Microporous and Mesoporous Materials 62 (2003) 2945 www.elsevier.

com/locate/micromeso

Preparation and characterization of novel CO2 molecular basket adsorbents based on polymer-modied mesoporous molecular sieve MCM-41
Xiaochun Xu, Chunshan Song *, John M. Andrsen, e Bruce G. Miller, Alan W. Scaroni
Clean Fuels and Catalysis Program, The Energy Institute, and Department of Energy and Geo-Environmental Engineering, The Pennsylvania State University, 209 Academic Projects Building, University Park, PA 16802, USA Received 22 October 2002; accepted 21 April 2003

Abstract Novel CO2 molecular basket adsorbents were prepared by synthesizing and modifying the mesoporous molecular sieve of MCM-41 type with polyethylenimine (PEI). The MCM-41-PEI adsorbents were characterized by X-ray powder diraction (XRD), N2 adsorption/desorption, thermal gravimetric analysis (TGA) as well as the CO2 adsorption/desorption performance. This paper reports on the eects of preparation conditions (PEI loadings, preparation methods, PEI loading procedures, types of solvents, solvent/MCM-41 ratios, addition of additive, and Si/Al ratios of MCM-41) on the CO2 adsorption/desorption performance of MCM-41-PEI. With the increase in PEI loading, the surface area, pore size and pore volume of the PEI-loaded MCM-41 adsorbent decreased. When the PEI loading was higher than 30 wt.%, the mesoporous pores began to be lled with PEI and the mesoporous molecular sieve MCM-41 showed a synergetic eect on the adsorption of CO2 by PEI. At PEI loading of 50 wt.% in MCM-41-PEI, the highest CO2 adsorption capacity of 246 mg/g-PEI was obtained, which is 30 times higher than that of the MCM-41 and is about 2.3 times that of the pure PEI. Impregnation was found to be a better method for the preparation of MCM-41-PEI adsorbents than mechanical mixing method. The adsorbent prepared by a one-step impregnation method had a higher CO2 adsorption capacity than that of prepared by a two-step impregnation method. The higher the Si/Al ratio of MCM-41 or the solvent/MCM-41 ratio, the higher the CO2 adsorption capacity. Using polyethylene glycol as additive into the MCM-41-PEI adsorbent increased not only the CO2 adsorption capacity, but also the rates of CO2 adsorption/desorption. A simple model was proposed to account for the synergetic eect of MCM-41 on the adsorption of CO2 by PEI. 2003 Elsevier Inc. All rights reserved.
Keywords: Adsorption separation; Carbon dioxide; Characterization; CO2 adsorbent; Flue gas; MCM-41; Mesoporous molecular sieve; Polyethylenimine; Preparation

1. Introduction
Corresponding author. Tel.: +1-814-863-4466; fax: +1-814865-3248. E-mail address: csong@psu.edu (C. Song).
*

Existing energy utilization system is largely based on combustion of fossil fuels in stationary and mobile devices. It is likely that the world will

1387-1811/03/$ - see front matter 2003 Elsevier Inc. All rights reserved. doi:10.1016/S1387-1811(03)00388-3

30

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

continue to rely on fossil fuels as the primary energy supply well into the 21st century [17]. Extensive studies have been conducted worldwide on controlling the emission of pollutants (from combustion) such as SOx , NOx , and particulate matter and trace elements. The availability of the technologies for using fossil fuels to provide clean, aordable energy is essential for the prosperity and the security of the world. On the other hand, the increased CO2 concentration in the atmosphere due to emissions of CO2 from fossil fuel combustion has caused the concerns for global warming [17]. Improving the eciency of energy utilization and increasing the use of low-carbon energy sources are considered to be potential ways to reduce CO2 emissions [6,7]. Recently, CO2 capture and sequestration have been receiving signicant attention and are being recognized as a third option [6,7]. Also, enriched CO2 streams can be an important starting material for synthetic clean fuels and chemicals [1,2,8,9]. For carbon sequestration, the costs for capture and separation are estimated to make up about three-fourths of the total costs of ocean or geologic sequestration [6,7]. It is therefore important to explore new approaches for CO2 separation [68]. There are chemical and physical methods for separation of CO2 from gas mixtures. Adsorption is one of the promising methods that could be applicable for separating CO2 from gas mixtures [1040]. Various adsorbents, such as zeolites [10 23], activated carbons [13,16,17,2429,3134,36], carbon molecular sieves [16,33], pillared clays [37,38] and metal oxides [30,35,39,40], have been investigated. At lower temperatures (e.g., room temperature), the zeolite-based adsorbents and activated carbon have generally been found to show higher adsorption capacity. However, their selectivity to CO2 in the presence of other gases (N2 , etc.) is still low; their adsorption capacities rapidly decline with increasing temperature above 30 C, and become negligible at temperature in excess of 200 C. Siriwardane et al. [13] reported that the CO2 adsorption capacity for zeolite 13X, 4A and activated carbon were about 160, 135 and 110 mg/g-adsorbent respectively at 25 C and 1 atm CO2 partial pressure [30]. Van der Vaart et al. [28] investigated the single and mixed gas adsorp-

tion equilibrium of CO2 /CH4 on Nortit RBI activated carbon. The CO2 adsorption capacity was 108 mg/g-adsorbent at 21.5 C. With increasing temperature, the adsorption capacity decreased to 77 mg/g-adsorbent at 30 C, 56 mg/g-adsorbent at 56 C and 40 mg/g-adsorbent at 75 C respectively. Meanwhile, the activated carbon also adsorbed methane. The CO2 /CH4 selectivity was around 2. Berlier and Frere [24] and Heuchel et al. [27] also found that CO2 adsorption capacity of activated carbon decreased signicantly when the temperature slightly increased and the CO2 /CH4 selectivity was low. Under 0.1 MPa CO2 partial pressure, the adsorption capacity was 126 mg/g-adsorbent at 15 C and 78 mg/g-adsorbent at 45 C. The best CO2 / CH4 ratio (selectivity) was around 2. Since all the gases are physically adsorbed into/onto these carbon and zeolite adsorbents, the separation factors (e.g., CO2 /N2 ratio) are low. In order to operate at relatively high temperature and reach high separation selectivity, chemical adsorption has been studied. Unfortunately, the adsorbents only show a low CO2 adsorption capacity at high temperature although the selectivity for CO2 is high. Anand et al. [30] reported that MgO showed an adsorption capacity of 8.8 mg/g-adsorbent at 400 C. Ding and Alpay [37] investigated the adsorption performance of hydrotalcite and this material showed a CO2 adsorption capacity of 22 mg/gadsorbent at 400 C and 0.2 atm CO2 partial pressure. Both types of adsorbents need high temperature operation and have a low adsorption capacity, thus they are not suitable for practical use for CO2 separation. For practical application, selective adsorbents with high capacity are desired. Many of the separations should preferably be operated at elevated temperature, e.g., higher than room temperature and up to $150 C for the ue gas of power plants. Developing an adsorbent with high CO2 selectivity and high CO2 adsorption capacity, which can also be operated at elevated temperature, is desired for more ecient CO2 separation by adsorption method. Recently, a new concept of CO2 molecular basket has been proposed by our laboratory for developing high-capacity, highly-selective CO2 adsorbents [4143]. The CO2 molecular basket adsorbents can selectively pack CO2 in con-

