Reduction mechanism of iron titanium based oxygen carriers with H₂ for chemical looping applications – a combined experimental and theoretical study

Abstract

The reduction mechanisms of three iron titanium based oxygen carriers with H₂ have been investigated by experimental and theoretical methods for the chemical looping process. The oxygen carriers have been successfully prepared and the initial rate for the reduction reactions has been predicted as a function of initial molar concentration of Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ at different temperatures from 800 to 1000 °C. Density functional theory calculations have been used to explain the reduction mechanism of oxygen carriers with H₂. Our results show that the reduction of Fe₂TiO₅ with H₂ has the smallest activation energy compared with the others. The observed rate constant values indicate that the reduction mechanism of Fe₂TiO₅ is much faster than that of the others. Also, we found that the diffusion of a H atom from Fe is the rate determining step for the reduction of the three iron titanium oxygen carriers. Additionally, the experimentally predicted activation energies are in good agreement with the theoretical values.

---

1. Introduction

The chemical looping process (CLP) has received increasing research interest as a promising technology for increasing combustion efficiency of fossil fuels because of its environmental friendliness.^1–7 CLP not only increases the thermal efficiency in power generation stations, but also exhibits inherent advantages of CO₂ separation without need of any major external energy.⁸ A typical chemical looping system contains a fuel and air reactors for repeated reduction and oxidation of oxygen carriers and the final products are just CO₂ and water vapor. The overall efficiency of CLP is not only dependent on the reactor design but also on the performance of oxygen carriers, which is most important for large-scale application. The development of a proper oxygen carrier is considered as a challenging task for successful operation of the CLP system.

In general, a good oxygen carrier must have high reaction rate, high oxygen carrying capacity, great mechanical strength, long-term recyclability and the ability to fully convert the fuel. In recent years, several efforts have been made to develop efficient oxygen carriers based on the oxides of transition metals (CuO,^9–12 Fe₂O₃,^13–15 NiO,^16,17 Mn₃O₄ ^18,19 and CoO²⁰) because of their good reduction/oxidation properties. However, the previous studies on development of oxygen carriers showed that the reactivity and recyclability of these pure metal oxides are greatly reduced during the chemical looping operation due to their lack in the sintering resistance and mechanical strength.²¹ Adding inert support materials such as SiO₂, Al₂O₃, TiO₂, MgO, ZrO₂ with pure metal oxides have been evidenced as an effective way to enhance their reactivity and recyclability of oxygen carriers.

Further, among the others, TiO₂ has used as the most commercial inert support which can enhance the reactivity of oxygen carriers by means of increasing porosity, surface area and mechanical strength.²² Considerable attention has been given to natural minerals such as ilmenite as an efficient oxygen carrier due to its low cost. Moreover, in recent years, the possibility of application of ilmenite as an oxygen carrier with various fuels has been widely investigated in different studies. Previously, Burton et al.²³ predicted the equilibrium phase transition of ilmenite from hematite using theoretical approximation model. Recently, Ku et al.,²² reported that the reduction mechanism of Fe₂TiO₅ to Fe₂TiO₄, FeTiO₃ as oxygen carriers in CLP. Even though studies have been performed on the iron titanium oxides as oxygen carrier in CLP, the detailed reaction mechanism between iron titanium oxides and syngas components still remains unknown. Therefore, a detailed understanding of reaction kinetic study of iron titanium oxide with syngas components as a possible fuel is essential in order to justify the design of oxygen carriers.

In the present study, the reduction mechanism of iron titanium oxides with one of the primary syngas component H₂, has been investigated by both experimental and theoretical methods. The kinetic studies (reaction rate, rate constant, accurate activation energy) of Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ are carried out by thermogravimetric analyzer (TGA). The detailed reduction mechanism of three iron titanium oxides surfaces (Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃) was theoretically explored using density functional theory (DFT) methods. The results from the present experimental and theoretical calculations can help to usage of Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ as oxygen carriers in CLP system related to combustion of syngas.

