OIL & GAS SCIENCE AND TECHNOLOGY - REVUE DE L'IFP

Characteristics of various new oxygen carriers for CLC Title in French Lagerbom, J.*; Moilanen, A; Kanerva U.; Koskinen, P.; Saastamoinen, J. Pikkarainen, T.; and Kauranen, P.; VTT Technical Research Centre of Finland, P.O. Box 1000, Espoo, FI-02044 VTT, Finland [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]

Abstract -- Characteristics of various new oxygen carriers for CLC – Various materials prepared especially for the use of chemical looping combustion (CLC) were characterised by thermogravimetry (TG) and scanning electron microscopy (SEM). The materials tested were three types of nickel and iron based oxygen carriers: 1) nickel based materials; 2) iron based materials and 3) nickel alloyed iron based material in an alumina matrix. In the nickel based oxygen carriers, the nickel was mixed with an alumina (Al2O3) matrix, which had nominal particle sizes of 1 µm, 3-4 µm and 10 µm, respectively. A 60 %Fe 2O 3/ 40% Al2O3 mixture and a 57% Fe2O3 + 3% NiO/ 40 % Al2O3 mixture were also tested. The particle size of the nickel based oxygen carrier’s alumina support materials was found to affect the reactivity of nickel based oxygen carriers in TG. Packing in granulation and sintering was affected by the alumina particle size. In the case of iron based oxygen carriers NiO addition increased the reactivity. In comparison of nickel based and iron based materials, nickel based materials were found to be more potential, as expected. The alloying of iron based material with NiO increased the reactivity and reaction rate but not enough to make iron oxide based oxygen carriers comparable to NiO based. Resume – In French.

1st International Conference on Chemical Looping, Lyon 2010

INTRODUCTION Combustion of fossil fuels releases a massive amount CO2 into the atmosphere. The Kyoto protocol sets international goals for reducing carbon dioxide (CO2) emissions. As energy cannot substantially be produced presently based only on renewable energy sources, CO2 emissions must be controlled otherwise. Capture of CO2 from the exhaust gas of fossil fuel combustion is an alternative considered. Some processes have been suggested for CO2 capture, it can be absorbed for example by amine solution absorption or scrubbing [1] or by membrane processes [2]. Fossil fuel combustion can be done in a way that air is not used directly to combust the fuel, while capture of CO 2 gets simpler and cheaper compared to the case of conventional air combustion in which nitrogen gas should be separated in order to CO2 emission. In such respect, oxy-fuel combustion using pure oxygen as an oxidiser instead of air [3] and chemical looping combustion (CLC) using solid metal oxides as an oxidiser [4] are similar. [5] CLC consists of separate air and fuel reactors that keep the nitrogen from air separate from the CO2 of exhaust. Basically only water vapour must be condensed from the gas to make carbon dioxide capture possible. This is accomplished by transferring the oxygen from an air reactor to a fuel reactor with solid metals oxides. Metal or lower oxidation state oxide is oxidised by air and subsequently reduced by the fuel. The metal/metal oxide oxygen carrier is circulated between the reactors without mixing the gases. A general description of the chemical looping can be given by Equations 1 and 2, with 1 being the oxidation in air reactor and 2 being the reduction in fuel reactor. [5, 6] Eq. 1. MyOx-1 + ½ O2 (air) MyOx + (air: N2 + unreacted O2) Eq. 2.

(2n+m)MyOx + C nH2m (2n+m)MyOx-1 + mH2O + nCO2

In order to make this process operational and economically viable some characteristics of the oxygen carrier must be met. First the oxygen carrier must be reactive enough both in air and in fuel gas, as measured by the reaction rate, oxygen carrying capacity and through reactivity of the oxygen carrier material. Second it must completely combust the fuel (conversion of the fuel). Oxygen carriers must also be fluidisable. Furthermore the oxygen carrier must be mechanically and stable against chemical cycling and mechanical agitation. They must resist agglomeration and sintering in process conditions. Also oxygen carriers must be envi-

