Fast oxygen ion migration in Cu–In–oxide bulk and its utilization for effective CO₂ conversion at lower temperature

Abstract

Efficient activation of CO₂ at low temperature was achieved by reverse water–gas shift via chemical looping (RWGS-CL) by virtue of fast oxygen ion migration in a Cu–In structured oxide, even at lower temperatures. Results show that a novel Cu–In₂O₃ structured oxide can show a remarkably higher CO₂ splitting rate than ever reported. Various analyses revealed that RWGS-CL on Cu–In₂O₃ is derived from redox between Cu–In₂O₃ and Cu–In alloy. Key factors for high CO₂ splitting rate were fast migration of oxide ions in the alloy and the preferential oxidation of the interface of alloy–In₂O₃ in the bulk of the particles. The findings reported herein can open up new avenues to achieve effective CO₂ conversion at lower temperatures.

---

Introduction

Anthropogenic emissions of greenhouse gases are regarded as a cause of global warming, which is expected to lead to severe future climate change. Particularly, carbon dioxide (CO₂) derived from fossil fuels and industrial processes has long presented strong effects that continue to raise global mean temperatures.^1,2 Therefore, development of CO₂ capture and utilization (CCU) technologies^2–9 represents an urgent task for reducing CO₂ emissions into the atmosphere and for establishing a sustainable carbon cycle. Synthesis of fuels using electricity generation from renewable resources and using CO₂, most notably sun-to-fuel (STF)⁵ and power-to-liquid⁶ processes, has been proposed as a means of developing CCU technologies. In fact, CO₂-based fuels are anticipated as a technology that can increasingly incorporate renewable energy into the mobility sector as an “e-fuel”.^8,9 For these fuel production technologies, the potential process route is a Fischer–Tropsch (FT) process using syngas, which includes hydrogen and carbon monoxide (CO) converted from captured CO₂.^2–4 As a method of efficient conversion of CO₂ to CO through this process, reverse water–gas shift (RWGS) is a promising reaction that uses renewable H₂.^10,11 Mallapragada et al. proposed that the STF route consisting of RWGS followed by the FT reaction has higher efficiency than other routes with direct CO₂ conversion by photosynthetic bacteria and biomass conversion.⁵ The RWGS is an equilibrium-limited reaction (eqn (1)) with an endothermic nature.

CO₂ + H₂ ↔ CO + H₂O, ΔH⁰₂₉₈ = 42.1 kJ mol⁻¹ (1)

This reaction requires high reaction temperatures and separation of gas products to gain high conversion and a suitable H₂/CO ratio in syngas. Furthermore, conventional catalytic RWGS processes include side reactions, as presented in eqn (2) and (3).

CO + 3H₂ ↔ CH₄ + H₂O, ΔH⁰₂₉₈ = −206.5 kJ mol⁻¹ (2)

CO₂ + 4H₂ ↔ CH₄ + 2H₂O, ΔH⁰₂₉₈ = −165.0 kJ mol⁻¹ (3)

The CH₄ produced via these side reactions makes gas separation more complex. Additionally, it loses energy because of the exothermic reaction. S.-C. Yang et al. reported a Cu/CeO_x catalyst¹² which demonstrates 100% CO selectivity at 573 K. However, at low temperatures, CO₂ conversion is low because RWGS is an equilibrium-limited endothermic reaction, which results in a much higher H₂/CO ratio of the obtained syngas than the suitable ratio (H₂/CO = 2) for fuel synthesis by the FT process. One solution for these shortcomings is using chemical-looping reverse water–gas shift (RWGS-CL).^13,14 The overall concept of the solar-driven fuel (chemical) synthesis process with RWGS-CL is presented in Fig. 1. Through this process, CO₂ is converted to CO in two separate steps: reduction and re-oxidation of a metal oxide as an oxygen storage material (OSM). First, the OSM is reduced by H₂ (reaction (4)). Subsequently, the reduced OSM is re-oxidized by CO₂ to generate pure CO (reaction (5)).

