Energy & Fuels 2003, 17, 643-651
643
Reactivity of Some Metal Oxides Supported on Alumina with Alternating Methane and OxygensApplication for Chemical-Looping Combustion Tobias Mattisson,*,† Anders Ja¨rdna¨s,‡ and Anders Lyngfelt† Department of Energy Conversion and Department of Environmental Inorganic Chemistry, Chalmers University of Technology, S-412 96 Go¨ teborg, Sweden Received July 3, 2002
Chemical-looping combustion (CLC) is a combustion technology with inherent separation of the greenhouse gas CO2. The technique involves the use of a metal oxide as an oxygen carrier, which transfers oxygen from the combustion air to the fuel. Two reactors are used in the process: (i) a fuel reactor where the metal oxide is reduced by reaction with the fuel, and (ii) an air reactor where the reduced metal oxide from the fuel reactor is oxidized with air. The possibility of using oxides of Cu, Co, Mn, and Ni as oxygen carriers was investigated. Particles were prepared by deposition of the metal oxides on γ-Al2O3 particles by so-called dry impregnation. The reactivity of the oxygen carrier particles was evaluated in a thermogravimetric analyzer (TGA), where the alternating atmosphere which an oxygen carrier encounters in a CLC system was simulated by exposing the sample to alternating reducing (10% CH4, 5% CO2, 10% H2O) and oxidizing (10% O2) conditions at temperatures between 750 and 950 °C. The particles of Ni and Cu showed high reactivity at all temperatures and cycles, with reduction rates of up to 100%/min for CuO and 45%/min for NiO and oxidation rates of up to 25%/min for the oxidation of both reduced metals. Oxides of Mn and Co showed a limited extent of reaction, which was explained by the chemical reaction of the metal oxide with the alumina, with the formation of highly irreversible phases which do not react with methane and oxygen. Thus, Mn and Co are not suitable as oxygen carriers for CLC when supported on Al2O3. From the reactivity data of the nickel and copper oxygen carriers, it was estimated that 460-620 kg/MW oxygen carrier would be needed in the reactors. Similarly, from the oxygen transfer capacity of the particles, the solids circulation rate between the fuel and air reactor would need to be between 1 and 8 kg MW-1 s-1.
1. Introduction There is a need for a drastic reduction in the emissions of greenhouse gases to the atmosphere. A substantial amount of the carbon dioxide released to the atmosphere today comes from the combustion of fossil fuels, and thus it is from this source that the major reductions are necessary. In the long term, the reduction will be based on the combination of switching to renewable forms of energy as well as savings in energy consumption. However, to achieve major reductions it is likely that carbon dioxide will need to be captured from the combustion process and stored in depleted oil and gas fields, aquifers, and possibly the deep sea. Chemical-looping combustion (CLC) is a combustion technology with inherent separation of the greenhouse gas CO2.1-4 It is composed of two reactors, an air and a * Corresponding author. Tel: +46-31-7721425. Fax: +46-317723592. E-mail:
[email protected]. † Department of Energy Conversion. ‡ Department of Environmental Inorganic Chemistry. (1) Anheden, M.; Na¨sholm, A.-S.; Svedberg, G. Chemical-looping combustion-efficient conversion of chemical energy in fuels into work. In 30th Intersociety Energy Conversion Engineering Conference, Orlando, FL; ASME: New York, 1995; pp 75-81. (2) Richter, H.; Knocke, K. ACS Symp. Ser. 1983, 235, 71-86. (3) Ishida, M.; Yamamoto, M.; Ohba, T. Energy Convers. Manage. 2002, 43, 1469-1478.
