- Email: [email protected]
Between two worlds
Ind. Eng. Chem. Res. 2002, 41, 4035-4042
4035
Carbonation-Calcination Cycle Using High Reactivity Calcium Oxide for Carbon Dioxide Separation from Flue Gas Himanshu Gupta and Liang-S. Fan* Department of Chemical Engineering, 121 Koffolt Laboratories, The Ohio State University, 140 West 19th Avenue, Columbus, Ohio 43210
This study is focused on improving the reactivity of a CaO sorbent for its use in a reactionbased process for the separation of carbon dioxide (CO2) from flue gas. The separation process consists of cyclical carbonation (of a metal oxide) and calcination (of the metal carbonate formed) reactions to yield concentrated CO2 from flue gas. CaO sorbents synthesized from naturally occurring limestone and dolomite were microporous in nature. Pore filling and pore pluggage of these micropores limited the conversion of CaO in the carbonation reaction to about 45-50% of the stoichiometric limit. A wet precipitation process was tailored to synthesize high-surfacearea precipitated calcium carbonate (PCC). The pores of PCC predominantly lie in the mesoporous range (5-20 nm). The CaO sorbent obtained from PCC (PCC-CaO) was less susceptible to pore pluggage and attained over 90% conversion. PCC-CaO was also capable of maintaining its high reactivity (>90%) over two carbonation-calcination cycles. Introduction The increasing use of fossil fuels to meet energy needs has led to higher carbon dioxide (CO2) emissions into the atmosphere. It is widely accepted that, whereas the CO2 concentration was about 280 ppm before the industrial revolution, it increased from 315 ppmv in 1950 to 355 ppmv in 1990.1,2 Rising CO2 concentrations have been reported to account for half of the greenhouse effect that causes global warming.3 Although anthropogenic CO2 emissions are small compared to the amount of CO2 exchanged in natural cycles, the discrepancy between the long life of CO2 in the atmosphere (50-200 years) and the slow rate of natural CO2 sequestration processes leads to CO2 accumulation in the atmosphere. The IPCC (Intergovernmantal Panel on Climate Change) is of the opinion that “the balance of evidence suggests a discernible human influence on the global climate”.3 It is therefore necessary to develop cost-effective CO2 management schemes to curb its emission. Many CO2 management schemes consist of three parts: separation, transportation, and sequestration.4 The cost of CO2 separation and compression to 110 bar (for transportation) is estimated at $30-50 per ton of CO2, and transportation and sequestration are estimated to cost about $1-3 per ton per 100 km and $1-3 per ton of CO2, respectively.5 The capture of CO2 imposes severe energy penalties on fossil-fuel-based power plants, thereby reducing their net electricity output by as much as 13-37%.6 The monumental costs associated with the current CO2 separation technologies necessitate the development of economical alternatives. Historically, CO2 separation was motivated by enhanced oil recovery.7,8 Current industrial processes such as the calcination of limestone, synthesis of ammonia, and production of hydrogen involve the separation of CO2. Numerous CO2 separation processes are currently * To whom correspondence may be addressed. Telephone: (614)-292-7907. Fax: (614)-292-3769. E-mail: fan@ che.eng.ohio-state.edu.
being tested for their deployment in fossil-fuel-based power plants. Absorption processes employ physical and chemical solvents such as Selexol, Rectisol, MEA, and KS-2.9-11 Adsorption systems capture CO2 on a bed of adsorbent materials such as molecular sieves or activated carbon.12 CO2 can also be separated from other gases by condensing it out at cryogenic temperatures. Polymers, metals such as palladium, and molecular sieves are also being evaluated for membrane-based separation processes.9 Reaction-based processes can also be used for separating CO2 from flue gas. This process is based on the carbonation reaction in which gaseous CO2 reacts with solid metal oxide (represented by MO) to yield the metal carbonate (MCO3). The reaction can be represented by
MO + CO2 f MCO3
(1)
Once the metal oxide has reached its ultimate conversion, it can be thermally regenerated to the metal oxide and CO2 by heating the metal carbonate beyond the calcination temperature. The calcination reaction can be represented by
MCO3 f MO + CO2
(2)
An actual installation of this reaction-based CO2 separation process in a fossil-fuel-fired utility would consist of a carbonation reactor and a regeneration reactor. The choice of the metal oxide would depend on the range of carbonation and calcination temperatures, as wellas the reactivity of the particular metal oxide, which governs both the ultimate sorption capacity and the carbonation kinetics. Reaction-based CO2 separation processes provide the following advantages: (a) CO2 separation can be performed under flue gas conditions. The temperature of flue gas generated from fossil-fuel-fired combustion sources is in the 800-120 °C range, and its pressure is subatmospheric. However, low temperatures and/or high pressures are required to enhance the capture/separation efficiency of absorption, adsorption, membrane separation, and cryogenic
10.1021/ie010867l CCC: $22.00 © 2002 American Chemical Society Published on Web 07/11/2002
4036
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
distillation processes. In most thermal power plant configurations, the CO2 concentration is only about 5-30% of the flue gas, with the remainder being mostly nitrogen, oxygen, and moisture. Any process that depends on high system pressure and/or low temperature to increase sorbent/solvent utilization necessitates compression and/or cooling of the entire flue gas stream, which can be expensive. In contrast, reaction-based processes provide high sorbent utilization at flue gas temperatures and pressures. Cooling, compression, and liquefaction can then be economically applied to the relatively pure CO2 stream, which is a small fraction of the entire flue gas. (b) Materials typically used for reaction-based CO2 separation processes have high equilibrium capacities. The sorption capacity of the sorbent/solvent (quantified in units of grams of CO2 per kilogram of solvent/sorbent) dictates the material requirements. Under ideal conditions, MEA captures 60 g of CO2/kg, silica gel adsorbs 13.2 g of CO2/kg, and activated carbon adsorbs 88 g of CO2/kg.13-15 In contrast, a CaO-based process would capture 393 g of CO2/kg, assuming a 50% conversion of CaO over repeated cycles. Employment of a reactionbased process would lead to lower sorbent requirements, smaller reactor sizes, and lower pressure drop acrosss the reactor. (c) Reaction-based separation processes leave low concentrations of CO2 in the treated flue gas. The sorbent/solvent should have a low equilibrium CO2 concentration so that CO2 can be removed from flue gas to very low concentrations, if necessary. Naturally occurring rocks containing MgO and CaO react with atmospheric CO2 at concentrations in the 250-370 ppm range, even at ambient temperature. These oxides are capable of scavenging CO2 to very low concentrations. (d) Reaction-based separation processes can generate pure streams of CO2. Upon regeneration, it is desirable that the only gas coming off the sorbent/solvent is CO2. If other gases such as SO2 or moisture also evolve in the regeneration, then the separation process would not be effective, as it would merely concentrate the CO2 but fail to yield pure CO2. For instance, when CaCO3 is calcined (650-890 °C), CaSO4 does not decompose to any significant extent. Many metal oxides exhibit the carbonation and calcination reactions. Metal carbonates that decompose beyond 1000 °C are not suitable for this purpose because their calcination would impose severe energy penalties. The calcination temperatures of candidate metal carbonates (CaCO3, ∼750 °C; MgCO3, ∼385 °C; ZnCO3, ∼340 °C; PbCO3, ∼350 °C; CuCO3, ∼225-290 °C; and MnCO3, ∼440 °C) lie within the temperature range of interest (200-800 °C). Because the carbonation temperature is lower than the calcination temperature, carbonation can occur at flue gas temperatures for these metal oxides. Apart from CaO, gas-solid carbonation of other metal oxides has not been widely studied. Sawada et al.16 observed the conversion of ZnO to ZnCO3 at 8-13 °C when ZnO powder was exposed to CO2 and H2O for over 100 days. They observed a comparable pickup of water and CO2 by ZnO. Shaheen and Selim17 conducted thermal stability studies on basic CuCO3 and found the decomposition temperature of CuCO3 to be 225-250 °C. They observed that MnCO3 undergoes a more complex thermal degradation process. MnCO3 first decomposes to MnO2 at 300 °C, which in turn changes to Mn2O3 at
440 °C. At higher temperatures (∼900 °C), the final thermal decomposition product was found to be Mn3O4. Different oxides of manganese provide the flexibility of investigating the carbonation-calcination reaction over a wider temperature range. Aqueous-phase MgO carbonation has been studied for its suitability for mineralbased CO2 sequestration.18 Butt et al.19 conducted the carbonation of Mg(OH)2 to investigate its sequestration properties. They obtained about 10% carbonate formation between 387 and 400 °C and 6% carbonate formation between 475 and 500 °C. They attributed the low conversions to the formation of a nonporous carbonate product layer. This layer hinders the inward diffusion of CO2 and the outward diffusion of H2O (a product of the carbonation reaction), leading to low conversions. An activation energy of 72 kcal/mol was estimated for carbonation using the contracting-sphere model. Kato et al. studied the carbonation of PbO as part of a chemical heat pump process.20,21 They reported 30% conversion in 1 h under 100% CO2 atmosphere at 300 °C. Furthermore, they found the reactivity of PbO to drop with the number of carbonation-calcination cycles. The carbonation of calcium oxide has been widely studied. Related applications of CaO carbonation and calcination include the storage of energy,22 a zeroemission coal alliance process consisting of hydrogasification of coal fueled by the heat of the carbonation reaction,23 and the inorganic chemical heat pump.20,21 The gas-solid CaO-CO2 reaction proceeds through two rate-controlling regimes. The first regime