Deliverable D6.2 Report on reaction mechanism and kinetic parameters

Transcription

1 Grant agreement no: Project acronym: SUCCESS Project full title: Industrial steam generation with 100% carbon capture and insignificant efficiency penalty - scale-up of oxygen carrier for chemical-looping combustion using Environmentally sustainable materials Collaborative project Theme: FP7 ENERGY Deliverable D6.2 Report on reaction mechanism and kinetic parameters Due delivery date: 31/08/2015 Actual delivery date: 11/04/2017 Lead beneficiary: Partner no. 3 - CSIC Dissemination level: Public (PU)

2 DELIVERABLE 6.2 SUCCESS - Industrial steam generation with 100% carbon capture and insignificant efficiency penalty - scale-up of oxygen carrier for chemicallooping combustion using Environmentally sustainable materials Work package: 6 Detailed reactivity investigation, modelling and design optimization Involved partners: CSIC CSIC- J. Adánez, A. Abad Chalmers- T. Mattisson SINTEF- Y. Larring, M. Pishanhang, M. Sunding IFPEN- A. Tilland, D. Chiche Objective Report on reaction mechanism and kinetic parameters Keywords: oxygen carriers, reduction kinetics, oxidation kinetics, Cu based oxygen carrier, Fe based oxygen carrier, TGA, fixed bed reactor, reaction mechanism Dissemination level: Public (PU) To determine reaction mechanism and detailed reduction and oxidation kinetics of developed oxygen carriers inside WP1 and WP2. C28 Abstract or description Fundamental kinetic data of selected oxygen carrier materials are needed for fuel and air reactor modelling and scale-up purposes. In this deliverable, detailed reduction and oxidation kinetics of Cu based, Fe based and C28 oxygen carriers with CH 4, H 2 and CO for oxidized carrier and O 2 for reduced carrier are determined by TGA and a fixed be reactor setup. Moreover, the oxygen uncoupling kinetics is also determined for C28 material. 2

3 Executive summary The objective of this investigation was to determine the reaction mechanism and the detailed kinetic parameters of the developed oxygen carriers inside WP1 and WP2. Overall fuel conversion kinetics for the main constituents CH 4, H 2, CO as well as re-oxidation kinetics with O 2 and O 2 generation will be determined using different methodologies allowing a critical cross checking. So CSIC carried out a complete kinetic determination by thermogravimetric analysis of a Cu based, Fe based and a perovskite type C28 oxygen carriers. At IFPEN it was carried out the determination of the main part of redox reactions for a Cu based and C28 oxygen carriers using a fixed bed kinetic device. The kinetics of oxygen release from particles by CLOU mechanism in oxygen free environment was also determined in a batch type fluidized device. At SINTEF was both the CLOU effect and combustion studied at original materials and compared with the large-scale fabrication in fixed bed, and the maximum CLOU capacity integrated over 20 min oxygen release into inert. The capacities were found to be around 3.5% for pure materials while for the large scale batch the CLOU capacity of 2% was found. In-situ XRD was also used to determine the quantitative phase content of the industrial scale sample both in air and under inert and reducing condition. The results shows that less perovskite phase was formed than expected, calling for further optimization of the sintering condition for the up-scaled material for achieving more homogeneity and more of the wanted perovskite phase. Pulsed reaction scheme to study the material ability to convert CH 4 as a function of its degree of reduction in a TPX module and in-situ XRD for investigating reactions related with to reduction of oxygen carriers and phase changes. It was obtained one increased understanding of the reaction pathways in the developed oxygen carriers for chemical-looping systems, including quantitative expressions for intrinsic kinetics necessary for predictive modelling of CLC systems. Developed oxygen carriers exhibited high reactivity in both reduction and oxidation reactions. The effect of temperature and reacting gas concentration on the reaction rate was determined. It was observed that the reaction rate increased with temperature and gas concentration. C28 material showed an activation process for reduction reaction with the redox cycles. During the activation process the reactivity with fuel gases was increased while the oxygen transport capacity decreased around a 7 %. The oxygen uncoupling property was maintained during redox cycles, although a decrease in its reactivity was observed after activation. The oxidation reactivity was not affected by the cycle number. The Shrinking Core Model (SCM) in the grains with plate-like geometry was used to determine the kinetic parameters for impregnated Cu14-γAl-SCCa and Fe20αAl-SG, while spherical geometry was used for the spray dried C28 material. Chemical reaction control and diffusion through the product layer were considered in the model. The chemical reaction was relevant during a first stage of the conversion, but the limiting step quickly changed to diffusion in the product layer at a determined conversion value, which mainly depended on the reacting temperature. The reaction models were able to predict the experimental results under different gas concentration, temperature or reaction type. The kinetic data obtained in this work will be useful to be included in mathematical models of CLC reactors with these materials as oxygen carrier. The choice of the kinetic parameters must be suited to the characteristics of the oxygen carrier used. Moreover, oxygen uncoupling process can be relevant for C28 and this mechanism should be considered in fuel reactor modelling. 3

4 Table of Contents 1 Report on reaction mechanism and kinetic parameters Characterization of developed oxygen carriers during redox reactions Reactivity and structural characterization of oxygen carriers... 6 Reactivity of oxygen carrier developed in WP1 for CLC and CLOU Determination of redox kinetics and reaction mechanism by a fixed bed reactor setup Introduction Experimental device Reactor modelling Gas phase modeling Solid phase modeling Kinetic modeling Study of reduction-oxidation cycles over the CaMn 0.775Mg 0.1Ti 0.125O 3-δ perovskite material (from WP1) Case of CO oxidation Case of H 2 oxidation Case of CH 4 oxidation Oxidation step Optimized kinetic parameters Study of reduction-oxidation cycles over the CuO/Al 2O 3 material (from WP2) Case of CO oxidation Case of H 2 oxidation Case of CH 4 oxidation Oxidation step Optimized kinetic parameters Comparison of the two materials Conclusions Kinetic determination using a thermogravimetric analizer setup Experimental Material Experimental setup Experimental procedure Determination of Cu14γAl-SCCa and Fe20αAl-SG redox kinetics Experimental results Kinetic model for impregnated particles

