Development of MgAl₂O₄-stabilized, Cu-doped, Fe₂O₃-based oxygen carriers for thermochemical water-splitting

Abstract

The commercially dominating technology for hydrogen production (i.e. steam methane reforming) emits large quantities of CO₂ into the atmosphere. On the other hand, emerging thermochemical water splitting cycles allow the production of a pure stream of H₂ while simultaneously capturing CO₂. From a thermodynamic point of view, the Fe₂O₃–Fe couple is arguably the most attractive candidate for thermochemical water splitting owing to its high H₂ yield and H₂ equilibrium partial pressure. However, the low reactivity of Fe₂O₃ with methane and the high activity of Fe for methane decomposition (leading to carbon deposition and in turn CO_x poisoning of the H₂ stream) are major drawbacks. Here, we report the development of MgAl₂O₄-stabilized, Cu-modified, Fe₂O₃-based redox materials for thermochemical water-splitting that show a high reactivity towards CH₄ and low rates of carbon deposition. To elucidate the effect of Cu doping on reducing significantly the rate of carbon deposition (while not affecting negatively the high redox activity of the material) extended X-ray absorption fine structure spectroscopy and energy dispersive X-ray spectroscopy was employed.

---

Introduction

An increasing global population in combination with the rapid industrial development of highly populous countries has resulted in a steep increase in the global electric power demand. Currently, fossil fuels have the largest share in the power generation mix (e.g. ∼68% in 2011) and will arguably remain a major energy carrier in the future.^1,2 However, the combustion of fossil fuels is associated with the emission of large quantities of greenhouse gases, in particular carbon dioxide (CO₂) into the atmosphere. The increasing concentration of greenhouse gases in the atmosphere is most likely linked to global warming and climate change.³ Thus, to mitigate climate change, there is an urgent need to reduce or at least stabilize anthropogenic CO₂ emissions. In this regard, a transition to hydrogen (H₂) as a clean energy carrier, could contribute appreciably to a reduction of CO₂ emissions. However, the currently available technologies for the large scale production of H₂, e.g. steam methane reforming and coal gasification, are very energy intensive and release large quantities of CO₂ into the atmosphere.⁴ Thus, for H₂ to become a major energy carrier, it must be produced in an efficient and ‘sustainable’ manner, i.e. if derived from hydrocarbon fuels, H₂ must be produced in combination with carbon dioxide capture and storage (CCS).

In this context, thermochemical water-splitting is a promising process to produce high purity H₂ (without the need of additional energy intensive H₂ purification steps) while simultaneously capturing CO₂.⁵ In the chemical looping based water-splitting cycle studied here, a hydrocarbon fuel is used to reduce an oxygen carrier. The reduced oxygen carrier is subsequently re-oxidized by steam to produce high-purity H₂. Depending on thermodynamics and process configurations an additional air oxidation step is included after steam oxidation, allowing for an additional heat output. Iron oxide (Fe₂O₃) is a very attractive candidate for thermochemical water-splitting owing to its low cost, wide availability, minimal environmental impact and high equilibrium partial pressure of H₂.⁶ The iron-oxide based thermochemical water-splitting cycle was originally referred to as the ‘steam-iron process’.⁷ A schematic diagram of the iron oxide mediated water-splitting process, using methane (CH₄) as a fuel, is shown in Fig. 1.

Fig. 1 Schematic diagram of the steam-iron process.

Depending on intrinsic kinetics and the ratio of CH₄ to Fe₂O₃, the reduction of Fe₂O₃ with CH₄ occurs via:

CH₄ + 3Fe₂O₃ → CO + 2H₂ + 2Fe₃O₄, ΔH_{900 °C} = 215.5 kJ mol⁻¹ (1a)

CH₄ + Fe₃O₄ → CO + 2H₂ + 3FeO, ΔH_{900 °C} = 262.8 kJ mol⁻¹ (1b)

CH₄ + FeO → CO + 2H₂ + Fe, ΔH_{900 °C} = 250.9 kJ mol⁻¹ (1c)

or

CH₄ + 12Fe₂O₃ → CO₂ + 2H₂O + 8Fe₃O₄, ΔH_{900 °C} = 145.1 kJ mol⁻¹ (2a)

CH₄ + 4Fe₃O₄ → CO₂ + 2H₂O + 12FeO, ΔH_{900 °C} = 334.3 kJ mol⁻¹ (2b)

CH₄ + 4FeO → CO₂ + 2H₂O + 4Fe, ΔH_{900 °C} = 286.6 kJ mol⁻¹ (2c)

Subsequently, Fe is re-oxidized using steam to produce a pure stream of H₂:

3Fe + 4H₂O → Fe₃O₄ + 4H₂, ΔH_{900 °C} = −103.1 kJ mol⁻¹ (3)

To close the cycle, Fe₃O₄ is oxidized back to Fe₂O₃ using air, viz.:

4Fe₃O₄ + O₂ → 6Fe₂O₃, ΔH_{900 °C} = −237.0 kJ mol⁻¹ (4)

Addition of reactions (1b), (1c) and (3) yields the conventional steam reformation of CH₄:

CH₄ + H₂O → CO + 3H₂, ΔH_{900 °C} = 228.1 kJ mol⁻¹ (5)

However, unlike in conventional steam reforming, in the iron oxide mediated process described above a pure stream of H₂ is produced inherently. Moreover, the synthesis gas produced during the partial oxidation of CH₄ (reaction (1a–c)) has a H₂ to CO ratio of 2 : 1 and is, thus, suitable for methanol or Fischer–Tropsch synthesis.⁸ Compared to steam methane reforming, the exergetic efficiency of the iron oxide mediated water-splitting process is promising,⁹ however, the poor cyclability of pure, i.e. unsupported Fe₂O₃ is currently a major drawback. It has been found that if reduced down to metallic Fe, unsupported Fe₂O₃ deactivates after only a few cycles, probably due to thermal sintering.¹⁰ To reduce its tendency for thermal sintering, Fe₂O₃ is commonly stabilized with a high Tammann temperature support, e.g. Al₂O₃, ZrO₂, TiO₂, SiO₂ or MgAl₂O₄.¹¹ Previous studies of the iron oxide mediated water-splitting process have largely focused on using CO or synthesis gas (CO + H₂) as the reducing agents.^12–17 The preference for CO (or synthesis gas) is due to the fact that carbon deposition via the Boudouard reaction (2CO → C + CO₂) is thermodynamically limited at temperatures above 800 °C (K_p = ∼0.15 at 800 °C), whereas Fe catalyzes CH₄ decomposition,¹⁸viz.:

CH_4(g) → C_(s) + 2H_2(g), ΔH_{900 °C} = 58.5 kJ mol⁻¹ (6)

Moreover, the oxygen carriers developed so far have shown a poor reactivity with CH₄ when compared to H₂ or CO.^19,20 However, it should be noted that CO and synthesis gas are not primary fuels. Therefore, it is important to develop Fe₂O₃-based materials that have a high reactivity with CH₄ and, at the same time, a low propensity for coke formation when using in-expensive CH₄ as a fuel.

