Use of chemically and physically mixed iron and nickel oxides as oxygen carriers for gas combustion in a CLC process

Abstract Different bimetallic Fe–Ni-based OCs have been prepared and evaluated in a TGA, a batch fluidised bed reactor, and a continuous CLC unit in order to analyse the effect of NiO content on the CLC performance when CH4 or a PSA-offgas was used as fuel. A set of experiments was conducted in continuous operation in a 500 Wth CLC unit, firstly working with a chemically mixed OC, with the iron and nickel oxides impregnated over the same alumina particle, and secondly working with a physical mixture of two impregnated Fe- and Ni-based OCs. The results were also compared with those obtained with an unmixed Fe-based OC. The effect on the combustion efficiency of different operating conditions, such as fuel composition, oxygen carrier to fuel ratio and fuel reactor temperature has been determined in the continuous unit. It was found that the use of a chemically mixed OC had a negative effect on the combustion efficiency since the formation of Fe–Ni compounds reduced the catalytic effect of Ni addition. On the other hand, a physically mixed OC with 2% of NiO increased significantly the combustion efficiency at low temperatures.


Introduction
Climate change is among the largest environmental, social and economic challenge currently facing mankind. In so far as climate change is concerned, there is, today, overall consensus on the need to reduce greenhouse gas emissions globally by 50% by 2050. This represents a cut of at least 80% in the industrialized world. This will mean, from now until 2050, considerable re-organization of the way in which society works (work, transport, leisure, city planning, housing, electricity production). The sectors responsible for emissionpower generation, industry, transport, buildings and constructionmust all prepare the transition to a low-carbon economy [1]. CO 2 capture and storage technologies (CCS) have been identified as one of the options necessary to overcome the anthropogenic emissions [2]. Among the different technologies for CO 2 capture, the chemical-looping combustion (CLC) process was suggested as a worthy alternative to reduce the economic cost of CO 2 capture [3].
CLC involves combustion of fuels with an oxygen carrier (OC), generally a metal oxide and a binder, which transfers oxygen from the air to the fuel by means of its exchange between two different reactors, avoiding in this way the direct contact between fuel and air. In conventional combustion the flue gas stream consists of carbon dioxide, steam and mostly nitrogen. Carbon capture in this combustion involves considerable energy for separating the CO 2 from the N 2 . In CLC, CO 2 separation is simply accomplished because the flue gas stream consists only of CO 2 and steam. By steam condensation, a pure CO 2 stream is produced. Therefore, CLC provides a sequestration ready CO 2 stream without the need for using costly gas separation techniques. Moreover, the net chemical reaction and energy release are similar to that of conventional combustion of the fuel.
A CLC system is generally composed of two interconnected fluidized bed reactors (Fig. 1) designated as air (AR) and fuel reactors (FR), where the solid metal oxide particles are circulated between the reactors. In the fuel reactor, the fuel gas (C n H 2m ) is oxidized to CO 2 and H 2 O by a metal oxide (MeO) that is reduced to a metal (Me) or a reduced form of MeO. The FR is typically a bubbling or circulating bed. The metal or reduced oxide is further transferred into the air reactor where it is oxidized with air, and the material regenerated is ready to start a new cycle. The flue gas leaving the air reactor contains N 2 and unreacted O 2 . The exit gas from the fuel reactor contains only CO 2 and H 2 O.
A large volume of knowledge on the CLC technology has been accumulated during recent years. CLC for gaseous fuels have successfully demonstrated in different CLC prototypes in the 10-140 kW th range using oxygen carriers based on nickel, cobalt, and copper oxides [4]. Many of the studies on CLC have focused on the development of appropriate oxygen carriers able to comply with the requirements imposed by the process. Tests with around 700 different materials based on transition metal oxides, mixed oxides, and low-cost materials have been reported [4].
An oxygen carrier for CLC should have the following properties: high fuel conversion to CO 2 and H 2 O; high reduction and oxidation rates; low tendency to agglomeration, fragmentation and attrition; low possibility of deactivation by carbon deposition or sulphur compounds; easy preparation to reduce costs; and it would be desired that the oxygen carrier will be environmental friendly.
Because of its low cost and environmental compatibility, Fe-based OC is considered an attractive option for CLC application. Other chemical characteristics are advantageous for the use of Fe-based oxygen carriers: low tendency to carbon deposition [5] and no risk of sulphide or sulphate formation at any sulphur containing gas concentration or operating temperature [4]. However, an important disadvantage is its low reactivity with methane [5]. But, it was observed that during the combustion of methane side reactions may occur, like the reforming reaction or the water-gas shift reaction, which generate CO and H 2 , gases which reacts faster with Fe-based oxygen carriers [5][6][7]. So, higher combustion efficiencies could be obtained increasing the reaction rate of the reforming reaction, and therefore, forcing the reaction to proceed via the intermediate products H 2 and CO.
