Release of Mn-Based Oxygen Carrier for Chemical Looping Air Separation ( CLAS ) : An Insight into Kinetic Studies

Chemical looping air separation (CLAS) is an air separation technology with a relatively small energy footprint. In this contribution, the kinetic study of Mn-based oxygen carriers has been carried out in a CLAS process. The O2 release behavior was investigated under Ar in a fixed bed. Results showed O2 release amounts increase gradually with an increasing temperature. Moreover, fitting various gas-solid reaction mechanisms with the experimental data was used to obtain the chemical reaction kinetics for MnO2 and Mn2O3 for O2 release. As for all involved gas–solid reaction mechanisms, the first-order chemical reaction model (C1) and Avrami–Erofe’ev random nucleation and subsequence growth model (A2) fitted well with O2 release experimental data for MnO2 and Mn2O3, respectively. Furthermore, activation energy, pre-exponential factor and reaction order were also determined for these models.


INTRODUCTION
Oxygen (O 2 ) is the most common element in the earth's crust, and the second largest industrial gas (after N 2 ) (Castle, 2002;Kerry, 2007;Häring, 2008).Oxygen is mainly used in the metal-mechanic industries, oxidation of the natural gas, medical unit, and integrated gasification combined cycle (IGCC) (Andersson and Johnsson, 2006), oxy-fuel combustion (Yin et al., 2008), solid oxide fuel cell (SOFC) (Hutchings et al., 2008), and so on.Several technologies are commercially employed to produce O 2 , including cryogenic distillation (the most common method) (Castle, 2002;Kerry, 2007;Häring, 2008), membrane separation (Schreiber et al., 2013), pressure swing adsorption (PSA) (Hassan and Ruthven, 1986;Jiang et al., 2003).Cryogenic distillation, with high investment costs, is a well-developed and established technique for large-scale production of pure oxygen.Membrane separation methods, in spite of less energy intensive, are expensive because of expensive and complicated their fabrication, installation, and integration (Pfaff and Kather, 2009).PSA processes are usually applied in smaller scale, when it is sufficient with lower purity, typically around 95 vol.% oxygen.Although conventional separation methods are mature technologies, they are limited because of high energy consumption and costs, facing global environmental change and energy shortage (Hashim et al., 2011;Moghtaderi et al., 2011).Chemical absorption process may be a rather competitive alternative if the facility corrosion could be reduced.Compared with these technologies, the chemical looping air separation (CLAS), proposed by Mogtaderi et al. (Moghtaderi, 2010;Moghtaderi et al., 2011;Shah et al., 2011;Moghtaderi, 2012;Shah et al., 2012), has the advantages of operationsimple, cost-efficient, and energy-saving.The average specific power of CLAS (about 0.08 kWh m -3 ) is only 26% of that of an advanced cryogenic air separation system (Moghtaderi, 2010).The CLAS involves two separate fluidized bed reactors, between which there is a loop seal to prevent gas leakage from one reactor to another, thus keeping N 2 from entering the O 2 release reactor.In the O 2 uptake reactor, the reduced oxygen carrier (OC) is oxidized by air, and then the oxide OC is transported to the O 2 release reactor.In the O 2 release reactor, the oxygen uncoupling occurs.The successful run of CLAS is greatly determined by the regenerative nature of OC.
However, the selection of effective OC materials is probably a challenge.In comparison with the OC for normal chemical looping combustion (CLC) where the fuel (gaseous or solid) reacts directly with the OC without any release of gas phase O 2 , there is a special requirement for the OC in the CLAS process.The OC materials include CuO, Mn 2 O 3 , Co 3 O 4 , Pb 3 O 4 , CrO 2 , SrO 2 , PdO 2 , and CeO 2 (Azimi et al., 2011;Imagawa et al., 2011;Adánez-Rubio et al., 2012;Wang et al., 2013).Among these oxides, Pb 3 O 4 is poisonous, PdO 2 is expensive, and the O 2 release rate of SrO 2 , CeO 2 , and CrO 2 is low.However, only three metal oxides (CuO, Mn 2 O 3 and Co 3 O 4 ) can be used as active compounds for CLAS (Mattisson et al., 2009) because of a suitable equilibrium O 2 partial pressure at 800-1200°C.The maximum oxygen transport capacity is 0.1, 0.03 and 0.066 (g O 2 )•(g OC) -1 for the pairs CuO/Cu 2 O, Co 3 O 4 /CoO and Mn 2 O 3 /Mn 3 O 4 , respectively.So far none of OC employed in the CLAS process can effectively satisfy the air separation requirements (Moghtaderi, 2010;Moghtaderi, 2012).
The main parameters of the CLAS reactors are determined based on the knowledge of O 2 release kinetics of OC.Reactivity data for a great deal of materials have been found, however, they have often been acquired for a single operational condition (Lyngfelt et al., 2008).A lot of experimental studies have been performed in a thermogravimetric analyzer (TGA).The temperature programmed reduction (TPR) or temperature programmed oxidation (TPO) technique also has been used for kinetic determination at a very low temperature.Others facilities have been also used to diminish mass transfer limitations (Bohn et al., 2009;Chuang et al., 2009Chuang et al., , 2010) ) or to account for the mass transfer processes (Iliuta et al., 2010).Chadda et al. (1989) found that CuO have a very high activation energy (313 kJ mol -1 ) for the O 2 release.However, few studies are focus on the O 2 release of Mn-based oxygen-carriers in CLAS.Further research on the O 2 release of Mn-based oxygen carriers for CLAS will be required in the future.
In a word, Mn-based oxygen carriers can actually meet the CLAS requirements.However, there is a limited capacity to optimize their performance since information on OC properties/characteristics in CLAS is quite inappropriate.Therefore, further research on Mn-based oxygen carriers is required to use in CLAS.This work focuses on reaction kinetics of Mn-based oxygen carriers for CLAS.