One approach to reduce CH₄ decomposition related carbon deposition is to limit the reduction of Fe₂O₃ to FeO, since FeO catalyzes only poorly reaction (6) when compared to Fe.²¹ However, this process modification reduces largely the yield of H₂ expressed as mol H₂ g⁻¹ oxygen carrier, since the H₂ yield for FeO is 3 times lower than for Fe. An alternative approach to suppress the formation of carbon during CH₄ oxidation while at the same time ensuring a high H₂ yield, is doping of Fe₂O₃ with a transition metal that is inactive for the dissociation of CH₄, e.g. Cu, Pt or Pd.

In this context, Cu is an attractive candidate because of its low cost and high activation energy for the dissociative chemisorption of CH₄ (compared to Pt and Pd).²² However, the main disadvantage of doping Fe₂O₃ with Cu is the possible formation of low Tammann temperature solid solutions, e.g. CuFe₂O₄ (Tammann temperature of only ∼400 °C).²³ So far only Kang et al.²⁴ have studied the redox characteristics of co-precipitated CuFe₂O₄, however, only five redox cycles were performed (CH₄ reduction at 900 °C for 50 min followed by steam oxidation at 800 °C for 120 min). It was reported that compared to unsupported Fe₃O₄, CuFe₂O₄ exhibited fast reduction kinetics, low carbon deposition and high CO selectivity. Nonetheless, large fluctuations in the CH₄ conversion (33–55 mol%) and H₂ yield (6.7–11.3 mmol g⁻¹) were observed over the 5 cycles tested (probably due to sintering). To stabilize the H₂ yield, Kang et al.²⁴ supported CuFe₂O₄ on ZrO₂ or CeO₂. However, this approach resulted in a reduction in the quantity of the active phase (CuFe₂O₄) to only 20 wt% CuFe₂O₄, a value that is arguably too small for practical applications. Nonetheless supported CuFe₂O₄ showed a higher CH₄ conversion when compared to pure CuFe₂O₄. However, somewhat surprisingly the oxidation kinetics of the new oxygen carrier were reduced substantially, viz. 80 min were required to fully re-oxidize ZrO₂ or CeO₂ supported CuFe₂O₄ (1 g oxygen carrier and gas space velocity ∼4 min⁻¹). In addition, the very low loading of CuFe₂O₄ in the material reduced the overall H₂ yield to only 2 mmol g⁻¹.

In this work, we demonstrate the potential of Cu-promoted, Fe₂O₃-based oxygen carriers for iron oxide mediated water-splitting. We were able to effectively stabilize CuO–Fe₂O₃ on MgAl₂O₄ using only 20 wt% support, thus, ensuring high H₂ yields. The cyclic redox stability and water-splitting characteristics were determined and interpreted in light of the morphological and phase characteristics of the material. In order to understand the role of Cu in inhibiting carbon deposition, the reduced oxygen carriers were studied using extended X-ray absorption fine structure spectroscopy and energy dispersive X-ray spectroscopy. The materials developed here show very little coke deposition while at the same time maintaining high and stable H₂ yields over many redox cycles.

Experimental

Oxygen carrier synthesis

MgAl₂O₄-stabilized, Cu-modified Fe₂O₃-based oxygen carriers were synthesized via a modified co-precipitation technique originally reported by He et al.²⁵ The quantity of support required to stabilize Fe₂O₃ is typically in the range of 30–60 wt%.¹¹ To maximize the quantity of Fe₂O₃ in the synthesized oxygen carriers, the quantity of Fe₂O₃ in the synthesized oxygen carriers was fixed to 70 wt% while the CuO content was varied from 0–30 wt%, the balance being MgAl₂O₄. In a typical synthesis, first appropriate amounts of Fe(NO₃)₃·9H₂O, Cu(NO₃)₂·2.5H₂O, Mg(NO₃)₂·6H₂O and Al(NO₃)₃·9H₂O were dissolved in 400 mL of deionized water (reverse osmosis, 15 MΩ cm). The molar ratio of Mg²⁺ to Al³⁺ was fixed to 1 : 2 while the total moles of cations were fixed to 0.3. Subsequently, 400 mL of an aqueous solution containing 24 g NaOH and 6 g Na₂CO₃ was added drop wise to the nitrate solution under continuous stirring. The pH of the resulting slurry was adjusted to 8.5 using NaOH. The mixture was kept at 80 °C for 15 h under total reflux. The resulting precipitate was filtered and washed several times with reverse osmosis water. The cake of the washed precipitate was dried in an oven at 70 °C for 24 h and subsequently calcined in a muffle furnace at 900 °C for 2 h using a temperature ramp of 5 °C min⁻¹. The calcined oxygen carriers were crushed and sieved into the size range of 300–425 μm. The following nomenclature is used to describe the composition of the oxygen carriers: the symbols Fe, Cu and MgAl are followed by the weight percentage of Fe₂O₃, Cu and MgAl₂O₄ in the oxygen carrier, respectively. For example, Fe₇₁Cu₀₂MgAl₂₇ is an oxygen carrier that contains 71 wt%, 2 wt% and 27 wt% of Fe₂O₃, Cu (after reduction of CuO) and MgAl₂O₄, respectively.

Characterization

The crystallinity and chemical composition of the fresh and cycled oxygen carriers was determined using a Bruker AXS D8 Advance X-ray diffractometer mounted with a Lynx eye super speed detector. The diffractometer was operated at 40 kV and 40 mA using CuK_α radiation (λ = 1.5418 nm). Each sample was scanned within the range of 2θ = 20–80° using a step size of 0.0275° per second. X-ray diffractograms were further refined using the General Structure Analysis System (GSAS).^26,27 Extended X-ray absorption fine structure (EXAFS) spectroscopy was performed at the Swiss-Norwegian beamline (SNBL, BM1B) of the European Synchrotron Radiation Facility (ESRF) to study the local structure of Fe and Cu in the reduced oxygen carriers. The oxygen carriers were mixed with cellulose (ratio 1 : 8) and pelletized. The pellets were scanned in transmission mode to acquire Fe and Cu K-edge EXAFS data. The acquired data were subsequently processed using the IFEFFIT package.²⁸

N₂ adsorption and desorption isotherms of the oxygen carriers were measured at −196 °C (Quantachrome NOVA 4000e analyser). Each sample was degassed for 3 h at 300 °C prior to the measurement. The surface area and pore volume were calculated using, respectively, the Brunauer et al.²⁹ and Barrett et al.³⁰ models. Scanning electron microscopy (Zeiss Gemini 1530 FEG) was applied to characterize the surface morphology of the fresh and cycled oxygen carriers. The average grain size was determined from electron micrographs using the ImageJ software package.³¹ The average grain size (and error bar) reported is based on the analysis of 50 individual grains. The elemental composition of the surface was mapped using energy dispersive X-ray (EDX) spectroscopy.

CH₄-temperature programmed reduction was performed in a Mettler Toledo TGA/DSC 1 thermo-gravimetric analyzer (TGA). In a typical experiment, a small amount (∼30 mg) of the oxygen carrier was heated from room temperature to 1000 °C at a rate of 10 °C min⁻¹ under a flow of 10 vol% CH₄ in N₂ (50 mL min⁻¹) and kept at 1000 °C for 30 min. In all experiments, a constant N₂ flow of 25 mL min⁻¹ was used as purge flow over the micro-balance. Raman spectroscopy (Thermo Scientific DXR Raman microscope) was performed to characterize the carbon deposited on the reduced oxygen carriers. Raman spectra were collected by exciting the samples with 10% power of a 780 nm laser (24 mW total power) for 10 s at 1 cm⁻¹ spectral resolution.

