The Effect of the Support Structure and Size of Cu-based Oxygen Carriers on the Performance of Chemical Looping Air Separation

Chemical looping air separation (CLAS) is a novel and efficient method of producing high-purity oxygen because of its low-energy demands. Cu-based materials are suitable oxygen carriers (OCs) for CLAS. In the current study, Cu-based OCs in different particle sizes were prepared using various methods (viz., through mechanical mixing, impregnation, and coprecipitation) and different supporting materials (viz., ZrO2, SiO2, and Al2O3) and porosities (viz., in the mesopore range). This study evaluated the reactivity, reaction kinetics, and recyclability of these OCs by measuring their conversion rates for reduction (oxygen release) and oxidation (regeneration) in a thermogravimetric analyzer. CuO OCs on ZrO2 nanoparticles prepared through impregnation (CuZr-IM) exhibited almost complete conversion and the fastest reaction rates of all the OCs for reduction and oxidation. These characteristics are primarily attributable to the fine particles (100–250 nm) of the OCs. Furthermore, the CuO on the surface of the ZrO2 particles was distributed in a uniform pattern, as these fine particles displayed greater oxygen mobility and more rapid diffusion than the micrometer-sized particles paired with bulk materials. Kinetic analysis revealed that Avrami-Erofe’ev random nucleation and the subsequent growth reaction model with n = 2 (A2), with an observed activation energy of 140.2 kJ mol–1 in our study, is the optimal fitting for CuZr-IM OC conversion during reduction, and a long-term-stability test indicated that this OC is an appropriate candidate for CLAS.