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

31

densed form in the mesoporous molecular sieve basket and therefore shows a high CO2 adsorption capacity and a high CO2 selectivity. To make the molecular basket adsorbent and to capture a large amount of CO2 gas, the adsorbent needs to have large-pore volume channels to serve as the basket. In order to make the basket to be a CO2 molecular basket, a substance with numerous CO2 -anity sites is loaded into the pores of the support to increase the anity between the adsorbent and CO2 and, therefore, to increase the CO2 adsorption selectivity and CO2 adsorption capacity. In addition, the adsorption anity to CO2 by the CO2 -philic substance increased in the conned mesoporous environment and therefore the mesoporous molecular sieve can have a synergetic eect on the adsorption of CO2 by the CO2 philic substance. In our previous work [4144], the sterically branched polymer of polyethylenimine (PEI), which has branched chains with numerous CO2 -capturing sites such as amine groups, was loaded into the large pore volume porous material of mesoporous molecular sieve MCM-41 type to make the molecular basket adsorbent. The CO2 adsorption capacity was signicantly increased after loading the PEI. The novel CO2 molecular basket adsorbent showed a CO2 adsorption capacity of 215 mg/g-PEI at 75 C and pure CO2 atmosphere, which was 24 times higher than that of the MCM-41 and was two times that of the pure PEI [4143]. It clearly showed that the mesoporous molecular sieve MCM-41 had a synergetic eect on the adsorption of CO2 by PEI. The adsorption of N2 was negligible and smaller than the apparatus measurement limit (<1.0 mg/g-adsorbent). The novel adsorbent can also eectively adsorb CO2 , even at low CO2 concentration, e.g., 0.5% CO2 in a CO2 /N2 gas mixture. Satyapal et al. [44] used the PEI as CO2 adsorbent by coating the PEI on the high surface area solid polymethyl methacrylate polymeric support. The composite material can eectively adsorb CO2 from gas mixtures and has been successfully used in the space shuttle for the removal of CO2 from the breathing air. By loading the PEI into the large pore volume material of MCM-41, the novel molecular basket adsorbent showed a higher adsorption capacity than that of the PEI/polymer composite adsorbent

[4143], which conrmed that it is the molecular basket concept, not the high surface area materials, increase the adsorption capacity largely. In this paper, we report the preparation approaches to further increase the CO2 adsorption capacity of the novel molecular basket adsorbent (i.e., inuence of PEI loading and method of loading on MCM-41, eect of polymer additive; eect of Si/Al ratio of MCM-41), and the synergetic eect of MCM-41 on the adsorption of CO2 by PEI.

2. Experimental 2.1. Preparation of the adsorbents Mesoporous molecular sieve MCM-41 was hydrothermally synthesized from a mixture with the following composition: xAl2 O3 :50SiO2 :4.32Na2 O: 2.19(TMA)2 O:15. 62(CTMA)Br:3165H2 O (x 0, 0.05, 0.25). The synthesis procedure was established in our laboratory [45,46], which is based on the method invented by researchers at Mobil [47,48]. Cab-O-Sil fumed silica (Cabot Corporation), tetramethylammonium silicate solution (0.5 TMA/SiO2 , 10 wt.% SiO2 , Sachem Inc.), sodium silicate (containing 14 wt.% NaOH and 27 wt.% SiO2 , Aldrich), cetyltrimethyl-ammonium bromide [(CTMA)Br, Aldrich], aluminum isopropoxide (Aldrich) and deionized water were used as raw materials. The synthesis was carried out at 100 C for 40 h. After the synthesis, the solid product was recovered by ltration, washed several times with deionized water, dried at 100 C overnight and calcined at 550 C for 5 h to remove the template. The polyethylenimine (PEI) modied MCM-41 was prepared by wet impregnation method. In a typical preparation, the desired amount of PEI (viscous liquid) was dissolved in 8 g methanol under stirring for about 15 min, after which 2 g calcined MCM-41 was added to the PEI/methanol solution. The resultant slurry was continuously stirred for about 30 min. Then the slurry was dried at 70 C for 16 h under reduced pressure (700 mmHg). The as-prepared adsorbent was denoted as MCM-41-PEI-X , where X represents the loading of PEI as weight percentage in the sample. By changing the preparation conditions, such as PEI

32

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

loadings, preparation methods, preparation procedures, types of solvents, solvent/MCM-41 ratios, Si/Al ratios of MCM-41, and adding additive of polyethylene glycol (PEG), dierent adsorbents were prepared. In one adsorbent, commercially available silica gel (SiO2 ) (Merck, surface area 550 m2 /g) was used as support instead of MCM41. Details on the adsorbents and their corresponding preparation conditions are summarized in Table 1. 2.2. Characterization of the adsorbents The mesoporous molecular sieve MCM-41, before and after modication, was characterized by X-ray diraction (XRD) and N2 adsorption/ desorption. The XRD patterns were obtained on a Rigaku Geigerex using CuKa radiation. The N2 adsorption/desorption was carried out on a Quantachrome Autosorb 1 automated adsorption

apparatus, from which the BET surface area, the pore volume and the pore size were obtained. The sample was out-gassed at 75 C for 48 h on a high vacuum line prior to adsorption. The pore volume of the mesoporous molecular sieve was calculated from the adsorbed nitrogen after complete pore condensation (P =P0 0:995) using the ratio of the densities of liquid and gaseous nitrogen. The pore size was calculated by using the BJH method. The thermal chemical and physical properties of MCM-41, PEI and MCM-41-PEI were characterized by thermal gravimetric analysis (TGA). The TGA was performed on a PE-TGA 7 thermal gravimeter. About 10 mg of the sample was heated at 10 C/min to 600 C in airow (100 ml/min). 2.3. Adsorption measurement The adsorption and desorption performance of the adsorbent was measured using a PE-TGA 7

Table 1 Preparation conditions and the adsorption/desorption performance of the adsorbents Sample name Si/Al of MCM-41 Pure 100 500 Pure Pure Pure Pure Pure Pure 100 500 Pure Pure Pure Pure Pure Pure silica PEI loading (wt.%) 5 15 30 50 75 50 50 50 50 30 50 50 50 30 50 100 Solvent Solution/ MCM-41 weight ratio 4:1 4:1 4:1 4:1 4:1 4:1 4:1 4:1 4:1 4:1 2:1 8:1 4:1 4:1 Adsorption capacitya (mg CO2 /g-adsorbent) 8.6 7.6 7.5 7.7 19.4 68.7 112 133 96.8 127 121 111 65.7 99 97 126 77.1 78.0 109 Desorption capacity (%) 103 101 100 105 103 98 99 101 97 100 99 100 99 97 98 99 100 98 56

Si-MCM-41 Al-MCM-41-100 Al-MCM-41-500 Si-MCM-41-PEI-5 Si-MCM-41-PEI-15 Si-MCM-41-PEI-30 Si-MCM-41-PEI-50 Si-MCM-41-PEI-75 Si-MCM-41-PEI-25-25b Al-MCM-41-100-PEI-50 Al-MCM-41-500-PEI-50 Si-MCM-41-PEI-50-H2 O PEI-30-Si-MCM-41c Si-MCM-41-PEI-50-Md Si-MCM-41-PEI-50-S2 Si-MCM-41-PEI-50-S8 Si-MCM-41-PEI-30PEG-20e SilicaGel-PEI-50f PEI
a b

silica silica silica silica silica silica

silica silica silica silica silica silica

Methanol Methanol Methanol Methanol Methanol Methanol Methanol Methanol Water Methanol Methanol Methanol Methanol Methanol

The adsorption capacity was measured by TGA under pure CO2 atmosphere at a ow rate of 100 ml/min at 75 C. The adsorbent was prepared by two-step impregnation method. In each step, 25 wt.% PEI was loaded. c The adsorbent was prepared by adding the PEI/methanol solution to the MCM-41 powder. d The adsorbent was prepared by mechanical mixing method. e PEG was added to the adsorbent of MCM-41-PEI-30. The weight percentage of PEG was 20 wt.%. f Silica Gel (SiO2 from Merck) was used as support material, instead of MCM-41.