involves a rapid, heterogeneous chemical reaction. In the second regime, the reaction slows because of the formation of an impervious layer of CaCO3. This product layer prevents the exposure of unreacted CaO in the particle core to CO2 for further carbonation. The kinetics of the second regime is governed by the diffusion of ions through the CaCO3 product layer. Bhatia and Perlmutter24 considered the formation of nucleation sites of carbonate as a separate rate-controlling step preceding the kinetically controlled regime. They estimated the activation energy to be 21 kcal/mol below 688 K and 43 kcal/mol above it for product-layer diffusion. They also suggested the countermigration of CO32- and O2- ions to explain diffusion through the product layer. Dedman and Owen25 mentioned two distinct mechanisms for the migration of reacting species through the CaCO3 layer depending on the Tamann temperature of CaCO3. Oakeson and Cutler26 also observed a two-step mechanism for the carbonation reaction carried out at 24.87 bar and 1044 °C. They calculated the diffusion activation energy to be 29 ( 6 kcal/mol. Mess et al.27 proposed grain-boundary diffusion and bulk diffusion to account for the diffusion of reacting species through the product layer. They calculated the activation energy to be 59 kcal/mol and suggested that bulk diffusion dominates the reaction rate at high temperatures and high conversions. The extent of the carbonation reaction reported in many studies has also shown considerable variation. Stoichiometrically, 56 g of CaO should react with 44 g of CO2 to form 100 g of CaCO3. This translates to about 0.786 g of CO2/g of CaO. However, structural limitations prevent the attainment of theoretical conversion. Dedman and Owen25 obtained a CO2 uptake of about 0.23 g of CO2/g of CaO (∼30% conversion) in 30 min at 600 °C. Bhatia and Perlmutter24 obtained approximately 70% conversion using 81-137-µm particles. Mess et al.27
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002 4037
reported 82% conversion at 1050 °C temperature and 11.74 atm CO2 pressure in 32 h using 15-20-µm particles. The limitation on total conversion stems essentially from the nature of the initial pore size distribution of the CaO sorbent. Microporous sorbents (pore size < 2 nm) are very susceptible to pore blockage and plugging through the formation of a higher-molarvolume product (molar volume of CaO ) 17 cm3/mol; molar volume of CaCO3 ) 37 cm3/mol). CaO sorbents obtained from naturally occurring precursors are usually microporous in nature. At the end of the kinetically controlled regime, diffusion processes through the product layer control the reaction rate. Similar structural limitations have prevented calcium-based sorbents from attaining theoretical conversions in the sulfation reaction as well.28 The results of this study suggested that a mesoporous structure that maximizes the porosity in the 5-20-nm pore size range would be less susceptible to pore pluggage. This structure would also be able to provide sufficient surface area to ensure rapid kinetics. The modified precipitation technique used in this study resulted in a mesoporous CaCO3 structure that also had a high BET surface area as determined by nitrogen (60 m2/g). A similar approach could also enhance the reactivity of CaO sorbents in the carbonation reaction, which is the focus of this study. Last, it is important that the CaO sorbents maintain their reactivity over many carbonation and calcination cycles. Barker22,29 observed that conversion of CaO dropped from about 73% in the first carbonation cycle to 43% at the end of the 5th cycle at 866 °C. He suggested that the CaCO3 layer is about 22-nm-thick, and his later work showed repeated 93% conversions over 30 cycles at 629 °C on 10-nm CaO particles. In another study, cyclical experiments conducted at a carbonation temperature of 880 °C and a calcination temperature of 860 °C led to a drop in conversion from 70% in the first carbonation to 38% in the 7th carbonation step.21 This study focuses on improving the reactivity of micron-sized CaO sorbents by altering their morphology and maintaining their reactivity over repeated cycles of carbonation and calcination. Experimental Section Chemicals, Sorbents, and Gases. Naturally occurring limestone (CaCO3) and hydrated lime [Ca(OH)2] synthesized from it were obtained from Linwood Mining and Minerals. Dolomite (CaCO3‚MgCO3) was procured from the National Dolomite Company. The purity of these ores was above 90%. High-purity metal oxides including ZnO, MgO, MnO2, NiO, CuO, PbO, and CaO were obtained from Aldrich Chemical Co. Precipitated calcium carbonate (PCC) was synthesized from Linwood hydrate by the procedure described in the next section. N40V dispersant, a sodium salt of a carboxylic acid, which was used in the synthesis of PCC, was obtained from Allied Colloid. N2 and CO2 used for calcination and carbonation experiments were 99.999% and 99.9% pure, respectively. Precipitated Calcium Carbonate (PCC) Synthesis. Structurally altered high-surface-area CaO precursors were synthesized using the procedure outlined by Agnihotri et al.30 A schematic diagram of the slurry bubble column used for this purpose is shown in Figure 1. The carbonator consisted of a 2-in.-o.d. Pyrex tube. A porous frit at the bottom provided a good distribution
Figure 1. Schematic diagram of a slurry bubble column used for the synthesis of precipitated calcium carbonate (PCC).
Figure 2. Change in the pH of the slurry as a function of Ca(OH)2 loading (500 mL water, 0.0575% N40V dispersant, 4 scfh CO2).
of CO2 through the slurry. A K-type thermocouple inserted in the slurry continuously recorded the slurry temperature. A pH probe monitored the pH of the slurry as the reaction medium changed from a basic to an acidic solution as the reaction proceeded. First, 500 mL of distilled water was poured into the carbonator, followed by 0.0575 g of N40V. Then, 12.8 g of Ca(OH)2 was added to the solution to provide a loading of 2.56 wt %. This corresponds to the concentration denoted as 16-sat [the concentration of Ca(OH)2 is 16 times its saturation solubility limit]. The solubility of Ca(OH)2 (∼0.16 g/100 g of water) leads to a pH of 12 at the start of the experiment. The remaining Ca(OH)2 remains suspended in the solution. The ratio of N40V to Ca(OH)2 was chosen to create a surface charge of zero on the incipiently formed CaCO3 particles.30 The flow of CO2 into the carbonator was then started, and the pH was continuously monitored. Figure 2 shows the change in pH with reaction time as a function of Ca(OH)2 loading. CO2 dissolved in water provides carbonate ions that react with Ca2+ ions to form CaCO3 according to the reaction
Ca2+ + CO32- f CaCO3
(3)
CaCO3 has a much lower solubility in water (∼0.0012
4038
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
g/100 g of water) than Ca(OH)2 and thus precipitates out. As the reaction proceeds, Ca2+ ions are consumed, but they are continuously replenished by the suspended Ca(OH)2. Hence, the pH remains at 12. As the reaction proceeds, Ca(OH)2 ultimately becomes depleted, and the concentration of Ca2+ ions cannot be maintained at its solubility limit. On the other hand, continued dissolution of CO2 gas leads to the accumulation of H+ ions, causing the solution to become acidic. Eventually, the pH settles at about 6.0, corresponding to the equilibrium solubility of CO2 in water at ambient temperature. This also signals the completion of the carbonation of all Ca(OH)2. The slurry is then removed from the precipitator, vacuum filtered, and stored in a vacuum oven at 90110 °C for 20 h to remove all remaining moisture. A higher Ca(OH)2 loading requires a longer reaction time, as is evident from Figure 2. Sorbent Reactivity Testing and Structural Analysis. The reactivity testing of CaO sorbents for carbonation was carried out in a Perkin-Elmer thermogravimetric analyzer (TGA-7) apparatus. The balance can measure accurately up to 1 µg. A small sample of the sorbent (5-20 mg) was placed in a quartz boat. The weight of the sample was recorded every second. The structural properties of CaO sorbents and their precursors were tested in a NOVA 2200 analyzer (Quantachrome Company). The BET surface area, pore volume, and pore size distribution were measured at -196 °C using nitrogen as the adsorbent. Results and Discussion Screening of Metal Oxides. Metal oxides such as ZnO, MgO, CuO, MnO2, NiO, PbO, and CaO that undergo the CCR scheme in the 800-200 °C temperature range were analyzed for their reactivity by a TGA. A powdered sample of these oxides was placed in a quartz pan, and pure CO2 was passed over the sample metal oxide. The temperature was then slowly raised, and the weight of the sample was continuously monitored. An increase in the weight of the sample is an indication of the formation of metal carbonate. However, no significant increase in the weight of these metal oxides was observed. In all cases except CaO, the change in weight was less than 0.2%. Hence, CaO was selected for further experimental analysis. Figure 3 profiles the change in weight percentage of CaO with increasing reaction temperature. CO2 flow was maintained over CaO throughout the reaction. The two regimes of control of this reaction, discussed in the Introduction section, are evident from this figure. A rapid increase in weight starts at about 225 °C corresponding to the kinetically controlled regime. The rapid reaction continues until fresh CaO surface is available for reaction. As soon as the available CaO surface is exhausted, further reaction can occur only by the diffusion of ionic species through the CaCO3 product layer. Beyond a weight increase of 50%, the rate of product-layer diffusion dictates the rate of the reaction. It is known that this regime is slower than the kinetically controlled regime, leading to an overall reduction in reaction rate. At about 900 °C, thermodynamics favors the CaO state, leading to calcination of the CaCO3 product and, consequently, to a drastic reduction in weight. The weight drop eventually tapers off when the sorbent is completely calcined to CaO. Preliminary insights into life-cycle performance were also obtained in the same experiment. When the weight stabilized, the temperature in the reactor was
Figure 3. Performance of CaO for the carbonation reaction in a TGA apparatus.