Fixed bed measurements

The redox performance of the synthesized oxygen carriers was evaluated in a packed-bed reactor at 900 °C. A schematic diagram of the experimental set-up is given in Fig. S1 of the (ESI†). In a typical experiment, the oxygen carrier was reduced with 10 vol% CH₄ in N₂ (1.5 L min⁻¹ at 25 °C and 1 bar). The reduced oxygen carrier was subsequently oxidized with 15 vol% steam in N₂. The flow rate of the steam/N₂ mixture was 2.83 L min⁻¹ (at 200 °C and 1 bar). To close the cycle, the oxygen carrier was oxidized back to Fe₂O₃ with 5 vol% O₂ in N₂ using a flow rate of 1.97 L min⁻¹ (at 25 °C and 1 bar). The duration of the CH₄ reduction, steam oxidation and air oxidation segments was fixed to 15 min, 8 min and 5 min, respectively. The reactor was purged for 1 min with N₂ (1.5 L min⁻¹) after every segment. In total 15 cycles were performed. The moles of H₂ produced during steam oxidation were calculated by integrating the molar flow rate of H₂ produced with respect to time.

Results

Characterization of the freshly calcined oxygen carriers

The composition of the freshly calcined oxygen carriers was determined using X-ray diffraction (Fig. 2). Hematite (Fe₂O₃) and MgAl₂O₄ were identified in fresh Fe₇₀MgAl₃₀. Rietveld refinement of the XRD data of Fe₇₀MgAl₃₀ gave ∼71.5 wt% Fe₂O₃ which is in good agreement with the theoretical value of 70 wt%. MgAl₂O₄-stabilized, Cu-modified Fe₂O₃-based oxygen carriers showed the formation of Fe₂O₃, CuFe₂O₄ and MgAl₂O₄. The X-ray diffractogram of fresh Fe₇₁Cu₀₂MgAl₂₇ contained peaks due to Fe₂O₃ and MgAl₂O₄ only, probably because the quantity of Cu²⁺ phases (CuO or CuFe₂O₄) was below the detection limit of the equipment. Rietveld refinement‡ of the XRD data of Fe₇₁Cu₀₂MgAl₂₇ yielded 70.9 wt% Fe₂O₃ and 29.1 wt% MgAl₂O₄, again in good agreement with the theoretically expected composition (70.0 wt% Fe₂O₃, 3.0 wt% CuO and 27.0 wt% MgAl₂O₄). The composition of Fe₇₁Cu₀₈MgAl₂₁ was determined (Rietveld refinement) as 53.7 wt% Fe₂O₃, 27.0 wt% CuFe₂O₄ and 19.3 wt% MgAl₂O₄. Assuming that all CuO in Fe₇₁Cu₀₈MgAl₂₁ forms CuFe₂O₄, ∼30 wt% CuFe₂O₄ would be expected theoretically in the calcined material. Interestingly, MgAl₂O₄ could not be observed in the diffractogram of Fe₇₃Cu₁₇MgAl₁₀, indicating its presence in an amorphous form. The predicted composition‡ of Fe₇₃Cu₁₇MgAl₁₀ (35.3 wt% Fe₂O₃ and 64.7 wt% CuFe₂O₄) was again in good agreement with its theoretical composition (30.0 wt% Fe₂O₃, 60.0 wt% CuFe₂O₄ and 10.0 wt% MgAl₂O₄, assuming the full solution of CuO into Fe₂O₃ forming CuFe₂O₄). Finally, only Bragg reflections corresponding to CuFe₂O₄ were identified in the diffractogram of fresh Fe₇₅Cu₂₅. Assuming a full solution of CuO into Fe₂O₃, 90.2 wt% CuFe₂O₄ and 9.8 wt% Fe₂O₃ would be expected for Fe₇₅Cu₂₅. Thus, our XRD measurements indicate that in Fe₇₅Cu₂₅, Fe₂O₃ may be present in an amorphous form.

Fig. 2 X-ray diffractograms of the freshly calcined (i.e. unreacted) oxygen carriers: (a) Fe₇₀MgAl₃₀, (b) Fe₇₁Cu₀₂MgAl₂₇, (c) Fe₇₁Cu₀₈MgAl₂₁, (d) Fe₇₃Cu₁₇MgAl₁₀ and (e) Fe₇₅Cu₂₅. The following compounds were identified: (■) Fe₂O₃, (▲) MgAl₂O₄ and (●) CuFe₂O₄.

Fig. S2 of the ESI† shows scanning electron micrographs of pure, i.e. unsupported Fe₂O₃, pure CuO and freshly calcined oxygen carriers. The surfaces of unsupported Fe₂O₃ and CuO are comprised of micrometre-sized polyhedrons (Fig. S2(g)† shows some sintering for unsupported CuO). On the other hand, the bimetallic oxygen carriers reveal different surface morphologies compared to unsupported Fe₂O₃ and CuO. The surfaces of Fe₇₀MgAl₃₀ and Fe₇₁Cu₀₂MgAl₂₇ are comprised of nanometre-sized grains, as seen in Fig. S2(a) and (b),† respectively. The average size of the grains was determined as 88 ± 14 nm and 124 ± 16 nm for, respectively, Fe₇₀MgAl₃₀ and Fe₇₁Cu₀₂MgAl₂₇. The surfaces of Fe₇₁Cu₀₈MgAl₂₁ and Fe₇₃Cu₁₇MgAl₁₀ showed some sintering and were, only in parts, comprised of small grains. The sintering tendency of the bi-metallic oxygen carriers (based on a visual inspection of the electron micrographs) was found to increase with increasing CuO content and can be explained by the low Tammann temperature of CuFe₂O₄ of only ∼400 °C.²³ Indeed, Fig. S2(e)† shows that Fe₇₅Cu₂₅ (almost pure CuFe₂O₄) has a very smooth surface texture possibly due to severe thermal sintering during calcination.