5 Determination of kinetic parameters Kinetics determination for fresh and activated C28 material Effect of gas concentration Effect of reacting temperature Kinetic model for C Determination of kinetics parameters Conclusions Nomenclature References Appendices Summary of the tests performed on CaMn 0.775Mg 0.1Ti 0.125O 3-δ oxygen carrier Summary of the tests performed on CuO/Al 2O 3 oxygen carrier

6 1 Report on reaction mechanism and kinetic parameters This deliverable corresponds to the investigations carried out inside Task 6.1. The aim of this investigation was to determine the reaction mechanism and the detailed kinetic parameters of the developed oxygen carriers inside WP1 and WP2. Overall fuel conversion kinetics for the main constituents CH 4, H 2, CO as well as re-oxidation kinetics with O 2 and O 2 generation will be determined using different methodologies that can allow a critical cross checking. So CSIC carried out a complete kinetic determination by thermogravimetric analysis and IFPEN carried out a determination in a fixed bed kinetic device. The kinetics of oxygen release from particles by CLOU mechanism in oxygen free environment has been determined in a batch type fluidized device at Chalmers. Pulsed reaction scheme to study the material ability to convert CH 4 as a function of its degree of reduction in a TPX module and in-situ XRD for investigating reactions related with to reduction of oxygen carriers and phase changes. It has been obtained one increased understanding of the reaction pathways in chemical-looping systems including quantitative expressions for intrinsic kinetics necessary for predictive modelling of CLC systems in Task Characterization of developed oxygen carriers during redox reactions. Reactivity and structural characterization of oxygen carriers SINTEF has performed analysis and characterization of the fresh CMTM samples. The samples prove spherical structures for samples from VITO (ELK-SAC small batches for analysis) and from EuroSupport (large upscaled batch). However, they differ in microstructure and grain sizes. The fresh VITO CMTM C28 samples' SEM-EDS analysis are presented in Figure 1. The material is spherical with minimal doughnut or necking. Higher resolution SEM reveals that the grain sizes are in the range of 5 to 10 μm. The EDS shows that Mg is not dissolved in the perovskite structure and separate Mg grains are available separate from the perovskite matrix structure (blue areas). The CMT matrix also consists of Ti rich and Mn rich areas. Both separation of Mg from the matrix and Ti / Mn rich areas in the CMT perovskites structures are known to be caused by thermodynamic properties of this oxide system. Figure 1: Fresh materials, CMTM-Vito SEM-EDS analysis of fresh EuroSupport CMTM C28 samples are presented in Figure 2. Although similar in nominal composition, this material differs from VITO's in microstructure and grain sizes. The material's sphericity is less than that of the previous one and some necking and doughnut particles are observed. The microstructure is also quite different and the grain sizes are much smaller. The grain 6

7 sizes are in the range of less than 1 μm. As well there are lacks of homogeneity in the chemical composition of the structure, for example there are unreacted CaO rich areas in the microstructure which has not taken part in the CMT perovskite phase formation. The small grain size and lack of complete perovskite structure formation is due to the low sintering temperature compared to the VITO sample. It should be noted that the mechanical strength is still the same the one carrier sintered at higher temperature (i.e. VITO), but the lack of homogeneity might reduce performance under redox. Figure 2: Fresh materials, CMTM-EuroSupport. SINTEF has performed pulsed and continues TPX in order to study the behavior of different CMTM samples. Continues TPX reactor tests of three different samples with different stoichiometry have been performed to check their CLOU performance and eventually differences in the reaction rates. The CLOU performance and capacity is very important for securing full combustion of the fuel used in CLC. The test was conducted on ELK-SAC 001 = CaMg0.1Ti0.125Mn0.775O3 => Ca1.1Ti0.14Mn6O3 + MgO ELK-SAC 002 = Ca0.9Mg0.1Ti0.125Mn0.775O3 => CaTi0.14Mn6O3 + MgO ELK-SAC 003= Ca5Mg0.1Ti0.125Mn0.775O3 => Ca0.95Ti0.14Mn6O3 + MgO In addition such test was performed on up-scaled CMTM from EuroSupport with the same nominal composition as ELK-SAC 001, and on Cu impregnated Al 2O 3 produced by JohnsonMattey. ELK-SAC 001 is the stoichiometric ratio of cations and the flow of gases were kept constant at 30 ml/min through reactor of ~1cm 2, and a bed mass of 0.5 g. Each test was run at four temperatures of 950, 900, 850, 800 C at each temperature a CLOU test was performed by 60 min 5% N 2 and 5% O 2 in He then, 32 min He before re-oxidation by 30 min 5%N 2 and 5%O 2 in He (bottle mixture). Figure 3 a) shows the CLOU effect under reduction in He while Figure 3 b) shows the re-oxidation of the oxide. This CLOU effect was calculated both on reduction (released oxygen) and oxidation (consumed oxygen). Figure 4 a) shows the oxygen released and consumed after subtraction of N 2 which was used as reference gas. CLOU capacity for the materials based on reduction was found to fit the oxidization when the MSbackground level of N 2 was subtracted from N 2, the N 2 was used to correct for oxygen produced. The oxidation part, which is much quicker and where background noise is not a part of the subtraction, is the best way to establish the CLOU capacity. The oxygen release as a function of time is shown in Figure 4.b for all the materials tested. The CMTM ES is equal to ESac001, the difference is that ESAC's materials are produced in laboratory using low grade chemicals and high firing temperature to get relatively phase pure samples, while CMTM ES (C28 Batch G00x) is made from raw materials and lower sintering temperature than intended leading to wanted leading to less homogeneous sample. This is also reflected in the much lower capacity for the CMTM ES than the ELC-SAC001 sample. The CMTM 7