2. Experimental and theoretical methods

2.1. Oxygen carriers preparation

We prepared the oxygen carrier Fe₂TiO₅ using the previously reported procedure with slight modification.²⁴ To synthesize the oxygen carrier Fe₂TiO₅, a 2 : 1 mole ratio of Fe₂O₃ (Sigma-Aldrich, 99%) and TiO₂ (Aldrich, 99%) powder was mixed well and then ultrapure water (>18.0 MW cm⁻¹) was slowly added into the mixture to form a paste with suitable viscosity. The mixture was wet milled and homogenized in a planetary ball miller (PM100, Retsch) for 10 h with a speed of 380 rpm. Then, the paste was dried at 80 °C for 24 h in an oven, and then sintered at desired temperatures (from 600 to 1075 °C, see Table S1 of ESI†) in a muffle furnace (Linderg BLUE/M UP550) with a heating rate of 5 °C min⁻¹ under air atmosphere for 2 h, resulting in the formation of pure Fe₂TiO₅ particle. Fe₂TiO₄ and FeTiO₃ particles were also prepared by the same synthesis steps but with heating under N₂ atmosphere and with different mole ratio of TiO₂ : FeO (Aldrich, 99.9%).²⁴ TiO₂ and FeO were taken in 1 : 2 and 1 : 1 mole ratio with respect to the chemical equations and the reaction conditions attempted in Table S2 and S3,† respectively, which resulted in the formation of pure Fe₂TiO₄ and FeTiO₃ structures. All prepared oxygen carriers were pulverized and double sieved to get a size range distribution of 153–293 μm.

Crystal phases of oxygen carriers were determined by X-ray diffraction (XRD, D2 PHASER Bruker). Incident radiation was generated using a Cu Kα line (λ = 1.5406 Å) at 30 kV and 10 mA. The samples were examined at range of 2θ from 10 to 80° with a sampling interval of 0.02° and scan rate of 0.05° min⁻¹. TOPAS 4.2 software from Bruker was used to refine the structures and to obtain the lattice parameters. The particle size distribution of prepared oxygen carriers after sieve and heat treatment was measured by laser scattering particle size distribution analyzer (LA-960, HORIBA). The sieved oxygen carrier particles were divided into 0.21–1.98 mmol (approximately 50–300 mg) batches in a hanging basket type of net-shaped platinum crucible (12 mm in both diameter and high) to study the reaction rates by a thermogravimetric analyzer (TGA STA 449 F3, NETZSCH). The experimental results from TGA were used to determine the rates of the reduction reaction of synthesized oxygen carriers with H₂ at different temperatures, while 20.4% of H₂ (balanced in N₂) was used for all the kinetic experiments. To optimize the gas flow rate, we initially tested the gas velocity by varying different flow rates for the reduction of oxygen carriers at 800 °C and found that a minimum gas flow rate of 200 mL min⁻¹ is sufficient to eliminate the external mass transfer resistance. Therefore, we used an optimum gas flow rate of 230 mL min⁻¹ for all TGA experiments. Since a high gas velocity was maintained with low solid sample weight, the TGA essentially acted as a differential bed reactor. In order to understand the thermodynamic properties of oxygen carriers, the differential thermal analysis (c-DTA) curve was obtained using 1.98 mmol portion of Fe₂TiO₅ at 800 °C. The thermal analysis was performed using Proteus 6.0.0 software from NETZSCH.

