Adsorptive Separation of Isobutene and Isobutane on Cu3(BTC)2

The metal organic framework material Cu3(BTC)2 (BTC = 1,3,5-benzenetricarboxylate) has been synthesized using different routes: under solvothermal ...
8 downloads 0 Views 323KB Size
8634

Langmuir 2008, 24, 8634-8642

Adsorptive Separation of Isobutene and Isobutane on Cu3(BTC)2 Martin Hartmann,* Sebastian Kunz, Dieter Himsl, and Oliver Tangermann AdVanced Materials Science, Institute of Physics, UniVersity of Augsburg, UniVersita¨tsstr. 1, D-86159 Augsburg, Germany

Stefan Ernst and Alex Wagener Fachbereich Chemie, Technische Chemie, TU Kaiserslautern, 67663 Kaiserslautern, Germany ReceiVed March 19, 2008. ReVised Manuscript ReceiVed May 9, 2008 The metal organic framework material Cu3(BTC)2 (BTC ) 1,3,5-benzenetricarboxylate) has been synthesized using different routes: under solvothermal conditions in an autoclave, under atmospheric pressure and reflux, and by electrochemical reaction. Although the compounds display similar structural properties as evident from the powder X-ray diffraction (XRD) patterns, they differ largely in specific surface area and total pore volume. Thermogravimetric and chemical analysis support the assumption that pore blocking due to trimesic acid and/or methyltributylammoniummethylsulfate (MTBS) which has been captured in the pore system during reaction is a major problem for the electrochemically synthesized samples. Isobutane and isobutene adsorption has been studied for all samples at different temperatures in order to check the potential of Cu3(BTC)2 for the separation of small hydrocarbons. While the isobutene adsorption isotherms are of type I according to the IUPAC classification, the shape of the isobutane isotherm is markedly different and closer to type V. Adsorption experiments at different temperatures show that a somewhat higher amount of isobutene is adsorbed as compared to isobutane. Nevertheless, the differential enthalpies of adsorption are only different by about 5 kJ/mol, indicating that a strong interaction between the copper centers and isobutene does not drive the observed differences in adsorption capacity. The calculated breakthrough curves of isobutene and isobutane reveal that a low pressure separation is preferred due to the peculiar shape of the isobutane adsorption isotherms. This has been confirmed by preliminary breakthrough experiments using an equimolar mixture of isobutane and isobutene.

1. Introduction The design and synthesis of novel materials with tailor-made properties for applications in catalysis or separation by selective adsorption remain an area of intensive research. Due to the quest for improved materials for several applications, metal organic frameworks (MOFs) have received considerable attention in recent years.1–3 The combination of bi- or trifunctional linkers with coordinating transition metal ions or clusters is an attractive method for the synthesis of coordination polymers which offer a porous, clearly defined, and highly symmetric structure. MOFs have already been shown to offer potential in gas storage,4 gas and liquid separation,5 drug release,6 and heterogeneous catalysis.7 However, most studies so far have focused on gas adsorption with particular emphasis on molecular hydrogen storage.8–13 * To whom correspondence should be addressed: Telephone: +49-821598-3559. Fax: +49-821-598-3227. E-mail: [email protected]. (1) Eddaoudi, M.; Moler, D. B.; Li, H. L.; Chen, B. L.; Reineke, T. M.; O’Keeffe, M.; Yaghi, O. M. Acc. Chem. Res. 2001, 34, 319. (2) Kitagawa, S.; Kitaura, R.; Noro, S. Angew. Chem., Int. Ed. 2004, 43, 2334. (3) Rowsell, J. L. C.; Yagi, O. M. Microporous Mesoporous Mater. 2004, 73, 3. (4) Panella, B.; Hirscher, M.; Pu¨tter, H.; Mu¨ller, U. AdV. Funct. Mater. 2006, 16, 520. (5) Alaerts, L.; Kirschhock, C. E. A.; Maes, M.; van der Veen, M. A.; Finsy, V.; Depla, A.; Martens, J. A.; Baron, G. V.; Jacobs, P. A.; Denayer, J. F. M.; De Vos, D. E. Angew. Chem., Int. Ed. 2007, 46, 4293. (6) Horcajada, P.; Serre, C.; Vallet-Regi, M. A.; Sebban, M.; Taulelle, F.; Ferey, G. Angew. Chem. 2006, 118, 1. (7) Alaerts, L.; Seguin, E.; Poelman, H.; Thibault-Starzyk, F.; Jacobs, P. A.; De Vos, D. E. Chem.sEur. J. 2006, 12, 7353. (8) Yang, Q.; Zhong, C. ChemPhysChem. 2006, 7, 1417. (9) Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Wachter, J.; O’Keefe, M.; Yaghi, O. M. Science 2002, 295, 469. (10) Skoulidas, A. I. J. Am. Chem. Soc. 2004, 126, 1356. (11) Vishnyakov, A.; Ravikovitch, P. I.; Neimark, A. V.; Bu¨1ow, M.; Wang, Q. M. Nano Lett. 2003, 3, 713.

