You are on page 1of 14

Chemical Engineering Science: X 8 (2020) 100071

Contents lists available at ScienceDirect

Chemical Engineering Science: X


journal homepage: www.elsevier.com/locate/cesx

CaCO3 decomposition for calcium-looping applications: Kinetic


modeling in a fixed-bed reactor
Athanasios Scaltsoyiannes, Angeliki Lemonidou ⇑
Department of Chemical Engineering, Aristotle University of Thessaloniki, University Campus, Thessaloniki GR-54124, Greece

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

Article history: Calcium looping can be used in various environmental applications such as post-combustion CO2 capture,
Received 25 April 2020 sorption-enhanced reforming and thermochemical energy storage. The calcination reaction mechanism
Received in revised form 8 June 2020 and the effect of CO2 partial pressure have not been fully clarified yet justifying further efforts; this work
Accepted 4 July 2020
focuses on kinetic modeling of the reaction. Herein, calcination kinetics of limestone were studied exper-
imentally in a fixed-bed reactor, and theoretically using different models previously reported in the lit-
erature. The Uniform Conversion Model (UCM) and the Changing Grain Model (CGM) described better the
Keywords:
conversion-time data than the Random Pore Model (RPM). A Langmuir-Hinshelwood mechanistic model,
Calcium looping
Calcination kinetics
which included the decomposition of CaCO3, the desorption of CO2 and the CaO* relaxation fitted well the
Limestone decomposition reaction rate values leading to an activation energy of 210 kJ/mol and the pre-exponential factor equal to
Langmuir-Hinshelwood mechanism 18,860 kmol/(m2 s). The model was validated by predicting adequately the decomposition rates mea-
Carbonate decomposition modeling sured by other authors under different conditions.
Ó 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction for intensification of thermodynamically limited reactions, such


as in reforming and water gas shift processes (Antzara et al.,
Calcium looping (CaL) has attracted interest as a process with 2014, 2016; Broda et al., 2013; Di Giuliano and Gallucci, 2018; Li
various potential environmental applications (Anthony, 2011; et al., 2006; Papalas et al., 2020a; Udomsirichakorn and Salam,
Hanak et al., 2018). The abundance of non-toxic materials as well 2014). During sorption-enhanced steam methane reforming, the
as the temperature range of carbonation and calcination reactions, produced CO2 is in-situ adsorbed shifting the equilibrium, resulting
render CaL suitable for coupling with other processes. During coal in high-purity hydrogen production under milder conditions and in
combustion, dry pulverized limestone is injected in a high temper- a single step. More recently, research community has also focused
ature region (600–1400 °C) of the furnace, where is heated rapidly on integration of CaL in thermochemical energy storage applica-
and calcined. At these conditions, the produced CaO adsorbs the tions. Concentrated Solar Power is a relatively new technology
SO2 emissions, responsible for acid rain formation (Basinas et al., for harnessing the solar energy in the form of heat by concentrat-
2014; Hu and Scaroni, 1996; Wang and Chen, 2016; Zhong and ing the sun’s radiation at a central receiver where temperatures up
Bjerle, 1993). CaL is also a rapidly developing technology for to 1500 °C can be developed. Commercially, this heat is converted
post-combustion CO2 capture from flue gases (Martínez et al., to electricity using molten salts as the storage medium, which suf-
2016; Perejón et al., 2016; Zhao et al., 2013). A proposed configu- fer from several drawbacks, such as temperature limits, low effi-
ration for this process includes two interconnected circulating ciency and short time storage (Edwards and Materić, 2012). As
fluidized-bed reactors (CFBs), a calciner and a carbonator, retro- recently reviewed (Ortiz et al., 2019; Yan et al., 2020), CaL is poten-
fitted in existing power plants (Martínez et al., 2012). The CO2 con- tially a means for storing solar energy in a chemical substance by
tent in the hot flue gases reacts with CaO particles in the driving the endothermic CaCO3 decomposition. Carbonation reac-
carbonator forming CaCO3, which afterwards is transferred to the tion can produce heat (DΗ0car = 178 kJ/mol) at high temperatures
calciner, wherein a pure concentrated CO2 stream ready for (~850 °C under pure CO2 atmosphere) increasing the power cycle’s
sequestration can be produced. CaO returns then into the carbon- efficiency, while the regeneration of CaCO3 is feasible within the
ator and the cycle is repeated. CO2 sorption can be applied also operational temperature range of CSP towers (Alovisio et al., 2017;
Chacartegui et al., 2016; Ortiz et al., 2018, 2019). These features,
along with materials’ abundance, high energy density and lack of
⇑ Corresponding author.
toxicity, render CaL a promising vehicle to store solar energy in
E-mail address: alemonidou@cheng.auth.gr (A. Lemonidou).

https://doi.org/10.1016/j.cesx.2020.100071
2590-1400/Ó 2020 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
2 A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071

Nomenclature

VM,CaCO3 CaCO3 molar volume, m3/kmol DS0a standard entropy of adsorption, kJ/(mol K)
r’ reaction rate, kmol/(m2 s) k1H kinetic constant of decomposition reaction according to
r reaction front velocity, m/s the Hyatt model, kmol/(m2 s)
dX/dt conversion change rate, 1/s k2H kinetic constant of carbonation reaction according to
dp mean particle diameter, m the Hyatt model, kmol/(m2 s kPa)
GHSV Gas Hourly Space Velocity, h1 k3H kinetic constant of structural transformation of CaO* to
SBET initial BET surface area, m2/g CaO according to the Hyatt model, kmol/(m2 s)
S0 initial total pore surface area, m2/(m3 of particle) k4H kinetic constant of structural transformation of CaO to
S’0 initial total pore surface area, m2/(m3 of CaCO3) CaO* according to the Hyatt model, kmol/m2 s
L0 initial total pore length, m/(m3 of particle) B kinetic parameter of the Hyatt model, s/(kPa m)
PV0 initial pore volume, cm3/g R0 reaction front velocity under zero PCO2 according to the
r0 initial mean pore radius, m Hyatt model, m/s
t reaction time, s K’ complex constant in L-H with structural transformation
(CaCO3) undecomposed calcium carbonate, mol model, 1/kPa
PCO2 CO2 partial pressure, kPa K00 pre-exponential term of K’ constant, 1/kPa
Pe equilibrium CO2 partial pressure, kPa DΗ00 enthalpy term of K’, kJ/mol
T reaction temperature, K DΗE enthalpy term of KE, kJ/mol
0
k1L-H kinetic constant of decomposition reaction according to DS0 entropy term of K’, kJ/(mol K)
L-H model, kmol/(m2 s) k1E kinetic constant of decomposition reaction according to
k1L kinetic constant of decomposition reaction according to the exponential model, kmol/(m2 s)
L-H model with Langmuir isotherm, kmol/(m2 s) k01E pre-exponential term of k1E constant, kmol/(m2 s)
k2L-H kinetic constant of carbonation reaction according to E1E activation energy term of k1E constant, kJ/mol
L-H model, kmol/(m2 s) k01L pre-exponential term of k1L constant, kmol/(m2 s)
kD kinetic constant of CO2 desorption according to L-H E1L activation energy term of k1L constant, kJ/mol
model, kmol/(m2 s) qb bulk density of calcite, g/cm3
kA kinetic constant of CO2 adsorption according to L-H
model, kmol/(kPa m2 s) Dimensionless parameters
KA equilibrium constant of CO2 adsorption according to L-H KE exponential constant according to exponential model
model, 1/kPa K0E pre-exponential term of KE constant, kmol/m2 s
KAL equilibrium constant of CO2 adsorption according to L-H h fraction of active sites occupied by CO2 molecules
model using the Langmuir isotherm, 1/kPa s dimensionless time used in RPM
KAF equilibrium constant of CO2 adsorption according to L-H W,r structural parameters used in RPM
model using the Freundlich isotherm, 1/kPa X calcination conversion
n number of active sites needed for the sorption of one e0 Initial porosity
CO2 molecule
DΗ0a standard enthalpy of adsorption, kJ/mol

