Combined chemical looping for energy storage and conversion

Combined chemical looping for energy storage and conversion

Journal of Power Sources 286 (2015) 362e370 Contents lists available at ScienceDirect Journal of Power Sources journal homepage: www.elsevier.com/lo...

2MB Sizes 46 Downloads 193 Views

Journal of Power Sources 286 (2015) 362e370

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Combined chemical looping for energy storage and conversion Vladimir V. Galvita*, Hilde Poelman, Guy B. Marin Laboratory for Chemical Technology, Ghent University, Technologiepark 914, B-9052 Ghent, Belgium

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

 Combined chemical looping as novel concept of energy storage: chemical looping combined with calcium looping.  CH4 induces metal reduction and surface carbon formation.  CaOeCaCO3 is used for storagerelease of CO2.  CO2 acts as mediation gas to oxidize metal and carbon deposits leading to CO production.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 12 February 2015 Received in revised form 27 March 2015 Accepted 30 March 2015 Available online 31 March 2015

Combined chemical looping was demonstrated as novel concept of energy storage in a laboratory scale test. The proposed technology is able to store and release energy from redox chemical looping reactions combined with calcium looping. This process uses Fe3O4 and CaO, two low cost and environmentally friendly materials, while CH4 þ CO2 serve as feed. During the reduction of Fe3O4 by CH4, both formation of carbon and metallic iron occur. CO2 acts as mediation gas to facilitate the metal/metal oxide redox reaction and carbon gasification into CO. CaO, on the other hand, is used for storage of CO2. Upon temperature rise, CaCO3 releases CO2, which re-oxidizes the carbon deposits and reduced Fe, thus producing carbon monoxide. The amount of produced CO is higher than the theoretical amount for Fe3O4, because carbon deposits from CH4 equally contribute to the CO yield. After each redox cycle, the material is regenerated, so that it can be used repeatedly, providing a stable process. © 2015 Elsevier B.V. All rights reserved.

Keywords: Methane Carbon dioxide Iron oxide Calcium oxide Fuel Carbon

1. Introduction Energy is one of the most important topics in the 21st century as it is the foundation of today's society [1,2]. With the rapid depletion of fossil fuels and increasing environmental pollution caused by immense fossil fuel consumption, there is a high demand to make more efficient use of energy. Novel renewable and clean energy sources can substitute fossil fuels to enable the sustainable development of our society. On the pathway to improved energy

* Corresponding author. E-mail address: [email protected] (V.V. Galvita). http://dx.doi.org/10.1016/j.jpowsour.2015.03.183 0378-7753/© 2015 Elsevier B.V. All rights reserved.

efficiency, energy storage is an intermediate step [3e7]. Energy can be stored in different forms: as mechanical energy; in an electric or magnetic field; as chemical energy of reactants and fuels [6]. With the rapid development of industries and the increase of global population, the rate of electrical energy consumption has dramatically increased and its consumption manner is diversified. Hence, energy storage becomes even more complex and important, and high-performance energy storage techniques are required to enable efficient, versatile, and environmentally friendly use of energy including electricity [4,5]. In a typical energy storage process, one type of energy is converted into another form which can easily be stored and converted for use when needed [5]. Various energy

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

storage systems are being developed aiming at proper utilization of different energy sources. The metaleair battery is one such advanced energy storage and conversion technology [8]. It converts the chemical energy in lithium (anode) and oxygen (cathode) into electric energy during discharge, and stores electric energy by splitting LieO2 discharge products during charging using electricity (like an electrolysis device or a reversible fuel cell to generate hydrogen and oxygen by splitting water). Recently, another type of metaleair battery was demonstrated combining a regenerative solid oxide fuel cell and a chemical looping redox cycle [9e11]. The air-electrode reactions in this battery involve only reduction and evolution of gaseous O2, making clogging of the air-pathway no longer an issue. In addition, solid oxide-ion electrolytes are known to be stable in a broad range of gas mixtures and in contact with a variety of oxides [12e15]. In the new battery, the solid oxide electrochemical cell serves as the “electrical functioning unit” operating alternately between fuel cell and electrolyzer modes to realize the discharge/charge cycles. On the other hand, the redox cycle unit acts as the “energy storage system” to carry out reversible chemicaleelectrical energy conversion via H2/H2O-mediated metal/metal oxide (Me/MeOx) redox reactions. In this way, chemical energy is stored in redox couples that are physically separated from the electrodes of the solid oxide electrochemical cell. The distinct advantages of this battery over conventional metaleair batteries include O2 transfer, state-of-charge independent electromotive force, high energy density and a design independent of power and energy [11,14,16]. The present novel energy storage unit proposes energy storage and conversion based on a combination of chemical looping and calcium looping processes. It uses a physical mixture of an oxygen storage material and a CO2 sorbent material. CO2 serves as mediation gas to facilitate metal oxidation and carbon gasification into CO by means of chemical looping [17e19], while the calcium looping process ensures storage and release of CO2 [20e25]. Chemical looping combustion (CLC) is an emerging combustion technology [18,26,27]. In this process, fuel is oxidized by a reducible metal oxide, e.g. Fe3O4, and the reduced metal oxide is re-oxidized by air in a separate step. CLC hence produces a pure CO2 stream, not diluted by N2. By replacing air with H2O or CO2 as oxidizer, the chemical looping analogue to steam and dry reforming is proposed [17,19,28,29]. This process converts CO2 to high-purity CO, providing an efficient path to CO2 conversion [17]. The overall stoichiometry of the process indeed demonstrates that the oxidation step converts more CO2 than was produced in the fuel combustion step. The calcium looping technology is a promising new technique for high-temperature scrubbing of CO2 from flue gas and syngas [30,31]. Calcium looping cycles have been intensively investigated in order to produce a concentrated CO2 stream from the utilization of fossil fuels and biomass. Efficient CO2 capture and storage using Ca-based sorbents can be achieved via the reversible reaction CaO þ CO2 #CaCO3 (DH298 K ¼ 178 kJ mol1), the so-called Calooping cycle. Combination of Fe3O4/Fe redox cycles with CaO/CaCO3 looping provides a number of important advantages. Where chemical looping dry reforming immediately oxidizes fuel to CO2, which is further converted to CO, the CO2 mediation gas is now stored to be used at will. The CaCO3 solid material acts both as CO2 reservoir and supplier, by cyclic CaO/CaCO3 calcination-carbonation. In addition, the adsorption of CO2 gas from the stream during iron oxide reduction promotes faster material reduction and carbon formation [32]. The working principle of the “combined chemical looping” system is schematically shown in Fig. 1. The operation of the

