibrium sulfur gas
increases slightly with increasing total sulfur gas
pressure over pyrite (valid from 598 to 1016 K):
pressure—for example, 1021 K at 33.5 MPa [2]. At
the incongruent melting point, pyrite melts to form log10 P Z 16:2ðG0:21ÞK15700ðG150Þ=T (4)
pyrrhotite and a sulfur-rich liquid. This point is the
invariant point of the system. The incongruent melting Here, P is the total sulfur gas pressure over pyrite in
point is also the maximum decomposition temperature bar (1 barZ0.1 MPa), T, the temperature in Kelvin.
of pyrite, as pyrite cannot exist at a temperature that is It is assumed in the calculation that allotropic
higher than its incongruent melting point. Under species with n greater than 8 are negligible. The figure
normal conditions, this maximum temperature should shows that under equilibrium conditions, S2 is the
be around 1016 K. absolutely dominant species for the decomposition of
During the thermal decomposition of pyrite, sulfur pyrite at temperatures lower than about 800 K (the
gas is released. Sulfur gas normally consists of a molar fraction of S2 is around 99%; the total
number of allotropic species [17,18] and should be equilibrium sulfur gas pressure is approximately
considered as a mixture rather than a pure gas. These 38 Pa). At higher temperatures, the molar fractions of
allotropic species can be represented by Sn, with n the larger allotropic species (mainly S3, S5 and S6)
varying from 1 to 8 or higher. Under equilibrium begin to increase to a significant level. The total molar
conditions, the fraction of each species in the mixture is fraction of these large allotropic species is around 6% at
determined by the chemical equilibrium. Low tem- 900 K (the total sulfur gas pressure is approximately
peratures and high-total sulfur gas pressures are 5700 Pa) and 35% at 1000 K (the total equilibrium
favorable to the formation of large allotropic species, sulfur gas pressure is approximately 0.3 MPa).
whereas high temperatures and low-total sulfur gas It is clear that, under equilibrium conditions, S2
pressures are favorable to the formation of small cannot be assumed to be the dominant species at
allotropic species, especially S2. However, under temperatures exceeding about 900 K. In practical
equilibrium conditions, the fraction of the large situations, the formation of the larger allotropic species
allotropic species increases with the temperature. This may need to be considered if such conditions are
is because the effect of the total sulfur gas pressure, present (i.e. with a temperature higher than about 900 K
which increases exponentially with the temperature, and with a total sulfur gas pressure that is close to the
surpasses the effect of the temperature increase. Fig. 3 equilibrium pressure).
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 299

According to Meyer [18], the establishment of reaction front between the unreacted solid reactant core
equilibrium between the different allotropic species and the formed product layer. This reaction front moves
can be slow when the temperature is lower than about inward with time and the progress of the reaction. The
673 K, and the sulfur gas composition can deviate from whole reaction process can be controlled by gas film/
the equilibrium values and be dependent on the sources product layer diffusion when the reaction involves
from which the sulfur gas is generated. gaseous reactants and/or gaseous products and when
Reactions (1)–(3) are endothermic. The existence of the diffusion process is slow, or by chemical reaction
non-stoichiometric pyrrhotite makes the individual when the reaction is slow compared with the diffusion,
calculation of enthalpy change of reactions (1) and or by a combination of these steps (mixed control)
(2) difficult. Hong and Fegley [20] summarized the when none of them is dominant.
enthalpy change of reaction (1) that was obtained by For endothermic gas–solid reactions, the reaction is
different authors. The values ranged from 234 to 449 kJ more or less influenced by heat transport through the
per mol sulfur gas (S2) formed. Hong and Fegley [20] gas film and/or product layer due to the requirement of
calculated it to be 284.5 kJ/mol S2 formed in the energy balance between the heat needed for the reaction
temperature interval from 600 to 900 K. The enthalpy and the heat transferred to the reaction interface. The
change for reaction (3) can be readily calculated by consequence of large heat transport resistance is the
using thermodynamic data of the relevant substances occurrence of a significant temperature gradient across
(166 kJ per mol of FeS at 298 K, calculated by using the gas film and the product layer. This means that the
data from NIST-JANAF Thermochemical Tables, reaction is proceeding at a lower temperature than that
Fourth Edition, by M.M. Chase, 1998). of the main gas phase. This can be shown by
performing an energy balance for the reaction.
Eq. (5) for the heat transport can be established by
2.2. Kinetics of pyrite decomposition assuming that:

The kinetics of the thermal decomposition of pyrite (a) Quasi-steady-state approximation is applicable
(reaction (1)) has been studied by a number of authors. (the temperature is assumed always to be in an
As observed by Hoare et al. [21], Fegley and Lodders equilibrium state in the particle due to the much
[11] and Hong and Fegley [20], the thermal decompo- faster heat transfer rate than the shrinking rate of
sition of pyrite (reaction (1)) always progresses with a the core).
clear boundary between the unreacted pyrite and the (b) Heat transport by radiation is negligible.
formed product layer of pyrrhotite. This is well (c) Heat transport by convection in the product layer
demonstrated by the scanning electron microscope is negligible.
(SEM) picture of a partially decomposed pyrite particle (d) The heat capacity of the particle is negligible when
taken by Hong and Fegley [20] (Fig. 4). compared to the reaction heat
This is a typical behavior when the reaction followed
the so-called ‘unreacted core model for particles of dXB ðTg KTc Þ
Z  : (5)
unchanging size’ [22]. The unreacted core model dt DH r R R2 K1=3
h C k ðð1KXB Þ
r B
3MB K1Þ
assumes that the reaction occurs at the interface of
the unreacted solid reactant core. There is a clear Eq. (5) shows that a significant temperature
difference across the gas film and the product layer
(TgKTc) can occur at a high reaction rate and/or for a
large particle size. Flow conditions, gas properties and
properties of the product layer can also affect this
temperature difference by influencing the heat transfer
coefficient h and the thermal conductivity k. When an
endothermic reaction is significantly influenced by heat
transport, the effect of an increase of the main gas
temperature on the reaction is expected to be weaker
than under true isothermal conditions (i.e. the main gas
Fig. 4. SEM picture of a partially decomposed pyrite particle (particle phase and the particles have the same temperature).
size: length!width!thicknessZ1–2 cm!1 cm!0.05 cm) (modi- This is because the core temperature will not increase
fied from Hong and Fegley [20]). proportionally with the increase of the main gas
300 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

