Reduction and Oxidation Kinetics of a CaMn0.9Mg0.1O3−δ OxygenCarrier for Chemical-Looping CombustionLuis F. de Diego, Alberto Abad,* Arturo Cabello, Pilar Gayan, Francisco García-Labiano, and Juan Adanez

Department of Energy and Environment, Instituto de Carboquímica (ICB-CSIC), Miguel Luesma Castan 4, Zaragoza 50018, Spain

*S Supporting Information

ABSTRACT: The objective of this study was to establish the reaction kinetics involved in redox cycles of the CaMn0.9Mg0.1O3−δmaterial to be used as an oxygen carrier in the chemical-looping-combustion process. The oxygen-transport capacity andreactivity of this material during consecutive reduction and oxidation steps with gaseous compounds (CH4, H2, CO, and O2)were studied in a thermogravimetric analysis apparatus. The oxygen uncoupling properties of this material were also analyzed. Itwas found that material reactivity increased with the number of redox cycles, whereas the oxygen-transport capacity decreasedwith the cycle number until it was stabilized or “activated”. Conversion versus time curves at different temperatures (973−1273K) and reacting gas concentrations (5−60 vol %) were used for kinetic determination. The grain model with control by chemicalreaction and diffusion through the product layer was used to obtain the kinetic parameters for fresh and “activated” particles.

1. INTRODUCTIONAccording to the Intergovernmental Panel on Climate Change(IPCC), warming of the climate system is unequivocal and isinduced by human activities, as a consequence of the so-calledgreenhouse gases (GHGs) emitted to the atmosphere.1 CO2emissions from burning fossil fuels in power generation areconsidered to be the largest source of GHGs contributing toglobal warming. Therefore, it is generally accepted that areduction in CO2 emissions is necessary as soon as possible. CO2capture and storage appears to be a necessary option if significantreductions in CO2 emissions are to be achieved. In this regard,chemical-looping combustion (CLC) is suggested as one of thebest alternatives for reducing the economic and energy costsassociated with the obtention of a highly concentrated CO2stream from processes where a carbon-based fuel is used inpower generation.2,3

The CLC process is based on the transfer of oxygen from air tothe fuel by means of a solid oxygen carrier, thereby avoidingdirect contact between fuel and air. The oxygen carrier consists ofa metal oxide, MexOy, which often is supported on an inertmaterial in order to provide a porous structure and to provideparticle stability throughout the cycles. Figure 1 shows a general

scheme of this process. In a first step, the fuel is oxidized to CO2and H2O by the oxygen carrier, which is reduced to a metal or areduced form (MexOy−1). The reduced oxygen carrier is furtheroxidized with air in a second step, and the regenerated material isready to start a new cycle. To carry out the CLC process, most ofthe experimental CLC plants that exist worldwide use a

configuration composed of two interconnected fluidized-bedreactors, similar to the circulating fluidized bed (CFB) concept.Thus, conversion of the fuel happens in the so-called fuel reactor,while the oxygen carrier is regenerated in the so-called air reactor.The gas produced in the fuel reactor primarily contains CO2 andH2O. After water condensation and purification, a highlyconcentrated stream of CO2 is achieved that is ready fortransport and storage. CLC is therefore a combustion processwith inherent CO2 separation. Fuels such as gases, liquids, andsolids can be processed in a CLC system. A review of thedevelopment status of this technology can be found elsewhere.4

A key issue for the implantation of CLC technology in athermal power plant for energy generation with CO2 capture isthe development of suitable oxygen-carrier materials. Mainly, theoxygen carrier must allow high fuel conversion toward CO2 andH2O in order to be able to transfer a high enough oxygen massfraction of material that is highly reactive during repeatedreduction and oxidation cycles and resistant to attrition,fragmentation, and agglomeration phenomena. Moreover,additional issues such as toxicity, environmental impact, andcost are relevant during material selection.Materials based on one metal oxide are frequently proposed

for use as the oxygen carrier in CLC.4,5 Consequently, most ofthe oxygen-carrier materials to date are based on nickel, copper,iron, or manganese oxides. Copper and manganese materialshave received attention of late because they can generate gaseousoxygen at conditions existing in a fuel reactor. The chemicallooping with oxygen uncoupling (CLOU) process takesadvantage of this property in order to reach completecombustion of the fuel.6 A proof of concept for the CLOUprocess has been successfully demonstrated burning coal orbiomass with a copper-based oxygen carrier.7−9

Received: May 17, 2013Revised: August 23, 2013Accepted: November 22, 2013Published: November 22, 2013

Figure 1. General scheme of a CLC system for gaseous fuels.

Article

pubs.acs.org/IECR

© 2013 American Chemical Society 87 dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−103

Manganese-based particles can be considered a promisingmaterial because they fulfill most of the basic requirements to beconsidered an oxygen carrier.10 Thus, manganese oxidesupported on magnesia-stabilized zirconia was shown to operatesuccessfully with syngas as the fuel. However, this materialsuffered from lower reactivity with methane. Moreover, theoxidation state Mn2O3 could not be exploited for this kind ofmaterial because only oxidation from MnO to Mn3O4 waspossible at the temperature of interest to the CLC process.11 Itwould be very interesting to increase the oxidation state to MnIII

or even MnIV because the oxygen uncoupling property ofmanganese could be used; additionally, the oxygen-transportcapacity would be increased.6 ABO3−δ perovskite-type structures(B = Mn) have shown to improve the oxidation degree ofMn,12,13 and oxygen uncoupling behavior was observed.14

CaMnO3‑δ perovskite-type materials showed high reactivitywith methane.15,16 Nevertheless, CaMnO3−δ doped with severalcations, e.g., La, Mg, Ti, Fe, or Cu, is believed to increase thestability of the perovskite structure under CLC conditions.16

A pe rov sk i t e - t ype mat e r i a l w i t h compos i t i onCaMn0.875Ti0.125O3−δ was seen to have the desired thermody-namic properties during continuous operation in a CLC unit.17

Oxygen uncoupling behavior was evidenced, which was notpreviously found for Mn3O4/Mg−ZrO2 particles.10 Similarconversion of methane with CaMn0.875Ti0.125O3−δ and Mn3O4/Mg−ZrO2 particles was found in the same CLC unit at a similarfuel reactor temperature and for a similar amount of solids perthermal fuel energy produced. Thus, a combustion efficiency of99% was achieved in both cases at 1223 K and with an inventorycorresponding to 600 kg MWth

−1. However, changes in theperovskite structure were found that could result in a loss of theoxygen uncoupling property of this material.17

Results recently obtained at the laboratory scale with aperovskite-typematerial doped withmagnesium showed that thismaterial is suitable for an oxygen carrier.18,19 Tests with aCaMn0.9Mg0.1O3−δ material in continuously operating unitsshowed near-complete methane conversion.19,20 A combustionefficiency of 98% at 1173 K with a solids inventory of 600 kgMWth

−1 was achieved. Moreover, the oxygen carrier showedexcellent stability of the perovskite structure and goodfluidization behavior under a nonsulfurous environment. Forthese reasons, a CaMn0.9Mg0.1O3−δ material can be considered apromising oxygen-carrier material.Further development is needed in order to estimate the

relevant design parameters for use of the CaMn0.9Mg0.1O3−δmaterial in CLC, e.g., solids circulation flow rate or solidsinventory in the fuel and air reactors. These parameters dependon the oxygen-transport capacity and reactivity of the oxygencarrier. The use of proper kinetic data for oxygen-carrierreduction with reacting gases resulted in the successful validationof reactor models for CLC or chemical-looping-reformingsystems with copper-, nickel-, iron-, manganese-, and CaSO4-based oxygen carriers for both methane and coal combus-tion.21−34 It must be considered that the oxygen carrier will reactin different environments during its stage in the fluidized-bed fuelreactor. Therefore, to determine the kinetic parameters fromexperimental data obtained under different operating conditionsof temperature and gas concentration, it is necessary to makeproper use of the reaction kinetics in a reactor model.35

The objective of this study was to determine the kinetics ofboth reduction and oxidation reactions taking place in the CLCprocess using CaMn0.9Mg0.1O3−δ oxygen-carrier particles. Theselected material was studied in a thermogravimetric analysis

(TGA) apparatus, analyzing its reactivity with CH4, H2, and COas the reducing gases and O2 for oxidation reaction. The effect ofthe temperature and gas concentration on the reaction rate wasexamined. In addition, the oxygen uncoupling kinetics wasdetermined. In the future, kinetic data obtained in this work willbe implemented in a reactor model for the prediction, design, andoptimization purposes of a CLC system with CaMn0.9Mg0.1O3−δas the oxygen carrier.

2. EXPERIMENTAL SECTION

2.1. Material. The oxygen-carrier particles were manufac-tured by VITO (Flemish Institute for Technological Research,Belgium) using the spray-drying method. Raw materials were46.8 wt %Mn3O4, 50.5 wt % Ca(OH)2, and 2.7 wt %MgO. Afterpreparation, particles were calcined at 1573 K for 4 h. Thegeneral formula of the calcined particles was CaMn0.9Mg0.1O3−δ.The 125−200 μm size fraction was used for kineticdetermination. Particles did not show any gain in mass duringadditional thermal treatment in air at temperatures in the interval973−1273 K; therefore, the calcined material was in the oxidizedstate.