363

Fig. 1. Schematic working principle of the “combined chemical looping” for energy storage and conversion.

“chemical charge” and “discharge” cycles can be described as follows. During the reduction step (“charge process”), CH4 þ CO2 is fed into a mechanical mixture of Fe3O4 and CaO. Interaction of methane with Fe3O4 leads to formation of metallic iron and surface carbon as well as CO2, H2O and H2 (Eqs. (1) and (2)). The H2 (Eq. (2)) obtained can be used for fuel cells. At the same time carbonation of calcium oxide occurs by interaction of CO2 with CaO (Eq. (3)). CaO particles react with CO2 with typical high temperatures (600e700  C) to form CaCO3 [20,23,24,33]. CH4 þ Fe3O4 / 2H2O þ CO2 þ 3Fe

(1)

CH4 / C þ 2H2

(2)

CaO þ CO2 / CaCO3

(3)

The material can be kept in this “charged” condition if storage is required or be put to immediate use. For the oxidation step (“discharge”), the temperature of the sample is increased by 50e150  C which leads to decomposition of calcium carbonate into CaO and CO2 (Eq. (4)) [26,30,34]. CaCO3 / CaO þ CO2

(4)

At the same time, the interaction of CO2 with metallic iron as well as with carbon produces CO [17,35,36] via the following chemical reactions: 4CO2 þ 3Fe / Fe3O4 þ 4CO

(5)

C þ CO2 / 2CO

(6)

364

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

The thus generated CO could for example be used in a solid oxide fuel cell (SOFC) where it is electrochemically oxidized, CO þ O2 ¼ CO2 þ 2e, producing electricity and CO2 [12,37e40]. CO essentially contains as much energy as hydrogen. Energetics of CO and H2 oxidation, calculated from thermochemical data at an operating temperature of 1123 K, exhibit similar values for the standard Gibbs free energy, i.e. 185 kJ mol1 for CO2 versus 186 kJ mol1 for H2O. Also the CO oxidation reaction is slightly more exothermic than the oxidation of H2, i.e. 282 kJ mol1 for CO2 versus 249 kJ mol1 for H2O [41]. The objective of the present study is the experimental validation of the proposed concept of combined chemical looping. The proof of concept was performed in two steps. In the first, a single NiOeFe2O3/CeO2 material was used and CH4 was fed for iron oxide reduction and carbon formation. Interaction of this material with CO2 led to CO formation by iron and carbon oxidation by CO2. In the second step, the NiOeFe2O3/CeO2 material was mixed with CaO to perform a complete combined



chemical looping test with a feed of CH4þCO2. 2. Experimental 2.1. Materials preparation Samples of 10 wt% Ni/CeO2eFe2O3 were investigated in this study [36]. Fe2O3/CeO2 previously showed improved performance and high stability in CO2 conversion compared to pure Fe2O3 or CeO2 [17]. The addition of Ni to Fe2O3/CeO2 has a beneficial effect upon the iron oxide reduction, because the Ni catalyst converts the feed of CH4 þ CO2 into a mixture of CO and H2, which both reduce Fe3O4 and produce CO2 and H2O [36]. The following chemicals were used in the preparation of the mixed oxides: Fe(NO3)3$9H2O (99.99 þ %, SigmaeAldrich), and Ce(NO3)3$6H2O (99.99%, SigmaeAldrich). CeO2eFe2O3 was prepared in a 1:1 ratio by coprecipitation through addition of excess ammonium hydroxide. This mixture was maintained at room temperature for 12 h. Next,

