Development of Iron Oxide Carriers for Chemical Looping Combustion

Apr 30, 2010 - The results were as follows: (1) For reduction to the FeO phase, unsupported Fe2O3 gave stable conversions over 40 cycles and no Al2O3 ...
0 downloads 0 Views 2MB Size
Ind. Eng. Chem. Res. 2010, 49, 5383–5391

5383

Development of Iron Oxide Carriers for Chemical Looping Combustion Using Sol-Gel A. M. Kierzkowska,†,‡ C. D. Bohn,*,† S. A. Scott,‡ J. P. Cleeton,‡ J. S. Dennis,† and C. R. Mu¨ller¶ Department of Chemical Engineering and Biotechnology, UniVersity of Cambridge, Pembroke Street, Cambridge, CB2 3RA, U.K., Department of Engineering, UniVersity of Cambridge, Trumpington Street, Cambridge, CB2 1PZ, U.K., and Institute of Energy Technology, Department of Mechanical and Process Engineering, ETH Zu¨rich, Leonhardstraβe 27, 8092 Zu¨rich, Switzerland

Composite particles with different mass ratios of Fe2O3 and Al2O3 were prepared using a sol-gel method and were examined for use in chemical looping combustion through repeated reduction and oxidation cycles in a packed bed reactor at 850 °C. Unlike traditional chemical looping combustion which reduces an oxygen carrier in methane and oxidizes it in air, the reducing gas here was a mixture of CO and N2. Oxidation was performed in a mixture of steam and N2 to produce H2, followed by oxidation in air in some cases. The results were as follows: (1) For reduction to the FeO phase, unsupported Fe2O3 gave stable conversions over 40 cycles and no Al2O3 support was needed. (2) For reduction to the Fe phase over 10 cycles, 10 wt % Al2O3 was sufficient to give stable conversions above 0.9. Over 30-40 cycles, however, the conversion for particles with 10-20 wt % Al2O3 dropped below 0.35. (3) For reduction to the Fe phase over 40 cycles, 40 wt % Al2O3 was required and gave stable conversions near 0.75. The formation of FeO · Al2O3 was confirmed using X-ray diffraction. Steam in N2, followed by air, is the recommended sequence for oxidizing the composite carriers, since temperature excursions and agglomeration of particles could be avoided and higher conversions could be achieved. Introduction For the combustion of carbonaceous fuels to be a viable option for the future generation of energy, capture of the resulting CO2 for storage in geological formations is necessary.1 Typical flue gas streams have too high a ballast of N2 to be efficiently compressed to the pressures required for sequestration. Techniques for the generation of a pure stream of CO2 such as amine solvent scrubbing, oxyfuel combustion and chemical-looping combustion have therefore attracted considerable attention. In chemical looping combustion, a metal oxide provides the oxygen for combustion and permits the inherent separation of CO2 without a significant energy penalty. Reviews on chemical looping combustion2,3 and details on oxygen carriers being investigated can be found in refs 3 and 4. Because of its ability to produce hydrogen, physical strength, low cost, and minimal environmental impact, iron oxide is a good candidate for the oxygen carrier in chemical looping combustion. The species of iron and its oxides involved are Fe2O3, Fe3O4, FeO, and Fe. Here for simplicity, the Fe2+ oxide is written as FeO rather than Fe(1-y)O. It is noted, however, that the stoichiometric ratio of Fe/O ) 1 is never achieved, since typically Fe(1-y)O exists, with 0.05 < y < 0.17.5 In the proposed process, coal or biomass is used as the fuel and is first gasified with steam to generate synthesis gas with a typical composition of 40 vol % CO, 40 vol % H2, 10 vol % CO2, and 10 vol % H2O. The reduction of the metal oxide to a lower oxidation state with this synthesis gas * To whom correspondence should be addressed. E-mail: [email protected]. Phone: +44 (0)1223 762962. Fax: +44 (0)1223 334796. † Department of Chemical Engineering and Biotechnology, University of Cambridge. ‡ Department of Engineering, University of Cambridge. ¶ Institute of Energy Technology.

Fe2O3 + CO a 2FeO + CO2

o o ∆H850 C ) -0.3 kJ/mol (1)

Fe2O3 + 3CO a 2Fe + 3CO2

o o ∆H850 C ) -33.3 kJ/mol (2)

generates mostly pure CO2, for reduction with CO as shown by reactions 1 and 2, and H2O for similar reduction in H2. Subsequent condensation of the H2O yields the desired product, a pure stream of CO2 suitable for sequestration. Oxidation of the reduced carrier in steam generates H2: 3FeO + H2O(g) a Fe3O4+H2

o o ∆H850 C ) -47.1 kJ/mol (3)

o o 3Fe + 4H2O(g) a Fe3O4 + 4H2 ∆H850 C ) -105.3 kJ/mol

(4)

Since, it is not thermodynamically possible to produce H2 in measurable quantities by oxidizing Fe3O4 to Fe2O3 with steam,6 additional oxidation of the carrier with air can also be performed: 2Fe3O4 + 1/2O2 a 3Fe2O3

o o ∆H850 C ) -237.2 kJ/mol (5)

Combining the above reduction and oxidation reactions gives a process which is exothermic overall and which could generate separate streams of CO2 and H2. If the CO2 is sequestered, this scheme results in a net zero release of CO2 for coal. For biomass, a net removal of CO2 from the atmosphere is possible, since the carbon in the plant mass which originated from the atmosphere is sequestered as CO2 in, e.g., saline aquifers. For industrial implementation, the three reaction stages could be carried out in three interconnected fluidized beds with solids cycling or in a single unit by dynamic switching of the inlet gas.

10.1021/ie100046f  2010 American Chemical Society Published on Web 04/30/2010

5384

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010

Table 1. Fe2O3 and Al2O3 Composite Carriers Used in Chemical Looping Combustiona calcining ref Li et al., 20099 Ishida et al., 200510 Jin et al., 199911 Ishida et al., 199812 Mattisson et al., 200413 Abad et al., 200717 Mattisson et al., 200413 Adanez et al., 200414

preparation

composition (wt % Fe2O3)

sol-gel mech mix Fe2O3, Al2O3 dissolution Fe(NO3)3 · 9H2O, Al(NO3)3 · 6H2O

60% 25%

mech mix and freeze granulation

60%

Fe(NO3)3 · 9H2O impregnated γ-Al2O3 mech mix and 10 wt % graphite for pores

22% 80% 60% 40%

60%

temp

time

900 °C 800 °C 1370 °C 1300 °C 1300 °C 1100 °C 1300 °C 1100 °C 1100/1300 °C 1100/1300 °C 1100/1300 °C

12 h 10 h 10 h 6h 6h 6h 6h 6h 6h 6h 6h

crushing stress (MPa) 10 2 54 44.3 44.3 18 167 6/27 5/25 5/29

a Preparation methods include mechanical mixing of powders of Fe2O3 and Al2O3, freeze granulation of mechanically mixed oxide powders, mixing aqueous solutions of iron and aluminum nitrates, and wet-impregnation of γ-Al2O3 support. The final composition of the oxygen carrier is listed as the weight percent of Fe2O3 after calcining at the specified conditions; the remaining mass fraction is Al2O3. The crushing stress is given for the fresh oxygen carrier. Details on calculating the crushing stress are given in the discussion.

