Enhancing the catalytic performance of cobalt oxide by doping on ceria in the high temperature water–gas shift reaction

Abstract

The high temperature water–gas shift (HT-WGS) reaction was performed using a Co–CeO₂ catalyst, prepared through a co-precipitation method. The catalyst showed stable activity performance at 400 °C with 90% CO conversion without any side reactions (methanation) at a very high GHSV of 143 000 h⁻¹, which is the highest value reported for the HT-WGS reaction. This remarkable performance is attributed to the superior reducible nature of ceria and the preferential exposure of (220) and (112) facets of CeO₂ and Co₃O₄, respectively. The time-on-stream study substantiates that ceria stabilizes the surface area of the Co–CeO₂ catalyst during the WGS reaction compared to the bulk Co₃O₄.

---

Introduction

The water–gas shift (WGS) reaction (CO + H₂O → H₂ + CO₂) is critical in the production of hydrogen.¹ Commercially the WGS reaction employed in industry consists of two steps: a high temperature shift and a low temperature shift. The high temperature water–gas shift (HT-WGS) catalysts are mainly composed of Fe₃O₄–Cr₂O₄, whereas the low temperature water–gas shift (LT-WGS) catalyst is composed of Cu–Zn–Al₂O₃.^2,3 Cr₂O₄ has been used with Fe₃O₄ as a textural promoter to prevent losses in specific surface area of Fe₃O₄ caused by thermal sintering at higher temperature (>400 °C).⁴ However, due to the potential environmental concerns related to Cr, much research concerning Cr-free catalysts for HT-WGS reaction has been reported in the last two decades. In one such approach, Fe–Al–Cu catalyst was investigated for use in the HT-WGS reaction and showed high catalytic activity. However, the Fe–Al–Cu catalyst suffered from deactivation at higher Cu loadings.⁵ Additionally, Cu–CeO₂ was employed as a catalyst for the WGS reaction and showed rapid deactivation at high temperature.⁶ Senanayake et al. explored the activity of Ni/CeO₂ catalyst in the HT-WGS reaction and observed the production of methane as an undesirable side product (CO + 3H₂ → CH₄ + H₂O).⁷ This might be due to the higher CO adsorption efficiency of Ni compared to that of Cu.⁸ This indicates a potential drawback of using Ni-based catalysts in the HT-WGS reaction. Co/γ-Al₂O₃ was also tested for use in the WGS reaction. Although Co/γ-Al₂O₃ catalysts have been shown to be effective for the WGS, 17% methanation (side reaction) was observed during the reaction.⁹ Because the generation of one mole of CH₄ requires three moles of H₂, the H₂ yield in the output gas is greatly diminished. Thus, the development of highly selective and thermally-stable catalyst is desirable for the industrialization of the WGS reaction.

Co₃O₄ is known to be effective for catalyzing the oxidation of CO.¹⁰ The valence states of cobalt in Co₃O₄ are +2 and +3. The coexistence of Co²⁺–Co³⁺ ion pairs in the same material was found to be essential for catalytic activity. However, the catalytic performance of Co₃O₄ decreases in the presence of water.¹¹ This might be caused by the adsorption of water onto the active sites (Co³⁺) of Co₃O₄. Ceria has been demonstrated as an active material for water dissociation in the WGS reaction.¹² Thus, doping cobalt oxide onto ceria is expected to overcome this problem and would be a promising catalyst for use in the WGS reaction. In addition, ceria shows unique redox properties (Ce⁴⁺/Ce³⁺) and strongly promotes surface oxygen vacancy formation.¹³ At a structural level, a doped transition metal can introduce stress into the lattice of ceria, which in turn decrease the oxygen vacancy formation energy resulting in greater non-stoichiometry.¹⁴ The formation of oxygen vacancies has been proposed to be responsible for the enhancement of catalytic activity in WGS reaction by promoting the mobility of bulk oxygen species at the surface of ceria.¹⁵