temperature due to the fact that reaction rate increases Lambert et al. [10] studied the thermal decompo-
exponentially with the temperature, whereas heat sition of pyrite in inert atmospheres (vacuum, argon and
transport increases linearly with the temperature helium) in a thermogravimetric analysis (TGA) in the
difference. temperature interval from 620 to 973 K. Particles of
The thermal decomposition of pyrite is an endother- 0.21–0.25 mm were used for the experiments. Based on
mic reaction with both solid and gaseous products but the observed linear relationship between the mass loss
without gaseous reactants. The experimental obser- per unit core surface area and the reaction time, it was
vations by a number of authors indicate that the thermal concluded that the process followed the unreacted core
decomposition of pyrite is often controlled by chemical model and was controlled by chemical reaction. The
reaction. reaction was assumed to be first-order in the surface
Schwab and Philinis [23] studied the thermal area of the unreacted core.
decomposition of pyrite by the thermogravimetric Fegley and Lodders [11] and Hong and Fegley [20]
method in the temperature interval from 873 to 923 K in the same research group studied the thermal
in CO2 and with pyrite particles of 0.01–0.1 mm. decomposition of pyrite in CO2 in a tube furnace in
Based on the observed apparent activation energy of the temperature interval from 663 to 863 K and with a
approximately 130 kJ/mol and the observation that single pyrite particle. The pyrite particle was in slice
the reaction rate was not sensitive to the change of form and with dimensions of 10–20 mm!10 mm!
the gas flow rate, it was concluded that the process 1 mm and 10–20 mm!10 mm!0.5 mm, respectively.
was controlled by chemical reaction. It was also Based on the shrinking core phenomenon observed by
observed that the conversion versus time curve in the SEM and the linear relationship between the conversion
position of 20–30% conversion showed a break in the and the reaction time, it was concluded that the
temperature interval from 873 to 888 K (the decreas- decomposition was controlled by chemical reaction
ing conversion rate was increasing at the break point). and that the reaction was zero-order with the rate
It was supposed that this phenomenon was caused by constant having the unit of length/time (this is
the surface catalytic effect by the formed pyrrhotite equivalent to ‘first-order in the surface area of the
phase. unreacted core’).
Coats and Bright [24] studied the thermal decompo- However, under unfavorable flow conditions, the
sition of pyrite in a quartz reactor in the temperature thermal decomposition of pyrite can also be signifi-
interval from 873 to 926 K with a single pyrite particle cantly influenced by the diffusion, as shown by the
fixed in argon. The pyrite particle was cylindrical and studies of Boyabat et al. [25].
was formed by pressing pyrite powder under high Boyabat et al. [25] studied the thermal decompo-
pressure. The pyrite pellet was 6.25 mm in both sition of pyrite in a horizontal tube furnace in the
diameter and length. Based on the observed linear temperature interval from 673 to 1073 K in nitrogen.
relationship between the unreacted core radius and the Different particle sizes ranging from 0.5 to 1.3 mm
reaction time, it was concluded that the decomposition were used. The conversion rate was observed to be
process was controlled by chemical reaction. The significantly influenced by the gas flow rate, the
reaction was assumed to be first-order in the surface temperature and the particle size. Based on the bent
area of the unreacted core (equivalent to assuming zero- form of the conversion versus time curves, they
order surface reaction with the rate constant expressed concluded that the process was controlled by heat
per unit surface area and reaction time). transport through a gas film at the initial stage of the
Hoare et al. [21] studied the thermal decomposition reaction, and by product layer diffusion at the later
of pyrite in nitrogen by the thermogravimetric stage of the reaction. The conclusion of heat transport
method. A single pyrite particle of a mass ranging control at the low conversions was based on the
from 0.5 to 10 mg was used for the experiments. It observation that there was a linear relationship between
was concluded that the decomposition of pyrite to the conversion and the reaction time. The conclusion of
pyrrhotite followed the unreacted core model based on product layer diffusion control at the high conversion
the microscopic observation of the shrinking core stage was based on the fact that the model for product
phenomenon. The decomposition process was sup- layer diffusion control [22] fitted the experimental data
posed to be controlled by chemical reaction as this well. They determined the apparent activation energy to
made the best-fit for their experimental data. The be approximately 113 kJ/mol for the heat transport
reaction was assumed to be first-order in the surface control and approximately 96 kJ/mol for the product
area of the unreacted core. layer diffusion control.
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 301

However, the conclusion of product layer diffusion point can be defined. Heat transport is not a natural step
control as suggested by Boyabat et al. [25] is of the reaction process (gas phase diffusion/product
questionable, as the activation energy of 96 kJ/mol layer diffusion/reaction), but a parallel process which
that was obtained by them is much higher than the is connected with the reaction. It is, therefore, more
apparent activation energy of about 10–15 kJ/mol for accurate to say that the reaction is influenced by heat
gas phase diffusion in the relevant temperature interval transport rather than controlled by heat transport.
for pyrite decomposition. The activation energy that The activation energy is an important kinetic
they measured at the start of the decomposition is only parameter that indicates the significance of the
17 kJ/mol higher than the activation energy that was temperature’s influence on the reaction. Activation
measured at high conversions. This difference seems to energy is also useful for the judgment of the controlling
be too small to justify the supposed shift of the mechanism. Table 1 lists the apparent activation
controlling mechanisms. Boyabat et al. [25] observed in energies obtained by the authors shown above and of
their experiments that the conversion rate was some other references.
significantly affected by gas flow rate even at the initial The activation energies that were obtained by the
stage. This clearly indicates that the decomposition different authors vary in a wide range from the lowest
process was significantly influenced by diffusion, most value of 96 kJ/mol to the highest value of 310 kJ/mol. The
likely gas-film diffusion at the start and product-layer variation may be caused by the varying relative
diffusion at high conversions. The decomposition significance of the diffusion process under different
process was, therefore, probably a case of mixed reaction conditions, as in the case of Boyabat et al. [25].
control by both chemical reaction and the diffusion. It may also be caused by a shift of the controlling reaction
This also explains the relatively low activation energies step in a multi-step reaction mechanism, as suggested by
that they obtained. Hong and Fegley [20] and Lambert et al. [10].
The influence of the diffusion observed by Boyabat Hong and Fegley [20] explained the variation of the
et al. [25] also indicates that the apparent reaction order observed activation energies by suggesting a two-step
of the decomposition reaction is non-zero with respect mechanism for the reaction:
to the sulfur gas under conditions, in which the reaction
is significantly influenced by the diffusion. Otherwise, (a) Pyrite (FeS2) decomposes to form pyrrhotite
the diffusion is not supposed to influence the reaction. (FeSx) and liquid sulfur atoms.
However, under chemical reaction control, the reaction (b) Liquid sulphur atoms combine and evaporate to
was observed by Coats and Bright [24], Hoare et al. form S2 (g).
[21], Fegley and Lodders [11], Hong and Fegley [20]
and Lambert et al. [10] to be zero-order. The The process can thus be controlled by step (a) (in this
discrepancy can be explained by the reversibility of case, pyrite is decomposed to form sulfur gas directly)
the decomposition reaction of pyrite. As discussed when the dissipation of sulfur from the surface is fast.
earlier, the decomposition reaction of pyrite is The activation energy is supposed to be the observed
reversible. The decomposition reaction can, therefore, value of approximately 297 kJ/mol. The process can
be influenced by the sulfur gas pressure due to its effect also be controlled by step (b), the evaporation of sulfur
on the reverse reaction. However, in the case of from the pyrite/pyrrhotite interface, in cases in which
chemical reaction control, the diffusion resistance for dissipation of sulfur from the surface is slow. The
the dissipation of the formed sulfur gas must be activation energy for this step is approximately
insignificant. The influence of the reverse reaction 112 kJ/mol.
will thus be negligible due to the insignificant sulfur gas Lambert et al. [10] explained the variation of the
partial pressure at the reaction front. The observed apparent activation energies that they obtained by
apparent reaction order of zero is, therefore, the suggesting a four-step mechanism for the decompo-
reaction order of the forward reaction. sition of pyrite:
‘Heat transport control’, as suggested by Boyabat et
al. [25], seems to be an ambiguous concept. As (a) Pyrite decomposes to pyrrhotite and sulfur atoms,
discussed earlier, for an endothermic reaction, the with an activation energy of around 30 kJ/mol.
significant influence of heat transport will cause the (b) Sulfur atoms diffuse through the pyrrhotite layer to
particle temperature to be significantly lower than that the surface of the particle or the interface of the
of the main gas phase. However, there is not a natural two pyrrhotite phases, with an activation energy of
limit of this temperature drop by which the controlling around 90 kJ/mol.
302 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

Table 1
Apparent activation energies obtained by different authors for the thermal decomposition of pyrite

Author Temperature Atmosphere Particle size Apparent activation Suggested control-

energy (kJ/mol) ling mechanism
Udintseva and 673–1023 K Vacuum 110
Chuforov [26]a
Schwab and Philinis 873–923 K CO2 0.01–0.1 mm 121–138 Chemical reaction
[23] (crushed, irregular
particles)
Pannetier and 724–749 K Vacuum 310
Davignon [27]a
Zhukovskii et al. 723–963 K N2, vacuum 110
[28]a
Samal [29]a 759–827 K Vacuum 120
Coats and Bright 873–926 K Ar :6.25!6.25 mm 281 Chemical reaction
[24] (cylindrical form)
Hoare et al. [21] (TGA with heating Nitrogen Irregular single par- 286 Chemical reaction
rate 3 K/min.) ticle of 0.5–10 mg
(estimated to be 0.6–
0.8 mm in diameter)
Fegley and Lodders 663–804 K CO2, Ar 10–20!10!1 mm 150 Chemical reaction
[11] (slice form)
Hong and Fegley 673–863 K He, N2, CO2 10–20!10!0. 297 Chemical reaction
[20] 5 mm (slice form)
Lambert et al. [10] 626–973 K Vacuum, N2, Ar He 0.21–0.25 mm 200 (vacuum and Chemical reaction
(crushed, irregular 0.1 MPa)
particles) 275–294
(0.8 MPa)
Boyabat et al. [25] 673–1073 K N2 1.3–0.46 mm 96–113 Heat transport at
(crushed, irregular the start
particles) Product layer
diffusion at high
conversions
a
References in Hong et al. [20].