Cu3(BTC)2 (also known as HKUST-1, BTC ) 1,3,5-benzenetricarboxylate) has been studied intensively since the first report by Chui et al. in 1999.14 The main structural feature of Cu3(BTC)2 is a copper dimer with a copper-copper distance of 0.263 nm. Twelve carboxylate oxygen atoms from the two BTC ligands bind to the four coordination sites of each of the three Cu2+ ions. These so-called paddle-wheel units form a face centered crystal lattice with Fm3jm symmetry which possesses a three-dimensional channel system with a bimodal pore size distribution. The larger (hydrophilic) pores (diameter ca. 0.9 nm) are formed from 12 paddle-wheel subunits forming a cuboctahedron. In addition to the carboxylate ligands, one water molecule is coordinated to the copper center pointing towards the center of the pore. The coordinated water molecules are removed in vacuum, creating accessible Cu2+ centers which can act as Lewis acid sites.15 A second pore system of tetrahedron-shaped side pockets (diameter ca. 0.5 nm) formed by four benzene rings is accessible from the large pores through windows with a diameter of 0.35 nm.16 There have been continuous efforts to improve the synthesis and activation of Cu3(BTC)2.17,15 The favored route to the synthesis of high quality Cu3(BTC)2 still is the hydrothermal synthesis under autogenous pressure. However, this synthesis procedure requires long reaction times, high temperatures, and moderate pressures. Moreover, it should be mentioned that (12) Krawiec, P.; Kramer, M.; Sabo, M.; Kunschke, R.; Fro¨de, D.; Kaskel, S. AdV. Eng. Mater. 2006, 8, 293. (13) Panella, B.; Hirscher, M.; Pu¨tter, H.; Mu¨ller, U. AdV. Funct. Mater. 2006, 16, 520. (14) Chui, S. S.-Y.; Lo, S. M.-F.; Charmant, J. P. H.; Orpen, A. G.; Williams, I. D. Science 1999, 283, 1148. (15) Schlichte, K.; Kratzke, T.; Kaskel, S. Microporous Mesoporous Mater. 2004, 73, 81. (16) Yang, Q.; Zhong, C. J. Phys. Chem. B 2006, 110, 17776. (17) Liu, J.; Culp, J. T.; Natesakhawat, S.; Bockrath, B. C.; Zande, B.; Sankar, S. G.; Garberoglio, G.; Johnson, J. K. J. Phys. Chem. C 2007, 111, 9305.

10.1021/la8008656 CCC: $40.75  2008 American Chemical Society Published on Web 07/09/2008