During the last years, it has been found that the combination of different oxygen-carrier materials may result in positive synergy effects, taking advantage of the favourable characteristics of each of them [4]. Different methods have been reported on literature to produce mixed oxides, such as the simple mixing [8][9][10][11][12][13], the impregnation of a second active phase onto existing particles [14] or the direct preparation of a multiple active phase material [8,9,[15][16][17][18][19][20][21][22][23][24]. Considering this and in order to improve the reactivity of the Fe-based carriers with methane, some authors mixed them with a small amount of a high reactive material. Ni-based oxygen-carriers have shown very high reactivity and good performance working at high temperatures (900-1100°C). Several investigations [25,26] found that NiO particles reacted with CH 4 through CO and H 2 formation, because the metallic nickel formed on the particle surface enhanced the reforming of methane. The indication that nickel catalyzes the methane reforming and the fact that iron oxide reacts fast with CO and H 2 , suggest that a combination of both types of oxides may show synergy effects with an increased overall rate of reaction with respect to iron. This may have great implication in terms of the cost and safety of oxygen carriers since nickel is more expensive than other metal oxides, and the use of Ni-based oxygen-carriers may require safety measures because of its toxicity.
Bimetallic oxygen carriers of Fe-Ni have been prepared by different researchers. Materials containing Ni and Fe with a spinel structure were tested in a TGA by Lambert et al. [27]. They found that impregnating NiO on a spinel material increased both oxygen-carrier capacity and reactivity of the resulting material. Lagerbom et al. [28] tested in a TGA a bimetallic Fe-Ni/Al 2 O 3 oxygen-carrier and observed that addition of NiO to Fe 2 O 3 /Al 2 O 3 particles improved the activity but decreased the mechanical strength. Son and Kim [22] carried out experiments in a continuous CFB using different Fe-Ni/bentonite particles. They found that the reactivity of the oxygen-carrier particles increased with increasing NiO content. The optimum ratio of NiO/Fe 2 O 3 was found to be 3 (NiO/Fe 2 O 3 = x75:25). In addition to mixed oxides with the spinel structure and bimetallic oxygen-carriers, other more complex metal oxides with perovskite structure have been proposed to be used as oxygen-carriers for the CLC process [29]. However, long-term chemical and mechanical properties of perovskite particles are largely unknown and further investigation with these new materials is needed to know its behaviour in continuous fluidized-bed reactors.
The addition of Ni-based particles in a bed of Fe-based particles has also been investigated. Several studies were performed with the addition of certain amount of nickel oxide using different experimental configurations as fixed bed [8], batch fluidized-bed reactor [10][11][12], 300 W th circulating fluidized-bed reactor [13] and 500 W th unit [9]. Johansson et al. [10] found that a bed of iron oxides with only 3 wt.% nickel oxide was sufficient to give a very high CH 4 conversion. In addition, these researchers showed that the mixed-oxide system produced significantly more CO 2 than the sum of the metal oxides that run separately, thus giving evidence of the synergy in using nickel oxide together with iron oxide. Very similar findings were also observed by Ryden et al. mixing NiO60-MgAl 2 O 4 either in a bed of Fe 2 O 3 60-MgZrO 2 [12], in a bed of ilmenite [13] or waste products from the steel industry [11]. Ortiz et al. [9] reported an increase in the combustion efficiency in a continuous 500 W th CLC prototype using PSA-offgas as fuel. Moghtaderi and Song [30] carried out a theoretical analysis of the kinetic parameters when physically mixed-metal oxides are used. No clear effect on the resulting kinetics of the mixture was achieved.
So, physical and chemical addition of Ni to a Fe-based OC has been previously investigated in the literature. However, a comparison between both Ni addition methods was never carried out before. So, this works the effects of Ni addition over a Fe-based OC that has been evaluated and compared using two different strategies: the chemical and the physical mixtures. Thus, different bimetallic OCs have been prepared by impregnation on an alumina support and evaluated in a TGA and batch fluidized bed reactor to analyse the effect of NiO content on the reactivity and gas product distribution. Moreover, a physically mixed OC, using a Fe-based and a Ni-based OC, was also evaluated in the combustion of fuel gases.  The effect of this Ni addition on the combustion efficiency was analysed in a 500 W th CLC continuous unit using different combustion gases and operating conditions (temperature, oxygen carrier to fuel ratio and fuel composition). A first set of experiments were conducted working with a chemically mixed OC, with the iron and nickel oxides impregnated over the same alumina particle. A second set of experiments were performed working with a physical mixture of two impregnated Fe-and Ni-based OCs. The results were also compared with those obtained by the authors [31] with a highly reactive unmixed Fe-based OC developed and tested in a continuous CLC unit.