fuel reactor, as shown in Figure 1. The fuel is introduced to the fuel reactor in a gaseous form where it reacts with an oxygen carrier according to
(2n + m)MyOx + CnH2m f (2n + m)MyOx-1 + mH2O + nCO2 (1) At full conversion of the fuel gas, the exit gas stream from the fuel reactor contains only CO2 and H2O, which means that pure CO2 can be obtained when H2O is condensed. The reduced metal oxide, MyOx-1, is then circulated to the air reactor where it is oxidized according to
MyOx-1 + 1/2O2 f MyOx
(2)
The flue gas from the air reactor will contain N2 and any unreacted O2. The extent of the reactions above may vary depending on the metal oxide and the reaction conditions. The total amount of heat evolved from reactions 1 plus 2 is the same as for normal combustion where the oxygen is in direct contact with the fuel. However, the advantage with this system compared to normal combustion is that the CO2 and H2O are inher(4) Lyngfelt, A.; Leckner, B.; Mattisson, T. Chem. Eng. Sci. 2001, 56, 3101-3113.
10.1021/ef020151i CCC: $25.00 © 2003 American Chemical Society Published on Web 05/02/2003
644
Energy & Fuels, Vol. 17, No. 3, 2003
Mattisson et al.
Figure 1. Chemical-looping combustion.
Figure 3. Scanning electron microscope image of unreacted particles of (a) CuO/Al2O3 and (b) NiO/Al2O3.
reactivity under alternating conditions of methane and oxygen was measured in a thermogravimetric analyzer. 2. Experimental Section Figure 2. Chemical-looping combustion designed using interconnected fluidized beds.4
ently separated from the rest of the flue gases, and no energy is expended for this separation. Limited work has been done on the design of a CLC unit. Lyngfelt et al.4 proposed a design based on two interconnected fluidized beds as seen in Figure 2. Here, the air reactor (1) is a high-velocity fluidized bed where the oxygen carrier particles are transported together with the air stream to the top of the air reactor, and then transferred to the fuel reactor (3) via a cyclone (2). The fuel reactor is a bubbling fluidized bed reactor, with the reduced oxygen carriers transported back to the air reactor by an overflow pipe. Clearly, the oxygen carrier should have a high tendency to react with the fuel gas in the fuel reactor, and similarly, the reduced oxygen carrier should have a high tendency to react with air in the air reactor. In an earlier paper, Mattisson and Lyngfelt found that some metal oxides of the transition state metals Fe, Cu, Co, Mn, and Ni were feasible candidates to be used as oxygen carriers in a CLC system based on interconnected fluidized beds.5 In the present work these metal oxides, with the exception of iron oxide, were impregnated on alumina carrier particles and the (5) Mattisson, T.; Lyngfelt, A. Capture of CO2 using chemical-looping combustion. In Scandinavian-Nordic Section of Combustion Institute, Gothenburg, 2001; pp 163-168.
2.1. Preparation of Alumina-Supported Metal Oxides. Because of the extensive work previously performed with iron as an oxygen carrier,5-7 it was not included in the present study. Oxygen carriers of Co, Cu, Mn, and Ni on alumina were prepared by dry impregnation, where γ-Al2O3 particles of diameter 0.1-0.5 mm (NWA-155, Sasol Germany Gmbh) were exposed to an aqueous solution of the metal nitrate. During each impregnation a volume of the metal nitrate solution corresponding to the total pore volume of the sample was added. The oxygen carrier was then calcined at 550 °C for 5 h with the formation of the metal oxide, and with subsequent release of nitrogen oxides. This procedure was repeated until the loading of metal oxide on the alumina support reached a theoretical loading of 28-35 wt %. A scanning electron microscope image of a fresh particle of CuO/Al2O3 and NiO/ Al2O3 is shown in Figure 3. Table 1 shows the metal oxide content of the particles, in addition to the active area and pore volume of the fresh oxygen carriers. 2.2. Reactivity Investigation. All of the experiments were carried out in a thermogravimetric analyzer (TGA) (Setaram TAG 24S16). About 100 mg of the oxygen carrier was heated in a platina cup in a quartz reactor to the reaction temperature in a nitrogen atmosphere. The sample was then exposed to a reducing gas of 10% CH4, 10% H2O, 5% CO2, and 75% N2 for 10-20 min, after which the reactor was flushed with nitrogen for 5 min. H2O and CO2 are added to the reducing gas in order (6) Mattisson, T.; Lyngfelt, A.; Cho, P. Fuel 2001, 80, 1953-1962. (7) Cho, P.; Mattisson, T.; Lyngfelt, A. Reactivity of iron oxide with methane in a laboratory fluidized bed-application of chemical-looping combustion. In 7th International Conference on Circulating Fluidized Beds, Niagara Falls, Canada, 2002.