INTRODUCTION
Oxy-fuel combustion is used for thermal power generation and as one of the carbon capture and storage technologies (Scheffknecht et al., 2011). Oxy-fuel combustion is the process of burning fuel using pure oxygen instead of air. It produces exhaust, which primarily comprises CO 2 and H 2 O. Almost pure CO 2 can be obtained by condensing water vapor. In the oxy-fuel combustion process, the required oxygen is produced through air separation (Smith and Klosek, 2001). A cryogenic air separation (CAS) process is the most common and commercially available method for large-scale, highvolume (> 100 t d -1 ), and high-purity O 2 (99.5-99.8%). However, the CAS method is complicated and highly energy-consuming because it involves the liquefaction of air and distillation of liquid air. The oxygen-specific power of the CAS process is 0.3-0.6 (kWh)ꞏ(m 3 O 2 ) -1 (Fan and Zhu, 2015). Moreover, it may cause an explosion resulting from rough handling in the CAS process. The other methods of producing oxygen are pressure swing adsorption (PSA) and Chemical looping with oxygen uncoupling (CLOU) is one operating mode of CLP (Mattisson et al., 2009;Imtiaz et al., 2013). The gas-phase oxygen produced by some specific metal oxides, such as CuO or Mn 2 O 3 , is directly reacted with solid fuels in order to produce nearly pure CO 2 in the reducer. Chemical looping air separation (CLAS), which is another operating mode of CLP, is a novel alternative for oxygen production proposed by Moghtaderi (2010). In the CLAS process, oxygen carriers (OCs) containing metal oxides are circulated between an oxidation reactor (oxidizer) and a reduction reactor (reducer). In the reducer, OCs were reduced to produce gas-phase oxygen (Eq. (1)). In order to obtain higher-purity oxygen, steam or CO 2 can be used as carrier gas in the reducer. The reduced OCs are then transported to an oxidizer. Fresh air is provided to the oxidizer to regenerate reduced OCs (Eq. (2)). CLAS is a cost-effective and energyefficient method to produce oxygen compared with CAS. The average oxygen-specific power of CLAS is approximately 0.04 (kWh)ꞏ(m 3 O 2 ) -1 .
The selection of suitable OCs is the key issue in CLAS. In previous thermodynamic evaluation, some metal oxides, such as Cu-, Mn-, Os-, and Co-based oxides, could release gas-phase oxygen at high temperatures (1100-1400 K) and were used as OCs for CLAS (Wang et al., 2013b). Among these metal oxides, CuO was the most suitable material for CLAS because of its higher oxygen transport capacity (0.1 (g O 2 )ꞏ(g CuO) -1 ) and higher reaction rates. However, the low melting point of Cu (1085°C) caused sintering and agglomeration, thus limiting its application at high temperature (Ku et al., 2017). Therefore, some inert materials, such as Al 2 O 3 , SiO 2 , TiO 2 , ZrO 2 , YSZ, and MgAl 2 O 4 , were used as supports for Cu-based OC for increasing the reactivity, stability, mechanical strength, and sinter resistance. SiO 2 is an appropriate support for CuO because SiO 2 does not react with CuO at the oxygen-releasing temperature (Shah et al., 2012). However, agglomeration is the main problem of SiO 2 support. Therefore, CuO/MgO-SiO 2 OCs were synthesized (Song et al., 2014a). The results revealed that MgO addition on SiO 2 support may suppress the agglomeration and increase the reactivity and stability during redox cycles due to the formations of Cu 2 MgO 3 stabilizing the structure of OCs. Kinetics of Cu-based OCs prepared with different support materials (ZrO 2 , TiO 2 , and SiO 2 ) for CLAS was studied by using thermogravimetry (TG) (Wang et al., 2013a). The results show that reaction rates of Cu-based OCs with different supports were as follows: ZrO 2 > TiO 2 > SiO 2 . It may be attributed to the fact that a higher surface area of ZrO 2 facilitates gas diffusion and heat transfer, thus increasing the reaction rate.
Different methods, such as mechanical mixing (MM), freeze granulation (FG), sol-gel (SG), impregnation (IM), coprecipitation (CP), and spray drying (SD), were used to prepare Cu-based OCs. CuO/Al 2 O 3 OCs with high CuO loadings were developed through a coprecipitation (CP) method (Imtiaz et al., 2012). It shows that the pH value influences the oxygen-carrying capacity and reactivity of OCs. CuO/Al 2 O 3 OCs synthesized at a high pH value (i.e., 12.5) resulted in OCs with a higher CuO content (87.8 wt.%) and possessed a higher oxygen capacity. CuO/MgO-SiO 2 OCs prepared using dry IM exhibit a higher stability than CuO/SiO 2 (Song et al., 2014a). The addition of MgO to CuO/SiO 2 and OCs prepared through IM method may suppress agglomeration.
Most studies on the preparation of Cu-based OCs for CLAS have focused on the MM method because scaling up for the production of OCs is simpler and easier compared to the other methods. However, the particle size of most OCs has been on the micrometer scale (Wang et al., 2014;Ku et al., 2018). Some researches have been conducted on the effect of support microstructure and the particle size of OCs for CLAS. The relationship between the conversion rate and the open porosity of supports was also studied (Liu et al., 2016). It has been reported that the high open porosity of the support is beneficial for gas transfer, thus increasing the reaction rate. CuO/Al 2 O 3 OCs derived from layered double hydroxide (LDH) precursors were prepared through CP method, and the OC was a two-dimensional (2D) nanostructured matrix (Song et al., 2013). It shows that the CuO nanocrystals (30-50 nm) were well dispersed in porous LDH supports. The reaction rates were faster in both reduction and oxidation reaction because of CuO nanocomposites, and porous structures. Moreover, agglomeration was not observed for this nanostructured OC.
The objective of the current study is to prepare the Cu-based OCs for CLAS operation. Cu-based OCs with different support materials (e.g., ZrO 2 , SiO 2 , and Al 2 O 3 ), pore structures, and particle sizes were prepared using various methods, such as IM, CP, and MM. The reactivity, kinetics, and recyclability of these Cu-based OCs for CLAS reaction has been analyzed from the TGA results. It may lead to a better understanding of the correlation between the conversion rate and the particle size and porosity of supports.

Preparation of Oxygen Carrier
In this study, Cu-based OCs with different structures and particle sizes were prepared using the following methods: IM, CP and MM. The content of CuO in the OCs is 40 wt.% in this study.