Oxygen carrier materials
The behaviour of several chemically mixed Fe-Ni oxygen carriers was analysed in this work together with a physically mixed Fe-Ni OC, using a Fe-based and a Ni-based OC. Fe-based OC was prepared by incipient wetness impregnation over commercial γ-Al 2 O 3 (Puralox NWa-155, Sasol Germany GmbH) particles of 0.1-0.32 mm with density of 1.3 g/cm 3 and porosity of 55.4%. The details of the preparation of the Fe-based OC have been described elsewhere [31]. Ni-based OC was prepared over α-Al 2 O 3 (obtained by calcination of γ-Al 2 O 3 at 1150°C during 2 h) particles of 0.1-0.32 mm with density of 2.0 g/cm 3 and porosities of 47.3% respectively. The details of the preparation of the Ni-based OC have been described elsewhere [32].
Bimetallic oxides were prepared by applying successive impregnation steps after the iron impregnation, followed by calcination at 550°C in air atmosphere for 30 min, with a nickel nitrate solution in the exact concentration to produce carriers with different Fe:Ni ratios from 1.4 to 10. Finally the particles were sintered at 950°C in air atmosphere for 1 h. Table 1 shows the different oxygen carriers prepared. The samples were designated with the chemical symbols referred to the active metal oxides followed by their weight content and an indication of the mixture: WM refers to without mixed OC, CM is used for the chemical mixed bimetallic OC and PM is used for the Fe-Ni OC made by a physical mixture of the unmixed Fe-OC and the Ni-OC.

Oxygen carrier characterization
Several techniques have been used to characterize physically and chemically the fresh and after-used oxygen carrier particles. The metal active content for the CLC process was determined by complete reduction of the sample with hydrogen in TGA at 950°C.
The oxygen transport capacity, R OC , was calculated assuming that Fe 2 O 3 is reduced to FeO·Al 2 O 3 and NiO to Ni in the CLC process.
The force needed to fracture a particle was determined using a Shimpo FGN-5X crushing strength apparatus. The mechanical strength was taken as the average value of at least 20 measurements. The real density of the particles was measured with a Micromeritics AccuPyc II 1340 helium pycnometer. The surface area of the oxygen carrier was determined by the Brunauer-Emmett-Teller (BET) method by adsorption/desorption of nitrogen at 77 K in a Micromeritics ASAP-2020 (Micromeritics Instruments Inc.), whereas the pore volume was measured by Hg intrusion in a Quantachrome PoreMaster 33. The identification of crystalline chemical species was carried out by powder X-ray diffractometer Bruker AXS graphite monochromator. The chemically mixed Fe-Ni oxygen carrier particles were also analysed in a scanning electron microscope (SEM) ISI DS-130 coupled to an ultra thin window PGT Prism detector for energy-dispersive X-ray (EDX) analysis.

Reactivity tests in TGA
The reactivity of the different oxygen carriers was determined in a TGA, CI electronics type, described elsewhere [33]. For the experiments, the oxygen carrier was loaded in a platinum basket and heated to the set operating temperature in air atmosphere. After weight stabilization, the experiment was started by exposing the oxygen carrier to alternating reducing and oxidizing conditions. To avoid mixing of combustible gas and air, nitrogen was introduced for 2 min after each reducing and oxidizing period.
The reactivity of the oxygen carrier was determined with different reducing gases: CH 4 , CO and H 2 at different temperatures. The gas composition was 15 vol.% of the reducing gas. In the experiments with CH 4 , 20 vol.% H 2 O was introduced to avoid carbon formation by methane decomposition. Steam was incorporated to the gas stream by bubbling through a water containing saturator at the selected temperature to reach the desired water concentration. Similarly, 20 vol.% CO 2 was introduced together with CO to avoid carbon formation by the Boudouard reaction. In all cases, nitrogen was used to balance. For oxidation reaction, 100% air was used as reacting gas. The effect of temperature on the reactivity of the OCs was evaluated using the different fuel gases at 830 and 950°C.The conversion of solids for the reduction reaction was calculated as: being m ox the mass of the fully oxidized solids, m the instantaneous mass of the sample and R OC the oxygen transport capacity of solids for the transformation between Fe 2 O 3 and FeAl 2 O 4 + NiO/Ni, given in Table 1.
The conversion for the oxidation reaction was calculated as X o = 1 − X r .