the long term. The material that has been proposed and thoroughly accounted for their contribution in the models (Escardino et al.,
investigated for such applications is limestone, a cheap, abundant, 2013; Hills, 1968; Hu and Scaroni, 1996; Khinast et al., 1996;
non-toxic mineral, consisted of CaCO3 with minor impurities Okunev et al., 2008; Takkinen et al., 2012; Ying et al., 2000). The
(Benitez-Guerrero et al., 2018a, 2017; Papalas et al., 2020b). Dolo- surface area value before decomposition greatly affects the calcu-
mite is another low-cost mineral with high CaCO3 content (usually lated rate. CaCO3 has a low Tammann temperature, so sintering
50–60%wt. in CaCO3), usually characterized by higher surface area phenomena are not negligible at calcination conditions. Addition-
and better cyclic stability than limestone (Sarrion et al., 2016). Also ally, the experimental conditions have to be chosen properly in
many synthetic materials with improved properties have been order to achieve differential operation with respect to the CO2 flow
reported (Antzara et al., 2018; Benitez-Guerrero et al., 2018b; changes and to minimize the effects of transfer phenomena on the
Erans et al., 2016), but they are considered inefficient for large scale reaction rate. The steps of calcination reaction are (i) the external
applications due to their increased production cost. and internal heat transfer towards the reaction surface, (ii) the
All the aforementioned processes need a deep understanding of decomposition reaction and (iii) the mass transfer of the produced
CaL kinetic mechanisms. In most cases, calcination reaction has to CO2 from the reaction surface to the bulk gas passing through the
be conducted under pure CO2 atmosphere, which imposes also generated porous CaO (Hu and Scaroni, 1996). Transfer effects
high temperatures (>900 °C). Decomposition reaction of CaCO3 is should be minimized in a kinetic study, isolating the second step
a well-studied topic for many decades, but without agreement which is of interest.
between the findings (Borgwardt, 1985; Garcı´a-Labiano et al., The surface area of the consuming material usually changes
2002; Hyatt et al., 1958; Ingraham and Marier, 1963; Khinast with time, thus, a particle model which relates the conversion with
et al., 1996; Ying et al., 2000). Many factors contribute to these dis- the material’s structural properties (e.g. surface area) has to be
crepancies such as the different materials used (e.g. limestones, introduced. There are many particle models proposed for this kind
calcite crystals, precipitated CaCO3), experimental apparatus (e.g. of reactions and used also for the CaCO3 decomposition (Stanmore
thermogravimetric analysers, flow reactors), calcination conditions and Gilot, 2005), such as the Shrinking Core Model (SCM), the Grain
(e.g. gas velocities, temperature and PCO2 range) and the different Model (GM), the Random Pore Model (RPM) and the Uniform Con-
models used to describe the experiments. Transfer phenomena version Model (UCM). These models predict the evolution of the
may also play an important role and several investigators have reaction surface area as the reaction proceeds knowing the initial
A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071 3

value. The SCM is an extreme case where the particle is considered An effort is put to shed light on the factors led to the reported dif-
impervious and only the external surface takes part into the ferences in the literature. The effect of CaCO3 sintering before cal-
decomposition. It applies only on non-porous solids, as well as, cination on the material’s surface area and the chosen particle
in cases where the intra-particle diffusion of gas species is practi- model are highlighted as important factors leading to discrepan-
cally negligible. The other three models (GM, RPM and UCM) cies. The dependence of the decomposition rate on CO2 concentra-
describe the porosity of the sorbent, with a different way each tion in the gas phase is investigated by conducting experiments
one. The RPM was initially introduced by Bhatia and Perlmutter under various CO2 partial pressures and reaction temperatures,
(1980) and another version by Gavals (1980). The void fraction is carefully minimizing the transfer effects. The derived model is val-
represented by a set of randomly distributed cylindrical pores, idated by predicting the experimental results from other authors
which have a probability of intersection due to enlargement during under different experimental conditions and configurations.
decomposition. The GM, firstly proposed by Szekely and Evans
(1970), describes the particle’s porosity as the voids between a
2. Methodology
set of non-porous particles. During the decomposition reaction,
these particles follow the shrinking core model and thus the over-
2.1. Experimental apparatus and procedure
all porosity increases. Each of these models calculates the reaction
surface area evolution differently, adding one more factor that
The apparatus used for the kinetic experiments consists of the
makes the reported kinetic parameters incomparable (Stanmore
gas feed inlet section, the reactor and the product analysis section.
and Gilot, 2005). The pre-exponential factor (k0) of the decomposi-
The incoming gases are controlled by mass flow controllers and are
tion reaction constant is greatly affected by the model used and the
pre-mixed before entering the reactor. A fixed-bed quartz reactor
surface area value taken into consideration. Garcı́a-Labiano et al.
(10 mm internal diameter), equipped with a coaxial thermocouple
(2002) calculated k0 equal to 0.254 kmol/(m2 s) using the Changing
for temperature monitoring, was used for the testing. The reactor
Grain Model (CGM) for a limestone with high porosity, 0.029 kmol/
was heated electrically by a tubular furnace, with three indepen-
(m2 s) using the CGM for a dolomite with high porosity and 6700
dently controlled temperature zones. The hot gases exiting the
kmol/(m2 s) using the SCM for a limestone with low porosity, tak-
reactor were cooled down to room temperature and continuously
ing into consideration only the external surface area. Other
analyzed by a mass spectrometer (OmnistarTM GSD 320, PFEIFFER).
researchers have used cleaved calcite crystals to study the decom-
The limestone sample was firstly sieved in the range of
position reaction (Hyatt et al., 1958; Ingraham and Marier, 1963).
45 < dp < 75 lm. For each experiment 100 mg of the material were
In this case, if the aspect ratio (width to height ratio of the crystal)
mixed with 1.5 g of quartz with particle size of 100 < dp < 180 lm
is high, a constant surface area can be considered (Hyatt et al.,
and loaded in the reactor. According to these values of particles
1958). The values of the activation energy mostly referred in the
size and reactor length and diameter (L = 1.5 cm and dt = 1 cm)
literature are close to the reaction enthalpy (180 kJ/mol), but Hills
the dispersion criteria were met for ideal plug flow (L/dp > 50
argued that this result arises due to heat transfer control (Hills,
and dt/dp > 10). The reactor was heated at calcination temperature
1968). There are also large deviations such as 91.7 kJ/mol by Martí-
with a heating rate of 20 °C/min under CO2 flow (GHSV = 22500 h
nez et al. (2012) and 110 kJ/mol by Zhong and Bjerle (1993). 1
), to avoid decomposition. When the desired temperature was
Borgwardt (1985) conducted calcination experiments of different
reached, calcination was initiated by switching the flow from pure
limestones in a wide temperature range (475–1000 °C) and calcu-
CO2 to the preferred CO2/N2 mixture with the same flow. The CO2
lated an activation energy of 205 kJ/mol, a value that was adopted
flow during calcination contained 2 %vol. Ar as an internal stan-
later by other researchers (Okunev et al., 2008; Silcox et al., 1989).
dard. Given that the Ar flow was constant, the flow ratio of CO2/
Despite the voluminous research on decomposition of CaCO3,
Ar increased when calcination reaction was taking place. The flow
no consensus has been achieved regarding the effect of CO2 partial
changes were calculated based on the ratio of CO2/Ar signals
pressure (Garcı´a-Labiano et al., 2002; Khinast et al., 1996). It has
detected by the mass spectrometer (represented by mass charge
been generally reported that the presence of CO2 slows down the
(m/z) ratios 44 and 40), while temperature in the reactor was also
reaction rate, but the mechanism has not been identified yet.
recorded during the whole procedure.
Ingraham and Marier (1963) and Silcox et al. (1989) argued a first
order reaction with respect to CO2, while others (Darroudi and
Searcy, 1981; Hu and Scaroni, 1996) proposed a zero order reaction 2.2. Materials and structural characterization
for very low partial pressures (<102 Pe) and linear under higher
PCO2. Khinast et al. (1996) and Okunev et al. (2008) observed an A raw limestone (Granicarb 0.1/0.8, OMYA Company) was used
exponential decay of the reaction rate as increasing PCO2. A mech- as received in the kinetic experiments. The material contained
anism including an adsorption stage of CO2 molecules on CaO almost pure CaCO3 as shown in the XRD diagram of Fig. 1a. X-ray
active sites has also been studied (Garcı́a-Labiano et al., 2002; diffraction (XRD) patterns were obtained with a BRUKER D8
Wang and Thomson, 1995). Accordingly, Wang and Thomson ADVANCE diffractometer and Cu-Ka radiation with radiation
(1995) used a Langmuir-Hinshelwood type model, and based on wavelength of 0.15406 nm. An aluminum holder was used to sup-
that, Garcı́a-Labiano et al. (2002) found that Freundlich isotherm port the samples during the measurements. The intensity data
fits well their experimental results. Hyatt et al. (1958) proposed were collected over a 2h range of 10°  80° with a step of 0.04°
a two stage model, containing a solid state relaxation step of the and a scanning time of 2 s/point. CaCO3 has a low Tammann tem-
active CaO* produced from the decomposition reaction of CaCO3, perature (529 °C (Lide, 2005)), thus, the effect of thermal annealing
to the cubic CaO. Recently, Valverde et al. used a Langmuir Hinshel- before calcination on the materials structural properties was
wood two-stage model to describe Hyatt’s results (Valverde, 2015). tested. According to this, the limestone was heated up to 850 °C
The scope of the present work is to investigate the kinetics of (which is well above the Tammann temperature) and remained
limestone decomposition under various CO2 partial pressures and there for 10 min under pure CO2 atmosphere. Then, it was cooled
reaction temperatures in a lab scale fixed-bed reactor. Different down to room temperature and was subjected to BET/BJH mea-
particle models such as the RPM, CGM and UCM are applied to surement. BET surface area and pore volume were measured by
describe the conversion-time curves. Various decomposition using nitrogen physisorption at 77 K with an Autosorb-1 Quan-
mechanistic models found in the literature are also discussed and tachrome flow apparatus. Prior to these measurements, the sam-
their application to the experimental data is critically assessed. ples were degassed in vacuum at 250 °C overnight.
4 A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071

Fig. 1. XRD pattern of raw limestone (a) and pore volume distribution of raw limestone, limestone just before 1st calcination and lime after 1st calcination (b).