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

33

thermal analyzer. The weight change of the adsorbent was followed to determine the adsorption and the desorption performance of the materials. In a typical adsorption/desorption process, about 10 mg of the adsorbent was placed in a small sample cell, heated up to 100 C in N2 atmosphere at a ow rate of 100 ml/min and held at that temperature (about 30 min) until there was no weight loss. Then the temperature was decreased to 75 C and the 99.8% bone-dry CO2 adsorbate was introduced at a ow rate of 100 ml/min. After the adsorption, the gas was switched to 99.995% pure N2 at a ow rate of 100 ml/min to perform desorption at the same temperature. The time for both the adsorption and desorption was 150 min. The adsorption/desorption temperature of 75 C and the adsorption and desorption time of 150 min were selected because our previous investigation showed that the MCM-41-PEI exhibited the best adsorption/desorption performance at 75 C; and that the adsorption nearly reached equilibrium and the desorption was complete after 150 min [41,42]. Adsorption capacity in mg of adsorbate/gadsorbent and desorption capacity in percentage were used to evaluate the adsorbent and were calculated from the weight change of the sample in the adsorption/desorption process. Desorption capacity in percentage was dened as the ratio of the amount of the gas desorbed over the amount of gas adsorbed.

Si-MCM-41-PEI-75

X4

Si-MCM-41-PEI-50

X4

Si-MCM-41-PEI-30

X2

Si-MCM-41-PEI-15

Si-MCM-41-PEI-5

Si-MCM-41 1 2 3 4 5 6 7 8 9 10

Fig. 1. XRD patterns of Si-MCM-41-PEI with dierent PEI loadings as shown by the trailing digits. The top three curves were shown at dierent Y -scales.

3. Results 3.1. Preparation and characterization of MCM-41PEI Si-MCM-41-PEI with dierent PEI loadings were prepared and characterized by XRD, N2 adsorption/desorption and TGA. Fig. 1 shows the XRD patterns of Si-MCM-41 and Si-MCM-41PEI with dierent PEI loadings. From comparing the diraction patterns of Si-MCM-41 with those of Si-MCM-41-PEI with dierent PEI loadings, the degree of Bragg diraction angles were nearly identical, indicating that the structure of Si-MCM41 was preserved after loading the PEI. However, the intensity of the diraction patterns of Si-

MCM-41 decreased after the PEI was loaded. Fig. 2 shows the diraction intensity of Si-MCM-41 (100 plane) for Si-MCM-41-PEI with dierent PEI loadings. With increasing PEI loadings, the intensity of the diraction peaks decreased. The intensity of the diraction peak of Si-MCM-41-PEI-50 and Si-MCM-41-PEI-75 was only 11% of that of the Si-MCM-41 support. At the same time, the degree of the Bragg diraction angle of the 100 plane slightly increased from 2.265 for Si-MCM41 to 2.2952.325 for Si-MCM-41-PEI. These changes were possibly caused by the pore lling eect of the MCM-41 channels and the PEI coating on the outer surface of the MCM-41 crystals. Reddy and Song [45] reported that the XRD patterns of MCM-41, after removal of the template, exhibited peaks with increased intensity and a shift to lower Bragg diraction angle compared to the template-containing MCM-41. Since the pore volume of the Si-MCM-41 support is 1.0 ml/g and the density of PEI is about 1.0 g/ml, the maximum

34
8000

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

0.6 0.5
Si-MCM-41 Si-MCM-41-PEI-5 Si-MCM-41-PEI-15 Si-MCM-41-PEI-30 Si-MCM-41-PEI-50 Si-MCM-41-PEI-75

Diffraction intensity (CPS)

7000 6000

D(v) (ml//g)
0 15 30 45 60 75

5000 4000 3000 2000 1000 0

0.4 0.3 0.2 0.1 0 10 20 30 40

50

60

Fig. 2. The inuence of PEI loadings in Si-MCM-41-PEI on the diraction intensity of the (1 0 0) plane of MCM-41.

Fig. 3. The pore size distribution of Si-MCM-41-PEI with dierent PEI loadings.

PEI loading in the pores of Si-MCM-41 is 50 wt.%. More PEI should be coated on the outer surface of Si-MCM-41 crystals for Si-MCM-41-PEI-75 than for Si-MCM-41-PEI-50. However, the diraction intensity of Si-MCM-41 for Si-MCM-41-PEI-50 and Si-MCM-41-PEI-75 was nearly the same, which indicated that the PEI coating on the outer surface of the crystals hardly inuenced the diraction intensity of the Si-MCM-41 support. Therefore, the decrease in the diraction intensity and the shift of the Bragg diraction angle of the 100 plane to high degree can be ascribed mainly to the loading of PEI into the Si-MCM-41s pore channels. The pore size, surface area and pore volume of Si-MCM-41 before and after loading the PEI were obtained from the nitrogen adsorption/desorption isotherms. Fig. 3 shows the pore size distributions of Si-MCM-41-PEI with dierent PEI loadings and Fig. 4 shows their surface areas and pore volumes. The pore size of the Si-MCM-41 support was 2.75 nm. After the PEI was loaded into its channels, the pore size decreased. The pore size of Si-MCM-41-PEI-5 was 2.47 nm, smaller than that of the Si-MCM-41 support, which conrmed that PEI was loaded into the Si-MCM-41 pore channels. With increasing PEI loadings, the pore size further decreased. The pore sizes were 2.19 and 1.68 nm for Si-MCM-41-PEI-15 and Si-MCM-41PEI-30, respectively. When the PEI loading further increased to 50 wt.%, the mesoporous pores channels were completely lled with PEI, restrict-

100000 10000
Surface Area Pore Volume

10 1 0.1

Surface area (m2/g)

1000 100 0.01 10 1 0.1 0 10 20 30 40 50 60 70 80 0.001 0.0001

Fig. 4. The surface area and pore volume of Si-MCM-41-PEI with dierent PEI loadings.

ing the access of nitrogen into the pores at the liquid nitrogen temperature. Therefore, information on the pore size cannot be obtained from the N2 adsorption/desorption isotherms for PEI loading above 50 wt.%. This is also consistent with the estimated maximum PEI loading of 50 wt.% inside the pores of Si-MCM-41 (based on the pore volume of the Si-MCM-41 and the density of PEI). The surface area and the pore volume of Si-MCM-41, after loading the PEI, exhibited the same trends as the pore size. The Si-MCM-41 support had a surface area of 1486 m2 /g and a pore volume of 1.0 ml/g. The surface area and the pore volume decreased after PEI was loaded into its channels. For example, when the PEI loading was 5.0 wt.%, the adsorbent showed a pore volume of

Pore volume (ml/g)

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

35

0.79 ml/g, smaller than that of the Si-MCM-41 support, which conrmed that PEI was loaded into the Si-MCM-41s channels. With increasing PEI loadings, the surface area and the pore volume further decreased. When the PEI loading was higher than 30 wt.%, the surface area decreased sharply, which indicated that some of the mesoporous pores was completely lled with the PEI. The surface area was only 4.2 m2 /g and the residual pore volume was only 0.011 ml/g for Si-MCM41-PEI-50 at liquid nitrogen temperature. These results correlate with the pore lling eect of the PEI, which was also reected by the XRD characterization, and further conrms that PEI was loaded into the pore channels of Si-MCM-41. The thermal chemical and physical properties of Si-MCM-41, PEI and Si-MCM-41-PEI were measured by TGA. Fig. 5 shows the TGA results and Fig. 6 shows the dierential thermal gravimetric analysis (DTGA) results. As expected, there was no weight loss for Si-MCM-41 till 600 C, which indicated that the hydrophobic Si-MCM-41 was free from adsorbed water or other gases at room temperature. The PEI lost 3.8% of its original mass at 100 C, which can be mainly ascribed to the desorption of CO2 and moisture. This was conrmed by analyzing the euent gas by gas chromatograph (GC). This also indicated that PEI has a low vapor pressure and unlike the commercially used amines such as diethanolamine (DEA), which makes PEI suitable for long-term use at relatively high temperature. The PEI began to

Derivate weight loss (%/min)

0 -5 -10 -15 -20 -25 -30 100

Si-MCM-41 Si-MCM-41-PEI-5 Si-MCM-41-PEI-15 Si-MCM-41-PEI-30 Si-MCM-41-PEI-50 Si-MCM-41-PEI-75 PEI

140

180

220
o

260

300

Fig. 6. DTGA prole for Si-MCM-41-PEI with dierent PEI loadings.