Fluidized bed reactor
Several reduction-oxidation multicycles were carried out in a batch fluidized bed reactor to know the gas product distribution during reaction and the fluidization behaviour of the carrier. The experimental set-up has been described elsewhere [31].
The tests were carried out at 950°C with an inlet superficial gas velocity into the reactor of 10 cm/s. The composition of the gas was 25 vol.% CH 4 in N 2 during reduction and 10-15 vol.% O 2 in N 2 during oxidation. The reduction periods were varied between 60 and 300 s. The oxidation periods necessary for complete oxidation varied between 600 and 1200 s. To avoid mixing of CH 4 and O 2 , N 2 was introduced for two minutes after each reducing and oxidizing periods. The fluidized bed was fed with 260-300 g of oxygen carrier with a particle size of 0.1-0.3 mm.
The conversion of the oxygen carriers as a function of time during the reduction and oxidation periods was calculated from the gas outlet concentrations by the equations: Oxidation where X i is the conversion of the oxygen carrier, Q in is the molar flow of the gas coming into the reactor, Q out is the molar flow of the gas leaving the reactor, P tot is the total pressure, P i,in is the partial pressure of gas i incoming to the reactor, P i,out is the partial pressure of gas i exiting the reactor, n 0 are the moles of oxygen which can be removed from fully oxidized oxygen carrier, and t is the time. The last terms in Eq. (4) take into account the formation of CO and CO 2 during the oxidation period due to the oxidation of C coming from the decomposition of CH 4 in the reduction period.
The back-mixing in the system, which was illustrated by the transient changes in gas concentration during the first seconds of reaction, was considered in order to obtain the actual concentration of the gases in the bed. The correction was done using a method of deconvolution that takes into account the gas residence time distribution in the system [34]. Fig. 2 shows a schematic diagram of the continuous atmospheric CLC facility used in this work, which was designed and built at Instituto of Carboquimica (ICB-CSIC). The pilot plant was basically composed of two interconnected fluidized-bed reactors-the air and fuel reactors, a riser for solids transport from the air reactor to the fuel reactor, a solid valve to control the solids flow rate fed to the fuel reactor, a loop seal and a cyclone. This design allowed the variation and control of the solid circulation flow rate between both reactors.

ICB-CSIC-g1 facility
In the FR (1) the oxygen carrier particles are reduced by the fuel. Reduced oxygen carrier particles overflowed into the AR (3) through a U-shaped fluidized bed loop seal (2), to avoid gas mixing between fuel and air. The oxidation of the carrier took place at the AR. Secondary air could be introduced at the top of the bubbling bed to help particle entrainment. N 2 and unreacted O 2 left the AR passing through a highefficiency cyclone (5) and a filter (9) before the stack. The oxidized solid particles recovered by the cyclone were sent to a solids reservoir setting the oxygen carrier ready to start a new cycle. In addition, these particles avoid the leakage of gas between the FR and the riser. The regenerated oxygen carrier particles returned to the FR by gravity from the solids reservoir through a solids valve (7) which controlled the solids circulation flow-rate entering the FR. A diverting solids valve (6) located below the cyclone allowed the measurement of the solids flow rates at any time. Fine particles produced by fragmentation/attrition in the plant were recovered in the filters that were placed downstream of the FR and AR. The gas outlet streams of the FR and AR were drawn to respective on-line gas analysers to get continuous data of the gas composition. Detailed information about this experimental facility was described elsewhere [31]. The total solids inventory in the system was about 1.2 kg of solid material. The temperature in the air reactor was always kept constant at about 950 ± 15°C. The inlet flow of fuel was 170 Nl h −1 , which corresponds to an inlet gas velocity in the fuel reactor of 10 cm s −1 . The inlet air flow in the AR was 720 Nl h −1 as primary air, (46 cm s −1 at 900°C), and 150 Nl h −1 as secondary air. Nitrogen was used to fluidize the bottom loop seal (37.5 Nl h −1 ).
The effect of NiO addition on a Fe-based OC was analysed working with a chemically mixed Fe-Ni oxygen carrier (Fe15Ni2-CM), with both metals impregnated on the same alumina particle, and with a physically mixed Fe-Ni OC (Fe15Ni2-PM), adding a small amount of a Ni-based OC to the Fe-based OC bed to obtain a mixed OC with similar NiO content than the chemically mixed one (but maintaining constant the solids inventory in the system). The results obtained with the chemically and physically mixed OCs were compared with those obtained previously in the same facility using an unmixed Fe-based OC (Fe15-WM) [31].
Two different fuel gases were used with all the OCs, methane and a simulated PSA-offgas stream. The PSA-offgas was used in order to know the potential of the materials as oxygen carriers for a steam reforming process coupled with a CLC system (SR-CLC) [31,35]. The gas composition of the PSA-offgas was determined to be 12 vol.% of CH 4 , 18 vol.% of CO, 25 vol.% de H 2 and 45 vol.% of CO 2 taken from the final report of CACHET project (FP VI-019972) [36]. Table 2 shows a summary of the different operating conditions used in the experiments conducted with both fuels. In the test series of experiments carried out with both fuels, the fuel reactor temperature, the fuel flow, the solids circulation rate and the power of the plant varied.
The oxygen carrier-to-fuel ratio (ϕ) was defined by Eq. (6), as: Thus, the oxygen carrier-to-fuel ratio (ϕ) was defined as the ratio between the oxygen supplied and the oxygen needed to stoichiometrically react with the fuel flow. A value of the ϕ ratio equal to unity means that the oxygen supplied by the solids is exactly the stoichiometric oxygen to fully convert the fuel gas to CO 2 and H 2 O.
To analyse the effect of the ϕ ratio, the experiments were carried out varying the flow of the fuel, but maintaining roughly constant the solids circulation flow-rate in each set of experiments. To maintain the total flow of gas entering to the fuel reactor, the corresponding flow of nitrogen was added in every case. Under all operating conditions, the ratio of the constituent gases of the PSA off-gas, i.e. CH 4 , H 2 , CO, CO 2 was maintained constant. When the flow of the fuel was varied, the air to fuel ratio, the solids inventory per MW th (m FR * ) and the gas concentration were varied simultaneously.
To analyse the effect of FR temperature on combustion efficiency the experiments were carried out at two different FR temperatures, 830 and 880°C.
To evaluate the behaviour of the oxygen carrier during the combustion tests, the combustion efficiency η c , defined in Eq. (8), was used as a key parameter. The combustion efficiency (η c ) was defined as the ratio of the oxygen consumed by the gas leaving the FR to that consumed by the gas when the fuel is completely burnt to CO 2 and H 2 O. So, the ratio gives an idea about how the CLC operation is close or far from the full combustion of the fuel, i.e. η c = 100%.
F in being the molar flow of the inlet gas stream, F out the molar flow of the outlet gas stream, and x i the molar fraction of the gas in the inlet or outlet gas stream.