Impregnation (IM)
In this study, three Cu-based OCs were prepared using ZrO 2 , SiO 2 , and Al 2 O 3 as support materials. ZrO 2 support was prepared through modified sol-gel methods (Mokhtar et al., 2013). Zr precursors were prepared using an aqueous solution of ZrOCl 2 -8H 2 O (Solution A). Ammonia solution (28%) was mixed with cetyltrimethylammonium bromide and homogeneously stirred for 30 min (Solution B). Solution B was added dropwise into Solution A and the mixture was stirred in an ultrasonic bath for 2 h. NaOH was then slowly added to the solution until pH reached 9.0. The formed precipitate was transferred to an autoclave and heated at 85°C for 24 h. The precipitate was filtered and repeatedly washed using ethanol. The resulting sample was dried in an oven at 100°C for 4 h and calcined at 600°C for 6 h.
Mesoporous SiO 2 support was synthesized using the spray-drying (SD) method (Wang et al., 2012). Sodium silicate was added to deionized water and 6 N HCl. The mixture was stirred at room temperature for 30 min until a clean solution was obtained. In a separate beaker, Pluronic P123 was added to EtOH (95%) and thoroughly mixed. The Pluronic P123/EtOH solution was added to the sodium silicate mixture to obtain a precursor solution. The precursor was dried in a pilot-scale SD system. The drying temperature of a reaction tower for solvent removal from self-organized particles was 200°C. The resulting SiO 2 particles were collected using a cyclone and bag filter. The collected particles were finally calcined (600°C, 4 h) in a muffle furnace to eliminate the surfactant template.
Mesoporous Al 2 O 3 support was prepared through a modified sol-gel method. In a general synthesis, Pluronic P123 and anhydrous ethanol were mixed for 30 min (Solution A). Aluminum isopropoxide, HNO 3 solution, and anhydrous ethanol were mixed for 1 h (Solution B). Solution A was added dropwise into Solution B and the mixture was stirred for 1 h. The mixed sample was aged for 20 h at room temperature. The resulting sample was dried in an oven at 100°C for 4 h and calcined at 600°C for 6 h.
Three types of homogeneous Cu-based OCs were prepared through IM using ZrO 2 , SiO 2 , and Al 2 O 3 as supports and Cu(NO 3 ) 2 -3H 2 O as the copper precursor. In the IM procedure, the copper precursor solution was dissolved in deionized water and slowly added in the aforementioned support materials. The solution was then stirred for 1 h. The resulting sample was dried in an oven at 100°C for 12 h and calcined at 900°C for 3 h.

Coprecipitation (CP)
The OC was prepared through CP using Cu(NO 3 ) 2 -3H 2 O and ZrOCl 2 -8H 2 O as materials. Cu(NO 3 ) 2 -3H 2 O and ZrOCl 2 -8H 2 O were added in water and stirred for 1 h. Aqueous NaOH was gradually added until pH reached 9.0 and precipitation occurred. The resulting sample was dried in an oven at 80°C overnight and calcined at 900°C.

Mechanical Mixing (MM)
OCs were prepared through MM using commercial CuO as an active material for releasing O 2 and ZrO 2 powders as support materials (Ku et al., 2018). The particle sizes of these metal oxide materials were 1-10 µm. CuO powder was mixed with water and support materials, and 10 wt.% binder was added and mixed. OC samples were dried at 80°C for 12 h and calcined at 900°C for 3 h.
The resulting OCs were denoted as CuS-P, where letters S and P represent support materials, such as ZrO 2 (Zr), SiO 2 (Si), and Al 2 O 3 (Al), and the preparation method of the OC, such as MM, IM, and CP. For example, OCs was prepared through IM and ZrO 2 as support, the resulting Cu-based OCs were termed CuZr-IM OCs.

