i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 Available online at www.sciencedirect.com journal homepage: www.elsevier.com/locate/he Continuous hydrogen production via the steameiron reaction by chemical looping in a circulating fluidized-bed reactor Magnus Ryde´n a,*, Mehdi Arjmand b a Department of Energy and Environment, Division of Energy Technology, Chalmers University of Technology, SE-412 96 Goăteborg, Sweden Department of Chemical and Biological Engineering, Division of Environmental Inorganic Chemistry, Chalmers University of Technology, SE-412 96 Goăteborg, Sweden b article info abstract Article history: The steameiron reaction was examined in a two-compartment fluidized-bed reactor at Received 24 October 2011 800e900  C and atmospheric pressure In the fuel reactor compartment, freeze-granulated Received in revised form oxygen carrier particles consisting of Fe3O4 supported on inert MgAl2O4 were reduced to 24 November 2011 FeO with carbon monoxide or synthesis gas The reduced particles were transferred to Accepted December 2011 a steam reactor compartment, where they were oxidized back to Fe3O4 by steam, while at Available online 10 January 2012 the same time producing H2 The process was operated continuously and the particles were transferred between the reactor compartments in a cyclic manner In total, 12 h of Keywords: experiments were conducted of which h involved H2 generation The reactivity of the Steameiron process oxygen carrier particles with carbon monoxide and synthesis gas was high, providing gas Chemical-looping combustion concentrations reasonably close to thermodynamic equilibrium, especially at lower fuel Chemical-looping reforming flows The amount of H2 produced in the steam reactor was found to correspond well with Hydrogen the amount of fuel oxidized in the fuel reactor, which suggests that all FeO that was formed Iron oxide were also re-oxidized Despite reduction of the oxygen carrier to FeO, defluidization or stops in the solid circulation were not experienced Used oxygen carrier particles exhibited decreased BET specific surface area, increased bulk density and decreased particle size compared to fresh This indicates that the particles were subject to densification during operation, likely due to thermal sintering However, stable operation, low attrition and absence of defluidization were still achieved, which suggest that the overall behaviour of the oxygen carrier particles were satisfactory Copyright ª 2011, Hydrogen Energy Publications, LLC Published by Elsevier Ltd All rights reserved Introduction Hydrogen (H2) is an important feedstock with many applications such as in the production of ammonia and fertilizers, upgrading of fuels in the refining industry, methanol synthesis, manufacturing of electronics and metallurgic processes There is also an increasing interest in hydrogen as a future energy carrier, see for example the review by Ogden [1] The so-called hydrogen economy would be favourable in a number of ways When H2 is burnt, the only product is water vapour (H2O) Therefore vehicles using H2 as fuel rather than petroleum products would neither emit greenhouse gases such as carbon dioxide (CO2) and methane (CH4), nor other harmful carbon based pollutants such as carbon monoxide (CO), soot or particulate matter * Corresponding author Tel.: ỵ46 31 7721457 E-mail address: magnus.ryden@chalmers.se (M Ryden) 0360-3199/$ e see front matter Copyright ª 2011, Hydrogen Energy Publications, LLC Published by Elsevier Ltd All rights reserved doi:10.1016/j.ijhydene.2011.12.037 4844 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 Nomenclature ar, AR CLC CnHm F fr, FR Ln/min m n M NG Air reactor Chemical-looping combustion Generic hydrocarbon fuel Volumetric flow (Ln/min) Fuel reactor Normal litres per minute Mass (g) Number of moles Molar mass Natural gas The dominating industrial route for generation of H2 currently is steam reforming of natural gas However, in order for H2 to be a feasible and environmentally benign energy carrier it needs to be produced as cheap and efficient as possible, and preferably with CO2 capture and sequestration (CCS) For a general assessment of the different options for which this could be achieved, see Mueller-Langer et al [2] or Cormos [3] for an overview of hydrogen fuelled power generation schemes with CCS H2 production via the steameiron reaction involves oxidation of reduced iron oxides with steam Although outdated, recent technological advances such as the development of chemical-looping combustion could possibly make the steameiron process an attractive process yet again Fully developed, a combined process involving H2 production via the steameiron reaction and regeneration of the iron oxide via chemical-looping combustion would provide high purity H2 with inherent CO2 capture, without the need for wateregas shift reactors, gas purification or other costly downstream processing Background 2.1 Steameiron process The steameiron process is one of the oldest methods for industrial production of high purity H2 The process was developed in the beginning of the 20th century by pioneers such as Messerschmitt [4] and Lane [5], mainly for production of H2 for airships and balloons The conventional steameiron process uses iron oxide to reduce steam to hydrogen In the first step of the process, iron oxide is reduced from hematite (Fe2O3), to magnetite (Fe3O4), to wustite (FeO), and sometimes all the way to metallic iron (Fe) Typically, gasified coal was used to perform the reduction but a modern process could use a wide range of fuels such as gasified biomass, natural gas, petroleum products, industrial waste gas from blast furnaces or refineries etc With for instance CO as reducing gas, the product is reduced iron oxide and CO2, see reactions (1)e(3): 3Fe2O3(s) þ CO(g) / 2Fe3O4(s) þ CO2(g) DH900 C ¼ À35.3 kJ/mol (1) ox PSD s SG SIR sr, SR t wt.