chemical engineering research and design 8 6 ( 2 0 0 8 ) 1017–1026

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journa l homepage: www.e lsev ier .com/ locate /cherd

The use of ilmenite as an oxygen carrier inchemical-looping combustion

Henrik Leiona,∗, Anders Lyngfeltb, Marcus Johanssona,Erik Jerndala, Tobias Mattissonb

a Department of Environmental Inorganic Chemistry, Chalmers University of Technology, S-412 96 Goteborg, Swedenb Department of Energy and Environment, Chalmers University of Technology, S-412 96 Goteborg, Sweden

a b s t r a c t

The feasibility of using ilmenite as oxygen carrier in chemical-looping combustion has been investigated. It was found

that ilmenite is an attractive and inexpensive oxygen carrier for chemical-looping combustion. A laboratory fluidized-

bed reactor system, simulating chemical-looping combustion by exposing the sample to alternating reducing and

oxidizing conditions, was used to investigate the reactivity. During the reducing phase, 15 g of ilmenite with a particle

size of 125–180 �m was exposed to a flow of 450 mLn/min of either methane or syngas (50% CO, 50% H2) and during

the oxidizing phase to a flow of 1000 mLn/min of 5% O2 in nitrogen. The ilmenite particles showed no decrease in

reactivity in the laboratory experiments after 37 cycles of oxidation and reduction. Equilibrium calculations indicate

that the reduced ilmenite is in the form FeTiO3 and the oxidized carrier is in the form Fe2TiO5 + TiO2. The theoretical

oxygen transfer capacity between these oxidation states is 5%. The same oxygen transfer capacity was obtained in

the laboratory experiments with syngas. Equilibrium calculations indicate that ilmenite should be able to give high

conversion of the gases with the equilibrium ratios CO/(CO2 + CO) and H2/(H2O + H2) of 0.0006 and 0.0004, respectively.

Laboratory experiments suggest a similar ratio for CO. The equilibrium calculations give a reaction enthalpy of the

overall oxidation that is 11% higher than for the oxidation of methane per kmol of oxygen. Thus, the reduction from

Fe2TiO5 + TiO2 to FeTiO3 with methane is endothermic, but less endothermic compared to NiO/Ni and Fe2O3/Fe3O4,

and almost similar to Mn3O4/MnO.

© 2008 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Chemical-looping combustion (CLC); Oxygen carrier; Ilmenite; Iron oxide; Titanium oxide

1

CtatirTw(tc

0d

. Introduction

hemical-looping combustion (CLC) has been introduced as aechnique where the greenhouse gas CO2 is inherently sep-rated during combustion. The CLC-process is composed ofwo fluidized bed reactors, an air and a fuel reactor, shownn Fig. 1. The fuel is introduced to the fuel reactor where iteacts with an oxygen carrier to CO2 and H2O, reaction (1).he reduced oxygen carrier is transported to the air reactorhere it is oxidized back to its original state by air, reaction

2). In this paper, when oxidation and reduction is men-ioned, it refers to oxidation and reduction of the oxygenarrier.

∗ Corresponding author. Tel.: +46 31 7722886; fax: +46 31 7722853.E-mail address: Leion@chalmers.se (H. Leion).Received 16 November 2007; Accepted 31 March 2008

263-8762/$ – see front matter © 2008 The Institution of Chemical Engioi:10.1016/j.cherd.2008.03.019

CnH2m + (2n + m)MexOy ↔ nCO2 + mH2O + (2n + m)MexOy−1

(1)

O2 + 2MexOy−1 ↔ 2MexOy (2)

The fuel never mixes with air, resulting in a stream ofoxygen-depleted air from the air reactor, and a stream of com-bustion gases from the fuel reactor, which mainly consistsof CO2 and H2O. The water is easily condensed and, after

compression, the CO2 can be transported to a suitable under-ground storage location. The total amount of heat resultingfrom reactions (1) and (2) is the same as for a normal combus-

neers. Published by Elsevier B.V. All rights reserved.

1018 chemical engineering research and d

Nomenclature

m mass of oxygen carrier (g)mox mass of oxygen carrier for in its most oxidized

state (g)mred mass of oxygen carrier for reduced oxygen car-

rier (g)Ro the oxygen transfer capacityxCH4 fraction of CH4 in the outgoing gases after the

water has been removedxCO fraction of CO in the outgoing gases after the

water has been removedxCO2 fraction of CO2 in the outgoing gases after the

water has been removedX degree of conversion

Greek symbols� gas yield

ω mass-based conversion

tion where the fuel is in direct contact with air, thus a purestream of CO2 has been produced without any direct energypenalty for the separation of gases.

The CLC process has in a few years grown from a paper con-cept to a promising technique for CO2-capture and a numberof prototypes have been designed. The first successful demon-stration with 100 h of operation was in a 10-kW prototype atChalmers in 2003 (Lyngfelt and Thunman, 2005; Lyngfelt etal., 2004). Now the process has been demonstrated in reac-tor units ranging from 300 W to 50 kW using natural gas orsyngas as fuel (Abad et al., 2006, 2007a; Adanez et al., 2006;de Diego et al., 2007; Johansson et al., 2006a,b; Linderholmet al., accepted for publication; Lyngfelt and Thunman, 2005;Lyngfelt et al., 2004; Ryu et al., 2004). Furthermore a 10-kW unitat Chalmers designed for solid fuel has been operated withcoal and petroleum coke (Berguerand and Lyngfelt, accepted

for publication, 2008). Various aspects of the oxygen carriersneeded in the CLC-process have been investigated such as

Fig. 1 – Schematic picture of the CLC-process. Twointerconnected fluidized bed reactors, an air and a fuelreactor, with circulating oxygen-carrying particles.

esign 8 6 ( 2 0 0 8 ) 1017–1026

different combination of active oxides and support materialsand the effect of sintering temperatures (Adanez et al., 2004,2005; Johansson, 2007; Mattisson et al., 2004), different manu-facturing processes (Ishida et al., 1996) and thermodynamicrestrictions (Jerndal et al., 2006). An overview of literatureconcerning CLC including over 100 publications is given byJohansson et al. (2006c).