Separation of Isobutene and Isobutane on Cu3(BTC)2

hydrothermal syntheses are also problematic with respect to scaleup. Alternative synthesis procedures have been attempted in order to find routes to faster and more efficient syntheses under even less drastic conditions compared to the common hydrothermal syntheses. At the moment, microwave synthesis,18 electrochemical synthesis,19 and syntheses under reflux and stirring seem to be suitable candidates for fast and practical synthesis methods. There are only few reports on the electrochemical synthesis20 as well as the ambient pressure route under reflux and stirring.12,21 Moreover, it has been shown that the activation of the MOF material is important for obtaining materials with high specific surface areas and pore volumes.17 In addition to routine characterization of the copper MOF by nitrogen adsorption at 77 K, adsorption of the noble gases argon,10,11,22 xenon,23 and krypton as well as a continuous separation of Kr and Xe were studied.20 While the smaller noble gases argon and xenon are readily adsorbed on Cu3(BTC)2, krypton is too large to enter the MOF, which is the basis for efficient separation of Kr and Xe.20 Cu3(BTC)2 has been reported to be a suitable candidate for the adsorption of hydrogen,4,12,16,17 carbon dioxide,16 and NO21 as well as the adsorption and separation of smaller hydrocarbons. The adsorption of hydrocarbons on Cu3(BTC)2 is so far limited to methane, ethane, ethene, propane, and propene.16,21,24 It is shown that alkene adsorption is preferred over alkane adsorption. The continuous separation of propene and propane has been studied by Wagener et al.24 A separation factor Rpropene/propane of ∼2 has been achieved. The preferential adsorption of propene in Cu3(BTC)2 has been provisionally attributed to the stronger interaction of the olefin with copper(II) species. The separation of olefins from the corresponding alkanes is an important separation problem at the interface between refinery and petrochemistry.25,26 The propane/propene separation is a very energy intensive process due to the close boiling points (-42.1 °C vs -47.7 °C) and the required number of separation stages for low temperature distillation. A suitable alternative is the selective adsorption on microporous adsorbents such as zeolites (e.g., ITQ-12,27 Si-Chabasite,28 and AlPO4-1429) or metal organic framework materials.30 In this work, we have concentrated on Cu3(BTC)2, since this MOF possesses a free coordination site (after activation) at each copper center, which can be used for the reversible coordination of molecules. We have explored the influence of the synthesis route, namely, hydrothermal, ambient pressure, and electrochemical synthesis, on the adsorptive properties of Cu3(BTC)2. It is shown that a novel ambient pressure synthesis of Cu3(BTC)2 using ethanol as the solvent yields the material with the highest (18) Ni, Z.; Masel, R. I. J. Am. Chem. Soc. 2006, 128, 12394. (19) Mu¨ller, U.; Pu¨tter, H.; Hesse, M.; Wessel, H. WO 2005/049892, assigned to BASF AG, 2005. (20) Mu¨ller, U.; Schubert, M.; Teich, F.; Puetter, H.; Schierle-Arndt, K.; Pastre, J. J. Mater. Chem. 2006, 16, 626. (21) Wang, Q. M.; Shen, D.; Bu¨low, M.; Lau, M. L.; Deng, S.; Fitch, F. R.; Lemcoff, N. O.; Semanscin, J. Microporous Mesoporous Mater. 2002, 55, 217. (22) Lee, J. Y.; Li, J.; Jagiello, J. J. Solid State Chem. 2005, 178, 2527. (23) Bo¨hlmann, W.; Po¨ppl, A.; Sabo, M.; Kaskel, S. J. Phys. Chem. B 2006, 110, 20177. (24) Wagener, A.; Schindler, M.; Rudolphi, F.; Ernst, S. Chem.-Ing.-Tech. 2007, 79, 851. (25) Eldridge, R. B. Ind. Eng. Chem. Res. 1993, 32, 2208. (26) Ruthven, D. M.; Reyes, S. C. Microporous Mesoporous Mater. 2007, 104, 59. (27) Olson, D. H.; Yang, X.; Camblor, M. A. J. Phys. Chem. B 2004, 108, 11044. (28) Olson, D. H.; Camblor, M. A.; Villasescusa, L. A.; Ku¨hl, G. H. Microporous Mesoporous Mater. 2004, 67, 27. (29) Padin, J.; Rege, S. U.; Yang, R. T.; Cheng, L. S. Chem. Eng. Sci. 2000, 55, 4525. (30) Pan, L.; Olson, D. H.; Ciemnolonski, L. R.; Heddy, R.; Li, J. Angew. Chem., Int. Ed. 2006, 45, 616.

Langmuir, Vol. 24, No. 16, 2008 8635

specific pore volume without using any specific activation procedure. We have chosen isobutene and isobutane as model adsorptives which possess similar sizes and very close boiling points (-6.9 vs -11.7 °C) but rather different electronic properties due to the double bond of isobutene. Isobutene and isobutane possess a moderate dipole moment of 0.531 and 0.13 D,32 respectively. Moreover, the adsorption isotherms were recorded at different temperatures in order to calculate the isosteric heat of adsorption. We have found that the synthesis route has a significant effect on the adsorption and the related continuous separation of hydrocarbons such as isobutane and isobutene. It is moreover observed that the interaction of isobutene and isobutane with the Cu2+ ions is relatively weak. However, the significantly different shape of the respective adsorptions isotherms may allow an effective separation of alkanes and olefins over Cu3(BTC)2.