2.2. Computational details

All the DFT calculations were carried out using Vienna ab initio simulation package (VASP) and the projector augmented wave (PAW) method was used to describe the electron–ion interaction.^25–27 The exchange–correlation functional of GGA-PW91 approximation²⁸ was used for all the calculation. Bulk phase Fe₂TiO₅ was represented by a 2 × 1 × 1 supercell and both Fe₂TiO₄ and FeTiO₃ bulk phases were represented by a 1 × 1 × 1 supercell. The calculated lattice parameters for the bulk structures are summarized in the Table S4,† which are in agreement with the available experimental and previous theoretical values. Further, (0 2 2) surface of Fe₂TiO₅, (3 1 1) surface of Fe₂TiO₄ and (1 0 4) surface were cleaved from the respective bulk structures, since they are the dominant growth facets and then we considered the most stable surfaces based on their surface energies for the surface reduction mechanism. A 12 Å of vacuum layer was used to avoid the interactions between the adjacent slabs. The plane-wave cutoff energy of 450 eV was used and the Brillouin zones were sampled with a Monkhorst–Pack meshes of 4 × 2 × 1, 3 × 4 × 1 and 4 × 6 × 1 for Fe₂TiO₅ (0 2 2), Fe₂TiO₄ (3 1 1) and FeTiO₃ (1 0 4) surfaces, respectively. Spin polarized calculations have been performed in all cases. A conjugate-gradient algorithm was used to relax the ions until the forces on unconstrained atoms were less than 0.01 eV Å⁻¹. The overall energies were converged to below 5 meV per atom. The transition states (TSs) for H diffusion were identified using climbing image nudged elastic band (CI-NEB) method.²⁹ Further, the vibrational frequency calculations have been performed to validate the transition states of H diffusion.

3. Results and discussion

3.1. Characterization of iron titanium oxides oxygen carriers

The crystal phases of prepared samples were first investigated by XRD and Fig. 1 shows the typical XRD patterns of the prepared iron titanium oxide samples (Fe₂TiO₅, Fe₂TiO₄ or FeTiO₃). The experimental XRD patterns are compared with simulated one using Rietveld refinement.³⁰ It can be seen from Fig. 1 that all the diffraction peaks of three iron titanium oxides are well indexed with Joint Committee on Powder Diffraction Standards (JCPDS, 76-1158, 75-1380 and 79-1255), revealing that they possess orthorhombic, face centered cubic and rhombohedral phases, respectively. There are no diffraction peaks from any other impurities, indicating the high purity of the prepared samples. The relative intensities of observed XRD patterns of Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ are agreed well with simulated Rietveld results (see Fig. S1–S3†). As shown in Fig. 1, 2θ at 25.48°, 34.84° and 32.51° are indexed as (0 2 2), (3 1 1) and (1 0 4) for Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃, respectively. The observed XRD peaks for Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ are sharp due to their large crystallite and particle size. Further, in order to quantify the possible effect of particle size, all the prepared oxygen carriers were suspended in DI water and analyzed by a laser scattering particle size distribution analyzer. Fig. 2 shows the particle size distribution of the prepared oxygen carriers after the sieved and heat treatment. The results indicated that the particle size distribution of all the prepared oxygen carriers was very similar. The particle size distribution D₅₀ of Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ were 210.4, 227.7 and 207.7 μm, respectively.

Fig. 1 XRD profiles of FeTiO₃, Fe₂TiO₄ and Fe₂TiO₅. The upper symbols illustrate the observed data (black dot) and calculated pattern (blue line).

Fig. 2 Sieved particle size distribution of FeTiO₃, Fe₂TiO₄ and Fe₂TiO₅ oxygen carriers.