8 EuroSupport will though get a higher quality if better solution for heat treatment is found for largescale batches, which will be reflected in higher CLOU capacity if achieved. The amount CLOU capacity estimated from oxygen release and oxygen absorption is in the same range with a deviation varying between 1-8%. Figure 5 shows the CLOU capacity as a function of time (30 min) at different temperatures ( C). The CLOU capacity decreases with decreasing temperature and shows that at 800 C the CLOU capacity is reduced. In Figure 6 the CLOU capacity is collected and plotted vs temperature for oxygen release during 30 min. The ElKSac 001 material which is the stoichiometric composition has the highest CLOU capacity at all the temperatures of study. This is due to the higher amount of CMT perovskite structure in this sample. On the other hand, the CMTM EuroSupport sample exhibits much lower CLOU capacity compared to its nominal value, due to its lack of complete intermixing and unreactive CaO left over in the structure. The CLOU capacity of this material is comparable to that on Cu-Al 2O 3, which will be discussed below Figure 3 Shows the a) oxygen release (CLOU) in He, b) the oxygen consumed under oxidation Figure 4 a) show the released and consumed oxygen vs time, b) integration of oxygen released under reduction in He and consumed under oxidation 8

9 Figure 5 CLOU release capacity as a function of temperature in He up to 30 min for the temperatures from C Figure 6 The CLOU capacity of the selected OCMs at different temperatures obtained from the TPX experiments In order to compare the performance of both ESac001 which was selected material for up-scaling, and CMTM EuroSupport samples, a similar TPX study of reduction and oxidation was performed on both samples. Reduction was performed under CH 4 and oxidation under 5% N 2 and 5% O 2 in He was performed, shown for ESac002 in Figure 7. 9

10 Figure 7 Combustion of 10%CH4, shows that for this fixed bed approach is the oxygen release limiting for methane conversion. Very small reforming activities are observed. The concentration of CO 2 in outlet gases is shown in Figure 8. CO formation is very low during CH 4 combustion. On the right side of the figures at times above 100 minutes were oxidation is performed, it is interesting to observe that no CO 2 release is observed for ESac001. There is a very small CO 2 release bump during oxidation for the CMTM ES sample. This is due to the free CaO left in the micro-structure which is prone to carbonate formation during reduction and release of CO 2 during oxidation. Evidently, proper sintering and homogeneity in the structure such as observed in VITO and ESac001 sample hinders this unwanted effect. On the other hand, the performance of the ESac001 sample exceeds CMTM EuroSupport at all temperatures of study, due to larger capacity and faster exchange of oxygen in perovskite's structure. Figure 8 The concentration of CO 2 in outlet gases. Reduction was performed under CH 4 and oxidation under 5% N 2 and 5% O 2 in He was performed. a) CMTM EuroSupport sample. b) ESac001 sample SINTEF also performed in-situ XRD analysis on the best performing CMTM material, i.e. ELK-SAC 001. The experiments were performed at room temperature (Table 1), at 900 C in air (Table 2), and at 900 C in N 2 (Table 3). The goal was to observe any phase transformations in air and during the CLOU performance. Approving the SEM-EDS results, the fresh sample contains unreacted MgO, and magnesium does not intermix into the perovskite structure. The fresh sample contains around ~8wt% of Ca 2MnO 4 Ruddlesden-Popper (RP) phase and around 12wt% of CaMn 2O 4 spinel phase. Heating up to the 900 C service temperature combines some of these structures back into the perovskite structure (Table 2), which is a more stable phase at this thermodynamic condition. During the CLOU (Table 3), some part of the perovskite structure is dissociated back into these two phases again. Table 1: The in-situ XRD results for phase composition of ELK-SAC 001 at room temperature 10

11 Phase Content Composition Unit Cell Cell Volume Perovskite (Orthorhombic) Perovskite (Orthorhombic) ~68 wt% ~CaMn 0.9Ti 0.1O 3 a= Å, b=7.4963å, c=5.2888å ~8 wt% ~CaMn Ti O 3 a=5.4042å, b=7.6323å, c=5.3475å RP phase ~8wt% ~Ca 2MnO Å 3 22Å 3 Spinel ~12wt% ~CaMn 2O 4 Magnesium Oxide ~4wt% ~MgO Table 2: The in-situ XRD results for phase composition of ELK-SAC 001 at 900 C in air Phase Content Composition Unit Cell Cell Volume Perovskite (Rhombohedral) ~75 wt% ~CaMn 0.9Ti 0.1O 3 a=5.3603å, c= å Å 3 Perovskite (Cubic) ~8 wt% ~CaMn Ti O 3 a=3.8524å 22Å 3 RP phase ~5wt% ~Ca 2MnO 4 a=5.2967å, c= å Spinel ~9wt% ~CaMn 2O 4 a=3.2076å, b= å, c=9.7377å Magnesium Oxide ~4wt% ~MgO Table 3: The in-situ XRD results for phase composition of ELK-SAC 001 at 900 C in N2 Phase Content Composition Unit Cell Cell Volume Perovskite (Cubic) ~62 wt% ~CaMn 0.9Ti 0.1O 3 a= Å 3 Perovskite (Cubic) ~8 wt% ~CaMn Ti O 3 a=3.8527å 22Å 3 RP phase ~9wt% ~Ca 2MnO 4 a=5.3054å, c= å Spinel ~16wt% ~CaMn 2O 4 a=3.2091å, b= å, c=9.7419å Magnesium Oxide ~4wt% ~MgO a=4.2714å The SEM-EDS analysis of the fresh Cu-Al 2O 3 samples from JohnsonMattey are presented in Figure 9. This batch is of OCM is the small batch fabricated for analysis. The material is fairly spherical and homogenous in structure indicating appropriate intermixing, and the particle size range from 80 to 220 μm. There are however Cu rich areas, as well as elemental unreacted copper islands on the surface. 11