Herein, we investigate a cobalt-doped ceria catalyst (Co–CeO₂) for use in the HT-WGS reaction in an attempt to attain high catalytic activity, selectivity, and thermal stability during the reaction. To the best of our knowledge, the use of a cobalt-doped ceria catalyst in the HT-WGS reaction has not yet been reported. Doping of the ceria sub-lattice with Co²⁺ ions allows for the accommodation of more oxygen vacancies compared to the pure ceria and enhances the redox properties of the resulting mixed oxide.¹⁶ Co–CeO₂ and bulk Co₃O₄ catalysts were synthesized through a co-precipitation method and evaluated for the HT-WGS reaction in order to understand the role of CeO₂ for an enhancement in catalytic activity and stability over time-on-stream in the case of Co–CeO₂ catalyst.

Results and discussion

The physico-chemical properties of the prepared catalysts are listed in Table 1 and Fig. 1. The crystallinity and phase identification of the reduced Co₃O₄, Co–CeO₂ and pure CeO₂ samples were measured using XRD. Pure CeO₂ shows XRD peaks at 2θ = 28.5, 33.0, 47.4, 56.2, 69.3, 76.7, and 78.9°, indicating the presence of fluorite-structured CeO₂. The diffraction pattern of the Co–CeO₂ sample was similar to that of pure ceria; however, the peaks were slightly shifted toward higher 2θ values, as shown in Fig. 1(A). This shift was due to the difference in ionic radii of the dopant (Co²⁺ = 0.65 Å and Co³⁺ = 0.61 Å) and host Ce⁴⁺ ions (0.97 Å), which leads to contraction of the unit cell parameter of ceria, as shown in Table 1.¹⁷ The contraction of the ceria lattice indicated that Co²⁺ ions had been substituted for Ce⁴⁺ and were doped into the lattice of CeO₂. The XRD pattern of reduced Co₃O₄ displayed a sharp peak at 2θ = 44.2°, corresponding to the metallic phase of cobalt.¹⁸ The Co–CeO₂ sample exhibited a small peak at the same position, indicating the presence of the metallic phase of Co in the catalyst. The average crystallite size and lattice strain of CeO₂ and Co–CeO₂ were calculated based on broadening of the XRD peak using the Williamson–Hall procedure. The estimated values are presented in Table 1, demonstrating the increase in crystallite size and lattice strain upon doping cobalt into ceria. The increase of lattice strain is beneficial for the WGS reaction because it can enhance the reducibility of the catalyst by decreasing the strength of the metal–oxygen bond.¹⁹

Table 1 Structural characterization of catalysts by BET and XRD

Catalyst	SBET (m^2 g^−1)	Crystallite size (nm)	Lattice strain	Unit cell parameter (Å)
CeO2	150	8	0.137	5.42
Co–CeO2	122	13	0.554	5.37

---

Fig. 1 XRD (A) and H₂-TPR (B) patterns of Co₃O₄, Co–CeO₂ and pure CeO₂ catalysts.

H₂-TPR analysis was performed to investigate the reducibility of Co–CeO₂ catalyst. The comparative H₂-TPR results of the bulk Co₃O₄, Co–CeO₂, and pure CeO₂ samples are shown in Fig. 1(B). The H₂-TPR profile of Co₃O₄ exhibited two reduction peaks: a low temperature peak (326 °C) attributed to the reduction of Co³⁺ to Co²⁺ and a high temperature peak (547 °C) corresponding to the reduction of Co²⁺ to metallic Co.²⁰ Pure CeO₂ showed two broad peaks at 458 and 597 °C, corresponding to the reduction of adsorbed and surface lattice oxygen of ceria, respectively.²¹ However, the H₂-TPR profile of Co–CeO₂ showed three reduction peaks at 287, 391, and 643 °C. Based on earlier reports, the first peak was assigned to the reduction of Co³⁺ to Co²⁺, the second peak was caused by the reduction of surface oxygen of CeO₂, and the third peak could be assigned to the reduction of Co²⁺ to metallic Co.²² Interestingly, the reduction peak of Co³⁺ to Co²⁺ was shifted to a lower temperature while reduction of Co²⁺ to metallic Co was delayed by the stabilization effect of ceria on the Co²⁺ ions in the Co–CeO₂ sample.²² This might be due to the interaction between cobalt and ceria components. A similar observation was reported by Luo and Martínez-Arias et al. for the CuO/CeO₂ system.²³ In addition, the peak related to the reduction of lattice oxygen of surface CeO₂ was shifted to a lower temperature (391 °C) in Co–CeO₂ catalyst relative to that of pure ceria. This phenomenon suggests a synergistic effect between cobalt and ceria oxide, attributed by their easier reducibility.

