Kinetics study of the SO2 sorption by Brazilian dolomite using thermogravimetry

Article abstract of DOI:10.1016/j.tca.2006.05.023

Sulfur emission in coal power generation is a matter of great environmental concern. Limestone sorbents are widely used for reducing such emissions. This work applies thermogravimetry to determine apparent activation energy and frequency factor on the sorption of SO₂ by limestone. The kinetic parameters were determined from Arrhenius plots generated from TG/DTG measurements. The experiments were carried out under isothermal conditions between 600 and 900 °C. A natural dolomite with a mean size of 650 μm was used. The Arrhenius plot shows that there is a clear change on reaction mechanism in the range of temperatures between 800 and 875 °C. Supposedly, beyond 850 °C sintering comes to increasingly restrain reaction. For temperatures up to 850 °C the frequency factor and the apparent activation energy resulted, respectively, 1.410 s^-1 and 8.8 kJ mol^-1.

Full text of DOI:10.1016/j.tca.2006.05.023

Thermochimica Acta 447 (2006) 161–166
Kinetics study of the SO2 sorption by Brazilian dolomite
using thermogravimetry
P.M. Crnkovic∗, F.E. Milioli, J.D. Pagliuso
School of Engineering of Sa˜o Carlos, University of Sa˜o Paulo, Sa˜o Carlos 13560-970, Brazil
Received 23 March 2006; received in revised form 25 May 2006; accepted 26 May 2006
Available online 3 June 2006
Abstract
Sulfur emission in coal power generation is a matter of great environmental concern. Limestone sorbents are widely used for reducing such
emissions. This work applies thermogravimetry to determine apparent activation energy and frequency factor on the sorption of SO2 by limestone.
The kinetic parameters were determined from Arrhenius plots generated from TG/DTG measurements. The experiments were carried out under
isothermal conditions between 600 and 900 ◦C. A natural dolomite with a mean size of 650 ␮m was used. The Arrhenius plot shows that there
is a clear change on reaction mechanism in the range of temperatures between 800 and 875 ◦C. Supposedly, beyond 850 ◦C sintering comes to
increasingly restrain reaction. For temperatures up to 850 ◦C the frequency factor and the apparent activation energy resulted, respectively, 1.410 s−1
and 8.8 kJ mol−1
.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Sulfation; Kinetics; Thermogravimetry; Dolomite
1. Introduction
Thecalciumoxideisfrequentlyconsideredasorbenttodesul-
furization in the burning process. When a natural limestone is
injected into the high temperature ambience of a fluidized bed
combustor, it is thermally decomposed and calcination takes
place, i.e.
Fluidized bed desulfurization by limestone sorbents is an
efficient technology for sulfur removal in the combustion of
fuels of high sulfur content. The fluid bed process is suitable
for burning any material containing carbon and hydrogen, no
matter it is gaseous, solid, liquid, slurry, sludge, etc. Concern-
ing large-scale utilization, besides the obvious coal combustion,
fluidized beds have been considered for burning a range of high
sulfurpetrolderivatives(e.g. heavyoils, tars, coke, residues, etc.)
[1].
Desulfurization under combustion is a quite complex process.
Besides chemical kinetics, the rates of reaction are consider-
ably affected by transport both intra-particle and throughout the
combustion environment. This work aims to contribute to the
discussion regarding the determination of reaction rates con-
trolled by the combined effects of kinetics and intra-particle
transport. As composed to ambience transport conditions, the
intrinsic rates allow to establish the actual rate of reaction in a
complex process such as fluidized bed combustion.
CaCO3 + MgCO3 → CaO + MgO + 2CO2
(1)
After calcination CaO reacts with SO2 and O2 to form CaSO4
accomplishing the sulfation step. MgO may also sulfate to form
MgSO4 at moderate temperatures [2]. The global sulfation may
be given by
CaO + MgO + 2SO2 + O2 → CaSO4 + MgSO4
(2)
The actual mechanism leading to the formation of CaSO4 is
still a subject of controversy among researchers. However, it is
generally accepted that the reaction between the porous solid
CaO and the gas SO2 (in oxidizing atmosphere) is first order
related to SO2. It is also generally agreed that sulfation rate is
maximum at temperatures around 850 ◦C [3].