(c) Sulfur atoms combine to form sulfur gas mol- quickly. This can cause a significant temperature
ecules at the surface. difference between the main gas phase and the unreacted
(d) Desorption of S2 from the surface, with an core if the conditions for heat transport are not optimal.
activation energy of around 200 kJ/mol. For example, by using data from Boyabat et al. [25], it
can be calculated that the temperature of the pyrite core
The process was assumed to be controlled by step (a) may be about 19 K lower than that of the main gas phase
at the start of the reaction because the pyrrhotite layer is at the start of the decomposition process for a pyrite
not yet formed, but by step (d) after the pyrrhotite layer particle of the size of 1.3 mm in a nitrogen atmosphere of
has been formed. 923 K (the reaction heat (DHr) of 284.5 kJ/mol S2,
As well as the possible influence of diffusion and the calculated by Hong and Fegley [20], is used for the
shift of the supposed different reaction steps, heat calculation). 19 K is significant for the decomposition of
transport is another important factor that can affect the pyrite at such a temperature level. It can be expected that
determination of the apparent activation energies. this temperature difference will be much larger at higher
According to Eq. (5), a higher reaction rate will cause temperatures. The reaction will, therefore, be observed
a significant temperature difference between the main to proceed more slowly than expected due to the fact that
gas phase and the unreacted core. At a low temperature, the reaction is actually proceeding at a lower tempera-
the influence of heat transport may be limited, as the ture than that measured in the main gas phase. When the
reaction proceeds relatively slowly and the temperature activation energy of the reaction is determined, a lower
difference between the main gas phase and the unreacted value will be obtained. It can, therefore, be expected that
core will be small. At a high temperature (for example O the activation energy obtained at a lower temperature
873 K), the decomposition of pyrite will proceed will be higher than that obtained in a wide temperature
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 303

range and at higher temperatures if the effect of the heat Dunn et al. [31,32] are good illustrations of these two
transport is not accounted for. This may be part of the transformation processes.
reasons for the relatively low activation energies of Jorgensen and Moyle [30] performed pyrite oxi-
around 100 kJ/mol that were obtained by Udintseva and dation in air with simultaneous differential thermal
Chuforov [26], Zhukovskii et al. [28] and Boyabat et al. analysis (DTA) and TGA at a heating rate of 10 K/min.
[25], as listed in Table 1. It needs to be noted that pyrite The pyrite particles were 0.053–0.074 mm in diameter.
has an incongruent melting point of 1016 K. The true They observed by X-ray diffraction and SEM examin-
temperature of the pyrite must be much lower than the ation that the pyrite was directly oxidized to form
main gas temperature when the activation energy is hematite at a temperature lower than about 803 K. At a
determined beyond 1016 K, as in the cases of Udintseva higher temperature, pyrrhotite was observed to be
and Chuforov [26] and Boyabat et al. [25]. For those formed as an intermediate, which was successively
cases with activation energies of around 300 kJ/mol, as oxidized to form hematite. The SEM pictures in Fig. 5
listed in Table 1, the influence of heat transport is show this process.
probably limited. Fig. 5(a) shows a partially reacted particle with an
unreacted pyrite core (PY), a porous pyrrhotite layer
3. Pyrite transformation in an oxygen-containing (PO) in the middle and a rim of hematite (H) surrounding
atmosphere the particle. The texture of the pyrrhotite is similar to that
observed by Hong and Fegley [20] under inert
3.1. Transformation processes and products conditions (Fig. 4). Fig. 5(b) shows a completely
oxidized pyrite particle with two parts of different
In an oxygen-containing atmosphere, pyrite will be textures: the rim surrounding the particle from direct
oxidized to form a series of final products, such as oxidation of the pyrite at the earlier heating stage; and
hematite (Fe2O3), magnetite (Fe3O4), iron (ferric or the inner part from the oxidation of the pyrrhotite, which
ferrous) sulfate (Fe2(SO4)3, FeSO4) and sulfur dioxide has maintained the texture of the pyrrhotite that was
(SO2). The transformation process and the formation of formed from the thermal decomposition of the pyrite.
these products are influenced by the reaction con- Dunn et al. [31,32] studied the oxidation of pyrite at
ditions, such as temperature, oxygen concentration, different oxygen concentrations with simultaneous DTA
flow conditions and particle size. and TGA at different heating rates. The reaction products
were analyzed by X-ray diffraction. The experiments
3.1.1. Transformation processes were performed with two different particle sizes. With the
It was observed by a number of authors that the smaller particle size (!0.045 mm), it was observed that
transformation of pyrite in an oxygen-containing pyrite was directly oxidized to form hematite at a low
atmosphere can take place in two ways, depending on heating rate (2.5 K/min) in air atmosphere. The oxidation
the reaction conditions. One is the direct oxidation of was completed at a temperature of around 776 K. With
pyrite, and the other is a two-step process that involves the larger particle size (0.09–0.12 mm) and at a heating
thermal decomposition as in an inert atmosphere and rate of 2.5 K/min in air, they observed that pyrite was
then successive oxidation of the formed pyrrhotite. The directly oxidized to form hematite at a temperature lower
observations made by Jorgensen and Moyle [30] and than about 788 K. At a higher temperature, a porous oxide

Fig. 5. SEM pictures of pyrite particles oxidized in air (Jorgensen and Moyle [30]) (Py, pyrite; Po, pyrrhotite; H, hematite).
304 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

Fig. 6. SEM pictures of oxidized pyrite particles (particle size about 0.09–0.125 mm) (modified from Dunn et al. [32]).

layer was observed to be formed in the inner part of the low heating rate (2–6 K/min) up to a temperature of
particles. A similar phenomenon was observed with small 753 K. The phase change was monitored by a high
particles (!0.045 mm) that were heated at a high heating temperature diffractometer. Pyrrhotite was not observed
rate (40 K/min). Fig. 6 shows SEM pictures of cross- during the transformation process. They only observed
sections of the reacted particles. the formation of hematite by the direct oxidation of the
Fig. 6(a) shows partially reacted particles with an pyrite.
oxide rim formed at the lower temperatures (! about Prasad et al. [34] studied the transformation of pyrite
788 K), a porous oxide layer in the middle formed at a by roasting pyrite powder in a small porcelain boat in air
temperature higher than about 788 K and an unreacted at 883 K. It was observed that pyrite first decomposed to
core of the pyrite. The porous oxide layer in the inner form pyrrhotite. The formed pyrrhotite was then oxidized
part was supposed to be formed by the oxidation of the to form oxides. Their analysis showed the existence of
porous pyrrhotite that was first formed as an intermedi- pyrite, pyrrhotite and hematite/magnetite at the later stage
ate by the thermal decomposition of the pyrite at of the roasting process. This is an indication of the
temperatures higher than about 788 K. Fig. 6(b) shows occurrence of simultaneous thermal decomposition of the
a particle after completed oxidation that consists of pyrite and the oxidation of the formed pyrrhotite.
mainly hematite with two different textures, as in the Hong and Fegley [20] studied pyrite transformation in
observations by Jorgensen and Moyle [30]. a tube furnace in a CO2 atmosphere containing
As observed by Jorgensen and Moyle [30] and Dunn et 100–1009 ppmv O2 in the temperature interval from
al. [31,32], the texture of the product layer that was 665 to 811 K and with a single pyrite particle. The particle
formed by direct oxidation of the pyrite is different from was in slice form and with a dimension of 10–20 mm!
that formed by the successive oxidation of the pyrrhotite 10 mm!0.5 mm. They observed that in 100 and
formed by the thermal decomposition of the pyrite. 1009 ppm O2 atmospheres and at temperatures from
During the direct oxidation of the pyrite, sulfates were 757 to 811 K, only pyrrhotite was formed. The activation
observed to be formed [30–32]. The formation of sulfates energy obtained under these conditions was almost the
makes the product layer less porous due to the large molar same as that obtained in inert atmospheres. At lower
volume of the sulfates (there will be more detailed temperatures (from 665 to 733 K), both pyrrhotite and
discussion about this later). However, the fact that the hematite were observed, which indicates simultaneous
texture difference remains after the sulfates were thermal decomposition of the pyrite and oxidation of the
decomposed [30–32] indicates that the porous appearance formed pyrrhotite.
of the inner part is probably related to the special Hansen [35] studied the transformation of pyrite in an
morphology of the pyrrhotite phase. The dense appear- entrained flow reactor in an atmosphere containing 5%
ance of the rim formed by the direct oxidation of the pyrite oxygen and at 798 K. The particle size was 0.032–
may not be related to the formation of sulfates. 0.064 mm. The residence time of the particles in the
These two different ways of oxidation of pyrite reactor was about 3 s. Fig. 7 shows a partially reacted
under different reaction conditions are also observed by particle with an unreacted pyrite core, a porous pyrrhotite
a number of other authors. layer (determined by energy dispersive X-ray micro-
Schorr and Everhart [33] studied the oxidation of analysis (EDAX)) and a rim of hematite at the particle
0.125–0.5 mm pyrite particles in air in a furnace with a surface.
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 305