Results and discussion
The behaviour of several chemically mixed Fe-Ni oxygen carriers was analysed in this work together with a physically mixed Fe-Ni OC, using a Fe-based and a Ni-based OC. Table 1 shows the physical and chemical characteristics of the fresh OCs. As it can be seen, the NiO content affects the properties of the materials. The higher the NiO content, the higher the density and mechanical strength of the oxygen carriers prepared by impregnation and the lower the porosity and the superficial area.
XRD patterns of the fresh oxygen carriers showed the phase transformation of the alumina support during the calcination step and the formation of nickel aluminium spinel (NiAl 2 O 4 ) and nickel ferrite (NiFe 2 O 4 ) compounds for the nickel oxide. Moreover, in Fig. 3 SEM images of the fresh chemically mixed Fe-Ni OC particles are shown. The oxygen carrier particles exhibited an irregular shape, since impregnation of porous alumina particles has been carried out. The iron and nickel distributions inside the particles were also analysed by EDX in some particles embedded in resin epoxy, cut, and polished. A uniform distribution of iron and nickel through the particles was found.

Reactivity in TGA
The different samples of chemically mixed OCs were first characterized by TGA, in order to know their reactivity with different fuel gases (CH 4 , H 2 and CO), since methane and PSA-offgas (a mixture of H 2 + CO + CH 4 ) were used as fuel in the continuous unit. The influence of the temperature was also studied.
For Fe-based oxygen carriers, different reduction reactions are possible depending on the reducing gas composition and temperature. For the data presented here, it was assumed that the weight variations observed in the TGA were mainly associated with the transformation Fe 2 O 3 /FeO·Al 2 O 3 . This assumption was confirmed by the results obtained by the XRD patterns, as it can be seen in Table 1, which shows that the final reduced form in the experiments carried out using CH 4 was the form FeO·Al 2 O 3 . Fig. 4 shows the reduction and oxidation reactivities at 950°C using CH 4 as reduction gas and air for oxidation for the chemically mixed OCs, together with the Fe15-WM and the Ni18-WM OCs for comparison purposes. As it can be seen, the effect of Ni addition on the bimetallic OC is negative, as higher amount of Ni in the mixed OC implies lower reactivity of the OC. These results can be explained by the formation of a Fe-Ni spinel compound in the bimetallic OCs with lower methane reactivity than the corresponding to Fe 2 O 3 and NiO [37], as it could be seen in Table 1. XRD patterns show the formation of NiFe 2 O 4 in the oxidation state and FeNi 3 in the reduced stated in all the bimetallic OCS. This fact may lead to a solid conversion higher than 1, since it was assumed that the total iron content is reduced only to FeO·Al 2 O 3 using CH 4 as fuel gas.
The Fe15-WM and the Ni18-WM OCs had been previously confirmed as high oxygen carriers for fuel gases (CH 4 , H 2 and CO), [31,32], as it can be seen in Fig. 4.
In spite of the formation of Fe-Ni compounds, the oxidation reactivities for all the bimetallic OC were high and very similar to the Fe15-WM OC, as it can be seen in Fig. 4.
The effect of the fuel gas was analysed for the bimetallic OCs using H 2 and CO in the TGA. Fig. 5 shows the reduction and oxidation conversions versus time curves obtained using H 2 as reacting gas, and air for the oxidation at 950°C and for the Fe15-WM. As it can be seen, very high reactivities were obtained with all carriers. It must be pointed out that using H 2 different reduction states can be reached depending on the ratios H 2 O/H 2 used [38]. In the TGA conditions used dp (µm) a.u. dp (µm) a.u. in this work, reduction up to Fe 0 could be reached. Comparing with the values obtained with CH 4 as reacting gas, it can be seen that higher reactivity was obtained working with H 2 with all carriers. Fig. 6 shows the reactivity curves for the mixed OCs and for the Fe15-WM, using CO as reacting gas at 950°C. Similar results to CH 4 were obtained for the effect of Ni content on the reactivity: higher NiO content implies lower reactivity of the mixed OC. Comparing Figs. 4, 5 and 6, it can be seen that the highest reactivities were obtained with H 2 as reacting gas, followed by CH 4 . The lowest reactivities were observed working with CO as reacting gas. These results agree with the ones obtained by Abad et al. [38] and Dueso et al. [39], who found the highest reactivities working with H 2 and the lowest with CO using Ni-based oxygen carriers. In all cases, lower reactivities were achieved with those OCs with higher Ni contents.
The effect of temperature on the reactivity of the OCs was evaluated using the different fuel gases at 830 and 950°C. Fig. 7 shows the reactivities obtained with CH 4 as reacting gas at the low temperature for all the bimetallic OCs. As it can be seen comparing Figs. 4 and 7 an important effect of the reduction temperature was observed for the bimetallic OCs, especially as Ni content increases. This effect is due to the formation of Ni based compounds, NiAl 2 O 4 and NiFe 2 O 4 , with a slower reactivity than NiO [40,41].
Thus, the bimetallic OCs used in this work need high operation temperatures, about 950°C, to have reactivities as high as to the unmixed Fe15-WM, specially working with CO, and CH 4 as fuel gases.