As it will be pointed out later, the surface area of the material made. This means that the reaction front velocity (r in m/s) is the
just before calcination is an important parameter introduced to same at any point of the CaCO3 particle, both inside the pores and
the particle models, so it has to be accurate. Because of the thermal at the external surface. This, in turn, imposes homogeneous condi-
annealing of limestone the surface area of the material dropped tions of temperature and CO2 partial pressure throughout the whole
from 1.20 to 0.57 m2/g as well as the pore volume due to induced particle. Different models have been used in the past to describe the
sintering above the Tammann temperature (see Table 1). This decomposition reaction of CaCO3 particles. Each of them uses an ini-
effect is not well studied in the literature before. The conditions tial surface area value and predicts its evolution during the reaction
and the time of thermal annealing before calcination may affect in different way, leading also to different curves of conversion versus
in different extent the structural properties of the material and, time. In addition, each of the models contains a term of r  S, which
thus, the subsequent decomposition rate producing at the same is the reaction front velocity multiplied by the initial surface area of
time the discrepancies reported in the literature. The first decom- CaCO3 (in units of m2/m3) considered available for decomposition.
position of limestone created a surface area of 17 m2/g attributed The value of the initial surface area may be the external surface of
to pores in the range of 10–100 nm (Fig. 1b). The lime’s surface the particle, as for the Shrinking Core Model (SCM) or the SBET if
area is close to that predicted by Mai and Egdar (Mai and Edgar, the pores are considered to participate in the reaction. The SCM is
1989) after calcination and sintering, and similar pore volume dis- an extreme case, valid in cases such as for materials with very low
tributions were reported by Rodriguez-Navarro et al. (2009). porosity. On the contrary, when SBET is used, it is assumed that all
the pores’ surface area is available for reaction, as used in the Ran-
2.3. Particle models dom Pore Model (RPM) (Khinast et al., 1996), the Changing Grain
Model (CGM) (Garcı´a-Labiano et al., 2002), as well as, in the Uniform
The most commonly referred unit for the reaction rate is kmol/ Conversion model (UCM) (Borgwardt, 1985). In every case, the fitting
(m2 s) which can be transformed to m3m2s1 or m/s (by multiply- parameter was the reaction front velocity r which is a function only
ing with CaCO3 molar volume, VM,CaCO3) and this refers to the reac- of decomposition temperature and CO2 partial pressure. The three
tion front velocity (Stanmore and Gilot, 2005). It has been shown different models (CGM, UCM, RPM) are compared on their ability
by other researchers (Ingraham and Marier, 1963; Satterfield and to describe the experimental results obtained from the decomposi-
Feakes, 1959; Wang and Thomson, 1995) that decomposition reac- tion of limestone in the fixed-bed reactor.
tion takes place in a layer at the CaO/CaCO3 interface and this reac-
tion front propagates at a steady state velocity. This velocity is the 2.3.1. The Random Pore Model (RPM)
intrinsic reaction rate (r in m/s or m3CaCO3/m2s), which depends on The RPM developed by Bhatia and Perlmutter (1980) assumes
the CO2 partial pressure and temperature at the interface. The con- that the pores of the particles are an assembly of uniform size, ran-
version change rate (dX/dt in s1) is calculated by using the r value domly oriented cylinders. During the decomposition reaction the
and the available surface area (S in m2/m3) at each moment. For pore diameters start to increase, increasing also the total reaction
the description of the conversion evolution during reaction, a parti- surface area, until they start overlapping. At this point, the surface
cle model has to be used that can describe the reaction surface area area of the consuming material exhibits a maximum value and
changes. First of all, a basic assumption of constant reactivity is starts to decrease falling to zero when the entire reactant quantity
is consumed. The initial structural properties of the material are
introduced in the model by two dimensionless structural parame-
Table 1 ters (W and r), which determine the reactant surface area evolu-
Structural properties of raw limestone, limestone just before 1st calcination and lime
tion during reaction. The conversion versus time expression as
after 1st calcination.
given by Bhatia and Perlmutter (1980) is:
   
s 3
Property Raw Limestone just Lime after
Ws
limestone before 1st
X ðsÞ ¼ 1  1  exp s 1 þ ð1Þ
1st calcination calcination r 4
2
SBET surface area (m /g) 1.20 0.57 17.0
where s is the dimensionless time and r and W are dimensionless
Micropores surface area (m2/g) 0.57 0.39 1.70
Meso/macropores surface area 0.63 0.18 15.3 parameters:
(m2/g)
rS0 t
Pore Volume, PV0 (cm3/g) 0.013 0.0017 0.173 s¼ ð2Þ
Mean pore radius, r0 (nm) 23.0 8.8 20.3 1  e0
A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071 5