Oxygen Carrier Materials
Manganese dioxide (MnO 2 ) was purchased from Sinopharm Chemical Reagent Co., Ltd.The sample was analytical reagent (AR) and used directly.7 g of MnO 2 , and 100 g of stainless steel balls were put into the planet balling mill.After ball milling for 10 h, the crushed samples were sieved to obtain smaller particles with the particle size distribution is in the range of 140-160 mesh (average size is 100 µm).The sample prepared is referred to as MnO 2 .The MnO 2 powders obtained were further heated at 600°C for 6 h.The calcined samples were crushed, and sieved to obtain smaller particles with the particle size distribution is in the range of 140-160 mesh (average size is 100 µm).The sample prepared is referred to as Mn 2 O 3 .The oxygen carriers (about 100 mg) were analyzed using X-ray diffraction (XRD, Bruker D8 Advance).

Fixed BED SYSTEM
The kinetics of O 2 release of MnO 2 and Mn 2 O 3 were conducted in a laboratory scale fixed bed reactor.The schematic diagram of fixed bed reactor was shown in Fig. 1.A quartz tube (length = 300 mm, i.d.= 20 mm) is located inside an electric furnace.The temperature inside the reaction tube is monitored by K type thermocouple in real time.The O 2 release is tested under highly pure Ar atmosphere.300 mL min -1 Ar was flowed into the reactor through the porous plate.The mass of oxygen carrier is 500 mg.Before the test, Ar was introduced into the reaction tube to sweep the residual O 2 .Then the reactor was heated in Ar atmosphere.As the temperature was stable for 60 min, the quartz boat (filled with oxygen carrier particles) was pulled from left to center by platinum wire using Ar as protective gas.Then the gap was sealed with a sealant.The start and completion of the oxygen release were marked by the emergence and vanishing of oxygen in the gas products, respectively.

