A High Temperature Lithium Orthosilicate-Based Solid Absorbent for

E-mail: [email protected]. Cite this:Ind. Eng. Chem. Res. ... View: ACS ActiveView PDF | PDF | PDF w/ Links | Full Text HTML. Citing Articles; Re...
0 downloads 0 Views 435KB Size
Download PDF
Article pubs.acs.org/IECR

A High Temperature Lithium Orthosilicate-Based Solid Absorbent for Post Combustion CO2 Capture Robert Quinn,* Ronald J. Kitzhoffer, Jeffrey R. Hufton, and Timothy C. Golden Air Products and Chemicals, Inc., 7201 Hamilton Boulevard, Allentown, Pennsylvania 18195-1501, United States ABSTRACT: Capture of carbon dioxide from combustion processes presents a unique and challenging technical problem arising from low CO2 partial pressures, high flow rates, and the presence of water vapor and reactive contaminants such as SO2. A series of solid sorbents were evaluated to determine suitability for postcombustion capture. A lithium orthosilicate (Li4SiO4)based absorbent supplied by Toshiba Corporation was found to have the most promising properties. The absorbent reacts chemically with CO2 at elevated temperatures (550 °C) to form lithium carbonate (Li2CO3) and lithium metasilicate (Li2SiO3). The presence of water vapor was shown to greatly enhance CO2 absorption rates without affecting capacity. Breakthrough capacities of ∼5−6 mmol/g (22−26 wt %) were obtained using a “clean” synthetic flue gas containing 15% CO2 and 10% H2O in N2 at 550 °C. Experimental studies showed that the absorbent used in a fixed bed process will likely require a thermal swing process with absorption at 550 °C and regeneration at 650 °C. Even for the high capacity of the Li4SiO4-based absorbent, an alternative to conventional fixed bed technology will be required for practical postcombustion capture from coal-fired power plants. Processes that can shorten cycle times by rapidly heating and cooling vessels and/or sorbents will be needed, and possible alternatives are described.

■

INTRODUCTION The capture of carbon dioxide (CO2) from combustion processes presents a unique and challenging technical problem. Flue gas pressures are typically near ambient with CO2 concentrations less than 20 vol %. The resulting low CO2 partial pressure limits the range of suitable sorbents, and a material with a reasonably high CO2 affinity will likely be required. Such materials generally have high heats of CO2 sorption and thus high energy demands for regeneration. For example, the current technology of choice, aqueous monoethanolamine (MEA), can effectively capture CO2, but regeneration requires substantial heat input, 63−84 kJ/mol CO2, resulting in plant inefficiencies and higher costs1 even though recent efforts have achieved some cost and energy reductions.2 Flue gas flow rates are generally high and gas− sorbent contact times will likely be quite short, meaning that sorption must be fast for effective CO2 capture. The presence of water vapor, ∼10 vol %, further complicates identification of a reasonable sorbent. Many materials that sorb CO2 are adversely affected by H2O because it competes effectively for sorption sites. The reactivity of additional flue gas contaminants, particularly SO2, with potential sorbents also needs to be considered.3−5 Many CO2 sorbents react readily with SO2, some irreversibly. A final issue is the presence of residual O2 in flue gas that may result in undesirable oxidation of organicbased sorbents such as MEA. Perhaps the only advantage associated with postcombustion capture is that flue gas is available from the coal-fired boiler temperature (1100−1600 °C)6 to that of the stack gas, 55 °C,1 and thus a wide range of separation temperatures can be considered. In addition to these technical challenges, the scale of postcombustion capture is daunting. A 500 MW coal-fired power plant burns enormous quantities of coal, about 2 rail cars/h, and emits over 4 billion kg CO2 in one year.1 On the basis of the U. S. Department of © 2012 American Chemical Society

Energy’s (DOE) target of 90% capture, 3.6 billion kg CO2 must be separated each year. Solid sorbents have been investigated as an alternative to aqueous MEA for postcombustion CO2 capture.6−10 Possible sorbents are (1) physical adsorbents for which adsorption occurs via weak interaction between the solid and the gas, (2) chemisorbents that react chemically with CO2 on the adsorbent surface resulting in a relatively strong interaction, and (3) solid absorbents that undergo bulk chemical reaction with CO2. A screening effort involving such sorbents was conducted to identify viable candidates for postcombustion capture. A lithium orthosilicate (Li4SiO4)-based absorbent supplied by Toshiba Corporation was found to have very favorable properties, including high capacity and absorption rate, and it holds promise for a practical CO2 capture process. However, development of a process that takes advantage of the absorbent properties remains a challenge. Much of the initial development of a Li4SiO4-based absorbent was performed by Toshiba Corporation.11−16 The absorbent was evaluated for precombustion CO2 capture in steam methane reforming11, and a conceptual evaluation for CO2 capture in coal-fired power plant flue gas was reported.12 Fabrication techniques for the formed absorbent have appeared.17,18 Others have shown an enhancement of Li4SiO4 absorption properties by doping with Al or Fe.19 A study of different forms of silica and methods for doping with K2CO3 found optimal properties when crystalline quartz was used along with 10 mol % K2CO3.20 Samples doped with Na to yield various Li4‑xNaxSiO4 compositions exhibited increased rates of Received: Revised: Accepted: Published: 9320

January 17, 2012 May 7, 2012 June 11, 2012 June 11, 2012 dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327