MO_x + δH₂ → MO_x−δ + δH₂O (4)

MO_x−δ + δCO₂ → MO_x + δCO (5)

Fig. 1 Concept of solar-driven fuels and chemical synthesis processes with RWGS-CL.

Actually, RWGS-CL has benefits deriving from the absence of undesirable side reactions, which drives the equilibrium to CO production, and simplified gas separation, which leads to its improved energy efficiency. In fact, Wenzel et al. investigated solar-to-syngas efficiency for RWGS-CL processes and reported that the energy demand for separation is reduced by 77% compared to conventional RWGS processes.¹³ In the RWGS-CL process, the development of the OSM is an indispensable factor to make this process feasible. The rates of reduction and oxidation, CO yield, and stability are emphasized as key factors affecting the OSM for the RWGS-CL cycles. Furthermore, the operation temperature is an important factor. Isothermal operation at lower temperatures is desired to reduce heat loss and to increase its feasibility.¹⁴ In recent years, great effort has been put forth for the development of OSMs with high redox properties for RWGS-CL.^14–19 In previous studies, perovskite-type oxides^14,15 and iron-based oxides^16–18 have been well reported. Perovskite-type oxides are some of the promising OSMs thanks to their high stability and redox properties.²⁰ Maiti et al. investigated a series of lanthanum-based perovskites and found that LaCa_0.4Fe_0.4Mn_0.6O₃ showed high CO₂ splitting rate at low temperatures (723–823 K).¹⁵ Iron-based oxides are other candidates of OSMs for RWGS-CL because of their high oxygen capacity. Cu, Co, and Mn co-doped ferrites^16,17 and Fe₂O₃ supported on ion-conductive oxides¹⁸ were developed, which showed high performance and stability for the RWGS-CL cycle at 923 K. However, in these OSMs, isothermal operations have been investigated at temperatures above 823 K, and the performance below 773 K is unclear. Although Utsis et al. reported isothermal RWGS-CL at 623–723 K using Fe-substituted Ba-hexa-aluminates,¹⁹ the CO₂ splitting capacity was 0.4 mmol g⁻¹, so higher performance of CO₂ splitting is desired. This study has revealed that an indium-based oxide can be a novel OSM material able to exhibit extremely high isothermal RWGS-CL performance at low temperatures (673–773 K). Particularly, the performance of Cu-modified In₂O₃ formed from Cu₂In₂O₅ as a parent material was found to be promising for this purpose. Characterization of OSMs was conducted using powder X-ray diffraction (XRD), a field emission transmission electron microscope equipped with an energy-dispersive X-ray spectrometer (STEM-EDX), X-ray photoelectron spectroscopy (XPS), and in situ X-ray absorption fine structure (XAFS) measurements.