allowed to fall. As soon as the temperature decreased below about 900 °C, carbonation of CaO increased again. Physical Characterization of CaO Sorbents. XRD Analysis of CaO Obtained from Its Precursors. CaO was identified as a viable candidate for carbonationcalcination reactions. However, a variety of precursors can be calcined to obtain the CaO sorbents necessary for the carbonation reaction. Common and economical precursors include calcium carbonate, calcium hydroxide, and dolomite. Another important source of CaO is the calcination of synthesized high-surface-area precipitated calcium carbonate. To compare the crystal structure of the CaO sorbents obtained from these sources, XRD patterns were obtained for all of the CaO sorbents. Figure 4 shows these diffractograms. From this figure, we can conclude that the crystal structures of the CaO sorbents obtained from numerous sources are identical. Only the XRD pattern corresponding to dolomite-derived CaO shows extra peaks due to the presence of MgO in the calcined dolomite. Judging from the similarities in the CaO structures, it can be assumed that any difference in reactivity of CaO for carbonation is an artifact of the sorbent morphology and not a reflection of the chemistry of the gas-solid reaction that occurs on the CaO surface. Pore Structure of CaO Sorbents. CaO sorbents were synthesized by calcining various CaO precursors such as Linwood calcium carbonate (LC), dolomite (DL), Linwood calcium hydroxide (LH), and precipitated calcium carbonate (PCC). For convenience, the oxides derived from these sources are denoted LC-CaO, FCDCaO (for fully calcined dolomite-CaO), LH-CaO, and PCC-CaO, respectively. The procedure involved heating the precursor in flowing nitrogen beyond the calcination temperature (800-950 °C) for 1 h, followed by its storage in a desiccator. Structural properties such as surface area (SA) and pore volume (PV) of these chemicals are listed in Table 1, and their pore size distributions are shown in Figure 5. The SAs of the naturally occurring minerals, LC and dolomite, were very low, at 1.06 and 1.82 m2/g, respectively. LH was synthesized by first calcining the LC and then hydrating it. LH exhibited a considerably higher SA (13.19 m2/g) and PV compared to LC. The SA of PCC (38.3 m2/g), however, was the highest among all precur-
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002 4039
Figure 5. Morphological properties of CaO precursors.
Figure 4. XRD diffractograms of CaO derived from various precursors. Table 1. Morphological Propertiesa of Various CaO Sorbents and Their Precursors
a
sorbent name
surface area (m2/g)
pore volume (cm3/g)
LC LC-CaO dolomite FCD-CaO LH LH-CaO PCC PCC-CaO
1.1 17.8 1.8 29.8 13.2 33.3 38.3 12.8
0.003 0.078 0.004 0.08 0.0453 0.1 0.11 0.027
Surface area and pore volume.
sors. From Figure 5, we can infer that the structures of LC, DL, and LH are predominantly microporous in nature. Most of the porosity lies in pores below 5 nm in diameter. In contrast, the maximum in PV occurs at 15 nm for PCC, and most of its PV originates from mesopores in the 5-25-nm range. The structural properties of the CaO sorbents synthesized from these precursors also show an expected trend. The SAs and PVs of LC-CaO, FCD-CaO, and LH-CaO increased corresponding to their precursors. In contrast, the calcination of PCC provided an unexpectedly low SA of only 12.8 m2/g, probably as a reult of sintering under calcination conditions. Effect of the Ratio of Ca(OH)2 to Dispersant on PCC Morphology. Precipitated calcium carbonate can be obtained by the reaction between carbonate and calcium ions in solution. It is known that the CaCO3 nuclei that precipitate out have a positive surface charge that prevents agglomeration.30 The resulting structure is also microporous in nature. However, the structural properties of the synthesized PCC can be altered by the use of negatively charged dispersants that neutralize the surface charge. This makes the ratio between the Ca(OH)2 loading and the dispersant used very critical. In
Figure 6. Effect of Ca(OH)2 loading on the morphology of PCC (500 mL water, 0.0575% N40V dispersant, 4 scfh CO2). Table 2. Morphological Properties of PCC as a Function of Ca(OH)2 Loadinga
a
Ca(OH)2 loading (wt %)
surface area (m2/g)
pore volume (cm3/g)
1.28 2.56 3.84
13.8 38.3 36.8
0.03 0.14 0.11
500 mL of water, 0.0575% N40V dispersant, 4 scfh CO2.
addition, the effect of the Ca(OH)2 loading in the slurry was studied to enhance the productivity of the precipitation process by synthesizing more PCC from the same slurry volume. Ca(OH)2 loading levels of 8-sat, 16-sat, and 24-sat were used, with all other factors remaining constant. It can be seen from Figure 6 and Table 2 that, at a concentration of 8-sat, there is proportionally more dispersant in the slurry, causing the incipiently formed CaCO3 particles to be negatively charged. The negative charge prevents the agglomeration of these nuclei,
4040
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
Figure 7. Conversion of four CaO sorbents under pure CO2 at 650 °C.
eventually leading to the formation of microporous PCC, as shown in Figure 6. The surface area in this case is also relatively lower. At a Ca(OH)2 loading corresponding to 16-sat, the ratio of N40V to CaCO3 is balanced, and the surface charge on the nuclei is zero. This allows for an optimal association of these nuclei, leading to a predominantly mesoporous structure. The SA of PCC under these optimum conditions is also the highest at 38.3 m2/g. As the loading of Ca(OH)2 is raised to 24-sat, there is not enough N40V dispersant to neutralize the surface charge on all of the incipiently formed nuclei. There could possibly be some positively charged particles. This again creates nonoptimum conditions, leading to a loss in SA and PV compared to the 16-sat case. Another experiment was conducted to process a 32sat Ca(OH)2 slurry keeping the Ca(OH)2-to-N40V ratio constant. The SA and PV of PCC synthesized from a 32-sat slurry were 37.07 m2/g and 0.139 cm3/g respectively, lending support to the idea that a higher mass of PCC can be synthesized from the same amount of slurry. Carbonation of CaO Sorbents. The performance of these four CaO sorbents was tested by TGA. The experimental procedure consisted of placing 6-12 mg of the chosen CaO sorbent in a thin layer in a quartz pan to minimize external mass-transfer resistances. The sorbent was then heated in flowing nitrogen (5.0 grade, 99.999% pure) to the desired temperature. The representative temperatures used in these experiments were 550, 600, and 650 °C. Once the desired temperature had been reached, the flow was switched to 100% CO2. The increase in weight with time was recorded, and the conversion of CaO to CaCO3 was calculated from the increase in weight. Only the data obtained at 650 °C are reported here. The performance of the four CaO sorbents LC-CaO, FCD-CaO, LH-CaO, and PCC-CaO at 650 °C is depicted in Figure 7. Initially, CO2 diffuses into the pores of LCCaO, and the reaction takes place on the CaO surface provided by the pores. The figure shows that there is a rapid increase in weight in the first 1-2 min. The
Figure 8. Effect of temperature on the carbonation of PCC-CaO.
conversion attained in this kinetically controlled regime depends on the initial surface area of the CaO sorbent. LC-CaO and FCD-CaO attained 40-45% conversion, whereas LH-CaO and PCC-CaO attained about 60 and 54% conversion, respectively, in this regime. After this regime, conversion increases relatively slowly with time. The increase in conversion is only about 2-4% in the next hour for LC-CaO and FCD-CaO. This confirms the susceptibility of micropores to pore filling and pore pluggage described earlier resulting from the formation of a higher-volume product, CaCO3. The trend is not as dramatic for the case of LH-CaO because of its relatively higher initial surface area. The conversion for LH-CaO increases by another 18% in the diffusion-controlled regime. However, the increase in conversion for PCC-CaO is about 34-36% more in the second regime. Because the PCC-CaO structure is mesoporous, the building CaCO3 product layer is not able to plug all of the pore mouths. This, in turn, allows the heterogeneous reaction to occur on a larger CaO surface. Once the kinetically controlled regime is over, diffusion of ions occurs through a larger area, ultimately leading to a higher conversion of 88-90% for PCC-CaO. The high conversion attained by PCC-CaO is more remarkable considering the fact that its SA and PV were the lowest among the four CaO sorbents chosen. An optimized calcination procedure for PCC is under investigation to increase the SA and PV of PCC-CaO to further enhance the rate of its conversion. Figure 8 shows the effect of temperature on the carbonation of PCC-CaO. It can be seen that the extent of conversion in the kinetic regime is different at different temperatures. However, unlike LC-CaO, the conversion at any temperature does not seem to taper off, and given sufficient time, PCC-CaO is capable of attaining 90% or higher conversion at all of these temperatures. Cyclic Calcination and Carbonation. One of the possible hurdles in the utilization of metal oxides for the carbonation and calcination reaction scheme is their vulnerability to sintering from the thermal cycling imposed by the cyclical nature of these reactions. Cyclical studies were carried out to quantify any loss
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002 4041
adverse effect of sintering on the reactivity of the CaO sorbents at 700 °C. Acknowledgment The authors acknowledge the help extended by Dr. Rajeev Agnihotri, Dr. Raja Jadhav, and Mr. Puneet Gupta during experiments. The authors are also grateful to the Ohio Coal Development Office of the Ohio Department of Development for providing financial support for this project. Literature Cited
Figure 9. Carbonation-calcination cycles on Aldrich CaCO3 and PCC at 700 °C.