This method for the generation of H2, the steam-iron process, was first patented by Messerschmitt in 1910.7 It has been shown, however, that unsupported Fe2O3 does not permit repeated redox cycling when it is reduced fully to Fe, with poor yields of H2 observed after as few as three cycles.8,9 In contrast, if the reduction can be limited to FeO, stable yields of H2 can be obtained.8 Since oxidation of Fe to Fe3O4 with steam produces four times as much H2 as oxidation of FeO to Fe3O4 per unit mass of Fe2O3 initially present, c.f. reactions 3 and 4, it is desirable to find an oxygen carrier which can be repeatedly reduced to Fe rather than FeO. The most common support for Fe2O3 for chemical looping applications is Al2O3. Table 1 gives a summary of some composite particles of Fe2O3 and Al2O3 used and the corresponding reference. While methods of preparation range from mechanical mixing oxide powders10 to drying aqueous solutions of Fe and Al nitrate precursors,11,12 a single weight ratio of 60 wt % Fe2O3 and 40 wt % Al2O3 is typically used. Other supports for Fe2O3 include ZrO2,13-15 TiO2,11-14 MgO,11 MgAl2O4,13 and yttria-stabilized zirconia (YSZ).12 Notably, SiO2 is unsuitable, owing to the formation of unreactive silicates with low melting points.16 Conventional methods for preparing mixed metal oxides do not always produce materials with both (i) a high internal surface area and (ii) homogeneously mixed components on the molecular scale. For example, coprecipitation does not ensure homogeneity because hydroxides of different metallic cations generally do not precipitate at the same pH. Mechanically mixed oxides usually do not have high surface areas and suffer from inhomogeneities of the mixed materials.18 The sol-gel method has received much attention in catalysis19,20 because it allows the homogeneous mixing of components almost at the molecular level as well as the control of pore structure and surface area using a low temperature synthesis. While limited research on Ni-based21,22 oxygen carriers prepared using a sol-gel technique exists, only recently have similar Fe-based9 oxygen carriers been investigated for chemical looping combustion. The objective of this study is to compare the performance of iron oxide carriers made either (i) by mechanical mixing or (ii) using sol-gel methods, especially with respect to the ability to be reduced completely to Fe. This work is distinct from previous investigations in two respects. First, it highlights the ability of an iron based carrier to generate hydrogen, and so considers reduction to Fe, rather than simply to Fe3O4, the lowest oxide in conventional chemical looping with methane.13 Second, if

the redox process is to be used to convert a carbonaceous fuel, such as coal or biomass, into hydrogen, repeated reduction of the carrier in gases with significant CO content must be considered and previous investigations have performed reduction of sol-gel derived carriers in a mixture of CO and H2 over a single cycle only.9 Experimental Section Oxygen carriers with different mass ratios of Fe2O3 and Al2O3 (60:40, 80:20, 90:10 wt %) were prepared using a sol-gel technique according to the Yoldas process.23 In a typical synthesis, aluminum isopropoxide (Acros Organics; >98 wt % purity) was added to water which had been purified by reverse osmosis and heated to 75 °C; the mixture was allowed to hydrolyze slowly for 2 h with constant stirring. The resulting slurry was peptised with nitric acid (Fisher; 70 wt %, diluted). In all syntheses, the molar ratio between Al3+, water, and H+ was constant and equal to 0.5:50:0.07. The temperature was then raised to 90 °C and the sol was refluxed for 12 h. Subsequently, the appropriate amount of Fe(NO3)3 · 9 H2O (Acros Organics, >98 wt % purity) dissolved in water at 90 °C to obtain a ∼1 M aqueous solution was added. The resulting slurry was refluxed for another 12 h at 90 °C, followed by cooling to room temperature. The encapsulated solvents were removed by drying the gel in an oven at 100 °C overnight. The resulting xerogel was calcined for 3 h at 900 °C and then sieved to +300, -425 µm. For comparison, particles of unsupported iron oxide were prepared by spraying water on to Fe2O3 powder (Sigma-Aldrich, purity >99.9 wt %, size 99 wt.% purity) with a frit of 4 holes, each 1.5 mm diameter The bed was loaded by inserting (i) 2 g of inert +1.4, -1.7 mm Al2O3 (Boud Mineral, >99 wt % purity) on to the frit, followed by (ii) 2 g of +300, -425 µm Al2O3, (iii) 0.3 g of +300, -425 µm oxygen carrier, and (iv) 10 g of +1.4, -1.7 mm Al2O3. The bottom layers, i and ii, prevented the iron particles from passing through the holes in the perforated plate, and the top layer, iv, served to preheat the reactant gas. A tubular furnace maintained the temperature of the reactor at 850 °C, as measured by a type-K

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010

5385

Table 2. Measured BET Surface Area, BJH Pore Volume, Mean Pore Diameter, and Crushing Stress after Calcining at the Specified Conditions and Prior to Reaction for the 60, 80, and 90 wt % Fe2O3 Oxygen Carriers Prepared Using Sol-Gel, 100 wt % Al2O3 Prepared Using Sol-Gel, and 100 wt % Fe2O3 Prepared by Mechanical Mixinga

preparation

composition

BET (m /g)

BJH (cm /g)

djpore (nm)

sol-gel sol-gel sol-gel sol-gel sol-gel mech mix

60 wt % Fe2O3, 40 wt % Al2O3 80 wt % Fe2O3, 20 wt % Al2O3 90 wt % Fe2O3, 10 wt % Al2O3 100 wt % Al2O3 100 wt % Al2O3 100 wt % Fe2O3

15.5 8.5 10.5 168 299 1.5

0.09 0.06 0.06 0.35 0.40 0.00

24 27 22 6 4 7

2

3

calcining temp

time

900 °C 900 °C 900 °C 900 °C 500 °C 900 °C

3h 3h 3h 3h 3h 3h

crush stress (MPa) 0.2 0.6 0.3 3.3

Determination of the crushing stress involved particles of diameter +1700, -2060 µm; all other characterizations involved particles of diameter +300, -425 µm. a