Results and discussion

Isothermal RWGS-CL performance

After applying various oxides for RWGS-CL, we found that the Cu–In mixed oxide has high potential for RWGS-CL. The CO₂ splitting performance obtained in this study is presented in Fig. 2 and Table S1† along with results described in earlier reports of the literature. Compared to other perovskite-type oxides^14,15 and ferrites,^16,17 which were reported earlier, Cu–In₂O₃ showed higher CO₂ splitting rates and sufficient amounts of CO₂ splitting even at low temperatures such as 673–773 K. Among all the oxides, Cu–In₂O₃ derived from Cu₂In₂O₅ is the most promising material for low-temperature isothermal RWGS-CL. The cycling performance of isothermal RWGS-CL on Cu₂In₂O₅ at 773 K is presented in ESI Fig. S1.† The reduction and oxidation (redox) amounts are defined as the moles of oxygen released or restored per gram of Cu₂In₂O₅. In the first cycle, Cu₂In₂O₅ showed reduction of 9.50 mmol g⁻¹ and re-oxidation of 4.82 mmol g⁻¹. Then, in cycles 2–5, the reduction and the oxidation amounts were, respectively, 5.06 ± 0.41 mmol g⁻¹ and 4.80 ± 0.40 mmol g⁻¹. Excess reduction in the first cycle (4.68 mmol g⁻¹) was almost equivalent to the release of 2 mol oxygen atoms per mol of Cu₂In₂O₅. This finding suggests that Cu₂In₂O₅ has become Cu(0)–In₂O₃, which is confirmed later. Products of RWGS-CL were measured using a quadrupole mass spectrometer. The mass spectra of products are presented in ESI Fig. S2.† The products of reduction and oxidation steps were identified, respectively, as H₂O (m/z = 18) and CO (m/z = 28) without production of other by-products. Results confirmed that RWGS-CL on Cu–In₂O₃ formed from Cu₂In₂O₅ is a selective CO₂ conversion to CO. Structural characterization of this material was achieved using a multi-point BET method (results and discussion are presented in ESI Tables S2, S3 and ESI text†), powder XRD, STEM-EDX, and in situ XAFS measurements. The XRD patterns obtained from fresh and post-reaction samples are shown in Fig. 3A. The XRD pattern of the fresh sample showed the formation of Cu₂In₂O₅ (ICDD 01-070-1082) with no impurity phase. The XRD pattern of the reduced sample exhibited diffraction peaks assigned to Cu–In alloy and In₂O₃ (ICDD 01-089-4595). The main compositions of the alloy were Cu₇In₃ (ICDD 03-065-2249) and Cu₉In₄ (ICDD 00-042-1476) (Fig. 3B and S3†). From the phase diagram of the Cu–In system, these alloys are considered to be a solid phase during the RWGS-CL cycle.²¹ Then, after the oxidation step, diffraction peaks of the Cu–In alloy disappeared. Those of Cu metal (ICDD 01-071-4610) and In₂O₃ were observed. STEM-EDX images are presented in Fig. 3C, D and ESI Fig. S4.† The SEM micrographs of the reduced Cu–In₂O₃ (ESI Fig. S4A†) show the presence of ca. 1 μm particles. The elemental composition of the particles was identified using EDX mapping (copper, red; indium, green; oxygen, blue). Results of EDX mapping of the sample after reduction (Fig. 3C and ESI Fig. S4A†) show that indium is distributed evenly over the particles, whereas copper is localized and oxygen is scattered. These images indicate Cu–In alloy formation by reduction. For the re-oxidized sample, many particles with uniformly distributed copper, indium, and oxygen were observed as depicted in Fig. 3D and ESI Fig. S4B.† Using image processing and analysis of these EDX images with Python (see the ESI text and Fig. S5 for details†), it was estimated that 84% of In was overlapping with Cu, and 93% of Cu was overlapping with In. These results demonstrate that a fine matrix of Cu–In–O was formed in the re-oxidized particles. The Cu in re-oxidized particles was regarded as highly distributed on the surface or incorporated into the In₂O₃ structure. As additional information, in situ XAFS results are presented in Fig. 3E and F (see the ESI text, Table S4 and Fig. S6–S8 for details†). The sequential change of In K-edge XANES spectra indicates that the redox of In(III) ⇆ In(0) occurs during the RWGS-CL cycle. By contrast, the Cu K-edge XANES spectrum, which only slightly changed during the RWGS-CL cycle, represented the formation of the Cu–In alloy.^22,23 This structural information obtained from XRD, STEM-EDX, and XAFS measurements demonstrates that the RWGS-CL cycle on Cu₂In₂O₅, as a parent material, is based on the redox of indium with the formation and oxidation of the Cu–In alloy. This material during RWGS-CL is denoted hereinafter as Cu–In₂O₃.

Fig. 2 Average CO₂ splitting rates on the Cu–In mixed oxide (this study) and on various oxides in earlier reports.