From the XRD and H₂-TPR results, we observed that doping cobalt into ceria leads to a change in the intrinsic properties of ceria; this is further explored using Raman characterization. For comparison, the Raman spectra of pure CeO₂ and Co–CeO₂ are displayed in Fig. 2. Pure CeO₂ shows a characteristic strong band at 461 cm⁻¹ (F_2g) with a weak band near 597 cm⁻¹ (D). The F_2g-band arises from the symmetric stretching mode of Ce–O₈, while the D-band is associated with Frenkel-type anion defects in the ceria lattice.²⁴ These defects in the ceria lattice induced the formation of oxygen vacancies due to the mobility of the lattice oxygen. The concentration of oxygen vacancies in the oxides can be determined by measuring the ratio of the defect-induced band intensity (I_D) and the F_2g-band intensity (I_F2g) (i.e., I_D/I_F2g). Doping cobalt into fluorite-structured CeO₂ results in deformation of the lattice, as evidenced by the reduction in F_2g-band intensity and a major shift to a lower energy (447 cm⁻¹) compared to that of pure ceria.²⁵ Moreover, Co–CeO₂ showed a broad D-band in the range of 530–600 cm⁻¹, which can be assigned to the oxygen vacancies/defect sites originated by the non-stoichiometry of the sample. The I_D/I_F2g values were calculated from the Raman spectra of the CeO₂ and Co–CeO₂ samples. These results show that the Co–CeO₂ sample possessed a higher value (0.37) than pure CeO₂ (0.02), which reflects abundance of the oxygen vacancies in the former sample. Five active modes (A_1g (674 cm⁻¹), F_2g (610, 515, and 191 cm⁻¹), and E_g (467 cm⁻¹) bands) were observed in the Raman spectrum of bulk cobalt oxide, suggesting a Co₃O₄ spinel structure with Co²⁺ and Co³⁺ located at tetrahedral and octahedral sites, respectively (ESI, Fig. S1†).²⁶ These characteristic peaks of Co₃O₄ spinel were not observed in Co–CeO₂ due to the broad nature of the peak.

Fig. 2 Raman spectra of pure CeO₂ and Co–CeO₂.

The HR-TEM images of the bulk Co₃O₄ and Co–CeO₂ catalysts are presented in Fig. 3. Bulk Co₃O₄ showed a truncated octahedral morphology with particle sizes of 45–60 nm (Fig. 3(A)). The lattice resolved HR-TEM image of bulk Co₃O₄ (Fig. 3(B)) showed predominant exposure of the (111) crystal plane with 0.467 nm d-spacing.²⁷ On the contrary, the Co–CeO₂ catalyst showed a mixed morphology of nanorods and cube (Fig. 3(C)). The estimated length of the nanorods was 27 ± 1 nm, with an average diameter of 6.6 nm. The average particle size of the nanocubes was 7 ± 1 nm. The lattice resolved HR-TEM image of Co–CeO₂ indicates predominant exposure of the (111) planes of CeO₂ with a d-spacing of 0.312 nm (Fig. 3(D)). In addition, a highly disordered ceria surface with a d-spacing of 0.19 nm (220) was observed at the interface of Co₃O₄, which is indicated by the dotted red circle in Fig. 4. This disordered surface indicates the presence of defects in the ceria lattice, which are induced by cobalt oxide. Previous studies have shown that the (220) planes of ceria are active surfaces.²⁸ Fig. 4 shows a lattice spacing of 0.33 nm, which corresponds to the interplanar separation between the (112) lattice planes of Co₃O₄.²⁹ However, the XRD results of Co–CeO₂ revealed the absence of a Co₃O₄ peak, which might be due to the fine distribution of cobalt oxide over the ceria support. Previous results showed that the high-index (112) crystal planes of cobalt oxide were more exposed than the (001), (011), and (111) planes on the surface of the catalyst.³⁰ Thus, cobalt-based catalysts exposed with (112) planes have more reactive surface compared to those which have (001), (011) or (111) planes of Co₃O₄.