thermocouple (o.d. 1.5 mm) placed within the layer of particles of oxygen carrier. Gases of various compositions were supplied to the reactor, operated in a downflow arrangement, from cylinders of (i) 10 vol % CO, balance N2, (ii) pure N2, and (iii) pure CO2; (iv) laboratory air was also supplied. The flow rate of gas was controlled using rotameters for streams i and iii and mass flow meters (AWM5101VN, Honeywell) for streams ii and iv. Automatic switching between inlet gases was performed with solenoid valves (Burkert). Steam was supplied using a syringe pump to feed liquid H2O at a flow rate of 30 mL/h through hypodermic tubing (0.8 mm i.d.) into an electrically heated chamber at 150 °C, swept with N2. The flow rate, gas composition, and cycle time for each experiment is listed in Table 3. The sampled gas was dried prior to analysis using two impinger tubes in series immersed in an ice bath, followed by a tube filled with granular CaCl2. The composition of the effluent gas was determined using (i) nondispersive infrared analysers measuring [CO] and [CO2] in the range 0-20 vol % (ABB, EasyLine), (ii) a nondispersive infrared analyzer measuring [CO] in the range 0-1 vol % (ABB, EasyLine), and (iii) a thermal conductivity analyzer measuring [H2] in the range 0-30 vol % (ABB, Caldos27). Cylinders of pure gases with compositions of (i) 10 vol % CO, balance N2, (ii) 10 vol % CO2, balance N2, (iii) 1 vol % CO, balance N2, and (iv) 10 vol % H2, balance N2, were used for calibration. A field emission scanning electron microscope (FESEM, JSM-6340F) operated at 5 kV was used to obtain high resolution images of the morphology of the oxygen carriers prior to the redox experiments. X-ray diffraction (XRD, Philips model PW1830/00, Cu KR, 40 kV and 40 mA, 0.01° s-1, receiving and antiscatter slits 1°, divergence slit 0.3 mm, in air at 298 K) was used to detect the presence of FeO · Al2O3. The specific surface area of the samples was calculated from N2 adsorption (Micromeritics, Tristar 3000) using the BET model;24 the cumulative pore volume for pores between 1.7 and 200 nm in diameter was estimated using the BJH model.25 For a sphere subjected to a force, F, on opposite sides of a diameter, the crushing stress was calculated from the empirical relation:10,26 σT ) k

F R2

(6)

where σT is the crushing stress (N/m2), F is the applied force (N), R is the sphere’s external radius (m), and k is a dimensionless constant with a typical value of k ) 0.22. The average value of the crushing stress, σT (Stable Micro Systems, TA-XT2 Texture Analyzer with a 5 kg load cell), for 20 particles with R ) 950 µm, omitting the maximum and minimum value, is reported in Table 2.

Results To quantify the ability of the oxygen carrier to produce H2 over repeated redox cycles, a conversion was defined as XH2 )

( )

mH2 MFe2O3 3 xwmox

MH2 8



mH2 xwmox

× 30

(7)

where mH2 is the mass of the H2 produced, mox is the mass of the fully oxidized carrier, xw is the mass fraction of Fe2O3 in the fully oxidized carrier, and MFe2O3 and MH2 are the molecular weights of Fe2O3 and H2, respectively. Here, it should be noted that the conversion is based on the total mass of iron oxide in the sample, i.e. the assumption is made that all of the Fe2O3 is available for reaction and no provision for the formation of unreactive FeO · Al2O3 is included. Initially, 10 redox cycles with 0.3 g of the unsupported Fe2O3 oxygen carrier of size +300, -425 µm prepared by mechanically mixing were performed at 850 °C. Figure 1 shows XH2 versus cycle number. Reduction to FeO in a mixture of 10 vol % CO, 10 vol % CO2, balance N2 produced stable quantities of H2 near that predicted from reaction stoichiometry (reaction 3). It is noted that stable values of XH2 were obtained whether samples were oxidized in steam or a sequence of steam and air. In contrast to reduction to FeO, for unsupported Fe2O3 reduced to Fe, XH2 drops to 0.1 by cycle 2 and 0.03 by cycle 10. If additional oxidation in air (reaction 5) is performed, this drop is less precipitous; however, Figure 1 shows that a low conversion of XH2 ) 0.1 is obtained by cycle 10. Thus, for repeated production of H2, it seems preferable to reduce unsupported Fe2O3 only to FeO, since by cycle 10, experiments involving reduction to FeO generated higher quantities of H2 than experiments involving reduction to Fe. Nevertheless, reaction stoichiometry shows that reduction to Fe has the potential to produce four times as much H2 as reduction to FeO, as shown by the right-hand ordinate of Figure 1 and by comparing reactions 3 and 4. Since, it is not thermodynamically possible to produce H2 in appreciable quantities by oxidizing Fe3O4 to Fe2O3 with steam,6 the zero on the graph has been marked with the Fe3O4 phase. Next, reduction of the sol-gel oxygen carriers with xw ) 0.6, 0.8, and 0.9 Fe2O3, balance Al2O3, was performed at 850 °C; the Fe2O3 was reduced completely to Fe. Figure 1 demonstrates that the production of H2 using the redox of iron oxides is promising for sol-gel derived carriers with 80 and 90 wt % Fe2O3 over 10 cycles, with conversions of XH2 ) 0.9. Open symbols represent experiments where oxidation was performed with 25 vol % steam, balance N2 only (i.e., reaction 4); filled symbols represent experiments where oxidation was performed in 25 vol % steam, balance N2 (reaction 4), followed by oxidation in air (reaction 5). In Figure 1, for particles with 60

5386

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010

Figure 1. Measured conversion, XH2, for 0.3 g of oxygen carrier reacted at 850 °C over 10 cycles. Open symbols represent experiments where oxidation was performed in 25 vol % steam, balance N2 only for the 60 (4), 80 (O), and 90 (0) wt % Fe2O3 carriers prepared using sol-gel and for unsupported Fe2O3 ()) prepared via mechanical mixing. Closed symbols represent experiments where oxidation was performed in 25 vol % steam, balance N2 followed by air for the 60 (2), 80 (•), and 90 (9) wt % Fe2O3 carriers prepared using sol-gel and for the unsupported Fe2O3 (() prepared via mechanical mixing. These carriers were reduced in 10 vol % CO, balance N2. Conversion for unsupported Fe2O3 reduced to FeO in a mixture of 10 vol % CO, 10 vol % CO2, balance N2 and oxidized sequentially in steam and air is also shown (+). A conversion of XH2 ) 1 represents the theoretical quantity of H2 expected from a 0.3 g sample for complete reduction to Fe as shown by the right-hand ordinate; the expected conversion for reduction to the phase FeO or Fe3O4 is also indicated.

Figure 2. Field emission scanning electron microscope (FESEM) images of (a) the 100 wt % Fe2O3 prepared by mechanically mixing and (b) a 60 wt % Fe2O3, 40 wt % Al2O3 carrier prepared using sol-gel. The particles are shown in their fully oxidized state after sintering at 900 °C for 3 h and prior to any reaction. A difference in grain size is noticeable; the scale bars in each figure are of equal length, 100 nm.