% DH x gCO2 u Oxidized sample Particle size distribution Sample Synthesis gas Steameiron reaction Steam reactor Time (s, min) Percentage by weight Heat of reaction (kJ/mol) Dry-gas concentration (%) CO2 yield (%) Mass-based degree of reduction Fe3O4(s) ỵ CO(g) / 3FeO(s) ỵ CO2(g) DH900 C ẳ 10.1 kJ/mol (2) FeO(s) ỵ CO(g) / Fe(s) ỵ CO2(g) DH900 C ẳ À16.3 kJ/mol (3) Naturally, the reduction could as well be performed with H2, producing H2O as product, or with hydrocarbons producing a mix of CO2 and H2O Regardless of fuel choice, reduction proceeds until the desired amount of FeO and Fe is obtained, at which point the reducing gas is switched to steam In the second step of the process, H2 is produced by oxidizing FeO and Fe with steam in accordance with reactions (4) and (5): Fe(s) ỵ H2O(g) / FeO(s) ỵ H2(g) DH900 C ẳ 16.8 kJ/mol (4) 3FeO(s) ỵ H2O(g) / Fe3O4(s) ỵ H2(g) DH900 C ẳ À43.2 kJ/mol (5) It is necessary to provide steam in excess, nevertheless pure H2 is obtained when the product mixture is cooled down and the steam is condensed to liquid water Due to thermodynamic constraints, it is not possible to oxidize Fe3O4 to Fe2O3 with steam If desirable, this oxidation step has to be performed with oxygen provided for example with air, see reaction (6): 2Fe3O4(s) ỵ ẵO2(g) / 3Fe2O3(s) DH900 C ¼ À232.2 kJ/mol (6) Reaction (6) is strongly exothermic Since the reduction of Fe3O4 to FeO, reaction (2) is endothermic when a hydrocarbon containing gas is used, reaction (6) is necessary in this case in order to obtain a continuous and thermally balanced process Reaction (6) is also needed in order to burn away sulphur and coke, which otherwise may accumulate on the surface of the iron oxide particles The steameiron process has not been used commercially for several decades Typically, the reactions outlined above were performed batch-wise at temperatures in the range of 550e900  C Each reaction step release or requires certain amounts of heat and their reaction kinetics and thermodynamics are dependent upon temperature Hence it is preferred to conduct the different reactions at different temperature levels, a fact that made the batch-wise steameiron process hard to optimize and as a consequence not overwhelmingly efficient 4845 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 The petrochemical revolution in the second half of the 20th century resulted in the development of new methods for large scale H2 production, such as steam reforming of natural gas, which were seen as economically more attractive In this era, steameiron processes built up from fluidized beds were also developed [6e9] Although these processes did not emerge as commercially viable alternatives, they did demonstrate that continuous and efficient operation of the steameiron process is possible In later years, the interest in the steameiron process had grown The increased consciousness about the connection between emissions of CO2 and anthropogenic climate change is one reason for this surge of interest The steameiron process could be configured so that pure CO2 is delivered in a separate process stream, and thus seems as a convenient method for H2 production with CO2 capture Another reason for the increased interest could be the fact that the dominant technology used for H2 production today, which is steam reforming of CH4, requires light hydrocarbons such as natural gas as feedstock which makes the long-term prospects for current technology somewhat uncertain Further, new insights in research areas such as catalysis, thermodynamics, process integration and combustion in fluidized beds, as well as the development of entirely new technologies such as chemical-looping combustion, could help facilitate the realization of a new generation of steameiron process In recent years, several research groups have presented interesting results concerning processes for use of the steameiron reaction for H2 production Fan et al [10] have suggested various processes for conversion of coal to H2, some of which involves the steameiron reaction Chiesa et al [11] have conducted a detailed process study, examining a three-reactor chemical-looping process with H2 generation by the steameiron process On the experimental and the material development side, batch experiments in fixed-bed reactor [12e15] using both ordinary iron oxides and iron oxide supported on magnesium, silica, chromium, titania and aluminium have been conducted by several research groups Lorente et al [16] also performed thermogravimetric analysis using iron ores Further, Bleeker et al [17] used pyrolysis oil for reduction of iron oxide in a batch fluidized bed, followed by H2 generation by oxidation with steam Yang et al [18] presented similar experiments, using char for direct reduction of the iron oxide 2.2 Chemical-looping combustion Chemical-looping combustion (CLC) is an innovative method for oxidation of fuels with inherent CO2 separation In this process, two separate reactor vessels are used with a solid oxygen carrier performing the task of transporting oxygen between the reactors as shown in Fig In the fuel reactor (FR), the oxygen carrier is reduced by the fuel, which in turn is oxidized to CO2 and H2O In the air reactor (AR), the reduced oxygen carrier is re-oxidized to its initial state with O2 from air Different kinds of oxygen carrier materials can be used The most commonly proposed are iron oxide, manganese oxide, copper oxide and nickel oxide [19] Reactions (7) and (8) describe the chemical-looping combustion of methane, using iron oxide as oxygen carrier, in the fuel (FR) and the air reactor (AR), respectively: CH4(g) ỵ 12Fe2O3(s) / 8Fe3O4(s) ỵ CO2(g) ỵ 2H2O(g) DH900 C ẳ 184.0 kJ/mol (7) 2O2(g) ỵ 8Fe3O4(s) / 12Fe2O3(s) DH900 C ¼ À986.5 kJ/mol (8) Since reactions (7) and (8) are conducted in a cyclic manner, the sum of reactions is the combustion of the fuel with oxygen as per reaction (9): CH4(g) ỵ 2O2(g) / CO2(g) ỵ 2H2O(g) DH900 C ¼ À802.5 kJ/mol(9) From reaction (9), it can be seen that the chemical-looping combustion produces the same amount of heat as conventional combustion The difference is that the reaction is divided into two steps, thus obtaining two separate process streams Chemical-looping combustion has several attractive features Most importantly, the gas from the fuel reactor consists essentially of CO2 and H2O Hence cooling in a condenser is all that is needed to obtain almost pure CO2, which makes chemical-looping combustion an ideal technology for heat and power production with carbon sequestration The general principles of chemical-looping combustion were laid out as early as in the 1950’s by Lewis and Gilliland [20], who suggested to carry out the reactions in fluidized-bed reactors with particles of oxygen carrier particles as bed material This remains the favoured design and several prototype reactors have been constructed using this principle For a straightforward explanation of the basic principles of chemical-looping combustion, see Lyngfelt et al [21] In the Depleted air O2, N2 Products CO2, H2O Fe2O3(s) AR FR Fe3O4(s) Air O2, N2 Fuel CnHm, CO, H2 Fig e Schematic description of chemical-looping combustion using iron oxide as oxygen carrier 4846 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 past decade, research about chemical-looping combustion has taken pace and providing a complete overview is not within the scoop of this publication Progress within the area has been reviewed by Lyngfelt [19], Fang et al [22], Hossain and de Lasa [23] and Adanez et al [24] the volumetric increase involved in converting hydrocarbons to H2, a pressurized system would clearly be favourable Otherwise the size of the plant would be very large and the mechanical work needed for compression