Industrial & Engineering Chemistry Research

Article

Table 1. Elemental Compositions in Weight % by XRF analysis concentration, weight %

molar ratio, X/Si

sample

Si

Ti

K

Na

Ti

K

Na

LS 5 mm LS 2 mm

33.66 33.77

14.28 12.92

3.15 4.38

− 0.491

0.250 0.225

0.067 0.093

− 0.018

reaction with CO2.21 Doping with the alkali cations Na+, K+, and Cs+ was found to enhance absorption properties with Cs+ giving the largest improvement.22 On the basis of a study of particle size effects, Li4SiO4 reactivity was modeled as a reaction of CO2 at the solid surface followed by rate limiting diffusion of lithium.23,24

■

RESULTS AND DISCUSSION Samples of the Li4SiO4-based absorbents were provided by Toshiba Corporation as 2 and 5 mm diameter spheres and are referred to as LS 2 mm and LS 5 mm, respectively. XRD analysis indicated that both contained Li4SiO4 as the major phase, along with Li2SiO3 and Li2TiO3 phases. Elemental compositions as determined by XRF are listed in Table 1 with the exception of lithium, which is not detected by XRF. It is assumed that potassium was present as amorphous K2CO3 (see below) and thus not observed by XRD. Absorption of CO2 occurs by formation of lithium carbonate and lithium metasilicate as in reaction 1.13−16 On the basis of thermodynamic calculations,25 the Li 2SiO 3 and Li 2 TiO 3 components of the absorbent are inert with respect to CO2 at the anticipated use temperature of 550 °C or higher. Confirmation of the reaction products was obtained for the LS 5 mm absorbent. Exposure of the material to 1 atm CO2 at 650 °C resulted in a ∼23% weight increase, and XRD analysis indicated the expected product, Li2CO3, along with Li2SiO3 and Li 2 TiO 3 . Carbon dioxide capacities as determined by thermogravimetric analysis (TGA) at 650 °C and 1 atm CO2 were 5.51 mmol/g for LS 5 mm and 6.47 mmol/g for LS 2 mm. Absorption was quite fast and nearly complete after a 2 min exposure. Desorption was slower and a 45 min N2 purge at 650 °C was required for desorption of ∼85% of the bound CO2. Li4SiO4 + CO2 (g) = Li 2CO3 + Li 2SiO3

Figure 1. LS 5 mm CO2 absorption rates at 550 °C. Feed: 14.7% CO2 in N2, dry (dashed curve) or humidified to ∼2.6% H2O (solid curve). Bottom panel shows capacities through 10 min exposure.

significant, while at 450 °C, the effect was more dramatic. Absorption for the LS 2 mm absorbent showed a similar but more modest rate enhancement due to the presence of water vapor. Absorption rates in the presence or absence of water vapor were greater for LS 2 mm than those for LS 5 mm. Water vapor-enhanced CO2 absorption rates at room temperature have been reported for Li4SiO4,16 and the addition of Li2SiO3 to the absorbent resulted in improved absorption properties in the presence of water vapor.30 For LS 2 mm containing absorbed CO2, the presence of water vapor in a N2 purge gas had little effect on the rate of desorption relative to that for a dry purge. The enhanced absorption rate in the presence of water vapor does not result from water absorption. Exposure of LS 5 mm to humidified N2 (∼2.6% H2O) at 550 °C resulted in less than 0.02 wt % increase in agreement with the unfavorable thermodynamics of the reaction of Li4SiO4 with water to form LiOH. Comparable water sorption properties were obtained for LS 2 mm. Exposure to humidified N2 at 550 °C resulted in a rapid weight increase corresponding to a maximum of 0.15 mmol H2O/g. Continued exposure to humidified N2 resulted in a gradual weight loss and a final weight slightly lower than before exposure to water vapor. Figure 2 illustrates the effect of CO2 partial pressure on absorption by LS 5 mm at 550 °C using humidified gas mixtures. Carbon dioxide partial pressures of 0.041 atm (4.1% in the figure) and greater resulted in essentially the same capacity, 5.4 mmol/g. For exposure to 0.012 atm CO2, however, the capacity was 0.69 mmol/g. These results are

(1)

Added K2CO3 functions as a promoter to enhance CO2 absorption, and it has been proposed that it lowers the melting temperature of the carbonate product below that of pure Li2CO3, 723 °C.14,17,18 Use of 0.5−30 mol % of an alkali carbonate as a promoter has been reported in the patent literature.13,17 Li2ZrO3 absorbents absorb CO2 by chemistry analogous to that of Li4SiO4, and a K2CO3 additive led to formation of low melting alkali carbonate compositions and a liquid phase that promoted absorption.26−29 Characterization of CO2 Absorption/Desorption Properties. TGA was used to determine absorption properties for a feed gas containing approximately ∼15% CO2 in N2 at 1 atm total pressure. Although CO2 was absorbed to its equilibrium capacity, absorption was discouragingly slow at 550 °C. Quite surprisingly, however, the use of humidified gas mixtures led to a substantial enhancement in absorption rates. Figure 1 compares CO2 absorption as a function of time for LS 5 mm exposed to 14.7% CO2 in N2, either dry or containing ∼2.6% water. The dramatic effect of water vapor is illustrated, for example, by comparing the quantity of CO2 absorbed after 5 min at 550 °C: dry gas, 1.05 mmol/g; humidified gas, 3.67 mmol/g. At 600 °C, the effects of water vapor were much less 9321

dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327

Industrial & Engineering Chemistry Research

Article

mol Li2CO3, was close to the expected heat of fusion, 45 kJ/ mol. In the absence of observed heat changes between 615 and 650 °C, the cause of the step-change in rate remains uncertain. However, similar results have been reported for Li4SiO4 and Na2ZrO3.23,24,31 At relatively low temperatures, CO2 reacts at the solid surface of these materials resulting in carbonate. Because diffusion of lithium or sodium is relatively slow, a surface layer of carbonate forms leading to slower CO2 sorption. At higher temperatures, diffusion is faster but still rate limiting.23,24,31 The substantial increase in absorption rate above 615−625 °C for LS 5 mm is consistent with diffusion no longer being rate limiting, perhaps due to the presence of K2CO3. Utilization of a solid absorbent for postcombustion CO2 capture requires knowledge of CO2 uptake as a function of temperature. A high operating temperature is preferable so as to maximize absorption rates but lower temperatures favor absorption. For a flue gas containing 15% CO2, reduction of the CO2 partial pressure from 0.15 to 0.015 atm is necessary. Hence, the absorption temperature must be lower than the equilibrium temperature associated with absorption at 0.015 atm CO2 partial pressure. TGA was used to determine the temperature dependence of absorption by heating the CO2loaded absorbent under a purge gas containing various CO2 concentrations. The temperature at which desorption began (sample weight loss) was taken to be the equilibrium temperature for the purge gas CO2 partial pressure thus providing equilibrium constants values for reaction 1. For example, heating the CO2-loaded LS 5 mm absorbent under 24.1% CO2/N2 purge resulted in a weight loss beginning at 676 °C corresponding to a Keq = 1/PCO2 value of 4.2. Table 2 lists

Figure 2. CO2 absorption rates at 550 °C for exposure to humidified gas feeds containing various CO2 concentrations. Top: LS 5 mm, 1.2, 4.1, 14.7, 24.1, 45.3, and 100% CO2. Bottom: LS 2 mm, 1.0, 4.1, 14.7, 24.1, 52.1, and 100% CO2.

consistent with two modes of absorption or two different CO2reactive species, one of higher affinity and another of higher concentration and slightly lower affinity. Analogous results were obtained for LS 2 mm, and partial pressures of 0.041 atm and greater gave final capacities of about 6.2 mmol/g (Figure 2). The absorption rates for the LS 5 mm absorbent exhibited an unusual temperature dependence. In contrast to the normal smooth transition from low to high rates with increasing temperature, a step-change increase over a relatively narrow temperature range, 615−625 °C (Figure 3) occurred upon

Table 2. Experimental Equilibrium Constants Values for CO2 Adsorption by LS 5 mm and LS 2 mm As Determined by TGA LS 5 mm

LS 2 mm

T, °C

Keq

T, °C

Keq

573 611 657 676 700 714

100 24 6.9 4.2 2.2 1.0

564 601 647 692 703

83 24 6.8 1.9 1.0

the equilibrium constant values obtained by this method. Experimental values were greater than those calculated from the thermodynamics of reaction 1. For example, the calculated equilibrium constant value at 676 °C is 1.48. The reason for this discrepancy is unclear. Nonetheless, the experimental results can be used to obtain the maximum temperature to achieve 90% postcombustion capture as illustrated in Figure 4 (Keq = 1/0.015 or 67): 585 °C for LS 5 mm and 571 °C for LS 2 mm. It is important to note that the weight losses upon heating in the TGA, and thus the equilibrium constant values, are due to desorption from the lower affinity of the two CO2 reactive species described above. This is noted as incomplete desorption of all the CO2 previously absorbed on the solid. For example, when the CO2-loaded absorbents were heated to 580 °C under 1.01% CO2, much but not all of the bound CO2 was desorbed. On the basis of these weight changes, the capacities of the two species can be obtained: lower affinity species, LS 5 mm, 4.94

Figure 3. LS 5 mm CO2 absorption at the indicated temperatures. Feed: 1 atm 100% CO2, dry.

exposure to a 100% CO2 feed. It was assumed that this dramatic increase near 615 °C was due to formation of a molten phase, but differential scanning calorimetric (DSC) analysis showed that this was not the case. The CO2-loaded LS 5 mm sample was heated under a CO2 atmosphere to suppress desorption and permit observation of any phase change endotherms. The only endotherm below 650 °C was observed at 470 °C with a heat of 1.3 kJ/mol and no sample weight change. An additional endotherm observed at 710 °C was consistent with Li2CO3 melting and perhaps some minimal CO2 desorption. The observed energy change associated with this endotherm, 38 kJ/ 9322

dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327

Industrial & Engineering Chemistry Research

Article

mol CO2 at 650 °C. The origin of the difference between the experimental and calculated heats is unclear. Experimental heat capacities for LS 2 mm under N2 were greater than those of Li4SiO4, the most abundant compound in the absorbent. At 550 °C, a value of 251 J/mol-K was obtained. The heat capacity of the CO2-loaded absorbent under 1 atm CO2 was 237 J/mol-K. Breakthrough Experiments. Breakthrough capacities were determined using a feed gas consisting of 14.7% CO2 in N2, either dry or containing water vapor, generally 10 vol%. Typical feed flows were 40−50 sccm with contact times generally near 4−6 s. Because of the relatively small vessel diameter, the LS 5 mm spheres were crushed and sieved to 25− 35 mesh (0.6 mm average diameter) prior to testing, while LS 2 mm was used as intact 2 mm spheres. Breakthrough capacities are reported as (1) a dynamic capacity equal to the quantity of CO2 absorbed at initial breakthrough and (2) an equilibrium capacity equal to the quantity of CO2 absorbed at 100% breakthrough. Quantities absorbed were determined by integration of the breakthrough curves. Figure 6 compares

Figure 4. Equilibrium constant values for CO2 absorption by LS 5 mm (squares) and LS 2 mm (circles) as a function of temperature. The horizontal line indicates an equilibrium constant value of 67 required for 90% CO2 capture.

mmol/g; LS 2 mm, 5.62 mmol/g; higher affinity species, LS 5 mm, 0.50 mmol/g; LS 2 mm, 0.62 mmol/g. Cycling Stability As Determined by TGA. The cyclic stability of LS 5 mm was evaluated by repeated exposure to 14.7% CO2/2.6% H2O/balance N2 for 10 min followed by a 30 min dry N2 purge (Figure 5). The first cycle capacity of 4.12