The BET surface area and BJH pore volume of the calcined oxygen carriers obtained from N₂ adsorption isotherms are summarized in Table 1. As expected from the electron micrographs, Fe₇₀MgAl₃₀ possessed the highest surface area and pore volume of all oxygen carriers synthesized. For the remaining oxygen carriers, a significant reduction in surface area and pore volume was observed for increasing Cu contents, most likely due to thermal sintering during calcination, as visualized in the electron micrographs shown in Fig. S2.†

Table 1 BET surface area and BJH pore volume of the calcined oxygen carriers

Material	BET surface area [m^2 g^−1]	BJH pore volume [cm^3 g^−1]
Fe70MgAl30	12	0.11
Fe71Cu02MgAl27	5	0.05
Fe71Cu08MgAl21	2	0.02
Fe73Cu17MgAl10	2	0.01
Fe75Cu25	<1	<0.01

---

CH₄-temperature programmed reduction

CH₄-temperature programmed reduction (TPR) was performed to compare the reactivity of the oxygen carries synthesized with CH₄ (Fig. 3, the CH₄-TPR profiles of pure Fe₂O₃, CuO and CuFe₂O₄ are provided in Fig. S3†). The onset of reduction was defined as the temperature at which the oxygen carrier had lost 2% of the total weight loss. The reduction was considered to be completed once 98% of the total weight loss had been lost. Fig. S3† shows that pure CuO was reduced in a single step in the temperature range 480–900 °C. On the other hand, the reduction of pure Fe₂O₃ and CuFe₂O₄ started at ∼650 °C and ∼750 °C, respectively, and did not reach completion during the measurement time. Compared to unsupported Fe₂O₃, the reduction of MgAl₂O₄-supported Fe₂O₃ (Fe₇₀MgAl₃₀) started at a considerably lower temperature, viz. at ∼460 °C. For Fe₇₀MgAl₃₀, a three step reduction mechanism was identified in the CH₄-TPR profile. The weight loss observed in the first step (temperature range 460–580 °C) agrees with the Fe₂O₃–Fe₃O₄ transition (see Fig. S4 of ESI†). Similarly, the weight losses recorded during the second and third steps correspond to the Fe₃O₄–FeO and FeO–Fe transitions, respectively. The reduction of Fe₃O₄ and FeO started at ∼620 °C and ∼800 °C, respectively. The increase in sample weight during the holding time (1000 °C) was due to carbon deposition via CH₄ decomposition. The diffractogram of reduced Fe₇₀MgAl₃₀ (Fig. S5 of the ESI†) revealed the presence of Fe, MgAl₂O₄ and Fe₃C.

Fig. 3 CH₄-TPR profiles of the synthesized oxygen carriers: () Fe₇₀MgAl₃₀, () Fe₇₁Cu₀₂MgAl₂₇, () Fe₇₁Cu₈MgAl₂₁, () Fe₇₃Cu₁₇MgAl₁₀ and () Fe₇₅Cu₂₅.

The reduction of Fe₇₁Cu₀₂MgAl₂₇ and Fe₇₁Cu₀₈MgAl₂₁ also occurred in three steps. The weight loss in the first step is in agreement with the CuO–Cu and Fe₂O₃–Fe₃O₄ transitions. The weight losses recorded in the second and third steps match well the theoretically predicted weight losses for the Fe₃O₄–FeO and FeO–Fe transitions. For Fe₇₁Cu₀₂MgAl₂₇, the first, second and third reduction steps started at ∼480 °C, ∼690 °C and ∼800 °C, respectively. Interestingly, the weight increase of Fe₇₁Cu₀₂MgAl₂₇ due to carbon deposition was appreciably lower compared to Fe₇₀MgAl₃₀. Indeed, peaks due to Fe₃C could not be observed in the diffractogram of reduced (and further exposed to methane) Fe₇₁Cu₀₈MgAl₂₁ (Fig. S5†), indicating that the presence of Cu inhibited notably carbon deposition. The reduction of Fe₇₁Cu₀₈MgAl₂₁ started at ∼510 °C, with the second and third reduction steps occurring at ∼710 °C and ∼820 °C, respectively. For Fe₇₁Cu₀₈MgAl₂₁, the tendency for carbon deposition was reduced further when compared to Fe₇₀MgAl₃₀ and Fe₇₁Cu₀₂MgAl₂₇. Only Fe and Cu were identified in the diffractogram of reduced (and further exposed to methane) Fe₇₁Cu₀₈MgAl₂₁.

The reduction of Fe₇₃Cu₁₇MgAl₁₀ was found to occur in four steps. The weight losses recorded for the first, second, third and fourth step correspond to the CuFe₂O₄–Cu + Fe₂O₃, Fe₂O₃–Fe₃O₄, Fe₃O₄–FeO and FeO–Fe transitions, respectively. The reduction of CuFe₂O₄, Fe₂O₃, Fe₃O₄ and FeO in Fe₇₃Cu₁₇MgAl₁₀ started at ∼580 °C, ∼660 °C, ∼760 °C and ∼890 °C, respectively. Interestingly, no increase in sample weight and, thus, no carbon deposition was observed for Fe₇₃Cu₁₇MgAl₁₀. The diffractogram of the reduced Fe₇₃Cu₁₇MgAl₁₀ showed peaks due to Fe and Cu only. The CH₄-TPR profile of Fe₇₅Cu₂₅ revealed a single step reduction mechanism in the temperature range 650–930 °C. The weight loss recorded is in agreement with the reductions of CuFe₂O₄ to Cu + Fe₂O₃ and Fe₂O₃ to FeO. The diffractogram of reduced Fe₇₅Cu₂₅ showed peaks due to FeO, Fe and Cu, indicative of a very slow reduction of FeO to Fe in Fe₇₅Cu₂₅ (as confirmed in Fig. 3).

Redox performance of the synthesized oxygen carriers

The cyclic redox performance of the synthesized oxygen carriers was assessed in a fixed bed reactor at 900 °C (reduction in 10 vol% CH₄ in N₂). The yield of H₂ during steam oxidation is plotted as a function of cycle number in Fig. 4. Fig. 5(a and b) plot, respectively, the molar ratio of H₂ to CO in the synthesis gas and the moles of CO_x produced during the oxidation of CH₄ as a function of cycle number. We observe that the H₂ yield of Fe₇₀MgAl₃₀ was ∼3.5 times higher than the theoretically expected value (11.6 mmol H₂ g_{oxygen carrier}⁻¹ assuming a full reduction to Fe) with some fluctuations during the first 10 cycles. For Fe₇₁Cu₀₂MgAl₂₇, Fe₇₁Cu₀₈MgAl₂₁ and Fe₇₃Cu₁₇MgAl₁₀ a stable H₂ yield of ∼16.2 mmol H₂ g_{oxygen carrier}⁻¹, ∼14.2 mmol H₂ g_{oxygen carrier}⁻¹ and ∼13.6 mmol H₂ g_{oxygen carrier}⁻¹, respectively, was obtained over the 15 redox cycles tested. Finally, the reference material Fe₇₅Cu₂₅ showed a stable, but low H₂ yield of ∼4.2 mmol H₂ g_{oxygen carrier}⁻¹, which is only ∼36% of the theoretically expected value.

Fig. 4 H₂ yield during re-oxidation using 15 vol% steam in N₂. The experiments were performed in a packed bed at 900 °C and the duration of the steam oxidation reaction was fixed to 8 min: (●) Fe₇₀MgAl₃₀, (×) Fe₇₁Cu₀₂MgAl₂₇, (■) Fe₇₁Cu₀₈MgAl₂₁, (♦) Fe₇₃Cu₁₇MgAl₁₀ and (▲) Fe₇₅Cu₂₅. The dashed line corresponds to the theoretical H₂ yield of 11.6 mmol H₂ g⁻¹ oxygen carrier for an oxygen carrier containing 70 wt% Fe₂O₃.

Fig. 5 (a) H₂ to CO ratio and (b) moles of CO (closed black symbols) and CO₂ (open red symbols) produced during reduction in 10 vol% CH₄ in N₂. The experiments were performed in a packed bed at 900 °C. The duration of the reduction reaction was fixed to 15 min: (●) Fe₇₀MgAl₃₀, (×) Fe₇₁Cu₀₂MgAl₂₇, (■) Fe₇₁Cu₀₈MgAl₂₁, (♦) Fe₇₃Cu₁₇MgAl₁₀ and (▲) Fe₇₅Cu₂₅. The dashed line corresponds to the theoretical H₂ to CO ratio of 2 : 1 assuming the partial oxidation of CH₄ to H₂ and CO.