ronmentally sound and the oxygen carrier must be cheap enough and long-lasting. Transition metals like Ni, Cu, Co, Fe and Mn oxides are thermodynamically possible materials for this use. Co, (Co 3O4 and CoO) Cu (CuO) and Ni (NiO) oxides have the highest oxygen carrier capacities [7]. In the case of Fe, only the reaction between magnetite and hematite (Fe2O3/Fe3O4) can be used, as the total reduction to Fe is too slow and full fuel conversion can not be gained. Co and Mn oxides tend decompose to unwanted oxides or volatile compounds in process environments. Cu oxide has the same tendency with slightly higher decomposition temperature but its melting point is considered too low and sintering may occur, causing defluidisation. High enough fuel conversion can be gained by Fe2O3/Fe3O4, NiO/Ni, Cu 2O/Cu and Mn 3O4/MnO, with NiO/Ni being slightly lower than the rest. Other materials than these have been judged to have insufficient conversion. NiO/Ni has the best reactivity amongst the tested oxygen carriers, as a large amount of oxygen can be carried with a fast reaction rate and nearly 100 % conversion [5]. On the other hand, NiO is a toxic material especially in fine powder form causing allergy, lung disease or even cancer. Dusting and risk of human exposure to NiO may not be avoided in CLC-facilities. Less harmful Fe 2O3 has a lower reactivity and low oxygen carrying capacity R0 (Equation 3) than most other oxygen carriers considered, especially NiO. [4]. R0 for Fe 2O3/Fe3O4 is 0.03 but for Ni/NiO it is 0.2. By small NiO alloying, the reactivity of of Fe2O3 may be increased catalytically [9]. [5, 9, 10] Eq. 3. R0 = (m ox - m red )/ m ox where m ox and m red are the masses of oxidised and reduced oxygen carrier. The basic concept of oxygen carriers includes fine precursor particles of active oxygen carrier and varying amount of inert material granulated to larger 80-500 µm granules. Usually active and inert powders are mixed together prior to granulation. Also impregnation of active material to the surfaces of inert granules has been tested [11]. The function of the inert part of the oxygen carrier is to give the granules mechanical stability, long time durability against sintering and even higher activity by maintaining the pore structure. Usually the inert part has a higher sintering temperature than the active part and it rules the sintering. As the inert component is not participating in the oxidation/reduction reactions, its volume must be considered as loss load which must be kept hot and circulating between reactors. It is intended that the inert part stays unreacted with the active part, thus maintaining the whole active inventory for carrying oxygen. On the other hand, it has been reported to be advantageous to allow some active material to react with the inert part, leaving the whole oxygen carrier more stable

and even more reactive. Examples of this are nickel aluminate NiAl3O4 and magnesium aluminate MgAl3O4 spinels. The active NiO or MgO can react with Al 2O3 and as they form the spinel, the portion of active metal can no longer carry oxygen as it is attached to the inert part. The spinels are more stable than pure Al2O3 and therefore oxygen carriers may be more resistant to chemical and thermal cycling. [5] Production method, particle size, granule density and other characteristics of the oxygen carrier may affect the reaction rate and therefore the total amount of the mass needed to keep the process temperature and electrical efficiency high enough. The most often reported production methods are freeze granulation and spray drying. Both methods produce spherical granules by nature. Due to flowability and granule fragmentation, a round shape is preferred. A large amount other usable granulation methods exist that can produce spherical granules, also in large quantities needed for CLC to be used economically in practice [12]. Achieving an active surface area, specific surface area or particle size can be approached from different directions. These granules should have near an optimal density, (active) surface area, particle size, pore volume and crushing strength. These factors are not independent factors. For example surface area, pore volume and crushing strength are connected. Usually higher density (higher sintering temperature) gives lower activity and higher strength. These factors must be optimised together. The crushing strength of oxygen carriers is commonly measured by the force needed to fracture the particles using a Shimpo FGN-5 crushing strength apparatus or by the ASTM D-4179 method. Usually only force is reported since every measurement uses the same area of a crushing tool (presumably 12 mm diameter, area 113 mm 2). Typical values are below 15 N, and often even below 10 N. It is advantageous to have a large surface area to increase reactivity and reaction rate. [10] Another test method is to crush the granules in a mould piston fixture using a materials testing machine. A shallow bed of tested powder is poured into the mould and pressed. The relative density against logarithmic pressure curve is analysed for sections of different slopes of the curve [13, 14]. High surface area powders with high activity are pursued. High surface area granules are however prone to sinter more easily in the reactor and their activity may decrease as a function of time. Vice versa, activity may increase if the active surface area is increasing by chemical cycling and subsequent fragmentation of the active part. However, this can cause dusting and loss of inventory to the small particle separation cyclone. The apparent density and crushing strength of particles increases as a function of an increased sintering temperature. Particles with a high density and