A majority of the publications concerning CLC have usedgaseous fuel such as natural gas or methane. Since solidfuels are considerably more abundant and often less expen-sive than natural gas, it would be highly advantageous ifthe CLC-process could be adapted for solid fuels. Lyon andCole (2000) studied the conversion of coal with Fe2O3 in asmall fluidized bed reactor and found that SO2 enhancesthe conversion rate. Pan et al. (2004) proposed a design fora CLC unit with solid fuel. Cao et al. (2005) performed TGAexperiments showing that it is possible to reduce CuO usingcoal as fuel. Moreover, Dennis et al. (2006) and Scott et al.(2006) demonstrated that lignite char could be oxidized usingFe2O3 as oxygen carrier in a small bed reactor fluidized withsteam and CO2. Leion et al. (2007) used Fe2O3 supported withMgAl2O4 as oxygen carrier and petroleum coke as fuel toinvestigate the effect of different reaction parameters suchas temperature and different composition of the fluidizinggas. Mattisson et al. has also proposed a novel combustiontechnique utlilizing chemical-looping, i.e. chemical-loopingwith oxygen uncoupling (CLOU), for combustion of solids fuel(Mattisson et al., accepted for publication). In this technol-ogy, the solid fuel is actually burnt with gas-phase oxygen,released from an oxygen carrier in the fuel reactor. There arealso a few publications on CLC using syngas as fuel (Abadet al., 2006; Copeland et al., 2002; Johansson et al., 2006b;Mattisson et al., 2006) which are of relevance, because solidfuels during gasification produce CO and H2, which are impor-tant reaction intermediates in CLC with solid fuel. Moreover,Gupta et al. (2005) have performed TGA experiments whereFe2O3 supported with TiO2 is used to produce syngas fromcoal.

In all these publications, except Berguerand and Lyngfelt(accepted for publication, 2008), the oxygen carrier has beenmanufactured using pure chemicals. These are high-costmaterials and may not be well suited for solid fuels since life-time of the oxygen carrier in a CLC-system with solid fuelsmay be restricted by deactivation caused by ash or by loss ofmaterial with the ash when separated from the oxygen car-rier. Therefore, low-cost materials with sufficient reactivitytowards H2 and CO are of interest when using solid fuels inCLC. An example of such an oxygen carrier is the natural min-eral ilmenite which will be the focus of this paper. Ilmenite hasrecently been tested by Berguerand and Lyngfelt (accepted forpublication, 2008) and Leion et al. (2008).

In this paper, the reactivity of ilmenite towards methaneand syngas as well as the fluidization properties of ilmenitewere investigated experimentally in a laboratory setup. Thisincluded both long-term experiments to verify that particlescan sustain multiple cycles of oxidation and reduction, as wellas test where the length of the reducing period was graduallyincreased in order to investigate whether the degree of reduc-tion has any effect on fluidization of particles. The ilmeniteused in these experiments has been compared with unusedilmenite using X-ray diffraction (XRD) as well as with scan-

ning electron microscope (SEM). Thermodynamic propertiesof ilmenite were calculated and compared with results fromexperiments.

design 8 6 ( 2 0 0 8 ) 1017–1026 1019

2

2

Ttlgpppitstapwto

wihinrTCe

tma

ttmts

4iphbs

chemical engineering research and

. Experimental

.1. Laboratory setup

he experiments were conducted with a fluidized-bed reac-or of quartz placed in an oven. The reactor had a totalength of 870 mm and an inner diameter of 22 mm. The oxy-en carrying particles were placed on a porous quartz platelaced 370 mm from the bottom of the reactor. The tem-erature was measured 5 mm under and 10 mm above theorous quartz plate, using 10% Pt/Rh thermocouples enclosed

n quartz shells. The temperature under the bed was usedo regulate the temperature of the oven and was held con-tant at 950 ◦C. The temperature measured above the plate inhe actual ilmenite sample was affected by the reaction heatnd therefore varied from 970 to 980 ◦C, with the highest tem-eratures during oxidation and the lowest during reductionith CH4. Temperatures in this work refer to the tempera-

ure in the bed over the porous plate and have an accuracyf ±5 ◦C.

A sample of 15 g of ilmenite particles of size 0.125–0.180 mmas placed on the porous plate and was then initially heated

n an inert atmosphere to the reaction temperature. The bedeight was approximately 20 mm when the bed was not flu-

dized. The sample was then alternately exposed to 5% O2 initrogen or fuel which was introduced from the bottom of theeactor, thus simulating the cyclic conditions of a CLC-system.he fuel was either pure methane or syngas, a mixture of 50%O and 50% H2. Nitrogen gas was introduced for 180 s betweenach reducing and oxidizing period.

The gas from the reactor was led to an electric cooler, wherehe water was removed, and then to a gas analyzer (Rose-

ount NGA-2000) where the concentrations of CO2, CO, CH4,nd O2 were measured in addition to the gas flow.

The exothermic nature of the oxidation reaction meanshat there will be release of heat and therefore a subsequentemperature rise. To limit this temperature increase, a gas

ixture with 5% O2 in N2 was used instead of air. Thus, largeemperature increases were avoided since there was no pos-ibility to cool the reactor in the present setup.

The experiments were conducted with a gas flow of50 mLn/min (normalized to 1 bar and 25 ◦C), for the reduc-ng periods and 1000 mLn/min for the oxidizing periods. It wasossible to establish whether the bed was fluidized or not by

igh-frequency measurements of the pressure drop over theed (Cho et al., 2006). A schematic layout of the laboratoryetup is presented in Fig. 2.

Fig. 2 – Schematic layout o

Fig. 3 – SEM images of unused ilmenite.