100 90 80 70 60 50 40 30 20 10 0 0 100 200 300 400 500 600

PEI Si-MCM-41-PEI-75 Si-MCM-41-PEI-50 Si-MCM-41 Si-MCM-41-PEI-5 Si-MCM-41-PEI-15 Si-MCM-41-PEI-30

Fig. 5. TGA prole for Si-MCM-41-PEI with dierent PEI loadings.

decompose above 150 C and a sharp weight loss appeared at 205 C. When the temperature was increased above 225 C, the rate of weight loss decreased, indicating that a dierent decomposition process took place. At 600 C, the PEI was completely decomposed and removed as volatiles. After the PEI was loaded into the Si-MCM-41s channels, the sharp weight loss of PEI took place at a lower temperature, which indicated that the decomposition temperature of PEI decreased. The weight loss also took place in a narrower temperature range than that of the pure PEI. These phenomena can be explained by the uniform dispersion of PEI into the nano-porous support pores, since the melt or decomposition temperature will decrease when the particle size of a substance decreases [49]. In addition, when the PEI loading was below 50 wt.%, Si-MCM-41-PEI showed a $10.5% weight loss at 100 C, which was higher than that of the pure PEI and can be also ascribed to the desorption of CO2 and moisture. When the PEI loading was 75 wt.%, the weight loss was only $3.5% at 100 C, similar with that of the pure PEI, which indicated that PEI was coated on the outer surface of the molecular sieve crystals and the composite material behaved like the pure PEI particles. The dierent weight loss at 100 C between the Si-MCM-41-PEI-50 and Si-MCM-41PEI-75 further conrmed that the PEI was loaded into the MCM-41s channels. The weight loss of SiMCM-41-PEI also indicated that there was no PEI loss during the preparation process. For example,

Weight (%)

36

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

the total weight loss for Si-MCM-41-PEI-50 was $56% at 600 C. If the adsorbed moisture or CO2 is excluded from the total weight, the PEI loading is calculated to be about 50 wt.%, which is in accordance with the designed PEI loading. 3.2. Adsorption performance of MCM-41-PEI 3.2.1. Eect of PEI loadings The inuence of PEI loading on the CO2 adsorption performance of Si-MCM-41-PEI was investigated in pure CO2 atmosphere at 75 C and the results are shown in Fig. 7. Before the PEI was loaded, the Si-MCM-41 support alone showed a CO2 adsorption capacity of 8.6 mg/g-adsorbent. The low adsorption capacity was caused by the weak interaction between CO2 and Si-MCM-41 at relatively high temperature. In order to strength the interaction between CO2 and Si-MCM-41, branched polymeric substances PEI with numerous CO2 -capturing sites, was loaded into the SiMCM-41 channels. Surprisingly, the PEI had no contribution on the CO2 adsorption capacity at low PEI loading. When the PEI loading was 5.0 wt.%, the CO2 adsorption capacity was only 7.7 mg/g-adsorbent, nearly identical to that of the SiMCM-41 support. When the PEI loading further increased, the adsorption capacity increased. The adsorption capacities were 19 mg/g-adsorbent and 69 mg/g-adsorbent for the PEI loading of 15 and 30 wt.%, respectively. At 50 wt.% PEI loading, the adsorption capacity was 112 mg/g-adsorbent,

which is higher than that of the pure PEI of 110 mg/g-adsorbent. The mesoporous molecular sieve of Si-MCM-41 showed a synergetic eect on the adsorption of CO2 by PEI. The highest adsorption capacity of 133 mg/g-adsorbent was obtained when the PEI loading was 75 wt.%. The desorption was complete for all the SiMCM-41-PEI adsorbents as well as the Si-MCM41 support. However, the desorption for pure PEI was slow and was not complete compared to the desorption time of the Si-MCM-41-PEI absorbents. The fast desorption of CO2 from the SiMCM-41-PEI absorbents can be explained by the high dispersion of the PEI into the Si-MCM-41 channels as shown by the XRD, N2 adsorption/ desorption and TGA characterizations. 3.2.2. Eect of preparation methods and preparation procedures Two methods, i.e., wet impregnation and mechanical mixing, were employed for the preparation of Si-MCM-41-PEI. For the mechanical mixing method, PEI was solidied, grinded into powder and mixed uniformly with Si-MCM-41 at liquid nitrogen temperature. The powder mixture was heated up to 70 C and held at that temperature for 16 h under reduced pressure (700 mmHg). The PEI loading was 50 wt.%. The adsorption performance of this absorbent was compared with that of the adsorbent prepared by the wet impregnation method in Table 1. The adsorption capacity was 99 mg/g-adsorbent for the adsorbent prepared by the mechanical mixing method (Si-MCM-41-PEI-50-M) and 112 mg/gadsorbent for the adsorbent prepared by the wet impregnation method (Si-MCM-41-PEI-50). The adsorption capacity of Si-MCM-41-PEI-50-M was lower than that of Si-MCM-41-PEI-50. The higher adsorption capacity for the wet impregnation method was possibly caused by the uniform dispersion of PEI into the Si-MCM-41 channels. However, if the PEI and the Si-MCM-41 were simply mechanically mixed in Si-MCM-41-PEI-50M, the adsorption capacity of MCM-41-PEI-50-M should be the linear sum of the adsorption capacity contributed from Si-MCM-41 and PEI, and was calculated to be 58.9 mg/g-adsorbent. The adsorption capacity of Si-MCM-41-PEI-50-M was

200 180 160 140 120 100 80 60 40 20 0 0 20 40 60 80

Adsorption capacity Desorption capacity

110 90 80 70 60 50 40 30 20 10 100

Fig. 7. The inuence of PEI loadings on the CO2 adsorption and desorption performance of Si-MCM-41-PEI.

Desorption capacity (%)

100

Adsorption capacity (mg/g-adsorbent)

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

37

higher than that value, which indicated that part of the PEI may also be loaded into the Si-MCM41 channels or be coated on the external surface of the Si-MCM-41 crystals during the thermal treatment. The desorption for both of the adsorbent was complete. Further, two dierent impregnation procedures were examined. In the rst set of experiments, the adsorbent was prepared by adding the PEI-methanol solution to Si-MCM-41 (PEI-30-Si-MCM41). Its adsorption performance was compared with that of the adsorbent prepared by adding SiMCM-41 to the PEI-methanol solution (Si-MCM41-PEI-30) in Table 1. The adsorption/desorption capacities of these two adsorbents were virtually similar, which indicates that this impregnation procedure did not inuence the dispersion of PEI into the Si-MCM-41 channels nor the adsorption performance of the adsorbent. In the second set of experiments, the adsorbent was prepared by the following procedure: Si-MCM41 was rst impregnated with 25 wt.% of PEI, then dried under vacuum at 70 C, and another 25 wt.% of PEI was then loaded. The adsorption performance of this adsorbent (Si-MCM-41-PEI-25-25) was compared with that of the adsorbent loaded with 50 wt.% PEI in one step (Si-MCM-41-PEI-50) in Table 1. The adsorption capacity of Si-MCM-41PEI-50 was higher than that of Si-MCM-41-PEI25-25, which indicates that one-step impregnation is a better method than two-step impregnation. Since PEI is hydrophilic in nature, the PEI impregnated in the rst step may adsorb the methanol and prevent the PEI from entering the MCM-41 channels in the next step of impregnation. Therefore, PEI may coat the external surface of the support crystals in the second impregnation and the adsorption capacity decreased. 3.2.3. Eect of solvent type and methanol/MCM-41 weight ratio Since the Si-MCM-41 that has been used so far is pure silica and is hydrophobic in nature, the polarity of solvent for the dissolution of PEI may inuence the dispersion of PEI into the Si-MCM41 channels. Two solvents, i.e., methanol and water, were used to examine the eect of solvent on adsorbent performance. The PEI loading for