Kinetic Mechanism of Oxygen Release
In order to explore the kinetic characteristics of OC, the conversion rate α of OC at different temperatures was calculated by Eq. ( 1).The conversion rates of MnO 2 and Mn 2 O 3 are represented by α.
where t (s) is the end time of oxygen release, x (mol%) is molar concentration of O 2 in the gas products, and Q (mol/s) is the molar flow rate of gas products.
Reaction kinetics focuses on pre-exponential factor A (s -1 ), kinetic constant k, activation energy E (kJ mol -1 ), the integrated form of mechanism function G(α) or the differential form f(α).The relationship between E, A and k can be expressed by Arrhenius formula [Eq.( 2)].f(α) and k can mutually convert through Eq. ( 3), and the relationship between f(α) and G(α) is described by Eq. ( 5).In Eq. ( 5), k is constant at constant temperature and G(α) is linearly related to t, which is used to determine the G(α) of the reaction process.The linear correlation between G(α) and t is described by coefficient r.When r is closer to 1, the linearity is better.And the corresponding G(α) is more possibly the mechanism function that we seek.We calculated the coefficient r by bringing fourteen G(α)s (Wang et al., 2013) (shown in Table 1) into Eq.( 5), which resulted in many G(α)s that exhibit good linear correlation with t.Therefore, the intercept of the line formed by G(α) and t was further calculated to determine the mechanism function.The G(α) with intercept near to zero was chosen as the mechanism function of this reaction.The kinetic constant k at various temperatures was calculated by Eq. (3).By putting k and temperature T into Eq.( 3), E and A were obtained through the linear relationship of lnk and T -1 .Reaction mechanism (1 -α) -1 -1 C2 Avrami-Erofe'ev random nucleation and subsequence growth (n = 2) [-ln(1 -α)] 1/2 A2 Avrami-Erofe'ev random nucleation and subsequence growth Different reaction mechanisms (Halikia et al., 1998;Pineau et al., 2006;Perkins et al., 2007;Pineau et al., 2007;Li et al., 2009) was used to obtain expressions of G(α), summarized in Table 1.Fourteen kinetic equations are divided into five groups: the diffusion controlled, chemical reaction controlled, random nucleation and subsequent growth of nuclei models, boundary-controlled, and Mampel power law models.By analyzing some curve points in the set range of conversion, we can choose the best fitting model using the least square method.

Thermodynamic Investigation
The Gibbs free energy minimization method was used to study thermodynamic of manganese oxides for the CLAS process.Fig. 2 shows the O 2 release for MnO 2 and Mn 2 O 3 for CLAS as a function of the temperature.The release of O 2 is assumed to be rapid under any conditions.From Fig. 2, it can be observed that, for MnO 2 and Mn 2 O 3 , the increasing temperatures promote the O 2 release.The lowest oxygen releasing temperature is about 290°C and 780°C for MnO 2 and Mn 2 O 3 , respectively, under inert atmosphere.The required temperatures are much lower than the reaction of CuO into Cu 2 O.For example, the minimum oxygen releasing temperature for the reaction of Co 3 O 4 into CoO is 918°C and of CuO into Cu 2 O is 1048°C (Wang et al., 2013).The low operation temperature can decrease the energy intensity and the operation difficulty of CLAS.

Non-Isothermal Experimental Studies
Conversion (α) versus time during the O 2 release is presented in Fig. 3.The decomposition in Ar takes place mainly in two stages.In the first stage, the thermal decomposition of MnO 2 to Mn 2 O 3 and O 2 takes place below 700°C, as indicated by a peak at 645°C.XRD patterns show that the phases in Mn-based oxygen carriers were only MnO 2 , Mn 2 O 3 and Mn 3 O 4 (Fig. 4).It is indicated that two O 2 release reactions can be described by Eqs. ( 6) and ( 7).The kinetics of these reactions has been determined in the following part.It is found that the main phases in Mnbased oxygen carriers are only MnO 2 before O 2 release reaction (Eq.( 6)).However, the main phases in the reduced oxygen carriers are only Mn 2 O 3 .The results show that the MnO 2 has been completely reduced to Mn 2 O 3 in the O 2 release reaction.The second stage is the decomposition of Mn 2 O 3 to Mn 3 O 4 and formation of O 2 .The peak is at 935°C and the decomposition is complete at 1000°C.Similarly, XRD patterns show that the main phases in the initial samples are only Mn 2 O 3 before O 2 release reaction (Eq.( 7)), but the main phases in the reduced samples are only Mn 3 O 4 .It is indicated that Mn 2 O 3 has been completely reduced to Mn 3 O 4 in the O 2 release reaction.