of the products would reduce the overall process efficiency considerably 2.3 2.4 H2 production via chemical-looping combustion The chemical-looping concept has also been proposed as a method to generate H2 Using the same general setup as in Fig and by introducing the air to the air reactor in understoichiometric proportions, partial oxidation of the fuel can be achieved, i.e synthesis gas (CO and H2) is produced in the fuel reactor rather than CO2 and H2O The synthesis gas can then be used for generation of H2, or other products This concept, typically referred to as chemical-looping autothermal reforming, has been demonstrated by Ryde´n et al [25e27] in a small circulating fluidized bed-reactor, by Kolbitsch et al [28] in a dual circulating fluidized bed, and by Ortiz et al [29,30] in a pressurized semi-continuous fluidized bedreactor Another approach would be to use steam reforming and pressure swing adsorption for generation of H2, with chemical-looping combustion used for production of heat by combustion of the resulting waste gases, as has been proposed by Ryde´n and Lyngfelt [31] and Ortiz et al [32] A third option would be to add a third reactor to the chemical-looping combustion process, in this paper referred to as a steam reactor (SR) In this reactor, reduced iron based oxygen carrier would be oxidized with steam producing pure H2 in a similar way as in the conventional steameiron process In fact, the resulting three-reactor process could be described as a hybrid of chemical-looping combustion and the steameiron process, as is shown in Fig In Fig 2, reactions (1) and (2) take place in the fuel reactor and reaction (5) is performed in the steam reactor, while reaction (8) is carried out in the air reactor The iron oxide circulates continuously through the system providing stable flows of solids and gases Similar processes to the one shown in Fig have been suggested recently by several authors, see for example Fan et al [10], Chiesa et al [11], Yang et al [33] and Chen et al [13] The detailed process study by Chiesa et al [11] shows that a process as the one in Fig could achieve similar efficiency as conventional processes for H2 production, without the need for downstream separation systems, wateregas shift reactors, cryogenic distillation of air or other expensive and energy consuming equipment Furthermore, high purity H2 would be produced and CO2 for sequestration would be obtained simply by cooling of the stream from the fuel reactor, condensing steam to water An analysis of oxygen carrier selection criteria for such a three-reactor chemical-looping process can be found in the work by Kang et al [34] Naturally, the concept described in Fig involves some technical difficulties as well The fuel reactor needs to be arranged in counter-current fashion with Fe2O3 added from the top and fuel added from the bottom; else the fuel conversion will be limited by thermodynamic constraints, as will be explained below Counter-current flow could be achieved for example by using moving bed reactor, see Li et al [35], or possibly by using a staged fluidized bed Further, due to Aim of this study The main objective of this study is to examine H2 generation via the steameiron reaction in a continuously operating reactor consisting of two interconnected fluidized beds The study aims to cover the current lack of experiments examining the steameiron reaction during such conditions A secondary objective is to examine whether continuous operation with iron oxide reduced to FeO is feasible, since the general experience from chemical-looping combustion experiments with iron oxide as oxygen carrier is that there is a strong correlation between reduction to FeO and defluidization of the particle bed, see Cho et al [36] and Ryde´n et al [37,38] Experimental 3.1 Manufacturing of oxygen carrier particles Synthetic oxygen carrier particles consisting of 60 wt.% Fe2O3 supported on 40 wt.% MgAl2O4 manufactured by freezegranulation were used MgAl2O4 has some attractive features which suggest that it should be a good support material for oxygen carriers for chemical looping applications, such as high melting point and high thermal and chemical stability In a previous study by Mattisson et al [39] concerning the Fig e Schematic description of hydrogen generation via the steameiron reaction configured in accordance to the principles of chemical-looping combustion 4847 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 development of iron oxide based oxygen carrier particles for chemical-looping combustion in fluidized bed-reactors, MgAl2O4 was identified as a suitable inert support material for this application, exhibiting higher reactivity than iron oxide supported on Al2O3, ZrO2 and TiO2 The manufacturing procedure was as follows: A waterbased slurry of Fe2O3 and support powder (MgAl2O4) with weight ratio of 60/40 along with small amount of dispersant (acrylic acid) was prepared The mixture was then ball milled for 24 h Subsequently polyvinyl alcohol was added as binder prior to granulation The slurry was pumped through a spray nozzle and into liquid nitrogen to form spherical particles upon instantaneous freezing The particles were initially calcined at 1100  C for h at a ramp rate of  C/min However, this did not result in particles of the desired strength and density, thus the particles were calcined for another h at 1150  C This resulted in particles with a bulk density of approximately 1000 kg/m3, which deemed suitable for the experiments The particles were then sieved through stainless steel screens to yield particles in the size range of 90e250 mm 3.2 Characterization of oxygen carrier particles The oxygen carrier particles were analysed before and after the experiments using powder X-ray diffraction (Siemens D5000 Diffractometer) with CuKa radiations The morphological investigation was carried out with an environmental scanning electron microscope (ESEM) fitted with a field emission gun (FEI, Quanta 200) The BET surface area of the particles was evaluated with TriStar 3000 (Micromeritics) The particle size distribution (PSD) before and after the experiments was determined using a light microscope (Nikon SMZ800) and using ImageJ [40] software to measure the area of an ellipse fitted to a large number of particles The crushing strength of the particles was measured as the strength needed to fracture the particles ranging within 180e250 mm for an average of 30 tests per sample using a digital force gauge (Shimpo FGN-5) The crushing strength was found to be approximately 0.6 N In some cases particles with a crushing strength below N are considered too soft [41] However, this did not cause any problem such as defluidization in the reactor, as determined by the pressure drop in the bed Nonetheless, for use in a full-scale plant, the crushing strength may need to be increased further which could be done by either increasing the sintering time or temperature Table summarizes the physical properties of the oxygen carrier used in this investigation 3.3 Two-compartment fluidized-bed reactor The experiments were carried out in a small-scale laboratory reactor constructed of 253 MA steel, which is a temperature, creep and deformation resistant stainless steel with the approximate composition 67.9% Fe, 21% Cr, 11% Ni and 0.1% C The reactor is similar to but not identical with a system previously used for various chemical-looping experiments [25,26,42e44] A schematic description of the reactor is shown in Fig The reactor is designed for chemical-looping combustion experiments using gaseous and liquid fuels, but steameiron Table e Properties of fresh oxygen carrier particles Oxygen carrier Theoretical Fe2O3 content [wt.