Fig. 2. (a): Full XRD scan of as prepared Ni/FeeCe sample. (◄ e CeO2; D e NiO; e Fe2O3) (b): STEM BF image (c): EDX element mapping of Fe (blue) and Ce (green) and Ni (red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

365

2.2. Characterization

Fig. 3. 2D XRD pattern recorded during (a): H2-TPR for Ni/CeeFe; TPR measuring conditions: 20  C/min; 5% H2/He; (b): CO2-TPO for Ni/CeeFe; CO2-TPO measurement conditions: 20  C/min, 100% CO2; (c) Full XRD scans of Ni/FeeCe sample (1) reduced by H2 (5% H2 in He at 750  C); (2) reduced by CH4 (50% CH4 in He at 750  C); (3) oxidized by CO2 after CH4 reduction (100% CO2 at 750  C). (◄ e CeO2; C e FeNi3; - e Fe; + e graphite; A e Fe3C; B e Ni; , e Fe3O4).

the sample was separated as precipitate from the solution, washed with water and dried overnight in an oven at 110  C. Finally, the materials were calcined at 750  C. A CaO based material mixed with a thermally stable Al2O3 matrix was used for the calcium looping process. A 80wt.%CaO-20wt.% Al2O3 sample was prepared via wet physical mixing [21]. Citric acid and aluminium nitrate were dissolved in distilled water and stirred at room temperature. Then CaCO3 powder was added and reacted with citric acid in the solution. After drying, the sample was calcined at 750  C. The addition of Al2O3 ensured structural stabilisation of CaO.

The BrunauereEmmetteTeller (BET) surface area of the samples was determined by N2 adsorption at 196  C (five point BET method using Gemini Micromeritics). Prior to analysis the sample was outgassed at 300  C for 6 h to eliminate volatile adsorbates from the surface. The crystallographic phase of the prepared materials was determined using a Siemens Diffractometer Kristalloflex D5000, with Cu Ka radiation. The powder XRD patterns were collected in a 2q range from 10 to 80 with a step of 0.02 and 30 s counting time per angle. By fitting a Gaussian function to a diffraction peak, the crystallite size was determined from the peak width via the Scherrer equation while the peak position gives information about the lattice spacing based on the Bragg's law of diffraction: 2dsin(q) ¼ nl. Crystallographic analyses of the tested catalysts were performed by means of in situ X-ray diffraction (XRD) measurements in qe2q mode using a Bruker-AXS D8 Discover apparatus with Cu Ka radiation of wavelength 0.154 nm and a linear detector covering a range of 20 in 2q with an angular resolution of approximately 0.1 2q. While the minimal capturing time is 0.1 s, a collection time of 10 s was typically used during these experiments. The evolution of the catalyst structure during temperature programmed reduction or re-oxidation was investigated by in situ XRD in a flowing gas stream (5 vol% H2/He or 100% CO2). These in situ experiments were carried out using a home-built reactor chamber with a Kapton foil window for X-ray transmission. A 10 mg sample was evenly spread on a single crystal Si wafer. Interaction between the catalyst material and the Si holder was never observed. The chamber atmosphere was evacuated and flushed before introducing the reducing gas flow. The sample was heated from room temperature to 750  C at a rate of 20  C/min. In situ XRD during CO2-TPO immediately followed H2-TPR. Temperatures were measured using a K-type thermocouple and corrected afterwards with a calibration curve of the heating device, based on the eutectic systems AueSi, AleSi and AgeSi. Morphological, structural and local chemical analyses were carried out using transmission electron microscopy (TEM)-based methods: conventional TEM, high resolution (HRTEM), scanning transmission bright field (STEM BF) and energy dispersive X-ray spectrometry (EDX). The instrument employed in this study was a JEOL JEM-2200FS: Cs-corrected, operated at 200 kV, and equipped with a Schottky-type field-emission gun (FEG), EDX JEOL JED2300D and JEOL in-column omega filter (EELS). Specimens were prepared by immersion of a lacey carbon film on a copper support grid into the as prepared powder followed by blowing off the excess powder. Particles sticking to the carbon film were subjected to microscopy. A beryllium specimen retainer was used to eliminate secondary X-ray fluorescence in EDX spectra originating from the specimen holder. XPS was performed using a S-Probe monochromatized XPS spectrometer from Surface Science Instruments (VG) with AlKa xray (1486.6 eV) source (10 kV voltage and 200 W power). The base pressure in the main chamber was 2  107 Pa. The take-off angle was 45 and a pass energy of 107.8 eV with step ¼ 0.05 eV was used. 2.3. Reaction setup and procedures Activity measurements were carried out at atmospheric pressure in a quartz tube microreactor (i.d. 10 mm), placed in an electric furnace. Typically, 100 mg of sample was packed between quartz wool plugs. The samples were diluted 1:10 with quartz. The temperature of the catalyst bed was measured with K-type thermocouples touching the outside and inside of the reactor at the