12 Figure 9. Fresh materials, Cu-Al2O3 JohnsonMattey Figure 10 The concentration of different gasses at 950 C (left) and CO 2 in outlet gases at different temperatures (right). Reduction was performed under CH 4 and oxidation under 5% N 2 and 5% O 2 in He was performed for the Cu-Al 2O 3 JohnsonMattey sample. The TPX outlet gases concentration over Cu-Al 2O 3 JohnsonMattey sample during CH 4 combustion is shown in Figure 10, and shows full combustion until break-through of CH 4 where reforming and then coking starts. This shows that the Cu material is also important for the combined catalytic conversion and combustion of CH 4. The right side of Figure 10 shows the CH 4 conversion as a function of temperature. The conversion of CH 4 is fast in the temperature range C where full conversion of the CH 4 is achieved. On the right side of the figures at times above 100 minutes were oxidation is performed, there exists some unwanted CO 2 release. This is due to incomplete oxidation of methane during reduction and coke formation during the reduction step. The coke is oxidized during the oxidation step, and produces CO 2. This will not be the case in CLC where the full capacity will not be utilized. Comparison of ESac002 and Cu/Al 2O 3 is shown in Figure 11. It shows that the ESAC has higher oxygen transport capacity but since it is inert for CH 4 the break-through is faster for this sample compared to Cu/Al 2O 3. The Cu/Al 2O 3 sample has less capacity but have excellent combustion efficiency. Cu has though a tendency to agglomerate at the surface and therefore is slightly removed by time from the surface through attrition. ESac samples could have benefited from better catalytic properties and higher surface area/smaller grains for faster release of its capacity. Given the slower combustion rate compared to Cu/Al 2O 3, CMT required higher solid to gas ratio (or circulation rate) to convert the same amount of gas. 12

13 Figure11 Comparison between ElkSac and Cu/Al 2O 3 at 950 C. Reduction was performed under CH 4 and oxidation under 5% N 2 and 5% O 2 in He. Reactivity of oxygen carrier developed in WP1 for CLC and CLOU At Chalmers the oxygen carrier selected for upscaling in WP1 was selected for determination of reactivity using syngas, methane and char. The study was performed in a batch fluidized bed reactor with these fuels, with the main ambition to determine the main reaction mechanism for oxygen release from the particle bulk, i.e. whether the oxygen is transferred via CLC or CLOU. As part of the project, additional two materials based on the C28 material were made, but here containing somewhat less Ca in the formulation compared to the reference material, see Table 4 The idea was based on the fact that the original material actually is over stoichiometric with respect to Ca, due to the migration of the Mg out of the perovskite structure. This has been confirmed in several studies. By reducing the amount of Ca, the propensity to react with sulphur is likely lower, as was investigated in T1.5. Table 4 shows the composition of the three spray-dried materials which were prepared at VITO. Again the C28E1S2_Ref is the material which was selected for up-scaling based on the research conducted in WP1. Table 4 Composition of the oxygen carrier tested at Chalmers. Notation Composition Ca/(Mn+Ti) Sintering Crushing strength [N] Tap density [kg/m 3 ] Attrition index [wt%/h] C28E1S2_Ref CaMn Ti Mg 0.1 O 3-δ (C28E1S2) >1 4h1335C C28E1S2_Stoich C28E1S2_Under CaMn 6 Ti 0.14 Mg 0.11 O 3-δ 1 CaMn 0.96 Ti 0.15 Mg 0.12 O 3-δ <1 4h1335C h1335C All materials from Table 4 were tested in a laboratory fluidized bed reactor using both gaseous fuel and solid char, see D1.3 for further details of the experimental procedure. Figure 4 shows the results of the tests conducted at 950 C using methane as fuel. Here using a 15 g sample of oxygen carrier and a fuel flow of 900 ml/min, normalized at 1 bar and 0 C. Evidently, the reactivity of the material as a function of solids conversion, i.e. ω, is similar for the materials, with the reference material having somewhat higher reactivity together with the under-stoichiometric oxygen carrier. 13

14 In order to gauge the oxygen release rates, or CLOU properties, experiments were conducted using devolatilized wood char, which was prepared through devolatilization in a TGA at 1000 C. For a detailed review of the procedure see for example (Arjmand et al. 2014). A sample of 0.1 g of char was introduced to a fluidized 10 g bed of oxygen carrier particles ( µm) at several temperatures. As it is not expected that the solid-solid reaction between oxygen carrier and char will occur at any appreciable rate, the released CO 2 is expected to be a good gauge of the rate of oxygen uncoupling. Figure 13 a) shows the mass based conversion for the three materials in Table 4. As was expected the oxygen release on a mass-basis corresponded to roughly 1 wt%, with highest rate for the reference material. The switch in rate between inert and char, clearly indicated in the figure, illustrates the higher driving force for CLOU Gas yield, γ CO2 0.7 C28E1S2_Ref C28E1S2_Stoich C28E1S2_Under Mass based conversion, ω Figure 12 Reactivity of C28 oxygen carriers in the batch fluidized bed reactor at 950 C using methane as fuel. For details of experimental procedure, see D1.3. when char is introduced to the well-mixed bed. Figure 13 b) shows a similar plot, but for 900 and 1000 C. Both the rates of release and oxygen capacity increase with reaction temperature, and the oxygen capacity approaches 1.5 wt% at the highest temperature for the reference material. 1 N 2 Char 1 Mass-based conversion, ω Mass-based conversion, ω Time(s) Time(s) a) b) Figure 13 Reactivity of C28 oxygen carriers with char in the batch fluidized bed reactor at a) 950 C and at b) 900 C and 1000 C. 14