crushing strength are believed to be advantageous since such particles are assumed to be more capable of withstanding fragmentation and attrition in a chemical-looping combustor. [15] To get final proof of durability of an oxygen carrier, it must be subjected to actual environments in a fluidised bed. Crushing strength measurements certainly give a good estimate of the attrition strength although fracture mechanics differ. The fracture mechanism can be different due to deformation rate, compression vs. impact, grinding and shear stresses and due to temperature difference. The strength of sintered powder granules is mostly a function of bridge cross-section between particles and that increases the strength in every deformation form. Coke formation, carbonisation or carbon formation can occur in the fuel reactor by pyrolysis (Equation 4) or the Boudouard reaction (Equation 5). An endothermic pyrolysis reaction is favoured at high temperatures and exothermic Boudouard reaction at lower temperatures. Both reactions are not thermodynamically favoured in the absence of a catalyst. Ni and Fe of the oxygen carriers can act as catalyst, yet it is even more likely with Ni since reduction to metallic Fe is not applicable in CLC due to the slow rate of the reduction reactions, as mentioned earlier. In CLC, coke formation can limit the efficiency of the combustion by carrying carbon to the air reactor. The presence of water vapour may help to avoid carbon formation. Also higher process temperatures may help. [5] Eq. 4. CH4

C + 2H 2

at 700 ºC G -16.3 kJ/mol, at 700 ºC G -38.0 kJ/mol Eq. 5. 2CO C + CO2 at 700 ºC G -8.75 kJ/mol, at 900 ºC G 17.5 kJ/mol In this work some oxygen carrier alternatives were investigated in order to test the potential of Fe2O3 as an oxygen carrier material compared to NiO based materials. Also the effect of Al2O3 particle size in NiO based oxygen carriers was tested. Especially thermogravimetric studies were utilised to reveal the reactivity of the oxygen carriers. Post TG test SEM studies were carried out to reveal the reaction characteristics and stability against chemical cycling.

1 EXPERIMENTAL 1.1 Oxygen carrier powder preparation NiO based granules were produced by spray drying with polyvinyl alcohol (PVA) binder. Prior to spray drying Ni

and Al2O3 were mixed with a dispergator mixer (homogenisator) to ensure thorough mixing but still avoiding particle size decrease. The mixing was performed in water for one hour. PVA was added to the resulting slurry and finally it was granulated by spray drying. The Al2O3 matrix had nominal particle sizes of 1 µm, 3-4 µm and 10 µm, respectively. The Ni power particle size was 2.2 to 2.8 µm. The materials were sintered at two temperatures of 1175 °C and 1225 °C. The nickel contents of the samples were between 35.7 - 40.3 wt%. The original nominal Ni content was 40 wt% but some sepaoccurred during the spray drying and sintering due to different alumina particle sizes. The granule size of the Ni-based oxygen carriers was between 10 and 80 µm. Two iron based oxygen carriers had nominal compositions of Fe2O3/Al2O3 60 wt%/40 wt% and Fe 2O3/NiO/Al 2O3 57 wt%/3 wt%/40 wt%. The iron based samples were prepared by dry mixing with a drum ball mill. Granulation was done by a rotary blade mixer with water/polyvinyl acetate PVAC mixture as both the granulation agent and binder. Several steps of were repeated of the sequence: water-based organic binder addition - granulation – drying – sieving to 125–500 µm – crushing. The Fe-based samples were subsequently sintered at temperatures of 1100 °C, 1170 °C and 1200 °C to gain sufficient strength and then it was sieved to a finale granule size between 90-500 m. A finer sieve was used to compensate the sintering shrinkage. The iron content was 39.6 wt%. NiO addition was used in one oxygen carrier in order to increase the reactivity of fuels. The crushing strength of the iron based granules was measured by a materials testing machine with a 20 mm piston-cup tool /6, 7/. Table 1 presents the tested materials. TABLE 1 Tested oxygen carrier materials.