Turning now to the production of a synthesis gas during the reduction of the oxygen carriers with methane, Fig. 5(a) shows that the H₂ to CO ratio for Fe₇₀MgAl₃₀ fluctuated somewhat during the first 10 cycles. From the eleventh cycle onwards, the H₂ to CO ratio stabilized at ∼9 : 1. On the other hand, Cu modified oxygen carriers showed a significantly lower H₂ to CO ratio. For Fe₇₁Cu₀₃MgAl₂₇, the H₂ to CO ratio was ∼6 : 1 in the first cycle which decreased to ∼5 : 1 in the fifteenth cycle. In cases of Fe₇₁Cu₀₈MgAl₂₁ and Fe₇₃Cu₁₇MgAl₁₀ the H₂ to CO ratio remained stable at ∼3 : 1 over repeated redox cycles. Finally, for the reference material Fe₇₅Cu₂₅ a synthesis gas with a H₂ to CO ratio of ∼4 : 1 was obtained in the first cycle. However, the H₂ to CO ratio gradually increased, reaching ∼5 : 1 in the fifteenth cycle. Among the synthesized oxygen carriers, Fe₇₅Cu₂₅ possessed the lowest CO yield of only ∼1.2 mmol CO g_{oxygen carrier}⁻¹, see Fig. 5(b). On the other hand, the CO yield of the MgAl₂O₄-stabilized oxygen carriers was 6–7 times higher than that of unsupported Fe₇₅Cu₂₅. It is also worth noting that the CO₂ yield of all oxygen carriers tested was ∼1.5 mmol CO₂ g_{oxygen carrier}⁻¹, independent whether Fe₂O₃ was supported or promoted with Cu.

Carbon deposition

Carbon deposition via CH₄ decomposition during reduction reduces the purity of the H₂ produced during steam oxidation due to the simultaneous formation of CO_x. Fig. 6 plots the moles of carbon deposited during CH₄ reduction as a function of cycle number. The quantity of carbon deposited was calculated from the moles of CO_x produced during re-oxidation (the mole fraction of CO_x in the product gas is plotted in Fig. S6† as a function of cycle number). The activity for CH₄ decomposition over the 15 redox cycles tested decreased in the following order: Fe₇₀MgAl₃₀ > Fe₇₁Cu₀₂MgAl₂₇ > Fe₇₁Cu₀₈MgAl₂₁ > Fe₇₃Cu₁₇MgAl₁₀ > Fe₇₅Cu₂₅. The average quantity of carbon deposited on the oxygen carriers over 15 redox cycles was 28.3 mmol g⁻¹, 3.7 mmol g⁻¹, 2.7 mmol g⁻¹, 1.8 mmol g⁻¹ and 0.4 mmol g⁻¹ for Fe₇₀MgAl₃₀, Fe₇₁Cu₀₂MgAl₂₇, Fe₇₁Cu₀₈MgAl₂₁, Fe₇₃Cu₁₇MgAl₁₀ and Fe₇₅Cu₂₅, respectively.

Fig. 6 Carbon deposition during CH₄ reduction. The experiments were performed at 900 °C in a packed bed: (●) Fe₇₀MgAl₃₀, (×) Fe₇₁Cu₀₂MgAl₂₇, (■) Fe₇₁Cu₀₈MgAl₂₁, (♦) Fe₇₃Cu₁₇MgAl₁₀ and (▲) Fe₇₅Cu₂₅.

To understand better the inhibiting effect of Cu for carbon deposition, reduced Fe₇₀MgAl₃₀ and Fe₇₃Cu₁₇MgAl₁₀ were studied using EXAFS spectroscopy. The acquired Fe and Cu K-edge EXAFS functions of the reduced materials are plotted in Fig. 7(a and b), respectively. The EXAFS functions of the reference Fe and Cu foils are also provided in Fig. 7. The Fe K-edge EXAFS functions (k²-weighted) of both oxygen carriers examined are similar to the reference Fe foil. Fitting of the first shell around the central Fe atom confirms that the experimentally determined Fe–Fe distances in the reduced oxygen carriers (Table S1†) are similar to the Fe–Fe distances in the reference Fe foil. A slightly higher Debye–Waller factor (σ²), which indicates disorder, was determined for the reduced oxygen carriers when compared to the reference Fe foil. Similarly, the k²-weighted Cu K-edge EXAFS function of reduced Fe₇₃Cu₁₇MgAl₁₀ resembles well the EXAFS function of the reference Cu foil. Fitting of the first shell around the central Cu atom yielded similar Cu–Cu distances as in the reference Cu foil (Table S2†). The Debye–Waller factor of reduced Fe₇₃Cu₁₇MgAl₁₀ was again higher when compared to the Cu foil. Importantly, the acquired Fe and Cu K-edge EXAFS functions of the reduced oxygen carriers confirmed that Fe and Cu did not form an alloy. The surface of reduced Fe₇₃Cu₁₇MgAl₁₀ was also probed via EDX spectroscopy. The Fe L-edge EDX spectrum of reduced Fe₇₃Cu₁₇MgAl₁₀ (Fig. 8(b)) indicates that Fe is present on the entire surface of the reduced material. However, the Cu L-edge EDX spectrum (Fig. 8(c)) and the contrast in Fig. 8(b) suggest that Cu covers Fe only partially on the surface of reduced Fe₇₃Cu₁₇MgAl₁₀.

Fig. 7 (a) Fe K-edge EXAFS functions (k²-weighted) of () Fe foil, () reduced Fe₇₀MgAl₃₀ and () reduced Fe₇₃Cu₁₇MgAl₁₀. (b) Cu K-edge EXAFS functions (k²-weighted) of () Cu foil and () reduced Fe₇₃Cu₁₇MgAl₁₀.

Fig. 8 (a) Scanning electron micrograph of reduced Fe₇₃Cu₁₇MgAl₁₀, (b) Fe L-edge, (c) Cu L-edge and (d) Fe and Cu L-edge EDX spectra.

Characterisation of the cycled oxygen carriers

X-ray diffractograms, scanning electron micrographs and N₂ physisorption isotherms of the cycled oxygen carriers (oxidized form) were acquired to probe compositional and morphological changes over repeated redox cycles. Fig. 9 shows that, compared to the freshly calcined materials, additional crystalline phases appeared in the cycled samples. For example, magnesium aluminium iron oxide (MgAl_0.6Fe_1.4O₄) was identified in the diffractograms of cycled Fe₇₀MgAl₃₀, Fe₇₁Cu₀₂MgAl₂₇ and Fe₇₁Cu₀₈MgAl₂₁. Owing to the formation of the MgAl_0.6Fe_1.4O₄ spinel, some amorphous Al₂O₃ might also be present in the cycled materials. Importantly, the formation of MgAl_0.6Fe_1.4O₄ indicates that MgAl₂O₄ is not inert as it has been assumed generally in previous studies.^17,32,33 Cycled Fe₇₃Cu₁₇MgAl₁₀ was found to be comprised mainly of CuFe₂O₄ and Fe₂O₃, however, weak Bragg reflections due to delafossite, CuFeO₂, were also identified. Intriguingly, MgAl₂O₄ could not be observed in cycled Fe₇₃Cu₁₇MgAl₁₀. Finally, cycled Fe₇₅Cu₂₅ contained CuFe₂O₄, CuFeO₂ and Fe₂O₃.