2.2. Oxygen carrier used

Ilmenite is an iron–titanium mineral, FeTiO3, and when usedas oxygen carrier FeTiO3 is the most reduced form. Ilmeniteis the most abundant of all titanium minerals and mined inlarge quantities. The term ilmenite is commonly used also forvarious ores containing more or less of the mineral ilmenite.Various compounds formed in oxidation of ilmenite are dis-cussed later in this paper. In this work, the term ilmenitenormally refers to the material used in the testing althoughin some cases it refers to the oxygen carrier system as such,i.e. ilmenite and its different chemical forms.

The ilmenite used in these experiments was supplied byTitania A/S. It was 94.3% pure and concentrated from a natu-rally occurring ore containing 40% ilmenite, 37% plagioclase,8.6% ortopyroxene, 6.5% klinopyroxene, 4.2% biotitt and someminor other phases. The iron and titanium molar ratio isclose to 1:1. Unused ilmenite had a BET area of 0.11 m2/gand SEM images of unused ilmenite particles are presentedin Figs. 3 and 4.

2.3. Data evaluation

In order to quantify the amount of converted gas the gas yield(�) is used, defined here as the fraction of CO2 in the outgoing

f the laboratory setup.

1020 chemical engineering research and d

slowly throughout the whole cycle. Worth noting is that mostof the CH4 passes through the system without reacting withthe ilmenite.

Fig. 4 – SEM images of unused ilmenite.

gas divided with the sum of the fractions of carbon contain-ing gases in the outgoing gas. A � of 1 corresponds to totalconversion of the fuel. With CH4 as reducing gas this gives

� = xCO2

xCO2 + xCO + xCH4

gas yield for experiments with CH4 (3)

xi is the fraction of component i in the outgoing gases after thewater has been removed. If CO is used as fuel the xCH4 = 0.

When H2 is used as reducing gas, it is also possible to definethe fraction of H2O in the outgoing gas divided by the sum ofthe fractions of hydrogen containing gases in the outgoing gas.But since neither the amount of H2 nor the amount of H2O wasdirectly measured in these tests this was not calculated.

In CLC the degree of conversion X of the oxygen carrier isused to quantify the conversion of the oxygen carrier and isdefined as the fraction of the difference between the mass ofthe oxygen carrier and the mass of the oxygen carrier in itsmost reduced state, and the difference between the mass ofthe oxygen carrier in its most oxidized state and in its mostreduced state.

X = m − mred

mox − mreddegree of conversion (4)

Often the degree of mass-based conversion ω is used as it isconvenient for comparisons of different materials. ω is definedas the mass of the oxygen carrier divided by the mass of theoxygen carrier in its most oxidized state.

ω = m

moxdegree of mass-based conversion (5)

ω has been calculated as the time integral of exhaust gas con-centration times the gas flow (Mattisson et al., 2004).

Since H2 could not be measured it was assumed to berelated to the measured faction of CO and CO2 through anempirical relation used in earlier work (Mattisson et al., 2001).This calculated fraction of H2 agreed reasonably well with thefraction calculated from the difference 1 − xCO2 − xCO − xCH4

in the part of the reduction period where there is little or

no back-mixing. It has previously been established that H2 ismore easily converted than CO (Abad et al., 2007b) and thegiven assumption about the conversion of H2 would then give

esign 8 6 ( 2 0 0 8 ) 1017–1026

an underestimation of the total degree of mass-based conver-sion of the oxygen carrier. However, since the conversion ofCO is high, in many cases close to 100%, the possible error inthis estimation is small.

X can easily be converted to ω via:

ω = 1 + Ro(X − 1) (6)

where Ro is the oxygen transfer capacity, i.e. the fraction ofavailable oxygen in the oxygen carrier.

Ro = mox − mred

mox(7)

2.4. Thermodynamic equilibrium calculations

Equilibrium data were calculated using HSC Chemistry® 5.0for Windows (Outokumpu Research Oy, 2002). The equi-librium composition is calculated using the Gibbs energyminimization method. The program finds the most stablephase combination and seeks the phase composition wherethe Gibbs energy of the system reaches its minimum at afixed mass balance at constant pressure and temperature. Oneassumption made is that all substances in the solid phase arein pure form and not mixtures. Further, it is assumed that H2

can be converted to H2O and CO to CO2. O2 is also included inthe equilibrium calculations.

3. Results from laboratory experiments

3.1. Concentration profiles

Fig. 5 shows the outlet gas concentrations after condensa-tion of water as a function of time for the eighth reducingperiod when methane was used as fuel. A sample of 15 g ofilmenite was used at a temperature of 970 ◦C. In Fig. 5, theCH4 is turned on at time 0 but the residence time in the sys-tem delays the response with 20–25 s before the CO2 and CH4

rapidly increase. CO2 reaches a maximum a few second laterwhereas CH4 increases through the whole cycle. CO increases

Fig. 5 – Concentration profile for the eighth reductionperiod with CH4 as reducing gas at a temperature of 970 ◦C.

chemical engineering research and design 8 6 ( 2 0 0 8 ) 1017–1026 1021

Fig. 6 – Concentration profile for the eighth reduction periodw

owttdt4cct

tdTmiTr

Ffldri

Fig. 8 – Gas yield (�) as a function of mass reduction of

ith syngas as reducing gas at a temperature of 975 ◦C.

In Fig. 6, the outlet gas concentrations after condensationf water is shown for the eighth reducing period when syngasas used as fuel. A sample of 15 g of ilmenite was used at a

emperature of 970 ◦C. After a period of 20–25 s into the cyclehe CO2 rapidly increases. Again the delay is due to gas resi-ence time in the system. Initially almost all CO is convertedo CO2 but some small amounts of unreacted CO are detected0 s after the CO2 and then CO increases through out the wholeycle. This means that the conversion in the beginning of theycle is close to 100%. However, CO2 does not reach 100% dueo back-mixing of nitrogen from the previous inert period.