%] Support phase Size interval of particles [mm] Bulk density [kg/m3] BET specific surface area [m2/g] Crystalline phase Crushing strength [N] Fe2O3/MgAl2O4 60 MgAl2O4 (S30CR, Baikowski) 90e250 1000 5.25 Fe2O3, MgAl2O4 0.6 reaction experiments could be conducted simply by replacing air normally fed to the air reactor with steam The reactor is 300 mm high The fuel reactor is 25 mm  25 mm The base of the air reactor is 25 mm  42 mm, while the upper narrow part is 25 mm  25 mm Fuel and air enter the system through separate wind boxes, located in the bottom of the reactor Porous quartz plates act as gas distributors In the air reactor the gas velocity is sufficiently high for oxygen carrier particles to be thrown upwards Above the reactor there is a particle separation box in which the crosssection area is increased and gas velocity reduced so that particles fall back into the reactor A fraction of these particles falls into the downcomer, entering a J-type loop-seal From the loop-seal, particles overflow into the fuel reactor via the return orifice The fuel reactor is a bubbling bed In the bottom particles return to the air reactor through a U-type slot and thus a continuous circulation of oxygen carrier particles is obtained The downcomer and the slot are fluidized with inert gas such as argon, which is added via thin pipes perforated by small holes, rather than through porous plates In order to make it possible to reach and sustain a suitable temperature, the reactor is placed inside an electrically heated furnace The furnace also makes it possible to conduct continuous steameiron reaction experiments even if the net reaction for the fuel reactor and the steam reactor is endothermic, thus omitting the air reactor from Fig The temperature in each reactor section is measured with thermocouples located inside the particle beds, a few centimetres above each bottom plate The reactor is operated roughly at atmospheric pressure However, a water-seal is located downstream the fuel reactor which makes it possible to apply an overpressure of z250 Pa to the fuel reactor, in order to inhibit leakage of air into the fuel reactor Along the reactor sections there are thirteen separate pressure measuring taps By measuring differential pressures between these spots, it is possible to estimate where particles are located in the system, and to detect abnormalities in the fluidization For gas analysis, roughly 0.50 Ln/min gas was extracted downstream of the air reactor and fuel reactor respectively Each of these flows passed through separate particle filters, coolers and water traps Hence all measurements were made on dry gas CO2, CO and CH4 were measured using infrared analysers while O2 was measured with paramagnetic sensors The gas from the air/steam reactor was also examined with a gas chromatograph which measured H2 and N2, in addition to the gas components mentioned above The gas chromatograph provided measuring points every 3e4 Excess gas that was not needed for analysis passed through a textile filter 4848 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 CO(g) ỵ H2O(g) CO2(g) ỵ H2(g) DH900 C ẳ 33.1 kJ/mol (10) Synthesis gas and natural gas was used mainly since they represent possible fuels for a real-world steameiron process 3.5 Assumptions and data evaluation The conversion of fuel in the fuel reactor is expressed as carbon dioxide yield gCO2: gCO2 ¼ xCO2 xCO2 ỵ xCH4 ỵ xCO (11) Here xi denotes the composition of the respective gas, obtained from measured concentration in the gas analyser When CO is used as fuel, gCO2 provides and accurate description of the degree of fuel conversion and the combustion efficiency However, when synthesis gas or natural gas is used as fuel, gCO2 only provides an adequate estimation since the conversion of H2 to H2O in the fuel reactor may differ slightly from the conversion of CO to CO2, due to differing thermodynamic properties of H2 and CO However, at the temperature levels investigated in this work (800e950  C) the difference should be rather small, as is shown in Fig Natural gas contains small amounts of higher hydrocarbons such as ethane and propane but those are expected to be much more reactive with the oxygen carrier than methane and have not been included in gCO2 The mass-based degree of reduction, u, can be used to describe the reduction of the oxygen carrier particles and is defined in expression (12): u¼ m mox (12) u describes the amount of oxygen that has been removed from the oxygen carrier compared to the oxidized state and can be calculated with a species balance as follows: Fig e Schematic description of the two-compartment fluidized-bed reactor in order to catch elutriated fines and particles, prior to release in a chimney For the experiments presented in this paper, 300 g of the particles were added to the reactor This corresponds to a bed height in the air and fuel reactor of roughly 10 cm, taken into consideration that a considerable share of the particles was located in the downcomer during operation 3.4 Fuel gases Three different fuel gases were used for reduction of the oxygen carrier namely pure carbon monoxide (CO), synthesis gas (SG) consisting of 50% CO and 50% H2, and natural gas (NG) with a composition equivalent to C1.14H4.25O0.01N0.005 Using CO as fuel has the advantage that solving the species balance for the reactor system becomes trivial Hence most of the figures included the experiments with carbon monoxide as fuel With CO as fuel the sum of reactions will be the wateregas shift reaction: Fig e Equilibrium composition of the gas phase for mixtures of CO/CO2 and H2/H2O in presence of Fe3O4eFeO, as calculated using thermodynamic data from FactSage 6.1 4849 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 Zt1 ui ¼ ui1 t0 n_ out MO 4xCO2 ỵ 3xCO À xH2 dt mox 4.2 ui is the instantaneous conversion at time i, uiÀ1 is the conversion in the preceding instant, t0 and t1 are the initial