Figure 6. Breakthrough curves for LS 5 mm at 550 °C. Feed: 14.7% CO2 dry or humidified to 10% H2O as indicated. Figure 5. Cyclic CO2 capacity normalized to the 1st cycle capacity for LS 5 mm (squares) at 550 °C (humidified 14.7% CO2 in N2 feed, 10 min; dry N2 regeneration, 30 min) and CaO powder (circles) at 750 °C (dry 100% CO2 feed, 30 min; dry N2 regeneration, 30 min). First cycle capacities: LS 5 mm, 4.12 mmol/g; CaO, 14.5 mmol/g.

breakthrough curves at 550 °C for LS 5 mm obtained with a dry gas feed and one containing 10% water vapor. Breakthrough capacities were about 4 times greater in the presence of water than in its absence, 4.18 mmol/g versus 0.90 mmol/g, respectively. The LS 2 mm absorbent exhibited larger breakthrough capacities with a similar enhancement of capacity in the presence of water vapor: 4.45 mmol/g for a dry feed and 6.77 mmol/g for the same feed containing 10% water. Following each feed gas exposure, the absorbent was regenerated by purging with 45% H2O/N2 at 650 °C to ensure complete desorption of CO2. Subsequent exposure to feed gas at 550 °C resulted in no significant change in capacity over a limited number of these cycles (Table 3). Decreasing contact times for LS 2 mm resulted in only modest decline in equilibrium breakthrough capacities and a more substantial decrease in dynamic breakthrough capacities suggesting that mass transfer rates impacted the shape of the mass transfer zone (Table 4). Breakthrough capacities as a function of temperature were evaluated between 350 and 650 °C for LS 5 mm. Two feed gas exposures were completed at each temperature, and average capacities are listed in Table 5. A maximum capacity was achieved near 550 °C. The LS 2 mm absorbent was further examined using a range of gaseous water concentrations in the feed gas (Table 6). After each exposure, regenerations were carried out using a dry N2 purge at 650 °C. For 4.0−20.0% water, breakthrough capacities were nearly constant at about 6 mmol/g.

mmol/g was, of course, lower than the equilibrium capacity due to the limited exposure time. A modest capacity decline to 3.81 mmol/g after 5 cycles was obtained, but this was followed by a gradual increase in capacity and a final capacity of 4.31 mmol/g after 45 cycles. Compared to other high temperature absorbents such as CaO,7,32−34 the LS 5 mm cyclic stability is superior even though recent efforts have led to improved CaO stability.7 For example, a CaO nanopowder at 750 °C (Figure 5) showed a >70% capacity decrease over just 45 cycles. Solid sorbents often undergo a decrease in crush strength with repeated cycling, but this was not the case for the Li4SiO4based absorbents. In a separate experiment, LS 5 mm pellets were exposed to repeated cycles of a 100% CO2 feed at 650 °C and regenerated under a N2 purge, also at 650 °C. After 50 cycles, pellet crush strengths were found to be greater than before cycling, 45.1 lb (8.50 lb/mm) versus 16.2 lb (3.1 lb/ mm) for the fresh absorbent. Heat of Absorption. Heats of absorption for LS 2 mm were determined by DSC using dry and humidified 100% CO2 feeds at 650 °C: dry CO2, 82.0 kJ/mol CO2; humidified CO2, 71.1 kJ/mol CO2. The calculated heat for reaction 1 is 107 kJ/ 9323

dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327

Industrial & Engineering Chemistry Research

Article

cycles utilize a lower regeneration pressure as the driving force for CO2 desorption. In the current application, the feed gas is at roughly atmospheric pressure, so a subatmospheric pressure would be required or the feed gas would need to be compressed for regeneration at atmospheric pressure. Both approaches were eliminated from consideration because of the capital and operating costs associated with compression and vacuum systems. In a concentration swing process, regeneration is carried out at the same temperature as sorption by purging with steam to reduce the CO2 partial pressure above the sorbent. Steam purge is the enabler in this approach, as a relatively high purity CO2 product can be easily obtained by simply condensing liquid water. The concentration swing process has the advantage that no heating or cooling is required, and in principle, cycle times can be short (less than hours). The final approach is a temperature swing process in which the sorbent is heated above the sorption temperature resulting in CO2 desorption. This can be coupled with a steam purge. Relative to a concentration swing process, temperature swing has the disadvantage that sorbent beds must be heated and cooled. This typically takes quite a long time because of the high thermal mass associated with the vessel and solid sorbent, resulting in long cycle times (hours) and thus larger vessels and sorbent inventory. Data presented in Table 3 clearly show that a temperature swing process with absorption at 550 °C and regeneration at 650 °C is sufficient to maintain capacity with cycling. For both LS 2 mm and LS 5 mm, the dynamic and equilibrium capacities were largely unchanged over three cycles. The same was not true for a concentration swing regeneration. Purging CO2loaded LS 5 mm with 45% H2O in N2 at 550 °C for 2 h resulted in decreased capacities over three cycles (Table 7). Similar results were obtained for LS 2 mm with capacities near 6 mmol/g initially, but regeneration at 550 °C led to lower capacities, less than 4.5 mmol/g after five regenerations at 550 °C. Higher purge flows at 550 °C, as expected, led to improved capacities, but the resulting flows were quite large and regeneration times were long. These results imply that regeneration was incomplete under the experimental conditions and that higher gas flows or longer purge times would be required for a concentration swing regeneration. Figure 7 shows why a concentration swing regeneration is unlikely to be practical. Desorption at 550 °C was quite slow, while it was much faster at 650 °C. The scale of postcombustion capture using a Li4SiO4-based absorbent in a fixed bed temperature swing process was calculated on the basis of the most optimistic assumptions. Considered was a 500 MW coal-fired power plant with 90% CO2 capture corresponding to about 419,000 kg CO2/h.1 It was assumed that the equilibrium breakthrough capacity could be achieved, and this value was used to calculate the minimum quantity of absorbent required. A 4.27 m diameter by 6.10 m high vessel was used (volume 87.2 m3), and the number of vessels needed was taken to be twice that required for absorption. A total cycle time of 8 h was assumed with 4 h each for absorption and regeneration. Calculated absorbent weight and number of vessels based on these assumptions were quite large, greater than 5,700,000 kg and greater than 200 vessels. In reality, with less than the theoretical capacity utilized, even greater absorbent quantities would be required. A fixed bed process is unlikely to be practical for postcombustion capture even for the most effective absorbent. This does not mean that solid sorbents cannot be applied to