2.2. Experimental Setup. The reactivity and oxygen-transport capacity of the CaMn0.9Mg0.1O3−δ material wasdetermined in a thermogravimetric analyzer (CI ElectronicsLtd.). Reduction with CH4, H2, and CO and oxygen-carrieroxidation with O2 were performed by exposing the sample toCH4, H2, CO, and O2, respectively, diluted in nitrogen. To avoidcarbon formation in the reactor, 20 vol % H2O was used in thereacting gas mixture when CH4 was the reducing gas, whereas 20vol % CO2 was present, together with CO. Also, the oxygenrelease rate of a fully oxidized material was studied in an inertenvironment using high-purity N2. Every experimental conditionwas evaluated starting with a new fresh sample. About 50 mg ofoxygen-carrier particles was loaded on a platinum basket, and agas flow of 25 LN h

−1 was fed. External and interparticle diffusionwere of low relevance under these conditions. Moreover,previous calculations showed a high enough rate of mass andheat transfer inside and outside of similar particles, so theparticles could be considered isothermally and homogeneouslyconverted.36

Temperature-programmed reductions (TPRs) were carriedout to determine the maximum oxygen-transport capacity of theoxygen-carrier material. Later, 10 redox cycles at fixed temper-ature (1173 K) were carried out to evaluate evolution of theredox properties with the cycle number. These experiments werefocused on analysis of the oxygen-transport capacity andreactivity for both oxygen uncoupling and reaction with a gas(CH4, H2, O2, or N2). Experiments with CH4, H2, O2, or N2 in thereacting atmosphere were subsequently performed at severalreacting temperatures and gas concentrations to determine thekinetics of reduction and oxidation of the oxygen carrier. Moreinformation about the experimental setup is found in theSupporting Information.

2.2.1. Determination of the Oxygen-Transport Capacity(TPR). TPRs were carried out in three different environments:high-purity N2, 15 vol % H2, and 15 vol % CH4. The oxygen-carrier sample was heated to 473 K in an air atmosphere. Noreaction occurred during this preheating step. After thistemperature was reached, the desired gas composition wasintroduced into the reactor and the weight loss was recorded,while the temperature was increased up to 1273 K at a constantrate of 20 K min−1.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10388

2.2.2. Evolution of the Redox Properties with the CycleNumber. The oxygen-carrier sample was heated to the desiredtemperature in an air atmosphere. After the set temperature wasreached, the experiment started by exposing the oxygen carrier toalternating reducing and oxidizing conditions. In order toprevent fuel and air mixing, a nitrogen flow was passed after theoxidizing and reducing processes. In general, the inert period was10 s in order to minimize decomposition of the oxygen carrierbefore reaction with the fuel gas.To evaluate the stability of the CaMn0.9Mg0.1O3‑δmaterial with

the redox cycle number, two series of 10 redox cycles each werecarried in N2 or CH4. The reduction period was 350 s, whereasoxidation was continued until no change in mass was observed,which was usually confirmed in 300 s. A temperature of 1173 Kand 15 vol % CH4 were chosen for these experiments becausethese conditions are often used during evaluation of a material asan oxygen carrier. Thus, the results obtained can be compared tothe results available in the literature for other materials.Three additional series of 10 cycles were carried out at 1173 K

with 15 vol % CH4 in order to evaluate the stability of the oxygenuncoupling property through redox cycles. In this case, thereducing period was different in each series (40, 90, and 600 s),thus achieving a different conversion of solids.2.2.3. Redox Cycles for Kinetic Determination. Different gas

concentration and temperature values were used to obtain theconversion versus time curves necessary for kinetic determi-nation. To analyze the effect of the fuel gas concentration (CH4,CO, or H2) on the reduction reaction rate, four differentconcentrations were used: 5, 15, 30, and 60 vol %. To determinethe oxidation kinetics, oxygen concentrations of 5, 10, 15, and 21vol % were used. These oxidation experiments were performedusing 15 vol % H2 during the reduction step. All of theseexperiments were carried out at 1073 K. To observe the effect ofthe temperature on the reaction rate, reduction and oxidationreactions were carried out at the following temperatures: 973,1023, 1073, 1123, 1173, 1223, and 1273 K. The fuel gasconcentration during these experiments was 15 vol %, while theoxygen concentration was 10 vol %. Moreover, oxygen release ina nitrogen atmosphere was analyzed at temperatures in the sametemperature interval.For kinetic determination, five redox cycles were carried out

under every condition in TGA. Variation in the reactivity withredox cycles was observed. Thus, the data corresponding to thefirst cycle were used to determine the kinetics of fresh particles,whereas the data corresponding to the fifth cycle were consideredfor kinetic determination of “activated” particles.

3. RESULTS

The oxygen-transport capacity and reactivity with severalreacting gases (CH4, H2, CO, and O2) as well as the oxygenuncoupling properties of the CaMn0.9Mg0.1O3−δ material were

analyzed in a TGA apparatus. In addition, evolution of theseproperties with the cycle number was determined. The kineticsof reduction and oxidation reactions of CaMn0.9Mg0.1O3−δparticles was obtained from evolution of solids conversion withtime, under different operating conditions, i.e., temperature andgas concentration.

3.1. Determination of the Oxygen-Transport Capacity.A perovskite structure is characterized by a unit cell ABO3−δ, withA being a large cation and B a small cation. The oxygen vacanciesin the solid are given by the parameter δ, which is related to theoxygen-transport capacity. However, in ABO3−δ perovskitematerials, such as CaMnO3−δ, the stoichiometric amount ofoxygen depends on the temperature and oxygen concentration.37

Thus, the mass of a fully oxidized oxygen carrier will depend onthese parameters. Discussion regarding the equilibriumcomposition in the Ca−Mn−O system and its application tothe CaMn0.9Mg0.1O3−δ material is given in the SupportingInformation. It should be pointed out here that the degree ofoxidation depends on the temperature. Thus, forCaMn0.9Mg0.1O3−δ oxidation in air, the value of 3 − δ decreasesfrom 2.9 to 2.83 when the temperature increases from 1073 to1273 K. The lower limit for the fraction of oxygen in perovskite,i.e., 3− δ, is taken as the boundary between the cubic structure ofperovskite and Ca2MnO4 or CaMn2O4 formation.

37 In this case,δ < 0.25 for CaMn0.9Mg0.1O3−δ would be required in order tomaintain the perovskite structure. A further loss of oxygen wouldproduce a change in the solid structure and the appearance ofother compounds.Mass losses corresponding to the oxygen uncoupling property

of this material, i.e., generation of gaseous oxygen, according tothe reaction

γ δ↔ + −δ γ− −CaMn Mg O CaMn Mg O

2O0.9 0.1 3 0.9 0.1 3 2

(1)

Nevertheless, the material is able to transfer oxygen to a fuel,e.g., CH4, H2, or CO, until CaMn0.9Mg0.1O2 formation; i.e., themaximum value allowed for γ is 1. When CH4, H2, or CO is usedas the reducing gas, the chemical process corresponds to thegeneral reactions

γ δ

γ δ γ δ

− +

→ + − + −

δ

γ

−

−

( )CH 4CaMn Mg O

4CaMn Mg O ( )CO 2( )H O

4 0.9 0.1 3

0.9 0.1 3 2 2

(2)

γ δ

γ δ

− +

→ + −

δ

γ

−

−

( )H CaMn Mg O

CaMn Mg O ( )H O

2 0.9 0.1 3

0.9 0.1 3 2 (3)

Table 1. Oxygen-Transport Capacity for Different Redox Systems with CaMn0.9Mg0.1O3−δ

stoichiometric formula descriptive formula Mn oxidation state accumulated ROC (wt %)a accumulated ROC (wt %)b

CaMn0.9Mg0.1O2.9 CaO·(MgO)0.1·(MnO2)0.9 MnIV 0.0CaMn0.9Mg0.1O2.83 0.0CaMn0.9Mg0.1O2.73 2.0 1.2CaMn0.9Mg0.1O2.45 CaO·(MgO)0.1·(Mn2O3)0.45 MnIII 5.2 4.4CaMn0.9Mg0.1O2.3 CaO·(MgO)0.1·(Mn3O4)0.3 MnII + MnIII 6.9 6.2CaMn0.9Mg0.1O2 CaO·(MgO)0.1·(MnO)0.9 MnII 10.4 9.7

aExpected oxygen-transport capacity considering reduction from CaMn0.9Mg0.1O2.9 at 1273 K. bExpected oxygen-transport capacity consideringreduction from CaMn0.9Mg0.1O2.83 at 1273 K.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10389

γ δ

γ δ

− +

→ + −

δ

γ

−

−

( )CO CaMn Mg O

CaMn Mg O ( )CO

0.9 0.1 3

0.9 0.1 3 2 (4)

with the condition that γ > δ. For these cases, a descriptiveformula is also shown in Table 1 as the Ca−Mg−Mn−Omaterialis being reduced. The fully oxidized perovskite corresponds tothe formula CaMn0.9Mg0.1O2.9, whereas the limit in oxygentransference is found for a value of γ − δ = 0.9, corresponding tothe reduced compound CaMn0.9Mg0.1O2.The mass difference between the oxidized and reduced forms

defines the oxygen-transport capacity of the oxygen carrier, ROC,which is calculated as