2. Experimental Section The synthesis of Cu3(BTC)2 was carried out using different procedures as described below: (1) Hydrothermal synthesis: A total of 1.75 g of Cu(NO3)2 · 3H2O and 0.84 g of trimesic acid (H3BTC) were dissolved in 48 mL of EtOH. The suspension was filled into a Teflon container and placed into an autoclave. The reaction was performed under autogenous pressure for 14 h at 120 °C (sample A). In a second experiment, 1.75 g of Cu(NO3)2 · 3H2O and 0.84 g of trimesic acid were dissolved in a mixture of 24 mL of water and 24 mL of EtOH. The suspension was heated under autogenous pressure for 14 h at 120 °C in a Teflonlined autoclave (sample B). At the end of the reaction, the autoclave was cooled down to room temperature, and the resulting blue powder was filtered off and washed once with water and then a second time with EtOH. Thereafter, the sample was dried under vacuum at 100 °C for 12 h and then kept under nitrogen atmosphere until further use. (2) Synthesis under ambient pressure: A solution of 0.84 g of trimesic acid and 1.75 g of Cu(NO3)2 · 3H2O in 50 mL of EtOH was refluxed under rigorous stirring. After a few hours, a blue solid started to precipitate. The synthesis mixture was kept under these conditions for 48 h and was subsequently cooled to room temperature. The obtained blue powder was recovered by filtration, washed once with water and then a second time with EtOH. Thereafter, the sample was dried under vacuum at 100 °C for 12 h and then kept under dry nitrogen until further use (sample C). (3) Electrochemical synthesis: The electrochemical synthesis was adopted from the patent literature and optimized in our laboratory. A typical synthesis was performed as follows: A total of 0.5 g of trimesic acid was dissolved in 60 mL of EtOH. An amount of 0.5 g of methyltributylammoniummethylsulfate (MTBS) was added to the solution in order to increase the conductivity of the reaction mixture for electrolysis. Two copper electrodes were used in the electrolysis cell, and a voltage of 15 V and current of 0.05 mA were applied. A few minutes after the start of the electrolysis, a blue solid started to precipitate from the solution. After 2 h, the solid precipitate was filtered off and washed one time with water and then a second time with EtOH. Afterwards the sample was dried under vacuum at 100 °C for 12 h and then kept under nitrogen atmosphere (sample D). In a second electrolysis experiment, 0.5 g of trimesic acid was dissolved in a mixture of 30 mL of EtOH and 30 mL of water. To this solution, 0.5 g of MTBS was added in order to increase the conductivity of the reaction mixture for electrolysis. The solution was electrolyzed at 60 °C using copper electrodes, a voltage of 15 V, and a current of 0.05 mA. At the end of the reaction, the sample was washed and dried as described above (sample E). All samples were routinely characterized by powder X-ray diffraction using a Siemens D5000 diffractometer operated at 40 kV (31) van Duin, A. C. T.; Baas, J. M. A.; van de Graaf, B. J. Chem. Soc., Faraday Trans. 1994, 90, 2881. (32) Laurie, V. W. J. Chem. Phys. 1961, 34, 1516.

8636 Langmuir, Vol. 24, No. 16, 2008

Figure 1. XRD patterns of Cu3(BTC)2 samples A, C, and D synthesized in pure ethanol.

Hartmann et al.

Figure 3. Simulated XRD pattern of Cu3(BTC)2.

Figure 2. XRD patterns of samples B and E synthesized in a 1:1 mixture of ethanol and water.