Fig. 3 Structural characterization of Cu–In₂O₃ derived from Cu₂In₂O₅ during each step of the RWGS-CL cycle. (A) XRD diffraction patterns of Cu₂In₂O₅ as prepared, after reduction, and after oxidation. (B) Main diffraction peaks for the Cu–In alloy on the sample after reduction. EDX image of Cu–In₂O₃ after reduction (C) and re-oxidation (D) at 773 K (copper, red; indium, green; oxygen, blue). Cu and In K-edge XANES spectra of Cu–In₂O₃ during the reduction (E) and the re-oxidation (F).

To elucidate the role of Cu in Cu–In₂O₃, we compared the performance of isothermal RWGS-CL as a function of temperature on Cu–In₂O₃ (derived from Cu₂In₂O₅), 10 wt% Cu supported on In₂O₃ (10 wt% Cu/In₂O₃) and pure In₂O₃ without Cu, as shown in Fig. 4. Results indicate that Cu–In₂O₃ is feasible for conducting RWGS-CL, even at 673 K. The performance at each temperature was much higher than that of 10 wt% Cu/In₂O₃ and pure In₂O₃ without Cu. Actually, the amount of redox on 10 wt% Cu/In₂O₃ was slightly higher than that of In₂O₃, although the CO₂ splitting rate was improved only slightly. Results of structural characterization of 10 wt% Cu/In₂O₃ (ESI text and Fig. S9–S13† provide additional details) show that, because of the poor contact of Cu species with the In₂O₃ surface, 10 wt% Cu/In₂O₃ showed lower performance than Cu–In₂O₃ derived from Cu₂In₂O₅.

Fig. 4 Dependence of RWGS-CL performance of each oxide on the reaction temperature. Amounts of redox and average CO₂ splitting rate during the re-oxidation step on Cu–In₂O₃ (A), 10 wt% Cu/In₂O₃ (B), and In₂O₃ (C). Reduction was conducted in a 10% H₂ atmosphere for 30 min; oxidation was conducted in a 10% CO₂ atmosphere.

Kinetics investigations of oxides during the RWGS-CL cycle

Reduction step by H₂. Temperature-programmed reduction by H₂ (H₂-TPR) measurements, the results of which are presented in Fig. 5A, were performed to investigate the reducibility of oxides. The ESI presents additional details.† For Cu-containing oxides (Cu–In₂O₃ and 10 wt% Cu/In₂O₃), the H₂-TPR profiles showed three notable reduction areas (ca. 400–550 K, 580–700 K, and 700–920 K). In the moderate temperature region of 580–700 K, the amount of reduction was equivalent to about 26% of oxygen contained in Cu₂In₂O₅. Ploner et al. have observed that the Cu–In alloy formed at 640–723 K on the Cu–In₂O₃ system in an H₂ atmosphere using in situ XRD measurement.²⁴ Therefore, the reduction of In(III) with the formation of the Cu–In alloy, the composition of which finally approached Cu₉In₄, was regarded as proceeding in this moderate temperature region. Regarding the results of H₂-TPR measurements, results show that Cu species promote the reduction of indium oxide at low temperatures. To investigate more details of the mechanism, the kinetics model for the reduction of these oxides was verified using the method presented by Hancock and Sharp for the isothermal solid-state reaction.^25–27 The ESI text, Fig. S14–S17 and Tables S6–S8† present additional details. The kinetics models used for model fitting are shown in ESI Table S5.† As a result of the investigations, the phase-boundary-controlled reaction models (R2 and R3 models) were found to be suitable to describe the experimentally obtained data for all the oxides. These models are categorized as a classic shrinking core model in which the rate-controlling step is the surface chemical reaction.^26,27 The activation energy was investigated using the reaction rate constant k values obtained from the model fitting described above (ESI Table S9†). Plots of ln (k_red/min⁻¹) vs. 1/T (Arrhenius plot) for the three oxides are presented in Fig. 5B. The respective activation energies for Cu–In₂O₃ (45.2 kJ mol⁻¹) and 10 wt% Cu/In₂O₃ (53.4 kJ mol⁻¹) were much lower than that for In₂O₃ (76.6 kJ mol⁻¹), confirming that the Cu species supported on In₂O₃ decreased the activation barrier for reactions between the surfaces of In₂O₃ and H₂. The slight difference in the activation energy between Cu–In₂O₃ and 10 wt% Cu/In₂O₃ suggests that the effect of the Cu species on the reduction does not depend much on these morphologies.

