Oxidation kinetics of YBaCo4O7+δ and substituted oxygen carriers

In this paper, the relaxation kinetics of the oxidation process of the YBaCo4O7+δ, Y0.95Ti0.05BaCo4O7+δ and Y0.5Dy0.5BaCo4O7+δ oxygen carriers is studied with isothermal reaction data. XRD analysis for fresh samples shows that all the samples have YBaCo4O7+δ structure. Scanning electron microscopy images of samples show that the samples consist of porous agglomerates of primary particles. Isothermal TG experiments are conducted with temperatures of 290°C, 310°C, 330°C and 350°C, respectively. It is found that the Avrami-Eroféev model describes solid-phase changes in the oxygen absorption process adequately. The results show that the distributed activation energies of the oxidation process obtained by the Avrami-Eroféev model are 42.079 kJ mol−1, 42.944 kJ mol−1 and 41.711 kJ mol−1 for the YBaCo4O7+δ, Y0.95Ti0.05BaCo4O7+δ and Y0.5Dy0.5BaCo4O7+δ oxygen carriers, respectively. The kinetic model was obtained to predict the oxygen carrier conversion of oxygen absorption for different time durations. The kinetic parameters obtained here are quite vital when this material is used in reactors.


Introduction
It is generally accepted that carbon dioxide (CO 2 ) emission is the main contributor to global warming. Oxygen-enriched combustion, one of the possible options to reduce CO 2 emission, is not applied widely in industry due to the high cost of oxygen production. The process of chemical looping air separation (CLAS) was developed by Moghtaderi & Song in 2010 [1]. The process saves 74% of the power of the cryogenic air separation process [1]. The schematic of the CLAS process is described elsewhere [2]. The oxygen carrier circulates between the oxidation reactor and the reduction reactor. In the oxidation reactor, the oxygen carrier is fully oxidized by oxygen. In the reduction reactor, the oxygen carrier is fully reduced by steam or CO 2 . from the bulk. The oxygen ions are involved in dissociative adsorption and chemical adsorption. The diffusion of oxygen ions is one of the processes which might control the rate of the oxidation reaction. The different adsorption steps and the possible surface migration of these adsorbed species to the reaction sites might also be rate-controlling [43]. In the case of the chemical reaction-controlled process, action there proceeds uniformly throughout the solid particles. In this case, different models need to be employed. The mechanism and kinetic parameters obtained here are quite vital when this material is used in the reactors.

Experimental section 2.1. Preparation of materials
Samples of YBaCo 4 O 7+δ , Y 0.95 Ti 0.05 BaCo 4 O 7+δ and Y 0.5 Dy 0.5 BaCo 4 O 7+δ were synthesized by a solidstate reaction. Mixed appropriate stoichiometric amounts of the starting materials, Y 2 O 3 , TiO 2 , Dy 2 O 3 , BaCO 3 and Co 3 O 4 , were ground thoroughly and then calcined at 1000 • C for 15 h. The calcined samples were reground and calcined at 1100 • C for 30 h. After calcination, all the samples were ground with a mortar and sieved with a 400-mesh sieve (average particle size less than or equal to 37.5 µm) for experiments and kinetic analysis.

Characterization of materials
Phase composition was studied by a powder X-ray diffraction technique (Panalytical, PW 3040/60; X'Pert Pro system with Cu Kα radiation). X-ray data were recorded with a step scan of 0.02°for 2θ between 10°and 70°, and the cell parameters were determined with JADE software. The microstructure of the synthesized samples was observed with scanning electron microscopy (SEM) on an ultra plus field emission scanning electron microscope. The oxygen absorption behaviour was observed with isothermal TG experiments in a thermogravimetric analyzer-TGA (STA409PC). During the TG experiment, a powder sample, with a mass of 10 mg, was heated to the target temperature (290, 310, 330 and 350 • C) in a N 2 atmosphere to prevent the occurrence of oxygen absorption. Then the atmosphere was changed to an air flow of 40 ml min −1 , keeping the target temperature for 2 h to investigate the oxygen absorption behaviour. Before the kinetic experiments, the internal and external diffusion were eliminated by the experiments by varying the gas flow rate and the sample loading weight in the ranges of 20-40 ml min −1 and 10-20 mg, respectively.

Characterization
The phase composition of the YBaCo 4 O 7+δ , Y 0.95 Ti 0.05 BaCo 4 O 7+δ and Y 0.5 Dy 0.5 BaCo 4 O 7+δ samples is shown in figure 1. The cell parameters of the samples are refined from the data in space group P6 3 mc, and the refined cell parameters are presented in table 1. By combining with the XRD patterns and refined cell parameters, the present samples are indexed to be of YBaCo 4 O 7+δ structure. Typical SEM images of samples are shown in figure 2. It is seen that the samples consist of porous agglomerates of primary particles. The differences in morphology with different substituting ions of the oxygen carriers are very small.
The 114 phase oxygen carriers can absorb certain amounts of oxygen at different temperatures. The percentage change in mass m (%) and total stoichiometric change (δ) obtained at different oxidation temperatures are presented in table 2. The amount of oxygen absorption increases with increase in the oxidation temperature lower than 330 • C. The amount of oxygen absorption obtained at 350 • C is lower than that of the value obtained at 330 and 310 • C. Furthermore, at a given oxidation temperature, the amount of oxygen absorption of Ti and Dy substituting samples is larger than that of the unsubstituted sample. Figure 3a-