Fig. 9 X-ray diffractograms of the cycled materials: (a) Fe₇₀MgAl₃₀, (b) Fe₇₁Cu₀₂MgAl₂₇, (c) Fe₇₁Cu₀₈MgAl₂₁, (d) Fe₇₃Cu₁₇MgAl₁₀ and (e) Fe₇₅Cu₂₅. The following compounds were identified: (■) Fe₂O₃, (♦) MgAl_0.6Fe_1.4O₄, (▲) MgAl₂O₄, (◊) CuFe₂O₄ and (●) CuFeO₂.

In addition, also the morphology of the materials changed over multiple redox cycles as confirmed by electron microscopy (Fig. S7†) and N₂ physisorption (Table 2). For example, the electron micrographs of cycled Fe₇₀MgAl₃₀ (Fig. S7(a)†) and Fe₇₁Cu₀₂MgAl₂₇ (Fig. S7(b)†) show that the nanometre-sized grains, originally present in freshly calcined Fe₇₀MgAl₃₀ and Fe₇₁Cu₀₂MgAl₂₇, had sintered, viz. the average grain size of cycled Fe₇₀MgAl₃₀ (135 ± 20 nm) and Fe₇₁Cu₀₂MgAl₂₇ (210 ± 29 nm) increased by a factor of 1.5. This observation is in line with the significantly reduced BET surface area and BJH pore volume of cycled Fe₇₀MgAl₃₀ and Fe₇₁Cu₀₂MgAl₂₇ (Table 2). The surface morphology of cycled Fe₇₁Cu₀₈MgAl₂₁ (Fig. S7(c)†) and Fe₇₃Cu₁₇MgAl₁₀ (Fig. S7(d)†) was also altered during repeated redox cycles. Cycled Fe₇₁Cu₀₈MgAl₂₁ and Fe₇₃Cu₁₇MgAl₁₀ comprised of large grains that were often fused together. The average grain size was 198 ± 54 nm and 172 ± 41 nm for Fe₇₁Cu₀₈MgAl₂₁ and Fe₇₃Cu₁₇MgAl₁₀, respectively. However, comparing the data listed in Tables 1 and 2, the surface area and pore volume of Fe₇₁Cu₀₈MgAl₂₁ did not change appreciably over 15 redox cycles. On the other hand, the surface area of Fe₇₃Cu₁₇MgAl₁₀ decreased from 2 m² g⁻¹ to 1 m² g⁻¹. Finally, the electron micrograph of cycled Fe₇₅Cu₂₅ (Fig. S7(e)†) shows a surface comprised of well-defined edges and steps. This change in morphology of Fe₇₅Cu₂₅ was accompanied by an increase in surface area (from <1 m² g⁻¹ to 3 m² g⁻¹).

Table 2 BET surface area and BJH pore volume of the cycled materials

Material	BET surface area [m^2 g^−1]	BJH pore volume [cm^3 g^−1]
Fe70MgAl30	2	0.02
Fe71Cu02MgAl27	2	0.01
Fe71Cu08MgAl21	2	0.01
Fe73Cu17MgAl10	1	0.01
Fe75Cu25	3	0.01

---

Discussion

Reduction characteristics of the synthesized oxygen carriers

CH₄-TPR measurements of the synthesized oxygen carriers (Fig. 3) show that supporting Fe₂O₃ on MgAl₂O₄ shifts its reduction to considerably lower temperatures (compared to unsupported Fe₂O₃). With regard to the lower reduction temperature of MgAl₂O₄-supported Fe₂O₃, it is conceivable that the MgAl₂O₄ support influences the reducibility of Fe₂O₃ by modifying CH₄ chemisorption and charge transport characteristics of the oxygen carrier. Cui et al.³⁴ studied the influence of α-Al₂O₃ and MgAl₂O₄ supports on the dissociative chemisorption of CH₄ on Ni. Using CH₄ pulse experiments, Cui et al.³⁴ demonstrated that the MgAl₂O₄ support lowered the activation energy of CH₄ dissociation on Ni sites when compared to α-Al₂O₃ supported and unsupported Ni. The effect of the support on the reduction temperature of Fe₂O₃ was also studied by Galinsky et al.³⁵ who reported that the reduction temperature of La_0.8Sr_0.2FeO₃-supported Fe₂O₃ is roughly 250 °C lower than for TiO₂-supported Fe₂O₃. The high reactivity of La_0.8Sr_0.2FeO₃-supported Fe₂O₃ was attributed to a faster solid state O²⁻ and electron/hole transport from the bulk to the particle surface. On the other hand, the addition of Cu shifts the reduction of Fe₂O₃ to higher temperatures, viz. the reduction temperature of Fe₂O₃ increased in the following order: Fe₇₀MgAl₃₀ < Fe₇₁Cu₀₂MgAl₂₇ ≈ Fe₇₁Cu₀₈MgAl₂₁ ≈ Fe₇₃Cu₁₇MgAl₁₀ < Fe₇₅Cu₂₅. CH₄-TPR profiles of unsupported Fe₂O₃ and CuFe₂O₄ showed that CuFe₂O₄ and Fe₂O₃ were reduced at ∼750 °C and ∼650 °C, respectively. Therefore, the shift to higher reduction temperatures for Cu containing materials may be explained by the formation of CuFe₂O₄ in the oxygen carriers. The CH₄-TPR profiles also revealed that metallic Cu formed through the reduction of CuFe₂O₄, viz.

3CuFe₂O₄ + 4CH₄ → 4CO + 8H₂ + 3Cu + 2Fe₃O₄ (7)

decreases slightly the (apparent) rate of reduction of Fe₃O₄ (to FeO) and FeO (to Fe). These observations are in line with the fact that the activation of CH₄ (CH_4,s ↔ CH_3,s + H_s) on Fe is exothermic (activation energy = −62.8 kJ mol⁻¹) but endothermic on Cu (activation energy = 123.1 kJ mol⁻¹).³⁶ As a consequence, the formation of Cu decreases the (apparent) rate of reduction of Fe₃O₄ and FeO with CH₄.