Table 3. Dynamic and Equilibrium CO2 Breakthrough Capacities for LS 5 mm and LS 2 mm Absorbentsa CO2 capacity, mmol/g absorbent

bed T, °C

H2O conc., %

LS 5 mm

550 550 550 550 550 550 550 550

0 10 10 10 0 10 10 10

LS 2 mm

dynamic

equilibrium

contact time, s

0.77 3.99 3.76 3.82 0.34 6.27 6.02 6.03

0.90 4.18 3.93 3.95 4.45 6.77 6.55 6.63

4.7 4.6 4.7 4.8 2.4 3.5 3.5 3.5

a

Feed 14.7% CO2 in N2, dry or humidified to 10% H2O. Regeneration after each feed exposure: 45% H2O in N2 at 650 °C for 4 h.

Table 4. Effect of Contact Time on Breakthrough Capacities for LS 2 mm Absorbent at 550 °Ca CO2 capacity, mmol/g contact time, s

dynamic

equilibrium

3.5 1.8 0.89 0.45

6.27 5.37 4.82 2.87

6.77 6.50 6.12 6.06

a Feed 14.7% CO2, 10% H2O in N2. Regeneration at 650 °C, 45% H2O in N2 for 4 h prior to each run.

Table 5. CO2 Breakthrough Capacities as a Function of Temperature for LS 5 mm Absorbentsa CO2 capacity, mmol/g T, °C

dynamic

equilibrium

contact time, s

350 400 450 500 550 600 650

0.31 0.75 1.71 3.57 4.27 4.23 3.18

0.43 0.85 2.06 3.88 4.53 4.59 3.41

5.4 5.1 4.6 3.9 3.9 3.6 3.6

a

Feed 14.7% CO2 in N2, 10% H2O. Regeneration after each feed exposure: 45% H2O in N2 at 650 °C for 4 h.

Critical to the successful implementation of any sorbent for CO2 capture is its regeneration properties. Typical modes of regeneration for fixed bed adsorption systems involve either pressure, concentration, or temperature swing. Pressure swing Table 6. Effects of Gaseous Water Concentration on CO2 Breakthrough Capacities for LS 2 mm at 550 °Ca CO2 capacity, mmol/g feed H2O conc., %

dynamic

equilibrium

contact time, s

3.99 5.74 10.00 15.09 20.01

5.54 5.68 5.67 5.78 5.56

6.23 6.05 6.31 6.20 6.02

3.9 3.7 3.6 3.4 3.1

a Feed 14.7% CO 2 in N2 with the indicated gaseous water concentration. Regeneration at 650 °C, dry N2 for 4 h prior to each run.

9324

dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327

Industrial & Engineering Chemistry Research

Article

Table 7. Effects of Regeneration Conditions on Breakthrough Capacities at 550 °Ca absorption experiment CO2 capacity, mmol/g sample/particle size

run

dynamic

equilibrium

contact time, s

T, °C

flow, sccm

time,h

LS 5 mm

1 2 3 1 2 3 4 5 6

3.97 3.68 1.87 4.97 4.16 3.01 3.34 3.94 3.54

4.15 3.88 3.38 6.45 5.52 5.33 4.89 5.01 4.47

3.9 3.6 3.3 1.7 1.8 1.8 1.9 1.9 2.1

550 550 550 650 550 550 550 550 550

180 180 180 1300 1680 670 1320 1600 1600

2 2 2 4 4 6 6 4 4

LS 2 mm

a

pretreatment or regeneration prior to run

Feed 14.7% CO2 in N2, humidified to 10% H2O. Regeneration using ∼45% H2O/N2 and conditions as listed.