In situation B, oxygen has contact with the particle,

but is not in contact with the pyrite core. The pyrite
undergoes thermal decomposition as in an inert
atmosphere, but part of the formed pyrrhotite is
oxidized alongside the gas phase oxidation of sulfur
to SO2. The observations made by Prasad et al. [34],
Dunn et al. [31,32], Hong and Fegley [20] and Hansen
[35] may be such cases. This situation can occur when
the reaction temperature is not very high and the
oxygen concentration is relatively high.
In situation C, oxygen is in direct contact with the
pyrite core. The pyrite is now oxidized directly. This
Fig. 7. SEM picture of a partially reacted pyrite particle (particle size
situation can occur when the reaction temperature is low
about 0.06 mm) (modified from Hansen [35]). and the oxygen concentration is high. The hematite layer
that is formed at lower temperatures in the TGA
This is clear evidence for the thermal decomposition, experiments by Jorgensen and Moyle [30] and Dunn et
instead of the direct oxidation, of the pyrite. The hematite al. [31,32], and the observation made by Schorr and
rim at the surface of the particle may be formed by direct Everhart [33], are probably due to this process. The direct
oxidation during the heating-up stage, but may also be oxidation could, in practice, occur to a high or low degree
formed by the simultaneous oxidation of the pyrrhotite. in situations A and B during the earlier period of the
The above observations by the different authors heating-up. Slow heating-up can cause significant
show that the way by which pyrite is transformed in an oxidation of the pyrite. During the direct oxidation,
oxygen-containing atmosphere is influenced by par- small amounts of iron sulfates may be formed. This will
ameters such as temperature, particle size, oxygen result in a denser product layer since the molar volumes of
concentration, heating rate and flow conditions. As iron sulfates are much higher than those of iron oxides
shown in Fig. 8, three different situations can be (this will be discussed in detail later). This dense layer will
generalized from the above observations: restrict the inward diffusion of the oxygen and the
In situation A, oxygen has no contact with the outward diffusion of the sulfur gas, and then influence the
particle. Pyrite undergoes thermal decomposition, as in overall transformation process. With fast heating-up, this
an inert atmosphere, as observed by Hong and Fegley direct oxidation will be greatly limited.
[20]. This situation can occur when the oxidation of the Which kind of situation occurs in practice will
sulfur gas is able to consume all oxygen during its depend on the relative rates of the transport of oxygen
diffusion to the particle surface. This is normally the to and into the particle and the thermal decomposition
case when the reaction temperature is high and/or the of pyrite under given conditions (the oxidation of
oxygen concentration is low. The pyrrhotite will not be sulfur gas can be assumed to be infinitively fast).
oxidized until the pyrite is fully decomposed. Thermal decomposition of pyrite will take place if the

Fig. 8. Illustration of pyrite transformation in an oxygen-containing atmosphere.

306 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

rate of the oxygen transport through the gas film and

the product layer toward the interface of the unreacted
pyrite core is slower than what is needed for the
oxidation of the sulfur gas generated by the
decomposition of the pyrite. In this case, the oxygen
will be consumed before it reaches the interface of the
unreacted pyrite core. Direct oxidation of pyrite will
take place if the rate of oxygen transport through the
gas film and the product layer is faster than that
needed for the oxidation of the released sulfur gas. In
this case, the oxygen is able to reach the interface of
the unreacted pyrite core.
The thermal decomposition of the pyrite will be the
dominant process at a high temperature due to the high
Fig. 9. Molten and partially oxidized pyrrhotite particles (the large
activation energy—for example, when the temperature particle is approximately 0.05 mm in diameter) (modified from
is higher than about 800 K [30–32,35]. At a much Hansen [35]).
higher temperature—for example, higher than about
923 K—the high thermal decomposition rate of the
pyrite will cause a significant temperature difference hematite can exist as the stable iron oxide up to about
between the main gas phase and the pyrite particle, as 1730 K.
discussed earlier. A number of authors [36–40] Investigations by a number of authors showed that
observed that the pyrite particles remained solid until the final oxide formed from pyrite/pyrrhotite oxidation
they were fully decomposed in combustion environ- was hematite at the highest temperature of up to about
ments with temperatures up to 1700 K. This means that 1473 K [20,23,30–33,35,39,46–53] and magnetite at a
the pyrite particles never exceeded their incongruent temperature higher than about 1773 K [38,54]. This is
melting point of 1016 K even at such high gas in good accordance with Fig. 10.
temperatures. Therefore, the melting of pyrite will, However, there are also investigations [34,36,40,
practically, normally not take place in an oxygen- 54–60] that showed that magnetite was present with
containing atmosphere, because the high decompo- hematite and, in some cases, magnetite was even the
sition rate at temperatures close to the melting point of dominant oxide under the conditions where hematite
pyrite and the large endothermic enthalpy change of
this process will keep the core temperature below the
melting point.
The successive oxidation of the formed pyrrhotite in
the case of situations A and B can be vigorous at high
O2 concentrations and at high temperatures [41–43].
Melting of the pyrrhotite can take place even though the
main gas temperature is lower than pyrrhotite’s melting
point of around 1261–1460 K [44]. This phenomenon
was observed by Hansen [35] in the oxidation of pyrite
that was performed in an entrained flow reactor at
798 K in a 21% oxygen atmosphere. Fig. 9 shows the
molten and partially oxidized pyrrhotite particles from
his experiment.

3.1.2. The formation of iron oxides

Iron oxides are the main final products of pyrite
oxidation. According to the phase diagram of the Fe–O2
system presented by Darken and Gurry [45] (Fig. 10),
hematite will be the stable iron oxide up to about Fig. 10. Phase equilibrium between iron oxides and oxygen (1 atmZ
1600 K in an atmosphere containing a low percentage 0.101 MPa) (modified from Darken and Gurry [45]).
of oxygen. In a pure oxygen atmosphere (0.101 MPa),
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 307