NiO/Al 2O3 1 µm

Metho d

wt% Active

Sintering T ºC

SD

30

1175 1225

NiO/ Al2O3 3-4 µm

SD

29

1175 1225

NiO/ Al2O3 10 µm

SD

31

1175 1225

Fe 2O3/ MG Al2O3 Fe 2O3 + MG NiO / Al2O3 SD Spray Drying

69 57 + 3

1170 and 1200 1170 and 1200

Granule size µm D10 D 90 12.9 59.6 9.8 63.9 5.7 60.4 8.8 67.9 12.9 67.3 16.9 67.3 90 50 90 500

MG Mixer Granulation

1.2 Thermogravimetry setup A thermobalance (TG, Mettler Toledo TGA/SDTA851) was used for testing. The gases were fed into the furnace through an automatic gas valve, in which the reducing and oxidising gases could be switched at the desired time interval. The characterisation tests were carried out at two temperatures: 850 °C and 900 °C. The gases were switched at 5 minute interval and 100 oxidation/reduction cycles were performed. The oxidising gas was air and the reducing gas was 4% CH4 in N2. After the TG-tests, the samples were examined by SEM (BSE) and EDS in order to reveal if the granules had eroded by the chemical cycling. Also the difference between the oxidised and reduced state of the granules was studied. Table 2 presents the TG-test parameters.

2 RESULTS AND DISSCUSSION 2.1 Powder granulation The tested oxygen carriers were characterised both before and after the TG tests in order to examine morphological changes such as fracturing during the thermal cycling. The structural deterioration only due to temperature cycling would be unacceptable. Mechanical agitation tests will be conducted later in a fluidised bed in order to simulate real process conditions. A spherical shape of the granules was targeted for the best flowability. Figures 1-6 present the morphology of tested powders before the tests. In Figure 1 NiO /1 µm Al2O3 had a typical granule shape of spray dried powder granules. The granule shape is mostly spherical although some non-spherical granules exist. The granule size was between 64 µm (d90) and 10 µm (d10). The NiO-bases oxygen carrier with 10 µm sized alumina did not form identically shaped granules due to too large of a alumina particle size. In spray drying the viscosity and other rheological characteristics of the slurry determine the droplet formation and subsequent particle size, shape and density. The smaller the primary particles are, the higher is the density inside the granulated particles. The Fe-based oxygen carriers granulated by rotary blade mixing had granule sizes clearly larger that the nickel based oxygen carriers. Iron based granules were sieved to 125 – 500 µm. Fe-based powders did not have as spherical shape as spray dried powders. This is due to the different droplet formation mechanism. In spray drying, slurries form by nature (surface tension) into spherical droplets. On the other hand, in the mixer granulation moist particles with binder as an adhesive are granulated by gathering the particles on each other. Using commercial granulators, spherical granules can

be formed by optimising the parameters. Here reactivity was prioritised over perfect shape. Especially in thermogravimetry, the exact shape does not noticeably affect the result as the tests are stationary. In the fluidization bed tests particle shape has a bigger effect. Figure 2 presents the morphology of Fe-based powder.

mixed the components together randomly. A much more stabile structure would be a skeleton of an inert component on which the active component could be coated. In this way the inert part would have a continuous structure with high strength and the active part would be optimally visible to air and fuel. To construct this kind of granules might be difficult and expensive.

Figure 2 SEM-picture of granulated Fe2O3 / Al2O3 powder prior to testing. Magnification

Figure 1 250x. SEM-picture of spray dried NiO /1 µm Al2O3 powder prior to testing. Magnification 500x. ___________________________________________ __ Figure 3 presents a cross-section of iron based oxygen carrier particle granules. It can be seen that the particles are macroscopically dense through granules, with only some voids or larger pores. Also the composition was found to be homogenous through the granules, mixing of the Fe2O3 and Al2O3 was successful and no separation occurred during granulation. Figure 4 presents the particles on the surface of the granules after sintering. During sintering the edges of the primary particles rounded and bridges formed between particles but no considerable grain growth was noticed. In spite of sintering the structure was still quite porous and it was considered that gas could penetrate the granules. In sintering of powders in any form, quite close green (unsintered) packing density is necessary in order to gain full density. Because of that, reasonably sized primary particles (0.5 to 4 µm) tend to result in powder particles with pore sizes that can not be sintered to their full density. For the best possible reactivity, it is thought to be beneficial for the oxygen carriers to have some porosity for gases to penetrate and also have quite a large surface area of the active particles. Together with reactivity (as a a thermodynamical quantity), the reactivity rate is also an important factor which is regulated, at least partly, by free surface area. Yet both methods used in this study

Figure 3 SEM-picture of granulated Fe2O3 / Al2O3 powder prior to testing. Cross-section, magnification 100x.