in reactivity of these sorbents upon multiple cycles. The temperature chosen for cyclical studies was 700 °C. This temperature is sufficient to achieve carbonation in the presence of pure CO2 and also to calcine the CaCO3 thus formed after the gas is switched from CO2 to N2. A variety of precursors were first calcined in nitrogen at 700 °C. The gas was then switched to pure CO2, and the weight gain was continuously tracked. After the ultimate conversion had been reached, the gas was switched back to N2. This process was repeated for 2-3 cycles. The data obtained for Aldrich CaCO3 and PCC undergoing this cyclical study are shown in Figure 9. It can be seen that the reactivity of Aldrich CaCO3 exhibited a gradual decrease even after the first cycle. In contrast, PCC completely regained its mass after the first calcination and carbonation cycle. At 700 °C, we can deduce that the conversion is almost complete (>95%). This figure also shows that the reactivity also did not decrease in the second cycle. Under the reaction conditions chosen, any sintering that might have occurred did not seem to adversely affect the sorbent morphology. This finding provides support for the use of CaO in this carbonation-calcination reaction-based CO2 separation process. Conclusions Preliminary screening of metal oxides for their application in the carbonation and calcination reaction scheme identified CaO to be a viable candidate. Experimental data were obtained on the carbonation of a variety of CaO sorbents between 550 and 700 °C. The two regimes of control of the carbonation reaction were verified. It was observed that CaO sorbents from natural sources failed to achieve complete conversion because of their microporous structure. The synthesis procedure for precipitated calcium carbonate was modified to obtain mesoporous CaO sorbents that achieved high (>90%) conversions in the carbonation reaction. This enhancement was attributed to the lower susceptibility of mesoporous sorbents to pore filling and pluggage. Life-cycle studies based on 2-3 cycles did not show an
(1) Cooper, C. D.; Alley, F. C. Air Pollution Control: A Design Approach, 2nd ed.; Waveland Press: Prospect Heights, IL, 1994. (2) Keeling, C. D.; Bacastow, R. B.; Whorf, T. P.; Mook, W. G. Evidence for Accelerated Releases of Carbon Dioxide to the Atmosphere Inferred from Direct Measurements of 13C/12C Ratio. In Proceedings of the 82nd Annual Meeting of the Air and Waste Management Association; Air and Waste Management Association: Pittsburgh, PA, 1989; Vol. 7, Paper 89-119.2. (3) Climate Change 1995: The Science of Climate Change, Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change; Houghton, J. T., Meira Filho, L.G., Callander, B.A., Harris, N., Kattenberg, A., Maskell K, Eds.; Cambridge University Press: Cambridge, U.K., 1996. (4) FETC Carbon Sequestration R&D Program Plan: FY 19992000; National Energy Technology Laboratory, U.S. Department of Energy: Washington, DC, 1999; www.fetc.doe.gov. (5) Wallace, D. Capture and Storage of CO2. What Needs To Be Done. Presented at the 6th Conference of the Parties, COP 6, to the United Nations Framework Convention on Climate Change; The Hague, The Netherlands, Nov 13-24, 2000; www.iea.org/ envissu/index.htm. (6) Herzog, H.; Drake, E.; Adams, E. CO2 Capture, Reuse, and Storage Technologies for Mitigating Global Climate Change. A White Paper; Final Report No. DE-AF22-96PC01257; U.S. Department of Energy: Washington, DC, Jan 1997. (7) Kaplan, L. J. Cost-Saving Processes Recovers CO2 from Power-Plant Fluegas. Chem. Eng. 1982, 89 (24), 30-31. (8) Pauley, C. P.; Smiskey, P. L.; Haigh, S. N-ReN Recovers CO2 from Flue Gas Economically. Oil Gas J. 1984, 82 (20), 8792. (9) Reimer, P.; Audus, H.; Smith, A. Carbon Dioxide Capture from Power Stations, IEA Greenhouse R&D Programme: Cheltenham, Gloucester, U.K., 2001; (www.ieagreen.org.uk). (10) Blauwhoff, P. M. M.; Versteeg, G. F.; van Swaaij, W. P. M. A study on the reaction between CO2 and alkanoamines in aqueous solution. Chem. Eng. Sci. 1984, 39 (2), 207-225. (11) Mimura, T.; Simayoshi, H.; Suda, T.; Iijima, M.; Mitsuake, S. Development of Energy Saving Technology for Flue Gas Carbon Dioxide Recovery by Chemical Absorption Method and Steam System in Power Plant. Energy Convers. Manage. 1997, 38 Suppl. P, S57-S62. (12) Kikkinides, E. S.; Yang, R. T.; Cho, S. H. Concentration and Recovery of CO2 from flue gas by pressure swing adsorption. Ind. Eng. Chem. Res. 1993, 32, 2714-2720. (13) Song, J.-H.; Yoon, J.-H.; Lee, H. Solubility of Carbon Dioxide in Monoethanolamine + Ethylene Glycol + Water and Monoethanolamine + Poly(ethylene glycol) + Water. J. Chem. Eng. Data 1996, 41, 497-499. (14) Zhang, Z.; Guan, J.; Ye, Z. Separation of Nitrogen-Carbon Dioxide Mixture by Rapid Pressure Swing Adsorption. Adsorption 1998, 4, 173-177. (15) Heuchel, M.; Davies, G. M.; Buss, E.; Seaton, N. A. Adsorption of Carbon Dioxide and Methane and Their Mixtures on an Activated Carbon: Simulation and Experiment. Langmuir 1999, 15, 8695-8705. (16) Sawada, Y.; Murakami, M.; Nishide, T. Thermal analysis of basic zinc carbonate. Part 1. Carbonation process of zinc oxide powders at 8 and 13 °C. Thermochim. Acta 1996, 273, 95-102. (17) Shaheen, W. M.; Selim, M. M. Effect of thermal treatment on physicochemical properties of pure and mixed manganese carbonate and basic copper carbonate. Thermochim. Acta. 1998, 322 (2), 117-128.
4042
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
(18) Fernandez, A. I.; Chimenos, J. M.; Segarra, M.; Fernandez, M. A.; Espiell, F. Kinetic study of carbonation of MgO slurries. Hydrometallurgy 1999, 53, 155-167. (19) Butt, D. P.; Lackner, K. S.; Wendt, C. H.; Conzone, S. D.; Kung, H.; Lu, Y.-C.; Bremser, J. K. Kinetics of Thermal Dehydroxylation and Carbonation of Magnesium Hydroxide. J. Am. Ceram. Soc. 1996, 79 (7), 1892-1898. (20) Kato, Y.; Saku, D.; Harada, N.; Yoshizawa, Y. Utilization of High Temperature Heat from Nuclear Reactor using Inorganic Chemical Heat Pump. Prog. Nucl. Energy 1998, 32 (3-4), 563570. (21) Kato, Y.; Harada, N.; Yoshizawa, Y. Kinetic feasibility of a chemical heat pump for heat utilization from high-temperature processes. Appl. Therm. Eng. 1999, 19, 239-254. (22) Barker, R. The Reversibility of the Reaction CaCO3 ) CaO + CO2. J. Appl. Chem. Biotechnol. 1973, 23, 733-742. (23) Tinkler, M. J.; Cheh, C. Towards a Coal-Capable Solid Oxide Fuel Cell System. In Proceedings of the 26th International Technical Conference on Coal Utilization and Fuel Systems; Coal Technology Association: Gaithersburg, MD, 2001; pp 569-570. (24) Bhatia, S. K.; Perlmutter, D. D. Effect of the Product Layer on the Kinetics of the CO2-Lime Reaction. AIChE J. 1983, 29 (1), 79-86. (25) Dedman, A. J.; Owen, A. J. Calcium Cyanamide Synthesis,
Part 4. The reaction CaO + CO2 ) CaCO3. Trans. Faraday Soc. 1962, 58, 2027-2035. (26) Oakeson, W. G.; Cutler, I. B. Effect of CO2 Pressure on the Reaction with CaO. J. Am. Ceram. Soc. 1979, 62 (11-12), 556558. (27) Mess, D.; Sarofim, A. F.; Longwell, J. P. Product Layer Diffusion during the Reaction of Calcium Oxide with Carbon Dioxide. Energy Fuels 1999, 13, 999-1005. (28) Wei, S.-H.; Mahuli, S. K.; Agnihotri, R.; Fan, L.-S. High Surface Area Calcium Carbonate: Pore Structural Properties and Sulfation Characteristics. Ind. Eng. Chem. Res. 1997, 36 (6), 21412148. (29) Barker, R. The Reactivity of Calcium Oxide Towards Carbon Dioxide and Its Use for Energy Storage. J. Appl. Chem. Biotechnol. 1974, 24, 221-227. (30) Agnihotri, R.; Chauk, S.; Mahuli, S.; Fan, L.-S. Influence of Surface Modifiers on the Structure of Precipitated Calcium Carbonate. Ind. Eng. Chem. Res. 1999, 38, 2283-2291.