Reactivity of Some Metal Oxides Supported on Alumina Table 1. Characterization of Fresh Oxygen Carrier Particles metal oxidea
theoretical loadingb (wt %)
actual loadingc (wt %)
pore volumed (mL/g)
BETd (m2/g)
CuO Co3O4 Mn2O3 NiO
33.0 34.5 28.4 33.0
33.8 34.8 29.4 33.0
0.29 0.25 0.26 0.27
87.8 80.2 80.8 91.7
a
Indicates the stable metal oxide after calcination at 550 °C. Based on the amount of metal nitrate solution added to the sample. c Determined using ICP-OES. d After final calcination at 550 °C. Determined using Micromeritics ASAP 2010 (nitrogen adsorption).
Energy & Fuels, Vol. 17, No. 3, 2003 645 of 10% CH4 corresponds to an inlet concentration of 100% and an outlet concentration of 0.005%; that is an almost complete conversion of the methane to carbon dioxide and water. Similarly, a 10% average O2 concentration corresponds to a concentration of 4% in the outlet from the air reactor, assuming an inlet oxygen concentration of 21%. The degree of oxidation, or conversion, was calculated as
X)
m - mred mox - mred
(4)
b
Figure 4. Mass as a function of time for an experiment conducted with NiO/Al2O3 at 850 °C. to better simulate the atmosphere in a fuel reactor. The presence of steam and carbon dioxide in the reactant gas should also prevent the formation of carbon on the oxygen carrier.8 After the inert period, the reduced sample was oxidized with 10% O2 in N2. The sample was exposed periodically to reducing and oxidizing conditions for between 3 and 7 cycles. The total flow of gas was 80 mLn/min for all periods and experiments. Figure 4 shows the mass as a function of time for an experiment conducted with NiO at 850 °C for seven reduction and oxidation cycles. The samples were cooled in nitrogen following the experiments. The reacted oxygen carriers were analyzed for crushing strength, phase composition, and surface appearance. The reactant gas concentrations during the reducing (10% CH4) and oxidizing periods (10% O2) were chosen to simulate an average concentration which the particles would be exposed to in a real CLC system. Assuming that the reaction of methane and metal oxide is first order with respect to methane and, similarly, that the reaction of reduced metal oxide is first order with respect to oxygen, then a suitable measure of the reactant gas concentration to which the particles are exposed in the reactors is given by the logarithmic mean volume fraction of methane or oxygen, xa,
where m is the actual mass of sample, mox is the mass of the sample when fully oxidized, and mred the mass of the sample in the reduced form. The reduced form is either the metal or a metal oxide of lower oxygen content. Thus, the difference between mox and mred in eq 4 is the amount of active oxygen in the oxygen carrier, i.e., the maximum amount of oxygen which can be removed through reaction with methane. In this work, the difference mox - mred is calculated on the basis of the metal oxides: (i) CoO - Co, (ii) CuO - Cu, (iii) Mn3O4 MnO, and (iv) NiO - Ni. Even though CuO is the stable species in air at the temperatures studied, some decomposition to Cu2O was noted during the inert periods, i.e., with 100% N2 at 850 °C and 950 °C. 2.3. Characterization of Oxygen Carriers. To determine the chemical transformations which occurred in the samples during reaction, all of the reacted samples were characterized using X-ray powder diffraction (Siemens D5000 Powder Diffractometer utilizing Cu KR radiation). Also the shape and morphology of the fresh and reacted oxygen carrier particles were studied using a light microscope as well as an analytical scanning electron microscope (Electroscan 2020 equipped with a Link eX1 EDX system). Because of the need for the oxygen carrier to be resistant toward fragmentation and attrition, the force needed to fracture the particles was measured using a Shimpo FGN-5 crushing strength apparatus.