Oxygen Carrier Characterization
The physical properties of the OCs were determined by N 2 adsorption/desorption at 77 K via a Micromeritics ASAP 2020 volumetric sorption analyzer (Norcross, GA). Before measurement, OCs were degassed under high vacuum at 100°C for 3 h until water vapor and impurities on the surface of the OCs have been desorbed. N 2 adsorption/desorption isotherms were measured at a relative pressure (P N2 /P 0 ) range of 0.0001-0.99 and employed to determine surface area, pore volume, and average pore diameter via the Barrett-Johner-Halenda (BJH) equation for pore size of 1.7-100 nm. The particle size and surface morphologies of the OCs were characterized via a field emission scanning electron microscope (SEM, JEOL JSM-6500F/CL (Gatan MonoCL3 plus HSPMT)) and a high-resolution transmission electron microscopy, energy-dispersive X-ray spectroscopy (TEM, JEM-2100F/EDS (Oxford X-MAX 80 mm 2 )). The crystal phase of the OCs was characterized by a powder X-ray diffractometer (XRD, Mac Science Co., Ltd., Japan) using Cu K radiation (40 kV, 30 mA). The actual CuO content was analyzed via an inductively coupled plasma-optical emission spectrometer (ICP-OES, Agilent 5100).

Reactivity Test Conversion of OCs
Reactivity tests of the oxygen carriers were carried out in a thermogravimetric analyzer (TGA, SETSYS Evolution 24, SETARAM). The temperature of TGA chamber was elevated with a ramping rate of 10°C min -1 from room temperature to 900°C. The carry gas used for oxygen release in the reducer was pure N 2 , and the gas used for the regeneration of OCs in the oxidizer was air. Previous study (Song et al., 2014b) indicated that no obvious difference was observed in the reactivity of oxygen carriers with different inert reducing environments (N 2 or CO 2 ). Therefore, N 2 was usually used as the regeneration gas. The flow rate of the reacting gas was controlled at 200 mL min -1 . The OC (20-40 mg) was loaded in a platinum basket and heated to 900°C. The duration was set at 60 min for reduction and oxidation. The data of weight changes can be used for the conversion of OCs using the following equations: where X r and X o are the conversions of OCs during reduction and oxidation, respectively; and m o , m(t), and m r are the masses of the sample in a completely oxidized state, during the test, and in a completely reduced state, respectively.
In addition to conversion, the oxygen transport capacity (OTC) and reaction rate of solid conversion (ROC) could be determined (Song et al., 2014b). The OTC and ROC of OCs were defined as follows:

Reaction Kinetics in Reduction
The reduction kinetics of Cu-based OCs for CLAS were calculated using the procedure reported by the previous study (Song et al., 2014c;Wang et al., 2016). In this study, the reaction rate for reduction of Cu-based OCs was evaluated using the following equations: where X, f(X), and t are the conversion of OCs during reduction, kinetic mechanism functions, and reaction times, respectively; C and C eq are oxygen concentrations used for oxidation and equilibrium oxygen concentration used for CuO decomposition, respectively; n and k are the reaction order and reaction rate constant following the Arrhenius equation; and: where A, E a , R, and T represent the pre-exponential factor, activation energy, gas constant, and reaction temperature, respectively. The value of n is zero because the reduction of CuO is a thermal decomposition process and oxygen is not present in the reactive gas (N 2 as the carrier gas). Therefore, Eq. (7) can be simplified as follows: The integral form of Eq. (9) is as follows: where G(X) is a kinetic mechanism function, which can be obtained from different gas-solid reaction mechanisms. Table 1 summarizes G(X) equations for different reaction mechanisms (Song et al., 2014c). For example, G(X) under the D1 reaction mechanism is determined by the value of X 2 . Therefore, the reaction rate constant (k) can be calculated using the slope of the fitting line in the plots of G(X) of X 2 versus reaction time.