Received for review October 23, 2001 Revised manuscript received April 12, 2002 Accepted April 15, 2002 IE010867L
4035
Carbonation-Calcination Cycle Using High Reactivity Calcium Oxide for Carbon Dioxide Separation from Flue Gas Himanshu Gupta and Liang-S. Fan* Department of Chemical Engineering, 121 Koffolt Laboratories, The Ohio State University, 140 West 19th Avenue, Columbus, Ohio 43210
This study is focused on improving the reactivity of a CaO sorbent for its use in a reactionbased process for the separation of carbon dioxide (CO2) from flue gas. The separation process consists of cyclical carbonation (of a metal oxide) and calcination (of the metal carbonate formed) reactions to yield concentrated CO2 from flue gas. CaO sorbents synthesized from naturally occurring limestone and dolomite were microporous in nature. Pore filling and pore pluggage of these micropores limited the conversion of CaO in the carbonation reaction to about 45-50% of the stoichiometric limit. A wet precipitation process was tailored to synthesize high-surfacearea precipitated calcium carbonate (PCC). The pores of PCC predominantly lie in the mesoporous range (5-20 nm). The CaO sorbent obtained from PCC (PCC-CaO) was less susceptible to pore pluggage and attained over 90% conversion. PCC-CaO was also capable of maintaining its high reactivity (>90%) over two carbonation-calcination cycles. Introduction The increasing use of fossil fuels to meet energy needs has led to higher carbon dioxide (CO2) emissions into the atmosphere. It is widely accepted that, whereas the CO2 concentration was about 280 ppm before the industrial revolution, it increased from 315 ppmv in 1950 to 355 ppmv in 1990.1,2 Rising CO2 concentrations have been reported to account for half of the greenhouse effect that causes global warming.3 Although anthropogenic CO2 emissions are small compared to the amount of CO2 exchanged in natural cycles, the discrepancy between the long life of CO2 in the atmosphere (50-200 years) and the slow rate of natural CO2 sequestration processes leads to CO2 accumulation in the atmosphere. The IPCC (Intergovernmantal Panel on Climate Change) is of the opinion that “the balance of evidence suggests a discernible human influence on the global climate”.3 It is therefore necessary to develop cost-effective CO2 management schemes to curb its emission. Many CO2 management schemes consist of three parts: separation, transportation, and sequestration.4 The cost of CO2 separation and compression to 110 bar (for transportation) is estimated at $30-50 per ton of CO2, and transportation and sequestration are estimated to cost about $1-3 per ton per 100 km and $1-3 per ton of CO2, respectively.5 The capture of CO2 imposes severe energy penalties on fossil-fuel-based power plants, thereby reducing their net electricity output by as much as 13-37%.6 The monumental costs associated with the current CO2 separation technologies necessitate the development of economical alternatives. Historically, CO2 separation was motivated by enhanced oil recovery.7,8 Current industrial processes such as the calcination of limestone, synthesis of ammonia, and production of hydrogen involve the separation of CO2. Numerous CO2 separation processes are currently * To whom correspondence may be addressed. Telephone: (614)-292-7907. Fax: (614)-292-3769. E-mail: fan@ che.eng.ohio-state.edu.
being tested for their deployment in fossil-fuel-based power plants. Absorption processes employ physical and chemical solvents such as Selexol, Rectisol, MEA, and KS-2.9-11 Adsorption systems capture CO2 on a bed of adsorbent materials such as molecular sieves or activated carbon.12 CO2 can also be separated from other gases by condensing it out at cryogenic temperatures. Polymers, metals such as palladium, and molecular sieves are also being evaluated for membrane-based separation processes.9 Reaction-based processes can also be used for separating CO2 from flue gas. This process is based on the carbonation reaction in which gaseous CO2 reacts with solid metal oxide (represented by MO) to yield the metal carbonate (MCO3). The reaction can be represented by
MO + CO2 f MCO3
(1)
Once the metal oxide has reached its ultimate conversion, it can be thermally regenerated to the metal oxide and CO2 by heating the metal carbonate beyond the calcination temperature. The calcination reaction can be represented by
MCO3 f MO + CO2
(2)
An actual installation of this reaction-based CO2 separation process in a fossil-fuel-fired utility would consist of a carbonation reactor and a regeneration reactor. The choice of the metal oxide would depend on the range of carbonation and calcination temperatures, as wellas the reactivity of the particular metal oxide, which governs both the ultimate sorption capacity and the carbonation kinetics. Reaction-based CO2 separation processes provide the following advantages: (a) CO2 separation can be performed under flue gas conditions. The temperature of flue gas generated from fossil-fuel-fired combustion sources is in the 800-120 °C range, and its pressure is subatmospheric. However, low temperatures and/or high pressures are required to enhance the capture/separation efficiency of absorption, adsorption, membrane separation, and cryogenic
10.1021/ie010867l CCC: $22.00 © 2002 American Chemical Society Published on Web 07/11/2002
4036
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
distillation processes. In most thermal power plant configurations, the CO2 concentration is only about 5-30% of the flue gas, with the remainder being mostly nitrogen, oxygen, and moisture. Any process that depends on high system pressure and/or low temperature to increase sorbent/solvent utilization necessitates compression and/or cooling of the entire flue gas stream, which can be expensive. In contrast, reaction-based processes provide high sorbent utilization at flue gas temperatures and pressures. Cooling, compression, and liquefaction can then be economically applied to the relatively pure CO2 stream, which is a small fraction of the entire flue gas. (b) Materials typically used for reaction-based CO2 separation processes have high equilibrium capacities. The sorption capacity of the sorbent/solvent (quantified in units of grams of CO2 per kilogram of solvent/sorbent) dictates the material requirements. Under ideal conditions, MEA captures 60 g of CO2/kg, silica gel adsorbs 13.2 g of CO2/kg, and activated carbon adsorbs 88 g of CO2/kg.13-15 In contrast, a CaO-based process would capture 393 g of CO2/kg, assuming a 50% conversion of CaO over repeated cycles. Employment of a reactionbased process would lead to lower sorbent requirements, smaller reactor sizes, and lower pressure drop acrosss the reactor. (c) Reaction-based separation processes leave low concentrations of CO2 in the treated flue gas. The sorbent/solvent should have a low equilibrium CO2 concentration so that CO2 can be removed from flue gas to very low concentrations, if necessary. Naturally occurring rocks containing MgO and CaO react with atmospheric CO2 at concentrations in the 250-370 ppm range, even at ambient temperature. These oxides are capable of scavenging CO2 to very low concentrations. (d) Reaction-based separation processes can generate pure streams of CO2. Upon regeneration, it is desirable that the only gas coming off the sorbent/solvent is CO2. If other gases such as SO2 or moisture also evolve in the regeneration, then the separation process would not be effective, as it would merely concentrate the CO2 but fail to yield pure CO2. For instance, when CaCO3 is calcined (650-890 °C), CaSO4 does not decompose to any significant extent. Many metal oxides exhibit the carbonation and calcination reactions. Metal carbonates that decompose beyond 1000 °C are not suitable for this purpose because their calcination would impose severe energy penalties. The calcination temperatures of candidate metal carbonates (CaCO3, ∼750 °C; MgCO3, ∼385 °C; ZnCO3, ∼340 °C; PbCO3, ∼350 °C; CuCO3, ∼225-290 °C; and MnCO3, ∼440 °C) lie within the temperature range of interest (200-800 °C). Because the carbonation temperature is lower than the calcination temperature, carbonation can occur at flue gas temperatures for these metal oxides. Apart from CaO, gas-solid carbonation of other metal oxides has not been widely studied. Sawada et al.16 observed the conversion of ZnO to ZnCO3 at 8-13 °C when ZnO powder was exposed to CO2 and H2O for over 100 days. They observed a comparable pickup of water and CO2 by ZnO. Shaheen and Selim17 conducted thermal stability studies on basic CuCO3 and found the decomposition temperature of CuCO3 to be 225-250 °C. They observed that MnCO3 undergoes a more complex thermal degradation process. MnCO3 first decomposes to MnO2 at 300 °C, which in turn changes to Mn2O3 at
440 °C. At higher temperatures (∼900 °C), the final thermal decomposition product was found to be Mn3O4. Different oxides of manganese provide the flexibility of investigating the carbonation-calcination reaction over a wider temperature range. Aqueous-phase MgO carbonation has been studied for its suitability for mineralbased CO2 sequestration.18 Butt et al.19 conducted the carbonation of Mg(OH)2 to investigate its sequestration properties. They obtained about 10% carbonate formation between 387 and 400 °C and 6% carbonate formation between 475 and 500 °C. They attributed the low conversions to the formation of a nonporous carbonate product layer. This layer hinders the inward diffusion of CO2 and the outward diffusion of H2O (a product of the carbonation reaction), leading to low conversions. An activation energy of 72 kcal/mol was estimated for carbonation using the contracting-sphere model. Kato et al. studied the carbonation of PbO as part of a chemical heat pump process.20,21 They reported 30% conversion in 1 h under 100% CO2 atmosphere at 300 °C. Furthermore, they found the reactivity of PbO to drop with the number of carbonation-calcination cycles. The carbonation of calcium oxide has been widely studied. Related applications of CaO carbonation and calcination include the storage of energy,22 a zeroemission coal alliance process consisting of hydrogasification of coal fueled by the heat of the carbonation reaction,23 and the inorganic chemical heat pump.20,21 The gas-solid CaO-CO2 reaction proceeds through two rate-controlling regimes. The first