3.2. Kinetic analysis for Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ oxygen carriers

The reduction kinetics of iron titanium oxides, Fe_xTiO_y−1 (if x = 1, y = 3 and if x = 2, y = 4 and 5), both forward and reverse reactions have considered as follows;where k_obs and k^′_obs are the observed rate constants of both forward and reverse reactions, respectively. In the reduction reaction, the oxygen carriers Fe_xTiO_y reacts with H₂ gas and produced Fe_xTiO_y−1, and H₂O. At sufficiently a high gas velocity of H₂, the overall reduction rate of iron titanium oxides particle is limited solely by the effects of chemical reaction, and the H₂ gas is excess in the ambient. Under the parameters given in experimental method, excess concentration of H₂ gas can eliminate the external mass transfer resistance when gas flow rates were higher than 230 mL min⁻¹, thus the reversed reaction can be neglected. In such cases, the concentration of H₂ is assumed as a constant and is merged to observed rate constants during conversion from Fe_xTiO_y to Fe_xTiO_y−1. Hence, the overall reduction rate is given by,where N is the mole of releasable oxygen in different stage from oxygen carriers. According to eqn (2), the k_obs(T) can be regressed based on the rate of weight loss during the reaction at specific temperature. Then, the Arrhenius plot is used to obtain the activation energy as described in eqn (3).

k_obs(T) = A_FexTiOye^(−E_a/RT) (3)

where A is the collision frequency or pre-exponential factor, E_a is the activation energy for the reduction of oxygen carriers and R is a gas constant.

Generally, reaction temperature should be high enough to provide the heat energy for the reduction reaction. Thus, effect of reaction temperature in the reactor caused by the heat of reaction, either endothermic or exothermic, can be moderated by the selected oxygen carrier without resorting. For this reason, the experiments of kinetic analysis have been conducted by TGA from 800 to 1000 °C. The initial rates for the reduction reactions were predicted as a function of initial molar concentration of Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ at different temperatures using the eqn (2) in the time interval of 0–1 min and are shown in the Fig. S4 of ESI.† We found that the plots for initial rate as a function of the initial molar concentration of Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃ oxygen carriers are linear. The predicted values of rate constants and reaction orders for the reduction of iron titanium oxygen carriers are listed in Table S5 of ESI.† As can be seen from the Table S5,† k_obs values of Fe₂TiO₅ are in the range of 8.36 × 10⁻³ to 1.84 × 10⁻²; k_obs values of Fe₂TiO₄ are in the range of 6.16 × 10⁻⁴ to 2.66 × 10⁻³; k_obs values of FeTiO₃ are in the range of 3.03 × 10⁻⁵ to 2.04 × 10⁻⁴. Also, our results show that the reduction of Fe₂TiO₅ occurred much faster than the others.

Further, the activation energies for the reduction of three iron titanium oxides with H₂ molecule are predicted by applying the linear regression of the slopes in Fig. 3 and the values are summarized in Table 1. The predicted activation energies are 44.39, 83.52, and 108.73 kJ mol⁻¹ for Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃, respectively. It has been observed that the activation energy for the reduction of FeTiO₃ is higher than that of Fe₂TiO₅ and Fe₂TiO₄, which indicate that the reduction of FeTiO₃ with H₂ is difficult. The predicted activation energy for FeTiO₃ is slightly lower than the previous study of Abad et al.³¹ and higher than the other iron based oxygen carriers.^32–34 The differential thermal analysis (DTA) was used to investigate the thermodynamic properties of the Fe₂TiO₅ oxygen carrier and the observed DTA is shown in Fig. 4. DTA pattern shows three peaks; first one is the exothermic peak which is associated with the weight loss due to the reduction of Fe₂TiO₅. The second and third peaks are endothermic, which are associated with the reduction of Fe₂TiO₄ and FeTiO₃, respectively.

Fig. 3 Arrhenius plots for the reduction of (a) Fe₂TiO₅, (b) Fe₂TiO₄ and (c) FeTiO₃ oxygen carriers with H₂ at various reaction temperatures.

Table 1 The experimentally predicted and theoretically calculated activation energies (E_a in kJ mol⁻¹) for the reduction of three iron titanium oxides with H₂ molecule^a

Structure	Ea
	Expt.	Theory
a A and B are Fe–H bond breaking and H atom diffusion.
Fe2TiO5	44.39	A	21.71
		B	45.01
Fe2TiO4	83.52	A	48.62
		B	87.35
FeTiO3	108.73	A	49.13
		B	122.23

---

Fig. 4 Reactivity curves by DTA at 800 °C using 20.4% of H₂ as fuel. (Continuous lines correspond to results predicted the thermodynamic behavior by Proteus 6.0.0 software from Fe₂TiO₅ reduction.)