dp S0 rate of CaCO3 at each time is proportional to the SBET of the unde-


r¼ ð3Þ
2ð1  e0 Þ composed material ((CaCO3) in mol) as follows:
Z Z
1a
dðCaCO3 Þ t
4pL0 ð1  e0 Þ ¼ rS0 0 dt ð10Þ
W¼ ð4Þ 1 ðCaCO3 Þ 0
S0 2
The initial total pore length L0, the initial pore surface area S0 which leads to the expression of calcination conversion as a func-
and the initial porosity e0 of the material are calculated using the tion of time:
pore volume distribution vo(r) obtained from the BJH measurement X ðtÞ ¼ 1  exp ðrS0 0 t Þ ð11Þ
as follows:
Z The surface area S’0 refers to the surface area per volume of
1 1
v 0 ðrÞ CaCO3 and not per particle’s volume as for S0. It is calculated by
L0 ¼ dr ð5Þ
p 0 r2 S’0 = 106  SBET  qb.
Z 1
v 0 ðrÞ
S0 ¼ 2 dr ð6Þ 3. Results and discussion
0 r
Z 3.1. Effect of transfer phenomena
1
e0 ¼ v 0 ðrÞdr ð7Þ
0 The study of the intrinsic kinetics of a reaction should be
Note that S0 and L0 refer to the surface area and pore length per accompanied with an evaluation of transfer phenomena in order
unit volume of particle and not only of CaCO3. to define appropriate experimental conditions for minimizing their
The estimated values of the above parameters used in the RPM effect. The parameters examined are the total gas flow rate, the
model are summarized in Table 2 for the material just before cal- limestone particles’ size, as well as, the bed dilution. The gas flow
cination reaction: rate and the particle size affect the Reynolds number, which in turn
The RPM has rarely been used for the decomposition reaction of is used for the calculation of mass and heat transfer coefficients
CaCO3. Khinast et al. (1996) modified it to model their experimen- through the Sh and Nu numbers respectively. By increasing the
tal results. flow rate and decreasing the particle size, the external mass and
heat transfer processes become faster. Additionally, decreasing
the particles’ size, the species concentration gradient inside the
2.3.2. The Changing Grain Model (CGM)
pores also decreases, until it is minimized and can be approxi-
The CGM firstly proposed by Szekely and Evans (1970) mated by using the value at the particle’s external surface. The
describes the porous structure as the cavities between a set of
bed density affects the bulk concentration of CO2. During calcina-
impervious grains of uniform initial radius calculated using the SBET tion, the evolved CO2 can increase the total CO2 concentration in
of the material. During the reaction, the grains follow the shrinking
the bulk gas affecting the decomposition rate of the neighboring
core mode and the total surface area of the reactant decreases particles. This can be diminished by diluting the sample using an
monotonically. Garcı́a-Labiano et al. (2002) used this model for
inert material. The dilution can maintain isothermal operation
the description of the decomposition of limestone and dolomite. inside the reactor bed, a necessary condition for endothermic reac-
According to CGM, the conversion of calcination is expressed as:
tions, such as limestone decomposition.
 3 Fig. 2 shows the effect of these three parameters on the reaction
rS0 t
X ðt Þ ¼ 1  1  ð8Þ rate. Increasing the GHSV from 22,500 to 45,000 h1 (Fig. 2a), has
3ð1  e0 Þ
negligible effect on the decomposition of limestone at 885 °C and
The initial porosity can be calculated using the initial total pore under 10.1 kPa of CO2 partial pressure. Increasing the bed dilution
volume PV0 of the material and the theoretical density of calcite qb by decreasing the mass ratio of sample/quartz from 1/15 up to 1/60
using the formula: produced the same result regarding the rate (Fig. 2b). This implies
that the sample density inside the reactor bed does not affect the
PV 0 qb
e0 ¼ ð9Þ rate and so the reaction does not substantially influence the bulk
1 þ PV 0 qb gas CO2 concentration. Finally, testing the effect of limestone par-
The surface area S0 in m2/m3 is estimated from the specific sur- ticles’ size on the rate it was concluded that the reaction was faster
face area by S0 = 106  SBET  qb  (1  e0), where SBET is the specific when decreasing the diameter from 500 down to 100 lm. Beyond
surface area (in m2/g) of limestone before calcination and that limit (dp < 100 lm) the internal resistances are minimized and
qb = 2.71 g/cm3 the bulk density of calcite. the reaction rate remains constant (Fig. 2c). Other authors have
also reported that internal mass transfer is negligible for particles
smaller than 90 lm (Escardino et al., 2013). The relative impor-
2.3.3. The Uniform Conversion model (UCM)
tance of the transfer phenomena compared to the reaction depends
This model was used by Borgwardt (1985) for the calcination of
on the applied conditions, which determine the limiting stage. Hu
1–90 lm limestone particles. According to UCM the decomposition
and Scaroni (Hu and Scaroni, 1996) examined the calcination of 6–
90 lm limestone in a drop tube furnace and found significant resis-
Table 2 tances due to mass and heat transfer. They observed a conversion
Parameters values used in the RPM model for the material just before calcination. gradient inside 63 lm particles after partial decomposition at
Parameter Limestone just before 1st calcination
1200 °C. On the contrary, Borgwardt (Borgwardt, 1985) did not find
any influence on the rate of 1–90 lm particles’ decomposition
Initial pore length, L0 (m/m3) 1.94  1013
Initial surface area, S0 (m2/m3) 652,600
under pure nitrogen and up to the 1000 °C. Accordingly, the exper-
Initial porosity, e0 (m2/m3) 0.0043 imental design parameters for the investigation of the intrinsic cal-
Particle diameter, dp (m) 6  105 cination rate in the present work were chosen as follows: particle
r (–) 19.7 size range = 45–75 lm, GHSV = 22,500 h1 and sample/quartz
W (–) 570
mass ratio = 1/15.
6 A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071

Fig. 2. Effect of the GHSV (a), bed dilution (b) and particle size (c) on calcination rate of raw limestone.

3.2. Particle models comparison The conversion-time curve is fitted well by the CGM and UCM
with CGM being slightly better, while the RPM produces significant
The three different particle models, namely the RPM (Eq. (1)), discrepancies. This is also evident when plotting the conversion
the CGM (Eq. (8)) and the UCM (Eq. (11)), are assessed on how well change rate (dX/dt) vs time in Fig. 3b, where although the experi-
they can describe the calcination reaction of limestone in the fixed- mental curve as well as the RPM exhibit a maximum of the rate,
bed reactor. For each model, the fitting parameter was the reaction the CGM and UCM fit better the data during almost the whole reac-
front velocity r. Fig. 3a shows the conversion versus time curve of tion time. According to the RPM, the reaction surface increases as
the experimental data along with the fitting curves of each of the the pores grow during the first moments of the decomposition
aforementioned particle models. until the pore overlap starts to dominate decreasing the available

Fig. 3. Comparison of RPM, CGM and UCM on describing the decomposition of raw limestone at T = 885 °C and PCO2 = 30.4 kPa. Conversion (X) vs time (a) and conversion
change rate (dX/dt) vs time (b).
A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071 7

Table 3 surface area of the initial material was calculated using the BJH
Comparison of r values derived from fitting of RPM, CGM and UCM on the data, while in the CGM and UCM the BET value was introduced.
experimental data of raw limestone’s decomposition (T = 885 °C and PCO2 = 30.4 kPa).
In the next sections, both CGM and UCM are used for the kinetic
RPM CGM UCM modeling of conversion-time data.
r (nm/s) 8.3 37.4 47.4

3.3. Effect of temperature

reaction surface area. The latter two models (CGM, RPM) predict The effect of reaction temperature on the rate was assessed for
that the surface area is decreasing during the whole reaction time. limestone’s decomposition under different CO2 partial pressures of
The experimental data exhibit a maximum of the rate, but this 10.1, 20.3 and 30.4 kPa as shown in Fig. 4. For each case the CGM
increase happens for only few seconds in the beginning of reaction, and UCM were fitted to the experimental data, denoted by the solid
and subsequently the rate follows a monotonic decrease, which and dashed lines respectively in the same graph. As expected, by
CGM and UCM fit better. None of the assessed particle models increasing the temperature, the decomposition reaction needs less
describe perfectly the conversion-time experimental curve of lime- time to be completed under the same CO2 partial pressure. Regard-
stone. This is probably produced from the assumptions made by ing the fitting to the experimental data, the two models adequately
each model. The RPM supposes cylindrical pores, while the Grain follow the evolution of conversion in all cases.
Model spherical grains. The UCM implies that the surface area of
the reactant is proportional to the conversion. In reality the grains 3.4. Effect of CO2 partial pressure
or the pores do not follow these ideal shapes. A more sophisticated
model taking into account the pores or grains shape would possi- The influence of different CO2 concentrations in the bulk gas on
bly have better fitting. Nevertheless, the development of such a the limestone’s calcination rate was also evaluated experimentally
model is not in the scope of the present work. at various temperatures, as depicted in Fig. 5. It should be noted
Each of the three models uses a different value of reaction front here that no induction period was detected at the beginning of
velocity r for the same experimental data. The values of r obtained the reaction under all the studied conditions. This comes in con-
from the fitting procedure of each model are summarized in trast with Valverde (2015) who observed around 4 min of induc-
Table 3. tion period at 885 °C and 60 kPa of CO2.
These values indicate that for the same experimental data the It is obvious that the time needed for complete decarbonation
obtained value of r can significantly vary depending on the model does not vary linearly with the partial pressure. Instead, as the
used. The lower value of r derived by the RPM is due to that the PCO2 increases, the time for complete conversion is more than that

Fig. 4. Effect of temperature on the decomposition rate under 10.1 kPa (a), 20.3 kPa (b) and 30.4 kPa (c) CO2 partial pressure. Experimental values (symbols) and the CGM and
UCM fitting curves (solid and dashed lines respectively).
8 A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071

Fig. 5. Effect of CO2 partial pressure on the decomposition rate under different temperatures: 825 °C (a), 845 °C (b), 870 °C (c) and 885 °C (d). Experimental values (symbols)
and the CGM and UCM fitting curves (solid and dashed lines respectively).

expected from the linear trend. The fitting of CGM and UCM to the 3.5.1. Langmuir-Hinshelwood model
experimental results is satisfactory for all the cases and during the The step of CO2 desorption in the reaction sequence of CaCO3
whole reaction time period, while UCM exhibits small deviations decomposition mechanism has been mentioned by other authors
for high conversion values. in the past (Dennis and Hayhurst, 1987; Garcı´a-Labiano et al.,
2002; Wang and Thomson, 1995). Based on that, a Langmuir–Hin-
shelwood mechanistic model was used to describe the experimen-
3.5. Mechanistic models tal results. This includes the chemical decomposition of CaCO3
during the first stage, wherein the produced CO2 molecule is firstly
The CGM and UCM expressions (Eqs. (8) and (11) respectively) in an adsorbed state on n active sites (S):
were fitted (as shown in Figs. 4 and 5) for each experimental result
(X vs t curve) and the reaction rate r value was obtained as the only ð14Þ
fitting parameter. The reaction rate r is a function of temperature
and CO2 partial pressure as follows: where n is the number of active sites needed for one molecule of
CO2 to adsorb and it usually varies between 0.5 and 2 (Garcı́a-
    Labiano et al., 2002). In a second step, desorption of CO2 takes place
Eai PCO2
r ðT; PCO2 Þ ¼ V M;CaCO3 k0i exp  f ð12Þ as follows:
RT Pe
ð15Þ
where Pe is the equilibrium CO2 partial pressure at the reaction
temperature and is calculated by: According to the above mechanism and assuming that the
decomposition (Eq. (14)) is the rate determining step, the expres-
20474 sion that gives the reaction front velocity is:
Pe ¼ 413:7  107 e T ð13Þ
 