366

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

position of the catalyst bed. In order to examine the reactivity and stability of these Ni/CeeFe materials, experiments were conducted in a fixed-bed reactor at 600e700  C using CH4 and CH4 þ CO2 (1:1) as model fuel. CO2 was used as the oxidant for oxidation of the chemical looping material. The total flow rate of the feed gas into the reactor was constantly maintained at 45 mmol/s using Brooks mass flow controllers. The feed and product gas streams were monitored on-line using MS, using argon as internal standard. The instrument was calibrated beforehand for different mixtures of feed and products. CH4 was followed at 16 AMU, CO2 at 44, 28 and 16 AMU, CO again at 28 AMU, H2 at 2 AMU and Ar at 40 AMU. Concentrations were determined taking into account the fragmentation patterns of the compounds. The response of the mass spectrometer detector was regularly verified with calibration gases. A carbon balance with a maximum deviation of 10% was obtained. 3. Results and discussion 3.1. Material characterization In the diffraction pattern of as prepared Ni/CeeFe only Fe2O3, CeO2 and NiO diffractions were observed (Fig. 2a). The peaks of NiO, CeO2 and Fe2O3 were used to determine the crystallite size, yielding 14 ± 5 nm for NiO, 45 ± 7 nm for Fe2O3 and 22 ± 9 nm for CeO2. The latter values were in line with the TEM images of the Ni/CeeFe sample, which are shown in Fig. 2b. Elemental distribution of the material was determined using energy-dispersive X-ray spectroscopy (EDX)-STEM mapping. Both Fe (blue) and Ce (green) elements were distributed uniformly in the sample. In contrast, Ni (red) was preferably located close to or on the Fe2O3 particles (Fig. 2c). Pure 10%Nie50%CeO2e50%Fe2O3 (Ni/CeeFe) had a BET surface area ~15 m2/g. In situ XRD measurements were performed during H2-TPR in order to understand the transformation of different components of the sample during the reduction process. The result for the Ni/ CeeFe sample is presented in Fig. 3a. Reduction of the sample was observed only at temperature above 377  C, where Fe2O3 evolves into Fe3O4, then further into FeO at 570  C and finally into Fe above 630  C. NiO starts reducing above 330  C and a Ni peak appears. Between 500  C and 630  C, the Ni peak shifts downward and passes into a NieFe alloy peak at a slightly lower angle.

CO2-TPO with in situ XRD monitoring was performed after cool down following H2-TPR (Fig. 3b). The re-oxidation of Fe to Fe3O4 was observed around 480  C without intermediate FeO phase. The further oxidation of Fe3O4 to Fe2O3 can only be achieved by application of gaseous oxygen [42]. The NieFe alloy remained stable up to 530  C in CO2 atmosphere and then shifted above 44 around 630  C, where it indicates a remaining Ni contribution. Oxidation of Ni to NiO by carbon dioxide is not obvious and was therefore not observed in this temperature window. The full XRD patterns of the sample reduced by H2, by CH4 and after CO2 re-oxidation of the methane reduced sample are shown in Fig. 4. Reduction of the sample by H2 leads to formation Fe and NieFe alloy (Fig. 3c pattern (1)). Based on the diffraction positions the alloy was identified as FeNi3 (PDF: 03-065-3244). After reduction of the sample by CH4, the peaks of Fe disappeared completely and were replaced with Fe3C diffractions. In addition, a new diffraction peak appeared at 2q ¼ 26 . This peak was attributed to growth of graphitic-like structures on the surface of the sample. Further distinction was not possible as the XRD patterns are not suitable to differentiate microstructural details between carbon nanotube and other similar graphitic-like structures, since the carbon nanotube diffraction peaks are close to those of graphite [43,44]. Literature results indicated before that Fe2O3 in the as prepared catalyst is reduced stepwise with methane during the reaction, i.e., Fe2O3 / Fe3O4 / FeO / Fe metal / Fe3C [45]. CO2 oxidizing of the CH4 reduced sample removed the FeC and graphitic structure diffractions and led to separation and transformation of the FeNi3 alloy into metallic Ni and Fe3O4 (Fig. 3c pattern 3). XPS measurements were performed on two samples after CH4 isothermal reduction and after CO2 isothermal oxidation. Due to the ex situ nature of these measurements, and the surface sensitivity of this technique, limiting the information depth to ~3 nm, strong contributions of metal oxides were observed, possibly related to air exposure. For the reduced sample, no clear alloy contribution was observed in either Ni or Fe window, but these could still be present as the resolution is limited and alloying typically introduces only small shifts [46e48]. The XPS wide scans showed photoemission lines for C, O, Ce and Fe as well as weak ones for Ni, the latter two with surface concentrations of 8.5 at% and 2.8 at%, respectively (not shown). Fig. 4a and b shows the detailed multiplex windows for Ni2p3/2 and Fe2p3/2 for both samples. After reduction with methane, both XPS lines have a

Fig. 4. XPS detail windows for 10 wt% Ni/CeO2eFe2O3 after CH4 isothermal reduction (grey lines) and after CO2 isothermal oxidation (black lines) with deconvolution for a: Ni2p3/2 and b: Fe2p3/2. The Fe peak around 706.7 eV (Fe0 lumped) contains metal and alloy contributions; the Fe oxide peak combines Fe2þ and Fe3þ oxide contributions (Fe oxides lumped).