Reactivity measurements

The molar H₂ to CO ratio of the synthesis gas produced during the reduction of iron oxide was higher than the theoretically expected value of 2 : 1. The average molar H₂ to CO ratio (over 15 redox cycles) increased in the following order: Fe₇₃Cu₁₇MgAl₁₀ (3 : 1) ≈ Fe₇₁Cu₀₈MgAl₂₁ (3 : 1) < Fe₇₁Cu₀₂MgAl₂₇ (5 : 1) ≈ Fe₇₅Cu₂₅ (5 : 1) < Fe₇₀MgAl₃₀ (9 : 1). Fig. 5 indicates that all synthesized oxygen carriers show similar CO₂ yields during CH₄ oxidation. In contrast, the CO yields of all MgAl₂O₄-stabilized materials were similar and significantly higher than for unsupported Fe₇₅Cu₂₅. Thus, the high H₂ to CO ratio of MgAl₂O₄-stabilized materials can be explained only by the deposition of carbon during the reduction step. Previously, Neal et al.³⁷ have observed four regimes during CH₄ reduction of Fe₂O₃-based oxygen carriers viz. (i) full oxidation of CH₄ to CO₂, (ii) competing full oxidation to CO₂ and partial oxidation to CO, (iii) partial oxidation of CH₄ to CO and (iv) CH₄ decomposition. For MgAl₂O₄-supported materials, the reaction between CH₄ and the oxygen carriers synthesized was sufficiently fast to reach regime (iv) within 15 min, resulting in carbon deposition which in turn led to a high H₂ to CO ratio. Galinsky et al.²⁰ studied the effect of three different supports, viz. Ca_0.8Sr_0.2Ti_0.8Ni_0.2O₃, CeO₂ and MgAl₂O₄, on the methane conversion characteristics of Fe₂O₃ and found that MgAl₂O₄ has the highest activity for CH₄ decomposition. Among the materials tested here, Fe₇₅Cu₂₅ has the lowest reactivity with CH₄ probably due to the absence of a MgAl₂O₄ support (see Fig. 3). As a consequence, the reduction of Fe₇₀Cu₂₅ was still in regime (ii) after 15 min. Thus, for Fe₇₅Cu₂₅ the high H₂ to CO ratio is due to a high CO₂ mole fraction (but low CO mole fraction) in the off gas.

Turning now to the steam oxidation step, the high propensity of Fe₇₀MgAl₃₀ for coke formation led to a substantial CO_x contamination of the H₂ produced according to:

C + H₂O → CO + H₂, ΔH_{900 °C} = 169.6 kJ mol⁻¹ (8)

C + 2H₂O → CO₂ + 2H₂, ΔH_{900 °C} = 136.9 kJ mol⁻¹ (9)

Owing to severe coke deposition, the H₂ yield of Fe₇₀MgAl₃₀ was 3.5 times higher than the theoretically expected value. For Cu-modified materials, the quantity of carbon deposited decreased with increasing Cu content. For example, for Fe₇₁Cu₀₂MgAl₂₇, Fe₇₁Cu₀₈MgAl₂₁ and Fe₇₃Cu₁₇MgAl₁₀ the H₂ yield was, respectively, only 1.4 times, 1.2 times and 1.1 times higher than the theoretically expected value, indicating that Cu significantly reduced coke deposition. Finally, the low H₂ yield of only ∼32% of the theoretically expected value for Fe₇₅Cu₂₅ is due to the incomplete reduction of FeO to Fe with CH₄. The yield of H₂ obtained through the oxidation of FeO is 3 times lower than that from Fe. For all Cu modified materials, the CO_x content in the hydrogen produced during steam oxidation is >5 vol% (Fig. S6†), however, with easily-implementable process modifications the purity of the H₂ produced can be increased further. Fig. S8† plots the composition of the effluent gas during steam oxidation (15^th cycle) for Fe₇₀MgAl₃₀ and Fe₇₁Cu₀₈MgAl₂₁. For Fe₇₀MgAl₃₀, CO is produced in large quantities during the entire steam oxidation step. On the other hand, for Fe₇₁Cu₀₈MgAl₂₁ the H₂ produced after the initial 30 s is essentially CO_x free. Thus, by a simple switching operation, high purity hydrogen can be produced with Fe₇₁Cu₀₈MgAl₂₁.

Influence of Cu on CH₄ decomposition

CH₄-TPR (Fig. 3) and reactive measurements (Fig. 6) confirm that the addition of Cu to MgAl₂O₄-stabilized, Fe₂O₃-based oxygen carriers improved appreciably the materials' resistance to carbon deposition via CH₄ decomposition. The propensity of the oxygen carriers to carbon deposition decreased with increasing Cu content. CH₄-TPR in combination with Raman spectroscopy (Fig. S9†) showed that the decomposition of CH₄ started once Fe₂O₃ was fully reduced to metallic iron. Previous studies of Bengaard et al.³⁸ and Besenbacher et al.³⁹ reported that (i) the dissociation of CH₄ molecules (CH_4,s ↔ C_s + 4H_s) on a metal surface and (ii) the nucleation of graphite on steps and edges play a key role in carbon formation. The reaction enthalpy for CH₄ dissociation over Fe is exothermic (−216.9 kJ mol⁻¹) but endothermic (251.2 kJ mol⁻¹) for Cu.³⁶ Our EXAFS (Fig. 7) and EDX (Fig. 8) measurements confirm that in the reduced oxygen carriers Fe and Cu do not form a solid solution, in line with an experimentally determined Fe–Cu equilibrium phase diagram (using thermal analysis and electron microscopy).⁴⁰ Owing to the phase separation of Cu and Fe a sharp interphase on the surface of the reduced oxygen carriers can be expected, as illustrated in Fig. 10. We speculate that the, at least partial, covering of surface Fe (largely steps) with Cu decreases the rate of CH₄ dissociation, which in turn reduces carbon deposition. Furthermore, our preliminary kinetic data show that the rate of oxidation of Fe with steam is not significantly affected by the presence of Cu. The fact that Fe and Cu occur in separate phases in the reduced state ensures that Cu-promoted Fe₂O₃ maintains the excellent redox characteristic of Fe₂O₃ such as a high H₂ yield and reaction kinetics during the steam oxidation step.

Fig. 10 A schematic representation of carbon deposition on (a) MgAl₂O₄-stabilized Fe₂O₃ and (b) Cu-modified, MgAl₂O₄-stabilized Fe₂O₃.

Conclusions

In this work, we have demonstrated that MgAl₂O₄-stabilized, CuO–Fe₂O₃-based oxygen carriers possess a high and stable H₂ yield over many redox cycles while maintaining a low propensity for coke deposition. CH₄-temperature programmed reduction experiments revealed that the propensity for coke formation decreased with increasing Cu content. Furthermore, Cu shifted the reduction of Fe₃O₄ and FeO to slightly higher temperatures (using CH₄). At 900 °C (using 10 vol% CH₄ in N₂) MgAl₂O₄-stabilized Fe₂O₃ produced a synthesis gas with a high H₂ to CO ratio of 9 : 1 owing to substantial carbon deposition via CH₄ decomposition. However, increasing the Cu content in the oxygen carrier decreased the H₂ to CO ratio (compared to MgAl₂O₄-stabilized Fe₂O₃). On the other hand, unsupported Cu-modified Fe₂O₃ showed low H₂ yields reaching only ∼32% of the theoretically expected value. This was due to the incomplete reduction of FeO to Fe using CH₄. During steam oxidation, MgAl₂O₄-stabilized, Cu-modified, Fe₂O₃-based oxygen carriers exhibited H₂ yields close to the theoretically expected value owing to a low CH₄ decomposition activity for these materials. Extended X-ray absorption fine structure spectroscopy and energy dispersive X-day spectroscopy confirmed that Fe and Cu did not form an alloy in the reduced oxygen carrier. The high resistance to carbon deposition of Cu modified oxygen carriers was, therefore, attributed to the (partial) coverage of surface Fe with Cu. Owing to their low tendency for coke deposition Cu-modified, MgAl₂O₄-stabilized Fe₂O₃ based oxygen carrier, produced hydrogen with very little CO_x contamination. The oxygen carrier that contained 71 wt% Fe₂O₃, 8 wt% Cu and 21 wt% MgAl₂O₄ was identified as a highly promising material for CH₄-based water-splitting cycles for H₂ production.