wt % Fe2O3, the observed conversion is lower than the value predicted from stoichiometry in reaction 4. An average over 10 cycles gives XH2 ) 0.78 and XH2 ) 0.56 for experiments with and without additional oxidation in air, respectively. Notably, the filled symbols in Figure 1 have higher values of XH2 than their open symbol counterparts indicating that oxidation of the carrier with air, via reaction 5, after oxidation in steam, reaction 4, has a beneficial effect. An explanation for this observation is provided in the discussion. The gray lines in Figure 1 indicate least-squares fits to the data and are shown only to guide the eye. Figure 2 shows scanning electron microscope images for (a) the unsupported Fe2O3 particles prepared by mechanical mixing and (b) the 60 wt % Fe2O3, 40 wt % Al2O3 particles prepared using the sol-gel method. The important difference between these images is the disparity in the size of the individual grains. Unmodified Fe2O3 has grains with an average diameter of 500 nm, while the sol-gel carrier has grains with an average diameter of 80 nm, nearly 1 order of magnitude smaller. The morphological difference between these two samples could contribute toward the large difference in conversion, XH2, observed in Figure 1. Calcination can significantly change the pore structure of dried gels. Table 2 shows results of the BET surface area and mean pore diameter, djpore, for sol-gel prepared alumina dried in an oven at 100 °C, followed by calcination at either 500 or 900

Figure 3. Measured conversion, XH2, for 0.3 g of oxygen carrier over 30 to 40 cycles at 850 °C. Results for carriers with 60 (2), 80 (•), and 90 (9) wt % Fe2O3 prepared using sol-gel and reduced to Fe, as well as for unsupported Fe2O3 prepared by mechanically mixing and reduced only to FeO (+) are shown. Oxidation was performed in 25 vol % steam, balance N2, followed by air. Final conversions for modified particles on various supports from the literature with 38 wt % Fe2O335 at 740 °C (×) and 76-78 wt % Fe2O336 at 650 °C (-, ), *) are also shown for comparison. Increasing the reduction time from 10 to 20 min with a gas composition of 10 vol % CO, balance N2, resulted in a higher conversion for the 90 wt % Fe2O3 particles (9), shown by the point within the dashed circle. A conversion of XH2 ) 1 represents the theoretical quantity of H2 expected from a 0.3 g sample for complete reduction to Fe as shown by the right-hand ordinate; the expected conversion for reduction to the phase FeO or Fe3O4 is also indicated.

°C. Under thermal processing, well-crystallized boehmite (AlOOH) will form γ-alumina in the range 450-600 °C and δ-alumina in the range 600-1050 °C.27,28 The average pore diameter for pure Al2O3 samples increased with calcination temperature with an average pore radius of 4 and 6 nm for samples calcined at 500 or 900 °C, respectively. Calcining the composite particles at 900 °C resulted in reduced surface areas (10 m2 /g) and pore volumes (0.06 cm3/g) compared to the pure δ-Al2O3 calcined at 900 °C (170 m2/g, 0.4 cm3/g). The decrease in surface area and increase in pore diameter with increasing calcination temperature has been observed for iron oxide and silica29 composites as well as cobalt aluminate particles30 prepared using sol-gel. These results could be due to sintering of particles. In addition, the nitrate salt could cause aggregation, resulting in formation of more crystalline structure with lower surface area. Figure 1 demonstrated a marked increase in XH2 for composite particles of Fe2O3 and Al2O3 prepared using sol-gel compared to that for unsupported Fe2O3 over 10 cycles. However, for chemical looping combustion on an industrial scale, the oxygen carrier must withstand hundreds or even thousands of redox cycles. The sol-gel carriers with xw ) 0.6, 0.8, and 0.9 Fe2O3 were tested at 850 °C over 30-40 cycles for complete reduction on each cycle to Fe. In all cases, air was introduced after the steam to oxidize the Fe3O4 to Fe2O3 via reaction 5; the reducing and oxidizing gases and the flow rates were identical to those used in the experiments shown in Figure 1 by filled symbols and are given in Table 3. Figure 3 shows the conversion of the oxygen carrier, XH2, calculated from eq 7, versus cycle number. For the 80 and 90 wt % Fe2O3 oxygen carriers, XH2 > 0.9 for the initial 7 cycles. From cycle 8-25, shown between the dashed vertical lines, however, the conversion decreases to XH2 ) 0.43 and XH2 ) 0.34 for the 80 and 90 wt % Fe2O3 carriers, respectively. From cycle 25-36, a further reduction of the conversion is observed. In Figure 3, conversions for the 90 wt % Fe2O3 carrier are consistently lower than those for the 80 wt % Fe2O3 carrier. Overall, in cycle 35 for the samples with 80 and 90 wt % Fe2O3, respectively, approximately 4.7 and 4.4 mmol H2/g carrier was obtained, where the mass is for the fully oxidized carrier, i.e. a composite of Fe2O3 and Al2O3.

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010

5387

Table 3. Redox Cycling: Gas Mixtures (vol %), Flow Rates as Measured at 298 K and 101 kPa, and Corresponding Flow Durations (s) for Experiments with an Initial Charge of 0.3 g of Oxygen Carriera purge time (min) flow (mL/s) flow (mL/h) Fe3O4-Fe Fe2O3-FeO Fe2O3-Fe Fe2O3-Fe Fe2O3-Fe

reduction

0-2 33

2-12 33

N2 N2 N2 N2 N2

10% 10% 10% 10% 10%

CO/90% CO/10% CO/90% CO/90% CO/90%

N2 CO2/80% N2 N2 N2 N2

purge

oxidation

purge

oxidation

12-13 33

13-16 33 N2, 30 H2 O(l) 75% N2/25% steam 75% N2/25% steam 75% N2/25% steam N2, air air

16-17 33

17-20 33

N2 N2

air air

N2 N2 N2 N2 N2

figures

symbol

1 1, 3 1, 3, 4, 5 5 5

4, O, 0, ) + 2, b, 9, ( ( )

a Reoxidation to Fe2O3 required air. The total cycle time was 20 and 16 min for experiments with and without oxidation in air. The figure number and symbol corresponding to the experiment in which each sample was used is listed.