and final time of measurement MO is the molar mass of oxygen, n_ in and n_ out are the dry molar flow rates of the gas at inlet and outlet of the reactor, respectively Results 4.1 Chemical-looping combustion experiments Steameiron reaction experiments (13) The aim of the chemical-looping combustion experiments was to examine the reactivity of the oxygen carrier with potential fuel gases, i.e carbon monoxide, synthesis gas and natural gas The furnace was heated to a temperature slightly above the desired fuel-reactor temperature, which in this case was 900e950  C During this period both reactor sections were fluidized with air, while the particle seals were fluidized with argon When the desired temperature was reached, the air stream that was introduced to the fuel reactor was replaced initially with N2, and after a few minutes, by fuel The reactor was then operated for h with more or less stable process parameters Following that, the oxygen carrier particles were reoxidized according to the following procedure First, the fuel gas was replaced with an equally large flow of N2 Other parameters were not changed Thus it is believed that the solids circulation should not have been affected since the gas flows were essentially the same Reduced particles present in the fuel reactor were eventually transferred to the air reactor, where they were oxidized with oxygen from air Therefore the time for the O2 concentration in the air reactor to reach a stable value corresponds to the particles residence time in the fuel reactor, a parameter that could be used to estimate the solid circulation between the reactor sections Once stable O2 concentration in the air reactor was obtained, the inert flows to the fuel reactor and particle seals were switched to air to make sure that all active materials were properly oxidized back to Fe2O3 A summary of conducted chemical-looping combustion experiments can be found in Table The operation of the chemical-looping combustion experiments was satisfactory Almost complete conversion of carbon monoxide and synthesis gas to CO2 and H2O was obtained As can be seen in Table 2, gCO2 was higher than 99.9% for these fuels, leaving only traces of unconverted CO in the products For natural gas, gCO2 was limited to 75e90%, leaving excess CH4, CO and H2 in the product gas The results with complete conversion of synthesis gas and much lower conversion for other hydrocarbons is in accordance with the expected behaviour of freeze-granulated iron oxide particles [45] Following the chemical-looping combustion experiments, it was decided to use CO and synthesis gas as fuel for the steameiron experiments This was due to the high reactivity and simple chemistry of these fuels, which facilitates the interpretation of the experimental data, compared to natural gas The steameiron reaction experiments were conducted with the same experimental setup as the chemical-looping combustion experiments The only difference was that instead of air, steam was added to the air reactor This way oxidation to Fe2O3 was not possible due to thermodynamic constraints and FeO would eventually be generated in the fuel reactor The solids circulation between the reactor compartments were quite high, as will be explained in section 4.3 below Hence any FeO produced in the fuel reactor would quickly be transferred to the air reactor, where it would be oxidized back to Fe3O4 by steam according to reaction (5), producing H2 Formation of metallic Fe in the fuel reactor seems unlikely, as will be further discussed in Section 4.3 below The steam flow added to the steam reactor was 4.0e5.0 Ln/ and the conversion of steam to H2 in presence of FeO could be expected to be 26e36% depending on temperature, according to Fig Since the fuel flow was in the order of 1.0 Ln/min or lower, this means that steam was always added in excess Once the flow of reducing gas was removed from the fuel reactor generation of H2 in the steam reactor ceased, which indicates that FeO did not accumulate within the reactor system during operation Hence the H2 production during stable operation should be determined simply by the amount of FeO produced in the fuel reactor by reduction with fuel A summary of conducted steameiron experiments can be found in Table The steameiron reaction experiments were initiated in the same way as the chemical-looping combustion experiments, but steam was added to the air reactor instead of air Hence the oxygen carrier would gradually become reduced to Fe3O4 according to reaction (1) At the point where Fe2O3 was no longer present in the system, the fuel added to the fuel reactor would become partially oxidized according to reaction (2) The degree of oxidation of the fuel could be expected to be confined by thermodynamics Measured data for an experiment with CO as fuel is shown in Fig In Fig 5, it can be seen that there was almost complete conversion of CO to CO2 as long as there was Fe2O3 present in the oxygen carrier This is in accordance with theory and with the chemical-looping combustion experiments Once Fe2O3 was depleted, the process shifted to a new equilibrium corresponding to oxidation of CO with Fe3O4 From this point onwards, about 75% of the added CO appears to have been converted to CO2 Table e Summary of conducted chemical-looping combustion experiments Ffuel,fr is the flow of fuel, Fair,ar is the flow of air, Tfr is the fuel reactor temperature and gCO2 is CO2 yield Fair,ar Experiment Operation Ffuel,fr (Ln/min) (Ln/min) (min) CLC CO CLC SG CLC NG 60 60 60 1.00e1.15 1.00 0.42 5.0 5.0 5.0 Tfr ( C) gCO2 (%) 900 >99.94 900 >99.95 900e950 75e90 4850 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 Table e Summary of conducted steameiron reaction experiments gCO2 ;fr and FH2 ;produced are representative average values over the period Experiment SIR SIR SIR SIR SIR COI COII COIII SGI SGII Operation (min) Tsr ( C) Fsteam,sr (Ln/min) FN2 ;sr (Ln/min) Ffuel,fr (Ln/min) gCO2 ;fr (%) FH2 ;produced (Ln/min) 60 60 60 120 240 900 850 800 900 900 5.0 5.0 5.0 4.0e5.0 5.0 0.60 0.60 0.60 1.00 0.60 0.87 0.89 0.92 0.78e0.87 0.50e1.25 75.0 71.5 67.0 77.5e75.5 81.0e60.5 z0.53 z0.52 z0.49 z0.46e0.53 z0.33e0.58 Any FeO produced in the fuel reactor would eventually be transferred via the solid circulation to the air reactor, where it was oxidized by steam according to reaction (5), producing H2 in the process Measured data for a one-hour experiment is shown in Fig In Fig 6, it can be seen that H2 was not produced during the initial few minutes, where Fe2O3 was still present Once the particles were deprived of Fe2O3, H2 was produced continuously, as could be expected The presence of CO2 and CO is due to gas leakage through the slot in the bottom of the reactor which is discussed in Section 4.3 below The balance is the Ar gas which was used for