regime involves a rapid, heterogeneous chemical reaction. In the second regime, the reaction slows because of the formation of an impervious layer of CaCO3. This product layer prevents the exposure of unreacted CaO in the particle core to CO2 for further carbonation. The kinetics of the second regime is governed by the diffusion of ions through the CaCO3 product layer. Bhatia and Perlmutter24 considered the formation of nucleation sites of carbonate as a separate rate-controlling step preceding the kinetically controlled regime. They estimated the activation energy to be 21 kcal/mol below 688 K and 43 kcal/mol above it for product-layer diffusion. They also suggested the countermigration of CO32- and O2- ions to explain diffusion through the product layer. Dedman and Owen25 mentioned two distinct mechanisms for the migration of reacting species through the CaCO3 layer depending on the Tamann temperature of CaCO3. Oakeson and Cutler26 also observed a two-step mechanism for the carbonation reaction carried out at 24.87 bar and 1044 °C. They calculated the diffusion activation energy to be 29 ( 6 kcal/mol. Mess et al.27 proposed grain-boundary diffusion and bulk diffusion to account for the diffusion of reacting species through the product layer. They calculated the activation energy to be 59 kcal/mol and suggested that bulk diffusion dominates the reaction rate at high temperatures and high conversions. The extent of the carbonation reaction reported in many studies has also shown considerable variation. Stoichiometrically, 56 g of CaO should react with 44 g of CO2 to form 100 g of CaCO3. This translates to about 0.786 g of CO2/g of CaO. However, structural limitations prevent the attainment of theoretical conversion. Dedman and Owen25 obtained a CO2 uptake of about 0.23 g of CO2/g of CaO (∼30% conversion) in 30 min at 600 °C. Bhatia and Perlmutter24 obtained approximately 70% conversion using 81-137-µm particles. Mess et al.27
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002 4037
reported 82% conversion at 1050 °C temperature and 11.74 atm CO2 pressure in 32 h using 15-20-µm particles. The limitation on total conversion stems essentially from the nature of the initial pore size distribution of the CaO sorbent. Microporous sorbents (pore size < 2 nm) are very susceptible to pore blockage and plugging through the formation of a higher-molarvolume product (molar volume of CaO ) 17 cm3/mol; molar volume of CaCO3 ) 37 cm3/mol). CaO sorbents obtained from naturally occurring precursors are usually microporous in nature. At the end of the kinetically controlled regime, diffusion processes through the product layer control the reaction rate. Similar structural limitations have prevented calcium-based sorbents from attaining theoretical conversions in the sulfation reaction as well.28 The results of this study suggested that a mesoporous structure that maximizes the porosity in the 5-20-nm pore size range would be less susceptible to pore pluggage. This structure would also be able to provide sufficient surface area to ensure rapid kinetics. The modified precipitation technique used in this study resulted in a mesoporous CaCO3 structure that also had a high BET surface area as determined by nitrogen (60 m2/g). A similar approach could also enhance the reactivity of CaO sorbents in the carbonation reaction, which is the focus of this study. Last, it is important that the CaO sorbents maintain their reactivity over many carbonation and calcination cycles. Barker22,29 observed that conversion of CaO dropped from about 73% in the first carbonation cycle to 43% at the end of the 5th cycle at 866 °C. He suggested that the CaCO3 layer is about 22-nm-thick, and his later work showed repeated 93% conversions over 30 cycles at 629 °C on 10-nm CaO particles. In another study, cyclical experiments conducted at a carbonation temperature of 880 °C and a calcination temperature of 860 °C led to a drop in conversion from 70% in the first carbonation to 38% in the 7th carbonation step.21 This study focuses on improving the reactivity of micron-sized CaO sorbents by altering their morphology and maintaining their reactivity over repeated cycles of carbonation and calcination. Experimental Section Chemicals, Sorbents, and Gases. Naturally occurring limestone (CaCO3) and hydrated lime [Ca(OH)2] synthesized from it were obtained from Linwood Mining and Minerals. Dolomite (CaCO3‚MgCO3) was procured from the National Dolomite Company. The purity of these ores was above 90%. High-purity metal oxides including ZnO, MgO, MnO2, NiO, CuO, PbO, and CaO were obtained from Aldrich Chemical Co. Precipitated calcium carbonate (PCC) was synthesized from Linwood hydrate by the procedure described in the next section. N40V dispersant, a sodium salt of a carboxylic acid, which was used in the synthesis of PCC, was obtained from Allied Colloid. N2 and CO2 used for calcination and carbonation experiments were 99.999% and 99.9% pure, respectively. Precipitated Calcium Carbonate (PCC) Synthesis. Structurally altered high-surface-area CaO precursors were synthesized using the procedure outlined by Agnihotri et al.30 A schematic diagram of the slurry bubble column used for this purpose is shown in Figure 1. The carbonator consisted of a 2-in.-o.d. Pyrex tube. A porous frit at the bottom provided a good distribution
Figure 1. Schematic diagram of a slurry bubble column used for the synthesis of precipitated calcium carbonate (PCC).
Figure 2. Change in the pH of the slurry as a function of Ca(OH)2 loading (500 mL water, 0.0575% N40V dispersant, 4 scfh CO2).
of CO2 through the slurry. A K-type thermocouple inserted in the slurry continuously recorded the slurry temperature. A pH probe monitored the pH of the slurry as the reaction medium changed from a basic to an acidic solution as the reaction proceeded. First, 500 mL of distilled water was poured into the carbonator, followed by 0.0575 g of N40V. Then, 12.8 g of Ca(OH)2 was added to the solution to provide a loading of 2.56 wt %. This corresponds to the concentration denoted as 16-sat [the concentration of Ca(OH)2 is 16 times its saturation solubility limit]. The solubility of Ca(OH)2 (∼0.16 g/100 g of water) leads to a pH of 12 at the start of the experiment. The remaining Ca(OH)2 remains suspended in the solution. The ratio of N40V to Ca(OH)2 was chosen to create a surface charge of zero on the incipiently formed CaCO3 particles.30 The flow of CO2 into the carbonator was then started, and the pH was continuously monitored. Figure 2 shows the change in pH with reaction time as a function of Ca(OH)2 loading. CO2 dissolved in water provides carbonate ions that react with Ca2+ ions to form CaCO3 according to the reaction
Ca2+ + CO32- f CaCO3
(3)
CaCO3 has a much lower solubility in water (∼0.0012
4038
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
g/100 g of water) than Ca(OH)2 and thus precipitates out. As the reaction proceeds, Ca2+ ions are consumed, but they are continuously replenished by the suspended Ca(OH)2. Hence, the pH remains at 12. As the reaction proceeds, Ca(OH)2 ultimately becomes depleted, and the concentration of Ca2+ ions cannot be maintained at its solubility limit. On the other hand, continued dissolution of CO2 gas leads to the accumulation of H+ ions, causing the solution to become acidic. Eventually, the pH settles at about 6.0, corresponding to the equilibrium solubility of CO2 in water at ambient temperature. This also signals the completion of the carbonation of all Ca(OH)2. The slurry is then removed from the precipitator, vacuum filtered, and stored in a vacuum oven at 90110 °C for 20 h to remove all remaining moisture. A higher Ca(OH)2 loading requires a longer reaction time, as is evident from Figure 2. Sorbent Reactivity Testing and Structural Analysis. The reactivity testing of CaO sorbents for carbonation was carried out in a Perkin-Elmer thermogravimetric analyzer (TGA-7) apparatus. The balance can measure accurately up to 1 µg. A small sample of the sorbent (5-20 mg) was placed in a quartz boat. The weight of the sample was recorded every second. The structural properties of CaO sorbents and their precursors were tested in a NOVA 2200 analyzer (Quantachrome Company). The BET surface area, pore volume, and pore size distribution were measured at -196 °C using nitrogen as the adsorbent. Results and Discussion Screening of Metal Oxides. Metal oxides such as ZnO, MgO, CuO, MnO2, NiO, PbO, and CaO that undergo the CCR scheme in the 800-200 °C temperature range were analyzed for their reactivity by a TGA. A powdered sample of these oxides was placed in a quartz pan, and pure CO2 was passed over the sample metal oxide. The temperature was then slowly raised, and the weight of the sample was continuously monitored. An increase in the weight of the sample is an indication of the formation of metal carbonate. However, no significant increase in the weight of these metal oxides was observed. In all cases except CaO, the change in weight was less than 0.2%. Hence, CaO was selected for further experimental analysis. Figure 3 profiles the change in weight percentage of CaO with increasing reaction temperature. CO2 flow was maintained over CaO throughout the reaction. The two regimes of control of this reaction, discussed in the Introduction section, are evident from this figure. A rapid increase in weight starts at about 225 °C corresponding to the kinetically controlled regime. The rapid reaction continues until fresh CaO surface is available for reaction. As soon as the available CaO surface is exhausted, further reaction can occur only by the diffusion of ionic species through the CaCO3 product layer. Beyond a weight increase of 50%, the rate of product-layer diffusion dictates the rate of the reaction. It is known that this regime is slower than the kinetically controlled regime, leading to an overall reduction in reaction rate. At about 900 °C, thermodynamics favors the CaO state, leading to calcination of the CaCO3 product and, consequently, to a drastic reduction in weight. The weight drop eventually tapers off when the sorbent is completely calcined to CaO. Preliminary insights into life-cycle performance were also obtained in the same experiment. When the weight stabilized, the temperature in the reactor was
Figure 3. Performance of CaO for the carbonation reaction in a TGA apparatus.