Fig. 7 shows the eighth oxidation period, thus the oxidationhat follows the reduction in Fig. 5, and the last oxidation (oxi-ation with defluidized bed) when using CH4 as reducing gas.he inlet O2 content was 5%. During the first part of a nor-al oxidation all incoming oxygen reacts with the reduced

lmenite resulting in only inert nitrogen in the outgoing flow.

hen there is a rapid increase in oxygen concentration thateaches the inlet O2 content of 5%. There is no CO2 in the gas

ig. 7 – Concentration profile for the eighth (oxidation withuidization) and last oxidation period (oxidation withefluidization) with 5% O2 in nitrogen. In the previouseductions CH4was used as reducing gas. The temperatures 980 ◦C.

ilmenite (ω) for a number of cycles. The fluidizing gascontained pure methane and the temperature was 970 ◦C.

which means that no carbon was formed on the oxygen carrierin the reduction period. The oxidation looked very much thesame, independent of the previous reduction, as long as thebed fluidizes. If the bed stopped fluidizing it resulted in initiallyhigher O2 concentration and longer oxidation periods, as alsocan be seen in Fig. 7. Defluidization leads to channelling andto bypass of the gas stream that makes the contact betweengas and particles less efficient which explains the higher O2

concentration and the longer oxidation period.

3.2. Experiments with increased length of thereduction

In Fig. 8 the gas yield, �, is shown as a function of mass reduc-tion of ilmenite, ω, with methane as fuel for a number ofcycles. It is obvious that the reactivity of the oxygen carrierincreases with every cycle. The length of the reduction periodwas increased with each cycle until the bed stopped fluidiz-ing in the end of the last reducing period. The second lastreducing period, which was the last reducing period with nor-mal fluidization, can be seen in Fig. 5. Defluidization of oxygencarriers in CLC batch experiments has previously been inves-tigated for Ni-, Fe- and Mn-based oxygen carriers by Cho et al.(2006).

In Fig. 9, � is shown as a function of ω when syngas is usedas fuel for a number of cycles with increasing cycle length.Although the reactivity towards CO is initially much higherthan for CH4, the increase is still remarkable. As an exam-ple there is an increase in � from 90% to approximately 100%for ω = 0.99. Just as in Fig. 8 the length of the reduction wasincreased with every cycle until the bed defluidized in orderto investigate how the degree of reduction affected the flu-idization properties. The last cycle with defluidization is notshown in Fig. 9. The second last reducing period which wasthe last reducing period with normal fluidization can be seenin Fig. 6. Just as in the experiments with CH4 the bed stoppedfluidizing in the end of the last reducing period.

Defluidization occurred at an ω of 0.975 for methane and0.945 for syngas probably due to sudden particle cohesion or

other effects that lead to channelling and to a bypass of thegas stream. However, tendencies of defluidization were seenalready at an ω of 0.96 for syngas. Also, defluidization occurred

1022 chemical engineering research and design 8 6 ( 2 0 0 8 ) 1017–1026

Fig. 9 – CO conversion (�) as a function of mass reduction ofilmenite (ω) for a number of cycles. The fluidizing gas

Fig. 10 – CO conversion (�) as a function of number of cyclesfor ilmenite. The fluidizing gas contained 50% H2 and 50%CO and the temperature was 975 ◦C.

Fig. 11 – SEM images of reduced oxygen carrier used inlong-term experiments.

Table 1 – Phases indicated by XRD

XRD phases indicated

Unused particles FeTiO3, Fe3Ti3O10, possibly someFe2O3 and Ti2O3

Particles after oxidation Fe2TiO5, Fe3Ti3O10, Fe2O3, possiblysome FeTiO3

contained 50% H2 and 50% CO and the temperature was975 ◦C.

in the beginning of experiments when unused ilmenite wasoxidized for the first time. However, after this first oxidationdefluidization was no longer a problem except for the last longreducing cycles in Figs. 8 and 9. Since unused ilmenite, just asthe oxygen carriers in the end of the last long reducing cycles,is in a reduced state, it can be concluded that defluidizationof ilmenite is linked with highly reduced oxygen carriers. Thisis a phenomenon previously seen for iron oxide particles (Choet al., 2006). It should be mentioned that the same ilmeniteparticles have been used in a 10-kW CLC-unit for long periodswithout difficulties of defluidization (Berguerand and Lyngfelt,2008).

3.3. Long-term experiment

A long-term experiment was made using syngas to investi-gate the effect on ilmenite after many cycles. Thirty-sevencycles were performed during 3 days, during approximately25 h under hot conditions. In the night-time the oven wasturned off and so was the flow when the sample had cooleddown to room temperature. During experiments the temper-ature was 975 ◦C, the flow of syngas was 450 mLn/min andthe reducing periods were 50 s. In Fig. 10, the fraction of COconverted in each cycle is presented versus cycle number.The conversion of CO increased during the first cycles beforestabilizing, giving good reproducibility for the rest of the exper-iment.

3.4. Particle analysis

SEM images were obtained on the particles used during thelong-term experiment and are presented in Figs. 11 and 12.The long-term experiment ended with a reduction and theparticles are therefore in a reduced state. These imagescan be compared to the images of unused particles inFigs. 3 and 4. It looks like the porosity of the ilmenite particleshas increased with clear cracks seen in the used particle. TheBET-measurements showed an increase from 0.11 to 0.28 m2/g

for used particles.

Table 1 presents the results from X-ray diffraction (XRD)made on unused ilmenite particles, fully oxidized particles

and reduced particles used in the long-term experiment. Asseen several different species can be identified. To be noted isthat the XRD shows peaks that have not been possible to iden-tify. Thus, the XRD data should be viewed with some caution.