Oxygen carrier behaviour in batch fluidized bed
Several reduction-oxidation multicycles were done with the chemically mixed oxygen carriers in the batch fluidized bed reactor, using CH 4 as reducing gas, to determine the gas product distribution and to analyse the fluidization behaviour of the oxygen carriers with respect to agglomeration phenomena. The carriers tested were the Fe15Ni2-CM, Fe15Ni5-CM, and Fe15Ni12-CM. The results obtained were compared with the one achieved working with the Fe-based OC without mixed, Fe15-WM, at similar conditions. Fig. 8 shows the outlet product gas distribution for a typical reduction period of 2 min working with the chemically mixed OCs and with the Fe15-WM at 950°C using CH 4 as fuel in the batch fluidized bed reactor. It was found, in all cases, a first period of full conversion of CH 4 , where CO 2 and H 2 O were formed just immediately after the introduction of the reducing gas to the reactor. After this first period, CO 2 and H 2 O concentrations begin to decrease as a result of the decrease of the oxygen transference rate. Because of this, the CH 4 , CO and H 2 concentrations start to increase because to the oxygen carrier is unavailable to convert fully CH 4 into CO 2 and H 2 O, and the partial oxidation of methane takes place. In this period, it can be observed that as NiO content in the bimetallic OC increases, the CH 4 concentration at the outlet of the reactor decreases, due to the capacity of the Ni active sites to catalyze the reforming reaction of methane. Thus, the selectivity of methane reaction to CO 2 + H 2 O increased as the nickel content in the oxygen carrier increases. Small amounts of nickel oxide (b5 wt.%) in the OC are sufficient to fully convert all the methane inlet flow. Similar results were found during the whole batch experiments indicating that the OC maintains its reactivity during cycling operation.
The multi-cycle tests carried out in the batch fluidized bed reactor were useful to determine the fluidization behaviour of the oxygen carrier with respect to the agglomeration phenomena. Although a high degree of conversion was reached in the cycling tests, neither of the bimetallic OCs show any agglomeration behaviour during operation. These results agreed with previous works carried out in continuous operation using the unmixed OC Fe15-WM, and other Fe-based materials [7,31,42] where the absence of agglomeration in the use of Fe-based materials for the CLC process was establish.
Once the reactivity and selectivity to CH 4 combustion of the different bimetallic OCs have been determined in the TGA and in the batch fluidized bed a few conclusions can be drawn. By one hand, the higher Ni content in the chemically mixed FeNi OCs, the lower CH 4 , and CO reactivity. On the other hand, the selectivity to full conversion of methane increases as Ni content increases due to the catalytic effect of the reduced nickel. According to this, the chemically mixed FeNi OC, Fe15Ni2-CM, was selected to evaluate its behaviour in the continuous unit, since it has shown a high reactivity with hydrogen, methane, and CO, at 950°C and full conversion of CH 4 was reached in the fluidized bed facility during reduction periods.