and 30 mA using monochromatic Cu KR radiation ( λCu KR ) 0.15405 nm). Scattering experiments were conducted at 2θ values from 4° to 50° with a step size of 0.1°. Nitrogen adsorption isotherms were recorded at 77 K on a Micromeritics ASAP 2010 sorption analyzer. Prior to the measurements, the samples were outgassed for 12 h at 100 °C under vacuum. Thermogravimetric analysis was performed using a Q 500 instrument manufactured by TA Instruments. Experiments were conducted with a constant heating rate of 20 K/min in oxygen flow. The chemical analyses have been performed with a Vario EL 3 elemental analyzer and vista-MPX CCD simultaneous ICP-OES instrument both from Varian. The adsorption experiments with isobutane and isobutene were conducted in a home-built volumetric adsorption apparatus. All samples were outgassed at 100 °C prior to the measurements using a turbo molecular pump (p < 10-7 hPa). UV-vis spectroscopy was performed on a Perkin Elmer Lambda 18 spectrometer in the diffuse reflectance mode using a “Praying Mantis” accessory. Barium sulfate was employed as the standard. The breakthrough curves were measured in a fixed-bed flow-type apparatus under atmospheric pressure. Prior to the breakthrough experiments, the Cu3(BTC)2 powder was carefully pelletized, crushed, and sieved to obtain particles with diameters between 230 and 350 µm. Thereafter, the sample was activated at 100 °C in flowing nitrogen. As feed, an equimolar mixture of isobutene and isobutane (p ) 5 kPa) in nitrogen (flow rate ) 36 cm3/min) was employed. The product stream at the outlet of the fixed-bed adsorber was analyzed on-line by gas chromatography using a flame ionization detector (FID).

3. Results and Discussion 3.1. Physicochemical Characterization. In Figures 1 and 2, the powder diffraction patterns of the samples A-E are collected. All patterns are in nice agreement with powder X-ray diffraction (XRD) patterns published in the literature15,20 and with a simulated X-ray powder diffraction pattern which was calculated using single crystal data published by Chui et al.14 The simulated X-ray powder diffraction pattern is shown in Figure 3. Notably, two additional reflections at 2θ ) 5.8° and 12.7° are observed for

Figure 4. Adsorption of nitrogen at 77 K on different Cu3(BTC)2 samples.

the hydrothermally synthesized sample A and at 2θ ) 11.0° for sample E synthesized electrochemically in an ethanol/water mixture. The two additional reflections observed for the sample prepared by hydrothermal synthesis are also visible in a number of literature XRD patterns15,24 but are absent in the XRD pattern of sample C. This is indicative for the higher quality of the sample prepared under ambient conditions under reflux. It is important to point out that reflections attributable to CuO (2θ ) 35.5 and 38.7°) or Cu2O (2θ ) 36.43°) are absent from the XRD patterns of the samples employed in this study. The good agreement of the powder X-ray powder diffraction patterns measured for our samples with the literature patterns as well as with the simulated pattern calculated from the structure proposed by Chui et al.14 confirms the successful synthesis of Cu3(BTC)2 in all five cases. Thus, we assume that it is possible to synthesize the skeletal structure of Cu3(BTC)2 using different procedures. Scanning electron microscopy revealed that the sample is polycrystalline with particle sizes below 1 µm (Figure S1, Supporting information) for both the electrochemically synthesized material and the sample synthesized under ambient pressure. In agreement with the literature,7,24 the particles from the hydrothermal synthesis are much larger (particle size of ca. 10 µm). The significantly smaller particle sizes of samples C and E are considered advantageous for envisaged application in separation and catalysis in order to reduce diffusion limitations. The high-resolution (viz. in the low pressure region) nitrogen adsorption isotherms at 77 K on the different samples are compared in Figure 4. Despite their seemingly similar structure as indicated by the XRD patterns, the nitrogen adsorption data yielding specific (Langmuir) surface areas ranging from 1150 to 1620 m2/g and specific pore volumes between 0.43 and 0.62 cm3/g are indicative of significant differences between the five samples (Table 1). As the specific surface areas are largely dependent on choosing a suitable part of the adsorption isotherm,

Separation of Isobutene and Isobutane on Cu3(BTC)2

Langmuir, Vol. 24, No. 16, 2008 8637

Table 1. Textural Properties of the Cu3(BTC)2 Samples Employed Is in the Present Study syntheses procedure sample A (autoclave synthesis in EtOH) sample B (autoclave synthesis in EtOH/H2O) sample C (synthesis in refluxing EtOH) sample D (electrochemical synthesis in EtOH) sample E (electrochemical synthesis in EtOH/H2O)