The capability of sorption of a sorbent is ultimately deter-
mined by its porous structure, that allows reactive gases to
access internal active sites and make room for accumulating
the product sulfate. The sulfation reaction goes fast in the first
∗
Corresponding author. Tel.: +55 16 33739557; fax: +55 16 33739402.
E-mail address: paulam@sc.usp.br (P.M. Crnkovic).
0040-6031/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.tca.2006.05.023

162
P.M. Crnkovic et al. / Thermochimica Acta 447 (2006) 161–166
few seconds. Then, it becomes limited by growing resistances to
gas diffusion, either external or intra-particle, or a combination
of both [4]. Pore plugging accounts for growing intra-particle
gas diffusion resistances. The blockage of pores is favored by
the fact that the molar volume of CaSO4 is 172% higher than
that of CaO, and the formed product layer is essentially non-
porous [5]. As the thickness of the product layer increases,
gas diffusion becomes slower and the reaction rate decreases
exponentially [6]. Of course, the higher the temperature the
faster the kinetics, and the higher the sulfation rate. However,
at temperatures sufficiently high the rate of reaction turns to
decrease. Such is commonly attributed to either sintering or re-
emission of SO2 due to a possible decomposition of sulfates [7].
The last effect is relevant in combustion reductive atmospheres
[8].
Many authors have studied sulfation in thermogravimetric
analysis systems. Some of them apply atmospheres with frac-
tionsofCO2 highenoughtoinhibitCaCO3 calcination. Sulfation
under such conditions has been referred to as direct sulfation
(CaCO3–SO2) [5,9–12]. In some cases, high concentrations
of CO2 are used to simulate the ambience of a fluidized bed
combustor. In those conditions direct sulfation of CaCO3 is
showedtoproceedtohighconversion, incontrasttotherelatively
lower sulfation extent of previously calcined limestones. Snow
et al. [5] reported thermogravimetric non-isothermal studies of
CaCO3 sulfation for six different particle sizes, in 95% CO2
atmosphere. The authors found an apparent activation energy of
64.1 kJ mol−1. Hajaligol et al. [6] developed experiments simi-
lar to those of Snow et al. [5] but under isothermal conditions.
Notwithstanding, they found results in accordance to those of
Snow and co-workers. The authors came to the expected conclu-
sion that diffusivities depend on porosity. Their research showed
that under direct sulfation the product sulfate layers become
more porous owing to the simultaneous generation of CO2. The
passage of CO2 through the sulfate layers leads to a more open
structure, thereby allowing an easier access of SO2 and O2 to
under layer CaO active sites. As a consequence, direct sulfation
turned out to be more effective that the sulfation of previously
calcined limestone.
In this work isothermal thermogravimetry is applied to deter-
mine apparent activation energy and frequency factor on the
sorption of SO2 by a particular dolomite, considering Arrhenius
kinetic.
1.1. Basic theory and method
Following the global sulfation reaction in Eq. (2), reaction
rate results
dm
∝ −mCSaO COb
(3)
2
2
dt
where m stands for the mass of Ca plus Mg available for sulfa-
tion; CSO and CO are the concentrations of the reactive gases
2
2
in the atmosphere; a and b are reaction orders related to each
reactant gas. The reaction is assumed first order related to the
sample mass. Differential conditions [4,14,15] are imposed by
applying high concentrations of the reactant gases in the atmo-
sphere (20%), so that transport effects external to the particles
are eliminated and the reaction results controlled by intrinsic
or intra-particle kinetics and diffusion mechanisms. As a conse-
quence, thereactionbecomesindependentofgasconcentrations,
i.e., pseudo zeroth order related to both SO2 and O2 (a = b = 0).
The ultimate rate of reaction results
dm
∝ −m
(4)
dt
Introducing a reaction rate coefficient (k), it comes that
dm
= −km
dt
(5)
Arrhenius kinetics is followed to account for reaction rate depen-
dence on temperature, i.e.