=−

Rm m

mOCo r

o (5)

where mo and mr are the masses of the oxidized and reducedforms of the material, respectively. In this work, two differentdefinitions of the oxygen-transport capacity, ROC,t and ROC,ou, areused depending on the fraction of oxygen involved in thechemical process. Thus, ROC,t is used to describe the solidsconversion for the total fraction of oxygen that can be transferredto a fuel, e.g., CH4, H2, or CO. ROC,ou is used to describe theconversion during the oxygen uncoupling reaction, which canoccur without the presence of a reducing gas. A detaileddescription of the calculations of ROC,t and ROC,ou can be found inthe Supporting Information.Table 1 shows variations in the oxidation state of Mn and the

oxygen-transport capacity as the material is reduced fromCaMn0.9Mg0.1O3−δ to CaMn0.9Mg0.1O2. The results shown inthis section are based on the thermodynamic properties of theCaMnO3−δ material37 (see the Supporting Information).Because of the special thermodynamic properties of theperovskite material, the value of δ in the oxidized state dependson the oxygen concentration and temperature at which theoxidation is carried out. The fully oxidized state, i.e.,CaMn0.9Mg0.1O2.9, can be reached during oxidation in air at atemperature lower than 1100 K. When oxygen is evolved innitrogen, the perovskite structure is stable up to aminimum valueof 3− δ. For example, the boundary of 3− δ is 2.73 at 1273 K buthas a low dependence on the temperature. Thus, the oxygen-transport capacity for oxygen uncoupling is ROC,ou = 2.0 wt % at1273 K, with the perovskite considered to be completely oxidized(see Table 1). Nevertheless, if a complete oxidation−reductioncycle (oxidation in air and reduction in nitrogen) is considered atisothermal conditions, the oxygen-transport capacity for oxygen

uncoupling depends on the temperature. Thus, ROC,ou = 1.7 wt %at 1073 K but decreases to 1.2 wt % at 1273 K.The total oxygen-transport capacity at 1273 K is ROC,t = 10.4

wt % in the case of reduction of CaMn0.9Mg0.1O2.9 toCaMn0.9Mg0.1O2. Nevertheless, the oxygen-transport capacityalso depends on the oxidation temperature as the non-stoichiometry of pervoskite decreases with temperature. Thus,if the temperature in a complete oxidation−reduction cycle is1273 K, then ROC,t = 9.7 wt % for reduction fromCaMn0.9Mg0.1O2.83 to CaMn0.9Mg0.1O2.TPR experiments were conducted in the TGA apparatus to

determine the mass in the reducing state. The oxidized state wasconsidered to be fresh particles as received after calcination in air.Thus, themass loaded in the sample holder of the TGA apparatuswas considered to be mo as CaMn0.9Mg0.1O2.9, i.e., the fullyoxidation degree for the perovskite material. This assumption issupported by thermodynamic equilibrium at the initialconditions.37

Figure 2 shows the normalized mass-loss variation, ω, duringTPR experiments in N2, H2, or CH4. The parameterω representsthe fraction of mass lost during reduction of the fully oxidizedoxygen carrier and was calculated as

ω = mmo (6)

According to eqs 5 and 6, the oxygen-transport capacity can becalculated as

ω= −R 1OC r (7)

where ωr is the normalized mass loss when the sample was fullyreduced. Here, the oxidized compound is CaMn0.9Mg0.1O2.9.When only N2 was fed to the reactor, oxygen generation

started at a temperature of 930 K and slowly continued at themaximum temperature of 1273 K. In addition, the maximumnormalized mass-loss variation corresponded to an oxygen-transport capacity for the oxygen uncoupling property ofROC,ou =1.92 wt %. In this case, a value of γ − δ = 0.166 was deduced,indicating that only a fraction of the oxygen atoms participated inthe oxygen uncoupling behavior. The oxygen released wasslightly lower than that required to induce a change in theperovskite structure (see Table 1). After that, the reaction ratewas slow because of the low driving force under this condition:the oxygen concentration at equilibrium conditions for thecomposition CaMn0.9Mg0.1O2.734 is 0.03 vol % at 1273 K (see theSupporting Information).

Figure 2. Normalized mass-loss profiles for TPR (20 K min−1) in TGA in () 100 vol % high-purity N2, (−−) 15 vol % CH4 + 20 vol % H2O, and(−·−) 15 vol % H2 as well as temperature evolution (···).

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10390

TPR tests were also carried out with 15 vol % CH4 or H2 (seeFigure 2). This reduction is representative of what would happenin the fuel reactor of a CLC system. Thus, the oxygen-transportcapacity in a reducing environment was determined. In thesecases, the reaction started at 610 and 580 K with CH4 or H2,respectively. The reaction rate with CH4 or H2 was faster thanoxygen generation. Moreover, reduction with H2 was faster thanthat with CH4; in fact, reduction with H2 finished when thereaction temperature was 1100 K, whereas reduction with CH4proceeded until the maximum temperature, i.e., 1273 K, wasreached. Nevertheless, the normalized mass loss of the fullyreduced sample was the same in both cases (ωr = 0.895). Thisindicates that the total oxygen-transport capacity was the sameregardless of the reducing gas used. The oxygen-transportcapacity of the CaMn0.9Mg0.1O3−δ material to oxidize a fuel gaswas ROC,t = 10.5 wt %, which corresponds to a value of γ − δ =0.92. So, approximately one atom of oxygen can be transferred tothe fuel per 1 mol of CaMn0.9Mg0.1O3−δ. This value correlateswell with the theoretical reduction from CaMn0.9Mg0.1O2.9 toCaMn0.9Mg0.1O2 (see Table 1).3.2. Evolution of the Redox Properties with the Cycle

Number. 3.2.1. Oxygen-Transport Capacity and ReactivityRelated to the Oxygen Uncoupling Property. TheCaMn0.9Mg0.1O2.9 material has a maximum capacity to releasegaseous oxygen in an inert environment of about 2 wt % (seeTable 1 and Figure 2). The stability of this capability with thecycle number was analyzed by performing two different testseries of 10 redox cycles at 1173 K. At this temperature, the fullyoxidized state is CaMn0.9Mg0.1O2.87, and the theoretical capacityto release oxygen is 1.5 wt %. In the first series, the reductionstage consisted of pure N2 for 350 s, and therefore the mass lossduring reduction was solely due to gaseous oxygen evolutionfrom the oxygen carrier following reaction (1). Figure 3 shows

variation of the normalized mass with nitrogen and oxygenalternated during the stages of reduction and oxidation,respectively. The oxygen carrier showed good stability regardingthe regeneration capability during the oxidation step. The massloss during the reduction step was about 1.1 wt % and wasconstant throughout the cycles. The reduction reaction rate wasslow at the end of every cycle because of the low driving forceunder these conditions: the oxygen concentration at equilibriumwas 0.01 vol %,37 and the total mass loss expected, i.e., 1.5 wt %,was not reached.

Nevertheless, the oxygen carrier should react with a fuel gas ina real case. A second test series of 10 redox cycles was carried outwith the reduction step consisting of 350 s in N2, followed by 600s with 15 vol % CH4 to analyze the behavior of the its oxygenuncoupling property when the oxygen carrier reacted with a fuelgas (see Figure 3). In this case, the capability of the oxygen carrierto transport oxygen decreased through the cycles. However,most remarkable was the loss of the oxygen uncoupling capacitywith the cycle number.To evaluate variation in the reactivity with the cycle number,

the normalized conversion of solids was calculated for everyreaction, i.e., reduction (XN,r) or oxidation (XN,o) as

=−

Xm mR mN,r

o

OC o (8)

=−

Xm mR mN,o

r

OC o (9)

In this case, the oxygen-transport capacity to calculate thenormalized conversion was ROC,ou = 1.5 wt %. Parts a and b ofFigure 4 show evolution ofXN,r versus time curves obtained in the10 consecutives cycles for the cases of reduction in N2 andreduction in N2 followed by reduction in CH4. In the case ofreduction with N2 only, the reactivity was maintained throughoutthe redox cycles. On the contrary, the reactivity and oxygenuncoupling capacity were seriously reduced when reduction inN2 was followed by reaction with CH4 (see Figure 4b).The maintenance of the oxygen uncoupling property may be

connected to the stability of the perovskite structure, which, inturn, is related to the reduction degree reached during reactionwith CH4. A deep reduction will produce a change in the solidstructure and the appearance of other compounds, e.g.,CaMn2O4,

37 with a lower capacity for oxygen uncoupling andlikely a different reactivity.