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

367

clear metal-like contribution: Ni at 852.5 eV with plasmon loss satellite around 858.7 eV, and Fe at 706.7 eV. The presence of a NieFe alloy together with the metallic Fe phase, as indicated by XRD, is not easily discerned here. According to XPS literature on NieFe alloys, both Ni0 and Fe0 can shift over þ0.3e0.5 eV upon alloying [46e48]. Thus, the Ni0 line should be regarded as stemming mainly from the alloy phase. For Fe on the other hand, the photoemission intensity mainly results from metallic Fe, while the presence of alloyed Fe cannot be resolved. Hence, the photoemission line for Fe0 is indicated as ‘lumped’ in Fig. 4b and should be regarded as combining an alloy contribution with the pure metal intensity. Next to the metal lines, oxide contributions can be distinguished in both reduced samples: Ni2þ at ~855 eV with corresponding multiplet intensity at ~862 eV (Fig. 4a) [49], and a broad oxide peak in Fig. 4b, where Fe2þ and Fe3þ oxide contributions are lumped together [46]. In the same range of Fe binding energy, a Ni Auger line is positioned (712 eV for Ni LMM1), but given the overall low intensity of the Ni signal (not shown), this small Auger intensity was ignored here. The oxide contributions in Ni2p and Fe2p can be ascribed to an oxide shell resulting from partial oxidation of the outer surface of the reduced samples, which were exposed to air prior to XPD measurement [47]. After oxidation with CO2, the Ni2þ intensity has grown with respect to Ni metal (Fig. 4a). indeed, the Ni0/Ni2þ intensity ratio goes from 0.65 for the reduced state to 0.32 after oxidation. The latter will result from the combined effect of CO2 re-oxidation and further post-reaction air exposure. As for the Fe photoemission (Fig. 4b), all metal-like Fe and alloy intensity has gone, leaving only Fe oxide contributions, representative of Fe3O4, as found from XRD. The calcium oxide based CO2 sorbent material was characterized by means of SEM (Fig. 5a). The material showed an open porous structure. In the diffraction pattern of the as prepared sample material (Fig. 5b), the contribution of crystalline calcium oxide resulted in distinctive and intense 2q-signals at 32 , 37, 53 , 64 and 67. The presence of calcium aluminum oxide (Ca3Al2O6) was equally observed. The peaks at 2q ¼ 30 and 2q ¼ 40 confirm the presence of calcite CaCO3. The average particle size of calcium oxide is determined by applying Scherrer's equation to the most intense characteristic peaks of CaO. The estimated particle size is ~50 nm. The results from N2 adsorption show a specific surface area of roughly 15 m2/g.

Fig. 5. (a) SEM images of the as prepared CaOeAl2O3 sample; (b) XRD pattern of as prepared alumina-promoted calcium oxide material with peaks assigned to calcium oxide (CaO, :), calcium aluminum oxide (Ca3Al2O6, -), and calcite (CaCO3, C).

3.2. Activity test For the single material test, a CH4 feed was used for the sample reduction and graphite formation, while oxidation was performed using CO2. The outlet space time yields (STY) of H2, CO2, CO and CH4 as a function of time are shown in Fig. 6a when CH4 was introduced

Fig. 6. (a) Space-time yield vs. time on stream during one cycle of reduction of the NiOeFe2O3/CeO2 by CH4 at 700  C; (b) space-time yield of CO during the re-oxidation phase with CO2/He. Solid line: reduction by H2; dotted line: reduction by CH4 for 5 min; dashed line: reduction by CH4 for 8 min; dash-dot line: reduction by CH4 for 20 min.

368

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

Fig. 7. SEM images of the NiOeFe2O3/CeO2 sample (a) as prepared, after (b) 5 min reduction, (c) 20 min reduction (d) CO2 re-oxidation after 20 min reduction time. Reduction conditions: 50% CH4 in Ar, 750  C.