capture technologies. Operation in the required temperature range remains a significant technical challenge. Another alternative is a moving bed process in which the sorbent is circulated between a sorption vessel and a regeneration vessel. The process is analogous to a liquid absorption process but solid rather than liquid is circulated from one vessel to another. Reports of moving bed processes for postcombustion have appeared.38,39 In one approach, a Na2CO3-based material sorbs CO2 from flue gas producing a CO2 lean flue gas.39 The CO2-loaded sorbent is moved to a regenerator vessel where CO2 and H2O are desorbed and exit the reactor as a low pressure gas stream. Following condensation of water vapor, the gas is ready for compression and storage. Assuming a composition of 30 wt % Na2CO3 on an inert support and a residence time of 10 min in each vessel, a solid circulation rate of 561,000 kg/min would be required for 90% capture from a 500 MW power plant under the most optimistic conditions. The Li4SiO4-based absorbent could be used in a moving bed process with the absorber at 550 °C and the regenerator at 650 °C. The calculated sorbent inventory based on a capacity of 6 mmol/g and a residence time of 10 min is 530,000 kg, a value substantially smaller than that for the fixed bed process. In addition to the challenges of operating at 550 °C and higher, sorbent attrition is likely to be a major issue for a moving bed process. Circulating solid between vessels results in physical degradation and potential formation of very small particles. This can be a serious problem, especially for more expensive materials, because continuous removal and replacement with fresh sorbent may be required. Perhaps the most promising approach is a rapid thermal swing chemisorption (RTSC) process that utilizes shell and tube type vessels to minimize heating and cooling times.40−42 The sorbent is contained within the tubes, and the shell side is used for heating and cooling. The chemisorbent Na2O/Al2O3 has been evaluated for sorption at 150 °C and regeneration at 450 °C using a steam purge.41,42 Simulations showed that heating and cooling times of 3 min were feasible. At the regeneration temperature of 450 °C, a relatively high pressure (18.5 atm) CO2 product can be obtained. Calculations indicated a 24% reduction in capital cost and a 78% reduction in operating cost relative to conventional MEA technology. The use of the Li4SiO4-based absorbent with its much higher capacity would provide a significant advantage over Na2O/ Al2O3, but higher operating temperatures would be required. Absorption would be carried out at 550 °C, and regeneration at >700 °C would be required for a ≥1 atm CO2 product. Higher

Figure 7. CO2 concentration (%, dry basis) in reactor exit for regeneration of LS 2 mm at 550 and 650 °C as indicated; 700 sccm 45% H2O/N2 purge gas.

postcombustion capture. It means that alternatives to a conventional temperature swing fixed bed process are required. For example, if the Li4SiO4-based absorbent could be used with a cycle time of 30 min rather than 8 h, the number of vessels and quantity of absorbent become much more reasonable, 357,000 kg and 14 vessels, on the basis of the same optimistic assumptions. A process that heats and cools vessels and contents rapidly could result in shorter cycle times, thus minimizing absorbent inventory and number of vessels. Three such alternatives are described below. One option is an “adsorbent wheel” in which a sorbent is contained within a rotating wheel or disk. The wheel rotates into a sorption zone where flue gas flows through the sorbent and CO2 is sorbed. The wheel continues to rotate into a desorption zone where the sorbent is regenerated by purging, heating, or both. The advantage is that shorter cycle times can be achieved versus a fixed bed temperature swing process and thus the quantity of sorbent can be minimized. This technology has been applied on a commercial scale for removal of water vapor from air and organic vapors from process streams. Removal of CO2 from gas streams has been cited in the patent literature.35,36 Commercial adsorbent wheels operate at ≤175 °C, a temperature adequate for solid sorbents such as alkali carbonates but not high enough for Li4SiO4-based absorbents. The possibility of an adsorbent wheel using a Li4SiO4-based absorbent has been considered previously.37 It was concluded that capture of 169 tons/h or 63% of the CO2 from a 250 MW coal plant would require a 20 m diameter × 1.5 m thick wheel. The projected cost of capture was less than $20/ton or a potential cost saving of 30−40% over existing alternative 9325

dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327

Industrial & Engineering Chemistry Research

Article

by XRF analysis using an Axios Spectrometer. Samples were ground into powder and placed in a sample cup covered with Prolene film in a helium atmosphere. XRD analysis was carried out on the MPD form 10−85° using X’Celerator optics, Co radiation, and a 0.03° step size. Breakthrough Experiments. The absorbent was typically contained in a stainless steel tube measuring 0.37 in internal diameter and 7.25 in long. The absorbent vessel was heated to the desired operating temperature using heat tape. Following an initial purge with N2, the absorbent was exposed to 14.7% CO2 in N2 generally humidified by passage through a water bubbler at 46 °C or about 10% gaseous water. Gas flows were controlled using Brooks 5850 mass flow controllers. Gases exiting the absorbent vessel were passed through a chilled trap to condense water vapor prior to analysis using a Horiba CO2 analyzer. Quantities of CO2 absorbed were calculated using gas flows and inlet and outlet CO2 concentrations. Dynamic capacities correspond to the quantity absorbed at initial breakthrough of CO2. Equilibrium capacities were obtained from mass balance as the difference between the total quantity of CO2 in the inlet and effluent gas. Regeneration was carried out by purging with either a dry N2 or ∼45% H2O/55% N2 stream. The latter was generated by vaporizing liquid water supplied at a known flow rates using a HPLC pump. Concentrations of desorbed CO2 were determined after passage through a cold trap as described above.

regeneration temperatures would be required for a higher pressure CO2.

■

CONCLUSIONS A lithium orthosilicate (Li4SiO4)-containing absorbent supplied by Toshiba Corporation was found to have promising properties for postcombustion CO2 capture. The absorbent reacts chemically with CO2 at elevated temperatures (550 °C) to form lithium carbonate (Li2CO3) and lithium metasilicate (Li2SiO3). The presence of water vapor was shown to greatly enhance the rate of CO2 absorption. Breakthrough capacities of ∼5−6 mmol/g were obtained using a “clean” synthetic flue gas containing 15% CO2 and 10% H2O in N2 at 550 °C. Experimental studies showed that the absorbent used in a fixed bed process will require a thermal swing process with absorption at 550 °C and regeneration at 650 °C. A concentration swing process will likely be impractical because desorption was quite slow at 550 °C. In this regard, other sorbents that react chemically with CO2 will also likely require a temperature swing regeneration. Even for the high capacity demonstrated for the Li4SiO4-based absorbent, calculations based on the most optimistic assumptions revealed that the scale of a fixed bed temperature swing process would be impractical for postcombustion at a 500 MW coal-fired power plant. The very large scale is attributable to the process and not the sorbent. Potential alternatives to fixed bed technology are an adsorbent wheel process, a moving bed process, or most promising, a rapid thermal swing chemisorption process.