both adsorbents was 50 wt.% and the adsorption results are shown in Table 1. The adsorption capacity was nearly the same for the two adsorbents, which indicated that these two solvents had limited inuence on the adsorption performance of the resulted adsorbent. The results also indicate that PEI is more hydrophobic than both the water and the methanol, and was therefore preferentially adsorbed into the Si-MCM-41 channels from the solution. Because methanol is easier to volatilize than water and the Si-MCM-41 structure is not stable under steam atmosphere at high temperature, methanol was selected as the solvent for the dissolution of PEI in the subsequent study. The inuence of methanol/Si-MCM-41 weight ratio on the adsorption performance of the resultant adsorbents was also examined and the adsorption results are shown in Table 1. From the adsorption results, it can be concluded that the adsorption capacity increased with increasing methanol/Si-MCM-41 weight ratio. The highest adsorption capacity of 126 mg/g-adsorbent was obtained when the methanol/Si-MCM-41 weight ratio was 8, which was 23% higher than that of the adsorbent prepared with a methanol/Si-MCM-41 weight ratio of 2. When the methanol/Si-MCM-41 weight ratio was 2, the Si-MCM-41 cannot be wetted well by the PEI/methanol solution. In this case, the PEI cannot be dispersed well and the adsorption capacity was nearly equal to that of the adsorbent prepared by mechanical mixing method. When the methanol/Si-MCM-41 weight ratio was increased to 4, MCM-41 can be wetted by the PEI/ methanol solution, allowing more PEI to be loaded into the Si-MCM-41 channels and thus increasing the adsorption capacity. 3.2.4. Eect of polyethyleneglycol as additive The inuence of PEG on the adsorption performance of Si-MCM-41-PEI was investigated. Fig. 8 compares the adsorption and desorption performance of Si-MCM-41-PEI-30 and Si-MCM41-PEI-30-PEG-20. The CO2 adsorption capacity and the CO2 adsorption/desorption rate for Si-MCM-41-PEI changed after adding the PEG to Si-MCM-41-PEI. The adsorption capacity of Si-MCM-41-PEI-30-PEG-20 was higher than that of Si-MCM-41-PEI-30. In addition, the

38
100

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

Weight change (mg/g-adsorbent)

90 80 70 60 50 40 30 20 10 0 0 50 100 150 200 250 300

Si-MCM-41-PEI-30-PEG-20 Si-MCM-41-PEI-30

Fig. 8. The eect of PEG on the CO2 adsorption/desorption performance of Si-MCM-41-PEI (PEI loading: 30 wt.%, PEG loading: 20 wt.%).

Fig. 9. The inuence of Si/Al ratio of the MCM-41 support on the CO2 adsorption performance of MCM-41 and MCM-41PEI (PEI loading: 50 wt.%).

adsorption/desorption rates for Si-MCM-41-PEI30-PEG-20 were faster than those of for Si-MCM41-PEI-30. Satyapal et al. [44] also found that adding PEG to the PEI/polymer adsorbent can accelerate the CO2 adsorption and desorption rates. The authors suggested that the enhancement of the adsorption and desorption rate is related to the preponderance of OH ions from the PEG molecules. They also suggested that the PEG attracts more water to the adsorbent due to the hydroscopic nature of the chemical. Since the adsorbate was pure CO2 in our experiment and there was no moisture in the gas, the enhanced adsorption capacity in this study must be from the CO2 alone. The increase in the adsorption capacity may result from modifying the reaction route after adding the PEG as discussed later. 3.2.5. Eect of the Si/Al ratio of the MCM-41 support The above results were derived using pure silica of MCM-41. To clarify whether aluminum incorporation into MCM-41 framework can aect the property of the MCM-41-PEI adsorbent, we further examined the inuence of the Si/Al ratio of the MCM-41 support on the adsorption performance of the resultant adsorbents. Fig. 9 shows the adsorption capacity of MCM-41 and MCM-41-PEI-50 with dierent support Si/Al ratio. The adsorption capacity was nearly identical for the MCM-41 support with dierent Si/Al ratio. However, the

adsorption capacity of MCM-41-PEI was inuenced by the Si/Al ratio of the MCM-41 support. The lower the Si/Al ratio of the MCM-41 support, the higher the adsorption capacity of the resultant adsorbent. The adsorption capacity of Al-MCM41-100-PEI-50 was 127 mg/g-adsorbent and was 12% higher than that of Si-MCM-41-PEI-50.

4. Discussions 4.1. Eect of mesoporous molecular sieve of MCM-41 In order to reveal any synergetic eects between MCM-41 and PEI on CO2 adsorption, the linear adsorption capacity and the synergetic adsorption gain were dened as follows:
Linear adsorption capacity mg adsorbate=g-adsorbent MCM-41 weight percentage in the adsorbent adsorption capacity of pure MCM-41 PEI weight percentage in the adsorbent adsorption capacity of pure PEI

Synergetic adsorption gain mg adsorbate=g-adsorbent adsorption capacity of the adsorbent linear adsorption capacity 2

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

39

The linear adsorption capacity and the synergetic adsorption gain were calculated and are shown in Fig. 10. For Si-MCM-41-PEI-5 and SiMCM-41-PEI-15, the synergetic adsorption gain was close to zero, indicating that the loading of PEI into the Si-MCM-41 channels had no eect on the increase in the CO2 adsorption capacity at these low PEI loadings. When the PEI loading was higher than 30 wt.%, a synergetic adsorption gain was observed, which indicated that PEI modied Si-MCM-41 began to act as a CO2 molecular basket. The highest synergetic gain was obtained at the PEI loading of 50 wt.%, while the synergetic adsorption gain decreased at the PEI loading of 75 wt.%. By correlating the adsorption performance and the pore structure of Si-MCM-41 and Si-MCM41-PEI, a simple model was proposed in Fig. 11 to account for the synergetic eect of Si-MCM-41 on the CO2 adsorption by PEI. Since the adsorption was operated at a relatively high temperature and the pore size of Si-MCM-41 is in the mesoporous region, the CO2 adsorption capacity of Si-MCM41 alone is low. When the PEI, which has branched chains with numerous CO2 adsorption sites of amine group, is loaded into the Si-MCM-41 channels, both the physical adsorption by capillary condensation and the chemical adsorption by reaction with PEI will contribute to the adsorption capacity. It may be expected that the adsorption capacity of Si-MCM-41-PEI would increase with increasing amount of PEI loaded. However, only
160 140

(A)

(B)

Fig. 11. Schematic diagram of PEI loaded in the mesoporous molecular sieve of MCM-41. (A) MCM-41 support; (B) low PEI loading; (C) high PEI loading; (D) extremely high PEI loading.

Adsorption capacity (mg/g-adsorbent)

120 100 80 60 40 20 0 -20 0 10

Synergetic adsorption gain Linear adsorption capacity Adsorption capacity

20

30

40

50

60

70

80

90

100

Fig. 10. Synergetic eect of Si-MCM-41 on the adsorption of CO2 by PEI as a function of PEI loading in MCM-41-PEI.

with high PEI loading did Si-MCM-41-PEI show a synergetic eect on the CO2 adsorption, indicating a strong interaction between Si-MCM-41 and PEI. When the PEI loading was low, the pore size of the adsorbent slightly decreased, from 2.75 nm for SiMCM-41 to 2.47 nm for Si-MCM-41-PEI-5. In this case, the PEI was mostly adsorbed on the inner pore wall of the Si-MCM-41 support (Fig. 11B). Both physical adsorption and chemical adsorption contribute to CO2 uptake in this case. The adsorption capacity was even smaller than that of the linear adsorption capacity. With increasing PEI loadings, the pore size further decreased. At the same time, because more PEI was loaded into the channels, the chemical adsorption of CO2 became much more dominant than that of with low PEI loadings. The physical adsorption on the unmodied pore wall of Si-MCM-41 (and the capillary condensation in the mesopore) became negligible when compared with the chemical adsorption force. In addition, mesoporous molecular sieve Si-MCM-41 had a synergetic eect on the adsorption of CO2 by PEI in the conned mesoporous pores [42]. Therefore, the CO2 adsorption capacity increased. The highest synergetic adsorption gain (Figs. 10 and 11C) was obtained