15 In order to judge the effect of CLOU on the reactivity with methane, the rates were compared in Figure 14. Included for comparison are the results using syngas, i.e. 50% H 2 and 50% CO. As the slope of the reaction, dω/dt, is similar for the reactions with char and methane initially, it is likely the rates of reaction are initially governed by the CLOU reactions at this temperature, i.e. 950 C, albeit the CLOU rates drop of in comparison to the overall rate as the degree of solids conversion approaches ω=0.995, after which the reaction is likely governed by the direct gas-solid reaction of methane with the oxygen carrier. With respect to syngas, the overall oxygen transfer is always greater than the oxygen transfer by CLOU, indicating that the rates are likely governed by direct reaction in the entire span of oxygen carrier conversion. 1 Mass-based conversion, ω CH 4 Char Syngas (CO,H2) Time, s Figure 14 Comparison of rates of oxygen transfer for the C28E1S2_Ref 3 Determination of redox kinetics and reaction mechanism by a fixed bed reactor setup. Introduction. Task 6.1 is focused on the determination of kinetics and reaction mechanisms of both selected oxygen carrier materials from WP1 and WP2. Kinetics of redox reactions of oxygen carriers has been determined by two independent methods: by thermogravimetric analysis (ICB-CSIC) and in a fixed bed reactor (IFPEN). The reactivity during redox cycling of oxygen carriers from WP1 and WP2 has been studied at IFPEN in a fixed bed reactor device (T062 experimental device). A coupled hydrodynamic and reaction model has been developed for the experimental data interpretation (information on reaction kinetics) and mechanism determination. Due to the configuration of the experimental set-up, mass transfer is controlled and is modelled accurately. Experimental device The experiments have been carried out on a IFPEN device (T062 installation), which consists in an inverse gas chromatography set-up. A schematic representation of the device is given in Figure 15. The oxygen carrier material to be tested is disposed in a fixed bed quartz reactor (diameter 14 mm). Experimental tests require between ~1 and ~5 grams of material as a function of material reactivity, i.e. oxygen capacity and kinetics. Oxidation reduction cycles can be programmed. Gaseous reactants and products are analysed by an on-line mass spectrometer (Hiden HPR20 model) for O 2, CO, CO 2, CH 4 analysis, and a µ-gc analyser (molsieve 5A separation column + Thermal Conductivity Detector) for H 2 analysis. 15

16 Introduction of water in the gas phase during the reduction step is possible, using a saturator device. A water condenser is also installed downstream the fixed bed reactor to remove water from the gas, that may interfere with gas analysis performed by mass spectrometry. Figure 15. Schematic representation of the T062 device. Operating parameters ranges explored for the present study are given in Table 5. Table 5. Operating parameters ranges explored for reduction-oxidation cycles (with GHSV the Gas Hourly Space Velocity, and i the interstitial porosity). Experiments operating conditions ranges Oxygen carrier CaMn0.775Mg0.1Ti0.125O3-δ Perovskite CuO/Al2O3 SUCCESS reference C28E1S2-1350_4_L_V1 S0397/1034 Reactor dimensions INT 43: Ø = 1.4 cm, hmax = 4 cm Amount of oxygen carrier (OC) 1 2 g 2 4 g Filling height cm cm Reducing gas H2, CO or CH4 in He 5%v 10%v Eventually mixed with CO2 20%, H2O ~18% Total gas flow rate (NL.h -1 ) 2 4 T ( C) P (bar) ~ 1 Residence time (= 3600 εi/ghsv(t,p)) ~5 0.1 s ~ ~ s GHSV, calculated for the operating conditions (T, P) ~ h -1 ~1800 ~8000 h -1 The experiments are operated through the following steps, as schematically represented in Figure 16: 1. Temperature increase under oxidizing conditions (O 2 in He), 2. Once operating temperature is reached, inerting under He flow, 3. Oxidation step, 16

17 4. Inerting step (under He), 5. Reduction step, 6. Repetition of steps 2 to 5, 7. After last cycle, an oxidation step is carried out, 8. Cooling in static conditions under He. Figure 16. Cycling red-ox experiments proceedings, and operating conditions. Table 11 and Table 12 with summary of tests performed on CuO/Al 2O 3 and perovskite oxygen carriers are presented in appendix. For each experiments, 10 consecutive reduction-oxidation cycles have been carried out to monitor any evolution of material reactivity over cycling. In the fixed bed device used, and for the operating conditions ranges evaluated, reactivity stabilization was noticed between 6 th and 10 th cycles. For both materials, and for all the experiments, results obtained for the 9 th cycle were considered for data modelling. This cycle was used as a reference to get the most stable results as possible (to lower ageing effects). Reactor modelling The fixed bed reactor used in this study is modeled as a plug flow reactor with axial dispersion. This model relies on some hypotheses: - there are no internal heat and mass limitations in the grains, - there are no external heat and mass limitations at the surface of the grains, - the reactor is supposed to be isotherm, - the grain volume is constant throughout the experiments, - the gas phase has a plug flow behavior with axial dispersion. The operating conditions have been chosen in order to avoid any limitation at the surface of the grains. Experiments with various grain sizes have confirmed the absence of internal limitations. Besides, the small height of the reactor and the low reactant concentrations used lead to small heat generation in the reactor. Gas phase modeling The variation of the gaseous components concentration along the reactor is predicted with a typical model considering an ideal plug flow reactor with axial dispersion (Schweich 2011) The axial dispersion 17