ꢀ
ꢁ
Ea
k = A exp
−
(6)
RT
Then, Eq. (5) becomes
ꢀ
ꢁ
1 dm
m dt
Ea
Van Houte and Delmon [11] investigated the CaCO3–SO2
reaction at low temperatures to avoid calcination. At around
550 ◦C and low conversions, the authors found an apparent acti-
vation energy of 210 kJ mol−1. For 60% conversion they found
110 kJ mol−1. Li and Sadakata [13] determined the apparent
activation energy for sulfation with both CaO and CaO·Al2O3,
thereby evaluating the effect of Al2O3 on SO2 removal. They
found apparent activation energies of 7.29 kJ mol−1 for pure
CaO, and 9.21 kJ mol−1 for CaO·Al2O3, and concluded that
CaO·Al2O3 is more active on the dessulfurization.
−
= A exp
−
(7)
(8)
RT
or
ꢂ
ꢃ
1 dm
m dt
Ea 1
ln −
= ln[A] −
R T
From empirical data of m, dm/dt and process temperature, and
applying Eq. (8), an Arrhenius’ plot can be constructed, which
allows to derive A and Ea by linear regression. The mass of Ca
plus Mg available for sulfation and its time rate can be deter-
mined from a mass balance. Assuming Ca and Mg to be sulfated
at the some rate, it comes that:
All the concerning sulfation literature presents apparent acti-
vation energies which are time averaged, since the physical
structure of limestones and thereby chemical kinetics change
on time. Reaction mechanisms change from kinetic control at
lower temperatures to diffusion control at higher temperatures,
which is intensified as pore plugging advances. It becomes quite
clear that the apparent activation energy considerably changes
during conversion.
ꢀ
ꢁ
ms − mc
m = ML(YCa + YMg) −
WSO + 0.5WO
2
2
ꢀ
ꢁ
Y
CaWCa + YMgWMg
×
(9)
YCa + YMg

P.M. Crnkovic et al. / Thermochimica Acta 447 (2006) 161–166
163
Table 1
and
Partial chemical analysis of the dolomite (wt.%)
ꢀ
ꢁ
dm
dt
YCaWCa + YMgWMg
dms
dt
= −
(10)
Element
wt.%
(WSO + 0.5WO )(YCa + YMg
)
2
2
Sr
0.0811
0.0888
0.0054
0.097
0.0554
0.3207
0.4233
17.07
where ms is the mass of sulfated material at a given reaction
stage (TG result) and dms/dt stands for its time rate (DTG
result); mc is the mass of calcined limestone (i.e. the mass of
the sample before sulfation); ML is the initial mass of the sam-
ple (i.e. the mass of the sample before calcination); Y accounts
for the mass fraction of the concerning species in the natu-
ral limestone; and W is the molecular mass of the concerning
species.
Mn
Ba
K
P
Fe
Al
Ca
Mg
11.72
The reaction rate on limestone sulfation is inherently vari-
able since the physical structure of the rock changes during
the process. The reaction experiments an enhancement in its
early stages while the reactive gases flow into the reactive solid
particles. Then, intra-particle diffusion resistances become pro-
gressively intensified notably owing to pore plugging, causing
the reaction rate to progressively decay. There is a stage along
the process on which a maximum reaction rate is found. Such
is identified by a pick on a DTG curve. In this work the kinetic
parameters are determined for this stage of maximum reaction
rate.
At the maximum reaction rate spot, the reaction rate coeffi-
cient is given by
ꢂ
ꢃ
1 dm
m dt
kmax = −
(11)
Fig. 1. X-ray diffraction of the natural Brazilian dolomite. ꢀ: CaMg(CO3)2.
max
or
ꢂ
ꢃ
((YCaWCa + YMgWMg)/((WSO + 0.5WO )(YCa + YMg)))(dms/dt)
2
2
kmax = −
(12)
ML(YCa + YMg) − ((ms − mc)/(WSO + 0.5WO )((YCaWCa + YMgWMg)/YCa + YMg))
2
2
max
2. Experimental
the arithmetic mean between the mesh apertures of two consec-
utive ASTM laboratory sieves. The particular 650 ␮m particle
size was chosen so that decrepitation effects were avoided [16],
and considering it is a size common to fluidized bed combus-
tion applications. Chemical analysis and RXD of the natural
dolomite are shown in Table 1 and Fig. 1, respectively.
Samples weighing 10.0 0.5 mg were used throughout the
experiments. This mass of dolomite allows minimizing inter-
particle mass transfer resistances by spreading the particles
in a monolayer on the bottom of the crucible (alumina, Ø
6 mm × 2.5 mm).