fluidization of the particle seals The volumetric H2 production could be estimated by comparison with the measured N2 concentration, which origins is a trace gas flow of 0.60 Ln/min The amount of H2 Fig e Dry-gas concentrations from the fuel reactor for the initiation period of steameiron experiments with 0.87 Ln/min CO as fuel at 900  C (SIR COIII) The dotted line at t z 10 describes the theoretical point for complete reduction of Fe2O3 to Fe3O4 Dilution with Ar is from the fluidization gas added to the particle seals produced was found to vary with the amount of fuel added to the fuel reactor, see Fig Fig shows the H2 production as function of fuel flow It can be seen that at the lowest fuel flow (i.e Fsg,fr ¼ 0.50 Ln/ min), the volumetric H2 generation corresponded well to the theoretical maximum, which is defined as when the fuel reacts with the oxygen carrier in the fuel reactor so that thermodynamic equilibrium is achieved, and all the resulting FeO is oxidized in the air reactor, producing H2 and Fe3O4 It is evident from Figs and that increasing the fuel flow did increase H2 generation, but that the increase was not as significant as expected Furthermore, the fuel flows for the experiments with CO (SIR COI-III) was specifically chosen so that 0.60 Ln/min H2 would be produced, if there was no gas leakage between the reactors and thermodynamic equilibrium was reached In reality, the amounts of H2 generated was slightly lower, i.e 0.49e0.53 Ln/min, as can be seen in Table Since no accumulation of FeO in the system was noticed during operation, this suggests that the conversion of fuel in the fuel reactor was to slow to achieve equilibrium when fuel flows higher than 0.50 Ln/min were used This is also supported by examining the CO2 yield, gCO2 , as a function of fuel flow as shown in Fig In Fig 9, it can be seen that the gCO2 is highly dependent on the fuel flow Higher flow results in lower conversion to CO2 and vice versa This clearly indicates that the reaction in the fuel reactor does not reach thermodynamic equilibrium Hence less FeO than expected is formed and less H2 than the theoretical maximum is produced, as shown in Figs and It can be observed in Fig that the expected conversion of CO to CO2 is approximately 70% at 900  C However, Figs and shows a conversion of about 75%, and in Fig 9, over 80% conversion is achieved for the lowest fuel flow Thus the fuel conversion is higher than what theoretically should be possible The likely explanation for this phenomenon is that there was a small leakage of steam from the steam reactor into the fuel reactor, for example via the particle seals When synthesis gas is used as fuel, steam is also formed in the reaction with the oxygen carrier The presence of steam in the gas from the fuel reactor could be expected to result in higher measured gCO2 because the gas from the fuel reactor has a residence time of several seconds in the particle separation box above the reactor, in which the temperature is about as high as in the reactor itself Hence steam may react with CO forming additional CO2 via the wateregas shift, see reaction (10) This means that the values of yCO2 ;fr presented in the figures and in Table are likely somewhat higher than what they are in reality In theory, a leakage as small as 3% of the steam added to the steam reactor could explain such high values of gCO2 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 Fig e Dry-gas concentration from the steam reactor for steameiron experiments with 0.87 Ln/min CO as fuel at 900  C (SIR COIII) 4.3 Reactor performance As have been explained above, inert argon was used as fluidization gas in the downcomer and in the slot The flow rates 0.70 Ln/min to the downcomer and 0.20 Ln/min to the slot The argon was fairly evenly distributed between the air and the fuel reactor, diluting the product gases somewhat This behaviour was expected No problems with defluidization or unwanted stops in the particle circulation were experienced, despite reduction of the oxygen carrier to FeO in the fuel reactor, which otherwise have been reported to propagate defluidization [36e38] Minor irregularities such as uneven particle circulation were observed when the fuel flow was reduced to 0.50 Ln/min, Fig e Dry-gas concentration from the steam reactor as function of fuel added to the fuel reactor for steameiron experiments with 0.50e1.25 Ln/min synthesis gas as fuel at 900  C (SIR SGII) 4851 Fig e Comparison of the theoretical (—) and actual (A) H2 production in the steam reactor as function of fuel added to the fuel reactor for steameiron experiments with 0.50e1.25 Ln/min synthesis gas as fuel at 900  C (SIR SGII) which could be expected due to the resulting low gas velocity These irregularities did not lead to any practical problems The solids circulation was estimated to be approximately 3e4 g/s using the following procedure Firstly, the amount of particles present in the fuel reactor was estimated from the particle density and reactor geometry, while the residence time of particles in the fuel reactor was estimated as the time to reach stable gas concentrations during reoxidation, see Fig e CO2 yield gCO2 ;fr as function of fuel added to the fuel reactor for steameiron experiments with 0.50e1.25 Ln/min synthesis gas as fuel at 900  C (SIR SGII) 4852 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 Fig 10 e Particle size distribution for fresh and used oxygen carrier roughly only 4e6% of the 4.0 wt.% oxygen available in reaction (2) was utilized, which corresponds to operation at u between 0.980 and 0.978 Although the reduction of iron oxides not necessarily proceed sequentially and formation of small amounts of metallic Fe in the fuel reactor can not be ruled out, the state of the iron oxide in the fuel reactor should be approximately 94e96 wt.% Fe3O4 and 4-6 wt.% FeO The gas leakage from the fuel reactor to the air reactor could be estimated by measuring the CO2 and CO concentrations after the air reactor, as is shown in Fig The leakage corresponded to 8e20% of the gas added to the fuel reactor, which is in the same order of magnitude as for earlier chemical-looping experiments in a similar two-compartment reactor [25,26,42e44] As stated above, gas leakage from the steam reactor to the fuel reactors could have affected measured CO2 yields to some extent, since CO and CO2 could possibly react with steam above the particle beds via the wateregas shift reaction The effect should be relatively small though, and should have no impact on the general conclusions of the study As can be seen in Table 3, continuous operation was possible and the volumetric H2 production typically reached 80e85% of the theoretical value For the lowest fuel flow examined, the H2 production was very close to the theoretical maximum, as is shown in Fig above 4.4 section 4.1 above The solids circulation could then be calculated simply as the mass of the particles present in the fuel reactor divided by the residence time The result is a rather rough estimation, but should be sufficient to conclude that the solids circulation was more than sufficient for the experiments conducted The chemical-looping combustion experiments required oxygen carrier particles corresponding to approximately 0.50 Ln/min O2 to be transferred to the fuel reactor via the solids circulation The amount of oxygen available in iron oxide for reaction (1) is 3.3 wt.