3. Results
where xin and xout are the inlet and outlet volume fractions of the reacting gas, CH4, or O2. Thus, an average concentration
Figure 5a-d shows the conversion, X, as a function of time for the second reduction and oxidation period for the four oxygen carriers at the temperatures 750 °C, 850 °C, and 950 °C. Because the mass was not always stable prior to the first reduction period for some of the experiments, the second cycle will be used as the reference cycle in the paper. The corresponding rates of reaction, dX/dt, are shown as a function of the conversion in Figure 6a-d. As is evident, the extent of reaction as well as the rate of reaction is a function of the metal oxide studied as well as the temperature. 3.1. Nickel Oxide. For NiO the reduction rate was as high as 45%/min for the trial at 950 °C and the extent of reaction was high, see Figure 6a. The rates were somewhat lower for the experiments at the two lower temperatures, up to 30%/min at 850 °C and 18%/min at 750 °C. The oxidation rates were rather similar for all three temperatures, reaching a maximum of about 25%/min. Minor amounts of the nickel aluminate NiAl2O4 were found in the reacted samples at 850 °C and 950 °C, see Table 2. Nickel aluminate is thermodynamically stable during oxidizing conditions, and the reaction between γ-Al2O3 and NiO to NiAl2O4 has been confirmed in other work in 20% O2/N2 at 1000 °C.9 Richardson and Twigg10 found interaction between impregnated nickel
(8) Mattisson, T.; Lyngfelt, A. Applications of chemical-looping combustion with capture of CO2. In Second Nordic Minisymposium on Carbon Dioxide Capture and Storage, Gothenburg, 2001; pp 4651.
(9) Bolt, P.; Habraken, F.; Geus, J. J. Solid State Chem. 1998, 135, 59-69. (10) Richardson, J.; Twigg, M. Appl. Catal. A: General 1998, 167, 57-64.
xa )
xin - xout xin ln xout
( )
(3)
646
Energy & Fuels, Vol. 17, No. 3, 2003
Mattisson et al.
Figure 5. The conversion as a function of time for the second reduction (solid line) and oxidation (dashed line) period for the four oxygen carriers; 750 °C ], 850 °C +, 950 °C 3. Note that the range of the x- and y-axes differ.
oxide and alumina support already at a calcination temperature of 650 °C. 3.2. Copper Oxide. The reduction rate is fast at all temperatures for CuO, approaching 100%/min at 950 °C. Even at 750 °C the reduction rate is high, with a maximum of 70%/min, see Figure 6b. After the fast initial reduction rate, the conversion actually increases somewhat, Figure 5b, which is believed to be due to some minor carbon formation on the particles. Minor amounts of carbon formation were verified in other tests with CuO/Al2O3 in 50% CH4 and 50% H2O, in contrast to tests with NiO/Al2O3 and Mn3O4/Al2O3, which were exposed to the same gas without carbon formation. This suggests that carbon is more readily formed on CuO, and supports the assumption that carbon formation causes the weight increase. However, carbon is not a thermodynamically stable end product under these conditions, but could possibly form if the carbon formation from CH4 is faster that the oxidation of carbon by H2O and CO2. The change in conversion after the reduction reaction at 950 °C is lower than for the two lower temperatures, see Figure 5b. This is explained by the decomposition