Characterization of OCs
The OCs comprising 40 wt.% CuO on various support materials (ZrO 2 , SiO 2 , Al 2 O 3 ) were prepared in this study. Table 2 lists the oxygen transport capacity, BET surface area, and pore volume of the OCs. For CuZr-IM OC, the precursor solution of copper was impregnated on the surface of ZrO 2 particles. The particle size and BET surface area of ZrO 2 supports were 15-25 nm and 37.8 m 2 g -1 , respectively. After IM, the size of CuZr-IM OCs was 100-250 nm. Therefore, CuZr-IM OCs can be classified as fine particles. Fig. 1 shows SEM images of OCs in general. The fresh OCs had a BET surface area and pore volume of 1.43 m 2 g -1 and 16.8 mm 3 g -1 , respectively. A crystalline phase from the XRD analysis shows CuO and ZrO 2 .
For CuZr-CP, the OC was prepared through CP. It reported that pH affects the reactivity and crushing strength of OCs (Imtiaz et al., 2012). In this study, the pH of 9.0 was selected because the OCs had high reactivity and mechanical strength. The particle size of the CuZr-CP was 20-50 µm, which is larger than that of CuZr-IM OC. The fresh OCs had a BET surface area and pore volume of 1.20 m 2 g -1 and 11.3 mm 3 g -1 , respectively. The crystalline phase showed CuO and ZrO 2 in XRD analysis.
For CuSi-IM, the precursor solution of copper was impregnated into the mesoporous SiO 2 particle. The particle size of the SiO 2 support was 1-10 µm. The average BET surface area and pore size of SiO 2 were approximately 250 m 2 g -1 and 3-6 nm, respectively. After the IM of copper nitrate, the particle size of CuSi-IM OCs was 3-10 µm. The fresh CuSi-IM OCs had a BET surface area of 6.05 m 2 g -1 and a pore volume of 35.0 mm 3 g -1 , which were larger than Table 1. G(X) equations for different reaction mechanisms.

