Studies in Surface Science and Catalysis 153 S.-E. Park, J.-S. Chang and K.-W. Lee (Editors) © 2004 Elsevier B.V. All rights reserved.
Novel Nanoporous "Molecular Basket" Adsorbent for CO2 Capture Chunshan Song*, Xiaochun Xu, John M. Andresen, Bruce G. Miller and Alan W. Scaroni Clean Fuels and Catalysis Program, The Energy Institute, and Department of Energy & Geo-Environmental Engineering, Pennsylvania State University, 209 Academic Projects Building, University Park, PA 16802, USA A CO2 adsorbent with high adsorption capacity and high selectivity was prepared based on a novel "molecular basket" concept consisting of mesoporous molecular sieve of MCM-41 type and an immobilized branched polymer with CCb-capturing sites. The novel CO2 "molecular basket" adsorbent was characterized by X-ray powder diffraction (XRD). N2 adsorption/desorption and CO2 adsorption/desorption measurement. The "'molecular basket" adsorbent can selectively capture CO2 in a gas mixture and was also applied in the separation of CO2 from simulated flue gas and boiler flue gas. 1. INTRODUCTION The continued use of fossil fuels to provide clean and affordable energy supply depends on the technology development to reduce their negative environmental impact. On the one hand, the emissions of particulate matter and the oxides of nitrogen (NOx) and sulfur (SOx) need to be further reduced [1, 2]. On the other hand, considerable increase in global atmospheric CO2 concentration has caused serious concern for climate change and led to worldwide effort in research and development on the control of CO2 emissions [3, 4]. Capture and separation of CO2 from stationary sources is considered an important option for the control of CO2 emission. The capture/separation cost constitutes about threefourths of the total cost of the control of CO2 emissions, e.g., carbon sequestration . It is therefore important to explore new cost-effective approaches for CO2 separation. A new concept called CO2 "molecular basket", has been proposed in our laboratory for developing a high-capacity, high-selective CO2 adsorbent to decrease the CO2 separation cost. The novel "CO2 molecular basket" is a solid adsorbent and can selectively capture CO2 from gas mixtures and "pack" CO2 in condensed form in nanoporous channels. In order to capture a large amount of CO2, the adsorbent needs to have large-pore channels filled with a CO2-capturing substance as the "basket". To make the "basket" CO2-selective, a substance with numerous CO2-affinity sites should be loaded into the pores of the support to increase the affinity between the adsorbent and CO2 and, therefore, to increase the CO 2 adsorption selectivity and CO2 adsorption capacity. In this paper, the preparation of the novel CO2 "molecular basket" adsorbent by using the large pore volume material of mesoporous molecular sieve MCM-41 and CO2-affinity substance of polyethylenimine (PEI) was studied. The separation of CO2 from simulated flue gas and boiler flue gas by using this novel "molecular basket" adsorbent are reported.
* Corresponding author, E-mail: csonaaDS.u.edu; Tel: 814-863-4466; Fax: 814-865-3248
2. EXPERIMENTAL The CO2 "Molecular Basket" adsorbent was prepared by loading 50 wt% PE1 into the mesoporous molecular sieves MCM-41 (MCM-41-PEI-50) and was used as adsorbent in the adsorption separation of CO2 from simulated flue gas mixture and boiler flue gas. The "molecular basket" adsorbents were characterized by X-ray powder diffraction (XRD), N2 adsorption/ desorption as well as the CO2 adsorption/desorption performance. The details on the preparation and characterization of this adsorbent were published elsewhere [6, 7]. The adsorption separation of CO2 from simulated flue gas mixture and boiler flue gas was carried out in a flow adsorption separation system . The simulated flue gas mixture contains 14.9% CO 2 , 4.25% O2 and 80.85% N 2 . In some experiments, the simulated flue gas mixture was mixed with 10% moisture. Typically, 2.0 g powder adsorbent was used for the adsorption separation. The adsorption separation was carried out at 75 °C and ambient pressure. The feed gas flow rate was 10 ml/min. After the adsorption, helium with a flow rate of 50 ml/min was used to purge the adsorbent bed to perform the desorption at 75 °C. The flow rate of the effluent gas was measured by a soap-film flowmeter. The concentration of the effluent gas was measured on-line using a SRI 8610 C Gas Chromatography (GC). The analysis was carried out every 5 minutes when the dry simulated flue gas mixture was used as adsorbate and every 15 minutes when the moist simulated flue gas mixture was used as adsorbate. Adsorption capacity in ml (STP) of adsorbate/g adsorbent and desorption capacity in percentage were used to evaluate the quality of the adsorbent. The adsorption and desorption capacities were calculated from the mass balance during the adsorption separation. The separation factor, a , j, was calculated from equation 1 as the ratio of the amount of gases adsorbed by the adsorbent. (nj/nj)adsorbed, over the ratio of the amount of gases fed into the adsorbent bed, (ni/nj)feed: \ni