of CuO to Cu2O during the inert period following the oxidation period at 950 °C, which means that there is less oxygen available for reaction with methane and consequently the change in conversion during the reduction will be lower compared to the experiments at the two lower temperatures. A small mass loss was even seen during the inert period at 850 °C. The formation of Cu2O was confirmed through X-ray powder diffraction of the reacted sample, see Table 2. Note that the reduced copper sample is oxidized to CuO during the oxidation, which explains the fact that the degree of conversion is above 1 in Figure 5b. Limited formation of copper aluminate was found in all of the reacted samples. Bolt el al.9 found a high reactivity between CuO and γ-Al2O3, and diffusion of copper ions into the alumina was noted at temperatures as low as 500 °C. 3.3. Manganese Oxide. During reduction of manganese oxide it was not possible to reach an X lower than 0.7 independent of temperature. This was due to the formation of the manganese aluminates, MnAl2O4 and Mn2AlO4, see Table 2. The change in conversion was greatest for the lower temperatures, which is explained by the greater degree of manganese-alumina reaction
Reactivity of Some Metal Oxides Supported on Alumina
Energy & Fuels, Vol. 17, No. 3, 2003 647
Figure 6. The reduction rate (solid line) and oxidation rate (dashed line) as a function of the degree of conversion; 750 °C ], 850 °C +, 950 °C 3. Note that the ranges of the x- and y-axes are different for all figures. Table 2. Phases Containing the Metals Co, Cu, Mn, and Ni Identified with X-ray Diffraction oxygen carrier
fresh sample
750 °Ca
850 °Cb
950 °Ca
CoO/Al2O3 CuO/Al2O3 Mn3O4/Al2O3 NiO/Al2O3
Co3O4 CuO Mn2O3, MnO2c NiO
Co3O4, CoAl2O4 CuO, CuAl2O4c Mn3O4, MnAl2O4/Mn2AlO4 NiO
Co, CoAl2O4 Cu, Cu2O, CuAlO2c, CuAl2O4c MnO, Mn3O4c, MnAl2O4/Mn2AlO4 Ni, NiAl2O4c
CoAl2O4 Cu2O, CuAlO2c, CuAl2O4c MnAl2O4 NiO, NiAl2O4c
a
Following an oxidation period. b Following a reduction period. c Minor phase.
at higher temperatures. The formation of manganese aluminate at temperatures below 1000 °C will most likely proceed through11
MnO + Al2O3 f MnAl2O4
(5)
and should thus occur mainly during the reducing periods where MnO is formed. Though the reoxidation of the manganese aluminate is thermodynamically feasible at temperatures below 820 °C, the reaction has been found to be very slow.11 At higher temperatures, the oxidation is thermodynamically hindered. After the oxidizing and reducing gases were replaced by nitrogen there was a small mass loss at all temper(11) Stobbe, E. Catalytic routes for the conversion of methane to synthesis gas. Ph.D. Thesis, University of Utrecht: Utrecht, 1999.
atures, which explains the deviation in the degree of conversion between the end of the reducing period and the beginning of the oxidation period, see Figure 5c. The mass loss is gradual and rather small, with a maximum of 0.25 mg. The reason for this mass loss is not known, but it has no significant effect on the reaction data obtained. 3.4. Cobalt Oxide. The samples are not reduced completely to Co at any temperature, see Figure 5d. The extent of the reaction is highest at 850 °C, reaching a conversion of about 52% after 500 s of reducing atmosphere. The samples were oxidized back to approximately the initial conversion after the reduction. The rates of reaction are dependent on the degree of conversion, with maximum rates at 850 °C where the rates are up to 20%/min for both reduction and oxidation, see
648
Energy & Fuels, Vol. 17, No. 3, 2003
Mattisson et al. Table 3. Oxygen Ratios of Different Oxygen Carriers MyOx/MyOx-1
a
Figure 7. The degree of conversion, X, as a function of time for an experiment conducted with CoO/Al2O3 at 850 °C.