Reactivity of OCs Effect of Support Material
In this study, three materials (i.e., ZrO 2 , SiO 2 , and Al 2 O 3 ) with different particle sizes and pore structures were used as a support for Cu-based OCs. All Cu-based OCs were prepared through IM because OCs have high reactivity and satisfactory mechanical strength (de Diego et al., 2004). Fig. 2 presents the effect of the support material on the reactivity of Cu-based OCs for CLAS. The reaction temperature of reduction and oxidation was controlled at 900°C. CuZr-IM exhibited superior conversion and faster reaction rate during reduction. This result may be attributed to the particle size of the ZrO 2 support. The ZrO 2 supports used for CuZr-IM were prepared through a modified sol-gel method. ZrO 2 supports were nanoparticles with a particle size of 15-25 nm (Fig. 3). After the IM of nano-sized ZrO 2 , the particle size of CuZr-IM OC was 100-250 nm, which is considerably smaller than that of other OCs. Fig. 4 shows TEM images and the corresponding EDX mapping of CuZr-IM; a high degree of dispersion of CuO was observed on the surface of ZrO 2 supports. The oxygen mobility, diffusion, and heat transfer of fine particles were superior to those of bulk materials with micrometer-sized particles, thus facilitating OC reactivity. These findings are consistent with those of a previous study (Song et al., 2013).
For the CuSi-IM OC, the mesoporous SiO 2 support was prepared through SD. The particle size of the prepared SiO 2 was 1.0-10.0 µm. The pore size and BET surface area of the SiO 2 support were 3-6 nm and approximately 250 m 2 g -1 , respectively. After IM of the copper precursor, the particle size of CuSi-IM was 3-15 µm. Although the BET surface  area of porous SiO 2 support was 250 m 2 g -1 . However, CuO will impregnate into the porous SiO 2 and then full up the pores due to high loading (40 wt.%). Moreover, the OC will be calcined at high temperature of 900°C. The surface area of CuSi-IM OC was therefore decreased dramatically. Fig. 5 shows TEM images of mesoporous SiO 2 support and CuSi-IM OCs; the particle size of CuO on the surface of the SiO 2 support ranged from 6-13 nm. The TEM images indicate a fine dispersion of copper oxide on the SiO 2 support (Fig. 6). The porosity of the SiO 2 support and CuO nanoparticles may enhance reactivity performance and provide high conversion and a fast reaction rate during reduction. However, the particle size of CuSi-IM OCs was 3-15 µm. The reactivity of CuSi-IM was slightly less than that of CuZr-IM. Moreover, the regeneration of CuSi-IM OC was incomplete during oxidation because of the slight agglomeration of CuSi-IM OCs during redox (Fig. 7).
The CuO supported on the mesoporous Al 2 O 3 support (CuAl-IM OC) showed lower conversion (approximately 25%) during the reduction process compared with CuZr-IM and CuSi-IM. This result can be attributed to the formation of copper aluminate (CuAl 2 O 4 ) between the CuO and Al 2 O 3 support during calcination (Eq. (11)), which significantly reduced the oxygen release of the OC. Fig. 8 shows XRD patterns of CuAl-IM, indicating that copper aluminate (CuAl 2 O 4 ) was present. This finding is in accordance with the results of previous studies (Abad et al., 2012;Song et al., 2014a). CuAl 2 O 4 is a very stable substance with a reduction temperature of about 900-1000°C. CuAl 2 O 4 will decomposed slowly to form Al 2 O 3 and O 2 in the anoxic environment (reduction reaction, Eq. (12)), but also CuAlO 2 is formed at the same time, so CuAl 2 O 4 can only be partially regenerated into CuO. Therefore, the amount of CuO capable of releasing oxygen reduced, which means Al 2 O 3 is not a suitable support material for the oxygen-releasing carrier despite having a mesoporous structure. The aforementioned analysis indicates  that nanosized ZrO 2 may be a good choice as the support material of OC.
CuO + Al 2 O 3 → CuAl 2 O 4 (11) 4CuAl 2 O 4 → 4CuAlO 2 + 2Al 2 O 3 + O 2 (12) Fig. 9 shows the conversions of three 40 wt.% Cu-based OCs using ZrO 2 as supports obtained through various preparation methods in one reduction and oxidation (redox) cycle in CLAS. The reaction temperature of redox was controlled at 900°C. OCs prepared through IM were more reactive than those prepared through CP and MM. CuZr-IM OC had a high conversion rate of nearly 100%. Furthermore, the average ROC was the fastest in the reduction process (7.40% min -1 ) ( Table 3). This result may be attributed to the particle size of the OCs. The aforementioned analysis indicates high conversion of CuZr-IM results from the CuO homogeneously dispersed on the nanosized ZrO 2 supports. Although CuZr-CP OCs have higher crushing strength, the particle size was larger than that of CuZr-IM. The particle size determines reactivity performance, including conversion and reaction rate, in CLAS.  For the regeneration of reduced OCs during oxidation in air at a temperature of 825-950°C, the average rates of solid conversion were 16.58%, 11.57%, and 5.78% min -1 for CuZr-IM, CuZr-CP, and CuZr-MM, respectively. CuZr-IM OC exhibited the fastest kinetics. Moreover, this result may be attributed to the particle size of OCs. Furthermore, the oxidation rate of the OC was considerably faster than the reduction rate. These results are consistent with previous study (Song et al., 2014c). CuZr-IM OCs exhibited the highest conversion and fastest reaction rate during redox in CLAS.