Acknowledgements

We are grateful to the Swiss National Science Foundation (SNF) (Project 406640_13670011) and the Swiss Federal Office of Energy (SFOE) (Projects SI/500652 and SI/500653) for financial support. We also thank Mrs Lydia Zender for her help with the XRD analysis and the Scientific Centre for Optical and Electron Microscopy (ScopeM) for providing access to and training on scanning electron microscopes. We acknowledge SNBL and ESRF for the provision of beamtime.

Notes and references

1. Key World Energy Statistics 2013, 2013.
2. Energy Technology Perspectives 2014: Harnessing Electricity's Potential, 2014.
3. B. Metz, Intergovernmental Panel on Climate Change. Working Group III, IPCC special report on carbon dioxide capture and storage, Cambridge University Press, for the Intergovernmental Panel on Climate Change, Cambridge, 2005.
4. A. Thursfield, A. Murugan, R. Franca and I. S. Metcalfe, Energy Environ. Sci., 2012, 5, 7421–7459.
5. M. A. Rosen, Energy, 2010, 35, 1068–1076.
6. R. D. Solunke and G. Veser, Ind. Eng. Chem. Res., 2010, 49, 11037–11044.
7. A. Messerschmitt, Germany Pat., 266863, 1911.
8. K. S. Go, S. R. Son, S. D. Kim, K. S. Kang and C. S. Park, Int. J. Hydrogen Energy, 2009, 34, 1301–1309.
9. J. P. E. Cleeton, C. D. Bohn, C. R. Muller, J. S. Dennis and S. A. Scott, Int. J. Hydrogen Energy, 2009, 34, 1–12.
10. C. D. Bohn, C. R. Muller, J. P. Cleeton, A. N. Hayhurst, J. F. Davidson, S. A. Scott and J. S. Dennis, Ind. Eng. Chem. Res., 2008, 47, 7623–7630.
11. J. Adanez, A. Abad, F. Garcia-Labiano, P. Gayan and L. F. de Diego, Prog. Energy Combust. Sci., 2012, 38, 215–282.
12. C. D. Bohn, J. P. Cleeton, C. R. Muller, S. Y. Chuang, S. A. Scott and J. S. Dennis, Energy Fuels, 2010, 24, 4025–4033.
13. A. M. Kierzkowska, C. D. Bohn, S. A. Scott, J. P. Cleeton, J. S. Dennis and C. R. Muller, Ind. Eng. Chem. Res., 2010, 49, 5383–5391.
14. P. Datta, L. K. Rihko-Struckmann and K. Sundmacher, Mater. Chem. Phys., 2011, 129, 1089–1095.
15. V. Galvita and K. Sundmacher, Chem. Eng. J., 2007, 134, 168–174.
16. A. Murugan, A. Thursfield and I. S. Metcalfe, Energy Environ. Sci., 2011, 4, 4639–4649.
17. S. Z. Wang, G. X. Wang, F. Jiang, M. Luo and H. Y. Li, Energy Environ. Sci., 2010, 3, 1353–1360.
18. Y. Nishiyam and Y. Tamai, J. Catal., 1974, 33, 98–107.
19. T. Kodama, H. Ohtake, S. Matsumoto, A. Aoki, T. Shimizu and Y. Kitayama, Energy, 2000, 25, 411–425.
20. N. L. Galinsky, A. Shafiefarhood, Y. G. Chen, L. Neal and F. X. Li, Appl. Catal., B, 2015, 164, 371–379.
21. X. Zhu, Y. G. Wei, H. Wang and K. Z. Li, Int. J. Hydrogen Energy, 2013, 38, 4492–4501.
22. M. S. Liao and Q. E. Zhang, J. Mol. Catal. A: Chem., 1998, 136, 185–194.
23. I. Barin, O. Knacke and O. Kubaschewski, Thermochemical properties of inorganic substances: supplement, Springer-Verlag, Berlin, New York, 1977.
24. K. S. Kang, C. H. Kim, W. C. Cho, K. K. Bae, S. W. Woo and C. S. Park, Int. J. Hydrogen Energy, 2008, 33, 4560–4568.
25. L. He, H. Berntsen, E. Ochoa-Fernandez, J. C. Walmsley, E. A. Blekkan and D. Chen, Top. Catal., 2009, 52, 206–217.
26. B. H. Toby, J. Appl. Crystallogr., 2001, 34, 210–213.
27. A. C. Larson and R. B. V. Dreele, General Structure Analysis System (GSAS), 2004.
28. B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12, 537–541.
29. S. Brunauer, P. H. Emmett and E. Teller, J. Am. Chem. Soc., 1938, 60, 309–319.
30. E. P. Barrett, L. G. Joyner and P. P. Halenda, J. Am. Chem. Soc., 1951, 73, 373–380.
31. C. A. Schneider, W. S. Rasband and K. W. Eliceiri, Nat. Methods, 2012, 9, 671–675.
32. W. Liu, M. Ismail, M. T. Dunstan, W. Hu, Z. Zhang, P. S. Fennell, S. A. Scott and J. S. Dennis, RSC Adv., 2015, 5, 1759–1771.
33. M. Rydén and M. Arjmand, Int. J. Hydrogen Energy, 2012, 37, 4843–4854.
34. Y. H. Cui, H. Y. Xu, Q. J. Ge, Y. Z. Wang, S. F. Hou and W. Z. Li, Chin. J. Catal., 2006, 27, 479–484.
35. N. L. Galinsky, Y. Huang, A. Shafiefarhood and F. X. Li, ACS Sustainable Chem. Eng., 2013, 1, 364–373.
36. M. J. Hei, H. B. Chen, J. Yi, Y. J. Lin, Y. Z. Lin, G. Wei and D. W. Liao, Surf. Sci., 1998, 417, 82–96.
37. L. M. Neal, A. Shafiefarhood and F. X. Li, ACS Catal., 2014, 4, 3560–3569.
38. H. S. Bengaard, J. K. Norskov, J. Sehested, B. S. Clausen, L. P. Nielsen, A. M. Molenbroek and J. R. Rostrup-Nielsen, J. Catal., 2002, 209, 365–384.
39. F. Besenbacher, I. Chorkendorff, B. S. Clausen, B. Hammer, A. M. Molenbroek, J. K. Norskov and I. Stensgaard, Science, 1998, 279, 1913–1915.
40. S. E. Amara, A. Belhadj, R. Kesri and S. Hamar-Thibault, Z. Metallkd., 1999, 90, 116–123.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ta06753g

‡ In calcined materials, only MgAl₂O₄ or CuFe₂O₄ was present in an amorphous phase. Thus, using Rietveld refinement, the composition of the samples was estimated on a MgAl₂O₄- or CuFe₂O₄-free basis and compared with the theoretically expected values.

---