Some Aspects of Calcium Looping Research at Imperial College, London J. Blamey, N.H. Florin, N. Paterson, D.R. Dugwell and P.S. Fennell* Department of Chemical Engineering, Imperial College, London *p.fennell@imperial.ac.uk 1st I.E.A. High Temperature Solid Looping Cycles Network Meeting INCAR, Oviedo, Spain, 17th September, 2009

Overview Introduction to calcium looping / using CaO-based sorbents for CO2 capture Current projects relating to calcium looping in our group Focus on a project related to hydration of spent sorbent Focus on a project related to co-precipitation of synthetic sorbents

Introduction to Calcium Looping Thank you to the previous speakers on the subject!

Current Projects Optimisation of reactivation strategies for exhausted sorbents for CO2, focusing on hydration [sponsored by EPSRC] Design of synthetic sorbents by co-precipitation in a slurry bubble column [sponsored by the Grantham Institute for Climate Change, Imperial College] Applications of the calcium looping cycle to cement manufacture [joint with industrial partner] Morphology changes of limestone sorbent particles during carbonation/calcination looping cycles in a TGA (also useful for sorbent enhanced H2 production) and reactivation with steam [joint project with CANMET, Canada, funded by KAUST] H2 production via sorbent-enhanced water-gas shift reactions [IC / KAUST] UK /China H2 production network (Imperial, Cambridge, Cranfield, Sheffield, Tsinghua, Taiyuan, NCEPU, TPRI, EPSRC funded) Many other projects in the field of CO2 capture, including amines, oxyfuel and chemical looping, as part of the Imperial College Centre for Carbon Capture and Storage (IC4S)

Reactivation of Sorbent using Hydration If a calcined sorbent is hydrated, upon calcination an improved uptake of CO2 can be observed Previous work has focused on TGA and low temperature fluidised bed environments Investigate reactivation of ‘spent’ sorbents using a small, bench scale, fluidised bed reactor more realistic conditions than previously studied potential to study attrition effects Sorbent reactivated [Fennell et al, 2007]

Design of Reactor Designed, built and tested for this project Small fluidised bed (ID = 21 mm) Resistance heated furnace Temperature range of up to 1000 °C at ambient pressure Capable of cycling between two temperatures to allow carbonation and calcination within same vessel Gas and fines vented to atmosphere

Experimental Work Creation of spent sorbent: Standard cycling experiments, 15% CO2, atmospheric pressure, 4.3 g Havelock limestone (500-710 µm) in 8 mL bed of sand (355-425 µm), flow rate ~ 8 U/Umf 13 cycles of carbonation for 900 s at 700 °C and calcination for 900 s with variation of calcination temperature. Tcalc = 840, 900, 950, 1000 °C Cycling experiments, varying calcination temperature Sorbent hydration Hydration: 38 hrs in a humid vessel at room temperature. Particles of Havelock limestone found to be fully hydrated under these conditions Further cycling experiments: Standard conditions, with constant Tcalc of 840 °C Further cycling experiments, constant calcination temperature Mass measurements: Sample is carefully weighed before and after each cycling experiment

Effect of Calcination Temperature Before Hydration Experimental data for Tcalc = 840 (◆), 900 (◼), 950 (▴) and 1000 (×) °C Fits of data to the Grasa Equation for the cycles before hydration of Tcalc = 840 (—) and 950 (---) °C After hydration, all limestones are cycled under the same conditions Error bars shown are one standard deviation on 5 experiments Tcalc = 840 °C after hydration Particles hydrated after 13th cycle Increasing Tcalc Conditions vary / conditions always the same Varied calcination temperature of first 13 cycles Increased sintering as you increase temperature, increasing reactivation After 13th cycle, sorbent is hydrated, after which sorbent is seen to be have be reactivated to variable extents As original calcination conditions become more severe, limestone is reactivated to a lesser extent. Highlight results of particles cycled at 950 and how reactivity slightly increased before decreasing to below that expected from Grasa’s equation when fitted to the first experiment PUT IN GRASA LINE FOR 840 C AS NEXT SLIDE… DRAW ON LINE MARKING OUT WHEN HYDRATION OCCURS MOVE LEGEND: LOSE BOX ALL GRAPHS CALL ‘EFFECT OF CALCINATION TEMPERATURE’