Test in ICB-CSIC-g1 facility
The effect of Ni addition on the combustion efficiency was analysed in the 500 W th CLC continuous unit using the unmixed Fe-OC, the chemically mixed Fe-Ni OC and the physically mixed Fe-Ni OC. A first set of experiments were conducted working with a chemically mixed OC, with the iron and nickel particles impregnated over the same alumina particle. A second set of experiments were performed working with a physical mixture of two impregnated Fe-and Ni-based OCs. Moreover, these results were compared with those obtained in the same facility using an unmixed Fe-based OC, Fe15-WM, described in detail in a previous work [31].
As it was commented above, the Fe15Ni2-CM OC was selected for the continuous CLC facility experiments. A total of about 38 h at hot conditions, of which 32 corresponded to combustion conditions were conducted in the facility with the bimetallic oxygen carrier. Experiments were performed with CH 4 and a simulated PSA-offgas as gas fuels for comparison purposes. The effect of the oxygen carrierto-fuel ratio and the fuel reactor temperature on the combustion efficiency, η c , was analysed.
The gas product concentrations of the fuel and air reactors were measured by on line analysers. These gas concentrations were used to make carbon, hydrogen and oxygen mass balances over the whole reactor system. For better comparison, the results are presented in N 2 free dry basis. CO and CO 2 concentrations at the outlet of the AR were never detected in any test. Thus, no losses in CO 2 capture were produced by carbon transfer to the AR, reaching 100% CO 2 capture in the process. Fig. 9a) shows the effect of ϕ on the combustion efficiency using the Fe15Ni2-CM as oxygen carrier, working with CH 4 as fuel at 830°C and at 880°C. As it can be observed full combustion of the fuel was reached at ϕ values higher than 4 working at 830°C. An important effect of the temperature can be seen also, especially at low ϕ values. Fig. 10a) shows the gas product distribution measured at the outlet of the FR in these tests. High CH 4 and low CO and H 2 concentrations were measured at low ϕ values for both temperatures, 830 and 880°C, when CH 4 was used as fuel, indicating a negligible catalytic effect of Ni present in the Fe15Ni2-CM OC. An increase in the oxygen carrier to fuel ratio produced an increase in the conversion of methane as more oxygen is available for combustion of the fuel.

Effect of the fuel composition
A simulated PSA-offgas stream (a mixture of CH 4 , CO, and H 2 ) was also used as fuel in the CLC continuous unit in order to prove the feasibility of the Fe15Ni2-CM oxygen carrier for a SR-CLC process. Fig. 9b) shows the effect of ϕ on the combustion efficiency using the Fe15Ni2-CM as oxygen carrier, working with PSA-offgas as fuel at 830°C and at 880°C. Full combustion of the fuel working at 880°C can be reached at ϕ values higher than 4. An important effect of the fuel reactor temperature on the combustion efficiency was found, as CO is present in the PSA-offgas and this OC had a slow reactivity with this gas, especially at low temperatures. Fig. 10b) shows the gas product distribution measured at the outlet of the FR in these tests. As it can be seen, at 830°C, similar low concentrations of CH 4 , CO and H 2 were obtained. At higher temperature, 880°C, the CH 4 concentration decrease, confirming the high effect of the temperature on the reactivity of the mixed OC.
The effect of the fuel composition on the combustion efficiency with the Fe15Ni2-CM can be analysed in Fig. 9. Lower combustion efficiencies were obtained using the PSA-offgas as fuel compared to CH 4 as fuel, especially at the lower operation temperature. These results were due to the low reactivity of the Fe15Ni2-CM OC with the CO, one of the gases contained in the PSA tail gas, as it could be seen in experiments conducted in TGA.

Effect of Ni addition on a Fe-based OC
In order to analyse the effect of Ni addition on the behaviour of Fe-based OC, a second set of experiments were performed working with a physical mixture of two impregnated Fe-and Ni-based OCs, Fe15Ni2-PM. The results were compared with those obtained in the first set of experiments, the chemically mixed Fe15Ni2-CM OC, and also with previous obtained with unmixed Fe-based OC, Fe15-WM [31]. It was added the necessary amount of Ni18-WM to obtain a mixed OC with a similar percentage of Ni than the chemically mixed OC. A total of about 56 h at hot conditions, of which 50 corresponded to combustion conditions were carried out in the facility using the Fe15Ni2-PM oxygen carrier. Fig. 11(a) and (b) shows a comparison of the combustion efficiency as a function of the ϕ value using the three different OCs, i.e. Fe15Ni2-PM, Fe15Ni2-CM and Fe15WM, using CH 4 as fuel at 830 and 880°C, respectively. As it could be seen no substantial improvement of the combustion efficiency was reached working with the chemically mixed bimetallic oxygen carrier, Fe15Ni2-CM, in respect to the results obtained with the Fe-based oxygen carrier without mixing, Fe15-WM. It can be seen that higher combustion efficiencies were obtained working at 830°C with the Fe15Ni2-PM in respect to the ones obtained with the Fe15-WM. Small differences between the combustion efficiencies values were observed at 880°C. Fig. 12 shows the effect of the oxygen carrier to fuel ratio on gas product concentration at the exit of the FR, using CH 4 as fuel, at 830°C, working with Fe15Ni2-PM. As it can be observed lower values of CH 4 and higher H 2 and CO concentrations were detected at the outlet of FR with the physically mixed material in respect to the ones obtained with the Fe15Ni2-CM, confirming the effect of NiO on the methane reforming when Ni is not bounded in the same particle.  10. Effect of oxygen carrier to fuel ratio on the composition of outlet gas for the Fe15Ni2-CM, using: a) CH 4 and b) PSA-offgas as fuels, at 830°C (empty dots) and 880°C (filled dots). Fig. 13 shows a comparison of the combustion efficiency as a function of ϕ using the three different materials (Fe15-WM, Fe15Ni2-CM and Fe15Ni2-PM) as oxygen carrier, working with PSA-offgas as fuel, at 830°C and at 880°C, respectively. Comparing with the results obtained using CH 4 as fuel it can be seen that higher combustion efficiency can be obtained working with CH 4 as fuel, due to the catalytic effect of Ni and the low reactivity of Ni with CO. It can be observed the similar combustion efficiencies for the physically mixed carrier and the Fe-based OC without mixed. Full combustion of the fuel was reached with these OCs at ϕ values of 2.5 at 880°C. It can be also observed that the lowest values of the combustion efficiency were achieved working with the bimetallic oxygen carrier Fe15Ni2-CM at both temperatures.