Surprisingly, the 60 wt % Fe2O3 oxygen carrier displays a nearly constant conversion with an average value of XH2 ) 0.75 over 40 cycles, as shown in Figure 3. The total moles of H2 produced per unit mass of particles is also higher for the carrier with 60 wt % Fe2O3, giving 7.5 mmol H2/g carrier. Figure 3 also shows XH2 for unsupported Fe2O3, reduced to FeO in a mixture of 10 vol % CO, 10 vol % CO2, balance N2. The conversion is again stable with XH2 ) 0.15 after 40 cycles. All of the sol-gel carriers, i.e. xw ) 0.6, 0.8, and 0.9, however, outperform the unsupported, mechanically mixed Fe2O3 reduced to FeO, which gave an average of 3.0 mmol H2/g carrier over 40 cycles. The point inside the dashed circle at cycle 42 in Figure 3 represents an experiment where the reduction time was increased from 10 to 20 min for the 90 wt % Fe2O3 carrier. The reducing gas composition was still 10 vol % CO, balance N2. Doubling the reduction time increased the conversion from XH2 ) 0.25 to 0.35, indicating that reduction had not reached completion after the 10 min reduction interval used here. The remaining values in Figure 3, labeled with temperatures of 740 and 650 °C were taken from other studies and will be highlighted in the discussion. The measured BET surface area for the 90 wt % Fe2O3 particles used for experiments in Figure 3 after 42 cycles was 0.1 m2/g; the surface area for the after 60 wt % Fe2O3 particles used for experiments in Figure 3 after 40 cycles was 0.4 m2/g. It should be mentioned that the volume fraction of CO in the effluent at the end of each reduction cycle in Figure 3 was g9.7 vol %; the inlet concentration of CO was 10 vol %. Therefore, no large discrepancy between the inlet and outlet concentration of CO existed prior to switching off the reducing gas. For use in polymeric electrolyte membrane (PEM) fuel cells, hydrogen must have a contamination of [CO] < 50 vol ppm so as not to poison the Pt on the anode. Other processes, e.g. ammonia synthesis or hydrocracking,31 have less stringent [CO] requirements. In this study, complete reduction in 10 vol % CO, balance N2, provided a “worst case” indication of the performance of the oxygen carriers since (i) it is generally expected to be slower than reduction in H2 and (ii) it permitted the deposition of solid carbon. Solid carbon was deposited on the iron oxide surface, owing to the Boudouard reaction, 2CO a C(s) + CO2, and was subsequently oxidized to CO when steam was introduced. The H2 in this study had an average contamination, calculated from results in Figures 1 and 3 for the 80 and 90 wt % Fe2O3 oxygen carriers, of [CO]/([H2] + [CO]) ) 1 vol % ) 10000 vol ppm, much in excess of the 50 vol ppm limit for PEM fuel cells. For the 60 wt % Fe2O3 oxygen carriers, experiments in Figures 1 and 3 gave [CO] ) 0.1 vol % ) 1000 vol ppm. By contrast, reduction of the unsupported, mechanically mixed oxygen carriers to FeO in a mixture of 10 vol % CO, 10 vol % CO2, balance N2, such that the Boudouard reaction

is thermodynamically unfavorable, has previously been shown to give H2 with a contamination of [CO] < 0.005 vol % ) 50 vol ppm.8 From the experiments shown in Figure 3, the conversion as a function of time for 0.3 g of the (a) 60, (b) 80, and (c) 90 wt % Fe2O3 oxygen carriers reduced in 10 vol % CO, balance N2, at 850 °C was derived and is displayed in Figure 4. Each line consists of 600 points and corresponds to a specific cycle from Figure 3. The conversion, XCO, was calculated from the amount of CO consumed by XCO )

(

)

mCO MFe2O3 1 mCO ≈ × 1.9 xwmox MCO 3 xwmox

(8)

where mCO is the mass of the CO consumed, mox is the mass of the fully oxidized carrier, xw is the mass fraction of Fe2O3 in the fully oxidized carrier, and MFe2O3 and MCO are the molecular weights of Fe2O3 and CO, respectively. Since the CO reacts to form a commensurate quantity of CO2, the conversion based on the CO2 signal, XCO2, provided a useful check; typically, XCO ) XCO2 ( 0.05 where XCO and XCO2 vary between 0 and 1. In Figure 4a, the final conversion for the 60 wt % Fe2O3 carrier at t ) 600 s decreased from XCO ) 0.87 in cycle 1 to XCO ) 0.68 in cycle 30. In Figure 4b, the final conversion of the 80 wt % Fe2O3 oxygen carrier decreased with cycle number and was XCO ) 1.0 and 0.27 for cycles 1 and 30, respectively. In Figure 4c, the final conversion for the 90 wt % Fe2O3 oxygen carrier decreases with cycle number and is given by XCO ) 0.95 and 0.22 for cycles 1 and 30, respectively. Black circles indicate that the time to reach 50% and 80% of the final conversion, a cursory measure of the rate of the reduction reaction, increased with cycle number for oxygen carriers with 80 and 90 wt % Fe2O3, but remained approximately constant for that with 60 wt % Fe2O3, as indicated by dashed arrows. If the production of hydrogen is not of primary concern, e.g. in traditional chemical looping combustion, the oxygen carrier can be fully oxidized in air rather than in steam by the following exothermic reaction: 2Fe + 3/2O2 a Fe2O3

o o ∆H850 C ) -814.0 kJ/mol (9)

It should be noted that in traditional chemical looping combustion, which is primarily concerned with the complete conversion of syngas to CO2 and H2O, the oxygen carrier will only be reduced to Fe3O4 and Fe will only arise locally, e.g. near the gas inlet of the fuel reactor. The effect of the choice of oxidizing gas on conversion, XCO, is shown in Figure 5 for 0.3 g of the 90 wt % Fe2O3 oxygen carrier reduced to Fe at 850 °C. The temperature of the layer of particles of oxygen carrier at the start of each oxidation cycle was also 850 °C. Figure 5 demonstrates that the initial reduction in cycle 1 gives XCO ≈ 1. This value

5388

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010

Figure 4. Measured conversion, XCO, versus time for 0.3 g of (a) 60, (b) 80, and (c) 90 wt % Fe2O3, balance Al2O3, carrier prepared using sol-gel for cycles 1, 5, 10, and 30 at 850 °C. In all cases, reduction was performed in 10 vol % CO, balance N2 for 600 s; oxidation was performed with 25 vol % steam, balance N2 followed by air. Black circles represent the time to reach 50%, 80%, and 100% of the final conversion at t ) 600 s.

Figure 5. Effect of oxidizing gas on conversion for 0.3 g of a 90 wt % Fe2O3, 10 wt % Al2O3, oxygen carrier prepared using sol-gel over six cycles at 850 °C. The oxygen carrier was reduced to Fe in 10 vol % CO, balance N2. Oxidation was performed in air with 21 vol % O2 ()), air diluted with N2 to give 4 vol % O2 ((), and sequentially with 25 vol % steam, balance N2 followed by air with 21 vol % O2 (9).

is independent of oxidizing gas, since no oxidation step had yet been performed. For oxidation in air with 21% O2, balance N2, the conversion drops to 0.2 in cycles 2-6. Here, the oxidation in air was accompanied by a 15 °C temperature rise and inspection of the 0.3 g bed of oxygen carrier after the six cycles showed agglomeration. Oxidation in a mixture of N2 and air giving 4 vol % O2, balance N2, showed a drop in conversion to XCO ) 0.65 in cycles 2-6; no agglomeration was observed, and oxidation was accompanied by a smaller temperature rise of ∼5 °C. Two identical experiments with oxidation in 25 vol % steam, balance N2, followed by oxidation in air (21 vol % O2, balance N2) are also shown and gave the highest overall conversion with an average of XCO ) 0.88 for cycles 2-6; no temperature rise and no agglomeration was observed. The furnace was not run at a constant power output, but rather controlled, so the observed temperature deviations were damped. Discussion The redox reactions of iron oxide enable synthesis gas, derived from either coal or biomass, to be upgraded to H2 with the simultaneous capture of CO2. Figure 1 showed that oxygen carriers prepared using a sol-gel method with 60, 80, and 90 wt % Fe2O3 could successfully produce H2 over 10 cycles at 850 °C. Samples with 60 wt % Fe2O3 gave lower values of XH2 compared to those with 80 or 90 wt % Fe2O3. Oxidation in air was also found to have a beneficial effect and resulted in higher values of XH2. These two observations can be explained by the formation of unreactive FeO · Al2O3. A thermodynamic analysis of the stability of FeO · Al2O3 was undertaken using published thermodynamic tables.6 A min-