18 coefficient is calculated as a function of the reactor diameter from the Gunn correlation(schweich 2011 b). Solid phase modeling The oxygen carriers used are progressively reduced and re-oxidized in this study. For both materials, three phases are considered depending on the reduction degree of the material (Table 6). Table 6: Phases considered for both materials depending on the reduction degree. Material CuO/Al 2O 3 CaMn 0.775Mg 0.1Ti 0.125O 3 (ABO 3- δ) Phase: Totally oxidized CuO ABO 3 Phase: Intermediate Cu 2O ABO 3- δclou Phase: Reduced Cu ABO 3- δclc ABO 3-δCLOU corresponds to the perovskite material when a part of its oxygen has been removed by oxygen uncoupling effect (CLOU), and ABO 3-δCLC corresponds to the solid totally reduced. Indeed, the oxygen carriers studied can release their oxygen in two different ways: - by oxygen uncoupling (Chemical Looping with Oxygen Uncoupling: CLOU): gaseous oxygen is spontaneously released by the material in reductive atmosphere. 2CCCCCC 1 2 OO 2 + CCCC 2 OO AAAAAA 3 δδ CCCCCCCC OO AAAAAA 3 δδcccccccc - by reduction (Chemical Looping Combustion: CLC): the reductive gases (H 2, CO, CH 4) take the oxygen directly from the solid. (2nn + mm)cccc 2 OO + CC nn HH 2mm nncccc 2 + mmhh 2 OO + 2(2nn + mm)cccc (2nn + mm) (δδ CCCCCC δδ CCCCCCCC ) AAAAAA 3 δδ CCCCCCCC + CC nn HH 2mm (2nn + mm) nncccc 2 + mmhh 2 OO + (δδ CCCCCC δδ CCCCCCCC ) AAAAAA 3 δδ CCCCCC The oxygen uncoupling effect can be observed even with inert gases, but cannot lead to a complete reduction of the material. When the material is totally oxidized the oxygen uncoupling effect is dominating. Thus, the materials will release gaseous oxygen before being directly reduced by the reacting gases. A comparison between the quantity of oxygen exchanged by the materials during the reduction and oxidation steps indicates that a portion of oxygen is released during the inert step performed between reduction and oxidation steps. Indeed, during this inert step, gaseous oxygen has been measured at the outlet of the reactor. This oxygen came from the oxygen uncoupling effect. An analysis of the quantities of oxygen exchanged indicates that copper oxide material releases 17% of its oxygen during these 15 minutes while the perovskite material releases 14% of its available oxygen. Experiments with longer inert duration indicate that the copper material released 3 more percent of oxygen while the perovskite does not release any more oxygen. Thus, in the model presented here it is considered that the reaction of oxygen uncoupling is not significant during the reduction step. The oxygen content of the materials at the beginning of the reduction step has to be calculated taking into account the quantity of oxygen lost by CLOU effect. For perovskite material, the δδ CCCCCCCC value takes into account this quantity of oxygen, so the ABO 3 phase will not be considered during the reduction of 18

19 the material. In the case of copper oxide material the composition of an initial mixture of Cu 2O and CuO is calculated (%CuO equation 2) and %Cu 2O (equation 3). During the reduction step of the tests presented in this work the material is totally reduced. As a consequence the quantity of oxygen recovered by the materials during the oxidation step (nn oooo ) determines the oxygen transport capacity (OTC) (equation 4) of the materials and the δδ CCCCCC value for perovskite (equation Error! No se encuentra el origen de la referencia.). The difference between δδ CCCCCCCC and δδ CCCCCC represents the quantity of oxygen exchanged by the perovskite during the reduction step (nn rrrrrr ). δδ CCCCCCCC = (nn oooo nn rrrrrr )MM oooo 1 mm oooo %CCCCCC = (2nn rrrrrr nn oooo ) nn rrrrrr %CCCC 2 OO = (nn oooo nn rrrrrr ) nn rrrrrr oooooo = nn oooomm OO 4 mm oooo δδ CCCCCC = nn oooomm oooo 5 mm oooo Being: δ CLOU the fraction of oxygen lost by oxygen uncoupling effect (-), n ox : the quantity of oxygen taken by the material during the oxidation step (mol), n red : the quantity of oxygen released during the reduction step (mol), M oc : the molecular mass of the oxygen carrier (g mol-1), m oc : the oxygen carrier mass (kg), %CuO: the fraction of CuO at the beginning of the reduction step (%), %Cu 2 O: the fraction of Cu2O at the beginning of the reduction step (%), otc: the oxygen transport capacity of the materials (kgo kgoc-1), M O : the molecular mass of oxygen (g mol -1 ) and δ CLC : the fraction of oxygen given in the reduced form of the perovskite (-). The variations of the solid phase (CuO, Cu 2O, Cu, AAAAAA 3 δδcccccccc and AAAAAA 3 δδcccccc ) concentrations depend only on the reaction rate. The reaction rates have to be chosen in order to predict correctly the experimental results. Kinetic modeling Choice of a kinetic scheme The kinetic scheme selected for the copper oxide material considers the partial oxidation of methane in carbon monoxide and hydrogen followed by the oxidation of the latter with both CuO and Cu 2O. Besides, reduced copper catalyzes methane reforming and water gas shift reactions in these experiments. In order to correctly predict the experimental results, a combination of the steam methane reforming reaction and water gas shift reaction is considered (r7) along with the reverse water gas shift reaction (r8). CCCC 4 + 2CCCCCC CCCC 2 OO + 2HH 2 + CCCC CCCC 4 + CCCC 2 OO 2CCCC + 2HH 2 + CCCC HH 2 + 2CCCCCC CCCC 2 OO + HH 2 OO HH 2 + CCCC 2 OO 2CCCC + HH 2 OO CCCC + 2CCCCCC CCCC 2 OO + CCCC 2 CCCC + CCCC 2 OO 2CCCC + CCCC 2 r1 r2 r3 r4 r5 r6 19