Fig. 5 Kinetics analysis of each oxide in the reduction step. (A) H₂-TPR profiles for Cu–In₂O₃ (derived from Cu₂In₂O₅), 10 wt% Cu/In₂O₃ and In₂O₃. (B) Arrhenius plots for the reduction of Cu–In₂O₃ (derived from Cu₂In₂O₅), 10 wt% Cu/In₂O₃ and In₂O₃.

Oxidation step by CO₂ for CO formation. Next, kinetics investigations were conducted for the oxidation step by CO₂ in the RWGS-CL cycle. The kinetics model for the oxidation of the oxide was also verified using the Hancock and Sharp method. As a consequence of the model fitting (details of which are presented in ESI Fig. S18–S21 and Tables S10–S12†), the most suitable models were concluded to be a zero-order model (R1 model)^25–27 for Cu–In₂O₃ and the nucleation model, known as the Avrami–Erofe'ev model (AE1 model), for the other two oxides.^28–30Fig. 6A presents the Arrhenius plot for the oxidation of these three oxides. The reaction rate constant k values obtained from model fitting are presented in Table S13.† The activation energy for Cu–In₂O₃ is 50.7 kJ mol⁻¹, which is markedly lower than that for either of the other two oxides. As the results of the model fitting show, the large difference in the activation energy is attributable to the different mechanisms of oxidation among these oxides. Considering that AE1 is suitable as a model for In₂O₃ and 10 wt% Cu/In₂O₃, the overall oxidation of these two oxides is determined by the rate of formation and growth of the indium oxide nuclei. At the beginning of the reaction, the rate of oxidation is high because of random nucleation on the surface of the reduced oxide. However, as the nucleation and growth of nuclei proceed, the rate of oxidation plummets because of the decrease of the reduced surface. As a result, a long time is necessary to complete oxidation by CO₂. Fig. 6B presents different rates of CO₂ splitting over these oxides during oxidation by CO₂. For all oxides, the degree of reduction before re-oxidation was fixed at 3.0 mmol g⁻¹. On In₂O₃ and 10 wt% Cu/In₂O₃, the CO₂ splitting rate decreased drastically with the progress of oxidation. The oxidation of Cu–In₂O₃ proceeded while maintaining a constant CO₂ splitting rate of about 340 μmol g⁻¹ min⁻¹ until conversion reached 60%. As a result, re-oxidation of the reduced Cu–In₂O₃ is completed twice as rapidly as that of the other two oxides. These results obtained from the kinetics investigation of Cu–In₂O₃ suggest that the amount of active sites for CO₂ splitting remains constant while oxidation is proceeding. The oxidation state of the Cu–In₂O₃ surface during the oxidation step was analyzed by XPS. Compared with the In 3d_5/2 peak of fully re-oxidized Cu–In₂O₃, that of reduced Cu–In₂O₃ was shifted to the lower binding energy (details in ESI Fig. S22 and ESI text†), which demonstrates the existence of In(0) species on the surface.³¹Fig. 6C shows the surface In/O ratio at each oxidation rate of the reduced Cu–In₂O₃ and 10 wt% Cu/In₂O₃, as measured by XPS. The In/O ratio of the Cu–In₂O₃ surface remained above 0.9, even at an oxidation rate of 75%, which indicates that bulk oxidation proceeds preferentially while the surface remains in a highly reduced state. This particular CO₂ splitting mechanism of Cu–In₂O₃ is explainable as shown in Fig. 6D. First, CO₂ splits on the surface of the Cu–In alloy; CO and oxide ions (O²⁻) are generated. Subsequently, the O²⁻ ions migrate to the bulk of the particle. They presumably oxidize the Cu–In alloy–In₂O₃ interface. As a result, the alloy surface, as the active site, remains in a highly reduced state. Rapid CO₂ splitting proceeds continuously.