3.3. H₂ adsorption and dissociation on iron titanium oxides surfaces

As a starting point, the adsorption behavior of H₂ on three titanium supported iron oxides was investigated. Here we considered nine possible adsorption modes of H₂ on the three surfaces, which includes top site of Fe atom, top site of O atom and bridge site between Fe and O atoms. The considered adsorption configurations are schematically illustrated in the Fig. S5–S7 of ESI.† We found that the H₂ molecule strongly adsorbs on the bridge site of all the three surfaces (Fe₂TiO₅ (0 2 2), Fe₂TiO₄ (3 1 1) and FeTiO₃ (1 0 4)), which is in agreement with the previous study of H adsorption on Fe₃O₄ surface.³⁵ We found that after the adsorption of H₂ on the surfaces, the H–H bond length is elongated to 0.93 Å from 0.74 Å. The calculated bond lengths of H–H, H–Fe and H–O are listed in the Table S6 in the ESI.†

3.4. Reduction of iron titanium oxides

The most stable H₂ adsorption sites were considered as an initial point to explore the possible reduction pathways for the three iron titanium surfaces. The complete potential energy profiles for the reduction reactions of Fe₂TiO₅ (0 2 2), Fe₂TiO₄ (3 1 1) and FeTiO₃ (1 0 4) surfaces with H₂ molecule and the initial states (IS), transition states (TS), intermediates (IM), and final states (FS) of the reduction reactions are shown in Fig. 5–7, respectively. In the Fe₂TiO₅ surface, H₂ molecule initially chemisorbed on the bridge site of Fe–O with elongated H–H bond length of 0.93 Å. The calculated activation barrier for H–H bond breaking is 21.23 kJ mol⁻¹ with an imaginary frequency of 831i cm⁻¹ in the TS₁. Subsequently, the H–H bond of H₂ molecule further stretched to 1.991 Å and stably adsorbed on the Fe and O atoms, resulting in the Fe–H and O–H bond formation in IM. Furthermore, the H atom of Fe–H bond stretched (from 1.49 to 1.52 Å) and diffused quite near to O–H bond in TS₂. The calculated activation barrier for this TS₂ is 45.01 kJ mol⁻¹. In the second step, H₂O formed on the Fe₂TiO₅ (0 2 2) surface and this reaction energy is exothermic by 213 kJ mol⁻¹. We found that the rate-determining step (RDS) for the formation of H₂O molecule from the adsorbed H₂, is the H atom diffusion from Fe–H bond. In the IS of Fe₂TiO₄ and FeTiO₃ surfaces, H₂ stably chemisorbed on the bridge site of Fe–O and the H–H distance is found to be 0.823 and 0.754 Å, respectively. The calculated activation barriers for H–H bond breaking in Fe₂TiO₄ and FeTiO₃ surfaces are 48.25 and 49.22 kJ mol⁻¹, respectively, which is comparatively higher than that of Fe₂TiO₅. After bond breaking, H atoms are stably adsorbed on the Fe and O atoms in the IM. Further, the Fe–H bond stretched and diffused quite near to O–H bond in TS₂ and produced the Fe₂TiO_4−x and FeTiO_3−x with H₂O molecule. The activation barriers for reduction of Fe₂TiO₄ and FeTiO₃ surfaces are found to be 87.35 and 122.23 kJ mol⁻¹, respectively, which are endothermic by 74.51 and 95.01 kJ mol⁻¹, respectively. From the above theoretical results, we found that the reduction reaction of Fe₂TiO₅ with H₂ molecule is exothermic, where as for Fe₂TiO₄ and FeTiO₃ is endothermic. This is in agreement with our differential thermal analysis (c-DTA) curve of iron titanium oxides (Fig. 4).