to the reactor at 700  C. CO and CO2 rise quasi-instantaneously up to ~0.06 mol s1 kg1 and after 4 min decrease again implying that all oxygen in Fe2O3 and NiO was consumed. The STY of H2 has two maxima: the first is attributed to methane decomposition over NiO. The second maximum corresponds to methane decomposition over metallic iron, followed by C deposition. The graphite formation indeed occurs after iron oxide reduction [50,51]. This includes CH4 decomposition (Eq. (6)), where CH4 dissociates completely to form graphite on the catalyst surface and produces H2. Graphite as well as Fe3C are present in the full scan XRD after CH4 interaction (Fig. 3c, pattern 2). SEM images of the as prepared sample and after CH4 reduction during 5 and 20 min are shown in Fig. 7aec, respectively. The reduced samples all showed carbon with a filamentous structure grown on the catalyst, as reported by many researchers [52]. The diameter range of the filamentous carbon was 40e100 nm. The visible amount of carbon significantly increased with increasing CH4 exposure time. Comparing the CO produced upon re-oxidation of NiOeFe2O3/ CeO2 reduced by H2 or by CH4, the re-oxidation process in the former case is significantly faster [36], as the CO production was close to completed after 2.5 min (Fig. 6b). For the sample reduced by CH4 during 5, 8 and 20 min, the CO production lasted for 7, 15 and 25 min respectively, due to the slow process of carbon oxidation. The amount of CO formed is tabulated in Table 1. CO produced after H2 reduction was close to 97% of the theoretical value based only on the oxygen storage capacity of Fe3O4 (Eq. (5)) as no carbon was deposited. When however carbon is formed upon reduction with CH4 (Eq. (2)), the CO yield from CO2 interaction with the “charged” sample is increased by 2.0, 4.7 and 8.8 times for the different methane exposure times compared to CO formation from iron re-oxidation only. The latter increase is due to the additional CO yield gained by oxidation of deposited C (Eq. (6)). This clearly demonstrates that interaction of CO2 with metallic iron and carbon filaments, formed upon CH4 reduction of the NiOeFe2O3/CeO2

Table 1 CO yield after “discharge” step. CO yield mol/kgFe

CO total CO from carbon COcarbon/COFe

Reduction gas/Reduction time H2 30 min

CH4 5 min

CH4 8 min

CH4 20 min

23.0 e e

69.5 46.5 2.0

131.0 108.0 4.7

225.0 202.0 8.8

material, is a viable approach to make a high capacity energy storage medium. A SEM image of the sample after 20 min CH4 reduction and subsequent 25 min of CO2 re-oxidation is shown in Fig. 7d. The morphology of the sample did not change compared to the reduced sample, Fig. 7c, but most of the carbon was indeed converted. Upon interaction with CO2, direct oxidation of the carbon filaments occurred through the Boudouard reaction (Eq. (6)) [53]. Further, all graphite and carbide diffractions disappeared from the XRD pattern (Fig. 3c pattern 3), indicating that lattice oxygen from the reoxidized Fe3O4 (Fig. 3b) in turn oxidized the surface carbon [17,19]. Finally, the concept combined chemical looping was tested. An equimolar mixture of CH4 and CO2, as a model biogas feed, was fed to a fixed bed reactor containing a mixture of NiOeFe2O3/CeO2 and CaOeAl2O3 materials with ratio 1:5 by weight during a sequence of isothermal charge/discharge cycles. In Fig. 8a the space-time yields of methane, carbon oxides and hydrogen are plotted as a function of time on stream at 650  C. During the initial 30 s of the experiment, CO2 goes through a minimum because it is taken up by CaO as well as consumed by dry reforming to form CO and H2. Once the available CaO is fully saturated with CO2, the space-time yield for CO2 increases and stabilizes with the other products. CH4 equally exhibits a minimum because given the low amount of CO2 methane decomposition over the Ni catalyst takes place and surface carbon and hydrogen are

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

369

Fig. 8. Space-time yield vs. time on stream for combined CO2 capture and Fe3O4 reduction over the mixture of NiOeFe2O3/CeO2 and CaOeAl2O3 (1:5). (a) CaO carbonation and Fe3O4 reduction with CH4 þ CO2 (1:1) at 650  C; (b) decarbonation of CaCO3 and CO production from oxidation of iron and carbon by CO2. Temperature increase from 650 to 720  C in He flow. Solid line: temperature profile.

formed. The space-time yield for H2 increases and goes through a maximum before reaching a constant value. CO and H2 both contribute to the reduction of Fe3O4. The space-time yields of reagent and products reach constant values after 200 s, which correspond to catalysed dry reforming. At this point, the process can be halted to store the energy, delaying the “discharge” at will. For CO2 release, the reactor is purged with He and the reaction temperature is increased from 650  C to 720  C. Carbon dioxide is now released by CaCO3 decomposition and CO is produced as a result of the oxidation of iron and carbon by CO2 (Fig. 8b). The CO production reaches a maximum at 200 s and then steadily decreases toward zero after 1000 s. After charging/discharging, the material is ready for reutilization, since it is regenerated in its original state. The full process can thus be applied as a combined chemical looping, which consists of the two steps of charging and discharging occurring in a cyclic and discontinuous manner. In addition to restoring the oxygen storage capacity, the re-oxidation step removes the surface carbon and eventually gives rise to a higher yield of CO compared to the theoretical value corresponding to complete oxidation of Fe to Fe3O4. In view of the process viability, the stability of the materials and their activity was examined during repeated charge and discharge cycles. The yield of carbon monoxide remained close to constant after the 7th cycle, which confirmed the stability of the materials' activity.