Fig. 3 HR-TEM images of bulk Co₃O₄ (A), (B) and Co–CeO₂ (C), (D).

Fig. 4 HR-TEM images of Co–CeO₂ with highly reactive (112) planes of cobalt oxide.

The surface composition and chemical states of the atoms of the cobalt-doped ceria catalyst were analyzed using XPS. The Ce 3d XPS spectra of Co–CeO₂ and pure CeO₂ are displayed in Fig. 5(A). The Ce 3d spectra indicate the co-existence of Ce³⁺ and Ce⁴⁺ ions in both samples (Ce³⁺ = dotted line and Ce⁴⁺ = solid line). Concentrations of Ce³⁺ ions in those samples were calculated by integrating the peak area of each component, which revealed that the cobalt-doped ceria sample had a higher concentration of Ce³⁺ ions (0.27) compared to pure ceria (0.23). This higher concentration was likely caused by the transfer of electrons from cobalt to ceria, which increases the concentration of Ce³⁺ ions in the Co–CeO₂ catalyst according to the following reaction:

2CeO₂ + 2Co₃O₄ → 3Co₂O₃ + Ce₂O₃ (1)

Fig. 5 XPS spectra of (A) Ce 3d of Co–CeO₂ and pure CeO₂, (B) Co 2p, and (C) O 1s of Co–CeO₂ and bulk Co₃O₄ catalysts.

In accordance with this observation, a higher Co³⁺ ions concentration was expected for the Co–CeO₂ catalyst. The Co 2p_3/2 spectra of Co–CeO₂ and bulk Co₃O₄ are presented in Fig. 5(B).

The Co 2p_3/2 spectrum was deconvoluted into two main peaks, located at 779.5 ± 0.4 and 781.8 ± 1.2 eV, which were attributed to Co³⁺ and Co²⁺ ions, respectively.³¹ The presence of satellite peaks at 865 eV is further evidence of the presence of Co²⁺ species. The evaluation of the peak area of each component in the Co 2p_3/2 spectrum indicates that the Co–CeO₂ sample had a relatively higher Co³⁺ ion concentration (Co³⁺ = 63%) than bulk Co₃O₄ (Co³⁺ = 59%), which is in accordance with our expectations.

The binding energy (BE) of the Co–O bond was obtained by comparing the O 1s spectra of Co–CeO₂ and bulk Co₃O₄. The O 1s spectra of both samples are presented in Fig. 5(C). Peak-fitting of the O 1s region in Fig. 5(C) illustrated that three different types of oxygen were present in Co–CeO₂ and bulk Co₃O₄. The first peak in the O 1s spectrum of bulk cobalt oxide (at 529.6 eV) was assigned to the lattice oxygen attached to cobalt (Co–O). The second peak (at 531.1 eV) was attributed to the oxygen present in the surface adsorbed –OH groups. Finally, the third peak (at 532.2 eV) was attributed to the multiplicity of physisorbed and chemisorbed water on or near the surface of Co₃O₄.³² It is worth noting that the BE of lattice oxygen in the Co–CeO₂ sample shifted to a lower value (529.2 eV) compared to that in bulk Co₃O₄ (529.6 eV). The lower BE of the lattice oxygen increased the oxygen availability at the reaction site, thereby enhancing the rate of reaction.