Fig. 6 Kinetics analysis of each oxide in re-oxidation steps. (A) Arrhenius plots for the oxidation of Cu–In₂O₃ (derived from Cu₂In₂O₅), 10 wt% Cu/In₂O₃ and In₂O₃. (B) Dependence of differential CO₂ splitting rate on conversion of oxides. (C) Surface In/O ratios on Cu–In₂O₃ and 10 wt% Cu/In₂O₃ during the re-oxidation step. (D) Presumed mechanism of CO₂ splitting on the reduced Cu–In₂O₃.

Conclusion

As reported herein, Cu–In₂O₃ was synthesized. Then its performance for RWGS-CL was investigated at the low temperatures of 673–773 K. Results of RWGS-CL cycle tests demonstrated that, even at low temperatures, Cu–In₂O₃ exhibited much higher CO₂ splitting performance than ever reported. Results of XRD, STEM-EDX, and in situ XAFS measurements showed that the oxide has a structure of reduced Cu supported on In₂O₃ (Cu–In₂O₃) under the RWGS-CL conditions, and showed that its redox properties are based on Cu–In alloy formation and re-oxidation. In addition, because low performance was exhibited by pure In₂O₃ and Cu supported In₂O₃ that were prepared using an impregnation method, the structured state of Cu–In₂O₃ has extremely important roles in its high RWGS-CL performance. H₂-TPR and kinetics investigations revealed that Cu–In₂O₃ shows high reducibility, even at low temperatures. We concluded that Cu species promote the reaction of H₂ with the oxide surface. Results of kinetics investigations in the oxidation step indicate that the Cu–In alloy particle surface conserved a highly reduced state even under oxidation condition, resulting in rapid completion of re-oxidation by CO₂. The interesting oxidation behavior is attributed to rapid O²⁻ migration from the surface to the bulk of the Cu–In alloy and preferential oxidation of the interface of alloy–In₂O₃. The combination of high reducibility and the specific re-oxidation mechanism engenders the high performance found for RWGS-CL on Cu–In₂O₃.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

In situ XAFS measurements were performed at the beamline BL07 of the SAGA Light Source (Proposal No. 1910093P/BL07). The authors thank Dr H. Setoyama for help with measurements of XANES spectra.