Figure 6d. The reason for the poor reaction at the higher temperature is believed to be associated with the interaction of the alumina with the cobalt oxide and the formation of CoAl2O4, which was found to be present in the reacted samples from X-ray powder diffraction, see Table 2. This spinel is stable under oxidizing conditions and the temperatures investigated. Though the reduction of CoAl2O4 to Co and Al2O3 is thermodynamically possible, the reaction rate is slow compared to the reduction of CoO. 3.5. Effect of the Number of Cycles. For the oxygen carriers of Cu, Ni, and Mn the rate of reduction and oxidation as a function of the number of cycles was relatively constant for six cycles. However, for cobalt oxide there was a clear effect of the number of cycles. Figure 7 shows the conversion as a function of time for the experiment conducted at 850 °C for six cycles with cobalt oxide. The extent of the reaction decreases as the number of cycles increases. It is believed that this is a result of the progressive reaction between alumina and cobalt oxide. A similar behavior is found at 750 and 950 °C. 3.6. Crushing Strength and Surface Morphology. Particles in the size range 0.355-0.5 mm were chosen to test the crushing strength. The crushing strength of fresh oxygen carrier particles was similar for all metal oxides and in the range 3.7-5.2 N. The original alumina particles had a crushing strength of 3.8 N. The strength of the metal oxide particles was observed to increase marginally after reaction, to between 4.5 and 6.5 N for particles reacted at 850 °C. It should be noted that all the measurements above were based on 6 measurements, and the standard deviation was between 0.5 and 1 N. To obtain information concerning physical changes of the particles during reaction, the surfaces of both the fresh and reacted samples were analyzed in a scanning electron microscope, see Figure 8. The surface of the fresh particles was granular for all oxides. However, the fresh cobalt and nickel particles had smaller grains, compared to manganese and copper. The latter had a considerably rougher texture than other particles, see Figure 3. For the nickel oxide particles, limited physical changes were observed when comparing the surface of fresh particles and particles reacted at 950 °C. The
Ro
Fe2O3/Fe3O4 Mn3O4/MnO CuO/Cu NiO/Ni CoO/Co
0.03 0.07 0.20 0.21 0.21
This worka Mn3O4/MnO with Al2O3 CuO/Cu with Al2O3 NiO/Ni with Al2O3 CoO/Co with Al2O3
0.02 0.08 0.08 0.07
Based on analysis of wt % of components.
reacted manganese oxide particles show some annealing of the individual grains, which may be due to the solidsolid reaction between the alumina and manganese oxide. Similarly, the surface of the reacted cobalt oxide particles seems to contain larger grains compared to the fresh sample. For the reacted copper sample, the surface texture was considerably different, with larger and smoother grains, suggesting the formation of a melt phase during the reaction. The melting point for Cu is 1089 °C, and though the reaction temperature is lower, the exothermic nature of both the reduction and oxidation could result in an increase in the temperature of the particles, thus approaching the melting point. The copper sample that reacted at 750 °C did not show any tendency to melt, and the surface remained similar to that of the original sample following reaction. 4. Application of Reactivity to Design Criteria The type of oxygen carrier and its corresponding reactivity has important implications for a CLC system. Below follows a discussion of some important aspects of the design of a chemical-looping combustor, based on two interconnected fluidized beds, in relation to the reactivity of the oxygen carriers in this work. 4.1. Oxygen Ratio. The mass fraction of oxygen that can be used in the oxygen transfer is called the oxygen ratio and is defined as
Ro)(mox - mred)/mox
(6)
where Ro shows the maximum amount of oxygen that can be transferred between the air and fuel reactor in CLC for a given mass flow of recirculating particles. Table 3 shows the oxygen ratio for some metal oxide pairs of Fe, Mn, Cu, Co, and Ni. The oxygen ratio is also shown for the particles used in the present study. To facilitate a comparison between the various metals a mass-conversion based on the mass of the oxygen carrier was defined as
ω)
m ) 1 + Ro(X - 1) mox
(7)
Figure 9a shows the rate of mass-conversion, dω/dt, as a function of ω for the second reduction period at 850 °C and 950 °C. Because of the low extent of reaction for the Co and Mn samples at the higher temperature, these are not included in the figure. The reduction rate is considerably faster for copper oxide than the other three oxides. Figure 9b shows dω/dt as a function of ω
Reactivity of Some Metal Oxides Supported on Alumina
Energy & Fuels, Vol. 17, No. 3, 2003 649
Figure 8. SEM images of the surface of unreacted and reacted oxygen carrier particles.