should be the stable oxide, according to Fig. 10. dominant oxide (O85%). Similar to the observation
Wustite was observed to form as an intermediate in a by Huffman et al. [57], it was observed that the fraction
few cases [39,40,57]. of hematite was increased with an increase of residence
Nishihara and Kondo [55] studied the oxidation of time in the furnace.
pyrite by thermomagnetic analysis at temperatures ten Brink et al. [60] investigated the slag formed
from 973 to 1173 K in atmospheres containing 1–5% during the combustion of pyrite rich coal in a power
oxygen. It was observed that hematite was the final plant. The gas temperature nearby the superheater was
oxidation product, but with magnetite as an intermedi- around 1473 K. Both hematite and magnetite were
ate. It was also observed that an increase in the oxygen identified in the formed slag.
concentration caused a decrease of the magnetite McLennan et al. [40] performed pyrite oxidation
proportion in the oxidation process. The conversion experiments in a drop-tube reactor at temperatures from
of the formed magnetite to hematite was observed to 1573 to 1873 K. It was observed that all three iron
take place at the end of the oxidation reaction and oxides—wustite, magnetite and hematite—were pre-
proceeded quickly. sent in the oxidation products. Pyrrhotite was first
Hubbard and McGill [54] analyzed, using SEM, the oxidized to form wustite, which was then oxidized to
product from pulverized coal combustion at 1773 K. magnetite and further to hematite. In the case of
Both hematite and magnetite were identified. reducing atmospheres, it was observed that only wustite
Prasad et al. [34] roasted pyrite particles in air at and magnetite were present in the oxidation products.
883 K. It was observed that hematite was the final The discrepancy between these experimental obser-
oxidation product. A small amount of magnetite was vations and the equilibrium state can probably be
observed before the completion of the oxidation reaction. explained by the fact that, in practical situations, there
Groves et al. [56] performed pyrite oxidation may be oxygen deficiency at the reaction front, which is
experiments in a drop-tube furnace in an atmosphere caused by the relatively fast reaction rate of the
containing 2% oxygen. They observed that hematite oxidation of pyrite/pyrrhotite and the slow diffusion
was the dominant oxide at temperatures lower than rate of oxygen under certain reaction conditions. The
about 1073 K. Magnetite began to form at a tempera- deficiency of oxygen at the reaction front favors the
ture higher than about 1073 K. This point moved to formation of magnetite and, in some cases, even
about 1223 K in an air atmosphere. wustite. The observations of the fast conversion of
Huffman et al. [57] performed pyrite oxidation magnetite to hematite at the end of the pyrite/pyrrhotite
experiments in a drop-tube furnace in an atmosphere oxidation, and the increase of hematite with the
containing 5% oxygen at 1311–1727 K. They observed increase of the oxygen concentration by the different
that magnetite was the dominant oxide in the whole authors, are good evidence of this.
temperature interval. It was observed that the fraction Thus, the formation of iron oxides (hematite or
of hematite was increased with an increase of the magnetite/wustite) is determined by the oxygen
residence time (in the level of milliseconds) in concentration at the reaction front and not by the
the furnace. This increase was most significant at the oxygen concentration in the main gas phase. Par-
lowest temperature tested (1311 K). ameters such as temperature, flow condition, particle
Helble et al. [58] investigated the oxidation products size and residence time that can affect the oxygen
of pyrite during pulverized coal combustion at 1500 K. concentration at the reaction front will all affect the
It was observed that both magnetite and hematite were formation of iron oxides. In practice, as shown above,
present. Magnetite was the dominant oxide. An the formation of magnetite as the main phase will take
increase in the oxygen concentration increased the place at temperatures higher than about 1173–1273 K
proportion of hematite. under conditions of sufficient oxygen supply in the
Komraus et al. [59] studied the transformation of main gas phase. This is often related to the oxidation of
pyrite contained in coal during coal oxidation exper- pyrrhotite. The formation of magnetite during the direct
iments in air. Hematite was identified as the oxidation oxidation of pyrite is expected to be limited, as the
product at temperatures from 573 to 1030 K. Magnetite oxidation of pyrite can only take place at temperatures
was found to be present as a minor component. lower than about 800 K.
Srinivasachar et al. [36] performed pyrite oxidation
experiments in an entrained flow reactor in an 3.1.3. The formation of sulfates
atmosphere containing 3% oxygen at 1500 K. The In the oxidation process of pyrite/pyrrhotite, sulfates
obtained results showed that magnetite was the (mainly ferrous sulfate (FeSO4) and ferric sulfate
308 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

Table 2 Allen et al. [63] studied the oxidation of pyrite in air

Molar volume and molar weight of different substances in a tube furnace at temperatures up to 523 K. Ferrous
Substance Molar volume Molar weight sulfate was determined by X-ray photoelectron spec-
(cm3/(mole Fe)) (g/(mole Fe)) troscopy to be the dominant phase formed.
FeS 18.2 [61, p. 2–15] 87.9 Eymery [64] observed the formation of ferrous
FeS2 24.0 [61, p. 2–16] 119.9 sulfate monohydrate (FeSO4$H2O) from pyrite oxi-
Fe2O3 15.6 [61, p. 2–14] 79.9 dation in air after it had been intensively ground. The
Fe3O4 14.8 [61, p. 2–14] 77.2
sulfate phase was determined by X-ray diffraction and
FeO 12.6 [61, p. 2–14] 71.9
Fe2(SO4)3 64.6 [61, p. 2–14] 199.9 Mössbauer spectrometry.
FeSO4 41.6 [62, p. 4–63] 151.9 Eneroth et al. [52] studied the oxidation of pyrite in
FeSO4$H2O 56.7 [62, p. 4–63] 170 air at 748–773 K in an oven. The formation of a small
amount of ferrous sulfate was determined by Möss-
bauer spectrometry.
(Fe2(SO4)3)) can be formed as minor products, as
The above observations by the different authors
observed by a number of authors. The formation of
show that both ferrous and ferric sulfates can be formed
sulfates will result in mass gain and volume increase as
during the oxidation of pyrite. The evidence for the
the molar weights and molar volumes of both sulfates
formation of ferrous sulfate is more pronounced. At
are higher than those of pyrite and iron oxides (as
lower temperatures (!500 K), hydrated ferrous sulfate
shown in Table 2). can be formed.
As shown in Table 2, the mass increase per mole For a better understanding of the observations made
oxidized pyrite will be 26.7% for the formation of by these authors, it is helpful to look at the thermal
ferrous sulfate and 66.7% for the formation of ferric stability of these two sulfates. Ferrous and ferric
sulfate. The formation of ferrous and ferric sulfate will sulfates will decompose according to the following
cause an increase of volume of 73 and 169%, overall reactions when heated [65,66]:
respectively.
Schwab and Philinis [23] studied the oxidation of 2FeSO4 ðsÞ4 Fe2 O3 ðsÞ C SO3 ðgÞ C SO2 ðgÞ (6)
pyrite by the thermogravimetric method in air at
673–773 K. The formation of ferric sulfate was SO3 ðgÞ4 SO2 ðgÞ C 0:5O2 ðgÞ (7)
identified.
Jorgensen and Moyle [30] observed the arrest of the Fe2 ðSO4 Þ3 ðsÞ4 Fe2 O3 ðsÞ C 3SO3 ðgÞ (8)
TGA curve (no visible weight loss) from about 803 to
873 K in their study of pyrite oxidation. The authors SO3 ðgÞ4 SO2 ðgÞ C 0:5O2 ðgÞ (9)
believed that it was caused by the formation of sulfate.
The further decomposition of SO3 to SO2 and O2 at
It was not identified which sulfate it was.
temperatures lower than about 923 K is limited due to
Cole et al. [47] observed the formation of ferrous
the thermodynamic equilibrium of this reaction [65,67].
sulfate monohydrate from pyrite oxidation at tempera-
Fig. 11 shows the total equilibrium gas pressures by the
tures of O523 K during their coal oxidation exper-
decomposition of ferrous sulfate and ferric sulfate. The
iments. The formed ferrous sulfate monohydrate was
converted to hematite, with ferric hydroxysulfate
(Fe(OH)SO4) and ferric oxysulfate (Fe2O(SO4)2) as
intermediates, at a temperature of around 673 K.
Komraus et al. [59] studied the transformation of
pyrite contained in coal during coal-oxidation exper-
iments in air in an oven at temperatures from 570 to
1030 K. It was observed that ferrous sulfate was formed
at temperatures up to around 590 K.
Dunn et al. [49,50] studied the formation of sulfates
with Fourier transform infrared spectroscopy. The pyrite
sample was roasted in air at different temperatures. They
observed that ferrous sulfate was the dominant sulfate Fig. 11. Equilibrium decomposition pressures of ferric sulfate and
formed at temperatures up to about 873 K. Ferric sulfate ferrous sulfate (1 atmZ0.101 MPa) (data from Greulich [65] and
was observed to be present as a minor phase. Warner and Ingraham [66]).
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 309