Catalyst performance in the HT-WGS reaction

The catalytic efficiencies of Co–CeO₂ and bulk Co₃O₄ were studied to investigate the doping effect of cobalt with ceria on the catalytic activity in the HT-WGS reaction. Fig. 6(A) and (B) show the CO conversion as a function of temperature and time-on-stream (TOS) at 400 °C, respectively, in Co–CeO₂ and bulk Co₃O₄. At 350 °C, both catalysts exhibited almost the same CO conversion (∼18%). However, at higher temperatures (>350 °C), Co–CeO₂ showed higher catalytic activity than bulk Co₃O₄. The lower activity of bulk Co₃O₄ at higher temperature can be attributed to the decrease of the surface area from 46 to 10 m² g⁻¹ or formation of carbonaceous residues over the catalyst surface, which decreased the catalytic activity.³³ However, the notable performance of the Co–CeO₂ could be related to the increased number of oxygen vacancies, which in turn increased the reducibility of the catalyst. This conclusion was also supported by Raman and H₂-TPR result. The bulk Co₃O₄ showed a lower surface area (46 m² g⁻¹) and exposed with (111) crystal planes on the surface. However, doping cobalt oxide into ceria increased the surface area (122 m² g⁻¹) and inducing the exposure of well-defined reactive faces between the ceria (220) and cobalt oxide (112) interface. These findings suggest the strong interaction between them. The high-index (112) planes possess more Co³⁺ ions compared to the (111) planes of cobalt oxide.³⁴ Because the Co³⁺ ion is regarded as a more active site than Co²⁺,³⁵ thus, higher density of the Co³⁺ ions in the Co–CeO₂ catalyst might be responsible for the higher catalytic performance in the WGS reaction.

Fig. 6 CO conversion (A) as a function of temperature and (B) time-on-stream over Co–CeO₂ and bulk Co₃O₄ catalysts.

The higher reducibility of Co–CeO₂ facilitates CO adsorption on the catalyst surface with concomitant oxidation by surface oxygen. This process creates oxygen vacancies on the catalyst surface. These oxygen vacancies are replenished by the oxygen from water dissociation, which accompanies the generation of H₂. Since the WGS reaction is an exothermic reaction, CO conversion decreased in both catalysts at higher temperatures (>450 °C); however, the Co–CeO₂ catalyst showed better activity than bulk Co₃O₄ even at high temperatures. The selectivity to CO₂ and CH₄ in both catalysts were measured to be 100 and 0%, respectively, over the entire temperature range (see ESI, Fig. S2†).

This result illustrates the potential of utilizing Co–CeO₂ catalysts in the WGS reaction. Co–CeO₂ catalyst do not have any side reactions and result in almost 0% H₂ loss in methanation. Additionally, catalytic stability tests of the Co–CeO₂ and bulk Co₃O₄ catalysts are presented in Fig. 6(B), at a temperature of 400 °C and a GHSV of 143 000 h⁻¹. The Co–CeO₂ catalyst exhibited stable performance with 90% CO conversion during a 30 h WGS reaction. After 30 h, the catalytic activity fluctuated, and CO conversion started to decline (from 90 to 85%). Conversion declined again after 40 h and finally reached a value of 76%. In contrast, bulk Co₃O₄ displayed severe deactivation within 10 h (25% loss compared to the initial CO conversion). This deactivation might be due to the catalyst sintering/coke formation during the time-on-stream study. The BET analysis results of spent catalysts show that the surface area decreases for both catalysts after the reaction. However, bulk Co₃O₄ exhibited 78%, while Co–CeO₂ showed a 50% loss in surface area (see ESI, Table S1†). Indeed, the above decline in activity and surface area underscores the necessity of integrating Co₃O₄ with ceria in order to retain both high activity and stable performance during the HT-WGS reaction. A plausible mechanistic pathway for the WGS reaction over a Co–CeO₂ catalyst was shown in Scheme 1. A bi-functional redox mechanism for the WGS reaction has been reported. In case of Co–CeO₂ catalyst both cobalt and ceria equally participate in the reaction. The CO molecule is activated by the cobalt and oxidized to CO₂, followed by the generation of an oxygen vacancy. The oxygen vacancy is expected to be the active sites for the water dissociation, which produced the H₂ and replenish the oxygen.³⁶