Discussion
According to the results obtained with the Fe15Ni2-CM using CH 4 or PSA-offgas as fuels, it could be said that worse results were obtained at both temperatures with the chemically mixed OC. Therefore, no improvement of the combustion efficiency was detected working with the chemically mixed OC in respect to the results obtained with a Fe-based OC without mixed. The formation of mixed compounds like the awaruite (FeNi 3 ) or the trevorite (NiFe 2 O 4 ) which could hinder the catalytic effect of Ni could be the cause of the poor results achieved with the Fe15Ni2-CM OC, as it could be seen in experiments conducted in TGA, where it was determined that the higher Ni content in the oxygen carrier the lower the reactivity of the carrier.
According to the results obtained with the physically mixed OC a positive effect was only measured when methane was used as fuel at low temperatures due to a Ni catalytic effect. Similar combustion efficiencies were found to respect with the results obtained with the Fe-based OC without mixed working with PSA-offgas.
The use of a specific OC has important implications for a CLC system. The reactivity of the solids determines the solids inventory in the system and the operating conditions needed. Table 3 shows the required values of ϕ and solid inventories in the FR, m FR * , to reach full combustion of the fuel, working with the different bimetallic OCs tested in the continuous 500 W th unit at 830 and 880°C, using CH 4 and PSA-offgas as fuels.
As it can be observed, using CH 4 as fuel, the highest solid inventory needed was obtained for the chemically mixed OC. A very low solid inventory (600 kg/MW) is required with the physically mixed OC at 880°C.
Using PSA-offgas as fuel, higher solid inventories are needed with the chemically mixed OC, due to the formation of Ni based compounds, NiFe 2 O 4 , and NiAl 2 O 4 , with a slower reactivity.
The considerable improvement of the combustion efficiency achieved working with the physically mixed OC at low temperature, and the limited improvement obtained working with the chemically mixed OC at any temperature, can be explained by the fact that in the physical mixture of Fe-OC and Ni-OC, the Ni-OC particles not only act as a catalyst but also as an oxygen carrier, providing bulk oxygen for the reactions with methane. The highest amount of oxygen transported, the lowest solid inventory is required to reach full combustion of the fuel.

Conclusions
The effect of Ni addition over a Fe-based OC has been evaluated and compared using two different strategies: the chemical and the  physical mixtures. Thus, different bimetallic OCs have been prepared with the iron and nickel oxides impregnated over the same alumina support particle. Moreover, a physically mixed OC, using a Fe-based and a Ni-based OCs, was also evaluated in the combustion of fuel gases. The effect of this Ni addition on the combustion efficiency was analysed in a 500 W th CLC continuous unit using methane and a PSA-offgas as fuel gases at different operating conditions. It was found that the use of a chemically mixed Fe-Ni OC had a negative effect on the combustion efficiency compared to the results obtained with an unmixed Fe-based OC since the formation of Fe-Ni compounds reduced the catalytic effect of Ni addition.
However, the addition of 2 wt.% of NiO by a physical mixture of an impregnated OC can improve significantly the combustion efficiency of methane at low temperatures.
A considerable reduction of the solid inventories needed to reach full CH 4 or PSA-offgas combustion was measured working with the Fe15Ni2-PM OC. In conclusion, in order to improve the behaviour of a Fe-based OC via NiO addition is preferable to mix physically a Ni-based OC and a Fe-based OC, with each metal supported on different particles, instead of the chemical addition by impregnation of NiO on Fe-based OC particles, supporting both metals on the same particle.