3.2.2. Oxygen-Transport Capacity and Reactivity withReacting Gases. Three series of 10 reduction−oxidation cycleseach were carried out in the TGA apparatus to analyze variationin the oxygen-transport capacity and reactivity of the oxygencarrier with the cycle number. Each cycle consisted of a reductionstep with 15 vol % CH4 at 1173 K followed by an oxidationperiod in air. The reduction period was different for each series.Times of 40, 90, and 600 s were used. The oxidation period lasted220 s in all cases, assuring the maximum oxidation degreepossible in every cycle, corresponding to the formulaCaMn0.9Mg0.1O2.87. The corresponding total oxygen-transportcapacity was ROC,t = 10.1 wt % at 1173 K.Parts a−c of Figure 5 show normalized mass variation during

each series of 10 cycles. Different behavior was observeddepending on the length of the reduction period. Obviously, theoxygen carrier became more reduced as the reduction periodincreased. When the reduction period was as short as 40 s (seeFigure 5a), the reduction degree of the oxygen carrier was lowand stayed constant through the 10 redox cycles. Also, theoxygen carrier was completely oxidized in air during theoxidation step, indicating good oxygen-carrier particle stabilityunder these conditions. Reduction proceeded up to thestoichiometry 3 − δ = 2.7. The results agree with the goodrecyclability found for CaMnO3−δ reduction to CaMnO2.5,

38

which corresponds to CaMn0.9Mg0.1O2.45 (see Table 1).After the 10 redox cycles under these conditions, a reduction

step was carried out until complete reduction was achieved. Itwas confirmed that the oxygen-transport capacity was main-tained constant at 10 wt %, which agrees with reduction to

Figure 3. Normalized mass variations during reduction−oxidationcycles in TGA starting from fresh particles. Reduction environment:() 100 vol % N2 during 350 s; (−−) 100 vol % N2 during 350 s,followed by 600 s in 15 vol % CH4 + 20 vol %H2O. Oxidation by air. T =1173 K.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10391

CaMn0.9Mg0.1O2. Figure 6a shows the normalized conversion forthe reduction reaction over time for consecutive cycles when thereduction period lasted 40 s. The CaMn0.9Mg0.1O3−δ material

presented a rather low reactivity, which was maintained constantunder these conditions.The behavior of the CaMn0.9Mg0.1O3−δ material was different

when the reduction period was longer. Figure 5b showsnormalized mass variation during 10 redox cycles with thereduction period lasting 90 s. Again, the oxygen carrier was notcompletely reduced, but the reduction degree increased with thecycle number. The value of 3− δwas 2.5 for the 1st reduction butdecreased to 3− δ = 2.3 for the 10th reduction. At the same time,the oxygen carrier was not oxidized up to the same initial level. Infact, the degree of oxidation decreased with the cycle number.This indicates that the effective oxygen-transport capacitydecreases with the cycle number under these conditions. Thefinal oxygen-transport capacity was evaluated by the completereduction of the sample after the 10th redox cycle, where ROC,t =8.5 wt %. With regard to the reactivity evolution (see Figure 6b),the oxygen carrier had a gradual gain in its reduction reaction ratefor a reduction period of 90 s.The changes in the reactivity and oxygen-transport capacity

were more intense when reduction was complete in every cycle(see Figure 5c). The oxidation degree after every complete redoxcycle clearly decreased through the cycles. This decrease was oflower relevance as of the fifth cycle. The oxygen-transportcapacity after 10 redox cycles was ROC,t = 7.5 wt %. The oxygen-transport capacity may have been affected by the formation of acompound with a lower oxidation state for Mn, i.e., CaMn2O4,which has been confirmed in cycled particles.18 In this case, theoxygen uncoupling property of this material decreased with thecycle number, but the reactivity increased up to the fourth or fifthcycle and then stabilized (see Figure 6c). From the fifth cycle, theCaMn0.9Mg0.1O3−δ material could be considered to haveachieved a maximum conversion rate in the reduction reaction.Therefore, an activation process was observed that made thereduction reactivity increase about 4 times during the first redoxcycles.Figure 7a shows the normalized conversion during the

oxidation reaction for 10 consecutive cycles in the series,where the oxygen carrier was completely reduced in thereduction step. From these data, it was concluded that theoxidation rate remained constant through consecutive redoxcycles. Nevertheless, the oxidation degree decreased with the

Figure 4. Evolution of normalized conversion versus time through 10 redox cycles of the CaMn0.9Mg0.1O3−δ material during the reduction period.Environment during the reduction step: (a) 100 vol % N2 during 350 s; (b) 100 vol % N2 during 350 s, followed by 600 s in 15 vol % CH4 + 20 vol %H2O. Oxidation by air. T = 1173 K.

Figure 5. Normalized mass variations during reduction−oxidationcycles in TGA starting from fresh particles. Reducing gas: 15 vol % CH4+ 20 vol % H2O. Oxidation by air. T = 1173 K.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10392

cycle number, and therefore the oxygen-transport capacitydecreased. The final values of the normalized conversion forreduction and oxidation reactions decreased as far as a value of0.75, which corresponds to an oxygen-transport capacity of ROC,t= 7.5 wt %. This value was lower than the initial ROC,t of freshparticles because the oxygen carrier was not able to recover theinitial oxidation state. Figure 7b shows evolution of the oxygen-transport capacity, ROC,t, with the cycle number. It is observedthat ROC,t mainly decreased during the first five cycles, but thedecrease in the ROC,t value was slower during subsequent cycles.The formula for the oxidized state corresponding to the ROC,tvalue obtained from the fifth cycle is CaMn0.9Mg0.1O2.63, which isa value between oxidation ofMn toMnIII andMnIV (see Table 1).This means that 60 wt % of Mn remains as MnIII, whereas 40 wt% is regenerated to the upper oxidation state MnIV.3.3. Redox Cycles for Kinetic Determination. As shown

above, the reactivity and oxygen-transport capacity ofCaMn0.9Mg0.1O3−δ particles differs depending on the operationalconditions and number of redox cycles. Thus, the reaction ratesof fresh and activated CaMn0.9Mg0.1O3−δ materials are obtainedunder different temperatures (973−1273 K) and gas concen-

trations (5−60 vol %). These results are presented as conversionversus time curves and used to determine the kinetic parametersof the reduction and oxidation reactions. The results obtained forthe first reduction or oxidation are considered for kineticdetermination of fresh particles. The fifth redox cycle wasselected for kinetic determination of activated particles because itwas previously shown that the reactivity barely changed after thiscycle (see Figures 6c and 7a).The conversion values are obtained from eqs 8 and 9 by using

the oxygen-transport capacity corresponding to the reactionconsidered and reacting temperature (see the SupportingInformation). Thus, the conversion values ranged from 0 to 1in all cases. When reduction by a fuel gas is considered, ROC,t

varies from 10.4 wt % at 973 K to 9.7 wt % at 1273 K for freshparticles. For activated particles, the ROC,t value is 7.5 wt %. Forthe oxygen uncoupling reaction, the ROC,ou value also depends onthe temperature because CaMn0.9Mg0.1O3−δ is more oxidized asthe temperature decreases; i.e., the δ parameter is higher with thetemperature. Thus, values for ROC,ou of 1.74 and 1.05 wt %correspond to 1073 and 1273 K, respectively. Kinetics of the

Figure 6. Evolution of normalized conversion versus time through 10 redox cycles of the CaMn0.9Mg0.1O3−δ material during the reduction period.Reducing gas: 15 vol % CH4 + 20% H2O. T = 1173 K.

Figure 7. (a) Evolution of normalized conversion versus time through 10 redox cycles of the CaMn0.9Mg0.1O3−δmaterial during the oxidation period inair. (b) Evolution of the oxygen-transport capacity with the redox cycle. T = 1173 K. Reduction period in the cycles with 15 vol % CH4 for 600 s.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10393

oxygen uncoupling reaction with activated particles are notcalculated because of its low relevance.3.3.1. Effect of the Fuel Type.Common reducing gases during

fuel combustion in a CLC system are CH4, H2, and CO. So, thereactivity with these gases was evaluated. To compare theoxygen-carrier reactivity with these gases, Figure 8a shows theconversion versus time curves for the different reducing gases.The reactivity of the CaMn0.9Mg0.1O3−δ material followed theorder H2 > CO >CH4 for fresh particles, while it was H2 > CH4 >CO for activated particles. Oxidation was relatively fast, reachinga conversion value of 80% in 60 s for fresh particles and in 45 s foractivated particles. Moreover, the oxidation rate was notdependent on the fuel gas used in the previous reduction stage.The rate index, calculated by eq 10, is often used for reactivitycomparison purposes.

= × RXt

rate index (%/min) 100 60ddOC (10)

The rate index at 1173 K and 15 vol % fuel gas or 10 vol %O2 isshown in Table 2. For the reduction reaction, higher rate index

values were found for activated particles. It should be noted thatthe rate index depended on both the reactivity and oxygen-transport capacity. Therefore, the increase in the reactivity couldhave been compensated for to some degree by the decrease in theoxygen-transport capacity, as was the case for CO. Mentionshould be made of the increase in the rate index shown by CH4after activation. Thus, it was necessary to activate the particles inorder to achieve high reactivity with methane, although in thiscase the oxygen uncoupling property was significantly reduced.