Mass Lost As Fines During 26 Cycles Note: Typically 10 % mass lost in 13 cycles before hydration Mass loss is calculated as mass loss of sample (calcined) as a percentage of theoretical maximum Error bars shown are one standard deviation on 5 experiments 95 % of sample lost for the sample calcined at 1000 degC. Only 30 % lost for that cycled at 840 degC. I say only, but this is in comparison with 10 % typically lost in first 13 cycles, rising to 12-15 % in across 26 cycles. CHANGE TITLE --> MASS LOST AS FINES AFTER 26 CYCLES

Carrying Capacity Normalised for Mass Lost When carrying capacity is normalised for mass loss during an experiment, similar deactivation curves are observed for each sample - experimental data shown for Tcalc = 840 (◆), 900 (◼) and 950 (▴) °C It is possible to conclude that: for particles large enough to remain within the bed, hydration has reactivated them to the same extent, independent of cycling conditions before hydration Point of interest: Reactivity of particles normalised for weight lost showed similar trends for all conditions studied--> suggests that hydrated particles which have the necessary size and lack of friability show similar properties. TGA & more regular mass measurements Similar for all curves --> suggests that reactivity of hydrated limestone is independent of original cycling.

Analytical Techniques Nitrogen adsorption analysis Investigation of nitrogen adsorption isotherms - varying pressure - yields information about surface area and porosity of a sample Pycnometry Skeletal (or absolute) density measured by helium displacement Measures density of particles excluding pores Envelope density measured by fine powder displacement Measures density of particles including pores From the above isotherm, the following can be calculated Brunauer-Emmett-Teller (BET) surface area Barrett-Joyner-Halenda (BJH) pore volume: yields estimates of pore size distribution 11

Gas Adsorption and Density Analysis Separate experiments were ran for gas adsorption and density analysis: all results are for calcined samples Tcalc = 840 (▴), 900 (◼), 950 (●) Pore volume associated with small pores decreases upon cycling and with increasing calcination temperature Pore volume increases after hydration, hence reactivation Densification of particles is observed upon cycling and to be greater for samples cycled at higher values of Tcalc [Manovic et al, 2009] Particles become considerably less dense after hydration: the least dense are the most friable After first calci-nation After cycling After hydra-tion 12

Optical Micrographs Before and After Hydration Examples of particles before (left column) and after (right column) hydration. Two particles cycled at 1113 K show small increase in size upon hydration. One particle cycled at 1273 K shows larger increase in size upon hydration and mild fragmentation, whereas the other shatters. Quartz cylinders (diameter 0.48 mm) used as reference distance.

Stress Distribution in the Growing Hydroxide Layer Balance on moles of Ca r2 Ca(OH)2 Initially CaO, subsequently Ca(OH)2 r0 r1 Oxide grows from the inner boundary Geometry

Stress Distribution in the Growing Hydroxide Layer (2)  increases with decreasing porosity of CaO, prior to hydration. relationship between conversion, stress in growing hydroxide layer and value  Shown left is the stress in the growing hydroxide layer as a function of conversion for three different initial envelope porosities of CaO The lower the initial porosity (i.e. the higher the temperature of calcination), the lower the conversion before a given stress level is reached: the more highly stressed the oxide; the greater the chance of fragmentation Ultimate tensile stress of concrete

Conclusions of Hydration Work Upon hydration, more highly sintered sorbents are reactivated to a lesser extent This is because of their increased friability upon hydration A model has been developed that establishes a link between the porosity of a particle (which decreases upon sintering) and the tensile strength The lower the initial porosity, the lower the conversion before a given stress level is reached