The coal-fired flue gas contains 12.5-12.8% CO2, -4.4% O 2 , 50 ppm CO, 420 ppm NOX, 420 ppm SO 2 , 6.2% H2O and 76-77% N 2 . Generally, 30 g adsorbent with particle size between 18 and 35 mesh was used. The beginning and the end of the adsorption column were filled with alumina (-170 g) to decrease the dead volume in the separation system. The adsorption separation was carried out at a feed flow rate of 5-6 1/min and temperature of 80±10 °C. After adsorption, gas was switched to helium at a flow rate of 5 1/min to perform the desorption at 80±10 °C. The flow rate of the effluent gas was measured by a rotameter. The concentrations of O2, CO, CO2, SO2 and NOX in the effluent gas were measured on-line using model NGA 2000 paramagnetic oxygen analyzer; model NGA 2000 nondispersive infrared CO analyzer; model NGA 2000 non-dispersive infrared CO 2 analyzer; model 890 ultraviolet SO2 analyzer; and model NGA 2000 chemiluminescence NOX analyzer, respectively. The analysis was carried out every 5-6 seconds. Since the alumina also adsorbed the gases, a blank separation test with the adsorption column only filled with the alumina (-210 g) was also carried out. Therefore, the adsorption/desorption capacity for the "molecular basket" adsorbent can be calculated by subtracting the adsorption/desorption capacity between the adsorption experiment and blank experiment. The adsorption/desorption capacity was calculated from mass balance during the adsorption separation. The separation factor was calculated from equation 1.
3. RESULTS AND DISCUSSIONS 3.1 Preparation and characterization of CO2 "Molecular Basket" adsorbent The structure of MCM-41 before and after loading 50 wt.% PEI was characterized by XRD and the results are compared in Figure 1. The diffraction patterns of MCM-41 did not change after PEI was loaded, which indicated that the structure of MCM-41 was preserved. However, the intensity of the diffraction patterns of MCM-41 changed. The intensity of the diffraction patterns of MCM-41 decreased after PEI was loaded, which was caused by the pore filling effect [6, 7]. The pore structure analysis by nitrogen adsorption/desorption confirms that PEI was loaded into the pore channels of the MCM-41 support. Completely degassed MCM-41 shows a type IV isotherm. The surface area, pore volume and pore diameter were 1480 m2/g, 1.0 ml/g and 2.75 nm respectively. After loading the PEI, the mesoporous pores were completely filled with PEI, resulting in a type II isotherm and restricting the access of nitrogen into the pores at the liquid nitrogen temperature. The residual pore volume of the MCM-41-PEI-50 was only 0.011 ml/g, the surface area was estimated to be 4.2 m2/g and the average pore diameter was smaller than 0.4 nm. The CO2 adsorption and desorption performance of MCM-41 and MCM-41-PEI-50 was measured by TGA at 75 °C under pure CO2 atmosphere. The results are shown in Figure 2. Before PEI was loaded, the MCM-41 support alone showed a CO2 adsorption capacity of 8.6 mg/g adsorbent. The low adsorption capacity was caused by the weak physical interaction between CO2 and MCM-41 at relatively high temperature. In order to strengthen the interaction between CO2 and MCM-41, the branched polymeric substance PEI with numerous CC^-capturing sites was loaded into the channels of the MCM-41. After loading the PEI, the adsorption capacity increased substantially. The MCM-41-PEI50 showed a CO2 adsorption capacity of 112 mg/g adsorbent, which was much higher than that of the MCM-41 support and higher than that of the pure PEI (109 mg/g-PEl). The desorption was complete for both the MCM-41 support and the MCM-41-PEI-50. However, the desorption for pure PEI was slow and was not complete compared to the desorption time of the MCM-41-PEI adsorbents, which indicated that the "molecular basket" adsorbent facilitates the CO 2 desorption. The fast desorption of CO 2 from the "molecular basket" adsorbent can be explained by the high dispersion of PEI into the MCM-41 channels.
Figure 1 Comparison of the XRD patterns of MCM-41 and MCM-41-PEI-50.