4. Conclusions A novel concept of energy storage was demonstrated in a laboratory scale test. The proposed combined chemical looping is able to store and release energy from redox chemical looping reactions combined with CO2 storage/release calcium looping. This process used Fe3O4 and CaO, two low cost and environmentally friendly materials. CO2 served as mediation gas to facilitate the metal/metal oxide redox reaction and carbon gasification into CO. Both formation of carbon and metallic iron occur during the reduction of Fe3O4 by CH4. On the other hand, CaO is used for storage and release of CO2. Upon temperature rise, CaCO3 releases CO2, which re-oxidizes the carbon deposits and reduced Fe, thus producing carbon monoxide. The resulting CO can be used e.g. as fuel in a solid oxide fuel cell. The amount of CO produced is higher than the theoretical amount for Fe3O4, because carbon deposits from CH4 equally contribute to the CO yield. After one loop, the material is regenerated, so that it can be used repeatedly, providing a stable process. The proposed process offers a wide flexibility both in energy

storage/release and fuel type. For charging, diluted gas streams can be applied, making it very attractive for present-day applications. On the other hand, the discharge or use of energy can easily be uncoupled in time from the charging. Acknowledgments This work was supported by the “Long Term Structural Methusalem Funding by the Flemish Government”, the Fund for Scientific Research Flanders (FWO, project number 3G004613), and the Interuniversity Attraction Poles Programme, IAP7/5, Belgian State e Belgian Science Policy. We also thank Dr. Vitaliy Bliznuk, Department of Materials Science and Engineering, Ghent University, for his support with the HRTEM measurements and acknowledge support and use of the in situ XRD equipment from Prof. C. Detavernier of the Department of Solid State Sciences, Ghent University. References [1] A.M. Omer, Renew. Sust. Energy Rev. 12 (2008) 2265e2300. [2] A. Antonino Salvatore, P. Bruce, B. Scrosati, J.-M. Tarascon, W. van Schalkwijk, Nat. Mater. 4 (2005) 366e377. [3] T.M.I. Mahlia, T.J. Saktisandan, A. Jannifar, M.H. Hasan, H.S.C. Matseelar, Renew. Sust. Energy Rev. 33 (2014) 532e545. [4] J.M. Bermudez, E. Ruisanchez, A. Arenillas, A.H. Moreno, J.A. Menendez, New Concept for Energy Storage: Microwave-induced Carbon Gasification with CO2, 2014, pp. 559e564. [5] T. Kousksou, P. Bruel, A. Jamil, T. El Rhafiki, Y. Zeraouli, Energy Storage: Applications and Challenges, 2014, pp. 59e80. [6] L. Schlapbach, A. Zuttel, Nature 414 (2001) 353e358. [7] C. Liu, F. Li, L.P. Ma, H.M. Cheng, Adv. Mater. 22 (2010) E28eE62. [8] Y. Shao, S. Park, J. Xiao, J.-G. Zhang, Y. Wang, J. Liu, ACS Catal. 2 (2012) 844e857. [9] X. Zhao, N. Xu, X. Li, Y. Gong, K. Huang, RSC Adv. 2 (2012) 10163e10166. [10] X. Zhao, X. Li, Y. Gong, K. Huang, Chem. Commun. 50 (2014) 623e625. [11] A. Inoishi, Y.W. Ju, S. Ida, T. Ishihara, J. Power Sources 229 (2013) 12e15. [12] J.C. Ruiz-Morales, D. Marrero-Lopez, M. Galvez-Sanchez, J. Canales-Vazquez, C. Savaniu, S.N. Savvin, Energy & Environ. Sci. 3 (2010) 1670e1681. [13] B.C.H. Steele, A. Heinzel, Nature 414 (2001) 345e352. [14] N. Xu, X. Li, X. Zhao, J.B. Goodenough, K. Huang, Energy & Environ. Sci. 4 (2011) 4942e4946. [15] D. Papurello, R. Borchiellini, P. Bareschino, V. Chiodo, S. Freni, A. Lanzini, F. Pepe, G.A. Ortigoza, M. Santarelli, Appl. Energy 125 (2014) 254e263. [16] X. Zhao, X. Li, Y. Gong, N. Xu, K. Romito, K. Huang, Chem. Commun. 49 (2013) 5357e5359. [17] V.V. Galvita, H. Poelman, V. Bliznuk, C. Detavernier, G.B. Marin, Industrial Eng. Chem. Res. 52 (2013) 8416e8426. [18] S. Bhavsar, M. Najera, R. Solunke, G. Veser, Catal. Today 228 (2014) 96e105. [19] M. Najera, R. Solunke, T. Gardner, G. Veser, Chem. Eng. Res. Des. 89 (2011) 1533e1543. [20] J.C. Abanades, D. Alvarez, Energy & Fuels 17 (2003) 308e315. [21] M. Zhang, Y. Peng, Y. Sun, P. Li, J. Yu, Fuel 111 (2013) 636e642. [22] G. Grasa, I. Martanez, M.E. Diego, J.C. Abanades, Energy & Fuels 28 (2014) 4033e4042. [23] S. Lin, M. Harada, Y. Suzuki, H. Hatano, Fuel 81 (2002) 2079e2085. [24] G. Itskos, P. Grammelis, F. Scala, H. Pawlak-Kruczek, A. Coppola, P. Salatino,