40

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

when the mesoporous channels were completely lled with PEI, i.e., about 50 wt.% of PEI loading. When the PEI loading further increased to 75 wt.% (Figs. 10 and 11D), the synergetic adsorption gain was smaller than that with PEI loading of 50 wt.%. Since the pore volume of Si-MCM-41 is about 1.0 ml/g and the density of PEI is about 1.0 g/ml, the largest amount of PEI that can be theoretically loaded into 1.0 g Si-MCM-41 is 1.0 ml, i.e., 50 wt.% PEI loading. For the adsorbent with PEI loading of 75 wt.%, theoretically, only one third of the PEI could be loaded into the Si-MCM-41 channels and two thirds of the PEI would be coated on the external surface of the molecular sieve particles. Interestingly, the highest adsorption gain weighed on PEI was also at the 50 wt.% PEI loading, which indicated that only when the PEI was loaded into the mesoporous molecular sieve did the Si-MCM-41 and PEI show a strong synergetic eect on the CO2 adsorption. When the PEI was coated on the external surface of the molecular sieve crystals, the synergetic eect was limited. In order to distinguish the synergetic eect of the PEI in the mesoporous channels and on the external surface of the molecular sieve crystals, the adsorption capacity of the PEI only in the molecular basket adsorbent was calculated as follows: PEI adsorption capacity mg adsorbate=g-PEI adsorption capacity of the adsorbent-MCM41 weight percentage in the adsorbent adsorption capacity of pure MCM-41 =PEI weight percentage in the adsorbent 3 The calculated PEI adsorption capacity is 215 mg/ g-PEI for Si-MCM-41-PEI-50 and 174 mg/g-PEI for Si-MCM-41-PEI-75. Since the Si-MCM-41PEI-75 was prepared by 25 wt.% of Si-MCM-41 and 75 wt.% of PEI, we can assume that 25 wt.% of the PEI is loaded into the pores of the Si-MCM41 and the other 50 wt.% of the PEI is coated on the external surface of the Si-MCM-41. If we assume that the PEI loaded into the Si-MCM-41 channel shows the same adsorption capacity as that of the Si-MCM-41-PEI-50, the adsorption capacity of the PEI that was coated on the external

surface of Si-MCM-41 for Si-MCM-41-PEI-75 can be calculated and is found to be only154 mg/gPEI. The adsorption capacity for the PEI coated on the external surface of the Si-MCM-41 was much lower than that of the Si-MCM-41-PEI-50 (215 mg/g-PEI) when the PEI was fully lled in the channels of the Si-MCM-41, which conrmed that only when the PEI was loaded into the channels of the mesoporous molecular sieve did the Si-MCM41 showed the highest synergetic eect and proved that the molecular basket concept resulted in the signicant improvement on the CO2 adsorption by PEI. The adsorption capacity for the PEI coated on the external surface of the crystal was slightly greater than that of the pure PEI of 110 mg/g-PEI. When the PEI was coated on the crystal surface, more adsorption sites will expose to the adsorbate. Therefore, the adsorption capacity will be higher than that of the bulk PEI. This can also be veried by the CO2 adsorption capacity of PEI coated on other high-surface-area materials such as silica gel. When the PEI was coated on a high-surface-area silica gel (550 m2 /g) and the PEI loading was 50%, the CO2 adsorption capacity was only 156 mg/gPEI, slightly higher than that of the bulk PEI and much lower than that of the molecular basket adsorbent with the same PEI loading, which also veried that only when the PEI was loaded into the channels of the mesoporous molecular sieve did the Si-MCM-41 showed the highest synergetic eect on the adsorption of CO2 . The CO2 adsorption capacity of SilicaGel-PEI was nearly identical to that of the calculated CO2 adsorption capacity on the external surface of the MCM-41 crystals. Satyapal et al. [44] also observed the same phenomenon when they coated the PEI on the high-surface-area solid polymethyl methacrylate polymeric support. By employing the molecular basket concept, the CO2 adsorption and desorption kinetics can also be improved signicantly. Fig. 12 compares the CO2 adsorption curves of Si-MCM-41-PEI-50, Si-MCM-41-PEI-75 and PEI in the rst minute (60 s) of adsorption. The CO2 adsorption rate decreases in the following order: Si-MCM-41-PEI50 > Si-MCM-41-PEI-75 > PEI. If the linear part of the adsorption curve was selected to calculate

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

160 140 120 100 80 60 40 20 0 0 6 12 18 24 30 36 42 48 54 60
Si-MCM-41-PEI-50 y = 5.60x - 19.66 2 R = 1.00 y = 3.23x - 15.84 Si-MCM-41-PEI-75 R2 = 1.00 PEI y = 2.67x - 16.87 R2 = 1.00

41

Fig. 12. Comparison of the CO2 adsorption kinetics of Si-MCM-41-PEI-50, Si-MCM-41-PEI-75 and PEI.

the CO2 adsorption rate, the CO2 adsorption rate was 5.6 mg/s.g-PEI, 3.23 mg/s.g-PEI and 2.67 mg/ s.g-PEI for Si-MCM-41-PEI-50, Si-MCM-41-PEI75 and PEI, respectively. The CO2 adsorption rate for Si-MCM-41-PEI-50 was much faster than that of the Si-MCM-41-PEI-75 and PEI, which conrmed that only when the PEI was loaded into the channels of the mesoporous molecular sieve did the Si-MCM-41 showed the highest synergetic ef-

fect and proved that the molecular basket concept resulted in the signicant improvement on the CO2 adsorption by PEI. Coating PEI on the high surface area materials only slightly increased the CO2 adsorption rate. The CO2 desorption rate showed the same trend as the adsorption. The CO2 desorption rate decrease in the following order: SiMCM-41-PEI-50 > Si-MCM-41-PEI-75 > PEI. The CO2 desorption was complete for Si-MCM-41PEI-50 and Si-MCM-41-PEI-75 in 150 min, whereas only 56% of the adsorbed CO2 was desorbed when the same desorption time as the Si-MCM-41-PEI adsorbent was used. Compared with conventional adsorbents, the molecular basket showed better CO2 adsorption performance at relatively high temperature. Table 2 lists the CO2 adsorption performance of zeolite, activated carbon, PEI/polymer composite and molecular basket adsorbent. The molecular basket adsorbent shows superior CO2 selectivity in terms of very high CO2 /N2 ratios (>1000 for MCM-41-PEI-50 vs 26 for aluminosilicates and activated carbons). The high selectivity for CO2 is particularly important when it was applied in the separation of CO2 from ue gas [51].

Table 2 Comparison of CO2 adsorption performance of molecular basket adsorbent and other adsorbents Adsorbents Temp (C) 25 75 75 75 75 25 75 75 75 75 25 25 25 21.5 75 20 25 75 $50 Pressure (atm) 1 1 0.149 1 1 1 1 0.149 1 1 1 1 1 1 1 1 1 1 0.02 Adsorption capacity (mg CO2 /g adsorbent) 27.3 8.6 6.3 7.6 7.5 32.9 112 89.2 127 121 168 135 110 108 40 88 140.8 78.1 $40 CO2 /N2 or CO2 / CH4 selectivity 2.9 (CO2 /N2 ) >1000 (CO2 /N2 ) $2 (CO2 /CH4 ) $2 (CO2 /CH4 ) $ 2 (CO2 /CH4 ) $1.9 (CO2 /CH4 ) Ref.