Figure 2 Comparison of CO2 adsorption and desorption performance of MCM-41 and MCM-41-PEI-50 (on 1 g adsorbent). In order to evaluate the effect of the "molecular basket", the adsorption capacity
weighed on PEI in the "molecular basket" adsorbent was calculated with equation (2): PEI adsorption capacity (mg adsorbate/g-PEI) = [Adsorption capacity of the adsorbent -(MCM-41 weight percentage in the adsorbent x Adsorption capacity of pure MCM-41)]/ (PEI weight percentage in the adsorbent) (2) The adsorption capacity weighted on PEI for MCM-4l-PEI-50 was 215 mg/g-PEl. which is two times that of the pure PEI. This clearly shows that MCM-41 has a synergetic effect on the CO2 adsorption when the PEI was loaded into its porous structure. There are two possible reasons for the synergetic effect of MCM-41, i.e., its high surface area and its uniform mesoporous channel. When the PEI was loaded on the materials with high surface area, more CO2 affinity sites were exposed to the adsorbate and thus the adsorption capacity increased. However, when the PEI was coated on a high-surface-area silica gel (550 m2/g) with the PEI loading of 50 wt.%, the CO 2 adsorption capacity was only 156 mg/g-PEI , which was slightly higher than that of the bulk PEI and much lower than that of the "molecular basket" adsorbent with the same PEI loading. Therefore, only when the PEI was loaded into the mesoporous channels of MCM-41 did the "molecular adsorbent" show a highest synergetic effect on the adsorption of CO2. 3.2 Separation of CO2 from simulated flue gas Figure 3 shows the concentrations of CO2, O2 and N2 during the adsorption separation of CO2 from a simulated flue gas mixture at 75°C by the "molecular basket" adsorbent. At the beginning of the separation, CO 2 was completely adsorbed by the adsorbent and the concentration of CO2 was below the detection limit of the gas chromatograph, i.e. < 100 ppm. After 50 minutes of adsorption, CO2 began to break through and was detected in the effluent gas. After 120 minutes of adsorption, the adsorbent can only adsorb 5% of the CO2 from the adsorbate. The CO2 adsorption capacity was 37.5 ml (STP)/g adsorbent before breakthrough and was 45.4 ml (STP)/g adsorbent after 120 minutes of adsorption. The adsorption capacity is comparable with that previously measured from TGA at about 15% CO2 concentration . The adsorption of O2 is much lower than that of the CO2 and the adsorption capacity of O2 is 0.07 ml (STP)/g adsorbent after 120 minutes adsorption. The CO2/O2 separation factor was calculated to be 185. The adsorbent hardly adsorbs any N 2 , with CO2/N2 separation ratio of > 1000. A similar experiment was carried out by using the MCM-41 as the adsorbent, which showed a CO2 adsorption capacity of 3.2 ml (STP)/g adsorbent; the separation factor for CO2/O2 and CO2/N2 was 3.3 and 2.9 respectively, which are much lower than those of the "molecular basket" adsorbent. These indicated that the high adsorption capacity and high selective CO2 adsorbent was prepared by using the "molecular basket" concept and by loading the PEI into the MCM-41 channels. 3.3 Separation of CO2 from moist simulated flue gas The influence of moisture on the adsorption separation of CO2 from simulated flue gas containing about 10% moisture by the novel "molecular basket" adsorbent (MCM-4l-PEI50) was investigated at 75 °C and ambient pressure. Figure 4 compares the CO2 breakthrough curve, where the amount of carbon dioxide was followed as the fraction of the CO2 concentration in the effluent gas from the adsorption column, C, over that of the CO2 concentration in the feed, Co, for the flue gas without moisture and with ~ 10% moisture. In the presence of moisture, the "molecular basket" adsorbent can still
Figure 3 Changes of gas concentration during the separation of CO2 from simulated flue gas. Operation condition: Weight of adsorbent: 2.0 g; Temperature: 75 °C; Feed flow rate: 10 ml/min. Temperature: 75 °C; Feed composition: 14.9% CO2, 4.25% O2 and 80.85% N2.
Figure 4 C O2 breakthrough curve with/without moisture in the simulated flue gas. Operation condition: Weight of adsorbent: 2.0 g; Temperature: 75 °C; Feed flow rate: 10 ml/min. Temperature: 75 °C. Dry feed composition: 14.9% CO2, 4.25% O2 and 80.85% N 2 ; Moist feed composition: 13.55% CO 2 , 3.86% O2, 72.72% N 2 and 9.87% H2O.