P CO2
Thus, the reaction rate contains an Arrhenius term with the pre- r ¼ V M;CaCO3 k1LH ð1  hÞ 1  ð16Þ
exponential factor k0i and the activation energy Eai of calcination
Pe
reaction, and a function of PCO2 and Pe. The subscript i is introduced where h is the fraction of active sites occupied by CO2 molecules.
for the different mechanisms assessed later. As mentioned before, Two different adsorption isotherms were used to calculate the sur-
various mechanisms have been proposed for the effect of CO2 on face coverage h, namely the Langmuir and the Freundlich isotherm.
the decomposition rate, but without consensus. Herein, we apply The h values for these two cases are
and compare all these mechanisms to the experimental results
KAL PCO2
obtained from decomposition of limestone in the fixed-bed h¼ for Langmuir isotherm ð17Þ
apparatus. 1 þ KAL PCO2
A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071 9

 k2HPCO2 k3H + k4H, transforming the rate expression as


follows:
  
k3H PCO2
r ¼ V M;CaCO3 k1H 1 ð24Þ
k3H þ k4H Pe
Eq. (24) is a linear relation between r and PCO2 which is not in
agreement with the experimental data, thus the hypothesis of
the decomposition reaction as the rate determining step is not
valid in that case. If the slow stage is the structural transformation
Fig. 6. Structural transformations during the decomposition reaction (adopted from (k1H,k2Hk3H,k4H) the reaction rate would be given by:
Valverde and Medina (2015)). ! 
k2H
P
k1H CO2 PCO2
r ¼ V M;CaCO3 k3H 1  k2H
1 ð25Þ
1 þ k1H PCO2 Pe
1=n
h ¼ KAF PCO2 for Freundlich isotherm ð18Þ
which is very similar to the one obtained using the Langmuir-
Hinshelwood model (Eq. (16)) with Langmuir adsorption iso-
3.5.2. The Hyatt model therm. Moreover, the structural transformation cannot be the
A two stage-model was proposed by Hyatt et al. (1958). Accord- limiting step in the whole temperature and PCO2 range. As
ing to this mechanism, during the first stage (Eq. (19)), decomposi- pointed out by Valverde (2015), the relaxation of active CaO*
tion of CaCO3 takes place leading to the formation of gaseous CO2 to the stable cubic form affects the rate at high CO2 partial pres-
and an active form of CaO, denoted as CaO*. This is an unstable sures and close to equilibrium, which is not the case in our
intermediate between the calcite’s initial rhombohedral form and experiments. Consequently, the Hyatt model cannot describe
the final cubic CaO. the calcination reaction mechanism convincingly. The mathe-
matical formula is similar to the one of L-H model but the mech-
ð19Þ anistic step of CO2 adsorption provides a more solid theoretical
explanation. The sorption/desorption of CO2 on CaO was recently
demonstrated experimentally using infrared reflection-
The concentration of CaO* on the reaction surface is supposed
absorption spectroscopy (IRAS) and computationally applying
to increase and reach a maximum value under steady state condi-
density functional theory (DFT) calculations (Solis et al., 2017;
tions. Thus, part of the reaction front (designated as u) is in the
Sun et al., 2016).
form of CaO* and the remaining (1-u) is occupied by CaCO3, indi-
cating that not all the CaCO3 is involved in the reaction. In a second
3.5.3. Langmuir-Hinshelwood model with structural transformation
stage (Eq. (20)), the relaxation of CaO* to cubic CaO proceeds, a
The combination of Langmuir-Hinshelwood model with the
transformation that refers to the solid’s structure change with no
structural transformation of CaO* to CaO during desorption of
accompanied weight loss. The ‘‘H” subscript is used here to denote
CO2 leads to the following reactions:
the Hyatt’s model.
ð26Þ
ð20Þ

As Hyatt et al. (1958) mentioned, reaction (20) is a diffusional ð27Þ


process whereby CaO* leaves its position at the reaction surface
and becomes part of the crystalline or bulk CaO away from the sur- Then, following the same procedure assuming the first step as
face. Since this process is assumed to be reversible, the opposite the rate controlling, the reaction rate is expressed as:
reaction may also occur.  
PCO2
The CaO* active form has been also mentioned later by other r ¼ V M;CaCO3 k1LH ð1  hÞ 1  ð28Þ
Pe
researchers (Manuel and Medina, 2017; Rodriguez-Navarro et al.,
2009; Valverde et al., 2015; Valverde and Medina, 2015) (see where h now is given by:
Fig. 6). If steady state is hypothesized during the reaction, the 0

activity of CaO* u will be constant and the net reaction rate will K PCO2 0 KAL
h¼ where K ¼ ð29Þ
be given by:
0
1 þ K PCO2 u
This leads to the same formula as for L-H model without struc-
1  PPCO2
r ¼ V M;CaCO3 e
ð21Þ tural transformation, but the apparent constant (K’) contains –
BPCO2 þ 1
R0 apart from the adsorption constant (KAL) – the activity of meta-
stable CaO* (u).
k1H k3H k2H k1H k3H The L-H model with the structural transformation of CaO* was
where R0 ¼ ; B¼ ; Pe ¼ ð22Þ
k1H þ k3H þ k4H k1H k3H k2H k4H used to describe the rate values derived from the fitting of UCM
and CGM on the experimental conversion-time curves. Both the
R0 is the reaction rate when decomposition takes place in the
Langmuir and Freundlich isotherms were assessed as shown in
absence of CO2. If the decomposition of CaCO3 is the slow step, then
Fig. 7.
As it was discussed earlier, the values of r derived from the UCM
 k1H k3H + k4H and R0 is simplified to:
are higher compared to that from CGM and this is the reason for
the differences observed in the data shown in Fig. 7 (symbols).
k1H k3H
R0 ¼ ð23Þ The Langmuir-Hinshelwood model with both adsorption modes
k3H þ k4H (using n = 1 for Freundlich isotherm) was fitted to the
10 A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071

Fig. 7. Reaction rate r derived from UCM (a) and CGM (b) fitting versus CO2 partial pressure at various calcination temperatures with the Langmuir-Hinshelwood model using
the Langmuir (n = 1, solid lines) and the Freundlich (n = 1, dashed lines) isotherms.

experimental data (Fig. 7). Langmuir isotherm follows adequately adsorption and CaO* formation) with values of 1.38  106 (kPa)1
the experiments in all cases, while the Freundlich isotherm fails and 4.52 * 106 (kPa)1 derived using UCM and CGM respectively.
to predict zero decomposition rate for PCO2 = Pe at 870 and Garcı´a-Labiano et al. (2002) found a decomposition activation
885 °C. The same result was derived using n = 2, thus only the energy of 166 kJ/mol and the enthalpy of adsorption equal to
Langmuir isotherm was used in the following calculations of the 93 kJ/mol using the Freundlich isotherm.
kinetic and thermodynamic parameters. The decomposition rate
constants and the adsorption constants of Eqs. (28) and (29) were
3.5.4. Exponential expression
obtained from the fitting procedure and their dependence on tem-
Other researchers (Khinast et al., 1996; Okunev et al., 2008)
perature is assessed in Fig. 8.
have found an exponential decay of the calcination rate when
The decomposition rate constants follow the Arrhenius depen-
increasing the CO2 partial pressure. This is an empirical relation
dence on temperature using the Langmuir isotherm (Fig. 7a), lead-
as it is not based on a certain reaction mechanism. The exponential
ing to decomposition activation energies of 210 ± 17 and
expression was also found to adequately fit our experimental data
217 ± 29 kJ/mol, which correspond to the UCM and CGM fittings
as shown in Fig. 9. It is obvious that this model cannot predict a
respectively. The pre-exponential factors were calculated equal
zero value for the reaction front velocity when PCO2 = Pe, as the
to 18,900 kmol/(m2 s) for the UCM and 31,900 kmol/(m2 s) for
exponential equation approaches asymptotically the x axis.
the CGM. The activation energy values are slightly higher than
The expression for the reaction front velocity using the expo-
other referred in the literature. A widely adopted value is 205 kJ/-
nential model is:
mol found by Borgwardt (1985), who studied decomposition of
limestone under pure nitrogen. The pre-exponential factor he  
PCO2
reported was 51,000 kmol/(m2 s), which is higher compared to that r ¼ V M;CaCO3 k1E exp K E ð30Þ
Pe
found in the present work but in the same order of magnitude. The
adsorption constants (K’) of the Langmuir model (Fig. 7b) is also The temperature dependence of the constants kE and KE are
consistent with Van’t Hoff expression giving an adsorption shown in Fig. 10:
enthalpy equal to 101 ± 25 kJ/mol and 91 ± 12 kJ/mol, from The activation energy of CaCO3 decomposition reaction was
UCM and CGM respectively. This enthalpy term contains, apart found equal to 216 ± 29 kJ/mol derived by the UCM and 181 ± 9
from the adsorption enthalpy, the enthalpy contribution of the for CGM fitting, according to the exponential model. The constant
metastable CaO* formation. The pre-exponential factor of K’ is KE also follows the Van’t Hoff behavior with an apparent enthalpy
0
equal to exp(DS0 ) (again including the entropy terms of CO2 term of 42 ± 12 kJ/mol and a pre-exponential term equal to 327 for