The thermogravimetric analysis was carried out in a Shi-
madzu 51H unit with a high-temperature furnace (up to
1500 ◦C). The procedureconsisted of heatingasorbentsampleat
30 ◦C min−1 until a desired temperature was reached. Synthetic
air was continuously blown through the furnace over the sam-
ple at a volumetric rate of 80 mL min−1 as a carrier gas. When
an isothermal condition was reached, 20 mL min−1 of pure SO2
(100% v/v) was added to ensure a volumetric fraction of the
gas in the furnace atmosphere of 20%. Under these conditions,
the reaction rate was found based on the shape of TG and DTG
curves (correct for buoyancy).
Isothermal TG sulfation experiments were performed at 600,
650, 700, 750, 800, 850, 875 and 900 ◦C. In the experiments
at 600, 650, 700 and 750 ◦C, the temperature was firstly raised
to 850 ◦C under synthetic air atmosphere in order to guarantee
a complete calcination. Then, the temperature was reduced at
a rate of 30 ◦C min−1 until a temperature desired for sulfation
was reached. Five replicate experiments were carried out on each
temperature.
The reaction solid products were analyzed through X-ray
diffraction in a 2θ range of 20–70◦. A supplementary study was
performed to clarify the reactivity to SO2 of both CaO and MgO.
In such study reagent-grade samples of CaCO3 and MgCO3
were submitted to calcination followed by isothermal sulfation
at 850 ◦C in TG experiments.
3. Results and discussion
˜
A Brazilian dolomite from Sao Paulo State was used in this
Fig. 2 shows TG and DTG curves for one of the performed
experiments of calcination followed by sulfation. In this case,
the isothermal condition was established at 850 ◦C. The data
study. Therockwasgrindedandsievedtoanaverageparticlesize
of 650 ␮m. The average size of the particles was determined as

164
P.M. Crnkovic et al. / Thermochimica Acta 447 (2006) 161–166
Fig. 2. TG and DTG curves for one of the performed experiments of calcination
followed by sulfation. The temperature of the isothermal step is 850 ◦C.
Fig. 4. TG curves of calcination followed by sulfation for reagent-grade samples
of CaCO3 and MgCO3.
Table 2 gives ML, ms, mc, dms/dt, and ln[(−1/m)(dm/dt)]max
averaged over five replicated experiments for each process tem-
perature. Fig. 5 shows an Arrhenius plot giving the dispersion
around the average values reported in Table 2.
show that sulfation presents two consecutive different behav-
iors. At first it advances very quickly turning to advance slowly
afterwards. In the first step the reaction is very fast owing to
kinetics control. The second step is slower owing to a growing
diffusion resistance to reaction imposed by pore plugging. As it
can be observed the mass gain in sulfation is not proportional to
the mass loss during the calcination.
The XRD spectrum of sulfated dolomite is shown in Fig. 3.
According to the JCPDS file the main peak for CaSO4 appears
at 2θ = 25.5◦, and the minor peaks appear at 2θ = 30.3◦, 36.0◦,
39.6◦ and 51.1◦. For MgSO4 the main peaks appear at 2θ = 24.9◦
and 25.1, and the minor peak, at 2θ = 21.4◦.
EventhoughCaSO4 andpossiblyMgSO4 havebeendetected,
it is quite difficulty to distinguish the main peaks of CaSO4 and
MgSO4 in the XRD spectrum. So, in order to prove the sulfa-
tion reaction of both CaO and MgO, investigations were made
through thermogravimetry. TG curves of calcination followed
by sulfation obtained with reagent-grade samples (CaCO3 and
MgCO3) are shown in Fig. 4. As it can be seen, the gain of mass
was obtained for both sorbents. From this observation, it was
assumed that the MgO reacts with SO2.
The relatively high dispersions are possibly due to the het-
erogeneity of the sample, since it deals with natural limestone
extracted directly from mine and size distribution. It can be
seen from Table 2 and Fig. 5 that reaction rate increases with
increasing temperatures up to 850 ◦C, owing to increasingly
faster kinetics. The reaction rate drops for temperatures above
875 ◦C. Asseen, thereisaclearchangeonreactionmechanismin
the range of temperatures between 800 and 875 ◦C. Supposedly,
beyond 850 ◦C sintering comes to increasingly restrain reaction.