imimization of the Gibbs free energy at 850 °C and 101 kPa subject to atomic and molecular constraints was performed by considering the solid species, Fe, FeO, Fe3O4, Fe2O3, Al, FeO · Al2O3, and Al2O3, for different partial pressures of mixtures of CO and CO2 or H2 and H2O. The formation of FeO · Al2O3 was found to be favorable for 8 × 10-3 < pCO2/pCO < 1.8 × 105 and 1 × 10-2 < pH2O/pH2 < 2.1 × 105. Thus, FeO · Al2O3 should be stable over the entire redox cycle, since during reduction incoming CO is readily oxidized to CO2 giving pCO2/pCO > 8 × 10-3 and during oxidation even trace amounts of H2 will cause pH2O/pH2 < 2.1 × 105. Only if oxidation in air is performed will FeO · Al2O3 be oxidized to Fe2O3 and Al2O3. It should be noted that reduction of FeO · Al2O3 to Fe and Al2O3 in CO will never occur if the species C(s), Fe3C, and Al4C3 are included from standard thermodynamic tables6 in the minimization problem. Here, some CO will react to form C(s) and CO2 via the Boudouard reaction, 2CO a C(s) + CO2, thereby maintaining 2 pCO2/pCO ≈ pCO2/pCO ≈ 0.06 > 8 × 10-3 at 850 °C and 101 6 kPa. Assuming that equilibrium is reached, the preceding experimental observations can now be explained as follows: (i) increasing the amount of labile Al2O3 in the oxygen carrier will result in more atomic Fe being bound as FeO · Al2O3 and unavailable for reaction and therefore lead to lower initial conversions in oxygen carriers with a higher mass fraction of Al2O3; (ii) since FeO · Al2O3 decomposes to Fe2O3 and Al2O3 in air, oxidation in air will the release reactive iron oxide once per cycle and could lead to higher values for XH2. To verify experimentally the formation of FeO · Al2O3, the oxygen carrier consisting of 60 wt % Fe2O3 which had been reduced in 10 vol % CO, balance N2, and oxidized in 25 vol % steam, balance N2 for 10 cycles was examined using X-ray diffraction (XRD). Figure 6 shows the resulting diffraction pattern. The full peak widths at half of the maximum intensity (fwhm) were typically 0.3-0.5° over 20° < 2θ < 80° and suggested peak broadening. For comparison, the instrument broadening estimated from a pure sample of Fe3O4 of known crystallite size gave ∼0.17° (fwhm) at 2θ ) 34.8°. Peak broadening is common in samples with either small crystallites below 1 µm or in which there is significant lattice stress. In the present case, peak broadening could not be fully explained by the Scherrer equation and small crystallite size. Lattice stress owing to the chemically heterogeneity must therefore also contribute and is likely, since the lattice dimensions of Fe3O4 and FeO · Al2O3 are different with a ) b ) c ) 8.40 and 8.15 Å for Fe3O4 and FeO · Al2O3, respectively.32 It is noted that both Fe3O4 and FeO · Al2O3 are cubic (R ) β ) γ ) 90°) with a space group of Fd3m j . The diffraction pattern for the sample also suggests a mixture of crystallites of Fe3O4 and

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010

Figure 6. Points (•) show the experimental X-ray diffraction pattern for a 60 wt % Fe2O3, 40 wt % Al2O3 oxygen carrier which had been reduced for 10 cycles in 10 vol % CO, balance N2, and oxidized in 25 vol % H2O, balance N2. The theoretical diffraction pattern from a Rietveld analysis is plotted as a line (-). The difference is plotted below the diffraction patterns and demonstrates the satisfactory agreement. Pairs of vertical lines show the peaks for identical Bragg reflection planes for pure Fe3O4 (red, left line in each pair) and FeO · Al2O3 (gray, right line in each pair); the experimental X-ray diffraction peak lies between the pairs of lines in each case suggesting that the sample is a mixture of Fe3O4 and FeO · Al2O3.

FeO · Al2O3.32 The maximum peak for the sample consistently lies within the bounds of 2Θ given by pairs of vertical lines corresponding to the peaks for identical Bragg reflection (hkl) planes of Fe3O4 (left, red) and FeO · Al2O3 (right, gray). Phase identification, using published information,32 revealed the presence of Fe3O4 and FeO · Al2O3; the peak at 41.90° was attributed to FeO which is also cubic and has a listed primary peak at 41.93°. The lack of intensity at 25.58° was used to rule out the presence of corundum, Al2O3. Next, a multiphase Rietveld refinement using standards from the ICSD database33 was performed. Figure 6 shows the experimental diffraction pattern for the oxygen carrier as points, the predicted diffraction pattern from the quantitative Reitveld analysis as a line and the difference, in counts per second, between the two. Good reproduction of peaks and peak shoulders was obtained; the final weighted profile R value and experimental R value were Rwp ) 12.7 and Rexp ) 8.9. The quantitative estimate of each phase gave 76 mol % FeO · Al2O3, 20 mol % Fe3O4, and 4 mol % FeO. The actual mole fraction of FeO · Al2O3 in the sample assuming it is composed entirely of Fe3O4 and FeO · Al2O3 was 75 mol %, in good agreement. The formation of FeO · Al2O3 is consistent with Topsøe et al.34 who suggested the formation of FeO · Al2O3 during reduction of an iron based ammonia catalyst. For increased numbers of cycles of operation, a noticeable drop in XH2 with cycle number was observed for oxygen carriers with 80 and 90 wt % Fe2O3, giving XH2 < 0.35 after 35 cycles as shown in Figure 3. By contrast, carriers with 60 wt % Fe2O3, 40 wt % Al2O3, gave stable quantities of H2 over 40 cycles with an average conversion of XH2 ) 0.75. Since in all cases oxidation in air was performed to oxidize the FeO · Al2O3 to Fe2O3 and Al2O3, it appears that a minimum amount of Al2O3 is required to stabilize the experimental yield of H2 in each cycle; current experiments suggest that this value lies between 20 and 40 wt % Al2O3. If only the iron that is capable of forming FeO · Al2O3 were able to react, 2 mol of Al would stabilize 1 mol of Fe, and the expected conversions for carriers with xw ) 0.9, 0.8, and 0.6 would be XH2 ) 0.09, 0.20, and 0.52, respectively. The observed conversions after 40 cycles are slightly higher than these values. So it seems that the formation