Fig. 5 The potential energy profile for the reduction of Fe₂TiO₅ (0 2 2) with H₂ molecule. The given values are the energy barriers (in kJ mol⁻¹).

Fig. 6 The potential energy profile for the reduction of Fe₂TiO₄ (3 1 1) with H₂ molecule (the given values are the energy barriers (in kJ mol⁻¹)).

Fig. 7 The potential energy profile for the reduction of FeTiO₃ (1 0 4) with H₂ molecule (the given values are the energy barriers (in kJ mol⁻¹)).

3.5. Electronic structure

In Fig. S8,† we have illustrated the total density of states (DOS) of H₂ on three designed oxygen carriers for three cases; when H₂ adsorbed on the surfaces in IS, when adsorbed H₂ dissociates into H atoms in IM and when H₂O formed on the surfaces in FS, respectively. From this figure, we observed that the intensity of peak at Fermi level for Fe₂TiO₅ is significantly higher than the others, which implies that the activity of Fe₂TiO₅ oxygen carrier is higher compared to others. We believe that this strong interaction is responsible for the faster reduction of Fe₂TiO₅ oxygen carriers than the others. We have also plotted the projected density of states for H atoms after adsorption for the above three cases and are shown in Fig. S9 of ESI.† The overlap between H atoms and Fe₂TiO₅ surface at just below the Fermi level (Fig. S9b†) validates the strong hybridization of Fe₂TiO₅ with H₂ in the intermediate state, which is the rate determining step for the overall reduction mechanism.

4. Conclusions

The combined experimental and theoretical calculations were performed to investigate the stepwise reduction mechanism of three iron titanium oxides with H₂ for the application of oxygen carriers in CLP. Kinetic studies were carried out at various initial concentration and reaction temperatures using TGA, indicating that the apparent activation energy for the reduction of Fe₂TiO₅ is smaller than the Fe₂TiO₄ and FeTiO₃. The observed rate constant values show that reduction of Fe₂TiO₅ occurs much faster than that of Fe₂TiO₄ and FeTiO₃. DFT results indicate that H diffusion from Fe–H bond in all the three iron titanium oxide surfaces is the rate determining step and the calculated energy barriers for these RDS are 45.01, 87.35 and 122.23 kJ mol⁻¹ for Fe₂TiO₅, Fe₂TiO₄ and FeTiO₃, respectively. Also, our theoretical results show that the reduction reaction of Fe₂TiO₅ with H₂ molecule is exothermic, whereas for Fe₂TiO₄ and FeTiO₃ is endothermic. The calculated energetics for the reduction of three iron titanium based oxygen carriers with H₂ are in agreement with predicted experimental values.

Acknowledgements

We acknowledge the financial support from the Ministry of Science and Technology, Taiwan (MOST 103-2221-E-011-002-MY3 and MOST 104-2113-M-011-002). We are also thankful to the National Center of High-Performance Computing (NCHC) for donating computer time and facilities.