Such decrease seems to be a consequence of sintering (SO2 re-
emission is discarded since no combustion reductive atmosphere
is applied).
An Arrhenius plot for sulfation up to 850 ◦C is showed in
Fig. 6. For the whole range of temperatures, up to 850 ◦C, a
linear regression provides a frequency factor of 1.410 s−1 and
an apparent activation energy of 8.8 kJ mol−1. In the range from
600 to 700 ◦C the frequency factor and the apparent activation
Fig. 5. Arrhenius plot showing the dispersion around the averaged values of
ln[(−1/m)(dm/dt)]max reported in Table 2. The bars point the maximum and the
minimum values that were found.
Fig. 3. X-ray diffraction pattern of the products CaSO4 (2θ = 25.5◦, 30.3◦, 36.0◦,
39.6◦ and 51.1◦) and MgSO4 (2θ = 24.9◦, 25.1◦, 21.4◦).

P.M. Crnkovic et al. / Thermochimica Acta 447 (2006) 161–166
165
Table 2
Values of ML, ms, mc, and dms/dt for each process temperature, and ln[(−1/m)(dm/dt)]max averaged over five replicated experiments
Temperature (◦C)
ML (mg)
ms (mg)
mc (mg)
dms/dt (mg min−1
)
Averaged ln[(−1/m)(dm/dt)]max
Tempature 10000/T (K−1
)
10.17
10.27
9.79
10.02
10.32
7.88
7.91
7.67
7.70
7.89
6.58
6.54
6.19
6.20
6.54
2.73
2.67
2.76
2.55
2.70
600
0.01
11.5
10.8
10.3
9.8
10.31
10.08
9.87
9.80
10.19
8.38
8.00
8.12
8.1
6.57
6.63
6.50
6.54
6.78
2.66
2.84
2.65
2.70
2.70
650
700
750
800
850
875
900
0.03
0.17
0.20
0.22
0.26
0.17
0.00
8.67
9.88
10.02
9.71
9.92
10.32
8.25
8.19
7.86
7.90
8.16
6.78
6.57
6.60
6.30
6.35
3.20
3.13
2.98
3.07
3.03
10.43
10.58
10.19
10.29
9.76
8.69
8.64
8.31
8.33
7.99
6.84
6.96
6.85
6.46
6.15
3.16
3.12
3.09
3.62
3.27
9.83
10.05
10.25
10.04
10.25
7.77
8.50
8.66
8.45
8.75
6.46
6.90
6.99
6.55
6.97
3.35
3.30
3.29
3.10
3.30
9.3
9.20
9.50
9.74
10.07
10.47
7.27
7.04
7.39
8.25
8.74
5.97
6.27
6.33
6.58
6.79
3.11
3.68
3.55
3.18
3.29
8.9
10.37
10.05
10.34
10.09
10.04
8.56
8.26
9.03
8.42
8.23
7.01
6.57
7.44
6.57
6.78
3.19
3.20
3.10
3.16
3.08
8.71
8.53
10.54
10.18
10.19
10.21
9.87
9.1
7.36
6.34
7.12
7.20
6.8
2.85
2.50
2.69
2.75
2.70
8.07
8.53
8.41
7.98
energy resulted, respectively, 2.037 s−1 and 11.6 kJ mol−1. In
the range from 700 to 850 ◦C the parameters resulted, respec-
tively, 0.957 s−1 and 5.3 kJ mol−1. This variation on the con-
cerning parameters suggests that at different temperatures the
maximum reaction rate finds the particles under different phys-
ical structures, and consequently at different balances between
kinetic and diffusion control. It should be noted that the present
experiments detected only combined effects of kinetics plus
intra-particle diffusion, so that the balance between these effects
could not be evaluated separately.
on apparent activation energy [5,9–12] considerably differ from
those derived in the present work (e.g. 64.1 kJ mol−1 [5]). How-
ever, the values presented in those reports refer to sulfation in
atmosphere with O2 concentration of 5%, CO2 concentration
ranged from 2 to 95% and SO2 varying from 600 to 3000 ppm.
The low concentrations of SO2 suggest that differential condi-
tions are not met, and mass transport external to the particles act
as an additional resistance to reaction. In addition, the presence
of CO2 in the atmosphere can also affect conversion since it can
affect the structure of the product layer.