effectively adsorb CO2. The CO2 breakthrough time was 60 minutes, which is longer than that under dry flue gas conditions and indicated that the moisture had a promotion effect on the adsorption of CO2 by the "molecular basket" adsorbent. The CO 2 adsorption capacity increased from 45.4 ml (STP)/g adsorbent for dry flue gas to 65.0 ml (STP)/g adsorbent for moist flue gas. Meanwhile, the adsorption of O2 was inhibited at the moist condition. The separation selectivity for CO2/O2 was -180 at the dry flue gas condition and ~ 600 at the moist flue gas condition. The adsorption of N 2 was below the detection limit of gas chromatography. The "molecular basket" adsorbent also adsorbed significant amount of water. The water adsorption capacity was 59.0 ml (STP)/g adsorbent in the moist flue gas condition. However, the CO2 adsorption capacity increased 40% in the presence of water and the separation of water and CO2 is rather easy in reality. The desorption of CO2 and moisture were complete. 3.4 Separation of CO2 from coal-fired boiler flue gas The adsorption separation of CO2 from a coal-fired flue gas was investigated and the CO2 breakthrough curves are shown in Figure 5. (Note that the analysis of the effluent gas composition was carried out after removing the moisture in the gas mixture, the analyzed concentrations of O2, CO, CO2, SO2 and NOX were slightly higher than those in the real flue gas mixture.) Clearly, both the alumina and the "molecular basket" adsorbent can adsorb CO 2 . However, the adsorption performance of the "molecular basket" adsorbent was much better than that of the alumina. The lowest CO2 emission concentration was ~ 2.5 % for alumina and was below 0.1% for the "molecular basket" adsorbent. The CO2 adsorption capacity was 1.4 ml (STP)/g adsorbent for the alumina and 36 ml (STP)/g adsorbent for the "molecular basket" adsorbent. In addition, the "molecular basket" adsorbent showed a better selectivity. The "molecular basket" adsorbent did not adsorb O2. N2 and CO, while the CO2/O2 selectivity was 3.5 for alumina. The SO2 and NOX adsorption capacity for the "molecular basket"
adsorbent were 0.11 ml (STP)/g adsorbent and 0.21 ml (STP)/g adsorbent, respectively. The separation selectivity for CO2/SO2 and CO2/NOX was 1.07 and 0.57 respectively. However, very little NOX and SO2 were adsorbed before CO2 breakthrough. The adsorption capacity for CO2, SO2 and NOX before CO2 breakthrough were 24 ml (STP)/g adsorbent, 0.0074 ml (STP)/g adsorbent and 0.028 ml (STP)/g adsorbent respectively. Therefore, the separation selectivity for CO2/SO2 and CO2/NOX were 10.7 and 2.86, respectively, before CO2 breakthrough. While the desorption of CO2 was complete, very little NOX and SO2 desorbed.
Figure 5 C0 2 breakthrough curve for coal-fired flue gas. Operation condition: Weight of adsorbent: 30 g; Feed composition: 12.5-12.8% CO2, 6.2% H2O, ~ 4.4% O2, 50 ppm CO, 420 ppm NOX, 420 ppm SOX and 76-77% N2; Feed flow rate: 6000ml/min; Temperature: 80±10°C.
4. CONCLUSIONS Novel CO2 "molecular basket" adsorbent with high adsorption capacity and high selectivity has been developed. The "basket" of MCM-41 channels displayed a synergetic effect on the adsorption of CO? by PEL The CO2 "molecular basket" adsorbent was successfully applied to the separation of CO2 from simulated flue gas and boiler flue gas. ACKNOWLEDGEMENTS
Financial support from U.S. Department of Defense (via an interagency agreement with U.S. Department of Energy) and the Commonwealth of Pennsylvania are highly appreciated (Cooperative Agreement No. DE-FC22-92PC92162). REFERENCE 1. J.J. Mooney, Annual Meeting of National Petrochemicals & Refiners Association, San Antonio. TX, March 26-28, 2000. 2. U.T. Turaga and C. Song, American Chemical Society Division of Petroleum Chemistry Preprints. 46(2001)275-279. 3. M. M. Maroto-Valer, C. Song and Y. Soong (Eds). Environmental Challenges and Greenhouse Gas Control for Fossil Fuel Utilization in the 21 st Century. Kluwer Academic/Plenum Publishers. New York, 2002, 447 pp. 4. C. Song, A. M. Gaffhey, K. Fujimoto (Eds). CO2 Conversion and Utilization. American Chemical Society (ACS), Washington DC, ACS Symp. Series, Vol. 809, 2002, 448 pp. 5. U.S. Department of Energy, Carbon Sequestration-Research and Development, 1999. http://www.fe.doe.gov/coal_power/sequestration/reports/ rd/index.html. 6. X.C. Xu, C. Song, J.M. Andresen, B.G. Miller and A.W. Scaroni, Energy & Fuels, 16 (2002) 1463-1469. 7. X.C. Xu, C. Song, J.M. Andresen, B.G. Miller and A.W. Scaroni, Microporous and Mesoporous Materials, 62 (2003) 29-45. 8. X.C. Xu, C. Song, J.M. Andresen, B.G. Miller and A.W. Scaroni, International Journal of Environmental Technology and Management, (2003) submitted for publication.