Isothermal Kinetics MnO 2
The O 2 release amounts versus time for MnO 2 are displayed in Fig. 5.It is found that the increase of temperature from 500 to 600°C promotes the O 2 release.It is attributed to the thermodynamic limitations of MnO 2 /Mn 2 O 3 .Generally, the O 2 release amounts (0.030, 0.049, 0.069, 0.075 and 0.072 (g O 2 )•(g MnO 2 ) -1 at 500, 520, 540, 560 and 600°C, respectively) are lower than the maximum oxygen transport capacity [0.123 (g O 2 )•(g MnO 2 ) -1 ] for the MnO 2 /Mn 2 O 3 system (Wang et al., 2013).Once the release amount of O 2 is close to the equilibrium value, a dense oxygen zone surrounding OC can produce diffusion effect, which restrains the release of O 2 .Similar trends were also found for the Co-based and Cu-based oxygen carriers (Li et al., 2008;Song et al., 2014).Our findings are in agreement with the results for Mn-Fe oxides (Azimi et al., 2013).
As mentioned above, Eq. ( 5) has been from analyzing the O 2 release kinetics.It presents the correlation that a possible reaction mechanism, G(α) in Table 1 and time t should follow.The most adequate G(α) can be obtained for the O 2 release from an isothermal experiment.Table 1 demonstrated G(α) equations for the O 2 release on the basis of fourteen mechanisms.Meanwhile, Table 2 also presented the R 2 values of the least-squares linear fitting for those mechanisms in the temperature range of 500 to 600°C From Table 2, it can be found that the first-order chemical reaction, i.e., C1, showed better overall fitting results than others in the temperature range of 500-600°C for the O 2 release.The reaction rate constant (k) was then obtained from the C1 model.Considering that, for the O 2 release process, the value of k is determined by the slope from fitting the C1 model with time t.
The C1 is associated with the first-order oxygen release reaction.It is assumed that C1 depends on the concentration of only one reactant (a unimolecular reaction).Therefore, it can be calculated through Eq. ( 8).It shows that a plot of ln (1 -α) versus t would be linear with a slope of k and an intercept of zero.Therefore, the O 2 uncoupling of MnO 2 is a reaction that proceeds at a rate that depends linearly only on the concentration of MnO 2 .The rate at which MnO 2 is consumed in a first-order process is proportional to its concentration in this case.Therefore, there is no induction period during the uncoupling of MnO 2 , as shown in Fig. 5.
-ln(1 -α) = kt (8) Fig. 6 shows the fittings results between the C1 model and time (t) for the typical tests at 500, 520, 540, 560 and 600°C.The reaction rate constants were acquired from the slope values.Then an Arrhenius type of plot is constructed in Fig. 7. Therefore, pre-exponential factor A and activation energy E were determined from the slope and intercept of the plot in Fig. 7. Thus, the O 2 release process is expressed by an apparent pre-exponential factor A=7.92 × 10 4 s -1 and activation energy of 130 kJ mol -1 in the temperature range of 500-600°C.In addition, a diffusion effect from MnO 2 thermodynamic properties may not be neglected.According to the thermodynamic results (Fig. 3), the equilibrium oxygen release amount will rise from 0.044 to 0.070 (g O 2 )•(g MnO 2 ) -1 (i.e., about 1.6 times higher) with the increasing from 500 to 600°C.For the O 2 release, the equilibrium O 2 release amount cannot be reached.Although the gas flow rate (300 mL min -1 ) is chosen as high as possible for the tests to diminish the diffusion effect, the equilibrium can still be obtained at the boundaries surrounding the unreacted grains inside the MnO 2 particles.The O 2 release rate will be effectively decreased once the equilibrium is reached because of the higher transfer resistance for the generated oxygen, viz., and diffusion effect.But the increasing temperature reduces the diffusion effect because of the raise in the equilibrium amounts of O 2 release.In addition, the ball mill method causes the agglomeration of the active metal oxide in MnO 2 , in particular, for higher temperatures, leading to a sintering of the grain.Consequently, the reduction reaction restrains the oxygen diffusion, which is reflected by the significant decrease of activation energy.As shown in Fig. 8, the typical prediction results (solid line) for MnO 2 at temperatures of 500, 520, 540, 560 and 600°C, along with    experimental data (dot points).There is an excellent agreement between the theoretical predicted and experiment values for all temperatures.