On the other hand, the apparent activation energies pre-
sented here are in good agreement with those presented in Li
and Sadakata [13] (7.29 kJ mol−1). Those authors performed
experiments in a thermogravimetric balance in absence of CO2,
applyingaSO2 concentrationof3000 ppm. Theyusedaveryfine
Any comparison among results from different works must
take into account differences on experimental conditions. One
of such conditions refers to the reaction atmosphere. In this work
reaction took place in absence of CO2, and the fractions of
SO2 and O2 were both 20%. Some literature reported figures

166
P.M. Crnkovic et al. / Thermochimica Acta 447 (2006) 161–166
the atmosphere can possibly affect conversion, contributing for
the observed differences.
Acknowledgements
This work is part of the research program on mineral
coal combustion of NETeF (Thermal and Fluids Engineering
˜
Group—University of Sao Paulo). The program is supported
˜
by FAPESP (Sao Paulo State Research Foundation) and CNPq
(Brazilian Council for Scientific and Technological Develop-
ment) through both scholarships and research grants. The first
author is grateful to FAPESP for the doctoral and post-doctoral
scholarships (processes 98/14393-3; 04/06894-4).
References
Fig. 6. Arrhenius plot for the sulfation reaction, for temperatures up to 850 ◦C.
[1] F.M. Okasha, S.H. Hemam, H.K. Mostafa, Exp. Therm. Fluid Sci. 27
(2003) 473–480.
[2] S.S. Tambe, S.S. Yerrapragada, K.L. Guari, J. Mater. Civil Eng. 6 (1994)
65–77.
[3] D. Allen, A. Hayhurst, J. Chem. Soc., Faraday Trans I. 92 (1996)
1227–1228.
CaO particulate (<45 ␮m) to reduce pore diffusion effects, and
dispersed the particles in quartz wool to minimize inter-particle
diffusion effects.
[4] M. Hartman, R.W. Coughlin, AIChE J. 22 (1976) 490–498.
[5] M.J.H. Snow, J.P. Longwell, A.F. Sarofim, Ind. Eng. Chem. Res. 27
(1988) 268–273.
[6] A. Lyngfelt, B. Leckner, Chem Eng. Sci 44 (1989) 207–213.
[7] T. Mattisson, A. Lyngfelt, Energy Fuels 12 (1998) 905–912.
[8] K. Lisa, M. Hupa, J. Inst. Energy 65 (1992) 201–205.
[9] M.R. Hajaligol, J.P. Longwell, A.F. Sarofim, Ind. Eng. Chem. Res. 27
(1988) 2203–2210.
[10] A.B. Fuertes, G. Velasco, M.J. Fernandez, T. Alvarez, Thermochim. Acta
242 (1994) 161–172.
[11] G. Van Houte, B. Delmon, J. Chem. Soc., Faraday Trans. I 75 (1979)
1593–1605.
[12] Z. Qin, J. Therm. Anal. 45 (45) (1995) 211–219.
[13] Y. Li, M. Sadakata, Fuel 78 (1999) 1089–1095.
[14] P. Daniell, Soltani-Ahmadi, H.O. Kono, Powder Technol. 55 (1988)
75–85.
[15] S.A. Carello, A.C.F. Vilela, Ind. Eng. Chem. Res. 32 (1993) 3135–3142.
[16] P.M. Crnkovic, W.L. Polito, C.G. Silva Filho, F.E. Milioli, J.D. Pagliuso,
Quim. Nova 27 (2004) 58–61.
4. Conclusions
The TG/DTG experiments on sulfur sorption by dolomite
showed a change on sulfation mechanism at 850 ◦C. Such was
attributed to sintering. SO2 re-emission was rejected since no
combustion reductive atmosphere was applied.
In this work differential conditions were applied so that the
concentration of the reactant gases did not affect reaction rates.
The relatively high dispersions around the average values over
five replicated experiments are possibly due to the heterogeneity
of the sample, since it deals with natural limestone extracted
directly from mine.
Significant differences were found as the present activation
energies were compared to literature available data. It is sug-
gestedthatmaybethosedatawerenotobtainedunderdifferential
conditions. Besides, it is recognized that the presence of CO2 in