for the second oxidation period. The oxidation rates are higher for CoO and NiO than for CuO in the span ω ) 0.94-1.0. 4.2. Amount of Oxygen Carriers in the Reactors. The amount of bed material necessary in each of the two reactors is inversely proportional to the reactivity
of the oxygen carrier and can be calculated from
mbed )
ωm ˘o dω/dt
(8)
where m ˘ o is the stoichiometric mass flow of oxygen
650
Energy & Fuels, Vol. 17, No. 3, 2003
Mattisson et al.
Figure 10. The temperature loss in the fuel reactor as a function of ∆ω assuming an inlet temperature of oxygen carrier of 950 °C and CH4 of 400 °C: CuO ], CoO +, Mn3O4 3, NiO ×.
4.3. Recirculation Rate of Oxygen Carriers. The rate of circulation of oxygen carriers between the air and fuel reactor is inversely proportional to the difference in the average degree of conversion of the oxygen carrier in the two reactors. The rate of solids flow, m ˘ sol, can be calculated from
m ˘ sol )
Figure 9. The mass-based rate dω/dt, as a function of ω for (a) the reduction and (b) oxidation at 850 °C (solid) and 950 °C (dashed): CuO ], CoO +, Mn3O4 3, NiO ×. Because of the low extent of reaction, the data for Mn3O4 and CoO are not included for 950 °C. Table 4. Amounts of Oxygen Carrier Necessary in Reactors
copper nickel
dω/dtreda (%/min)
dω/dtoxa (%/min)
mfr (kg/MW)
mar (kg/MW)
6.5 2.0
1.2 1.2
70 230
390 390
*Rates are based on measured rates at ω ) 0.98 in Figure 9a,b (950 °C).
needed for complete conversion of the methane to carbon dioxide and water. The relationship in eq 8 assumes that the average methane and oxygen concentrations in the fuel and air reactors are the same as those in the TGA experiments, i.e., 10% CH4 and 10% O2. From Figure 9 and eq 8 it is clear that for a certain power, considerably less copper oxide would be needed in the fuel reactor compared to the other three metals. However, because of the slower oxidation rate, compared to the reduction rate, of reduced copper oxide, more material would be needed in the air reactor for this oxygen carrier when compared to the fuel reactor. Table 4 shows the mass of the oxygen carriers CuO/Al2O3 and NiO/Al2O3 needed in the reactors in a CLC system, based on rates, dω/dt, at 950 °C taken at ω ) 0.98. The total mass varies between 460 and 620 kg/MW.