Fig. 8. Arrhenius and Van’t Hoff diagrams of decomposition rate (a) and adsorption constants (b) derived from the L-H model (Langmuir isotherm) on the values obtained
from the UCM and CGM fitting on the experimental conversion-time curves.
A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071 11

Fig. 9. Reaction rate r derived from UCM (a) and CGM (b) fitting versus CO2 partial pressure at various calcination temperatures with the exponential model (solid lines).

Fig. 10. Arrhenius and Van’t Hoff diagrams of decomposition rate constant k1E (a) and exponential constant KE (b) derived from the exponential model.

the UCM. On the other hand, KE from CGM is not consistent with chosen to be validated on predicting the experimental results of
the Van’t Hoff expression (R2 = 0.25) which may be due to the nat- other authors. Due to deviations both in the pre-exponential factor
ure of the model and the produced errors from the sequential fit- and the activation energy, it is more convenient for comparison to
ting procedures. It is reminded that this is an empirical model use the rate at a certain temperature and CO2 partial pressure. In
and these values have no theoretical background. Table 5, highly cited works on calcination kinetics are summarized.
The rate at 780 °C and in the absence of CO2, using our model (L-H
3.6. Validation of the model from UCM), was calculated and compared to the values proposed
by other researchers.
In the previous sections two particle models (UCM and CGM) The value is higher than the values presented by Khinast et al.
were used to fit the experimental data curves of conversion versus (1996) and Gracia-Labiano et al. (2002). However, it is five times
time obtained after calcination of limestone under different CO2 lower compared to the values reported by Borgwardt (1985)
partial pressures and temperatures in a fixed-bed reactor. The val- (Table 5). Variations can arise from incorrect measurement of the
ues of the rate derived from UCM fitting were slightly higher com- reaction temperature during the experiments. In most cases (in-
pared to that using CGM. These two sets of rate values were fitted cluding our work) it is very difficult to measure the temperature
by a Langmuir-Hinshelwood mechanistic model, which included on the reaction spot. This can produce errors and rate underestima-
the decomposition of CaCO3, the desorption of CO2 (described by tion as the actual reaction temperature is lower than the one mea-
a Langmuir type isotherm), and the CaO* solid state transformation sured due to the endothermic decomposition.
to stable CaO. Additionally, an empirical exponential model was The predictive ability of model was checked with literature data
also tested, as it has been used previously by other authors. A relevant to calcination in the presence of CO2. Two recent works
decomposition reaction rate constant and a K constant have been from the literature were chosen with different time scales of reac-
derived for each model and presented in Table 4 with their Arrhe- tion, materials and experimental configurations. Maya et al. (2018)
nius and Van’t Hoff parameters values. studied the decomposition of impervious limestone particles with
As discussed previously the Langmuir-Hinshelwood model with reaction times in the order of a few minutes. The conversion-time
a Langmuir isotherm and the transformation of CaO* to CaO was diagram that they obtained experimentally is shown in Fig. 11a
found to describe adequately the experimental results obtained (symbols), and the dotted and solid lines represent the prediction
here, as well as the literature experimental findings regarding of the UCM-L-H and CGM-L-H models. The model is in good agree-
the CaO* formation. On the contrary, the exponential model is ment with the experiment. Also, Fernandez et al. in their recent
not supported by a theory despite of fitting well the experimental work (Fernandez et al., 2019), studied the decomposition rate of
results obtained here in most cases. For this reason the former was porous limestone particles in a pilot-scale drop-down tube reactor.
12 A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071

Table 4
Values of constants for the calcination reaction derived from the L-H and the exponential model.

UCM CGM
2 2
L-H Langmuir isotherm with CaO* relaxation k1L (kmol/(m s)) k01L (kmol/(m s)) 18,860 31,940
E1L (kJ/mol) 210 ± 17 217 ± 29
0
K’ (1/kPa) K0 (106/kPa) 1.38 4.52
0
DΗ0 (kJ/mol) 101 ± 25 91 ± 13
Exponential model k1E (kmol/(m2 s)) k01E (kmol/(m2 s)) 34,000 640
E1E (kJ/mol) 216 ± 29 181 ± 9
KE (–) K0E (–) 327
DΗE (kJ/mol) 42 ± 12

Table 5
Comparison of the results with other works in the literature.

Ref Experimental apparatus Material Temperature PCO2 Particle CO2 dependence r’ (at 780 °C, Activation
range (°C) range model PCO2 = 0 kPa) energy
(kPa) (kmol (kJ/mol)
m2s1)
Khinast et al. TGA Limestone (5–100 lm) 780 <6.6 RPM Exponential 2.03 107 –
(1996)
Borgwardt Differential reactor and Limestone (1–90 lm) 475–1000 – UCM – 3.57 10 6
205
(1985) entrained flow reactor
6
Ingraham and TGA Crystals and compact 790–850 in air, <47.6 – Linear 4.9 10 169.9
Marier pellets of precipitated 850, 835 in CO2/
(1963) CaCO3 air
Garcı́a- TGA Limestone 775–900 <81.1 CGM Two stage model 6.4 108 114
Labiano with CO2 sorption
et al.
(2002))
This work Fixed-bed reactor Limestone (45–75 lm) 825–885 <60.8 UCM L-H Langmuir 0.72 106 210
isotherm

In their experiments the calcination time was of the order of a few The effect of thermal annealing before decomposition on CaCO3
seconds. The conversion-time data for calcination under 30% CO2 structural properties was proved significant decreasing the mate-
are depicted in Fig. 11b (symbols). Again, the UCM-L-H model rial’s surface area due to sintering above Tammann temperature.
developed in the present study predicts adequately the experimen- The effect of temperature and CO2 partial pressure on the reaction
tal results. rate was evaluated in a series of tests. It was found that the decom-
position rate is enhanced by increasing the reaction temperature
and varies non-linearly with CO2 partial pressure.
4. Conclusions
Three different particle models were assessed for their ability to
describe the conversion-time experimental curves, namely the
Calcination kinetics of limestone were studied experimentally
Random Pore Model (RPM), the Uniform Conversion Model
and theoretically. The decomposition experiments were conducted
(UCM) and the Changing Grain Model (CGM). It was found that
in a continuous flow fixed-bed reactor at temperatures between
the UCM and CGM follow better the experimental data, while
825 and 885 °C and CO2 partial pressures <60.8 kPa. In order to
RPM deviates. The values of the rate derived from UCM were
measure the intrinsic kinetics, the experimental conditions were
slightly higher compared to that using CGM. These two sets of rate
chosen so that the internal and external transfer phenomena to
values were fitted by a Langmuir-Hinshelwood mechanistic model,
be fast enough not affecting the overall rate.