5389

of FeO · Al2O3 at some point during the reduction and oxidation cycle is nearly, but not completely, sufficient to account for all of the H2 produced. Agglomeration between the Al2O3 packing material and the oxygen carrier was not observed in any experiment on emptying the tube containing the packed bed. The stable conversions observed for the reduction of a 60 wt % Fe2O3, 40 wt % Al2O3 oxygen carrier prepared using sol-gel are consistent with the work of Li et al.9 who used a carrier with identical mass fractions, but with iron chloride as a precursor, rather than iron nitrate, and who performed reduction in 37.5% H2, balance N2 using a thermogravimetric analyzer, rather than in 10 vol % CO, balance N2 using a packed bed. Overall, for chemical looping combustion on an industrial scale, the 60 wt % Fe2O3 carriers appear to be preferable. The force needed to crush a particle of oxygen carrier is one measure of its robustness for chemical looping combustion in a fluidized bed where interparticle and particle/wall collisions occur at high frequency. Table 2 demonstrates that the crushing stresses for the oxygen carriers prepared using sol-gel were lower than those for the mechanically mixed Fe2O3. In general, particles with higher surface area gave lower crushing stresses. Table 1 suggests that an increase in crushing stress can be achieved by increasing the calcining time to g6 h and increasing the temperature up to 1370 °C. More severe temperatures would likely result in the formation of larger grains and could be used in the future to produce carriers with higher crushing stresses. However, increased sintering would almost certainly adversely affect the activity of the oxygen carrier. The crushing stress for oxygen carriers from Mattison et al.13 was calculated using eq 6 with a particle radius of 112 µm and reported breakage forces for particles calcined at 1100 and 1300 °C of 1 and 9.5 N, respectively; this is seen in Figure 5 of ref 13. Abad et al.17 used an identical oxygen carrier to that of Mattisson et al.13 but with R ) 150 µm, rather than R ) 112 µm; this difference in radius was deemed insignificant and an identical value reported. Adanez et al.14 list the crushing stress for cylindrical extrudates of diameter D ) 2 mm in units of force per unit length extrudate. To convert this value in to an estimate for an equivalent spherical crushing stress, a length of extrudate of L ) 2 mm was assumed and eq 6 was used with the approximation R ) L/2 ) D/2 ) 1 mm. Recently, chemical looping in packed bed reactors has also been proposed,37 and the oxygen carriers with lower crushing stresses could be suitable for such a process. To compare the conversions achieved in this work to those in other studies, the definition of XH2 in this paper was adopted, i.e. it was assumed that all of the iron oxide present in the oxygen carrier was available for reaction. Composite particles from other groups often contain a variety of metals with different supports and additives to improve reduction and oxidation kinetics and overall carrier activity. From Figure 8 in the work of Galvita et al.,35 the quantity of hydrogen produced per unit mass of oxygen carrier with a composition of 5 wt % Cr2O3, 38 wt % Fe2O3, and 57 wt % of equimolar CeO2 and ZrO2 was calculated to be, on average, 1.7 mmol/g carrier. This group found that the addition of Cr improved reduction and oxidation kinetics and overall carrier activity by decreasing sintering. The conversion, XH2, was then determined by multiplying the quantity of hydrogen produced per unit mass of oxygen carrier by MFe2O3 × 3/8 and dividing by xw ) 0.38, the mass fraction of Fe2O3. The values obtained over the first 40 cycles are shown in Figure 3 and averaged XH2 ) 0.27. Assuming that the iron oxide reduces sequentially according to Fe2O3 f Fe3O4 f FeO f Fe, this value of XH2 suggests that the reduced state of the oxygen carrier was a mixture of FeO and Fe, as shown by the

5390

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010

reported values lying between the ticks for these phases on the right-hand ordinate of Figure 3. An identical procedure was applied to Figure 10a-c of the work of Galvita et al.36 for samples with (a) 2 wt % Mo, 78 wt % Fe2O3, 20 wt % CeO2 and ZrO2, (b) 5 wt % Mg, 76 wt % Fe2O3, 19 wt % CeO2, and ZrO2, and (c) 5 wt % Cu, 78 wt % Fe2O3, 19 wt % CeO2, and ZrO2. Here, the average amount of H2 produced was 0.8, 1.0, and 1.2 mmol H2/ g carrier, respectively, giving conversions between 0.06 < XH2 < 0.11. For sequential conversion, the most reduced state of the oxygen carrier was thus a mixture of Fe3O4 and FeO as shown in Figure 3. In all cases, the oxygen carrier prepared by the sol-gel technique compared favorably to those in other studies.35,36 However, it is noted that the purity of the H2 in studies by Galvita was significantly higher, [CO]/([H2] + [CO]) ) 10 vol ppm, compared to 1000 vol ppm for experiments with the 60 wt % Fe2O3 carrier prepared using sol-gel presented here. Conclusion Oxygen carriers with different ratios of Fe2O3 and Al2O3 by mass (60:40, 80:20, and 90:10 wt %) were prepared using a sol-gel technique and tested for use in chemical looping combustion. Samples were characterized using scanning electron microscopy and by measuring crushing stress, BET surface area, and BJH pore volume. A packed bed reactor at 850 °C was used to reduce 0.3 g of oxygen carrier in a mixture of CO and N2; oxidation in a mixture of steam and N2 was performed, followed by oxidation in air in some cases. It was found that: (1) For reduction to the FeO phase, no Al2O3 support was required for stable conversions over 40 cycles. (2) For reduction to the Fe phase, a drop in conversion, XH2, (Figures 1 and 3) and reactivity (Figure 4) with cycle number was observed for oxygen carriers with a loading of only 10-20 wt % Al2O3. (3) For reduction to the Fe phase, 40 wt % Al2O3 was sufficient to give stable conversions of XH2 ) 0.75 over 40 cycles. The higher conversion with higher mass fraction of Al2O3 after 40 cycles was linked to the formation of FeO · Al2O3. (4) The formation of FeO · Al2O3 was definitively confirmed with quantitative X-ray diffraction and was used to explain the experimental observation that for composite particles prepared using the sol-gel method, additional oxidation in air resulted in higher conversions (Figure 1). (5) Reduction to Fe in 10 vol % CO, balance N2, resulted in the deposition of solid carbon on the oxygen carrier’s surface. This carbon was subsequently oxidized to CO with the introduction of steam and gave an average value of [CO]/([H2] + [CO]) ) 0.001-0.01 vol %. The 60 wt % Fe2O3, 40 wt % Al2O3, oxygen carrier produced using sol-gel gave conversions competitive with those in other studies and seems promising for use in chemical looping applications. Acknowledgment Assistance from Mr. Simon Griggs with SEM, Dr. Mary Vickers with XRD, Mr. Z. Saracevic with N2 adsorption, and the Chemical Database Service at Daresbury is acknowledged. Financial support from the Engineering and Physical Sciences Research Council (EP/F027435/1) and the Gates Cambridge Trust is also recognized.