Mn 2 O 3
The O 2 release amounts versus time for MnO 2 are displayed in Fig. 8.The O 2 release amounts raised with increasing the temperature from 760 to 820°C.It is indicated that the higher temperature promote the O 2 release because of the thermodynamic limitations of Mn 2 O 3 /Mn 3 O 4 .Generally, the O 2 release amounts [0.018, 0.022, 0.023, 0.025 and 0.025 (g O 2 )•(g Mn 2 O 3 ) -1 ] are lower than the maximum oxygen transport capacity [0.034 (g O 2 )•(g Mn 2 O 3 ) -1 ] for the Mn 2 O 3 /Mn 3 O 4 system (Wang et al., 2013).Table 3 shows that the Avrami-Erofe'ev random nucleation and subsequence growth model with n=2 (A2), attain better fitting than other models for the O 2 release.k was then determined from the A2 model for the O 2 release.For the O 2 release process, the value of k is obtained by the slope from fitting the A2 model with time t.
The A2 is associated with two-dimensional growth of nuclei.Assumptions are made regarding the rate of change in the number of nuclei and the volume swept out by them as they react.Therefore, it can be written in logarithmic form as Eq. ( 9).Mn 3 O 4 nuclei may be present initially or they may grow in at certain locations by a process that is usually considered to be first order.It suggests that (i) Mn 3 O 4 particles will be nucleated and precipitated out from the Mn(O) intermediate/complex during heating and (ii) the growth of the fine nuclei of Mn 3 O 4 particles will be restricted to the surface only (two dimensional growth).[-ln(1 -α)] 1/2 = kt (9) Fig. 9 shows the fittings between the A2 model and t for the typical tests performed at 760, 770, 780, 800 and 820°C.The corresponding slope values were obtained for k.Then we construct an Arrhenius type of plot, as shown in Fig. 10.As a result, A and E can be obtained from the slope and intercept of the plot (Fig. 10).
Given all that, There is an E of 163 kJ mol -1 with A = 7.87 × 10 4 s -1 during the O 2 release process in the temperature range of 760-820°C.As shown in Fig. 3, the oxygen release amounts are higher than those of the corresponding  equilibrium values, because the gas flow rate is chosen as high as possible for the tests to diminish the diffusion effect.Once the equilibrium is broken through, the O 2 release rate will be significantly increased.
Fig. 12 shows the typical prediction values (solid line) for Mn 2 O 3 conversion during the O 2 release at temperatures of 760, 770, 780, 800 and 820°C, along with experimental data (dot points).There is an excellent agreement between the theoretical predicted and experimental values for all temperatures.

CONCLUSIONS
In this work, reaction mechanisms have been identified for the O 2 release reaction of MnO 2 and Mn 2 O 3 .The O 2 release behavior was investigated under Ar in a fixed bed reactor.The typical experimental conditions, viz., gas flow rate (300 mL min -1 ), sample loading (500 mg), and average particle size (100 µm), have been chosen for experiment to obtain the corresponding kinetic parameters.It is found that the O 2 release amounts raise significantly with increasing the temperature.The O 2 release mechanisms have been identified for MnO 2 and Mn 2 O 3 .Results indicate that the first-order chemical reaction model (C1) and Avrami-Erofe'ev random nucleation and subsequence growth model (A2) are appropriate for the O 2 release of MnO 2 and Mn 2 O 3 , respectively.For MnO 2 , the pre-exponential factor and activation energy were determined as 7.92 × 10 4 s -1 and 130 kJ mol -1 , respectively.For Mn 2 O 3 , the pre-exponential factor and activation energy are 7.87 × 10 4 s -1 and 163 kJ mol -1 , respectively.

Fig. 1 .
Fig. 1.Schematic diagram of the fixed bed reactor system.

Fig. 3 .
Fig. 3.The conversion of Mn-based oxygen carrier versus time during the O 2 release.

Fig. 5 .
Fig. 5.The O 2 release amounts as a function of time based on the MnO 2 conversion.

Fig. 6 .
Fig. 6.Plots for the reaction rate constant k determination from C1 model for MnO 2 .

TFig. 9 .
Fig. 9.The O 2 release amount as a function of time based on the Mn 2 O 3 conversion.

Fig. 10 .
Fig. 10.Plots for the reaction rate constant k determination from A2 model for Mn 2 O 3 .

Fig. 12 .
Fig. 12. OC conversion versus time for Mn 2 O 3 during the O 2 release.

Table 1 .
Summary of G(α) equations for different reaction mechanisms.

Table 3 .
R 2 values for fitting different reaction mechanisms based on the Mn 2 O 3 conversion during the O 2 release.