%, which for an oxygen carrier with 60 wt.% active material is reduced to 2.0 wt.% This suggests that for the chemical-looping combustion experiments, only about 15e20% of the oxygen available in reaction (1) was utilized, i.e that the oxygen carrier was operated at u between 1.000 and 0.996 For the steameiron reaction, approximately 0.30 Ln/min O2 needed to be transferred Thus Effect on oxygen carrier particles Following the experiments, the reactor was disassembled and the used oxygen carrier recovered Out of the 300 g oxygen carrier added to the reactor, 285 g was found inside the reactor system itself g of the material was found in the fuel reactor windbox, having somehow passed through the fitting between the porous plate and the reactor These particles had been extensively reduced during operation and had formed soft agglomerations g of the material had been blown out of the system and was collected in the filters downstream The blown out material was very fine, with g < 45 mm and g < 90 mm g of the material was missing and could have been spilled during the disassembling of the reactor system, or possibly stuck in one of the pressure measuring taps In general, the particles behaved satisfactory The analysis shows that the particle size had decreased somewhat compared to the fresh material Fig 10 shows the particle size Fig 11 e ESEM images of (a) fresh and (b) used Fe2O3/MgAl2O4 oxygen carrier i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 distribution (PSD) for fresh and used oxygen carrier The size range for the fresh particles, as measured using light microscope, is slightly higher than the original sieving due to the fact that the particles were not perfectly spherical However, the shift towards smaller sizes could clearly be observed for the used particles Moreover, the bulk density of the used particles increased to 1600 kg/m3 compared to 1000 kg/m3 for the fresh materials The BET specific surface area of the used particles also decreased to 2.55 m2/g, compared to 5.25 m2/g for the fresh oxygen carrier These facts indicate that the particles became densified during operation, possibly due to thermal sintering at high temperature The densification of the particles during the initial oxidation and reduction cycles was expected and had been accounted for in advance by adding a comparably large volume of particles to the reactor The formation of fines was very low, considering that most oxygen carrier particles tend to form dust during the initial hours of operation The X-ray diffraction spectra from used oxygen carrier did not differ from fresh material This indicates that the oxygen carrier was fully oxidized without formation of any unwanted ternary compound Fig 11 shows fresh and post-experiments scanning electron microscope images of the particles It can be observed that the surface morphology of the oxygen carrier was not affected by the redox operation Conclusions Freeze-granulated particles of 60 wt.% Fe2O3 and 40 wt.% inert MgAl2O4 proved to perform well as oxygen carrier both for chemical-looping combustion and for the steameiron reaction As long as there was Fe2O3 present, carbon monoxide or synthesis gas added to the fuel reactor was more or less completely oxidized to CO2 and H2O Once all Fe2O3 had been reduced to Fe3O4, carbon monoxide or synthesis gas added to the fuel reactor would become partially oxidized to a composition corresponding approximately to thermodynamic equilibrium at the relevant temperature, producing FeO in the process The fuel conversion was lower at high fuel flows though, but this could be expected The reduced oxygen carrier was transferred to the steam reactor where it was oxidized with steam forming Fe3O4 and H2 The amount of H2 produced in the steam reactor was found to correspond reasonably well with the amount of fuel oxidized in the fuel reactor, which suggests that all FeO was indeed oxidized in the air reactor Continuous operation of the process was achieved No problems with defluidization or unwanted stops in the particle circulation were experienced, despite reduction of the oxygen carrier to FeO in the fuel reactor This suggests that the MgAl2O4 supported oxygen carrier particles used have favourable properties, with respect to fluidization behaviour The oxygen carrier experienced densification during operation, but otherwise behaved satisfactory It is concluded that continuous operation of the steameiron reaction in a fluidized-bed reactors is feasible Since it is established that chemical-looping combustion using iron oxide as oxygen carrier is viable, it should be 4853 possible to realize a continuous three-reactor system as proposed in Fig Due to the potentially favourable characteristics of such process, providing pure H2 and CO2 without gas separation or wateregas shift reactor, this opportunity should be further examined references [1] Ogden MJ Prospects for building a hydrogen energy infrastructure Annual Review of Energy and the Environment 1999;24:227e79 [2] Mueller-Langer F, Tzimas E, Kaltschmitt M, Peteves S Techno-economic assessment of hydrogen production processes for the hydrogen economy for the short and medium term International Journal of Hydrogen Energy 2007;32:3797e810 [3] Cormos C-C Evaluation of power generation schemes based on hydrogen-fuelled combined cycle with carbon capture and storage (CCS) International Journal of Hydrogen Energy 2011;36:3726e38 [4] Messerschmitt A Process for producing hydrogen; 1910 U.S patent 971,206 [5] Lane H Process for the production of hydrogen; 1913 U.S patent 1,078,686 [6] Gasior SJ, Forney AJ, Field JH, Bienstock D, Benson HE Production of Synthesis gas and hydrogen by the steameiron process: pilot plant study of fluidized and free-falling beds United States Bureau of Mines; 1961 Report of investigations 5911 [7] Reed H Hydrogen process; 1953 U.S patent 2,635,947 [8] Biljetina R, Tarman PB Development status of the steameiron process for hydrogen production Proceedings of the Second Miami International Conference on Alternative Energy Sources 1981;8:3335e47 [9] Fukase S, Suzuka T Residual oil cracking combined with hydrogen production by steameiron reaction The Canadian Journal of Chemical Engineering 1994;72:272e8 [10] Fan