figure is constructed by using data from Greulich [65] for SO2 ðgÞ C 0:5O2 ðgÞ4 SO3 ðgÞ (11)
the decomposition of ferrous sulfate and Warner and
Ingraham [66] for the decomposition of ferric sulfate.
The existence of these two sulfates depends on the † Formation and decomposition of sulfates:
surrounding gas composition and temperature. It may be
incorrect to say that ferric sulfate is more stable than 2FeS2 ðsÞ C 7O2 ðgÞ/ Fe2 ðSO4 Þ3 ðsÞ C SO2 ðgÞ
ferrous sulfate during the oxidation of pyrite, just based on (12)
the lower equilibrium decomposition pressure of ferric
sulfate, as the gases in balance with these two sulfates are
different. The gas in balance with ferric sulfate is SO3, FeS2 ðsÞ C 3O2 ðgÞ/ FeSO4 ðsÞ C SO2 ðgÞ (13)
whereas the gases in balance with ferrous sulfate are SO2
and SO3. It is well known that SO2 is normally the
2FeSO4 ðsÞ4 Fe2 O3 ðsÞ C SO3 ðgÞ C SO2 ðgÞ (14)
dominant gaseous species from the oxidation/combustion
of diverse sulfur-containing substances, including pyrite,
and that the establishment of equilibrium between SO2 Fe2 ðSO4 Þ3 ðsÞ4 Fe2 O3 ðsÞ C 3SO3 ðgÞ (15)
and SO3 is slow at low temperatures without a catalyst
[67]. Ferric sulfate can be less stable than ferrous sulfate
during the oxidation of pyrite if the formed SO2 in the gas SO2 ðgÞ C 0:5O2 ðgÞ4 SO3 ðgÞ (16)
is not readily converted to SO3. This may be the In the case of thermal decomposition and successive
explanation for the many observations of the formation oxidation:
of ferrous sulfate, especially the observation of the
† Thermal decomposition of pyrite and oxidation of
existence of ferrous sulfate up to about 873 K that was
the formed sulfur gas:
made by Dunn et al. [49,50].
Both sulfates decompose at temperatures higher than
about 823–923 K, as observed by Jorgensen and Moyle FeS2 ðsÞ/ FeSx ðsÞ C ð1K0:5xÞS2 ðgÞ (17)
[30], Dunn et al. [31,32,49,50], as the partial pressures of
SO2 and SO3 under the normal pyrite oxidation conditions
are much lower than the equilibrium pressures. S2 ðgÞ C 2O2 ðgÞ/ 2SO2 ðgÞ (18)
Thus, during the oxidation of pyrite, the temperature
and the gas composition close to the solid surface are SO2 ðgÞ C 0:5O2 ðgÞ4 SO3 ðgÞ (19)
probably the two most important parameters that
determine whether sulfate can be formed and which
sulfate will be formed. The relatively high SO2 † Oxidation of pyrrhotite at temperatures lower than
concentration in the gas phase under normal reaction about 1173–1273 K and possible formation and
conditions seems to favor the formation of ferrous decomposition of sulfates at temperatures up to
sulfate over the formation of ferric sulfate. The about 923 K:
formation of ferric sulfate may, to a high degree,
depends on whether there are conditions for the ready 2FeSx ðsÞCð1:5 C2xÞO2 ðgÞ/Fe2 O3 ðsÞC2xSO2 ðgÞ
conversion of SO2 to SO3.
As shown above, the transformation of pyrite in an (20)
oxygen-containing atmosphere is a complicated pro-
cess, which involves many reactions and products. By 2FeSx ðsÞCð3 C2xÞO2 ðgÞ/Fe2 ðSO4 Þ3 ðsÞ
considering only the main reactions and the related final
products, the transformation of pyrite in an oxygen- Cð2xK3ÞSO2 ðgÞ (21)
containing atmosphere may be generalized and rep-
resented by the following overall reactions:
FeSx ðsÞCð1 CxÞO2 ðgÞ/FeSO4 ðsÞCðxK1ÞSO2 ðgÞ
In the case of direct oxidation: (22)
† Oxidation of pyrite:

2FeS2 ðsÞ C 5:5O2 ðgÞ/ Fe2 O3 ðsÞ C 4SO2 ðgÞ (10) 2FeSO4 ðsÞ4Fe2 O3 ðsÞ CSO3 ðgÞ CSO2 ðgÞ (23)
310 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

Fe2 ðSO4 Þ3 ðsÞ4Fe2 O3 ðsÞC3SO3 ðgÞ (24)

SO2 ðgÞC0:5O2 ðgÞ4SO3 ðgÞ (25)

† Oxidation of pyrrhotite at temperatures higher than

about 1173–1273 K:

3FeSx ðs=lÞCð2 C3xÞO2 ðgÞ/Fe3 O4 ðs=lÞ

Fig. 12. The calculated porosity of the product layer of an oxidized
C3xSO2 ðgÞ (26)
pyrite particle, with the fraction of FeS2 being converted to Fe2(SO4)3
as a variable (the rest is assumed to be Fe2O3).

t
Z 1K3ð1KXB Þ2=3 C2ð1KXB Þ (29)
2Fe3 O4 ðsÞC0:5O2 ðgÞ/3Fe2 O3 ðsÞ (27) t
Sulfates have a much larger molar volume than iron
oxides (as shown in Table 2). The formation of sulfates
3.2. Kinetics of pyrite oxidation during the oxidation of pyrite will therefore partially or
totally block the pores in the product layer, depending on
As shown earlier, the transformation of pyrite in an the fraction of the formed sulfate. Fig. 12 shows the
oxygen-containing atmosphere can proceed by direct influence of the formation of ferric sulfate on the porosity
oxidation or by a two-step process: first, thermal of the product layer by assuming that the product layer
decomposition of the pyrite and, second, successive consists of only Fe2O3 and Fe2(SO4)3. As shown in the
oxidation of the formed pyrrhotite. The reaction kinetics figure, the pores will be totally blocked if about 18% (by
of these different transformation processes are different. mass) of the oxidized pyrite is converted to ferric sulfate.
Schwab and Philinis [23] observed that the
3.2.1. Direct oxidation formation of ferric sulfate caused the gradual blocking
In the case of direct oxidation, the reaction follows of the pores and, eventually, the total cessation of the
the unreacted core model and may be controlled by the oxidation process. It was observed that the achieved
reaction and the diffusion of oxygen. final conversions were lower at lower temperatures than
Schwab and Philinis [23] performed pyrite oxidation at higher temperatures due to the earlier pore-blocking
experiments by the thermogravimetric method in air at at lower temperatures.
673–773 K. The oxidation process was probably direct Hansen et al. [53] performed oxidation experiments
oxidation of the pyrite, as judged by the reaction with shale particles containing pyrite in a fixed-bed
conditions (in air and at a temperature lower than reactor. It was observed that the transformation process
783 K). It was concluded that the process was controlled proceeded in a two-stage pattern: a fast initial stage and
by the diffusion of oxygen through the product layer, a slow later stage. It was also observed that lower
based on the observed insensitivity of the reaction rate to temperatures and higher oxygen concentrations caused
the change of temperature and the low apparent activation a lower final conversion, as shown in Fig. 13(a)–(c).
energy of the reaction process. The following empirical The following empirical equation was proposed to
equation was proposed to fit the experimental data: model the oxidation process of the pyrite contained in
shale particles. The model consists of two parts. The
t first-term on the right represents a fast process. The
1Kð1KXB Þ1=2 Z (28) second-term on the right represents a slow process
t
dXB
The gradual pore-blocking effect that was caused by Z fkr;1 ð1KXB;1 ÞCð1Kf Þkr;2 ð1KXB;2 Þ (30)
dt
the formation offerric sulfate was thought to be part of the
reason why the following theoretical equation for The observations were principally the same as the
spherical particles [22] did not fit the experimental data: observations made by Schwab and Philinis [23]. A
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 311

For the second step—that is, the oxidation of

pyrrhotite—there are two different situations: one
with pyrrhotite in a solid state at lower temperatures
(!1261–1460 K); and another with pyrrhotite in a
molten state at higher temperatures (O1261–1460 K).
Kinetic data and models related to this process are
limited in the literature.
Srinivasachar and Boni [68] proposed the following
seven-stage model for the transformation of pyrite in a
combustion environment:

Fig. 13. Influence of oxygen concentration and temperature on the (1) Heating up of pyrite to 870 K.
conversion of pyrite contained in shale particles (modified from (2) Decomposition of pyrite to pyrrhotite. This stage is
Hansen et al. [53]). assumed to be controlled by heat transport.
(3) Oxidation and heating up of solid pyrrhotite. The
oxidation of solid pyrrhotite is assumed to be a
reasonable explanation could be that direct oxidation of surface reaction with first-order dependence on the
pyrite had occurred, and the formation of sulfate blocked oxygen concentration.
the pores, which then caused the observed phenomena (4) Melting of pyrrhotite. The rate of melting is
(although this was not pointed out by the authors). assumed to be determined by heat generated by
the oxidation reaction and heat exchanged with the
3.2.2. The two-step process surrounding environment.
In the case of the two-step process, the kinetic (5) Oxidation of molten pyrrhotite. This stage is
models for the thermal decomposition of pyrite in an assumed to be controlled by the diffusion of
inert atmosphere should be applicable to the first step. oxygen through gas and liquid film.
However, compared to the same reaction in an inert (6) Cooling of the particle and crystallization of
atmosphere, the thermal decomposition of pyrite in an magnetite.
oxygen-containing atmosphere may proceed much (7) Oxidation of solid magnetite to hematite. This
faster due to the extra heating effect from the oxidation stage is assumed to be controlled by chemical
of the formed sulfur gas and the reduction of eventual reaction with first-order dependence on the oxygen
diffusion resistance of the outgoing sulfur gas through concentration.
the product layer and/or the gas film. This effect must
be significant in the case of the pyrite oxidation In this model, two general equations were used. One
experiments performed by Hansen [35] in an entrained is an equation for the energy balance accounting for the
flow reactor. The conversion reached approximately heating up of the particle, heat released/needed by the
60–80% under the conditions of 823 K, with approxi- reaction, heat transferred to the particle by convection
mately 3 s of residence time and with a particle size of and heat transferred to the particle by radiation. The
around 0.03–0.06 mm. The reaction is about two orders second is an overall rate equation for surface reaction
of magnitude faster than the calculated rate, as assessed based on external particle surface area. It is assumed
by using the following equation for reaction control for that the intrinsic oxidation reaction is of first-order
spherical particles [22] and the rate constants obtained dependence on the oxygen concentration.
by Hong and Fegley [20] and Lambert et al. [10] The oxidation of pyrrhotite in a molten state was
thought to be controlled by the diffusion of oxygen
t
Z 1Kð1KXB Þ1=3 (31) through gas and liquid film in this model, due to the
t relatively fast oxidation reaction at such high tempera-
This may be caused by the possibly under-estimated tures. ten Brink et al. [38] used a similar model for the
rate constants for the thermal decomposition of pyrite transformation of pyrite in a flame environment. It was
in an inert atmosphere (as discussed earlier) or by the observed that the resistance of the liquid boundary layer
erroneous assumption of equal temperatures of the was negligible. This was explained by the so-called
particles and the main gas phase or by a combination of ‘surface renewal mechanism’ with melts of lower
these two. viscosity which improves the mass transport efficiency.
312 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

4. Conclusions compared with the gas phase under these conditions is

also the reason for the existence of pyrite in solid form
An assessment of the mechanisms and a short at temperatures that are much higher than its melting
discussion of the kinetics of pyrite transformation in point.
inert and oxygen-containing atmospheres at elevated In an oxygen-containing atmosphere, pyrite can be
temperatures have been given based on a literature oxidized directly or oxidized after it is first decomposed
review. to pyrrhotite (the two-step process). This is determined
The transformation of pyrite in an inert atmosphere by the relative rates between oxygen diffusion to the
proceeds through a multi-step sequential process pyrite core and the thermal decomposition of the pyrite
(pyrite/pyrrhotite/troilite/iron), depending on depending on the actual conditions, such as tempera-
the temperature and the total sulfur gas pressure in ture, oxygen concentration, flow and particle size.
the system. The direct oxidation of pyrite normally takes place
The equilibrium composition of the primary pyr- at temperatures lower than about 800 K and at higher
rhotite formed by the decomposition of the pyrite is oxygen concentrations, whereas the two-step process
non-stoichiometric and is determined by the tempera- normally takes place under opposite conditions. The
ture. Pyrite has an incongruent melting point of 1026 K, direct oxidation of pyrite proceeds with a shrinking
which is also the maximum temperature for the unreacted core. The oxidation process may be
decomposition of pyrite. controlled by chemical reaction or the inward diffusion
The transformation of the primary pyrrhotite to of oxygen. The unreacted core model with unchanging
troilite proceeds gradually. Pyrrhotite with sulfur product layer structure is not applicable due to the
content between the primary pyrrhotite and troilite gradual pore-blocking effect that results from the
can exist as stable phases under certain conditions formation of ferric or ferrous sulfate.
(sulfur gas pressure and temperature). In the two-step process, the kinetics of the thermal
The sulfur gas formed from the decomposition of decomposition of pyrite in an inert atmosphere is
pyrite consists of different allotropic species. The expected to be applicable for the first step—that is, the
fraction of each allotropic species is influenced by the thermal decomposition of the pyrite. The second step—
temperature and the total sulfur gas pressure. High that is, the oxidation of the pyrrhotite—may proceed
temperatures and low sulfur gas pressures are favorable with pyrrhotite in a solid or a molten state, depending
for the formation of small allotropic species, whereas on the particle temperature. The pyrrhotite will be
low temperatures and high sulfur gas pressures are melted at temperatures higher than about 1261–1460 K.
favorable for the formation of large allotropic species. The oxidation of pyrrhotite in a molten state may be
Under equilibrium conditions (the total sulfur gas controlled by oxygen diffusion through the gas film and
pressure is equal to the equilibrium sulfur gas pressure the liquid boundary layer.
over pyrite), S2 is the dominant species at temperatures Iron oxides (mainly hematite Fe2O3 and magnetite
up to about 900 K. At higher temperatures, the fraction Fe3O4) are the main final products of the oxidation of
of larger allotropic species increases to a significant pyrite. The formation of iron oxides is determined by
level due to the exponential increase of the equilibrium the temperature and the oxygen concentration at the
sulfur gas pressure, which surpasses the effect of the reaction front. Hematite usually forms at lower
temperature increase. temperatures (!1173–1273 K) and at higher oxygen
The thermal decomposition of pyrite to pyrrhotite concentrations, whereas magnetite usually forms at
seems to follow the unreacted core model and may be higher temperatures and/or at lower oxygen
controlled by zero-order surface reaction, gas film/pro- concentrations.
duct layer diffusion, or a combination of these The formation of magnetite at temperatures lower
mechanisms, depending on the reaction conditions. than the equilibrium temperature for the Fe–O2 system
With large particle sizes and/or at high temperatures, is probably due to the oxygen deficiency at the reaction
the particle temperature will be significantly lower than front that is caused by the relatively high reaction rate
the temperature of the main gas phase due to the of the oxidation of pyrite/pyrrhotite and the low
significant resistance of heat transport under these diffusion rate of oxygen under certain reaction
conditions. This will result in a lower reaction rate than conditions.
expected and will also cause an underestimation of the Sulfates (mainly ferrous sulfate FeSO4 and ferric
activation energy if this temperature difference is not sulfate Fe2(SO4)3) may be formed as minor products of
taken into account. The lower particle temperature the oxidation of pyrite at temperatures lower than about
 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314 313