Kinetic Analysis of OCs during Reduction
As stated previously, CuZr-IM was the most suitable Cubased OC for CLAS because of its high conversion and reaction rate during redox. Therefore, CuZr-IM was selected as an OC for the determination of chemical kinetics during reduction in CLAS. Table 1 summarizes G(X) equations of different gas-solid reaction mechanisms. The reaction rate constant (k) can be calculated from the slope of the fitting line in a plot of G(X) versus time. Table 4 presents the results of R 2 values for fitting different reaction mechanisms based on CuZr-IM   OCs during reduction at a temperature of 825-950°C. It indicated that Avrami-Erofe'ev random nucleation and the subsequent growth reaction model with n = 2 (A2) are the optimal fit for CuO-based OC conversion during reduction in CLAS. Therefore, the oxygen release of OCs was determined using the rates of nucleation, nuclei growth, and nucleus formation.
The kinetic mechanism function G(X) under the A2 reaction model is determined by the value of [-ln(1 -X)] 1/2 . Fig. 10 shows the results of the fitting curves of G(X) versus reaction time at different reduction temperatures. Table 5 presents linear correlation coefficient (R 2 ) values of the least-squares linear fitting and rate constants for temperatures from 825 to 950°C. Rate constants were 0.0636, 0.0831, 0.1271, 0.1820, 0.23, and   Fig. 11 shows the fitting lines of ln(k) versus 1000/T for the reduction of CuZr-IM OCs. The values of activation energy (E a ) and pre-exponential factor (A) can be obtained by calculating the slope and intercept of the fitting line. The E a of 140.2 kJ mol -1 and A of 2.82 × 10 5 min -1 at a temperature of 825-950°C were lower than those of micrometer-sized CuO/SiO 2 OCs (176 kJ mol -1 ) at temperatures higher than 900°C (Song et al., 2014c). CuZr-IM OC with lower activation energy may be attributed to the Fig. 11. Plots of ln(k) versus 1000/T of CuZr-IM OCs at different temperatures during reduction.
decreasing particle size (Alalwan et al., 2018). This result is consistent with that of a previous study (Pang et al., 2009).
The conversion of CuZr-IM OCs during reduction can be obtained using kinetic parameters according to Eq. (10). Fig. 12 shows the prediction results, which indicated that the predicted value (line) was close to the experimental values (dot points) at reaction temperatures of 825-950°C.

Stability of OCs
Based on the aforementioned analysis, CuZr-IM OCs are better for CLAS. In addition to a high conversion and fast reaction rate, long-term recyclability and durability are necessary for OCs. These can reduce the fine-particle purging, thereby reducing the fresh OC makeup rate. To determine the long-term reaction stability, a 20-cycle redox test was performed on CuZr-IM OCs at 900°C using a TGA. The OTC of an OC is defined as the difference in the mass fraction between the completely oxidized and reduced forms. This can be a critical index to determine the favorability for oxygen transport (Song et al., 2014b). Fig. 13 shows the weight fraction of CuZr-IM OCs during 20 consecutive redox cycles, indicating CuZr-IM OCs with a stable OTC in CLAS. The reduced OCs were completely reoxidized during oxidation. Furthermore, the CuZr-IM was converted over 20 redox cycles. The CuZr-IM OCs maintained relatively stable conversion. Comparison of the fresh and used OCs after 20 redox cycles in CLAS revealed no evident agglomeration for the used OCs (Fig. 14).

CONCLUSION
In this study, the reactivity and conversion rates of novel Cu-based OCs prepared with various methods and different supporting materials for CLAS were studied. Cu-based OCs impregnated with supporting ZrO 2 nanoparticles (CuZr-IM OCs) exhibited excellent performance and can be employed in CLAS to produce oxygen at temperatures of 825-950°C. This result can be attributed to the small particle size of the OCs. The CuO nanoparticles on the surface of the nano-ZrO 2 supports were distributed in a uniform pattern and displayed  greater oxygen mobility and faster diffusion than micrometersized particles and bulk materials. The reaction kinetics of the CuZr-IM OCs during reduction were studied, and Avrami-Erofe'ev random nucleation and the subsequent growth reaction model with n = 2 (A2) were found to be the optimal fit for OC conversion during this reaction. Additionally, an activation energy of 140.2 kJ mol -1 for this OC indicates that reducing the particle size decreases the activation energy. Recyclability and durability tests also reveal that the CuZr-IM OC is suitable for CLAS reactions.
In the future, the fluidized-bed process should be used to evaluate the real-time oxygen concentration during reduction. A series of parametric tests will also be conducted.

ACKNOWLEDGMENT
This study was supported by the Bureau of Energy, Ministry of Economic Affairs of Taiwan.