References

1. J. Adanez, A. Abad, F. Garcia-Labiano, P. Gayan and L. F. De Diego, Prog. Energy Combust. Sci., 2012, 38, 215–282.
2. E. Jerndal, T. Mattisson and A. Lyngfelt, Energy Fuels, 2009, 23, 665–676.
3. A. Lyngfelt, Appl. Energy, 2014, 113, 1869–1873.
4. M. M. Hossain and H. I. de Lasa, Chem. Eng. Sci., 2008, 63, 4433–4451.
5. H. J. Richter and K. F. Knoche, in Efficiency and Costing, American Chemical Society, 1983, vol. 235, ch. 3, pp. 71–85.
6. A. A. Olajire, Energy, 2010, 35, 2610–2628.
7. H. Yang, Z. Xu, M. Fan, R. Gupta, R. B. Slimane, A. E. Bland and I. Wright, J. Environ. Sci., 2008, 20, 14–27.
8. M. Ishida, D. Zheng and T. Akehata, Energy, 1987, 12, 147–154.
9. I. Adánez-Rubio, M. Arjmand, H. Leion, P. Gayán, A. Abad, T. Mattisson and A. Lyngfelt, Energy Fuels, 2013, 27, 3918–3927.
10. S. Z. Wang, G. X. Wang, F. Jiang, M. Luo and H. Y. Li, Energy Environ. Sci., 2010, 3, 1353–1360.
11. Y. Zhang, F. Liu, L. Y. Chen, C. L. Han, L. R. Richburg, J. K. Neathery and K. L. Liu, Energy Fuels, 2013, 27, 5341–5351.
12. A. H. Sahir, J. S. Lighty and H. Y. Sohn, Ind. Eng. Chem. Res., 2011, 50, 13330–13339.
13. F. Lind, N. Berguerand, M. Seemann and H. Thunman, Energy Fuels, 2013, 27, 997–1007.
14. L. Liu and M. R. Zachariah, Energy Fuels, 2013, 27, 4977–4983.
15. S. C. Bayham, H. R. Kim, D. W. Wang, A. Tong, L. Zeng, O. McGiveron, M. V. Kathe, E. Chung, W. Wang, A. N. Wang, A. Majumder and L. S. Fan, Energy Fuels, 2013, 27, 1347–1356.
16. S. Bhavsar and G. Veser, Energy Fuels, 2013, 27, 2073–2084.
17. Q. Zafar, A. Abad, T. Mattisson and B. Gevert, Energy Fuels, 2007, 21, 610–618.
18. P. Hallberg, D. Z. Jing, M. Ryden, T. Mattisson and A. Lyngfelt, Energy Fuels, 2013, 27, 1473–1481.
19. Q. Zafar, T. Mattisson and B. Gevert, Energy Fuels, 2006, 20, 34–44.
20. T. Mattisson, A. Jardnas and A. Lyngfelt, Energy Fuels, 2003, 17, 643–651.
21. L. S. Fan, Chemical Looping Systems for Fossil Energy Conversions, Wiley, New York, 2010.
22. Y. Ku, Y. C. Liu, P. C. Chiu, Y. L. Kuo and Y. H. Tseng, Ceram. Int., 2014, 40, 4599–4605.
23. B. P. Burton, Am. Mineral., 1985, 70, 1027–1035.
24. S. Tan, T. Oers, M. K. Aydinol, T. Ozturk and I. Karakaya, J. Alloys Compd., 2009, 475, 368–372.
25. G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
26. G. Kresse and J. Furthmuller, Comput. Mater. Sci., 1996, 6, 15–50.
27. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 48, 13115–13118.
28. J. P. Perdew and Y. S. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 13244–13249.
29. G. Henkelman and H. Jonsson, J. Chem. Phys., 2000, 113, 9978–9985.
30. H. Rietveld, J. Appl. Crystallogr., 1969, 2, 65–71.
31. A. Abad, J. Adanez, A. Cuadrat, F. Garcia-Labiano, P. Gayan and L. F. de Diego, Chem. Eng. Sci., 2011, 66, 689–702.
32. A. Abad, J. Adanez, F. Garcia-Labiano, L. F. de Diego, P. Gayan and J. Celaya, Chem. Eng. Sci., 2007, 62, 533–549.
33. A. Abad, F. Garcia-Labiano, L. F. de Diego, P. Gayan and J. Adanez, Energy Fuels, 2007, 21, 1843–1853.
34. S. R. Son and S. D. Kim, Ind. Eng. Chem. Res., 2006, 45, 2689–2696.
35. T. Yang, X. D. Wen, C. F. Huo, Y. W. Li, J. Wang and H. Jiao, J. Mol. Catal. A: Chem., 2009, 302, 129–136.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra22922k

‡ These authors contributed equally to this work.

---