Furthermore, the rate index for oxidation was lower for activatedparticles, as a consequence of the unchanged reactivity with theredox cycles but of a decrease in the oxygen-transport capacity. Ingeneral, values for activated particles were in the medium-to-upper range of values of the rate index for oxygen-carriermaterials.39−41

3.3.2. Effect of the Gas Concentration. To determine theeffect of the gas concentration on the kinetics of the reductionreaction, several experiments were carried out at 1073 K and withdifferent gas concentrations. At this temperature, the materialwas able to be fully oxidized to CaMn0.9Mg0.1O2.9, with theoxygen-transport capacity ROC,t = 10.4 wt %. The conversionversus time curves obtained with CH4, H2, CO, or O2 as reactinggases are shown in Figures 9 and 10 for fresh and activatedparticles, respectively.There was a direct relationship between the fuel concentration

and reaction rate. As expected, an increase in the fuel gasconcentration produced an increase in the reaction rate. Inaddition, a decrease in the reaction rate was observed with theconversion of solids, particularly for CO. This may indicate achange in the controlling mechanism of the reaction, most likelyfrom chemical reaction control to control by diffusion in theproduct layer of the grains at high conversion values. This had tobe taken into account when the kinetic model was formulated.

3.3.3. Effect of the Reacting Temperature. The effect of thetemperature on the reaction rate of the CaMn0.9Mg0.1O3−δmaterial was investigated in the interval 973−1273 K. Parts a−d of Figure 11 show the conversion versus time curves forreduction of fresh material with CH4, H2, and CO as well asoxidation with O2. Similarly, Parts a−d of Figure 12 showconversion versus time curves for activated particles. In all cases,the reaction rate was affected by the temperature. An increase inthe temperature produced an increase in the reduction rate,although this effect was rather low for oxidation.A decrease was observed in the reaction rate as the solids

conversion increased, particularly for reduction with CO. Asmentioned previously, this may have been due to a change in thecontrolling mechanism from chemical reaction to diffusion in theproduct layer. Nevertheless, the decrease in the reaction rate wasless relevant at high temperatures (≥1173 K), suggesting thatdiffusion in the product layer had less importance at thesetemperatures.

Figure 8.Conversion versus time curves during the reduction period with CH4, H2, or CO and the oxidation period by O2 for (a) fresh and (b) activatedparticles. T = 1173 K; 15 vol % fuel; 10 vol % O2.

Table 2. Oxygen-Transport Capacity and Rate Index for theCaMn0.9Mg0.1O3−δ Material with Different Reacting Gases at1173 K

fresh particles activated particles

ROC (wt %) rate index ROC (wt %) rate index

CH4 10.1 1.9 7.5 11.2H2 10.1 13.7 7.5 16.7CO 10.1 6.3 7.5 6.6O2 10.1 8.2 7.5 7.8N2 1.5 0.8 0.6 n.a.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10394

The effect of the temperature on the oxygen uncoupling rate offresh CaMn0.9Mg0.1O3−δ particles was also analyzed. Figure 13shows the conversion versus time curves obtained duringreaction in pure N2. The need for high temperatures in orderto achieve high oxygen uncoupling reaction rates can beobserved.

4. KINETIC MODEL

The grain model was considered for reduction and oxidationreactions. The particle was assumed to be formed by a certainamount of spherical grains, each following the shrinking-coremodel during reduction or oxidation.42 Some information on thecontrolling mechanism can be extracted from the shape of theconversion versus time curves. It was observed that reductionwith a fuel gas or oxidation by oxygen showed a lineardependence on the reacting time at low conversion values andthat the reaction rate decreased over time at higher solidsconversion values. In fact, there was a sharp decrease in thereaction rate in some cases. The linear dependence of conversionon time can be attributed to a reaction in the gas−solidinterphase with chemical reaction control. Readman et al.43

proposed that in this case oxygen coverage on the surface of thegrains was constant during the reduction period. The oxygendiffused from the bulk metal oxide onto the surface where the

reaction took place, but this process did not limit the reactionrate under the applied conditions. Additionally, the decrease inthe reaction rate at higher conversion values was attributed to achange in the controlling mechanism from chemical reaction todiffusion through the product layer around the grains.With all of the above considerations and considering the

resistance to gas film mass transfer and the diffusion inside theparticle to be negligible, the equations that describe this modelare the following:44

τ τ= + − − + −t X X X[1 3(1 ) 2(1 )]ch pl2/3

(11)

where

τ =−k C C1

( )in

inch

s ,eq (12)

= −k k e E R Ts s,0

/ch g (13)

and

τ =−′ ′D C C1

( )in

inpl

pl ,eq (14)

= −D D e E R Tpl pl,0

/pl g(15)

Figure 9. Effect of the gas concentration on the reaction rate of fresh CaMn0.9Mg0.1O3−δ particles with (a) CH4, (b) H2, (c) CO, and (d) O2.T = 1073 K.Continuous line: model predictions. Gas concentration for reduction reactions: (gray■) 5 vol %; (red◆) 15 vol %; (blue ▲) 30 vol %; (green●) 60vol %. Oxygen concentration: (gray ■) 5 vol %; (red ◆) 10 vol %; (blue ▲) 15 vol %; (green ●) 21 vol %.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10395

where Ci,eq is the equilibrium concentration of the reacting gas.The equilibrium concentration of each fuel gas, i.e., CH4, H2, orCO, was assumed to be zero; in other words, theCaMn0.9Mg0.1O3−δ material was able to fully convert the fuelgas to CO2 and H2O.When a sharp decrease in the reaction rate with time was

observed, an additional term that modified the effectivediffusivity through the product layer became necessary, withDpl calculated as

= − −D D e eE R T k Xpl pl,0

/pl g X(16)

where

= −k k e E R TX X,0

/X g (17)

These two equations were only used for reduction of theCaMn0.9Mg0.1O3−δ material with CO.In the case of the oxygen uncoupling reaction (see eq 1), the

oxygen concentration at equilibrium conditions was considered.Only chemical reaction control was considered, and evolution ofconversion over time was defined by the following equation:

=−

tk C C

X1

( )n ns O ,eq O2 2 (18)

The oxygen concentration at equilibrium conditions dependson the temperature and sharply decreases with solids conversion.A detailed description of this dependence is given in theSupporting Information.

5. DETERMINATION OF THE KINETIC PARAMETERS

5.1. Reduction and Oxidation Reactions. The kineticmodel described in section 4 was used to determine the kineticparameters for the reduction reaction with CH4, H2, and CO andoxidation with O2. The kinetic parameters for the chemicalreaction for the reactions considered were obtained from analysisof the τch values obtained for different gas concentrations andtemperatures (see the Supporting Information for more details).Parts a and b of Figure 14 show a plot of ln(1/τch) versus ln(Ci)for the reaction with CH4, H2, CO, or O2 of fresh and activatedparticles, respectively. The reaction order, n, was obtained fromthe slope of the figure, with ln(ks) being the intercept. Thechemical reaction kinetic constant, ks, as a function of thetemperature was obtained from experiments carried out atdifferent temperatures.

Figure 10. Effect of the gas concentration on the reaction rate of activated CaMn0.9Mg0.1O3−δ particles with (a) CH4, (b) H2, (c) CO, and (d) O2. T =1073 K. Continuous line: model predictions. Gas concentration for reduction reactions: (gray■) 5 vol %; (red◆) 15 vol %; (blue▲) 30 vol %; (green●) 60 vol %. Oxygen concentration: (gray ■) 5 vol %; (red ◆) 10 vol %; (blue ▲) 15 vol %; (green ●) 21 vol %.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10396

Parts a and b of Figure 15 show the Arrhenius plots of theln(ks) values as a function of 1/T for fresh and activated particles.The slope of the curves is Ech/Rg, whereas the intercept value isln(ks,0). The calculated kinetic parameters for each reaction areshown in Tables 3 and 4. For chemical reactions of fresh particles,values for the activation energies ranged from 25 to 107 kJ mol−1.The activation energies determined for H2 and CO (85 and 40 kJmol−1, respectively) were lower than those for CH4 (107 kJmol−1). In general, the activation energies were lower foractivated particles, except CO, which showed a similar value forboth fresh and activated material. Also, the activation energy wasthe same for fresh and activated particles in the oxidation reactionbecause the reactivity was not affected by redox cycles. Thecalculated reaction order was n = 1 for all of the reacting gasesexcept CH4, which was n = 0.5, for both fresh and activatedparticles.Once the kinetic parameters for chemical reaction were

obtained, the kinetic parameters for diffusion through theproduct layer were obtained by fitting the conversion curvesobtained at different gas concentrations and temperatures. It wasfound that the diffusion mechanism was not dependent on thereacting gas concentration in the reduction reactions with CH4,H2, or CO. However, it was dependent on the oxygen

concentration during oxidation, with n′ = 1. This result suggeststhat the limiting process was the diffusion of oxygen anionstoward the outer surface of the particle, which agrees with theformulation of the kinetic model.Values were obtained for the effective diffusivity in the product

layer, Dpl, at each temperature. Furthermore, the decay constantfor reduction with CO, kX, was obtained at several temperatures.The values of the corresponding preexponential factor and theactivation energies were obtained assuming Arrhenius depend-ence forDpl and kX on the temperature (see eqs 15 and 17). Partsa and b of Figure 16 show the Arrhenius plots, and the kineticparameter values are shown in Tables 3 and 4 for fresh andactivated solids. A lower activation energy value can be observedfor activated particles. Also, the activation energy was higher forCO, which could be a consequence of a variable diffusivity withsolids conversion. All of the differences shown in the reductionwith CO, which includes slight variation in the activation energyfor ks after activation or the need for a variable effective diffusivityin the product layer, suggest a different reaction mechanismduring CO oxidation.