Notes and references

1. IPCC, www.ipcc.ch/site/assets/uploads/2018/02/SYR_AR5_FINAL_full.pdf.
2. S. J. Davis, K. Caldeira and H. D. Matthews, Science, 2010, 329, 1330–1333.
3. S. Perathoner and G. Centi, ChemSusChem, 2014, 7, 1274–1282.
4. J. Artz, T. E. Müller, K. Thenert, J. Kleinekorte, R. Meys, A. Sternberg, A. Bardow and W. Leitner, Chem. Rev., 2018, 118, 434–504.
5. D. S. Mallapragada, N. R. Singh, V. Curteanu and R. Agrawal, Ind. Eng. Chem. Res., 2013, 52, 5136–5144.
6. F. V. Vázquez, J. Koponen, V. Ruuskanen, C. Bajamundi, A. Kosonen, P. Simell, J. Ahola, C. Frilund, J. Elfving, M. Reinikainen, N. Heikkinen, J. Kauppinen and P. Piermartini, J. CO₂ Util., 2018, 28, 235–246.
7. I. Dimitriou, P. García-Gutiérrez, R. H. Elder, R. M. Cuéllar-Franca, A. Azapagic and R. W. K. Allen, Energy Environ. Sci., 2015, 8, 1775–1789.
8. M. Romero and A. Steinfeld, Energy Environ. Sci., 2012, 5, 9234–9245.
9. P. Kaiser, R. B. Unde, C. Kern and A. Jess, Chem. Ing. Tech., 2013, 85, 489–499.
10. Y. A. Daza and J. N. Kuhn, RSC Adv., 2016, 6, 49675–49691.
11. K. Oshima, T. Shinagawa, Y. Nogami, R. Manabe, S. Ogo and Y. Sekine, Catal. Today, 2016, 232, 27–32.
12. S.-C. Yang, S. H. Pang, T. P. Sulmonetti, W.-N. Su, J.-F. Lee, B.-J. Hwang and C. W. Jones, ACS Catal., 2018, 8, 12056–12066.
13. M. Wenzel, L. Rihko-Struckmann and K. Sundmacher, AIChE J., 2017, 63, 15–22.
14. Y. A. Daza, D. Maiti, R. A. Kent, V. R. Bhethanabotla and J. N. Kuhn, Catal. Today, 2015, 258, 691–698.
15. D. Maiti, B. J. Hare, Y. A. Daza, A. E. Ramos, J. N. Kuhn and V. R. Bhethanabotla, Energy Environ. Sci., 2018, 11, 648–659.
16. L. Ma, Y. Qiu, M. Li, D. Cui, S. Zhang, D. Zheng and R. Xao, Ind. Eng. Chem. Res., 2020, 59, 6924–6930.
17. Y. Qiu, L. Ma, D. Zeng, M. Li, D. Cui, Y. Lv, S. Zhang and R. Xiao, J. Energy Chem., 2020, 46, 123–132.
18. D. Zeng, Y. Qiu, L. Ma, M. Li, D. Cui, S. Zhang and R. Xiao, Environ. Sci. Technol., 2020, 54, 12467–12475.
19. N. Utsis, M. V. Landau, A. Erenburg and M. Herskowitz, Catalysts, 2020, 10, 1082.
20. X. Zhu, K. Li, L. Neal and F. Li, ACS Catal., 2018, 8, 8213–8236.
21. Z. Bahari, E. Dichi, B. Legendre and J. Dugué, Thermochim. Acta, 2003, 401, 131–138.
22. M. G. Kim, S. Sim and J. Cho, Adv. Mater., 2010, 22, 5154–5158.
23. W.-R. Lee, M. G. Kim, J.-R. Choi, J.-I. Park, S. J. Ko, S. J. Oh and J. Cheon, J. Am. Chem. Soc., 2005, 127, 16090–16097.
24. K. Ploner, L. Schlicker, A. Gili, A. Gurlo, A. Doran, L. Zhang, M. Armbrüster, D. Obendorf, J. Bernardi, B. Klötzer and S. Penner, Sci. Technol. Adv. Mater., 2019, 20, 356–366.
25. J. D. Hancock and J. H. Sharp, J. Am. Ceram. Soc., 1972, 55, 74–77.
26. K. Piotrowski, K. Mondal, H. Lorethova, L. Stonawski, T. Szymański and T. Wiltowski, Int. J. Hydrogen Energy, 2005, 30, 1543–1554.
27. Z. Zhou, L. Han and G. M. Bollas, Int. J. Hydrogen Energy, 2014, 39, 8535–8556.
28. M. Avrami, J. Chem. Phys., 1939, 7, 1103–1112.
29. M. Avrami, J. Chem. Phys., 1940, 8, 212–224.
30. M. Avrami, J. Chem. Phys., 1941, 9, 177–184.
31. J. Gao, F. Song, Y. Li, W. Cheng, H. Yuan and Q. Xu, Ind. Eng. Chem. Res., 2020, 59, 12331–12337.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental procedure, supporting data etc. See DOI: 10.1039/d0sc05340f

---