Adsorption capacity (mg /g-PEI)

Si-MCM-41 Si-MCM-41 Si-MCM-41 Al-MCM-41-100 Al-MCM-41-500 Si-MCM-41-PEI-50 Si-MCM-41-PEI-50 Si-MCM-41-PEI-50 Al-MCM-41-100-PEI-50 Al-MCM-41-500-PEI-50 Zeolite 13 X Zeolite 4A Activated carbon Norit RBI activated carbon Norit RBI activated carbon Activated carbon Norit RBI activated carbon PEI-silica gel PEI-polymer

This This [51] This This This This [51] This This [13] [13] [13] [28] [28] [24] [29] This [44]

study study study study study study study study

study

42

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

4.2. Eect of preparation method The adsorption capacity of the adsorbent prepared by the wet impregnation method was higher than that of the adsorbent prepared by the mechanic mixing method. For Si-MCM-41-PEI-50, the mesoporous pores were completely lled with PEI and high adsorption capacity was obtained because most of the PEI is uniformly dispersed into the Si-MCM-41 channels. In the case of Si-MCM41-PEI-50-M, whose morphology was similar to that of Si-MCM-41-PEI-75, it is likely that only some of the PEI was loaded into the support channels during the thermal treatment and the remaining PEI may coat the outer surface of the molecular sieve particles. As we discussed above, only when the MCM-41 channels are lled with PEI does the adsorbent show the highest synergetic eect on the adsorption of CO2 . When the PEI was coated on the external surface of the molecular sieve crystals, the synergetic eect was much lower than that when the PEI was in the MCM-41 channels. Hence, the low adsorption capacity of SiMCM-41-PEI-50-M can be ascribed to the inhomogeneous dispersion of the PEI into the channels. The low adsorption capacity for the adsorbent SiMCM-41-PEI-50-S2 and Si-MCM-41-PEI-25-25 can also be explained for the same reason. The homogeneous dispersion of PEI into the MCM-41 channels is critical for the preparation of this novel CO2 molecular basket adsorbent. 4.3. Eect of PEG addition CO2 adsorption capacity increased after adding 20 wt.% PEG into MCM-41-PEI-30. Chaee and co-workers [50] reported that the ratio of CO2 molecular per available N atom in the presence of hydroxyl group is approximately twice that without the hydroxyl group. They suggested that the CO2 chemical adsorption mechanism of the amine changed in the presence of hydroxyl group. Without the hydroxyl group, the formation of carbamate is favored in the manner shown in Eqs. (4)(6). Two moles of amine groups react with 1 mol of CO2 molecule. CO2 2RNH2 ( NH R2 NCOO ) 4 4

) CO2 2R2 NH ( R2 NH R2 NCOO 2 ) CO2 2R3 N ( R4 N R2 NCOO

5 6

In the presence of hydroxyl groups, the formation of carbamate type zwitterions is stabilized in a manner depicted in Eq. (7). One molar amine groups react with one mole CO2 molecule. Therefore, the adsorption capacity increased with the same amine group.

Although PEG cannot adsorb CO2 , the presence of OH groups in PEG may also inuence the chemical adsorption mechanism. In the presence of hydroxyl group, the formation of carbamate type zwitterions may be promoted and, therefore, amine group will adsorb more CO2 molecular. The change of adsorption mechanism is also suggested by the adsorption and desorption rate as shown in Fig. 9. The adsorption and desorption rates of SiMCM-41-PEI-30-PEG-20 increased after adding the PEG into the Si-MCM-41-PEI-30. 4.4. Eect of Si/Al ratio of MCM-41 support The Si/Al ratio of the MCM-41 support inuences the adsorption performance of the resultant MCM-41-PEI adsorbent. The lower the Si/Al ratio of the MCM-41 support, the higher the CO2 adsorption capacity of the MCM-41-PEI adsorbent. When the Si/Al ratio of the MCM-41 support was 100, the highest CO2 adsorption capacity was obtained and was 127 mg/g-adsorbent, which correspond to a CO2 adsorption capacity of 246 mg/ g-PEI. In order to clarify the eect of the Si/Al ratio of the MCM-41 support, the pore structures were measured for the MCM-41 with dierent Si/ Al ratio and their corresponding MCM-41-PEI adsorbent with PEI loading of 50 wt.%. The results are shown in Figs. 1315, respectively. For the MCM-41 support, the pore size and pore volume increase with the decrease in Si/Al ratio. After the PEI was loaded, the pore volume decreased. The

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

43

Fig. 13. The inuence of Si/Al ratio on the surface area and pore volume of MCM-41.

0.45 0.4 0.35

Si/Al=100 Si/Al=500 All silica

0.3 0.25 0.2 0.15 0.1 0.05 0 10 15 20 25 30 35 40 45 50 55 60

Fig. 14. The inuence of Si/Al ratio on the pore size distribution of MCM-41.

Fig. 15. Comparison of the BET surface area and the pore volume of the MCM-41-PEI-50 adsorbents with dierent Si/Al ratio of MCM-41.

lower the Si/Al ratio of the MCM-41 support, the higher the residual pore volume of the MCM-41PEI adsorbent. It was possibly caused by the large pore volume of the MCM-41 support with low Si/ Al ratio. Since the PEI is a branched polymer, its molecular size is relatively large compared to the simple molecule, e.g., N2 . Each of the pore channels of the MCM-41 support can only accommodate a limited number of PEI molecules. The pore volume of Si-MCM-41-PEI-30 was 0.23 ml/g, considerably smaller than that of the theoretical value of 0.4 g/ml when we assume the PEI molecules pack in a compact form in the channels as they would in a neat PEI liquid. In the pore channel of mesoporous molecular sieve, the PEI molecules cannot pack as compact as in the neat liquid or solid PEI, as suggested by the pore volume of Si-MCM-41-PEI-30 vs that of Si-MCM-41. Therefore, although the maximum PEI loading in the pore channel of Si-MCM-41 would be 50 wt.%, a part of the PEI will be inevitably coated on the external surface of the molecular sieve crystals. The adsorption capacity for the PEI coated on the external surface was much lower than that when the PEI was lled in the channels of the MCM-41. When the aluminum was incorporated into the framework of MCM-41, the pore volume of MCM-41 increased allowing more PEI loading into the channels of the mesoporous molecular sieve. The lower the Si/Al ratio of MCM-41, the higher the pore volume. For Al-MCM-41-100PEI-50, the pores were even not completely lled with PEI as reected by the residual pore volume (Fig. 15). This can explain why the higher CO2 adsorption capacity was obtained with lower Si/Al ratio of MCM-41 support. It should be noted that the total volume of adsorbed CO2 in MCM-41-PEI-50 was 114 ml (STP)/g-PEI for Si-MCM-41-PEI-50. Since the adsorption mechanism is chemical adsorption for the molecular basket adsorbent, CO2 reacts with the PEI and is not adsorbed in the free space of the pores. In addition, the pore size, pore volume, and surface area were measured at liquid nitrogen temperature. With the increase of temperature, the polymer becomes more exible and more CO2 anity sites will be exposed to the CO2 and the adsorption capacity increases. The pore size of

Dv(d) (ml//g)

44

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945

the MCM-41-PEI-50 should increase to some extent with increasing temperature (e.g., 75 C for the CO2 adsorption compared to liquid nitrogen temperature for pore size measurement). Therefore, the diusion of CO2 in the MCM-41 channel is faster at high temperature than at low temperature.

anol/MCM-41 weight ratio for adsorbent preparation can lead to higher CO2 adsorption capacity of MCM-41-PEI. 5. Adding PEG into the MCM-41-PEI adsorbent can increase not only the CO2 adsorption and desorption rate, but also the CO2 adsorption capacity.

5. Conclusions 1. Novel CO2 molecular basket adsorbents based on polyethylenimine (PEI)-modied mesoporous molecular sieve of MCM-41 type (MCM41-PEI) have been successfully developed. The highest CO2 adsorption capacity of 246 mg/gPEI was obtained, which is 30 times higher than that of the MCM-41 and more than twice that of the pure PEI. 2. When the PEI loading was higher than 30 wt.%, the mesoporous molecular sieve of MCM-41 showed a synergistic eect on the adsorption of CO2 by PEI. The highest synergistic eect was obtained when the PEI loading was 50 wt.%. When the PEI loading was 75 wt.%, the PEI was coated on the outer surface of the molecular sieve crystals and the synergistic eect decreased. The experiment results showed that, although the dispersion of PEI on the high surface area materials could increase the adsorption capacity, it was the synergetic eect between PEI and MCM-41 pore channels (CO2 molecular basket) that contributes to major increase in the CO2 adsorption capacity. 3. The uniform dispersion of the PEI into the pore channels of the MCM-41 support is critical for the preparation of this novel molecular basket adsorbent. The MCM-41-PEI adsorbents prepared by the one-step impregnation method showed a higher CO2 adsorption capacity than the adsorbent prepared by the mechanical mixing method and the two-step impregnation method. 4. Proper incorporation of Al into the MCM-41 framework in the synthesis step can signicantly enhance the performance of the nal MCM-41PEI adsorbents. The relatively lower Si/Al ratio of MCM-41 support and relatively higher meth-

Acknowledgements Funding for the work was provided by the US Department of Defense (via an interagency agreement with the US Department of Energy) and the Commonwealth of Pennsylvania under cooperative agreement no. DE-FC22-92PC92162.