allowed to fall. As soon as the temperature decreased below about 900 °C, carbonation of CaO increased again. Physical Characterization of CaO Sorbents. XRD Analysis of CaO Obtained from Its Precursors. CaO was identified as a viable candidate for carbonationcalcination reactions. However, a variety of precursors can be calcined to obtain the CaO sorbents necessary for the carbonation reaction. Common and economical precursors include calcium carbonate, calcium hydroxide, and dolomite. Another important source of CaO is the calcination of synthesized high-surface-area precipitated calcium carbonate. To compare the crystal structure of the CaO sorbents obtained from these sources, XRD patterns were obtained for all of the CaO sorbents. Figure 4 shows these diffractograms. From this figure, we can conclude that the crystal structures of the CaO sorbents obtained from numerous sources are identical. Only the XRD pattern corresponding to dolomite-derived CaO shows extra peaks due to the presence of MgO in the calcined dolomite. Judging from the similarities in the CaO structures, it can be assumed that any difference in reactivity of CaO for carbonation is an artifact of the sorbent morphology and not a reflection of the chemistry of the gas-solid reaction that occurs on the CaO surface. Pore Structure of CaO Sorbents. CaO sorbents were synthesized by calcining various CaO precursors such as Linwood calcium carbonate (LC), dolomite (DL), Linwood calcium hydroxide (LH), and precipitated calcium carbonate (PCC). For convenience, the oxides derived from these sources are denoted LC-CaO, FCDCaO (for fully calcined dolomite-CaO), LH-CaO, and PCC-CaO, respectively. The procedure involved heating the precursor in flowing nitrogen beyond the calcination temperature (800-950 °C) for 1 h, followed by its storage in a desiccator. Structural properties such as surface area (SA) and pore volume (PV) of these chemicals are listed in Table 1, and their pore size distributions are shown in Figure 5. The SAs of the naturally occurring minerals, LC and dolomite, were very low, at 1.06 and 1.82 m2/g, respectively. LH was synthesized by first calcining the LC and then hydrating it. LH exhibited a considerably higher SA (13.19 m2/g) and PV compared to LC. The SA of PCC (38.3 m2/g), however, was the highest among all precur-
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002 4039
Figure 5. Morphological properties of CaO precursors.
Figure 4. XRD diffractograms of CaO derived from various precursors. Table 1. Morphological Propertiesa of Various CaO Sorbents and Their Precursors
a
sorbent name
surface area (m2/g)
pore volume (cm3/g)
LC LC-CaO dolomite FCD-CaO LH LH-CaO PCC PCC-CaO
1.1 17.8 1.8 29.8 13.2 33.3 38.3 12.8
0.003 0.078 0.004 0.08 0.0453 0.1 0.11 0.027
Surface area and pore volume.
sors. From Figure 5, we can infer that the structures of LC, DL, and LH are predominantly microporous in nature. Most of the porosity lies in pores below 5 nm in diameter. In contrast, the maximum in PV occurs at 15 nm for PCC, and most of its PV originates from mesopores in the 5-25-nm range. The structural properties of the CaO sorbents synthesized from these precursors also show an expected trend. The SAs and PVs of LC-CaO, FCD-CaO, and LH-CaO increased corresponding to their precursors. In contrast, the calcination of PCC provided an unexpectedly low SA of only 12.8 m2/g, probably as a reult of sintering under calcination conditions. Effect of the Ratio of Ca(OH)2 to Dispersant on PCC Morphology. Precipitated calcium carbonate can be obtained by the reaction between carbonate and calcium ions in solution. It is known that the CaCO3 nuclei that precipitate out have a positive surface charge that prevents agglomeration.30 The resulting structure is also microporous in nature. However, the structural properties of the synthesized PCC can be altered by the use of negatively charged dispersants that neutralize the surface charge. This makes the ratio between the Ca(OH)2 loading and the dispersant used very critical. In
Figure 6. Effect of Ca(OH)2 loading on the morphology of PCC (500 mL water, 0.0575% N40V dispersant, 4 scfh CO2). Table 2. Morphological Properties of PCC as a Function of Ca(OH)2 Loadinga
a
Ca(OH)2 loading (wt %)
surface area (m2/g)
pore volume (cm3/g)
1.28 2.56 3.84
13.8 38.3 36.8
0.03 0.14 0.11
500 mL of water, 0.0575% N40V dispersant, 4 scfh CO2.
addition, the effect of the Ca(OH)2 loading in the slurry was studied to enhance the productivity of the precipitation process by synthesizing more PCC from the same slurry volume. Ca(OH)2 loading levels of 8-sat, 16-sat, and 24-sat were used, with all other factors remaining constant. It can be seen from Figure 6 and Table 2 that, at a concentration of 8-sat, there is proportionally more dispersant in the slurry, causing the incipiently formed CaCO3 particles to be negatively charged. The negative charge prevents the agglomeration of these nuclei,
4040
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
Figure 7. Conversion of four CaO sorbents under pure CO2 at 650 °C.
eventually leading to the formation of microporous PCC, as shown in Figure 6. The surface area in this case is also relatively lower. At a Ca(OH)2 loading corresponding to 16-sat, the ratio of N40V to CaCO3 is balanced, and the surface charge on the nuclei is zero. This allows for an optimal association of these nuclei, leading to a predominantly mesoporous structure. The SA of PCC under these optimum conditions is also the highest at 38.3 m2/g. As the loading of Ca(OH)2 is raised to 24-sat, there is not enough N40V dispersant to neutralize the surface charge on all of the incipiently formed nuclei. There could possibly be some positively charged particles. This again creates nonoptimum conditions, leading to a loss in SA and PV compared to the 16-sat case. Another experiment was conducted to process a 32sat Ca(OH)2 slurry keeping the Ca(OH)2-to-N40V ratio constant. The SA and PV of PCC synthesized from a 32-sat slurry were 37.07 m2/g and 0.139 cm3/g respectively, lending support to the idea that a higher mass of PCC can be synthesized from the same amount of slurry. Carbonation of CaO Sorbents. The performance of these four CaO sorbents was tested by TGA. The experimental procedure consisted of placing 6-12 mg of the chosen CaO sorbent in a thin layer in a quartz pan to minimize external mass-transfer resistances. The sorbent was then heated in flowing nitrogen (5.0 grade, 99.999% pure) to the desired temperature. The representative temperatures used in these experiments were 550, 600, and 650 °C. Once the desired temperature had been reached, the flow was switched to 100% CO2. The increase in weight with time was recorded, and the conversion of CaO to CaCO3 was calculated from the increase in weight. Only the data obtained at 650 °C are reported here. The performance of the four CaO sorbents LC-CaO, FCD-CaO, LH-CaO, and PCC-CaO at 650 °C is depicted in Figure 7. Initially, CO2 diffuses into the pores of LCCaO, and the reaction takes place on the CaO surface provided by the pores. The figure shows that there is a rapid increase in weight in the first 1-2 min. The
Figure 8. Effect of temperature on the carbonation of PCC-CaO.
conversion attained in this kinetically controlled regime depends on the initial surface area of the CaO sorbent. LC-CaO and FCD-CaO attained 40-45% conversion, whereas LH-CaO and PCC-CaO attained about 60 and 54% conversion, respectively, in this regime. After this regime, conversion increases relatively slowly with time. The increase in conversion is only about 2-4% in the next hour for LC-CaO and FCD-CaO. This confirms the susceptibility of micropores to pore filling and pore pluggage described earlier resulting from the formation of a higher-volume product, CaCO3. The trend is not as dramatic for the case of LH-CaO because of its relatively higher initial surface area. The conversion for LH-CaO increases by another 18% in the diffusion-controlled regime. However, the increase in conversion for PCC-CaO is about 34-36% more in the second regime. Because the PCC-CaO structure is mesoporous, the building CaCO3 product layer is not able to plug all of the pore mouths. This, in turn, allows the heterogeneous reaction to occur on a larger CaO surface. Once the kinetically controlled regime is over, diffusion of ions occurs through a larger area, ultimately leading to a higher conversion of 88-90% for PCC-CaO. The high conversion attained by PCC-CaO is more remarkable considering the fact that its SA and PV were the lowest among the four CaO sorbents chosen. An optimized calcination procedure for PCC is under investigation to increase the SA and PV of PCC-CaO to further enhance the rate of its conversion. Figure 8 shows the effect of temperature on the carbonation of PCC-CaO. It can be seen that the extent of conversion in the kinetic regime is different at different temperatures. However, unlike LC-CaO, the conversion at any temperature does not seem to taper off, and given sufficient time, PCC-CaO is capable of attaining 90% or higher conversion at all of these temperatures. Cyclic Calcination and Carbonation. One of the possible hurdles in the utilization of metal oxides for the carbonation and calcination reaction scheme is their vulnerability to sintering from the thermal cycling imposed by the cyclical nature of these reactions. Cyclical studies were carried out to quantify any loss
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002 4041
adverse effect of sintering on the reactivity of the CaO sorbents at 700 °C. Acknowledgment The authors acknowledge the help extended by Dr. Rajeev Agnihotri, Dr. Raja Jadhav, and Mr. Puneet Gupta during experiments. The authors are also grateful to the Ohio Coal Development Office of the Ohio Department of Development for providing financial support for this project. Literature Cited
Figure 9. Carbonation-calcination cycles on Aldrich CaCO3 and PCC at 700 °C.