20 CCCC 4 + 2HH 2 OO 4HH 2 + CCCC 2 HH 2 + CCCC 2 CCCC + HH 2 OO r7 r8 For perovskite a similar kinetic scheme has been selected. It takes into account the reduction of AAAAAA 3 δδcccccccc in AAAAAA 3 δδcccccc by methane, hydrogen and carbon monoxide. Even though the catalytic effect of perovskite material is really poor, the same two catalytic reactions are considered. AAAAAA 3 δδcccccccc + (δδ CCCCCC δδ CCCCCCCC )CCCC 4 (δδ CCCCCC δδ CCCCCCCC )HH 2 + (δδ CCCCCC δδ CCCCCCCC )CCCC + AAAAAA 3 δδcccccc AAAAAA 3 δδcccccccc + (δδ CCCCCC δδ CCCCCCCC )HH 2 (δδ CCCCCC δδ CCCCCCCC )HH 2 OO + AAAAAA 3 δδcccccc AAAAAA 3 δδcccccccc + (δδ CCCCCC δδ CCCCCCCC )CCCC (δδ CCCCCC δδ CCCCCCCC )CCCC 2 + AAAAAA 3 δδcccccc CCCC 4 + 2HH 2 OO 4HH 2 + CCCC 2 HH 2 + CCCC 2 CCCC + HH 2 OO r1 r2 r3 r4 r Kinetic laws selected Various models have been developed to describe reduction and oxidation of oxygen carriers. The most frequently used models are: - The Changing Grain Size model (CGS model), - The Shrinking Core model, - The nucleation and nuclei growth model (Avrami-Erofeev). During the oxygen carrier reduction the grains are composed of a layer of reduced phase and a core of oxidized phase. The gaseous reactants diffuse through the layer of reduced phase until the core of nonconverted solid to react. The properties (eg: density) of the reduced and oxidized phases differ from each other which can lead to a variation of the grain size. The Changing Grain Size model takes into account this variation of grain size and the diffusion of the gaseous species until the reacting core of the grain. This model predicts the evolution of the grain and un-reacted core radius. The Shrinking Core model is similar to the CGS model aside it does not take into account the change of grain size. These two models consider that the material is homogeneously consumed over its entire surface. From this point of view, the nucleation and nuclei growth model differs since it represents the formation of nuclei at the surface of the grain, which will form a layer of product more or less homogeneous while growing. In this work, the Shrinking Core model and the nucleation and nuclei growth model have been selected to represent the experimental results. According to the simulation data, the reaction of CuO reduction in Cu 2O is well represented by Avrami-Erofe ev equation. The reduction of Cu 2O and perovskite material are represented by the Shrinking Core model considering diffusion limitation, reaction limitation or both limitations depending on the material and the reducing gas. Typical kinetic laws are used to represent the catalytic reactions taking into account the equilibrium constants. In section and are presented the kinetic laws selected respectively for the perovskite material and the copper oxide material. The kinetic parameters of the various kinetic laws are determined by optimization. Their values are chosen so that the experimental data can be predicted with the model. 20

21 Study of reduction-oxidation cycles over the CaMn0.775Mg0.1Ti0.125O3-δ perovskite material (from WP1) Various operating temperatures have been studied in order to determine the kinetic parameter values of the kinetic laws of each reducing gas oxidation. The operating conditions of the tests used for the kinetic determination are presented in Table 11 in appendix. In Table 8 are summarized the kinetic laws selected and the values of the kinetic parameters obtained by optimization. Case of CO oxidation The carbon monoxide and dioxide breakthrough curves are predicted with the model using the kinetic parameters of carbon monoxide oxidation reaction optimized for this purpose. In Figure 17 are presented the breakthrough curves of carbon monoxide for three different temperatures. An experiment conducted at 775 C was not considered for the modeling, as the formation of carbonate observed invalids this temperature for the kinetic measurement. The three breakthrough curves are adjusted in order to obtain 50% of the maximal CO concentration value at the same time whatever the temperature in order to make easier the comparison. According to the experimental results presented in this Figure 17 the reactivity of the material increases with the temperature (increase of the stiffness of the curve). Figure 17. Breakthrough curves of CO at 3 temperatures (825, 875 and 925 C) for perovskite reduction. The kinetic law selected to represent the perovskite material reduction with CO can represent the competition between the rate of diffusion of the reactive species through the layer of reduced copper and the rate of reaction of the reactive species with the oxidized copper. The values of both diffusion and reaction rates are adjusted with the kinetic constant (kk 0 3 ) and the effective diffusion coefficient (DDDD ii,0 CCCC ) (See Table 8 for the detailed kinetics). Depending on the values determined by optimization it is possible to determine if the reaction rate is controlled by the diffusion rate or the kinetic rate. The comparison between the experimental data and the simulation results is presented in Figure 18. The time is not normalized for these figures. A good agreement is obtained between these results with the kinetic parameters presented in Table 8. 21

22 925 C 875 C 825 C Figure 18. Experimental and simulation results for perovskite reduction with carbon monoxide at 3 different temperatures. Case of H2 oxidation In the case of H 2 oxidation, the same methodology as the one developed for carbon monoxide oxidation is used. The breakthrough curves of hydrogen at four different temperatures are compared in Figure 19. The four breakthrough curves are adjusted in order to obtain 50% of the maximal H2 concentration value at the same time whatever the temperature in order to make easier the comparison Figure 19. Breakthrough curves of H 2 at 4 temperatures (775, 825, 875 and 925 C) for perovskite reduction. As for carbon monoxide oxidation, an increase of temperature induce to an increase of reactivity of the material marked by an increase of the stiffness of the breakthrough curves. 22

23 Note that the kinetic law presented in Table 8 only considers kinetic limitations. This is due to the fact that the small hydrogen molecules can diffuse faster in the material, which avoids any diffusion limitation. The comparison between the experimental data and the simulation results is presented in Figure 20. The normalized time has not been used for this figure. A good agreement is obtained with the kinetic parameters presented in Table C 875 C 825 C 775 C Figure 20. Experimental and simulation results for perovskite reduction with hydrogen at 4 different temperatures. Case of CH4 oxidation The 5 reactions of the kinetic scheme are taken into account in the model. The kinetic parameters of hydrogen oxidation (r2) and carbon monoxide oxidation (r3) determined in the previous parts are used in this model. Thus, only the kinetic parameters of the reactions of methane partial oxidation in carbon monoxide and hydrogen (r1) and the catalytic reactions (r4 and r5) are optimized. (a) (b) 23

24 Figure 21. Breakthrough curves of CH 4 (a), CO 2 (a), CO (b) and H 2 (b) at 3 temperatures (825, 875 and 925 C) for perovskite reduction. The experimental concentration profiles of carbon dioxide and methane at the outlet of the reactor are presented in Figure 21 (a). Hydrogen and carbon monoxide concentration profiles are presented in Figure 21 (b). The perovskite material shows a poor catalytic activity. Indeed, the maximal concentrations measured are around 0.7% for hydrogen and 6% for carbon monoxide. These extremely low values cannot be considered exact regarding the incertitude of the analytical devices. Nevertheless the simulations show that a part of the carbon dioxide measured at the end of the experiments comes from catalytic reactions. It has thus been decided to adjust the kinetic parameters of steam methane reforming reaction (r4) to correctly predict the carbon dioxide concentration and consider the reaction of reverse water gas shift at the equilibrium. This hypothesis is especially valuable since the concentration of water is close to 15% and the quantity of hydrogen produced is low which is unfavorable for the reaction of reverse water gas shift and leads to a really poor carbon monoxide production. The comparison between experimental and simulation results is presented in Figure 22. A relatively good agreement has been obtained between experimental and simulation results even if some differences can be observed. Nevertheless, the tendencies are well predicted. The kinetic constants selected are presented in Table C 875 C 825 C Figure 22. Experimental and simulation results for perovskite reduction with methane at 3 different temperatures. Oxidation step Figure 23 shows O 2 consumption curves obtained during the perovskite oxidation step (data obtained after H 2 reduction). One could notice stiffness of the breakthrough curves, related to high oxidation kinetics. Moreover, oxidation curves do not show significant difference as a function of operating temperature. 24