L, Li F, Ramkumar S Utilization of chemical looping strategy in coal gasification processes Particuology 2008;6: 131e42 [11] Chiesa P, Lozza G, Malandrino A, Romano M, Piccolo V Three-reactors chemical looping process for hydrogen production International Journal of Hydrogen Energy 2008; 33:2233e45 [12] Hacker V, Fankhauser R, Faleschini G, Fuchs H, Friedrich K, Muhr M, et al Hydrogen production by steameiron process Journal of Power Sources 2000;86:531e5 [13] Chen S, Shi Q, Xue Z, Sun X, Xiang W Experimental investigation of chemical-looping hydrogen generation using Al2O3 or TiO2-supported iron oxides in a batch fluidized bed International Journal of Hydrogen Energy 2011;36:8915e26 [14] Bohn CD, Cleeton JP, Muăller CR, Chuang SY, Scott SA, Dennis JS Stabilizing iron oxide used in cycles of reduction and oxidation for hydrogen production Energy & Fuels 2010; 24:4025e33 [15] Yamaguchi D, Tang L, Wong L, Burke N, Trimm D, Nguyen K, et al Hydrogen production through methaneesteam cyclic redox processes with iron-based metal oxides International Journal of Hydrogen Energy 2011;36:6646e56 [16] Lorente E, Pen˜a JA, Herguido J Cycle behaviour of iron ores in the steameiron process International Journal of Hydrogen Energy 2011;36:7043e50 [17] Bleeker MF, Kersten SRA, Veringa HJ Pure hydrogen from pyrolysis oil using the steameiron process Catalysis Today 2007;127:278e90 4854 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y ( 2 ) e4 [18] Yang J-B, Cai N-S, Li Z-S Hydrogen production from the steameiron process with direct reduction of iron oxide by chemical looping combustion of coal char Energy & Fuels 2008;22:2570e9 [19] Lyngfelt A Oxygen carriers for chemical looping combustion - 000 h of operational experience Oil & Gas Science and Technology - Revue d’IFP Energies nouvelles 2011;66:161e72 [20] Lewis KW, Gilliland ER, Sweeney MP Gasification of carbon metal oxides in a fluidized powder bed Chemical Engineering Progress 1951;47:251e6 [21] Lyngfelt A, Leckner B, Mattisson T A fluidized-bed combustion process with inherent CO2 separation; application of chemical-looping combustion Chemical Engineering Science 2001;56:3101e13 [22] Fang H, Haibin L, Zengli Z Advancements in development of chemical-looping combustion: a review International Journal of Chemical Engineering 2009;2009 [23] Hossain MM, de Lasa HI Chemical-looping combustion (CLC) for inherent CO2 separations e a review Chemical Engineering Science 2008;63:4433e51 [24] Adanez J, Abad A, Garcia-Labiano F, Gayan P, de Diego LF Progress in chemical-looping combustion and reforming technologies Progress in Energy and Combustion Science 2012;38:215e82 [25] Ryden M, Johansson M, Lyngfelt A, Mattisson T NiO supported on MgeZrO2 as oxygen carrier for chemicallooping combustion and chemical-looping reforming Energy & Environmental Science 2009;2:970e81 [26] Ryden M, Lyngfelt A, Mattisson T Chemical-looping combustion and chemical-looping reforming in a circulating fluidized-bed reactor using Ni-based oxygen carriers Energy & Fuels 2008;22:2585e97 [27] Ryde´n M, Lyngfelt A, Mattisson T Synthesis gas generation by chemical-looping reforming in a continuously operating laboratory reactor Fuel 2006;85:1631e41 [28] Kolbitsch P, Proăll T, Bolhar-Nordenkampf J, Hofbauer H Operating experience with chemical looping combustion in a 120 kW dual circulating fluidized bed (DCFB) unit Energy Procedia 2009;1:1465e72 [29] Ortiz M, de Diego LF, Abad A, Garcı´a-Labiano F, Gaya´n P, Ada´nez J Hydrogen production by auto-thermal chemicallooping reforming in a pressurized fluidized bed reactor using Ni-based oxygen carriers International Journal of Hydrogen Energy 2010;35:151e60 [30] Ortiz M, Abad A, de Diego LF, Garcı´a-Labiano F, Gaya´n P, Ada´nez J Optimization of hydrogen production by chemicallooping auto-thermal Reforming working with Ni-based oxygen-carriers International Journal of Hydrogen Energy 2011;36:9663e72 [31] Ryde´n M, Lyngfelt A Using steam reforming to produce hydrogen with carbon dioxide capture by chemical-looping combustion International Journal of Hydrogen Energy 2006; 31:1271e83 [32] Ortiz M, Gaya´n P, de Diego LF, Garcı´a-Labiano F, Abad A, Pans MA, et al Hydrogen production with CO2 capture by coupling steam reforming of methane and chemical-looping combustion: use of an iron-based waste product as oxygen carrier burning a PSA tail gas Journal of Power Sources 2011; 196:4370e81 [33] Xhang CH, Wan HJ, Yang Y, Xiang HW, Li YW Catalysis Communications 2006;7:733 [34] Kang K-S, Kim C-H, Bae K-K, Cho W-C, Kim S-H, Park C-S Oxygen-carrier selection and thermal analysis of the chemical-looping process for hydrogen production International Journal of Hydrogen Energy 2010;35:12246e54 [35] Li F, Zeng L, Velazquez-Vargas LG, Yoscovits Z, Fan L-S Syngas chemical looping gasification process: bench-scale studies and reactor simulations AIChE Journal 2010;56: 2186e99 [36] Cho P, Mattisson T, Lyngfelt A Defluidization conditions for a fluidized bed of iron oxide-, nickel oxide-, and manganese oxide-containing oxygen carriers for chemical-looping combustion Industrial & Engineering Chemistry Research 2005;45:968e77 [37] Ryde´n M, Cleverstam E, Johansson M, Lyngfelt A, Mattisson T Fe2O3 on Ce-, Ca-, or Mg-stabilized ZrO2 as oxygen carrier for chemical-looping combustion using NiO as additive AIChE Journal 2010;56:2211e20 [38] Ryde´n M, Cleverstam E, Lyngfelt A, Mattisson T Waste products from the steel industry with NiO as additive as oxygen carrier for chemical-looping combustion International Journal of Greenhouse Gas Control 2009;3: 693e703 [39] Mattisson T, Johansson M, Lyngfelt A Multicycle reduction and oxidation of different types of iron oxide particlesapplication to chemical-looping combustion Energy & Fuels 2004;18:628e37 [40] Rasband WS ImageJ Bethesda, Maryland, USA: National Institutes of Health; 1997e2004 [41] Johansson M, Mattisson T, Lyngfelt A Investigation of Fe2O3 with MgAl2O4 for chemical-looping combustion Industrial & Engineering Chemistry Research 2004;43:6978e87 [42] Ryde´n M, Lyngfelt A, Mattisson T CaMn0.875Ti0.125O3 as oxygen carrier for chemical-looping combustion with oxygen uncoupling (CLOU) e experiments in a continuously operating fluidized-bed reactor system International Journal of Greenhouse Gas Control 2011;5:356e66 [43] Ryde´n M, Lyngfelt A, Mattisson T Combined manganese/ iron oxides as oxygen carrier for chemical looping combustion with oxygen uncoupling (CLOU) in a circulating fluidized bed reactor system Energy Procedia 2011;4:341e8 [44] Ryde´n M, Johansson M, Cleverstam E, Lyngfelt A, Mattisson T Ilmenite with addition of NiO as oxygen carrier for chemical-looping combustion Fuel 2010;89:3523e33 [45] Abad A, Mattisson T, Lyngfelt A, Johansson M The use of iron oxide as oxygen carrier in a chemical-looping reactor Fuel 2007;86:1021e35