873–923 K. The formation of sulfates is probably [14] Meng L, Liu MS. Mater Sci Eng 1999;B60:168–72.
determined by the gas composition at the reaction [15] Meng L, Liu YH, Huang W. Mater Sci Eng 2002;B90:84–9.
[16] Meng L, Liu YH, Tian L. Mater Res Bull 2003;38:941–8.
front. Ferrous sulfate usually forms in a gas that is rich [17] West James R. Ind Eng Chem 1950;42(4):713–8.
in SO2, whereas ferric sulfate usually forms in a gas that [18] Meyer B. Elemental sulphur. In: Nickless G, editor. Inorganic
is rich in SO3. sulphur chemistry. Amsterdam: Elsevier; 1968 [Chapter 7].
In general, it can be concluded that, at present, much [19] Gurvich, LV, Veyts IV, Alcock CB. Thermodynamic properties
knowledge has been obtained about the thermodyn- of individual substances. Part 2, vol. 1, 4th ed. New York:
Hemisphere; 1989.
amics of pyrite decomposition and the general
[20] Hong Y, Fegley B. Ber Bunsen Phys Chem 1997;101(12):
mechanisms of the transformation processes of pyrite 1870–81.
at elevated temperatures and in inert and oxygen- [21] Hoare Ian C, Hurst Harry J, Stuart William I. J Chem Soc
containing atmospheres. In contrast to this, the kinetic Faraday Trans I 1988;84(9):3071–7.
knowledge is less satisfactory. For the thermal [22] Levenspiel O. Chemical reaction engineering. New York:
Wiley; 1962.
decomposition of pyrite, a lot of kinetic data have
[23] Schwab G-M, Philinis J. J Am Chem Soc 1947;69(11):2588–96.
been published in the literature. However, these data [24] Coats AW, Bright Norman FH. Can J Chem 1966;44:1191–5.
scatter widely, probably due to difficulties in ensuring a [25] Boyabat N, Özer AK, Bayrakceken S, Gülaboglu MS. Fuel
well-defined experimental condition. For the oxidation Process Technol 2003;85:179–88.
of pyrite, the available kinetic data are limited. For the [26] Udintseva VS, Chuforov GI. J Appl Chem USSR 1941;14:3.
[27] Pannetier G, Davignon L. Bull Soc Chim 1961;72:1513.
purpose of reasonable predictions of the transformation
[28] Zhukovskii VM, Montilo IA, Babadzhan AA. Trans Ural Nauch
of pyrite under different reaction conditions in Issled Proekt Inst Med Prom 1965;8:387.
industrial processes, new and better kinetic data on [29] Samal GI. Geterogennye Khim Reaktsii Inst Obshch i Neorgan
the decomposition and oxidation of pyrite are needed, Khim Akad Nauk Belorussk SSR 1965;93.
as well as more sophisticated models that can properly [30] Jorgensen FRA, Moyle FJ. J Therm Anal 1982;25:473–85.
take accounts of other factors such as heat transport and [31] Dunn JG, De GC, O’Connor BH. Thermochim Acta 1989;145:
115–30.
the variation of product layer structure. [32] Dunn JG, De GC, O’Connor BH. Thermochim Acta 1989;155:
135–49.
Acknowledgements [33] Schorr JR, Everhart JO. J Am Ceram Soc 1969;52(7):351–4.
[34] Prasad A, Singru RM, Biswas AK. Phys Status Solidi A 1985;
87:267–71.
This work was carried out in the Combustion and
[35] Hansen J P. PhD Thesis, Department of Chemical Engineering,
Harmful Emission Control (CHEC) Research Centre at Technical University of Denmark; August 2003.
the Department of Chemical Engineering, Technical [36] Srinivasachar S, Helble JJ, Boni AA. Prog Energy Combust Sci
University of Denmark, and financially supported by 1990;16:281–92.
the Technical University of Denmark and FL Smidth [37] Tuffrey NE, Richards GG, Brimacombe JK. Metall Mater Trans
A/S. B 1995;26B:959–70.
[38] ten Brink HM, Eenkhoorn Simon, Hamburg Gerrit. Fuel 1996;
75(8):945–51.
References [39] Vuthaluru HB, Eenkhoorn S, Hamburg G, Heere PGT. Fuel
Process Technol 1998;56:21–31.
[1] Bog S, Rosenqvist T. Trans Faraday Soc 1959;55:1565–9. [40] McLennan AR, Bryant GW, Stanmore BR, Wall TF. Energy
[2] Kullerud G, Yoder HS. Econ Geol 1959;54:533–72. Fuel 2000;14:150–9.
[3] Arnold RG. Econ Geol 1962;75:72–90. [41] Dunn JG, Mackey LC. J Therm Anal 1991;37:2143–64.
[4] Dickson FW, Shields LD, Kennedy GC. Econ Geol 1962;57: [42] Dunn JG, Mackey LC. J Ther Anal 1992;38:487–94.
1021–30. [43] Dunn JG, Mackey LC. J Therm Anal 1993;39:1255–71.
[5] Toulmin P, Barton PB. Geochim Cosmochim Acta 1964;28: [44] Kubaschewski O. Iron-binary phase diagrams. Iron–sulphur.
641–71. Berlin: Springer; 1982. p. 125.
[6] Scott SD, Barnes HL. Econ Geol 1971;66:653–69. [45] Darken LS, Gurry RW. J Am Chem Soc 1946;68:798–816.
[7] Chuang YY, Hsieh KC, Chang Y. Austin Metall Trans B 1985; [46] Kopp OC, Kerr PF. Am Mineral 1958;44:1079–97.
16B:277–85. [47] Cole DA, Simmons GW, Herman RG, Klier K, Czako-Nagy I.
[8] Barker WW, Parks TC. Geochim Cosmochim Acta 1986;50: Fuel 1987;66:1240–8.
2185–94. [48] Gao F, Li F. J Fuel Chem Technol 1989;17(2):175–82.
[9] Hong Y, Fegley B. Planet Space Sci 1998;46(6/7):683–90. [49] Dunn JG, Gong W, Shi D. Thermochim Acta 1992;208:293–303.
[10] Lambert JM, Simkovich G, Walke PL. Metall Mater Trans B [50] Dunn JG, Gong W, Shi D. Thermochim Acta 1993;215:247–54.
1998;29B:385–96. [51] Pelovski Y, Petkova V. J Therm Anal Calorim 1999;56:95–9.
[11] Fegley B, Lodders K. Icarus 1995;115:159–80. [52] Eneroth E, Koch C Bender. Miner Eng 2003;16:1257–67.
[12] Ferrer IJ, Sanchez C. J Appl Phys 1991;70(5):2641–7. [53] Hansen J-P, Jensen LS, Wedel S, Dam-Johansen K. Ind Eng
[13] de las Heras C, Martin de Vidales JL, Ferrer IJ, Sanchez C. Chem Res 2003;42:4290–5.
J Mater Res 1996;11(1):211–20. [54] Hubbard, FH, McGill, RJ. Mineral Mag 1984;48:251–6
314 G. Hu et al. / Progress in Energy and Combustion Science 32 (2006) 295–314

[55] Nishihara K, Kondo Y. Mem Fac Eng Kyoto Univ 1959;21:214–8. [62] David R Lide. CRC handbook of chemistry and physics. 84th ed.
[56] Groves SJ, Williamson J, Sanyal A. Fuel 1987;66:461–6. West Palm Beach, FL: CRC Press; 2003. p. 4–63.
[57] Huffman GP, Huggins FE, Levasseur AA. Fuel 1989;68:485–90. [63] Allen GC, Paul M, Hallam KR. Can J Appl Spectrosc 1995;
[58] Helble JJ, Srinivasachar S, Boni AA. Prog Energy Combust Sci 40(6).
1990;16:267–79. [64] Eymery JP. Eur Phys J AP 1999;5:115–21.
[59] Komraus JL, Popiel ES, Mocek R. Hyperfine Interact 1990;58: [65] Greulich E. Z Anorg Chem 1927;168:197–202.
2589–92. [66] Warner NA, Ingraham TR. Can J Chem 1960;38:2196–202.
[60] ten Brink HM, Smart JP, Vleeskens JM, Williamson J. Fuel [67] Schenk PW, Steudel R. Oxides of sulphur. In: Nickless G,
1994;73(11):1712–7. editor. Inorganic sulphur chemistry. Amsterdam: Elsevier; 1968
[61] Perry HR, Green DW, Maloney JO. Perry’s chemical engineers’ [chapter 11].
handbook. 7th ed. New York: McGraw-Hill; 1997. [68] Srinivasachar S, Boni AA. Fuel 1989;68:829–36.