Literature Cited (1) IntergoVermental Panel on Climate Change, Climate Change 2007: Synthesis Report; Cambridge University Press: Cambridge, 2007. (2) Anthony, E. J. Solid looping cycles: A new technology for coal conversion. Ind. Eng. Chem. Res. 2008, 47, 1747–1754. (3) Hossain, M. M.; de Lasa, H. I. Chemical-looping combustion (CLC) for inherent CO2 separations - a review. Chem. Eng. Sci. 2008, 63, 4433– 4451. (4) Song, Q.; Xiao, R.; Deng, Z; Zhang, H.; Shen, L.; Xiao, J.; Zhang, M. Chemical-looping combustion of methane with CaSO4 oxygen carrier in a fixed bed reactor. Energy ConVers. Manage. 2008, 49, 3178–3187. (5) v. Bogdandy, L; Engell, H-J. The Reduction of Iron Ores; SpringerVerlag: Berlin, 1971, 19-30. (6) Barin, I.; Knacke, O. Thermochemical properties of inorganic substances; Springer-Verlag: Berlin, 1973. (7) Messerschmitt, A. Process of producing hydrogen. U.S. Pat. 971,206, 1910. (8) Bohn, C. D.; Mu¨ller, C. R.; Cleeton, J. P.; Hayhurst, A. N.; Davidson, J. F.; Scott, S. A.; Dennis, J. S. Production of very pure hydrogen with simultaneous capture of carbon dioxide using the redox reactions of iron oxides in packed beds. Ind. Eng. Chem. Res. 2008, 47, 7623–7630. (9) Li, F.; Kim, H. R.; Sridhar, D.; Wang, F.; Zeng, L.; Chen, J.; Fan, L. S. Syngas chemical looping gasification process: Oxygen carrier particle selection and performance. Energy Fuels 2009, 23, 4182–4189. (10) Ishida, M.; Takeshita, K.; Suzuki, K.; Ohba, T. Application of FeOAlO composite particles as solid looping material of the chemical-loop combustor. Energy Fuels 2005, 19, 2514–2518. (11) Jin, H.; Okamoto, T.; Ishida, M. Development of a novel chemicallooping combustion: Synthesis of a solid looping material of NiO/ NiAl2O4. Ind. Eng. Chem. Res. 1999, 38, 126–132. (12) Ishida, M.; Jin, H.; Okamoto, T. Kinetic behaviour of solid particle in chemical-looping combustion: Suppressing carbon deposition in reduction. Energy Fuels 1998, 12, 223–229. (13) Mattisson, T.; Johansson, M.; Lyngfelt, A. Multicycle reduction and oxidation of different types of iron oxide particles - Application to chemical-looping combustion. Energy Fuels 2004, 18, 628–637. (14) Ada´nez, J.; de Diego, L. F.; Garcı´a-Labiano, F.; Gaya´n, P; Abad, A. Selection of oxygen carriers for chemical-looping combustion. Energy Fuels 2004, 18, 371–377. (15) Galvita, V.; Schro¨der, T.; Munder, B.; Sundmacher, K. Production of hydrogen with low COx-content for PEM fuel cells by cyclic water gas shift reactor. Int. J. Hydrogen Energy 2008, 33, 1354–1360. (16) Zafar, Q.; Mattisson, T.; Gevert, B. Integrated hydrogen and power production with CO2 capture using chemical looping reforming - Redox reactivity of particles of CuO, Mn2O3, NiO, and Fe2O3 using SiO2 as a support. Ind. Eng. Chem. Res. 2005, 44, 3485–3496. (17) Abad, A.; Mattisson, T.; Lyngfelt, A.; Johansson, M. The use of iron oxide as oxygen carrier in a chemical-looping reactor. Fuel 2007, 86, 1021–1035. (18) Ward, D. A.; Ko, E. I. Preparing catalytic materials by the sol-gel method. Ind. Eng. Chem. Res. 1995, 34, 421–433. (19) Ertl, G.; Kno¨zinger, H.; Weitkamp, J. Preparation of solid catalysts; Wiley: New York, 1999. (20) Brinker, C. J.; Scherer, G. W. Sol-gel Science; Academic Press: New York, 1990. (21) Ishida, M.; Jin, H.; Okamoto, T. A fundamental study of a new kind of medium material for chemical-looping combustion. Energy Fuels 1996, 10, 958–963. (22) Zhao, H.; Liu, L.; Wang, B.; Xu, W.; Jiang, L.; Zheng, C. Solgel-derived NiO/ NiAl2O4 oxygen carriers for chemical-looping combustion by coal char. Energy Fuels 2008, 22, 898–905. (23) Yoldas, B. E. Alumina gels that form porous transparent Al2O3. J. Mat. Sci. 1975, 10, 1856–1860. (24) Brunauer, S.; Emmett, P. H.; Teller, E. Adsorption of gases in multimolecular layers. J. Am. Chem. Soc. 1938, 60, 309–319. (25) Barrett, E. P.; Joyner, L. G.; Halenda, P. H. The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms. J. Am. Chem. Soc. 1951, 73, 373–380. (26) Jaeger, J. C. Failure of rocks under tensile conditions. Int. J. Rock Mech. Min. Sci. 1967, 4, 219–227. (27) Kasprzyk-Horden, B. Chemistry of alumina, reactions in aqueous solution and its application in water treatment. AdV. Colloid Interface Sci. 2004, 110, 19–48. (28) Lippens, B. C.; Steggerda, J. J. Physical and chemical aspects of adsorbents and catalysts.: Academic Press: New York, 1970, 189. (29) Khalil, K. M. S.; Makhlouf, S. A. High surface area thermally stabilized porous iron oxide/silica nanocomposites via a formamide modified sol-gel process. Appl. Surf. Sci. 2008, 254, 3767–3773.

Ind. Eng. Chem. Res., Vol. 49, No. 11, 2010 (30) Zayat, M.; Levy, D. Surface area study of high area cobalt aluminate particles prepared by the sol-gel method. J. Sol-Gel Sci. Technol. 2002, 25, 201–206. (31) Isalksi, W. H. Separation of Gases; Clarendon Press: Oxford, 1989. (32) Powder Diffraction Files (PDF), JCPDS-International Centre for Diffraction Data, Pennsylvania, 2005. Fe3O4 (PDF 019-0629), FeO (PDF 006-0615), FeO · Al2O3 (PDF 034-0192), Al2O3 (PDF 010-0173). (33) ICSD Crystal Structure Database. Daresbury Laboratory. Fe3O4 (40093-ICSD), FeO (633038-ICSD), FeO · Al2O3 (260410-ICSD). (34) Topsøe, H.; Dumesic, J. A.; Boudart, M. Alumina as a textural promoter of iron synthetic ammonia catalysts. J. Catal. 1973, 28, 477– 488. (35) Galvita, V.; Sundmacher, K. Cyclic water gas shift reactor (CWGS) for carbon monoxide removal from hydrogen feed gas for PEM fuel cells. Chem. Eng. J. 2007, 134, 168–174.

5391

(36) Galvita, V.; Hempel, T.; Lorenz, H.; Rihko-Struckmann, L. K.; Sundmacher, K. Deactivation of modified iron oxide materials in cyclic water gas shift process for CO-free hydrogen production. Ind. Eng. Chem. Res. 2008, 47, 303–310. (37) Noorman, S.; van Sint Annaland, M.; Kuipers, H. Packed bed reactor technology for chemical-looping combustion. Ind. Eng. Chem. Res. 2007, 46, 4212–4220.

ReceiVed for reView January 8, 2010 ReVised manuscript receiVed March 25, 2010 Accepted April 3, 2010 IE100046F