Particles after subsequentreduction

FeTiO3, Fe3O4, some TiO2 and FeO,possibly some Fe3Ti3O10

chemical engineering research and design 8 6 ( 2 0 0 8 ) 1017–1026 1023

Fig. 12 – SEM images of reduced oxygen carrier used inl

4e

Tiwm

tTlFeoieF

iFTtA

FF

Table 2 – Species included in equilibrium calculation inFig. 13

FeO*TiO2

TiO2

Fe2TiO5

Fe3O4

Fe0.945OFe0.947OFeOFeO1.056

FeO1.5(W)Fe2O3

Fe2O3(G)Fe2O3(H)Fe3O4(H)*2FeO*TiO2

FeTiO3

FeTi2O5

Fe2TiO4

TiOTiO(B)TiO1.01

TiO2(A)Ti2O3

Ti3O2

Ti3O5

Ti4O7

Ti5O9

Ti6O11

Ti7O13

Ti8O15

Ti9O17

Ti10O19

ong-term experiments.

. Results from thermodynamicquilibrium calculations

he mineral ilmenite has a molar ratio Fe:Ti of 1:1. The actuallmenite that has been studied also has a molar ratio Fe:Ti

hich is close to 1:1. The calculations below have all beenade using this molar ratio.Fig. 13 shows calculated equilibrium for the gradual oxida-

ion of ilmenite by oxygen (Outokumpu Research Oy, 2002).he species included in the equilibrium calculations are

isted in Table 2. According to Fig. 13 the oxidation goes viae3O4 + TiO2, to Fe2TiO5 + TiO2. The compounds noted in thequilibrium calculation are all found by the XRD. However, onef the important compounds noted in the XRD, i.e. Fe3Ti3O10

s not found in the database. Reduction of Fe2TiO5 + TiO2 by,.g. H2 or CO will be the reverse and proceed to FeTiO3 viae3O4 + TiO2.

It can be seen that the oxidation of ilmenite, as seenn Fig. 13, involves three distinct oxidation levels, FeTiO3,e3O4 + TiO2, and Fe2TiO5 + TiO2. These are indicated inable 3. Note that in level of oxidation FeTiO3 corresponds

o FeO + TiO2, and Fe2TiO5 + TiO2 corresponds to Fe2O3 + TiO2.lso added in the table is Fe3Ti3O10, found by XRD, and which

ig. 13 – Oxidation of 60 kmol of ilmenite by O2 at 950 ◦C.eO*TiO2 is equal to FeTiO3.

Ti20O39

corresponds to Fe3O4 + 3TiO2. The levels are here denominated0–2, where level 0 corresponds to the oxidation state 2+ foriron (FeII) and level 2 has the oxidation state 3+ for iron (FeIII),whereas level 1 is a mixture of 2+ and 3+, FeIII

2 FeII. The tablealso indicates the oxygen ratio, Ro. Note that the table includesall species identified by XRD as well as those found in theequilibrium calculations.

Assuming that the possible reactions proceeds between thelevels, the following reactions are possible:

• Oxygen release from level 1 to 0:

2Fe3O4 + 6TiO2 ⇒ 6FeTiO3 + O2 (8a)

2Fe3Ti3O10 ⇒ 6FeTiO3 + O2 (8b)

2Fe3Ti3O10 ⇒ 6FeO + 6TiO2 + O2 (8c)

2Fe3O4 ⇒ 6FeO + O2 (8d)

• Oxygen release from level 2 to 1:

6Fe2TiO5 + 6TiO2 ⇒ 4Fe3O4 + 12TiO2 + O2 (9a)

6Fe2TiO5 + 6TiO2 ⇒ 4Fe3Ti3O10 + O2 (9b)

6Fe2O3 + 12TiO2 ⇒ 4Fe3Ti3O10 + O2 (9c)

6Fe2O3 ⇒ 4Fe3O4 + O2 (9d)

• Oxygen release from level 2 directly to 0:

1024 chemical engineering research and design 8 6 ( 2 0 0 8 ) 1017–1026

Table 3 – Ilmenite, level of oxidation

Level of oxidation Compound Corresponding Fe/Ti compound Added oxygen Ro to level 0 (%) Ro to level 1 (%)

0 FeTiO3 FeO + TiO2 – – –1 (1/3) Fe3Ti3O10 (1/3) Fe3O4 + TiO2 1/3 3.4 –

2 (1/2) Fe2TiO5 + 1/2 TiO2 (1/2) Fe2O3 + TiO2

2Fe2TiO5 + 2TiO2 ⇒ 4FeTiO3 + O2 (10a)

2Fe2TiO5 + 2TiO2 ⇒ 4FeO + 4TiO2 + O2 (10b)

2Fe2O3 + 4TiO2 ⇒ 4FeTiO3 + O2 (10c)

2Fe2O3 ⇒ 4FeO + O2 (10d)

Here the reduction reactions are shown as oxygen release withone O2 in each, which corresponds to the reduction by two H2

or two CO.Based on the equilibrium calculations reactions, the reac-

tions that take place are (8a) and (9a), which togetherconstitute the total reaction (10a). However, based on whatis seen in XRD, it is likely that the reaction instead proceedsvia Fe3Ti3O10, i.e. following the compounds in the second col-umn in Table 3. This means that oxidation/reaction proceedsvia reactions (8b) and (9b), which again constitutes the totalreaction (10a). Note that the equilibrium calculations do notinclude this compound, and yet it is probably seen in the XRD.But the XRD-spectrum for Fe2TiO5 and Fe3Ti3O10 only differsin a few peaks and may be difficult to separate. Based on bothXRD and equilibrium data, it is unlikely that reactions (8c) and(8d) are important, which also applies to the total reactions(10b) and (10d). It is also believed that the reactions (9c) and(9d) are less important, which would also then apply to reac-tions (10c) and (10d). The reaction path is further discussed inthe section “oxidation of ilmenite” below.