Scheme 1 Proposed mechanism for the WGS reaction over a Co–CeO₂ catalyst.

Conclusions

Doping cobalt into ceria was found to be effective for enhancing the catalytic activity and stability in the HT-WGS reaction. The higher activity of the Co–CeO₂ catalyst relative to bulk Co₃O₄ was mainly due to the strong cobalt–ceria interaction. This led to the formation of a large number of oxygen vacancies, an increase in lattice strain, and enhancement of the reducibility of the catalyst. In addition, the (112) plane of Co–CeO₂ which contains a large amount of Co³⁺ active sites, is also responsible for the higher activity. The TOS results showed that the Co–CeO₂ catalyst was more stable than bulk Co₃O₄. The strong interaction between cobalt and ceria might be responsible for stabilizing cobalt against sintering during the WGS reaction.

Experimental

Catalyst preparation

The Co–CeO₂ catalyst was prepared using a co-precipitation method. In a typical preparation of Co (15 wt%)–CeO₂ (85 wt%) catalyst, 0.04 M of nitrate solution of cobalt and 15% KOH solution were simultaneously added drop-wise into 0.04 M Ce(NO₃)₃·6H₂O solution in a 1 L round-bottom flask. This was placed in a heating mantle and fitted with a pH electrode. During preparation, temperature of the flask was maintained at 80 °C at a constant pH of 10.5. The as-formed suspension was stirred and kept at 80 °C for 24 h. The end product was washed, collected, and dried in air in an oven at 100 °C for 12 h. The dried material was crushed into powder and annealed in a furnace at 500 °C for 5 h. The obtained final products were used for the HT-WGS reaction.

Catalyst characterization

X-ray diffractograms of the catalysts were recorded in the 2θ range of 20–80° using a Rigaku D/MAX-IIIC diffractometer (Ni-filtered Cu-Kα radiation, 40 kV, 100 mA). The crystallite size and lattice strain were estimated using the Williamson–Hall procedure, where the peak width, (β), is expressed as:

β cos θ = 2ε sin θ + 0.9 λ/D

here, ε is the lattice strain and D is the average crystallite size. By plotting β cos θ against sin θ for a number of XRD peaks at different 2θ values, the average crystallite size (D) and lattice strain (ε) were determined from the intercept and the slope of the linear regression to the data, respectively. The unit cell parameters were estimated by employing standard indexation methods using the intensity of the high 2θ peaks (111). The BET surface area was determined by the N₂ adsorption/desorption method at 77 K using an ASAP 2010 Micromeritics. Hydrogen-temperature programmed reduction (H₂-TPR) experiments were conducted on an Autochem 2910 (Micromeritics). Typically, 0.1 g of sample was loaded into a quartz reactor. H₂-TPR was performed using 10% H₂ in Ar with a heating rate of 10 °C min⁻¹ from room temperature to 800 °C. The sensitivity of the detector was calibrated by reducing a known weight of NiO. The Raman measurements were performed on a confocal micro-Raman spectrometer (model LabRAM ARAMIS Horiba Jobin-Yvon) with a laser excitation wavelength of 532 nm. XPS measurements were performed on a Kα Themo-Scientific spectrometer using a high-resolution monochromated Al Kα X-ray source. All of the measured spectra were corrected by setting the reference binding energy of carbon (1s) at 284.6 eV. HR-TEM images were obtained via a JEOL JEM-2100F microscope. All of the catalysts were reduced in 5% H₂/N₂ at 400 °C for 1 h before XRD, XPS, and Raman measurements.