■

■

AUTHOR INFORMATION

Corresponding Author

*Tel: 610-481-4306. E-mail: [email protected].

EXPERIMENTAL SECTION Materials. All Li4SiO4-containing absorbents were supplied by Toshiba Corporation as either 5 or 2 mm diameter spheres. For use in breakthrough experiments, the 5 mm spheres were crushed and sieved into 25−35 mesh particles under an inert atmosphere. The 2 mm spheres were used as received. For identification of CO2 reaction products, 157 g LS 5 mm was heated to 650 °C and purged with 150−200 cc/min CO2 for 20 h. Following cooling to room temperature under CO2, 193 g of material was recovered corresponding a CO2 loading of 23 wt %. XRD analysis was used to identify the product phases. Methods. Thermogravimetric analysis was performed using a TA Instruments TGA Q5000 instrument. The instrument was contained within a N2 purge box to minimize exposure to atmospheric water or oxygen. Humidified gases were obtained by passage through a water bubbler at room temperature. Water concentrations in such gas streams were not measured and were assumed to be equal to the water vapor pressure at room temperature, 20 Torr or 2.6% at ambient pressure. TGA experiments to determine equilibrium constant values for CO2 sorption were performed as follows. The absorbent sample was heated to 650 °C and exposed to 1 atm CO2 for 30 min. Following cooling to 40 °C under CO2, the sample was exposed to a CO2-containing purge gas while heating at 2 °C/ min to 750 °C. Two runs were performed for each CO2 composition to ensure reproducibility, and nearly identical temperatures corresponding to the initial weight loss were obtained. Heats of absorption were obtained using a TA Instrument 2960 SDT V3.0F by purging the sample with pure CO2 at the temperature of interest. Heat capacities were determined using a TA Instruments 2920 DSC V2.6A with the sample purged with N2 or pure CO2. Elemental compositions were determined

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank B. Wu and T. Imada of Toshiba Corporation for providing absorbent samples and for technical discussions. The technical contributions of F. Prozonic, D. Nelson, P. DeSanto, C. Mengel-Smith, S. Voth, and S. Gardner are gratefully acknowledged.

■

REFERENCES

(1) The Future of Coal. MIT, 2007. http://web.mit.edu/coal/. (2) Rochelle, G. T. Amine scrubbing for CO2 capture. Science 2009, 325, 1652−1654. (3) Ryu, H.; Grace, J. R.; Lim, C. J. Simultaneous CO2/SO2 capture characteristics of three limestones in a fluidized-bed reactor. Energy Fuels 2006, 20, 1621−1628. (4) Sun, P.; Grace, J. R.; Lim, C. J.; Anthony, E. J. Removal of CO2 by calcium-based sorbents in the presence of SO2. Energy Fuels 2007, 21, 163−170. (5) Pacciani, R.; Torres, J.; Solsona, P.; Coe, C.; Quinn, R.; Hufton, J.; Golden, T.; Vega, L. F. Influence of the concentration of CO2 and SO2 on the absorption of CO2 by a lithium orthosilicate-based absorbent. Environ. Sci. Technol. 2011, 45, 7083−7088. (6) Cheng, J.; Zhou, J.; Liu, J.; Zhou, Z.; Huang, Z.; Cao, X.; Zhao, X.; Cen, K. Sulfur removal at high temperature during coal combustion in furnaces: A review. Prog. Energy Combust. Sci. 2003, 29, 381−405. (7) Wang, Q.; Luo, J.; Zhong, Z.; Borgna, A. CO2 capture by solid adsorbents and their applications: current status and new trends. Energy Environ. Sci. 2011, 4, 42−55. (8) Sjostrom, S.; Kruta, H. Evaluation of solid sorbents as a retrofit technology for CO2 capture. Fuel 2010, 89, 1298−1306. (9) Davidson, R. M. Post-combustion carbon capture − Solid sorbents and membranes. IEA Clean Coal Centre, January 2009. 9326

dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327

Industrial & Engineering Chemistry Research

Article

(32) Wang, J.; Anthony, E. J. On the decay behavior of the CO2 absorption capacity of CaO-based sorbents. Ind. Eng. Chem. Res. 2005, 44, 627−629. (33) Lysikov, A. I.; Salanov, A. N.; Okunev, A. G. Change of CO2 carrying capacity of CaO in isothermal recarbonation-decomposition cycles. Ind. Eng. Chem. Res. 2007, 46, 4633−4638. (34) Grasa, G. S.; Abanades, J. C. CO2 capture capacity of CaO in long series of carbonation/calcination cycles. Ind. Eng. Chem. Res. 2006, 45, 8846−8851. (35) Gidaspow, D.; Onischak, M. Process for Regenerative Sorption of CO2. U.S. Patent 3,865,924, February 1975. (36) Baum, J.; Schimkat, A. Process for Removing Carbon Dioxide Form Gas Turbine Exhaust Gas and Apparatus for Conducting the Process. European Patent 1,138,369, 2007. (37) Shimomura, Y. The CO2 Wheel: a Revolutionary Approach to Carbon Dioxide Capture. Modern Power Systems, January 15−16, 2003. (38) Green, D. A.; Turk, B. S.; Gupta, R. P.; Portzer, J. W.; McMichael, W. J. Capture of carbon dioxide from flue gas using solid regenerable sorbents. Int. J. Environ. Technol. Manage. 2004, 4, 53−67. (39) Nelson, T. O.; Coleman, L. J. I.; Green, D. A.; Gupta, R. P. The dry carbonate process: Carbon dioxide recovery from power plant flue gas. Energy Proc. 2009, 1, 1305−1311. (40) Sircar, S. Rapid Thermal Swing Adsorption. U.S. Patent Application 2003/0037672, February 2003. (41) Lee, K. B.; Sircar, S. Removal and recovery of compressed CO2 from flue gas by a novel thermal swing chemisorption process. AIChE J. 2008, 54, 2293−2302. (42) Beaver, M. G.; Sircar, S. Adsorption technology for direct recovery of compressed, pure CO2 from a flue gas without precompression or pre-drying. Adsorption 2010, 16, 103−111.