Figure 4 Figure 6 SEM-picture of granulated of Fe2O3 / Al2O3 powder prior to testing. Sintered at 1200 ºC. Magnification 5000x. ___________________________________________ __ Figures 5 and 6 present the morphology of the particle granules of NiO alloyed iron based oxygen carriers. Particle shape was similar to the unalloyed iron based version. Sintering of the alloyed version was not as successful as without NiO alloying. Even higher sintering temperatures were tested to increase granules strength but it was not possible to generate a clear The reason for this might be a tendency of NiO alloying to change the sintering behaviour of iron oxide. Similar dispersion and granulation methods were used but iron oxide alloyed with NiO was found to be in larger groups thus not helping the sintering of alumina. For some reason sintering of NiO alloyed powder also did not make the particles round as in the case of unalloyed Fematerial.

Figure 5 SEM-picture of granulated Fe2O3+NiO/ Al2O3 powder prior to testing. Magnification 250x.

SEM-picture of granulated of Fe2O3+NiO/ Al 2O3 powder prior to testing. Sintered at 1200 ºC. Magnification 10 000x. ___________________________________________ __ 2.2 Thermogravimetry According to the thermogravimetric (TG) results, the behaviour of the nickel samples with particle sizes of 1 and 3-4 um were similar in their behaviour but the samples having a bigger alumina particle size of 10 µm significantly slower. The reason for lower activity is supposed to be inadequate dispersion of nickel to too coarse alumina powder. The surface area of coarse alumina powder might not be sufficient to disperse the nickel fully and subsequently larger agglomerates of nickel react slower than small. At the higher test temperature of 900°C, the reduction rates (DTG curve) a little higher than those measured at 850°C. No significant difference in the oxidation rates was detected. In all of the tested oxygen carriers, the oxidation rate was faster than the reduction rate. The capacities of the samples were approximately 90-95% of the theoretical capacity. It can be seen in Figure 7 that a fully reduced sample starts to gain weight as the fuel flow is continued. Approximately 0.1 mg weight gain is observed. As the oxidising cycle is started the weight first drops quite quickly, indicating the burning of carbon. After the carbon was totally burnt of, the weight started to rise due to nickel oxidation. Weight change rate (DTG) is also indicated in Figure 7. The influence of the carbon was verified when the reduction was carried out using 3% hydrogen mixed with N2, where no weight gain was noticed after full reduction. DTA curves were also measured but the curves are not presented here. The DTA data verified the endothermic reaction of the carbon deposition and the exothermic reaction of carbon combustion. In the TG-tests the amount of fuel is high

compared to an actual CLC-boiler, in which the carbon deposition may not be a problem due to residual oxygen in the end of the reduction cycle. Also in Figure 7 it is shown that the reaction rate is slower at the start of the reduction and at the end of the oxidation. All tested samples behaved similarly. The reason for the slow of reduction might be the burning of entrapped air in the oxygen carrier granules. The slow end of oxidation might be due to growth of oxide scale on nickel, which can physically block the oxygen and require longer diffusion lengths.

rate at the beginning of reduction and at end of the oxidation. No insignificant decrease in reduction rate can be seen with either Ni-based or Fe-based oxygen carriers in each cycle. If the reaction rate would be noticeably a function of granule size, the reaction would be much faster at the beginning of the reduction and more so with larger Fe-based materials. The steady reaction rate indicates that the pore size and volume are sufficient to enable fuel transport inside the granules. A larger particle size was enforced to enable fluidised bed experiments with actual full size granulate size. A large enough spray dryer was not available to continue the experimental program at this time. Some differences were found when comparing Fe 2O3 based material to Fe 2O3 with 3% NiO alloying. A considerable increase in reactivity and reactivity rate was found as the NiO alloying was done. Figure 11 presents the comparison between Fe2O3 / Al2O3 and Fe2O3 +NiO/ Al2O3. Even if taking in account the increased reactivity due to alloying and erosion of particles, Ni-based materials still had a considerably higher than Fe-based materials. 2.3. Crushing strength

Figure 7 th

5 cycle of TG-curve at 850 ºC of NiO /1 µm Al2O3 left vertical axis , red curve indicates the weight change rate (DTG differential thermogravimetry) right vertical-axis.