In Table 4 the ratio of the heat of reaction for variousreactions is compared to the corresponding reaction wheremethane is oxidized, per mole of oxygen. If the ratio is higherthan unity this means that more heat is produced in theair reactor per mole of oxygen compared to combustion ofmethane. The consequence is that the reaction in the fuelreactor is endothermic. As can be seen in Table 4 the reac-tion in the fuel reactor would be endothermic for all of thereactions where data are available. However, the fuel reactorreactions corresponding to reactions (8a), (9a) and (10a) are

less endothermic compared to what would be the case forthe oxygen carrier systems NiO/Ni and Fe2O3/Fe3O4. For theoverall reaction (10a) the ratio is 1.108, which means that the

Table 4 – Heats of reaction and equilibria

Reaction Comment �H/�HCH

(8a) 1.088(8b) Most important? ?(8c) Not important ?(8d) Not important 1.479(9a) 1.147(9b) Most important? ?(9c) Less important ?(9d) Less important 1.197(10a) 1.108(10b) Not important 1.363(10c) Less important 1.124(10d) Not important 1.379

1/2 5.0 1.67

endothermic reaction in the fuel reactor would correspond to10.8% of the total thermal power. For coal the heating value(per mole of oxygen) is estimated to be slightly more than 2%higher than for methane. This would mean that the endother-mic heat needed for coal corresponds to about 9% of the totalthermal power. This value could be off-set if, for instance, reac-tions going from level 2 to 1 dominate, and this gives a moreendothermic reaction. Thermodynamic data for Fe3Ti3O10 aremissing, which adds some uncertainty if it is an intermedi-ate.

Also shown in Table 4 are the equilibria. It is clear that bygoing from level 2 to 1 it is possible to achieve very high conver-sion, and high conversion is also possible when going to level0. The H2 and CO fractions are about 0.2 and 0.3% at equilib-rium. However, note the equilibrium is more favourable for thedirect reaction (10a).

The stability with respect to temperature of the com-pounds seen in Fig. 13 was also investigated. It was foundthat Fe2TiO5 is converted to Fe2O3 + TiO2 at temperaturesbelow 500–550 ◦C. It is therefore possible that Fe2O3 foundby XRD could form when temperature is lowered. Alsoat oxidation level 1 a change was seen at lower temper-atures, 150–200 ◦C, where Fe3O4 + 2TiO2, are converted toFeTiO3 + Fe2O3 + TiO2.

4.1. Oxidation of ilmenite

According to Borowiec and Rosenqvist (1991) the oxida-tion of ilmenite goes via hematite (Fe2O3), in solid solutionwith ilmenite, to pseudobrookite (Fe2TiO5), at temperaturesbetween 950 and 1000 ◦C. At 900 ◦C the end products werehematite, pseudobrookite and rutile (TiO2).

Briggs and Sacco (1993) oxidized ilmenite at 867 ◦C andfound a solid solution of ilmenite and hematite, max 7mol%, and upon further oxidation, Fe2TiO5 and FeTi2O5 solidsolutions and rutile. More of FeTi2O5 was found at lower tem-perature, 800 ◦C.

Gupta et al. (1991) reported that a new phase “Fe2O3•2TiO2”,

was identified as an intermediate product during heating ofilmenite in oxygen. This intermediate decomposes into pseu-dobrookite and rutile at temperatures above 800 ◦C.

4 CO/(CO + CO2) H2/(H2 + H2O)

0.00321 0.00215? ?? ?0.424 0.3300.00002 0.00001? ?? ?0.00001 0.000010.0006 0.00040.0222 0.01500.0005 0.00030.0175 0.0118

desi

pp

att

•

•

irt(F

2

4

2

lerfsm

5

WtItfsfs2trt

tCt(00

Rfs

a

chemical engineering research and

Zhang and Ostrovski (2002), also found the metastablehase “Fe2O3

•2TiO2” as an intermediary at 1000 ◦C, and an endroduct of Fe2TiO5 and TiO2.

Many papers have been written on the oxidation of ilmenitet different temperatures. The four cited above appears to behe most relevant for this work. The literature data indicatehe following:

There seems to be no intermediate oxidation step, level 1,with FeIII

2 FeII, i.e. compounds like Fe3O4 or Fe3Ti3O10 havenot been found. In fact it seems to be very difficult to find anyinformation on the compound Fe3Ti3O10 possibly identifiedby XRD. Instead the oxidation seems to go directly to level2, i.e., FeIII is formed directly.Reaction intermediates are either Fe2O3 or “Fe2O3

•2TiO2”.

Thus, assuming this is correct; the reaction sequencenvolving level 1 can be discarded. Instead we have a directeaction between level 2 and 0, either through the direct reac-ion (10a) or via the intermediate Fe2O3 according to reactions10c) and (11) or alternately via the reaction intermediatee2O3

•TiO2 according to reactions (12) and (13).

Fe2O3 + 2TiO2 ⇒ 2Fe2TiO5 (11)

FeTiO3 + O2 ⇒ 2(“Fe2O3•2TiO2

′′) (12)

(“Fe2O3•2TiO2

′′) ⇒ 2Fe2TiO5 + 2TiO2 (13)

Note that the reactions (10a) and (10c) have very simi-ar reaction enthalpies, the difference between the reactionnthalpies being approximately 1.5%. Thus, it appears thateaction (10a) can be used to represent the reaction enthalpyor oxidation of ilmenite with good accuracy. However, ithould be pointed out that intermediates seen in oxidationay not necessarily be intermediates in the reduction.

. Discussion

hen natural gas is used as fuel in CLC it is of great impor-ance that the particles have good reactivity towards methane.f on the other hand some kind of solid fuel is used the reac-ivity towards syngas is probably more important, since soliduels in chemical-looping combustion goes through a twotage mechanism: (1) pyrolysis/gasification of the solid fuelorming a syngas of mainly CO and H2, followed by (2) theubsequent reaction of the gases with metal oxide (Leion et al.,007). However, solid fuels normally also release volatiles, con-aining methane and other hydrocarbons, and therefore theeactivity of the oxygen carrier towards methane also needso be considered.