Catalyst activity measurement

Catalyst activity tests were performed between 350 and 550 °C at atmospheric pressure in a fixed-bed micro-tubular quartz reactor with an inner diameter of 4 mm. 0.03 g of catalyst was charged into the reactor, and quartz wool was used to hold the catalyst bed in a fixed position. MgAl₂O₄ was used as a diluent (0.3 g) to increase the bed height in order to eliminate the effects of axial dispersion. A t-union was employed to install a thermocouple at the exit of the quartz reactor. Prior to each catalytic measurement, the catalyst was reduced in 5% H₂/N₂ from room temperature to 400 °C at a heating rate of 4.6 °C min⁻¹, and then the temperature was maintained for 1 h. Afterward, the temperature was decreased to 350 °C. The composition of the simulated reformed gas is 17.02 vol% CO, 9.55 vol% CO₂, 1.03 vol% CH₄, 13.14 vol% H₂, 55.20 vol% H₂O, and 4.06 vol% N₂. The feed H₂O/(CH₄ + CO + CO₂) ratio was intentionally fixed at 2.0 to avoid coke formation. A GHSV of 143 000 h⁻¹ was used to screen all of the catalysts. Water was fed using a syringe pump and was vaporized at 180 °C upstream of the reactor. The outlet gas was chilled, passed through a trap to condense residual water, and then the reactant and the product were analyzed online using an Agilent micro-gas chromatograph. No appreciable WGS activity was observed in the reactor filled only with quartz wool and MgAl₂O₄ under reaction conditions.

Acknowledgements

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2013R1A1A1A05007370).