Figure 8 present a 100 cycles TG-curve for a Fe–based material. TG-curves of NiO based materials with two sintering temperatures are compared in Figures 9 and 10. The main difference between the Fe-based materials and the Ni-based materials seems to be that the oxygen carrying capacity of the Fe-based materials was significantly smaller and the reaction rate, even slower than that of the nickel material. Moreover, the capacity increased with the number of cycles while in the Nibased materials, no changes in the capacity as a function of cycle number was detected. A considerable increase in weight change was noticed after 50 oxidation reduction cycles. Chemical cycling resistance of Fe2O3 / Fe 3O4 seems not to be ideal. Chemical engineering by alloying or by purity of the iron oxide might solve the problem. In Fe2O3 +NiO/ Al2O3 materials, the carbon deposition was also strong when tested with methane. With Ni-based materials, the reaction rate was also somewhat faster that with Fe-based materials. This can partly be explained by the smaller particle size of the spray dried Ni-based materials compared to blade granulated Fe-based materials. The reaction rate of Febased material seems to be steady except for the lower

The crushing strength of samples Fe2O3 / Al2O3 and Fe 2O3 +NiO/ Al 2O3 were tested. The strength of unalloyed Fe2O3 granules was found to be more than three times higher than 3% NiO alloyed granules. The measured crushing strengths were 3.5 MPa for NiO alloyed and 11.4 MPa for unalloyed Fe-based oxygen carrier. A clear increase in crushing strength was not noticed by raising the sintering temperature of the NiO alloyed version of Fe-based material. The Fe-based materials low chemical cycling resistance was noticed by TG-tests, in which oxidation/reduction behaviour was not stable as was the case with Ni-based materials. The NiO alloyed version of Fe-based material was more stable than unalloyed Fe-based material. One explanation for the lower strength and still better chemical cycling resistance can be the NiO alloyed Fe 2O3 particles’ tendency to sinter better and form larger groups, thus leaving larger pure alumina sections. The difference between the Fe 2O3 morphology of pure Fe 2O3 and the NiO alloyed version is presented in Figures 15 and 16. As alumina has a melting point of 2054 ºC and Fe 2O3 1565 ºC, iron oxide can behave as a bridge former between alumina particles if it is finely dispersed, yet it is not as effective in larger groups as in the NiO alloyed case. SEM pictures (Figures 11 and 12) reveal that NiO-based materials were reasonable intact after 100 cycle of the TG-test. Also Fe2O3/ Al2O3 and Fe2O3 +NiO/ Al2O3 were found to only slightly erode.

Cycle number 10

20

30

40

50

60 0.012

10.3

0.01

10.25

0.008

10.2

0.006

10.15

0.004

10.1

0.002

10.05 0

10000

20000

30000

40000

50000

Ro

Sample mass (mg)

0 10.35

0 70000

60000

Time (s) Figure 8. Sample mass during cyclic operation as function of time (line, left y-axis) and the oxygen carrier capacity as tion of cycle number for Fe 2O3/AlO3 (dots, right y-axis).

1

1.12

0.98

1.1

0.96

1.08

m /m 0

m /m 0

func-

0.94

1.06

0.92

1.04

0.9

1.02

0.88

1 0

20

40

60

80

100

0

Time (s)

20

40

60

80

Time (s)

Figure 9.

Figure 10.

Relative mass (mass/initial mass) of different o materials at TG-experiments at 900 C during th reduction in 5 cycle. Ni-based materials with different alumina size and sintering temperature o o ( 1 µm & 1175 C; 1 m & 1225 C; 3-4 o o m & 1175 C; 3-4 m & 1225 C; 10 m & 1175oC; 10 m & 1225oC).

Relative mass (mass/initial mass) of different o materials at TG-experiments at 900 C during th oxidation in 5 cycle. Ni-based materials with different alumina size and sintering temperature o o ( 1 µm & 1175 C; 1 m & 1225 C; 3-4 o o m & 1175 C; 3-4 m & 1225 C; 10 m & 1175oC; 10 m & 1225oC).