Fig. 11. UCM-L-H and CGM-L-H model prediction curves (solid lines) on Maya et al. (2018) experimental results (dots) (a) and UCM-L-H prediction curve on Fernandez et al.
(2019) measurements.
A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071 13

which include the decomposition of CaCO3, the desorption of CO2 Bhatia, S.K., Perlmutter, D.D., 1980. A random pore model for fluid-solid reactions: I.
Isothermal, kinetic control. AIChE J. 26, 379–386. https://doi.org/10.1002/
represented by a Langmuir type isotherm, and the CaO* solid state
aic.690260308.
transformation to stable CaO. Additionally, an empirical exponen- Borgwardt, R.H., 1985. Calcination kinetics and surface area of dispersed limestone
tial model was also tested, as it has been used previously by other particles. AIChE J. 31, 103–111. https://doi.org/10.1002/aic.690310112.
authors. The L-H model, which is based on experimental evidences Broda, M., Manovic, V., Imtiaz, Q., Kierzkowska, A.M., Anthony, E.J., Müller, C.R.,
2013. High-purity hydrogen via the sorption-enhanced steam methane
regarding the CaO* formation and CO2 –adsorption, described ade- reforming reaction over a synthetic CaO-based sorbent and a Ni catalyst.
quately the experimental results obtained here. On the contrary, Environ. Sci. Technol. 47, 6007–6014. https://doi.org/10.1021/es305113p.
the exponential model is not supported by a theory despite of fit- Chacartegui, R., Alovisio, A., Ortiz, C., Valverde, J.M., Verda, V., Becerra, J.A., 2016.
Thermochemical energy storage of concentrated solar power by integration of
ting well the results in most cases. The activation energy calculated the calcium looping process and a CO2 power cycle. Appl. Energy 173, 589–605.
via the L-H model was equal to 210 ± 17 kJ/mol. The model was https://doi.org/10.1016/j.apenergy.2016.04.053.
validated by predicting the decomposition rates measured by other Darroudi, T., Searcy, A.W., 1981. Effect of carbon dioxide pressure on the rate of
decomposition of calcite (CaCO3). J. Phys. Chem. 85, 3971–3974. https://doi.org/
authors using different materials, experimental apparatus and 10.1021/j150626a004.
reaction time scales. The predictions were adequately accurate in Dennis, J.S., Hayhurst, A.N., 1987. the effect of CO2 on the kinetics and extent of
all cases. calcination of limestone and dolomite particles in fluidised beds. Chem. Eng. Sci.
42, 2361–2372. https://doi.org/10.1016/0009-2509(87)80110-0.
Di Giuliano, A., Gallucci, K., 2018. Sorption enhanced steam methane reforming
CRediT authorship contribution statement based on nickel and calcium looping: a review. Chem. Eng. Process. – Process
Intensificat. 130, 240–252. https://doi.org/10.1016/j.cep.2018.06.021.
Edwards, S.E.B., Materić, V., 2012. Calcium looping in solar power generation plants.
Athanasios Scaltsoyiannes: Investigation, Methodology, Soft- Sol. Energy 86, 2494–2503. https://doi.org/10.1016/j.solener.2012.05.019.
ware, Formal analysis, Validation, Writing - original draft. Angeliki Erans, M., Manovic, V., Anthony, E.J., 2016. Calcium looping sorbents for CO2
Lemonidou: Conceptualization, Methodology, Validation, Writing capture. Appl. Energy. https://doi.org/10.1016/j.apenergy.2016.07.074.
Escardino, A., García-Ten, J., Feliu, C., Saburit, A., Cantavella, V., 2013. Kinetic study
- review & editing, Resources, Supervision, Project administration, of the thermal decomposition process of calcite particles in air and CO2
Funding acquisition. atmosphere. J. Ind. Eng. Chem. 19, 886–897. https://doi.org/10.1016/j.jiec.2012.
11.004.
Fernandez, J.R., Turrado, S., Abanades, J.C., 2019. Calcination kinetics of cement raw
Declaration of Competing Interest meals under various CO 2 concentrations. React. Chem. Eng. 4, 2129–2140.
https://doi.org/10.1039/C9RE00361D.
The authors declare that they have no known competing finan- Garcı́a-Labiano, F., Abad, A., de Diego, L.F., Gayán, P., Adánez, J., 2002. Calcination of
calcium-based sorbents at pressure in a broad range of CO2 concentrations. Chem.
cial interests or personal relationships that could have appeared
Eng. Sci. 57, 2381–2393. https://doi.org/10.1016/S0009-2509(02)00137-9.
to influence the work reported in this paper. Gavals, G.R., 1980. A random capillary model with application to char gasification at
chemically controlled rates. AIChE J. 26, 577–585. https://doi.org/10.1002/
aic.690260408.
Acknowledgments
Hanak, D.P., Michalski, S., Manovic, V., 2018. From post-combustion carbon capture
to sorption-enhanced hydrogen production: a state-of-the-art review of
This research has received funding from the European Union carbonate looping process feasibility. Energy Convers. Manage. 177, 428–452.
HORIZON 2020 project SOCRATCES SOlar Calcium looping integRA- https://doi.org/10.1016/j.enconman.2018.09.058.
Hills, A.W.D., 1968. The mechanism of the thermal decomposition of calcium
tion for Thermo Chemical Energy Storage, under Grant Agreement carbonate. Chem. Eng. Sci. 23, 297–320. https://doi.org/10.1016/0009-2509(68)
Number: 727348. We thank Dr Mark Sceats and Dr Tom Hills of 87002-2.
Calix Europe Limited for critical reading of the manuscript and Hu, N., Scaroni, A.W., 1996. Calcination of pulverized limestone particles under
furnace injection conditions. Fuel 75, 177–186. https://doi.org/10.1016/0016-
valuable suggestions. 2361(95)00234-0.
Hyatt, E.P., Cutler, I.B., Wadsworth, M.E., 1958. Calcium carbonate decomposition in
carbon dioxide atmosphere. J. Am. Ceramic Soc. 41, 70–74. https://doi.org/
References 10.1111/j.1151-2916.1958.tb13521.x.
Ingraham, T.R., Marier, P., 1963. Kinetic studies on the thermal decomposition of
Alovisio, A., Chacartegui, R., Ortiz, C., Valverde, J.M., Verda, V., 2017. Optimizing the calcium carbonate. Canadian J. Chem. Eng. 41, 170–173. https://doi.org/
CSP-calcium looping integration for thermochemical energy storage. Energy 10.1002/cjce.5450410408.
Convers. Manage. 136, 85–98. https://doi.org/10.1016/j.enconman.2016.12.093. Khinast, J., Krammer, G.F., Brunner, C., Staudinger, G., 1996. Decomposition of
Anthony, E.J. Ben, 2011. Ca looping technology: current status, developments and limestone: the influence of CO2 and particle size on the reaction rate. Chem.
future directions. Greenhouse Gases Sci. Technol. 1, 36–47. https://doi.org/ Eng. Sci. 51, 623–634. https://doi.org/10.1016/0009-2509(95)00302-9.
10.1002/ghg3.2. Li, Z., Cai, N., Yang, J., 2006. Continuous production of hydrogen from sorption-
Antzara, A., Heracleous, E., Bukur, D.B., Lemonidou, A.A., 2014. Thermodynamic enhanced steam methane reforming in two parallel fixed-bed reactors operated
analysis of hydrogen production via chemical looping steam methane in a cyclic manner. Ind. Eng. Chem. Res. 45, 8788–8793. https://doi.org/10.1021/
reforming coupled with in situ CO2 capture. Energy Procedia 63, 6576–6589. ie061010x.
https://doi.org/10.1016/j.egypro.2014.11.694. Lide, D.R., 2005. CRC Handbook of Chemistry and Physics 86TH Edition 2005–2006.
Antzara, A., Heracleous, E., Lemonidou, A.A., 2016. Energy efficient sorption CRC Press.
enhanced-chemical looping methane reforming process for high-purity H2 Mai, M.C., Edgar, T.F., 1989. Surface area evolution of calcium hydroxide during
production: experimental proof-of-concept. Appl. Energy 180, 457–471. https:// calcination and sintering. AIChE J. 35, 30–36. https://doi.org/10.1002/aic.
doi.org/10.1016/j.apenergy.2016.08.005. 690350103.
Antzara, A.N., Arregi, A., Heracleous, E., Lemonidou, A.A., 2018. In-depth evaluation Manuel, J., Medina, S., 2017. Limestone calcination under calcium-looping
of a ZrO 2 promoted CaO-based CO 2 sorbent in fluidized bed reactor tests. conditions for CO 2 capture and thermochemical energy storage in the
Chem. Eng. J. https://doi.org/10.1016/j.cej.2017.09.192. presence of H 2 O : an in situ XRD analysis. PCCP. https://doi.org/10.1039/
Basinas, P., Wu, Y., Grammelis, P., Anthony, E.J., Grace, J.R., Jim Lim, C., 2014. Effect of C7cp00260b.
pressure and gas concentration on CO2 and SO2 capture performance of Martínez, I., Grasa, G., Murillo, R., Arias, B., Abanades, J.C., 2012. Kinetics of
limestones. Fuel 122, 236–246. https://doi.org/10.1016/j.fuel.2014.01.038. calcination of partially carbonated particles in a Ca-looping system for CO 2
Benitez-Guerrero, M., Sarrion, B., Perejon, A., Sanchez-Jimenez, P.E., Perez-Maqueda, capture. Energy Fuels 26, 1432–1440. https://doi.org/10.1021/ef201525k.
L.A., Manuel Valverde, J., 2017. Large-scale high-temperature solar energy Martínez, I., Grasa, G., Parkkinen, J., Tynjälä, T., Hyppänen, T., Murillo, R., Romano, M.
storage using natural minerals. Sol. Energy Mater. Sol. Cells. https://doi.org/ C., 2016. Review and research needs of Ca-Looping systems modelling for post-
10.1016/j.solmat.2017.04.013. combustion CO2 capture applications. Int. J. Greenhouse Gas Control 50, 271–
Benitez-Guerrero, M., Valverde, J.M., Perejon, A., Sanchez-Jimenez, P.E., Perez-Maqueda, 304. https://doi.org/10.1016/j.ijggc.2016.04.002.
L.A., 2018a. Low-cost Ca-based composites synthesized by biotemplate method for Maya, J.C., Chejne, F., Gómez, C.A., Bhatia, S.K., 2018. Effect of the CaO sintering on
thermochemical energy storage of concentrated solar power. Appl. Energy 210, the calcination rate of CaCO 3 under atmospheres containing CO 2. AIChE J. 64,
108–116. https://doi.org/10.1016/j.apenergy.2017.10.109. 3638–3648. https://doi.org/10.1002/aic.16326.
Benitez-Guerrero, M., Valverde, J.M., Sanchez-Jimenez, P.E., Perejon, A., Perez- Okunev, A.G., Nesterenko, S.S., Lysikov, A.I., 2008. Decarbonation rates of cycled CaO
Maqueda, L.A., 2018b. Calcium-Looping performance of mechanically modified absorbents. Energy Fuels 22, 1911–1916. https://doi.org/10.1021/ef800047b.
Al2O3-CaO composites for energy storage and CO2 capture. Chem. Eng. J. Ortiz, C., Romano, M.C., Valverde, J.M., Binotti, M., Chacartegui, R., 2018. Process
https://doi.org/10.1016/j.cej.2017.11.183. integration of Calcium-Looping thermochemical energy storage system in
14 A. Scaltsoyiannes, A. Lemonidou / Chemical Engineering Science: X 8 (2020) 100071

