X-ray diffractograms of the as-synthesized (calcined) and reduced materials (Figures S1 and S3, respectively) show that CuO (or Cu2O) and CaeO2 do not form solid solutions, confirming the high chemical stability of this mixture. Electrical conductivity measurements of the oxygen carriers and the references CuO and CeO2 reveal two distinct charge transport regimes, viz. a low temperature and a high temperature regime. At low temperatures (T < 450 ºC) the conduction in CuO is due to the hopping of charge carriers in metal deficient CuO, i.e. Cu1 − yO. Jeong and Choi determined the activation energy of charge carrier hopping in CuO as 0.1 ± 0.01 eV which is in good agreement with the values obtained here (0.13 eV – 0.15 eV). On the other hand, at low temperatures (T < 250 ºC) conduction in pure CeO2 is due to the hopping of polarons with an activation energy of ~ 0.40 eV, a value that agrees very well with the value of 0.38 eV determined in this work. At high temperatures (T > 550 ºC), the activation energy for conduction in unsupported CuO was determined previously as 0.7 ± 0.04 eV. This value is higher than our measurements (0.42 eV) and can most likely be attributed to morphological differences between commercial CuO and the material synthesized here. The activation energy for electrical conduction in CeO2 at high temperatures is a function of grain size and has been determined as 0.99 eV, 1.35 eV and 2.80 eV for 10 nm, 30 nm and 5 µm sized grains. The pristine CeO2 studied here had a grain size of ~ 100 nm and revealed an activation energy of 1.37 eV.
Turning now to the charge transport measurements of the synthesized oxygen carriers, Fig. 3(b) shows that at the typical operating temperatures of the CLC process (i.e. 800 ºC – 1000 ºC) the activation energy for charge transport in pure CuO and Cu60Ce is identical (0.42 eV) and ~ 3 times lower than in CeO2. EDX mapping (Fig. 1) and FIB-SEM tomography (Fig. 2) revealed that in Cu60Ce, CuO and CeO2 a percolation network is formed (sketched schematically in Fig. 6). The fact that the activation energy for charge transport in Cu60Ce and pure CuO is identical, suggests that charge transport occurs through CuO conduction bridges forming a conduction pathway with a low energy barrier. When the quantity of CuO is reduced to 30 wt. % the overall conductivity of the materials decreases (Fig. 3(a)). However, at the same time the activation energy for charge transport increases by only 0.11 eV (Fig. 3(b)), indicating that charge transport in Cu30Ce occurs still predominantly through CuO conduction pathways. Although the EDX map of Cu30Ce shows that at the particle surface the CuO grains are not connected with each other, our conductivity data suggest that CuO conduction bridges still exist in Cu30Ce, albeit being less effective than in Cu60Ce. Finally, when the CuO content in the material was decreased to 20 wt. %, the activation energy for conduction increased appreciably (from 0.53 eV to 0.80 eV) accompanied by a substantial decrease in conductivity. The sharp decrease in conductivity indicates that the quantity of CuO in Ce20Ce is below its percolation threshold, as confirmed by EDX mapping (Fig. 1) and FIB-SEM tomography (Fig. 2), and sketched in Fig. 6. As a consequence, in Cu20Ce charge transport occurs predominately through CeO2 conduction pathways, yielding in turn a high activation energy for charge transport (0.80 eV). Our conductivity measurements reveal that the percolation threshold of CuO is between 20–30 wt. %. The minimum quantity of CuO required to form a percolation network can be estimated using percolation theory, developed to determine the conductivity threshold in a binary composite of conducting and insulating particles. According to Kryuchkov, the critical volume fraction of the conducting material (here CuO) depends on the ratio of the size of the conducting (i.e., CuO) and non-conducting particles (here CeO2) and can be calculated according to:
Here, di is the diameter of the insulating particles (i.e., CeO2) and dc is the diameter of the conducting material (i.e., CuO). The size of pure CuO and CeO2 particles was ~ 1.24 µm and ~ 0.11 µm, respectively, leading to a threshold value of 21.1 vol. % equivalent to 18.2 wt. % CuO. This value is at the lower side of our experimentally determined percolation threshold in the range 20–30 wt. % CuO. The difference between the experimentally and theoretically estimated percolation threshold can be explained by the difference in the particles size of pure CeO2 (~ 0.11 µm) and CeO2 present in the oxygen carriers (average particle size ~ 1.47 µm).
From the N2-TPR profiles of the synthesized materials (Fig. 4) it is evident that the rate of oxygen release increased with increasing CuO content in the material. However, the TPR and conductivity experiments were performed in N2 and air, respectively. Therefore, it is not possible to establish a relationship between the rate of oxygen release during N2-TPR and the activation energy for charge transport. Thus, in this work attempts were made to relate the rate of oxidation to the conductivity performance of the materials. Our results (Figures S8 and 5) show that the normalized rates of oxidation decrease in the following order: Cu60Ce > Cu30Ce > Cu20Ce. The oxidation of the oxygen carriers can be divided into two regions, viz. surface oxidation and bulk oxidation, in accordance with the finding of Chuang et al. Chuang et al. showed that the oxidation of Al2O3-stabilized Cu follows a shrinking core model. Since all materials possessed a low surface area (< 1 m2/g), it is likely that the surface oxidation occurs at identical conditions for all cermets, leading in turn to very similar rates of (surface) oxidation. Once an initial CuO layer is formed around the particle, the oxidation of the unreacted bulk Cu takes place either by the diffusion of gaseous oxygen through the CuO layer or by the transport of oxygen ions from the surface to the unreacted Cu in the bulk via CuO and/or CeO2 conduction bridges. It is worth mentioning here that all materials have a low pore volume (< 0.01 cm3/g), see Table 1, and that the rate of oxidation was not limited by intra-particle mass transfer (Figure S4). Therefore, we can assume that the oxidation of bulk Cu was not controlled by the diffusion of gaseous oxygen through CuO. Indeed, using DFT calculations and inert marker experiments, Li et al. have demonstrated numerically that solid-state electronic and ionic conduction influences the rate of the redox reactions to a larger extent than intra-particle diffusion of the reactive (or product) gas. As outlined above, CuO-based conduction pathways have a lower energy barrier for charge transport when compared to CeO2-based conduction pathways. Therefore, the energy barrier for solid-state conduction is lowest for Cu60Ce due to the formation of a percolation network, yielding in turn the highest rate of oxidation. Decreasing the quantity of CuO in the material reduces the connectivity of CuO-CuO bridges, leading to an increase in the energy barrier for the transport of oxygen ions. Due to the increasing activation energy for charge transport, the rate of bulk oxidation decreases with a decreasing quantity of CuO. Based on these observations, we can conclude that for the oxidation of CeO2-stabilized CuO (containing up to 60 wt. % CuO) the counter-diffusion of oxygen ions and electrons from the surface to the bulk are the rate-determining steps. It is worth mentioning here that the results of this study cannot be extrapolated to the unsupported CuO system as both CuO and Cu have very low Tammann temperatures of 526°C and 405°C, respectively, and, hence, sinter already during the first CLC cycle. As a result, the morphology of the unsupported CuO and Cu under reaction conditions is very different to that of supported Cu. Secondly, the electrical properties of unsupported metals/metal oxides are very different to that of supported metals and metal oxides. For example, the ionic conduction in pure Cu and pure CuO is negligible compared to electronic conduction. Therefore, we speculate that for pure Cu the rate of inward diffusion of oxygen ions from the surface to the bulk is significantly lower than that of supported Cu.