(10) Samanta, A.; Zhao, A.; Shimizu, G. K. H.; Sarkar, P.; Gupta, R. Ind. Eng. Chem. Res. 2012, 51, 1438−1463. (11) Kato, M.; Maezawa, Y.; Takeda, S.; Hagiwar, Y.; Kogo, R. Precombustion CO2 capture using ceramic absorbent and methane steam reforming. J. Ceram. Soc. Jpn. 2005, 113, 252−254. (12) Nakgaki, T.; Yamashita, K.; Kato, M.; Essaki, K.; Iwahashi, T.; Fukuda, M. Performance prediction of high-temperature CO2 capture system utilizing lithium silicate for pulverized coal-fired power plant. Proc. Int. Conf. Power Eng., April 2005. (13) Kato, M.; Yoshikawa, S.; Nakagawa, K. Carbon dioxide absorption by lithium orthosilicate in a wide range of temperature and carbon dioxide concentrations. J. Mater. Sci. Lett. 2002, 21, 485− 487. (14) Kato, M.; Nakagawa, K; Essaki,K.; Yoshikawa, S. Method of Regenerating Carbon Dioxide Gas Absorbent. U.S. Patent Application 20060183628, August 2006. (15) Essaki, K.; Kato, M.; Uemoto, H. Influence of temperature and CO2 concentration on the CO2 absorption properties of lithium silicate pellets. J. Mater. Sci. 2005, 40, 5017−5019. (16) Essaki, K.; Nakagawa, K.; Kato, M.; Uemoto, H. CO2 absorption by lithium silicate at room temperature. J. Chem. Eng. Jpn. 2004, 37, 772−777. (17) Kato, M.; Nakagawa, K.; Ohashi, T.; Yoshikawa, S.; Essaki, K. Carbon Dioxide Gas Absorbent Containing Lithium Silicate. U.S. Patent 6,387,845, May 2002. (18) Essaki, K.; Kato, M.; Yoshikawa, S.; Imada, T. CatalystContaining Reaction Accelerator and Steam Reforming Method Using Hydrocarbon. U.S. Patent Application 20050214203, September 2005. (19) Gauer, C.; Heschel, W. Doped lithium orthosilicate for absorption of carbon dioxide. J. Mater. Sci. 2006, 41, 2405−2409. (20) Seggiani, M.; Puccini, M.; Vitolo, S. High-temperature and low concentration CO2 sorption on Li4SiO4 based sorbents: study of the used silica and doping method effects. Int. J. Greenhouse Gas Control 2011, 5, 741−748. (21) Mejia-Trejo, V. L.; Fregoso-Israel, E.; Pfeiffer, H. Textural, structural, and CO2 chemisorption effects produced on the lithium orthosilicate by its doping with sodium (Li4‑xNaxSiO4). Chem. Mater. 2008, 20, 7171−7176. (22) Korake, P. V.; Gaikwad, A. G. Capture of carbon dioxide over porous solid adsorbents lithium silicate, lithium aluminate and magnesium aluminate at pre-combustion temperatures. Front. Chem. Sci. Eng. 2011, 5 (2), 215−226. (23) Venegas, M. J.; Fregoso-Israel, E.; Escamilla, R.; Pfeiffer, H. Kinetic and reaction mechanison of CO2 sorption on Li4SiO4: study of particle size effects. Ind. Eng. Chem. Res. 2007, 46, 2407−2412. (24) Rodriguez-Mosqueda, R.; Pfeiffer, H. Thermokinetic analysis of the CO2 chemisorption on Li4SiO4 by using different gas flow rates and particle sizes. J. Phys. Chem A 2010, 114, 4535−4541. (25) Thermodynamic calculations were performed using HSC Chemistry for Windows, version 5.11, Outokumpu Research. (26) Ohashi, T.; Nakagawa, K.; Ohzu, H.; Akasaka, Y.; Tomimatsu, N. Method for Manufacturing a Carbon Dioxide Absorbent. U.S. Patent 6,271,172, August 2001. (27) Lin, J. Y. S.; Ida, J.; Xiong, R. Carbon dioxide sorption on lithium zirconate: mechanism, model and sorbent improvement. 8th Int. Conf. Fund. Adsorption 2004, 75. (28) Ida, J.; Xiong, R.; Lin, Y. S. Synthesis and CO2 sorption properties of pure and modified lithium zirconate. Sep. Purif. Technol. 2004, 36, 41−51. (29) Ida, J. I.; Lin, Y. S. Mechanism of high-temperature CO2 sorption on lithuium zirconate. Environ. Sci. Technol. 2003, 37, 1999− 2004. (30) Imada, T.; Kato, Y.; Kato, M. Carbon Dioxide Absorbent, Carbon Dioxide Separating Apparatus, and Reformer. U.S. Patent 7,588,630, September 2009. (31) Alcerreca-Corte, I.; Fregoso-Israel, E.; Pfeiffer, H. CO2 absorption on Na2ZrO3: A kinetic analysis of the chemisorptions and diffusion processes. J. Phys. Chem. C 2008, 112, 6520−6525. 9327

dx.doi.org/10.1021/ie300157m | Ind. Eng. Chem. Res. 2012, 51, 9320−9327