100

1

a

0.995

8 7

m/m 0

0% NiO

6

0.99

5 0.985

4

0.98

3

3% NiO

0.975

1

0.97 0

50

100

150

200

250

2 300

Time (s)

1.035

b

1

1.03

3

m /m 0

1.025

2 3 % NiO

4

Figure 12

1.02 1.015

6

5

1.01 0 % NiO 1.005 7

8

150

200

1 0

50

100

250

SEM (BSE) image of NiO/ 1 µm Al 2O3 after the 100 cycle TG-test, oxidised state, NiO white, magnification 1000x.

300

Time (s)

Figure 11. Relative mass (mass/initial mass) of Fe2O3 (5-8) and Fe2O3 with 3% NiO (1-4) in TG-experiments th o during 50 cycle at 850 (2, 3, 7, & 8) and 900 C (1, 4, 5 & 6) with different sintering temperatures o 1150 (5 & 8), 1170 (3 & 4) and 1200 C (1, 2, 6& 7) during reduction (a) and oxidation (b). ___________________________________________ __ 2.4 Morphology after TG-testing Figures 12-16 present SEM images of NiO /1 µm Al2O3, samples after 100 cycles. Figure 12 presents powder granules after chemical cycling. The spray dried spherical shape of the granules is intact after chemical cyFigure 13 presents the surface of Ni-based oxygen carrier in the oxidised state and Figure 14 in the reduced state. It can be seen that NiO (white phase) has a porous surface structure in the oxidised state but the surface is smooth due to carbonisation in the reduced It can be deduced that the activity of the oxygen carrier is decreased by the carbon until it is combusted away and the surface of the oxygen carrier is revealed. Adhesion of Ni/NiO to the alumina body appears to be good. The structure of the granules is porous which helps the reactant gases to pass through the oxygen carrier.

Figure 13. SEM (BSE) image of NiO/ 1 µm Al 2O3 after the 100 cycle TG-test, oxidised state, NiO white, magnification 10 000x.

Figure 14

Figure 16

SEM (BSE) image of NiO/ 1 µm Al 2O3 after the 100 cycle TG-test, reduced state, Ni/NiO white, magnification 10 000x.

SEM (BSE) image of Fe2O3+NiO/ Al2O3 after the 100 cycle TG-test, oxidised state, iron oxide white, magnification 1000x.

___________________________________________ __ Figures 15 and 16 present Fe 2O3 / Al2O3 and Fe 2O3 +NiO/ Al 2O3 samples after 100 TG-cycles. A large difference in the white Fe2O3 structure can be seen. A denser and more agglomerated structure of the NiO alloyed phase can be noticed. A higher increase in activity (compared to Figure 15) of the pure iron oxide version can be explained by the increase of surface area by chemical cycling. This indicates the possibility of major advances to be made by alloying the active part of the oxygen carrier. Also as mentioned earlier, sintering behaviour can be affected by alloying.

Figure 15 SEM (BSE) image of Fe2O3/ Al2O3 after the 100 cycle TG-test, oxidised state, iron oxide white, magnification 1000x.

___________________________________________ __ As spray drying is done using water based slurry, the starting powders can be dispersed better and the mixing of the active and inert parts can be homogeneous compared to the mixer granulation method. Furthermore, dispersing can be done in such a way that even some decrease in particle size can occur. The density and particle size determine the fluidization behavior of the oxygen carrier and the cyclone separation efficiency. Particle size also affects heat and mass transfer between gas and particle and inside the particle and subsequently the overall reaction rate of particles. According to literature suitable particle size range for oxygen carriers in CLC is from 0.08 to 2 mm [5]. Such large granules are hard to be produced using laboratory or pilot scale spray dryers. Mixer granulation was found to be a more convenient production method for test oxygen carriers, although shape and active material dispersion was somewhat affected. For larger quantities, a larger scale spray dryer or other kind of granulator should be applicable. For reactivity and gas permeation, the original particle size and sintering temperature probably play a greater role that the production method. It is possible to gain better reactivity by decreasing the original particle size but denser granules result simultaneously. Also affects the densification: higher sintering temperature and smaller original particle size may decrease the gas permeation. For the best possible result, using fine original particles with s reasonable sintering temperature (avoiding grain growth) is probably the best way but gas permeation pores might have to be generated by some kind of a pore former.