References
[1] M.M. Maroto-Valer, C. Song, Y. Soong (Eds.), Environmental Challenges and Greenhouse Gas Control for Fossil Fuel Utilization in the 21st Century, Kluwer Academic/ Plenum Publishers, New York, 2002, p. 447. [2] C. Song, A.M. Ganey, K. Fujimoto (Eds.), CO2 Conversion and Utilization, ACS Symp. Series, vol. 809, American Chemical Society (ACS), Washington, DC, 2002, 448 pp. [3] C. Azar, H. Rodhe, Science 276 (1997) 18181819. [4] H. Herzog, E. Drake, E. Adams, CO2 Capture, Reuse and Storage Technologies for Mitigating Global Climate Change, Report no. DOE/DE-AF22-96PC01257. [5] J.A. Ruether, FETC Programs for Reducing Greenhouse Gas Emissions, Report no. DOE/EFTC-98/1058, US Department of Energy, Pittsburgh, PA, 1999. [6] DOE/OS-FE. Carbon Sequestration. State of the Science, Oce of Science and Oce of Fossil Energy, US Department of Energy, Washington, DC, 1999. [7] US Department of Energy, Carbon SequestrationResearch and Development, 1999. http://www.fe.doe.gov/ coal_power/sequestration/reports/rd/index.html. [8] C. Song, Am. Chem. Soc. Symp. Ser. 809 (2002) 230. [9] C. Song, Chem. Innovat. 31 (2001) 2126. [10] Y.H. Ma, C. Mancel, AIChE J. 18 (1972) 11481153. [11] Y.H. Ma, A.J. Roux, AIChE J. 19 (1973) 10551059. [12] D.T. Hayhurst, Chem. Eng. Commun. 4 (1980) 729735. [13] R.V. Siriwardane, M.S. Shen, E.P. Fisher, J.A. Poston, Energy Fuels 15 (2001) 279284. [14] Y. Takamura, S. Narita, J. Aoki, S. Hironaka, S. Uchida, Sep. Purif. Technol. 24 (2001) 519528. [15] M. Ishibashi, H. Ota, N. Akutsu, S. Umeda, M. Tajika, J. Izumi, A. Yasutake, T. Kabata, Y. Kageyama, Energy Convers. Mgmt. 37 (1996) 929933. [16] F. Dong, H.M. Lou, A. Kodama, M. Goto, T. Hirose, Sep. Purif. Technol. 16 (1999) 159166.

X. Xu et al. / Microporous and Mesoporous Materials 62 (2003) 2945 [17] R.W. Triebe, F.H. Tezel, Gas Sep. Purif. 9 (1995) 223230. [18] A.Y. Khodakov, L.V.C. Rees, Gas Sep. Purif. 9 (1995) 253257. [19] V.R. Choudhary, S. Mayadevi, A.P. Singh, J. Chem. Soc. Faraday Trans. 91 (1995) 29352944. [20] G. Calleja, A. Jimenez, J. Pau, L. Dominguezl, P. Perez, Gas Sep. Purif. 8 (1994) 247256. [21] G. Calleja, J. Pan, J.A. Calles, J. Chem. Eng. Data 43 (1998) 9941003. [22] M. Katoh, T. Yoshikawa, K. Tomonama, T. Tomida, J. Colloid Interf. Sci. 226 (2000) 145150. [23] S.W. Rutherford, D.D. Do, Carbon 38 (2000) 13391350. [24] K. Berlier, M. Frere, J. Chem. Eng. Data 42 (1997) 533537. [25] F. Foeth, H. Bosch, A. Sjostrand, G. Aly, T. Reith, Sep. Sci. Technol. 31 (1996) 2138. [26] J.H. Chen, D.S.H. Wong, C.S. Tan, R. Subramanian, C.T. Lira, M. Orth, Ind. Eng. Chem Res. 36 (1997) 28052815. [27] M. Heuchel, G.M. Davies, E. Buss, N.A. Seaton, Langmuir 15 (1999) 86958705. [28] R. van der Vaart, C. Huiskes, H. Bosch, T. Reith, Adsorption 6 (2000) 311323. [29] F. Dreisbach, R. Staudt, J.U. Keller, Adsorption 5 (1999) 215227. [30] M. Anand, J. Hufton, S. Mayorga, S. Nataraja, S. Sircar, T. Ganey, APCI report for DOE, 1995. [31] H. Song, K. Lee, Sep. Sci. Technol. 33 (1998) 22392257. [32] B.K. Na, K.K. Koo, H.M. Eum, H. Lee, H.K. Song, Korean J. Chem. Eng. 18 (2001) 220227. [33] T.D. Burchell, R.R. Judkins, M.R. Rogers, A.M. Williams, Carbon 35 (1997) 12791294. [34] F. Foeth, M. Andersson, H. Bosch, G. Aly, T. Reith, Sep. Sci. Technol. 29 (1994) 93118. [35] H.P. Huang, Y. Shi, W. Li, S.G. Chang, Energy Fuels 15 (2001) 263268. [36] Z. Yong, V.G. Mata, A.E. Rodrigues, Adsorption 7 (2001) 4150.

45

[37] Y. Ding, E. Alpay, Chem. Eng. Sci. 50 (2000) 34613474. [38] Z. Yong, V.G. Mata, A.E. Rodrigues, Ind. Eng. Chem. Res. 40 (2001) 204209. [39] P.R. Pereira, J. Pires, M.B. Carvalho, Langmuir 14 (1998) 45844588. [40] Z. Yong, V.G. Mata, A.E. Rodrigues, J. Chem. Eng. Data 45 (2000) 10931095. [41] X.C. Xu, J.M. Andresen, C. Song, B.G. Miller, A.W. Scaroni, Proceeding of the 18th Annual International Pittsburgh Coal Conference, New South Wales, Australia, 37 December, 2001 (Paper no. 17-2). [42] X.C. Xu, C. Song, J.M. Andresen, B.G. Miller, A.W. Scaroni, Energy Fuels 16 (2002) 14631469. [43] X.C. Xu, J.M. Andresen, C. Song, B.G. Miller, A.W. Scaroni, Preprints of SymposiaAmerican Chemical Society, Division of Fuel Chemistry 47 (1) (2002) 6768. [44] S. Satyapal, T. Filburn, J. Trela, J. Strange, Energy Fuels 15 (2001) 250255. [45] K.M. Reddy, C. Song, Catal. Lett. 36 (1996) 103109. [46] K.M. Reddy, C. Song, Stud. Surf. Sci. Catal. 117 (1998) 291299. [47] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck, Nature 359 (1992) 710712. [48] J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T.W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B. Higgins, J.L. Schlenker, J. Am. Chem. Soc. 114 (1992) 1083410843. [49] P. Zeng, S. Zajac, P.C. Clapp, J.A. Rifkin, Mater. Sci. Eng. A-Struct. 252 (1998) 301306. [50] W.S. Delaney, G.P. Knowles, A.L. Chaee, Preprints of SymposiaAmerican Chemical Society, Division of Fuel Chemistry, 47 (1) (2002) 6566. [51] X.C. Xu, C. Song, J.M. Andresen, B.G. Miller, A.W. Scaroni, Proceeding of the Nineteenth Annual International Pittsburgh Coal Conference, Pittsburgh, Pennsylvania, USA, 2327 September, 2002 (Paper no. 39-2).