References

1. K. Mudiyanselage, S. D. Senanayake, L. Feria, S. Kundu, A. E. Baber, J. Graciani, A. B. Vidal, S. Agnoli, J. Evans, R. Chaang, S. Axnanda, Z. Liu, J. F. Sanz, P. Liu, J. A. Rodriguez and D. J. Stacchiola, Angew. Chem., Int. Ed., 2013, 52, 5101.
2. L. Y. Gan and Y. J. Zhao, J. Phys. Chem. C, 2012, 116, 16089.
3. M. Kusche, K. Bustillo, F. Agel and P. Wasserscheid, ChemCatChem, 2015, 7, 766.
4. F. Meshkani and M. Rezaei, RSC Adv., 2015, 5, 9955.
5. Y. Ye, L. Wang, S. Zhang, Y. Zhu, J. Shan and F. Tao, Chem. Commun., 2013, 49, 4385.
6. P. Djinović, J. Batista and A. Pintar, Appl. Catal., A, 2008, 347, 23.
7. S. D. Senanayake, J. Evans, S. Agnoli, L. Barrio, T. L. Chen, J. Hrbek and J. A. Rodriguez, Top. Catal., 2011, 54, 34.
8. J. H. Lin and V. V. Guliants, ChemCatChem, 2012, 4, 1611.
9. M. S. Batista, E. I. Santiago, E. M. Assaf and E. A. Ticianelli, J. Power Sources, 2005, 145, 50.
10. X. Xie, Y. Li, Z. Q. Liu, M. Haruta and W. Shen, Nature, 2009, 458, 746.
11. Y. Li and W. Shen, Chem. Soc. Rev., 2014, 43, 1543.
12. J. Paier, C. Penschke and J. Sauer, Chem. Rev., 2013, 113, 3949.
13. J. A. Rodriguez, J. C. Hanson, D. Stacchiola and S. D. Senanayake, Phys. Chem. Chem. Phys., 2013, 15, 12004.
14. J. B. Park, J. Graciani, J. Evans, D. Stacchiola, S. Ma, P. Liu, A. Nambu, J. F. Sanz, J. Hrbek and J. A. Rodriguez, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 4975.
15. S. Royer and D. Duprez, ChemCatChem, 2011, 2, 24.
16. C. Sun, H. Li and L. Chen, Energy Environ. Sci., 2012, 5, 8475.
17. D. Jampaiah, K. M. Tur, S. J. Ippolito, Y. M. Sabri, J. Tardio, S. K. Bhargava and B. M. Reddy, RSC Adv., 2013, 3, 12963.
18. I. I. Soykal, H. Sohn and U. S. Ozkan, ACS Catal., 2012, 2, 2335.
19. F. Zhao, Z. Liu, W. Xu, S. Yao, A. Kubacka, A. C. Johnston-Peck, S. D. Senanayake, A.-Q. Zhang, E. A. Stach, M. F. García and J. A. Rodriguez, J. Phys. Chem. C, 2014, 118, 2528.
20. W. J. Xue, Y. F. Wang, P. Li, Z. T. Liu, Z. P. Hao and C. Y. Ma, Catal. Commun., 2011, 12, 1265.
21. K. Li, H. Wang, Y. Wei and D. Yan, Appl. Catal., B, 2010, 97, 361.
22. J. Y. Luo, M. Meng, X. Li, X. G. Li, Y. Q. Zha, T. D. Hu, Y. N. Xie and J. Zhang, J. Catal., 2008, 254, 310.
23. A. Martínez-Arias, M. Fernández-García, O. Gálvez, J. M. Coronado, J. A. Anderson, J. C. Conesa, J. Soria and G. Munuera, J. Catal., 2000, 195, 207.
24. S. Agarwal, X. Zhu, E. J. M. Hensen, L. Lefferts and B. L. Mojet, J. Phys. Chem. C, 2014, 118, 4131.
25. B. M. Reddy, L. Katta and G. Thrimurthulu, Chem. Mater., 2010, 22, 467.
26. J. Liu, Z. Zhao, J. Wang, C. Xu, A. Duan, G. Jiang and Q. Yang, Appl. Catal., B, 2008, 84, 185.
27. A. Jha and C. V. Rode, New J. Chem., 2013, 37, 2669.
28. N. Qiu, J. Zhang and Z. Wu, Phys. Chem. Chem. Phys., 2014, 16, 22659.
29. J. Zhang, H. Kumagai, K. Yamamura, S. Ohara, S. Takami, A. Morikawa, H. Shinjoh, K. Kaneko, T. Adschiri and A. Suda, Nano Lett., 2011, 11, 361.
30. L. Hu, Q. Peng and Y. Li, J. Am. Chem. Soc., 2008, 130, 16136.
31. L. Óvári, S. K. Calderon, Y. Lykhach, J. Libuda, A. Erdőhelyi, C. Papp, J. Kiss and H. P. Steinrück, J. Catal., 2013, 307, 132.
32. A. Serov, N. I. Andersen, A. J. Roy, I. Matanovic, K. Artyushkova and P. Atanassov, J. Electrochem. Soc., 2015, 162, F449.
33. D. Gu, C. J. Jia, C. Weidenthaler, H. J. Bongard, B. Spliethoff, W. Schmidt and F. Schuth, J. Am. Chem. Soc., 2015, 137, 11407.
34. K. Song, E. Cho and Y.-M. Kang, ACS Catal., 2015, 5, 5116.
35. D. Su, S. Dou and G. Wang, Sci. Rep., 2015, 4, 1.
36. J. Vecchietti, A. Bonivardi, W. Xu, D. Stacchiola, J. J. Delgado, M. Calatayud and S. E. Collins, ACS Catal., 2014, 4, 2088.

---

Footnote

† Electronic supplementary information (ESI) available: Raman spectra of Co₃O₄, product selectivity results, surface area information of catalysts before and after reaction. See DOI: 10.1039/c5ra22704f

---