CONCLUSIONS In this research two different granulation methods were tested and compared. Spray drying was found to be more controllable and granule shape to be more spherical, i.e. suitable for fluidised beds. A Ni-based oxygen carrier has good oxygen carrying capacity, a stable structure with an Al2O3 body but carburisation can cause problems with methane fuel at least at the laboratory scale. Smaller NiO particles solve the slowness of the reaction in the end of the reduction cycle. Small NiO alloying seems to have a significant effect on the Fe-based oxygen carrier’s behaviour and characteristics. The strength is lower but the activity higher with the alloyed version. Erosion, or some other kind of fragmentation, increases the capacity of iron oxide based materials but not as much as with NiO alloying than without. In addition to the lower oxygen carrying capacity and lower reaction rate of Fe-based oxygen carriers, the tendency to fracture was noticed in TG-tests. Both types of materials suffer for carbon deposition in dry fuel. Yet materials sintered at reasonably low temperatures seem to have sufficient porosity to enable full conversion of oxygen carriers with a steady reaction rate. It seems possible to use even higher sintering temperatures without loosing full conversion. Alloying of iron oxide with, for example NiO, to modify its characteristics seems to be one possible development scheme for the future.

ACKNOWLEDGMENTS Tekes, The Finnish Funding Agency for Technology and Innovation, is acknowledged for the funding this work together with members from Finnish industry.

REFERENCES [1] Experience with CO2 capture from coal flue gas in pilot-scale: Testing of different amine solvents, Jacob N. Knudsen, Jørgen N. Jensen, Poul-Jacob Vilhelmsen, Ole Biede, Original Research Article Energy Procedia 1 (2009) 783-790. [2] A feasibility study of CO2 capture from flue gas by a facilitated transport membrane, Arshad Hussain, MayBritt Hägg, Department of Chemical Engineering, Norwegian University of Science and Technology, N-7491 Trondheim, Norway. [3] Maja B. Toftegaard, Jacob Brix, Peter A. Jensen, Peter Glarborg, Anker D. Jensen, Oxy-fuel combustion

of solid fuels, Progress in Energy and Combustion Science 3 (2010) 581-625. [4] /Richter H.J., Knoche K.F., Reversibility of combustion processes. ACS Symposium Series 235 (1983) 7185. [5] Hossain M. M., de Lasa H. L. Chemical-looping combustion (CLC) for inherent CO2 separations — a review, Chemical Engineering Science 63 (2008) 4433 – 4451. [6] Lyngfelt, A., Leckner, B, and Mattison, T. A fluidizedbed combustion process with inherent CO2 separation; application of chemical-looping combustion, Chemical Engineering Science 56 (2001) 3101-3113. [7] E. Jerndal, T. Mattisson, A. Lyngfelt, Thermal Analysis of Chemical-Looping Combustion, Chemical Engineering Research and Design 84 (2006) 795-806. [8] Rydèn M., Cleverstam E., Lyngfelt A., Mattisson T. Waste products from the steel industry with NiO as additive as oxygen carrier for chemical-looping combustion, International Journal of Greenhouse Gas Control 3 (2009) 693–703. [9] Johansson, E. et al A 300 W laboratory reactor system for chemical-looping combustion with particle circulation, Fuel 85 (2006) 1428-1438. [10] Johansson, M. Screening of oxygen-carrier partibased on iron-, manganese-, copper- and nickel oxides for use in chemical-looping technologies, Doctoral Thesis, Chalmers University of Technology, (2007) Göteborg, Sweden. [11] Hossain, M., Fluidized bed chemical-looping combustion: development of a bimetallic oxygen carrier and kinetic modelling, Ph.D. Dissertation, The University of Western Ontario, (2007) Ontario Canada. [12] Handbook of Powder Technology Vol 11 Granulation download, from: filefactory.com, size: 95 MB, date: 2009-06-14 . www.filestube.com/.../Handbook-ofPowder-Technology-Vol-11. [13] Reed J, S. Introduction to the principles of ceramic processing, first edition, USA, John Wiley & Sons Inc. (1988). [14] Song J. A die pressing test for the estimation of agglomerate strength, Journal of American Ceramic Society 77 (1994) 806-814. [15] Erik Jerndal, Tobias Mattisson, Ivo Thijs, Frans Snijkers, Anders Lyngfelt, Investigation of NiO/NiAl2O4 oxygen carriers for chemical-looping combustion pro-

duced by spray-drying, International Journal of Greenhouse Gas Control 4 (2010) 23–35.