in reactivity of these sorbents upon multiple cycles. The temperature chosen for cyclical studies was 700 °C. This temperature is sufficient to achieve carbonation in the presence of pure CO2 and also to calcine the CaCO3 thus formed after the gas is switched from CO2 to N2. A variety of precursors were first calcined in nitrogen at 700 °C. The gas was then switched to pure CO2, and the weight gain was continuously tracked. After the ultimate conversion had been reached, the gas was switched back to N2. This process was repeated for 2-3 cycles. The data obtained for Aldrich CaCO3 and PCC undergoing this cyclical study are shown in Figure 9. It can be seen that the reactivity of Aldrich CaCO3 exhibited a gradual decrease even after the first cycle. In contrast, PCC completely regained its mass after the first calcination and carbonation cycle. At 700 °C, we can deduce that the conversion is almost complete (>95%). This figure also shows that the reactivity also did not decrease in the second cycle. Under the reaction conditions chosen, any sintering that might have occurred did not seem to adversely affect the sorbent morphology. This finding provides support for the use of CaO in this carbonation-calcination reaction-based CO2 separation process. Conclusions Preliminary screening of metal oxides for their application in the carbonation and calcination reaction scheme identified CaO to be a viable candidate. Experimental data were obtained on the carbonation of a variety of CaO sorbents between 550 and 700 °C. The two regimes of control of the carbonation reaction were verified. It was observed that CaO sorbents from natural sources failed to achieve complete conversion because of their microporous structure. The synthesis procedure for precipitated calcium carbonate was modified to obtain mesoporous CaO sorbents that achieved high (>90%) conversions in the carbonation reaction. This enhancement was attributed to the lower susceptibility of mesoporous sorbents to pore filling and pluggage. Life-cycle studies based on 2-3 cycles did not show an
(1) Cooper, C. D.; Alley, F. C. Air Pollution Control: A Design Approach, 2nd ed.; Waveland Press: Prospect Heights, IL, 1994. (2) Keeling, C. D.; Bacastow, R. B.; Whorf, T. P.; Mook, W. G. Evidence for Accelerated Releases of Carbon Dioxide to the Atmosphere Inferred from Direct Measurements of 13C/12C Ratio. In Proceedings of the 82nd Annual Meeting of the Air and Waste Management Association; Air and Waste Management Association: Pittsburgh, PA, 1989; Vol. 7, Paper 89-119.2. (3) Climate Change 1995: The Science of Climate Change, Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change; Houghton, J. T., Meira Filho, L.G., Callander, B.A., Harris, N., Kattenberg, A., Maskell K, Eds.; Cambridge University Press: Cambridge, U.K., 1996. (4) FETC Carbon Sequestration R&D Program Plan: FY 19992000; National Energy Technology Laboratory, U.S. Department of Energy: Washington, DC, 1999; www.fetc.doe.gov. (5) Wallace, D. Capture and Storage of CO2. What Needs To Be Done. Presented at the 6th Conference of the Parties, COP 6, to the United Nations Framework Convention on Climate Change; The Hague, The Netherlands, Nov 13-24, 2000; www.iea.org/ envissu/index.htm. (6) Herzog, H.; Drake, E.; Adams, E. CO2 Capture, Reuse, and Storage Technologies for Mitigating Global Climate Change. A White Paper; Final Report No. DE-AF22-96PC01257; U.S. Department of Energy: Washington, DC, Jan 1997. (7) Kaplan, L. J. Cost-Saving Processes Recovers CO2 from Power-Plant Fluegas. Chem. Eng. 1982, 89 (24), 30-31. (8) Pauley, C. P.; Smiskey, P. L.; Haigh, S. N-ReN Recovers CO2 from Flue Gas Economically. Oil Gas J. 1984, 82 (20), 8792. (9) Reimer, P.; Audus, H.; Smith, A. Carbon Dioxide Capture from Power Stations, IEA Greenhouse R&D Programme: Cheltenham, Gloucester, U.K., 2001; (www.ieagreen.org.uk). (10) Blauwhoff, P. M. M.; Versteeg, G. F.; van Swaaij, W. P. M. A study on the reaction between CO2 and alkanoamines in aqueous solution. Chem. Eng. Sci. 1984, 39 (2), 207-225. (11) Mimura, T.; Simayoshi, H.; Suda, T.; Iijima, M.; Mitsuake, S. Development of Energy Saving Technology for Flue Gas Carbon Dioxide Recovery by Chemical Absorption Method and Steam System in Power Plant. Energy Convers. Manage. 1997, 38 Suppl. P, S57-S62. (12) Kikkinides, E. S.; Yang, R. T.; Cho, S. H. Concentration and Recovery of CO2 from flue gas by pressure swing adsorption. Ind. Eng. Chem. Res. 1993, 32, 2714-2720. (13) Song, J.-H.; Yoon, J.-H.; Lee, H. Solubility of Carbon Dioxide in Monoethanolamine + Ethylene Glycol + Water and Monoethanolamine + Poly(ethylene glycol) + Water. J. Chem. Eng. Data 1996, 41, 497-499. (14) Zhang, Z.; Guan, J.; Ye, Z. Separation of Nitrogen-Carbon Dioxide Mixture by Rapid Pressure Swing Adsorption. Adsorption 1998, 4, 173-177. (15) Heuchel, M.; Davies, G. M.; Buss, E.; Seaton, N. A. Adsorption of Carbon Dioxide and Methane and Their Mixtures on an Activated Carbon: Simulation and Experiment. Langmuir 1999, 15, 8695-8705. (16) Sawada, Y.; Murakami, M.; Nishide, T. Thermal analysis of basic zinc carbonate. Part 1. Carbonation process of zinc oxide powders at 8 and 13 °C. Thermochim. Acta 1996, 273, 95-102. (17) Shaheen, W. M.; Selim, M. M. Effect of thermal treatment on physicochemical properties of pure and mixed manganese carbonate and basic copper carbonate. Thermochim. Acta. 1998, 322 (2), 117-128.
4042
Ind. Eng. Chem. Res., Vol. 41, No. 16, 2002
(18) Fernandez, A. I.; Chimenos, J. M.; Segarra, M.; Fernandez, M. A.; Espiell, F. Kinetic study of carbonation of MgO slurries. Hydrometallurgy 1999, 53, 155-167. (19) Butt, D. P.; Lackner, K. S.; Wendt, C. H.; Conzone, S. D.; Kung, H.; Lu, Y.-C.; Bremser, J. K. Kinetics of Thermal Dehydroxylation and Carbonation of Magnesium Hydroxide. J. Am. Ceram. Soc. 1996, 79 (7), 1892-1898. (20) Kato, Y.; Saku, D.; Harada, N.; Yoshizawa, Y. Utilization of High Temperature Heat from Nuclear Reactor using Inorganic Chemical Heat Pump. Prog. Nucl. Energy 1998, 32 (3-4), 563570. (21) Kato, Y.; Harada, N.; Yoshizawa, Y. Kinetic feasibility of a chemical heat pump for heat utilization from high-temperature processes. Appl. Therm. Eng. 1999, 19, 239-254. (22) Barker, R. The Reversibility of the Reaction CaCO3 ) CaO + CO2. J. Appl. Chem. Biotechnol. 1973, 23, 733-742. (23) Tinkler, M. J.; Cheh, C. Towards a Coal-Capable Solid Oxide Fuel Cell System. In Proceedings of the 26th International Technical Conference on Coal Utilization and Fuel Systems; Coal Technology Association: Gaithersburg, MD, 2001; pp 569-570. (24) Bhatia, S. K.; Perlmutter, D. D. Effect of the Product Layer on the Kinetics of the CO2-Lime Reaction. AIChE J. 1983, 29 (1), 79-86. (25) Dedman, A. J.; Owen, A. J. Calcium Cyanamide Synthesis,
Part 4. The reaction CaO + CO2 ) CaCO3. Trans. Faraday Soc. 1962, 58, 2027-2035. (26) Oakeson, W. G.; Cutler, I. B. Effect of CO2 Pressure on the Reaction with CaO. J. Am. Ceram. Soc. 1979, 62 (11-12), 556558. (27) Mess, D.; Sarofim, A. F.; Longwell, J. P. Product Layer Diffusion during the Reaction of Calcium Oxide with Carbon Dioxide. Energy Fuels 1999, 13, 999-1005. (28) Wei, S.-H.; Mahuli, S. K.; Agnihotri, R.; Fan, L.-S. High Surface Area Calcium Carbonate: Pore Structural Properties and Sulfation Characteristics. Ind. Eng. Chem. Res. 1997, 36 (6), 21412148. (29) Barker, R. The Reactivity of Calcium Oxide Towards Carbon Dioxide and Its Use for Energy Storage. J. Appl. Chem. Biotechnol. 1974, 24, 221-227. (30) Agnihotri, R.; Chauk, S.; Mahuli, S.; Fan, L.-S. Influence of Surface Modifiers on the Structure of Precipitated Calcium Carbonate. Ind. Eng. Chem. Res. 1999, 38, 2283-2291.
Received for review October 23, 2001 Revised manuscript received April 12, 2002 Accepted April 15, 2002 IE010867L