ωair m ˘o ∆ω
(9)
where ωair is the conversion in the air reactor and m ˘ o is the rate of oxygen transfer between the two reactors. Thus, for a ∆ω of between 0.01 and 0.06 the needed recirculation rate would be between 1 and 8 kg MW-1 s-1 . 4.4. Heat Balance. For most metal oxides the reaction in the fuel reactor, i.e., reaction 1, is endothermic, and thus there will be a temperature drop in this reactor. It is important that the temperature in the reactor is sufficient for a rapid reduction of the metal oxide, and thus the temperature drop should be limited. The temperature drop in the reactor is dependent on the rate of circulation of bed material, which is connected to the mass conversion difference between the oxygen carrier in the two reactors, see eq 9. The temperature change in the fuel reactor was calculated as a function of the change in mass-based conversion, ∆ω, for the oxygen carriers tested, see Figure 10. For CuO the temperature in the reactor would increase due to an exothermic reaction between CuO and CH4. For the other three oxygen carriers the temperature would decrease with a maximum drop of 200 °C. Thus, to avoid a large temperature drop, a high degree of circulation of the oxygen carrier between the fuel and air reactor is needed, which in practice means a low ∆ω. 5. Discussion Nickel and copper oxide supported on alumina show a relatively rapid reactivity with methane. Further, the oxidation rate of the reduced metal oxide is also rather rapid. The formation of irreversible alumina spinels is
Reactivity of Some Metal Oxides Supported on Alumina
limited in the temperature range tested, and therefore most of the oxygen is active for reaction with methane. For CuO/Al2O3 particles reacted at 950 °C there were major changes on the surface of the particle, possibly due to the formation of a melt phase during reaction. Thus the use of such particles in fluidized beds at high temperatures may be questionable. For both the manganese oxide and cobalt oxide samples, the extent of reaction was found to be limited by a high degree of spinel formation at all temperatures. But both of these oxides do react rapidly with methane and oxygen at some temperatures and conversion levels, see Figures 5c,d and 6c,d. Thus, the use of support materials which are inert toward reaction with manganese and cobalt is likely to give higher extents of reaction. The total amount of bed material needed in the two reactors is estimated to 460-620 kg/MW for the metal oxide particles, see Table 4. This can be compared to a previous estimate for use of iron ore with a total bed mass of 650 kg/MW.4 Reactivity data for an oxygen carrier composed of 60% Fe2O3 and 40% Al2O3 showed considerably higher reactivities with methane compared to iron ore,5,7 suggesting that the total amount of bed material could be reduced to 400 kg/MW or less. The data for iron oxide may not be directly comparable, as they are obtained with fixed and fluidized bed reactor experiments, where the average gas concentrations to which the oxygen carrier is exposed may differ. One major difference between iron oxide and the investigated oxides is the much faster oxidation rate of reduced iron oxide, up to 90%/min.5,6 The advantage with a rapid oxidation reaction is that the reduced pressure drop will save power for the fans supplying the large air flows. The pressure drop of the small fuel flow is less important. The advantage with a rapid reduction reaction, on the other hand, is that very low outlet concentrations of CH4 more easily can be reached, which may reduce the power needed for compression of CO2. One way of reducing the needed bed mass is to increase the oxygen ratio, Ro, by increasing the fraction of metal oxide relative to support material. The fraction of inert
Energy & Fuels, Vol. 17, No. 3, 2003 651
material is rather high in all of the samples, about 6570%. If this could be reduced without affecting the reaction rate measured as dX/dt, this would increase the mass-based reaction rate, dω/dt, thus reducing the amount of bed material needed. 6. Conclusions Chemical-looping combustion is a promising combustion technique for the separation of CO2 with small losses in energy. Particles of NiO, CuO, CoO, and Mn3O4 were prepared on an alumina support by impregnation. The rates of reduction and oxidation with methane and air were investigated in a cyclic manner using a TGA at 750-950 °C. NiO and CuO were found to react to the highest extent with reduction rates up to 100%/min for CuO and 45%/min for NiO. The oxidation rates of the reduced metal oxides were up to 25%/min for both these oxides. The manganese and cobalt oxides were found to react to a much lower extent, which was explained by the chemical reaction between the alumina and metal oxide to nonreactive compounds, which meant that considerably less oxygen was present for the reaction with methane. This means that the use of CuO and NiO on an alumina support should be possible in a CLC system based on interconnected fluidized beds operating at temperatures below 950 °C. The total amount of bed material necessary in the two reactors was estimated to be in the range 460-620 kg/MW, and recirculation rates of oxygen carrier between the two reactors would need to be between 1 and 8 kg MW-1 s-1. Acknowledgment. The authors thank Preems Miljo¨stiftelse for their financial support. We also express our gratitude to Sasol Germany GmbH for supplying us with the alumina particles. A special thanks to Dr. Britt-Marie Steenari for her help with the X-ray diffraction. EF020151I