Kinetic models
The reaction rate of the process [45] can be written as follows: where α is the extent of conversion, k(T) is the reaction rate content and f (α) is the kinetic model function. Equation (3.1) can be modified as follows: Equation (3.2) can be transformed into equations (3.3a,b):

A4
Avrami The plots of G(α) versus t should be straight lines whose slope can be used to determine the reaction rate k(T). The model showing the best linear fitting is chosen as the favoured model. The reaction models used for describing the oxidation process of oxygen carriers are presented in table 3 [41,42,[45][46][47][48][49].
By linear fitting the mechanism functions against t (parameters were estimated in the 0.1-0.90 conversion range), the linear correlation coefficient R 2 and the residual sum of squares (RSS) of each function can be obtained. Figure 4 shows the fitting linear curves G(α) versus t under different oxidation temperatures.
Tables 4 and 5 list the R 2 and RSS values obtained by fitting functions, respectively. The discrimination among the models was based on the higher R 2 and lower RSS. The functions with the bigger R 2 and smaller RSS values are selected as the mechanism functions. For the YBaCo 4 O 7+δ and Y 0.5 Dy 0.5 BaCo 4 O 7+δ samples, given the R 2 and RSS values obtained, it was concluded that the A model and R better fitting were achieved with n values of 4 and 1, respectively. For the Y 0.95 Ti 0.05 BaCo 4 O 7+δ sample, it was concluded that the A model and R better fitting were achieved with n values of 3 and 1, respectively. For the Avrami-Eroféev random nucleation and the nuclei growth model, the overall conversion of the oxygen absorption reaction is determined by the relative rates of nucleation, nuclei growth and nucleus formation [24,[50][51][52]. Nucleation and crystal growth are a dynamic process which practically initiates the oxygen absorption reaction. Generally, for the unreacted shrinking-core model, the overall conversion of the oxygen absorption reaction is determined by the chemical process [24]. That is, the overall conversion of the reaction is dominated by the chemical reaction, not the diffusion process for the A models and R models. The determined models can be used to evaluate the reaction rate, apparent activation energy and pre-exponential factor of the oxygen absorption reaction. For the YBaCo 4 O 7+δ , Y 0.95 Ti 0.05 BaCo 4 O 7+δ and Y 0.5 Dy 0.5 BaCo 4 O 7+δ samples, the reaction rate constants are evaluated and presented in table 6.
From table 6, for the different mechanism functions and oxygen carriers, the reaction rate constant increases with increase in the reaction temperature, indicating that high temperature is propitious to the rate of oxygen adsorption. Low temperature may be one of the reasons accounting for the slow reaction rates shown in figure 3. Furthermore, the reaction rate (except for the reaction rate obtained at 290 • C for the Y 0.95 Ti 0.05 BaCo 4 O 7+δ sample) obtained by the A model is lower than that of the R model. After evaluating the reaction rate constant, the pre-exponential factor and apparent activation energy can be evaluated.
Along with the Arrhenius expression, the following is obtained: where A is the pre-exponential factor, E is the apparent activation energy, R is the gas constant and T is the reaction temperature. Along with the Arrhenius expression, the following form is obtained: where the ln k(T) has been evaluated above, the plots ln k(T) versus 1/T are straight lines whose slope and intercept can be used to evaluate the apparent activation energy and pre-exponential factor, respectively. Figure 5 shows the plots ln k(T) versus 1/T as a function of different mechanism functions. Table 7 lists the estimated apparent activation energy and pre-exponential factor as a function of reaction temperatures. The apparent activation energies obtained by the different mechanism functions remain close to constant levels for an oxygen carrier.
The activation energies for Y 0.5 Dy 0.5 BaCo 4 O 7+δ oxidation are found to be lower than those for YBaCo 4 O 7+δ and Y 0.95 Ti 0.05 BaCo 4 O 7+δ oxidation, thus confirming the favourable effect of Dy on the oxidizability of the YBaCo 4 O 7+δ oxygen carrier. This may be accounted for by the cell volume. The larger the cell volume, the easier is the absorption of oxygen. For an oxygen carrier, the pre-exponential factor obtained by the R model is larger than that of the A model.
For the purpose of further model discrimination between the A and R models, the A model is more favourable considering the higher unity of data values. In the case of the A model, the activation energies and the frequency factor remain close to constant levels at the different temperatures. Moreover, the activation energies with the R model vary in a much wider range. Thus, these results confirm the adequacy of the A model over the R model [24]. Thus, the nucleation and nuclei growth model is chosen as the most possible mechanism function.
The values of the established kinetic parameters, the apparent activation energies, the pre-exponential factors and the mechanism function were introduced into equation (3.1) and the differential equation was obtained to predict the oxygen carrier conversion of the oxidation process for different reaction time durations. The kinetic models are listed in table 8.