Design of Synthetic Sorbents by Co-precipitation in a SBC Objective: to create sorbents of high long-term reactivity and better mechanical stability For synthetic sorbents to compete with natural sorbents, it is necessary for the synthesis route to be straightforward and the reactants inexpensive Hence, precipitation in a Slurry Bubble Column (SBC) Literature shows that a highly reactive sorbent powder can be produced by a precipitation method using a SBC [e.g. Gupta and Fan, 2002] However, although precipitated sorbents show good short term reactivity, they are very prone to reactivity decay due to sintering [Florin and Harris, 2008] 17

Design of Synthetic Sorbents by Co-precipitation in a SBC Our current work focuses on achieving high long term reactivity and better mechanical strength Literature shows that a sorbent of high long-term reactivity can be achieved using inert supports, e.g. mayenite (Ca12Al14O33) and/or magnesium oxide (MgO) [e.g. Li et al., 2005, Pacciani et al., 2008] Can we use a SBC to co-precipitate calcium carbonate and aluminum hydroxide? Thermodynamic studies performed within the group suggest that yes, we can 18

Design of Synthetic Sorbents by Co-precipitation in a SBC: Exptl. Main hypothesis: Better distribution of CaO achievable via co-precipitation vs. wet mixing or hydrolysis methods Experiments performed in a SBC, varying: Ca(OH)2 loading and derivation amount of Al(NO)3.9H2O solution presence of dispersant (propan-2-ol) presence of inorganic additives carbonation initiated by bubbling through CO2 – precipitating CaCO3 seeded on finely dispersed Al(OH)3 Preliminary results indicate high capture capacity on a g-CO2/g- sorbent basis compared to Havelock; and method suited to using Ca2+ dissolved from natural limestone 19

Co-precipitation of CaCO3 and Al(OH)3 Figure 1 (left) Weight change associated with CO2 capture-and-release for PCC with 15 wt. % mayenite through 30 cycles – calcination 5 min at 900 °C, carbonation 10 min at 650 °C (15 % CO2); Figure 2 (right) Carrying capacity of four synthetic sorbents compared to Havelock

Optimisation and scale-up Figure 1 (left) slurry bubble column; Figure 2 (centre) preliminary extrusion experiment; Figure 3 (right) bench-scale FBR

Acknowledgements Engineering and Physical Sciences Research Council (EPSRC), UK King Abdullah University of Science and Technology (KAUST), Saudi Arabia National Resources, Canada The Grantham Institute for Climate Change, Imperial College Prof. Ben Anthony Prof. Rafael Kandiyoti Dr. Yinghai Wu Charles Dean Kelvin Okpoko

References Fennell, P.S., Davidson, J.F., Dennis, J.S., and Hayhurst, A.N., Regeneration of sintered limestone sorbents for the sequestration of CO2 from combustion and other systems. Journal of the Energy Institute, 2007. 80(2): p. 116-119. Florin, N.H. and Harris, A.T., Screening CaO-based sorbents for CO2 in biomass gasifiers. Energy and Fuels, 2008. 22(4): p. 2734-2742. Gupta, H. and Fan, L.S., Carbonation-calcination cycle using high reactivity calcium oxide for carbon dioxide separation from flue gas. Industrial & engineering chemistry research, 2002. 41(16): p. 4035-4042. Li, Z.-S., Cai, N.-S., Huang, Y.-Y., and Han, H.-J., Synthesis, experimental studies, and analysis of a new calcium-based carbon dioxide absorbent. Energy & Fuels, 2005. 19(4): p. 1447-1452. Manovic, V., Charland, J.-P., Blamey, J., Fennell, P.S., Lu, D.Y., and Anthony, E.J., Influence of calcination conditions on carrying capacity of CaO-based sorbent in CO2 looping cycles. Fuel, 2009. 88(10): p. 1893-1900. Pacciani, R., Muller, C.R., Davidson, J.F., Dennis, J.S., and Hayhurst, A.N., Synthetic Ca-based solid sorbents suitable for capturing CO2 in a fluidized bed. The Canadian journal of chemical engineering, 2008. 86(3): p. 356-366.