5.2. Oxygen Uncoupling Reaction. In this case, a τch valuewas not adequate to describe the reaction progress because theoxygen concentration at equilibrium conditions varied with

Figure 11. Effect of the temperature on the reaction rate of fresh CaMn0.9Mg0.1O3−δ material with (a) CH4, (b) H2, (c) CO, and (d) O2. 15 vol %reducing agent or 10 vol % O2. Continuous line: model predictions. Reacting temperature: (gray■) 973 K; (red◆) 1023 K; (blue▲) 1073 K; (green●) 1123 K; (pink ▼) 1173 K; (burgundy ⬢) 1223 K; (orange ○) 1273 K.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10397

solids conversion. So, a value of ks was determined for everytemperature by fitting the experimental conversion versus timecurves by means of eq 18. The reaction order affected thecurvature of the conversion versus time lines. The reaction orderwas also determined by the fitting method, with n = 0.5. Thedependence of the ks values on the temperature for the oxygenuncoupling reaction is shown for freshmaterial in Figure 15a, andthe kinetic parameters are shown in Table 3.

5.3. Confidence in the Kinetic Parameters. Theconversion versus time curves predicted by the reaction modelwith the kinetic parameters obtained are shown in Figures 9−13.As can be seen, the conversion curves are well predicted in theentire range of operating conditions studied, i.e., gas concen-tration, temperature, and reaction type. The kinetic dataobtained in this work will be useful for inclusion in mathematicalmodels for the design and optimization of CLC reactors. Thechoice of kinetic parameters must be suited to the characteristicsof the particles used. For example, activation was not seen during70 h of operation in a CLC unit operating at high oxygen-carriercirculation rates,19 so suitable kinetics for model validationshould be those for fresh particles. Moreover, an oxygenuncoupling process may be relevant, and this mechanism shouldbe considered.

Figure 12. Effect of the temperature on the reaction rate of activated CaMn0.9Mg0.1O3−δ material with (a) CH4, (b) H2, (c) CO, and (d) O2. 15 vol %reducing agent or 10 vol % O2. Continuous line: model predictions. Reacting temperature: (gray■) 973 K; (red◆) 1023 K; (blue▲) 1073 K; (green●) 1123 K; (pink ▼) 1173 K; (burgundy ⬢) 1223 K; (orange ○) 1273 K.

Figure 13. Effect of the temperature on the oxygen uncoupling reactionrate of fresh CaMn0.9Mg0.1O3−δ material in pure N2. Continuous line:model predictions. Reacting temperature: (gray ■) 973 K; (red ◆)1023 K; (blue ▲) 1073 K; (green ●) 1123 K; (pink ▼) 1173 K;(burgundy ⬢) 1223 K; (orange ○) 1273 K.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10398

6. IMPLICATIONSOF REACTION KINETICS ON THE CLCPERFORMANCE

Some indications can be drawn from the kinetic determinationabout the optimum operating conditions in a CLC unit. In orderto ensure high oxygen transference rate, high temperature wouldbe advisable in all cases to improve methane conversion in a CLCunit. It can be asserted that the oxygen uncoupling mechanism

should be dominant for fresh particles in order to achieve a highoxygen-transfer rate. This is due to the low reactivity of freshmaterial with CH4. The implication in a CLC unit is that a highsolids circulation flow rate would be necessary. Considering thatROC,ou is about 1/10th of ROC,t, the required oxygen carrier to fuelratio would be higher than 10. Here, the oxygen carrier to fuelratio, ϕ, is defined as the flow of oxygen available in circulatingsolids divided by the oxygen required to fully convert the fuel to

Figure 14. Plot to obtain the reaction order with respect to CH4 (gray■), H2 (red◆), CO (blue▲), and O2 (green●) for (a) fresh and (b) activatedparticles. T = 1073 K.

Figure 15. Arrhenius plots for the kinetic constant ks for the reactions of CH4 (gray■), H2 (red◆), CO (blue▲), and O2 (green●) for (a) fresh and(b) activated particles. The kinetic constant values for oxygen uncoupling reaction in N2 for fresh particles are also included (pink ▼).

Table 3. Kinetic Parameters for the Reaction of Fresh CaMn0.9Mg0.1O3−δ Particles with CH4, CO, H2, and O2 and OxygenUncoupling in N2

units CH4 H2 CO O2 N2

n order of the reaction 0.5 1 1 1 0.5ks,0 preexponential factor of ks m3n mol−n s−1 2.7 × 102 7.2 × 101 3.1 × 10−1 2.1 × 10−1 4.4Ech activation energy for ks kJ mol−1 107.3 84.7 39.7 25.1 55.6n′ order of diffusion 0 0 0 1Dpl,0 preexponential factor of Dpl m3m mol−n s−1 4.1 × 109 6.7 × 108 4.9 × 1018 3.1 × 10−2

Epl activation energy for Dpl kJ mol−1 291.2 234.2 424.8 13.4kX,0 preexponential factor of kX 5.4 × 102

EX activation energy for kX kJ mol−1 43.9

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−10399

CO2 and H2O. Lower oxygen carrier to fuel ratio values will limitthe relevance of the oxygen uncoupling, and poor methaneconversion in a CLC unit can be expected. Taking 1 MW of fuelas a reference, and assuming full conversion of gas, the solidscirculation flow rate can be calculated by40

ϕ =Δ °

mdM

R H2

OCO

OC,t c (19)

Table 5 shows the solids circulation flow rate assuming ϕ = 1or 10 and for CH4, H2, and CO as the fuel gas. A reactingtemperature of 1223 K was assumed; thus, ROC,t = 0.096. Ifnatural gas is used as the fuel, the real value may be close to thevalue for CH4. However, if syngas is used as the fuel, the real value

could be between those for CO and H2, depending on the syngascomposition.41 A solids circulation flow rate as high as 16 kg s−1

per MWth can be achieved with the current CFB technology.40

Thus, this oxygen carrier could be used inside the limits definedby the solids circulation flow rate required.4

However, with lower circulation rates, the oxygen carrier tofuel ratio is low enough to allow a high variation in the solidsconversion. The activation process can then take place. Thus,reactivity of the oxygen carrier with CH4 is highly increased and alower oxygen carrier to fuel ratio could be allowed in order toachieve high conversion of methane in a CLC unit.Complementary to the rate index shown in Table 2, the

minimum oxygen-carrier inventory in the fuel and air reactor is asuitable parameter to compare the quality of a material as anoxygen carrier. The minimum oxygen-carrier inventory isdetermined following the methodology shown in a previouswork.40 Thus, the solids inventory in every reactor can becalculated by

=Δ ° Φ

mdM

R H X t2 1

[d /d ]FRO

OC c red FR (20)

=Δ ° Φ

mdM

R H X t2 1

[d /d ]ARO

OC c ox AR (21)

where the parameters ΦFR and ΦAR are the characteristicreactivity in the fuel and air reactors, respectively. For the kineticmodel used in this work, the values for the characteristicreactivity were between 0 and 1.40 The minimum solidsinventory was calculated for the case of the upper limit of ΦFR

Table 4. Kinetic Parameters for the Reaction of Activated CaMn0.9Mg0.1O3−δ with CH4, CO, H2, and O2

units CH4 H2 CO O2 N2

n order of the reaction 0.5 1 1 1ks,0 preexponential factor of ks m3n mol−n s−1 2.4 × 101 2.0 1.0 2.8 × 10−1

Ech activation energy for ks kJ mol−1 66.1 42.9 44.9 25.1n′ order of diffusion 0 0 0 1Dpl,0 preexponential factor of Dpl m3m mol−n s−1 3.2 × 105 3.5 × 106 7.0 × 109 4.2 × 10−2

Epl activation energy for Dpl kJ mol−1 187.8 170.4 242.2 13.4kX,0 preexponential factor of kX 1.8 × 101

EX activation energy for kX kJ mol−1 16.2

Figure 16.Arrhenius plots for the effective diffusivity in the product layerDpl for the reaction of CH4 (gray■), H2 (red◆), CO (blue▲), andO2 (green●) for (a) fresh and (b) activated particles. kX for CO is also shown (pink ▼).