25 As a consequence, data obtained for O 2 breakthrough were not shown to be not sensitive enough to operating parameters, due to very high oxidation kinetic rate. Therefore, oxidation was not modeled in the present study. Figure 23. O 2 consumption curves (breakthrough curves) obtained for the 9 th reduction-oxidation cycle, as a function of temperature (T = 775 C / 825 C / 875 C / 925 C). Data obtained in the case of H 2 reduction oxidation cycles. 25

26 Optimized kinetic parameters Table8. Optimized kinetic parameters values for the perovskite material reduction by H 2, CO or CH 4 Fuel H 2 CO CH 4 (δδ CCCCCC δδ CCCCCCCC )CCCC 4 + AAAAAA 3 δδcccccccc Reaction (δδ CCCCCC δδ CCCCCCCC )HH 2 + AAAAAA 3 δδcccccccc (δδ CCCCCC δδ CCCCCCCC )CCCC + AAAAAA 3 δδcccccccc 2(δ CLC δ CLOU )H 2 + (δ CLC δ CLOU )CO + ABO 3 δclc (δ CLC δ CLOU )H 2 O + ABO 3 δclc (δ CLC δ CLOU )CO 2 + ABO 3 δclc CCCC 4 + 2HH 2 OO 4HH 2 + CCCC 2 HH 2 + CCCC 2 CCCC + HH 2 OO Kinetic law 3εε SS HH CC 2 AAAAAA 3 δδcccccccc gg CC ss RR gg kk 0 2 ee EE aa2 RRRR 1 XX AAAAAA3 δδcccccccc 2 3 3εε SS RR gg 1 1 XX AAAAAA3 δδcccccccc CC CCCC AAAAAA 3 δδcccccccc gg CC ss DDDD ii,0 CCCC ee EE dddddd RRRR 3 1 XX AAAAAA3 δδcccccccc 2 + RR gg kk 0 3 ee EE aa3 RRRR 1 XX AAAAAA3 δδcccccccc k j E aj (kj/mol) εε SS CC gg CCCC 4 CC ss RR gg 1 1 XX AAAAAA3 δδcccccccc AAAAAA 3 δδcccccccc DDDD ii,0 CCCC4 ee EE dddddd RRRR 3 1 XX AAAAAA3 δδcccccccc 2 + RR gg kk 0 1 ee EE aa1 RRRR 1 XX AAAAAA3 δδcccccccc kk 0 4 ee EE aa4 CCCC RRRR CC 4 HH gg CC 2 OO 2 gg CC HH 4 2 CCCC 2 gg CCgg AAAAAA 3 δδcccccc CC ss MM εε AAAAAA3 δδcccccc ss KK 4 kk 0 5 ee EE aa5 CCCC RRRR CC 2 HH gg CC 2 gg CC HH 2 OO CCCC gg CC gg AAAAAA 3 δδ CC CCCCCC KK ss MM AAAAAA3 δδcccccc εε ss 5 k 10 = k 40 = , r 5 at equilibrium E a1= E a4= i,0 D ej (m 2 /s) / E dif (kj/mol) / K i (-) / / KK 4 = TT KK 5 = TT 2 3

27 Study of reduction-oxidation cycles over the CuO/Al2O3 material (from WP2) For each reducing gas the kinetic parameters have been determined in order to predict the experimental results at various operating temperatures. The values obtained are presented in Table 9. Case of CO oxidation The kinetic parameters of the carbon monoxide oxidation with CuO and Cu 2O are optimized in order to predict the carbon monoxide and dioxide breakthrough curves. In Figure 24 are presented the breakthrough curves of carbon monoxide for four different temperatures. The four curves are adjusted in order to obtain 50% of the maximal CO concentration value at the same time whatever the temperature in order to make easier the comparison. Figure 24. Breakthrough curves of CO at 4 temperatures (800, 850, 900 and 950 C) for CuO reduction. According to the experimental data the reactivity of the material decreases (decrease of the stiffness of the curve) when the temperature increases. In order to explain this behavior, specific surface measurements have been done on four samples which were submitted to reduction/oxidation cycles at different temperatures. The measured surfaces are presented in Table 10 and show a decrease of the material specific surface when temperature increases. In order to take into account the reactivity loss, an activation coefficient has been added in the kinetic laws. This coefficient a is determined as the ratio of the specific surface area of the material at a chosen temperature to the specific surface area of this material at 900 C (equation 6). With: aa = SS ssssss(tt) SS sspppp (900 CC) 6 aa: activation coefficient (-) and SS ssssss (TT): specific surface area of the oxygen carrier at a temperature T (m 2 g -1 ) Depending on the temperature a diffusional limitation is observed at the end of the breakthrough curves presented in Figure 24 (longer return to the inlet concentration value). Lower the temperature is, higher the diffusional limitation is. This limitation occurs when the conversion of the solid is really high (more than 90% of the solid is already reduced) and corresponds to a limitation of the reduction of Cu 2O reduction. Indeed, the modeling study indicates that the beginning of the breakthrough curves is affected by the reduction of CuO while the end of the breakthrough curves is mainly controlled by the reduction of Cu 2O. Thus a correction factor has been added in the kinetic law of Cu 2O reduction ii correcting the effective diffusion coefficient of CO (DDDD CCCC ) according to equation Error! No se encuentra el origen de la referencia..