This work shows that ilmenite has a very good reac-ivity towards syngas. Calculated equilibrium ratios forO/(CO2 + CO) and H2/(H2O + H2) are 0.0006 and 0.0004, respec-

ively, assuming the reaction is direct according to reaction10a). Experiments yielded ratios for CO/(CO2 + CO) down to.007–0.002, with a measurement inaccuracy estimated to.005.

According to stoichiometric calculations ilmenite has an

o of 5.0% between the most reduced and the most oxidizedorms. The experiments with syngas in Fig. 9 indicate that

uch a high Ro is possible.

A key in the development of oxygen carriers is to have anwareness of particle cost. To motivate a high cost the parti-

gn 8 6 ( 2 0 0 8 ) 1017–1026 1025

cles have to be more reactive and/or have a longer lifetime.Ilmenite has properties that in many respects are comparableto the best synthetically produced particles (Johansson et al.,2006c) but is significantly cheaper.

For ilmenite the reaction enthalpy of the overall oxidationis 11% higher than for direct oxidation of methane with air ifcalculated per kmol of oxygen. A similar ratio can be expectedfor solid fuels which have heating values similar to or evenslightly higher than methane, calculated per kmol of oxygen.This means that the endothermic reaction in the fuel reactorof a CLC-unit would correspond to 11% of the total thermalpower. Thus, the reaction in the fuel reactor is endother-mic, but less endothermic compared to that of NiO/Ni andFe2O3/Fe3O4, and almost similar to Mn3O4/MnO. Suggested cir-culation rate of 380 kg/MWth min have been suggested whenusing Fe2O3/Fe3O4 supported on MgAl2O4 as oxygen carrier ina CLC-unit with solid fuel (Leion et al., 2008). With the sameassumption, but with ilmenite as oxygen carrier, a recircula-tion flow of 240 kg/MWth min would be sufficient, decreasingthe needed solids inventory in the fuel reactor of a CLC-unitby almost a third.

6. Conclusions

The main conclusions from this work are

• Ilmenite is an attractive and inexpensive oxygen carrier forCLC.

• Ilmenite gives high conversion of CO and moderate conver-sion of CH4.

• Ilmenite showed an oxygen ratio, Ro, of 5% and high reac-tivity towards CO in this interval, i.e. for an ω of 1–0.95.

• Defluidization occurred only when the ilmenite particleswere in a highly reduced state.

• Ilmenite showed no tendency of decreased reactivity after37 cycles or 3 days of experiments.

• The oxidation/reduction of ilmenite proceeds betweenthe reduced form ilmenite, FeTiO3, and the oxidizedform Fe2TiO5 + TiO2, with the likely intermediate Fe2O3

or Fe2O3•TiO2. Other intermediates cannot be entirely

excluded.

Acknowledgments

The work was made in the EU financed research projectEnhanced Capture of CO2 (ENCAP), SES6-2004-502666 andStatens Energimyndighet (STEM) Dnr 2006-04665 Projektnr21670-2. Titania A/S provided the ilmenite.

References

Abad, A., Mattisson, T., Lyngfelt, A. and Ryden, M., 2006,Chemical-looping combustion in a 300 W continuouslyoperating reactor system using a manganese-based oxygencarrier. Fuel, 85(9): 1174–1185.

Abad, A., Mattisson, T., Lyngfelt, A. and Johansson, M., 2007, Theuse of iron oxide as oxygen carrier in a chemical-loopingreactor. Fuel, 86(7–8): 1021–1035.

Abad, A., Adanez, J., Garcia-Labiano, F., de Diego, L.F., Gayan, P.and Celaya, J., 2007, Mapping of the range of operationalconditions for Cu-, Fe-, and Ni-based oxygen carriers inchemical-looping combustion. Chem Eng Sci, 62(1–2): 533–549.

Adanez, J., de Diego, L.F., Garcia-Labiano, F., Gayan, P., Abad, A.and Palacios, J.M., 2004, Selection of oxygen carriers forchemical-looping combustion. Energy Fuels, 18(2): 371–377.

and d

Zhang, G. and Ostrovski, O., 2002, Effect of preoxidation and

1026 chemical engineering research

Adanez, J., Garcia-Labiano, F. de Diego, L. F., Gayan, P., Abad, A.,Celaya, J., 2005, Development of oxygen carriers forchemical-looping combustion. Carbon Dioxide Capture forStorage in Deep Geologic Formations—Results from the CO2

Capture Project, 1, pp. 587–604.Adanez, J., Gayan, P., Celaya, J., de Diego, L.F., Garcia-Labiano, F.

and Abad, A., 2006, Chemical Looping Combustion in a10 kWth prototype using a CuO/Al2O3 oxygen carrier: effect ofoperating conditions on methane combustion. Ind Eng ChemRes, 45(17): 6075–6080.

Berguerand, N. and Lyngfelt, A., 2008, The use of petroleum cokeas fuel in a 10 kWth chemical-looping combustor. Int JGreenhouse Gas Control, 2(2): 169–179.

Berguerand, N., Lyngfelt, A., accepted for publication, Design andOperation of a 10 kWth chemical-looping combustor for solidfuels—testing with South African coal. Fuel.

Borowiec and Rosenqvist., 1991, Scand J Metall, 10: 217–224.Briggs and Sacco., 1993, Metall Trans A, 24A: 1257–1264.Cao, Y., Cheng, Z., Meng, L., Riley, J.T. and Pan, W.-P., 2005,

Reduction of solid oxygen carrier (CuO) by solid fuel (coal) inchemical looping combustion, In Preprints ofSymposia-American Chemical Society, Division of Fuel Chemistry(50)1 , pp. 99–102.

Cho, P., Mattisson, T. and Lyngfelt, A., 2006, Defluidizationconditions for fluidized-bed of iron, nickel, and manganeseoxide-containing oxygen-carriers for chemical-loopingcombustion. Ind Eng Chem Res, 45(3): 968–977.