Table 5. Minimum Solids Circulation Flow Rate andMinimum Solids Inventory in the Fuel and Air Reactors forFresh and Activated CaMn0.9Mg0.1O3−δwith CH4, H2, and CO

comments units CH4 H2 CO

mOC ϕ = 1 kg s−1 MW−1 0.8 0.7 0.6mOC ϕ = 10 kg s−1 MW−1 8.3 6.9 5.9

Fresh ParticlesmFR reaction with gases kg MW−1 109 27 65mFR oxygen uncoupling kg MW−1 680 564 482mAR kg MW−1 42 35 30

Activated ParticlesmFR reaction with gases kg MW−1 27 21 43mAR kg MW−1 40 34 29

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−103100

andΦAR, which meant that variation in the solids conversion waslow.4With this condition, chemical reaction control was assumedto be the controlling mechanism to calculate the reduction rate,[dXred/dt], and oxidation rate, [dXox/dt].Reduction with CH4, H2, or CO was considered in order to

calculate the solids inventory in the fuel reactor. The reactionrates were calculated for an average gas concentration, whichdepended on the reaction order, gas conversion, and possible gasexpansion during fuel conversion.40 The average gas concen-tration for a fuel conversion of 99.9% was used, which in this casewere 12 vol % CH4 and 14 vol % for H2 and CO. For freshparticles, the oxygen uncoupling rate, [dXou/dt], was alsoconsidered for fresh particles. In this case, ROC was ROC,ou in eq20, which corresponded to the oxygen-transport capacity tocalculate Xou. An average oxygen concentration of zero wasassumed to calculate [dXou/dt], whereas the oxygen concen-tration at equilibrium was given by the maximum oxidation statein the air reactor. Assuming that the oxygen concentration at theair reactor exit was 4 vol %, corresponding to an air excess of 20%,the perovskite material is oxidized to CaMg0.1Mn0.9O2.824 at 1223K. Thus, the oxygen concentration at equilibrium is 4 vol % at1223 K.To calculate the solids inventory in the air reactor, the average

concentration was 11 vol %, corresponding to a concentration atthe air reactor exit of 4 vol % O2. Note that ROC = ROC,t must beused to calculate the solids inventory in the air reactor (eq 21).Table 5 shows the minimum solids inventory in the fuel and air

reactors for fresh and activated particles. The solids inventory inthe fuel reactor for fresh particles is higher than that for activatedparticles owing to the lower reactivity of fresh particles. Also, thesolids inventory follows the reactivity order with different gases.The minimum solids inventory values for this material can beused for comparison purposes. Thus, these values are lower thanthose reported for other manganese-based materials.4,11

Note that the solids inventories corresponding only to theoxygen uncoupling mechanism are higher than those corre-sponding to reaction with gases. This is due to the low value ofROC,ou compared to ROC,t. Nevertheless, the oxygen uncouplingmechanism can be relevant for fuel conversion, where the fuel gasconcentration seeing the particles is low. This fact can be relevantin the bottom bed, where the fuel concentration in the emulsionphase can be low,21,22 or at the top of the reactor, where the fuel ishighly converted. In these sections of the reactor, the oxygentransference rate via oxygen uncoupling may be faster than thedirect gas−solid reaction between the fuel and oxygen carrier.

7. CONCLUSIONSThe oxygen-transport capacity and reactivity of theCaMn0.9Mg0.1O3−δ material for the reduction and oxidationreactions were determined by TGA using CH4, H2, CO, or O2 asreacting gases. The oxygen uncoupling property of this materialwas also analyzed.Fresh particles showed an oxygen uncoupling property, with

an oxygen loss of 1.1 wt % at 1173 K in N2, but showed lowreactivity against CH4. The oxygen uncoupling mechanism mayhave been dominant for fresh particles when a high solidscirculation rate was used.This material showed an activation process for the reduction

reaction when variation of the solids conversion in consecutiveredox cycles was higher than 50%. During activation, thereactivity of the oxygen carrier with CH4, H2, or CO increased,but the oxidation reactivity was not affected by the cycle number.Also, the oxygen-transport capacity decreased from an initial

value of 10.5 wt % to a stable value of 7.5 wt % after 10 redoxcycles, and the oxygen uncoupling property was stronglyreduced. The kinetics of activated particles could be relevant ifa low solids circulation rate were to be used in a CLC unit.The grain model, assumingmixed control by chemical reaction

and diffusion through the product layer, was able to predict theexperimental results under different gas concentrations, temper-atures, or reaction types. The results suggested that diffusion ofthe oxygen anions in the product layer was the controllingmechanism during the diffusion stage. For comparison purposes,the minimum solids inventory in the fuel and air reactors wasdetermined by using kinetic data. The total minimum solidsinventory in a CLC unit is 151 or 67 kg MWth

−1 for fresh oractivated particles, respectively. These values are lower than thevalues calculated for other manganese-based materials because oftheir higher reactivity.In the future, the kinetic model can be included in reactor

models to predict the performance of this perovskite material in aCLC unit.

■ ASSOCIATED CONTENT*S Supporting InformationDetailed description of the TGA apparatus, thermodynamicanalysis of the Ca−Mn−O system, and determination of thekinetic parameters. This material is available free of charge via theInternet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Phone: + 34 976 733 977. Fax: +34 976 733 318. E-mail: abad@icb.csic.es.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis paper is based on the work produced within the frameworkof the INNOCUOUS (Innovative Oxygen Carriers UpliftingChemical Looping Combustion) Project, funded by theEuropean Commission under the Seventh Framework Pro-gramme (Contract 241401). The authors thank Frans Snijkersfrom VITO and Prof. Anders Lyngfelt from Chalmers Universityof Technology for supplying the CaMn0.9Mg0.1O3−δ material.

■ NOMENCLATURECi = reacting concentration of gas i (mol m−3)

Dpl = effective product layer diffusivity (m3n′ mol−n′ s−1)Dpl,0 = preexponential factor for effective product layer

diffusivity (m3n′ mol−n′ s−1)Ei = activation energy of process i (J mol−1)ks = chemical kinetic constant (m3n mol−n s−1)ks,0 = preexponential factor for the chemical kinetic constant(m3n mol−n s−1)kX = decay constant for the product layer diffusivitykX,0 = preexponential factor for the decay constant for theproduct layer diffusivitym = instantaneous mass of the oxygen carrier (kg)mo = mass of the oxidized form of the oxygen carrier (kg)mr = mass of the reduced form of the oxygen carrier (kg)n = reaction order for the chemical reactionn′ = exponential constant for the gas concentration in theproduct layer diffusion process

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−103101

Rg = constant for ideal gases (Rg = 8.314 J mol−1 K−1)ROC = oxygen-transport capacityROC,ou = oxygen-transport capacity for the oxygen uncouplingprocessROC,t = total oxygen-transport capacity with a fuel gast = time (s)T = temperature (K)X = conversion of solidsXN = normalized conversion of solids

Greek Symbolsγ=parameter in the subscript for oxygen in the reduced state,according the formula CaMn0.9Mg0.1O3−γδ=parameter in the subscript for oxygen in the oxidized state,according the formula CaMn0.9Mg0.1O3−δϕ=oxygen carrier to fuel ratioτi=time for complete conversion when the reaction iscontrolled by process i (s)ω=normalized mass-loss variation

Subscriptsch=chemical reactioneq=equilibriumo=oxidationou=oxygen uncouplingpl=product layerr=reduction

■ REFERENCES(1) Climate Change 2007: Synthesis Report. Contribution of WorkingGroups I, II and III to the Fourth Assessment Report of theIntergovernmental Panel on Climate Change; Pachauri, R. K., Reisinger,A., Eds.; IPCC: Geneva, Switzerland, 2007(2) IPCC Special Report on Carbon Dioxide Capture and Storage.Prepared by Working Group III of the Intergovernmental Panel on ClimateChange; Metz, B., Davidson, O., de Coninck, H. C., Loos, M., Meyer, L.A., Eds.; Cambridge University Press: Cambridge, U.K., 2005.(3) Ekstrom, C.; Schwendig, F.; Biede, O.; Franco, F.; Haupt, G.; deKoeijer, G.; Papapavlou, D.; Røkke, P. E. Techno-economic evaluationsand benchmarking of pre-combustion CO2 capture and oxy-fuelprocesses developed in the European ENCAP project. Energy Proc.2009, 1, 4233.(4) Adanez, J.; Abad, A.; Garcia-Labiano, F.; Gayan, P.; de Diego, L. F.Progress in Chemical-Looping Combustion and Reforming technolo-gies. Prog. Energy Combust. Sci. 2012, 38, 215.(5) Hossain, M. M.; de Lasa, H. I. Chemical-looping combustion(CLC) for inherent CO2 separationsa review. Chem. Eng. Sci. 2008,63, 4433.(6) Mattisson, T.; Lyngfelt, A.; Leion, H. Chemical-looping withoxygen uncoupling for combustion of solid fuels. Int. J. Greenhouse GasControl 2009, 3, 11.(7) Abad, A.; Adanez-Rubio, I.; Gayan, P.; García-Labiano, F.; deDiego, L. F.; Adanez, J. Demonstration of chemical-looping with oxygenuncoupling (CLOU) process in a 1.5 kWth continuously operating unitusing a Cu-based oxygen-carrier. Int. J. Greenhouse Gas Control 2012, 6,189.(8) Adanez-Rubio, I.; Abad, A.; Gayan, P.; de Diego, L. F.; García-Labiano, F.; Adanez, J. Performance of CLOU process in thecombustion of different types of coal with CO2 capture. Int. J.Greenhouse Gas Control 2013, 12, 430.(9) Adanez-Rubio, I.; Abad, A.; Gayan, P.; de Diego, L. F.; García-Labiano, F.; Adanez, J. Biomass combustion with CO2 capture byChemical Looping with Oxygen Uncoupling (CLOU). Int. J. GreenhouseGas Control 2013, 19, 541−551.(10) Abad, A.; Mattisson, T.; Lyngfelt, A.; Ryden, M. Chemical-looping combustion in a 300 W continuously operating reactor systemusing a manganese-based oxygen carrier. Fuel 2006, 85, 1174.