370

V.V. Galvita et al. / Journal of Power Sources 286 (2015) 362e370

E. Kakaras, Appl. Energy 108 (2013) 373e382. [25] V. Galvita, K. Sundmacher, J. Mater. Sci. 42 (2007) 9300e9307. [26] I. Vorrias, K. Atsonios, A. Nikolopoulos, N. Nikolopoulos, P. Grammelis, E. Kakaras, Fuel 113 (2013) 826e836. [27] J. Adanez, A. Abad, F. Garcia-Labiano, P. Gayan, L.F. de Diego, Prog. Energy Combust. Sci. 38 (2012) 215e282. [28] M. Ryden, M. Arjmand, Int. J. Hydrogen Energy 37 (2012) 4843e4854. [29] C.L. Muhich, B.W. Evanko, K.C. Weston, P. Lichty, X. Liang, J. Martinek, C.B. Musgrave, A.W. Weimer, Science 341 (2013) 540e542. [30] E.J. Anthony, Greenh. Gases: Sci. Technol. 1 (2011) 36e47. [31] M. Ryden, P. Ramos, Fuel Process. Technol. 96 (2012) 27e36. [32] I. Alstrup, M.T. Tavares, J. Catal. 139 (1993) 513e524. [33] T. Shimizu, T. Hirama, H. Hosoda, K. Kitano, M. Inagaki, K. Tejima, Chem. Eng. Res. Des. 77 (1999) 62e68. [34] J.M. Valverde, P.E. Sanchez-Jimenez, L.A. Perez-Maqueda, Appl. Energy 126 (2014) 161e171. [35] A.C. Lee, R.E. Mitchell, T.M. Gür, AIChE J. 55 (2009) 983e992. [36] V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Appl. Catal. B: Environ. 164 (2015) 184e191. [37] V.V. Galvita, V.D. Belyaev, A.K. Demin, V.A. Sobyanin, Appl. Catal. A: General 165 (1997) 301e308. [38] T.I. Politova, V.V. Gal'vita, V.D. Belyaev, V.A. Sobyanin, Catal. Lett. 44 (1997) 75e81. [39] A. Utz, A. Leonide, A. Weber, E. Ivers-Tiffee, J. Power Sources 196 (2011) 7217e7224. [40] V.V. Gal'vita, V.D. Belyaev, V.N. Parmon, V.A. Sobyanin, Catal. Lett. 39 (1996)

209e211. [41] M. Homel, T.M. Gur, J.H. Koh, A.V. Virkar, J. Power Sources 195 (2010) 6367e6372. [42] J. Zielinski, I. Zglinicka, L. Znak, Z. Kaszkur, Appl. Catal. A: General 381 (2010) 191e196. [43] W.Z. Zhu, D.E. Miser, W.G. Chan, M.R. Hajaligol, Mater. Chem. Phys. 82 (2003) 638e647. [44] A.E. Awadallah, A.A. Aboul-Enein, D.S. El-Desouki, A.K. Aboul-Gheit, Catalytic Thermal Decomposition of Methane to COx-free Hydrogen and Carbon Nanotubes over MgO Supported Bimetallic Group VIII Catalysts, 2014, pp. 100e107. [45] C. Emmenegger, J.M. Bonard, P. Mauron, P. Sudan, A. Lepora, B. Grobety, A. Zuttel, L. Schlapbach, Carbon 41 (2003) 539e547. [46] M. Dickinson, T. Scott, R. Crane, O. Riba, R. Barnes, G. Hughes, J. Nanoparticle Res. 12 (2010) 2081e2092. [47] N. Moghimi, S. Bazargan, D. Pradhan, K.T. Leung, J. Phys. Chem. C 117 (2013) 4852e4858. [48] J. Ashok, S. Kawi, Steam Reforming of Biomass Tar Model Compound at ^V“iron Relatively Low Steam-to-carbon Condition over CaO-doped Nickela ^V“alumina Catalysts, 2015, pp. 24e35. Alloy Supported over Irona [49] Y. Han, W. Yan, Water Res. 66 (2014) 149e159. [50] D. Pakhare, J. Spivey, Chem. Soc. Rev. (2014). [51] X.E. Verykios, Int. J. Hydrogen Energy 28 (2003) 1045e1063. [52] Y. Li, D. Li, G. Wang, Catal. Today 162 (2011) 1e48. [53] A.L. Yergey, F.W. Lampe, Fuel 53 (1974) 280e281.