concentrating solar power plants. Energy. https://doi.org/10.1016/j.energy. CaO and CaO regeneration. RSC Adv. 6, 39460–39468. https://doi.org/10.1039/
2018.04.180. C6RA05152A.
Ortiz, C., Valverde, J.M., Chacartegui, R., Perez-Maqueda, L.A., Giménez, P., 2019. The Szekely, J., Evans, J.W., 1970. A structural model for gas-solid reactions with a
calcium-looping (CaCO3/CaO) process for thermochemical energy storage in moving boundary. Chem. Eng. Sci. 25, 1091–1107. https://doi.org/10.1016/
concentrating solar power plants. Renew. Sustain. Energy Rev. 113, 109252. 0009-2509(70)85053-9.
https://doi.org/10.1016/j.rser.2019.109252. Takkinen, S., Saastamoinen, J., Hyppänen, T., 2012. Heat and mass transfer in
Papalas, T., Antzaras, A.N., Lemonidou, A.A., 2020a. Intensified steam methane calcination of limestone particles. AIChE J. 58, 2563–2572. https://doi.org/
reforming coupled with Ca-Ni looping in a dual fluidized bed reactor system: a 10.1002/aic.12774.
conceptual design. Chem. Eng. J. 382, 122993. https://doi.org/10.1016/j. Udomsirichakorn, J., Salam, P.A., 2014. Review of hydrogen-enriched gas production
cej.2019.122993. from steam gasification of biomass: the prospect of CaO-based chemical looping
Papalas, T., Antzaras, A.N., Lemonidou, A.A., 2020b. Evaluation of calcium-based gasification. Renew. Sustain. Energy Rev. 30, 565–579. https://doi.org/10.1016/j.
sorbents derived from natural ores and industrial wastes for high-temperature rser.2013.10.013.
CO 2 capture. Ind. Eng. Chem. Res. 59, 9926–9938. https://doi.org/10.1021/acs. Valverde, J.M., 2015. On the negative activation energy for limestone calcination at
iecr.9b06834. high temperatures nearby equilibrium. Chem. Eng. Sci. 132, 169–177. https://
Perejón, A., Romeo, L.M., Lara, Y., Lisbona, P., Martínez, A., Valverde, J.M., 2016. The doi.org/10.1016/j.ces.2015.04.027.
calcium-looping technology for CO2 capture: on the important roles of energy Valverde, J.M., Medina, S., 2015. Crystallographic transformation of limestone
integration and sorbent behavior. Appl. Energy 162, 787–807. https://doi.org/ during calcination under CO 2. PCCP 17, 21912–21926. https://doi.org/10.1039/
10.1016/j.apenergy.2015.10.121. C5CP02715B.
Rodriguez-Navarro, C., Ruiz-Agudo, E., Luque, A., Rodriguez-Navarro, A.B., Ortega- Valverde, J.M., Sanchez-Jimenez, P.E., Perez-Maqueda, L.A., 2015. Limestone
Huertas, M., 2009. Thermal decomposition of calcite: mechanisms of formation calcination nearby equilibrium: kinetics, CaO crystal structure, sintering and
and textural evolution of CaO nanocrystals. Am. Mineral. 94, 578–593. https:// reactivity. J. Phys. Chem. C 119, 1623–1641. https://doi.org/10.1021/jp508745u.
doi.org/10.2138/am.2009.3021. Wang, C., Chen, L., 2016. The effect of steam on simultaneous calcination and
Sarrion, B., Valverde, J.M., Perejon, A., Perez-Maqueda, L., Sanchez-Jimenez, P.E., sulfation of limestone in CFBB. Fuel 175, 164–171. https://doi.org/10.1016/
2016. On the multicycle activity of natural limestone/dolomite for j.fuel.2016.02.028.
thermochemical energy storage of concentrated solar power. Energy Technol. Wang, Y., Thomson, W.J., 1995. The effects of steam and carbon dioxide on calcite
4, 1013–1019. https://doi.org/10.1002/ente.201600068. decomposition using dynamic X-ray diffraction. Chem. Eng. Sci. 50, 1373–1382.
Satterfield, C.N., Feakes, F., 1959. Kinetics of the thermal decomposition of calcium https://doi.org/10.1016/0009-2509(95)00002-M.
carbonate. AIChE J. 5, 115–122. https://doi.org/10.1002/aic.690050124. Yan, Y., Wang, K., Clough, P.T., Anthony, E.J., 2020. Developments in calcium/chemical
Silcox, G.D., Kramlich, J.C., Pershing, D.W., 1989. A mathematical model for the flash looping and metal oxide redox cycles for high-temperature thermochemical
calcination of dispersed calcium carbonate and calcium hydroxide particles. energy storage: a review. Fuel Process. Technol. 199, 106280. https://doi.org/
Ind. Eng. Chem. Res. 28, 155–160. https://doi.org/10.1021/ie00086a005. 10.1016/j.fuproc.2019.106280.
Solis, B.H., Cui, Y., Weng, X., Seifert, J., Schauermann, S., Sauer, J., Shaikhutdinov, S., Ying, Z., Chuguang, Z., Zhaohui, L., Xuefeng, S., 2000. Modelling for flash calcination
Freund, H.-J., 2017. Initial stages of CO 2 adsorption on CaO: a combined and surface area development of dispersed limestone particles. Develop. Chem.
experimental and computational study. PCCP 19, 4231–4242. https://doi.org/ Eng. Min. Process. 8, 233–243. https://doi.org/10.1002/apj.5500080305.
10.1039/C6CP08504K. Zhao, M., Minett, A.I., Harris, A.T., 2013. A review of techno-economic models for the
Stanmore, B.R., Gilot, P., 2005. Review—calcination and carbonation of limestone retrofitting of conventional pulverised-coal power plants for post-combustion capture
during thermal cycling for CO2 sequestration. Fuel Process. Technol. 86, 1707– (PCC) of CO 2. Energy Environ. Sci. 6, 25–40. https://doi.org/10.1039/C2EE22890D.
1743. https://doi.org/10.1016/j.fuproc.2005.01.023. Zhong, Q., Bjerle, I., 1993. Calcination kinetics of limestone and the microstructure
Sun, Z., Wang, J., Du, W., Lu, G., Li, P., Song, X., Yu, J., 2016. Density functional theory of nascent CaO. Thermochim Acta 223, 109–120. https://doi.org/10.1016/0040-
study on the thermodynamics and mechanism of carbon dioxide capture by 6031(93)80125-T.

You might also like