(11) Zafar, Q.; Abad, A.; Mattisson, T.; Gevert, B.; Strand, M.Reduction and oxidation kinetics of Mn3O4/Mg−ZrO2 oxygen carrierparticles for chemical-looping combustion. Chem. Eng. Sci. 2007, 62,6556.(12) Sarshar, Z.; Kleitz, F.; Kaliaguine, S. Novel oxygen carriers forchemical looping combustion: La1−xCexBO3 (B = Co, Mn) perovskitessynthesized by reactive grinding and nanocasting. Energy Environ. Sci.2011, 4, 4258.(13) Sarshar, Z.; Sun, Z.; Zhao, D.; Kaliaguine, S. Development ofSinter-Resistant Core−Shell LaMnxFe1−xO3@mSiO2 Oxygen Carriersfor Chemical Looping Combustion. Energy Fuels 2012, 26, 3091.(14) Leion, H.; Larring, Y.; Bakken, E.; Bredesen, R.; Mattisson, T.;Lyngfelt, A. Use of CaMn0.875Ti0.125O3 as Oxygen Carrier in Chemical-Looping with Oxygen Uncoupling. Energy Fuels 2009, 23, 5276.(15) Fossdal, A.; Bakken, E.; Øye, B. A.; Schøning, C.; Kaus, I.;Mokkelbost, T.; Larring, Y. Study of inexpensive oxygen carriers forchemical looping combustion. Int. J. Greenhouse Gas Control 2011, 5,483.(16) Arjmand, M.; Hedeyati, A.; Azad, A. M.; Leion, H.; Ryden, M.;Mattisson, T. CaxLa1−xMn1−yMyO3−δ (M = Mg, Ti, Fe, or Cu) asOxygen Carriers for Chemical-Looping with Oxygen Uncoupling(CLOU). Energy Fuels 2013, 27, 1473.(17) Ryden, M.; Lyngfelt, A.; Mattisson, T. CaMn0.875Ti0.125O3 asoxygen carrier for chemical-looping combustion with oxygen uncou-pling (CLOU)Experiments in a continuously operating fluidized-bedreactor system. Int. J. Greenhouse Gas Control 2011, 5, 356.(18) Hallberg, P.; Jing, D.; Ryden, M.; Mattisson, T.; Lyngfelt, A.Chemical Looping Combustion and Chemical Looping with OxygenUncoupling Experiments in a Batch Reactor Using Spray-DriedCaMn1−xMxO3−δ (M = Ti, Fe, Mg) Particles as Oxygen Carriers. EnergyFuels 2013, 27, 1473.(19) Internal communication in the INNOCUOUS project.Innovative Oxygen Carriers Uplifting Chemical-Looping CombustionProject, funded by the European Commission under the SeventhFramework Programme, Contract 241401.(20) Kallen, M.; Ryden, M.; Dueso, C.; Mattisson, T.; Lyngfelt, A.CaMn0.9Mg0.1O3−δ as oxygen carrier in a gas-fired 10 kWth chemical-looping combustion unit. Ind. Eng. Chem. Res. 2013, 52, 6923.(21) Abad, A.; Adanez, J.; García-Labiano, F.; de Diego, L. F.; Gayan, P.Modeling of the chemical-looping combustion of methane using a Cu-based oxygen-carrier. Combust. Flame 2010, 157, 602.(22) Abad, A.; Adanez, J.; García-Labiano, F.; de Diego, L. F.; Gayan,P.; Kolbitsch, P.; Proll, T. CLC Modeling: the fuel-reactor at fastfluidizationConversion of CH4 using a NiO-based oxygen-carrier in a120 kWth unit. Proceedings of the 1st International Conference on ChemicalLooping, Lyon, France, 2010.(23) Abad, A.; Adanez, J.; de Diego, L. F.; Gayan, P.; García-Labiano,F.; Lyngfelt, A. Fuel reactor model validation: assessment of the keyparameters affecting the chemical-looping coal combustion process.Submitted for publication, 2013.(24) Mahalatkar, K.; Kuhlman, J.; Huckaby, E. D.; O’Brien, T.Computational fluid dynamic simulations of chemical looping fuelreactors utilizing gaseous fuels. Chem. Eng. Sci. 2011, 66, 469.(25) Mahalatkar, K.; Kuhlman, J.; Huckaby, E. D.; O’Brien, T. CFDsimulation of a chemical-looping fuel reactor utilizing solid fuel. Chem.Eng. Sci. 2011, 66, 3617.(26) Mahalatkar, K.; Kuhlman, J.; Huckaby, E. D.; O’Brien, T.Simulations of a Circulating Fluidized Bed Chemical LoopingCombustion System Utilizing Gaseous Fuel. Oil Gas Sci. Technol.2011, 66, 301.(27) Deng, Z.; Xiao, R.; Jin, B.; Song, Q.; Huang, H. Multiphase CFDModeling for a Chemical Looping Combustion Process (Fuel Reactor).Chem. Eng. Technol. 2008, 31, 1754.(28) Deng, Z.; Xiao, R.; Jin, B.; Song, Q. Numerical simulation ofchemical looping combustion process with CaSO4 oxygen carrier. Int. J.Greenhouse Gas Control 2009, 3, 368.(29) Brown, T. A.; Dennis, J. S.; Scott, S. A.; Davidson, J. F.; Hayhurst,A. N. Gasification and Chemical-Looping Combustion of a Lignite Charin a Fluidized Bed of Iron Oxide. Energy Fuels 2010, 24, 3034.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−103102

(30) Yazdanpanah, M. M.; Forret, A.; Gauthier, T.; Delebarre, A.Modeling of CH4 combustion with NiO/NiAl2O4 in a 10 kWth CLCPilot Plant. Proceedings of 2nd International Conference on ChemicalLooping, Darmstadt, Germany, 2012.(31) Peltola, P.; Ritvanen, J.; Tynjala, T.; Proll, T.; Hyppanen, T. One-dimensional modelling of chemical looping combustion in dual fluidizedbed reactor system. Int. J. Greenhouse Gas Control 2013, 16, 72.(32) Cloete, S.; Johansen, S. T.; Amini, S. An assessment of the abilityof computational fluid dynamic models to predict reactive gas−solidflows in a fluidized bed. Powder Technol. 2012, 215−216, 15.(33) Iliuta, I.; Tahoces, R.; Patience, G. S.; Rifflart, S.; Luck, F.Chemical-Looping Combustion Process: Kinetics and MathematicalModeling. AIChE J. 2010, 56, 1063.(34) Ortiz, M.; de Diego, L. F.; Abad, A.; García-Labiano, F.; Gayan, P.;Adanez, J. Catalytic Activity of Ni-Based Oxygen-Carriers for SteamMethane Reforming in Chemical-Looping Processes. Energy Fuels 2012,26, 791.(35) García-Labiano, F.; de Diego, L. F.; Adanez, J.; Abad, A.; Gayan, P.Reduction and Oxidation Kinetics of a Copper-Based Oxygen CarrierPrepared by Impregnation for Chemical-Looping Combustion. Ind. Eng.Chem. Res. 2004, 43, 8168.(36) García-Labiano, F.; de Diego, L. F.; Adanez, J.; Abad, A.; Gayan, P.Temperature variations in the oxygen carrier particles during theirreduction and oxidation in a chemical-looping combustion system.Chem. Eng. Sci. 2005, 60, 851.(37) Leonidova, E. I.; Leonidov, I. A.; Patrakeev, M. V.; Kozhevnikov,V. L. Oxygen non-stoichiometry, high-temperature properties, andphase diagram of CaMnO3‑δ. J. Solid State Electrochem. 2011, 15, 1071.(38) Bakken, E.; Norby, T.; Stølen, S. Nonstoichiometry and reductivedecomposition of CaMnO3−δ. Solid State Ionics 2005, 176, 217.(39) Johansson, M.; Mattisson, T.; Lyngfelt, A. Comparison of oxygencarriers for Chemical-Looping Combustion. Therm. Sci. 2006, 10, 93.(40) Abad, A.; Adanez, J.; García-Labiano, F.; de Diego, L. F.; Gayan,P.; Celaya, J. Mapping of the range of operational conditions for Cu-,Fe-, and Ni-based oxygen carriers in chemical-looping combustion.Chem. Eng. Sci. 2007, 62, 533.(41) Abad, A.; García-Labiano, F.; de Diego, L. F.; Gayan, P.; Adanez, J.Reduction and oxidation kinetics of Cu-, Ni-, and Fe-based oxygencarriers using syngas (H2 + CO) for chemical-looping combustion.Energy Fuels 2007, 21, 1843.(42) Szekely, J.; Evans, J. W. A structural model for gas−solid reactionswith a moving boundary. Chem. Eng. Sci. 1970, 25, 1091.(43) Readman, J. E.; Olafsen, A.; Smith, J. B.; Blom, R. Chemicallooping combustion using NiO/NiAl2O4: mechanisms and kinetics ofreduction-oxidation (Red−Ox) reactions from in situ powder X-raydiffraction and thermogravimetry experiments. Energy Fuels 2006, 20,1382.(44) Levenspiel, O. Chemical Reaction Engineering; John Wiley andSons: New York, 1981.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie4015765 | Ind. Eng. Chem. Res. 2014, 53, 87−103103