Copeland, J.R., Alptekin, G., Cesario, M. and Gershkovich, Y., 2002,Sorbent energy transfer system (SETS) for CO2 separation withhigh efficiency, In Proceedings of the 27th International Conferenceon Coal Utilization & Fuel Systems

de Diego, L.F., Garcia-Labiano, F., Gayan, P., Celaya, J., Palacios, J.M.and Adanez, J., 2007, Operation of a 10 kWth chemical-loopingcombustor during 200 h with a CuO–Al2O3 oxygen carrier.Fuel, 86(7–8): 1036–1045.

Dennis, J.S., Scott, S.A. and Hayhurst, A.N., 2006, In situgasification of coal using steam with chemical looping: atechnique for isolating CO2 from burning a solid fuel. J EnergyInst, 79(3): 187–190.

Gupta, Rajakumar and Grieveson., 1991, Metall Trans B, 22B:711–716.

Gupta, P., Velazquez-Vargas, L.G., Thomas, T. and Fan, L.S., 2005,Chemical looping combustion of coal to produce hydrogen, InProceedings of the 30th International Technical Conference on CoalUtilization & Fuel Systems, vol. 1 , pp. 349–352.

Ishida, M., Jin, H. and Okamoto, T., 1996, A fundamental study ofa new kind of medium material for chemical-loopingcombustion. Energy Fuels, 10(4): 958–963.

Jerndal, E., Mattisson, T. and Lyngfelt, A., 2006, Thermal analysisof chemical-looping combustion. Chem Eng Res Des, 84(A9):795–806.

Johansson, M., 2007, Screening of oxygen-carrier particles basedon iron-, manganese-, copper- and nickel oxides for use inchemical-looping technologies, Dept. of Chemical andBiological Engineering, Environmental Inorganic Chemistry,PhD Thesis.

Johansson, E., Mattisson, T., Lyngfelt, A. and Thunman, H., 2006,A 300 W laboratory reactor system for chemical-loopingcombustion with particle circulation. Fuel, 85(10–11):1428–1438.

Johansson, E., Mattisson, T., Lyngfelt, A. and Thunman, H., 2006,

Combustion of syngas and natural gas in a 300 Wchemical-looping combustor. Chem Eng Res Des, 84(A9):819–827.

esign 8 6 ( 2 0 0 8 ) 1017–1026

Johansson, M., Mattisson, T., Ryden, M. and Lyngfelt, A., 2006,Carbon capture via chemical-looping combustion andreforming, In International Seminar on Carbon Sequestration andClimate Change

Leion, H., Mattisson, T. and Lyngfelt, A., 2007, The use ofpetroleum coke as fuel in chemical-looping combustion. Fuel,86(12–13): 1947–1958.

Leion, H., Mattisson, T. and Lyngfelt, A., 2008, Solid fuels inchemical-looping combustion. Int J Greenhouse Gas Control,2: 180–193.

Linderholm, C., Abad, A., Mattisson, T., Lyngfelt, A, accepted forpublication, 160 h of chemical-looping combustion in a 10 kWreactor system with a NiO-based oxygen carrier. Int JGreenhouse Gas Control.

Lyngfelt, A., Thunman, H. 2005, Construction and 100 h ofoperational experience of a 10-kW chemical-loopingcombustor. Carbon Dioxide Capture for Storage in DeepGeologic Formations—Results from the CO2 Capture Project, 1,pp. 625–645.

Lyngfelt, A., Kronberger, B., Adanez, J., Morin, J.-X. and Hurst, P.,2004, The GRACE project. Development of oxygen carrierparticles for chemical-looping combustion. Design andoperation of a 10 kW chemical-looping combustor, InProceedings of the Seventh International Conference on GreenhouseGas Control Technologies

Lyon, R.K. and Cole, J.A., 2000, Unmixed combustion: analternative to fire. Combust Flame, 121: 249–261.

Mattisson, T., Lyngfelt, A. and Cho, P., 2001, The use of iron oxideas an oxygen carrier in chemical-looping combustion ofmethane with inherent separation of CO2. Fuel, 80(13):1953–1962.

Mattisson, T., Johansson, M. and Lyngfelt, A., 2004, Multicyclereduction and oxidation of different types of iron oxideparticles-application to chemical-looping combustion. EnergyFuels, 18(3): 628–637.

Mattisson, T., Johansson, M. and Lyngfelt, A., 2006, CO2 capturefrom coal combustion using chemical-loopingcombustion—reactivity investigation of Fe, Ni and Mnbased oxygen carriers using syngas, In Clearwater CoalConference

Mattisson, T., Lyngfelt, A., Leion, H., accepted for publication,Chemical-looping with oxygen uncoupling for combustion ofsolid fuels. Int J Greenhouse Gas Control.

Outokumpu Research Oy, 2002, HSC Chemistry® 5.0 for Windows,Chemical Reaction and Equilibrium Software with ExtensiveThermochemical Database.

Pan, W.-P., Cao, Y., Liu, K., Wu, W., Riley, J.T., 2004, Application of acirculating fluidized-bed process for the chemical loopingcombustion of solid fuels. Abstracts of Papers, 228th ACSNational Meeting, Philadelphia, PA, United States, August22–26, 2004, FUEL-155.

Ryu, H.-J., Jin, G.-T., Bae, D.-H., Yi, C.-K., 2004. Continuousoperation of a 50 kWth chemical-looping combustor:long-term operation with Ni- and Co-based oxygen carrierparticles. Presented at the Fifth China-Korea Joint Workshopon Clean Energy Technology, Qingdao University, China,October 25–28, 2004, 221–230.

Scott, S.A., Dennis, J.S., Hayhurst, A.N. and Brown, T., 2006, In situgasification of a solid fuel and CO2 separation using chemicallooping. AIChE J, 52(9): 3325–3328.

sintering on properties of ilmenite concentrates. Int J MinerProcess, 64: 201–218.