specific pore specific surface volume (cm3/g) area (m2/g) 0.57

1510

0.47

1253

0.62

1624

0.50

1309

0.43

1153

we will use the specific pore volume for the following discussions. We have to note, however, that we calculated the specific pore volume at relative pressure p/p0 ) 0.2, while often relative high relative pressures p/p0 > 0.95 are used where the onset of condensation of nitrogen between the particles might give higher absolute values. In the literature, specific pore volumes between 0.3 and 0.83 cm3/g have been reported,17 while the free pore volume from simulation as reported by Liu et al.17 amounts to ∼0.82 cm3/g. It is, thus, concluded that the differences in adsorption capacity observed for the samples studied are due to defects and/or the presence of guest molecules located inside the pores. The presence of nitrates has been suggested by Mu¨ller et al.20 for hydrothermally prepared samples, while the occlusion of trimesic acid or solvent molecules such as dimethylformamide (DMF) also has to be considered. It is nevertheless clear that the choice of synthesis method and solvent significantly influences the product obtained and in particular the adsorption properties. Different specific pore volumes and specific surface areas are observed. The largest specific surface area is found for sample C which was prepared at ambient pressure in pure ethanol. In general, syntheses with ethanol as solvent yield materials with higher specific pore volumes compared to those materials prepared in ethanol/water mixtures. Comparing samples A, C, and D leads to the assumption that a longer reaction time is beneficial with respect to sample quality. Samples B and E are in agreement with this observation, although we have to point out that no particular activation procedure was employed, since only ethanol and water were used as solvents. In particular, the coprecipitation (and pore blocking) of trimesic acid and MTBS cannot be ruled out (vide infra). We would like to add that the synthesis of Cu3(BTC)2 under ambient pressure in an ethanol/water mixture was also attempted but did not yield appreciable quantities of product. All five materials employed in the present study were characterized using thermogravimetric analysis (TGA). In addition, the gas flow at the outlet of the TG oven was monitored by IR spectroscopy. Prior to the TG experiment, the samples were dried under vacuum at 100 °C for 24 h and subsequently rehydrated overnight in a controlled water atmosphere employing a desiccator in order to ensure that the pores are completely filled with water. The TG curves of the representative samples C and D are compared in Figures 5 and 6. Sample C shows a large weight loss between 50 and 120 °C, which is ascribed to the desorption of water and is used to determine the water content collected in Table 2. Thereafter, the sample weight is reduced only slightly with increasing temperature up to 300 °C. At this temperature, the structure collapses and the organic linker is burned, which is accompanied by the detection of CO2 by IR spectroscopy. For sample D, a first weight loss analogous to the first weight loss observed for sample C is recorded, although the relative amount of water desorbed is

Figure 5. TGA experiments of Cu3(BTC)2 synthesized under ambient pressure in refluxing EtOH (sample C).

Figure 6. TGA experiments of Cu3(BTC)2 synthesized electrochemically in EtOH (sample D). Table 2. Chemical Analysis and Water Content Determined by TGA for the Cu3(BTC)2 Samples Employed in the Present Study

material

Cu content (wt %)

C content (wt %)

H content (wt %)

molar ratio nC/nCu

water content (wt %)

sample A sample B sample C sample D sample E Cu3(BTC)2

34.80 31.72 31.56 23.50 24.33 31.51

34.44 33.14 36.90 40.97 39.61 35.73

1.66 1.76 0.95 3.82 3.19 0.99

5.2 5.5 6.2 9.2 8.6 6.0

15.55 28.85 26.75 10.71 12.38

significantly smaller (Table 2). Moreover, at temperatures above 150 °C, a second (creeping) weight loss is clearly visible until the structure collapses. Again, carbon dioxide is observed as a desorption product leaving the material. However, in addition to the conducting salt MTBS, the only organic compounds which have been used for the syntheses are ethanol and trimesic acid. Ethanol has a boiling point of 78 °C and, thus, is expected to leave the sample at temperatures lower than 160 °C even if it is adsorbed in the pores. Since ethanol is assumed to be desorbed at lower temperatures, we tentatively ascribe the weight loss to material which is located inside the pores of Cu3(BTC)2 but is not part of the framework itself. Since the only compounds used for the electrochemical synthesis besides ethanol are trimesic acid and MTBS, we assume that this additional weight loss is indicative for trimesic acid and MTBS occluded in the pores of the metal organic framework material. This assumption is also in line with the lower total pore volumes and lower specific surface areas detected for the electrochemically synthesized samples as compared to the other samples in the present study. A similar TG curve was observed by Liu et al.17 and ascribed to the slow release of DMF from the pore of Cu3(BTC)2.

8638 Langmuir, Vol. 24, No. 16, 2008

Preliminary proton NMR experiments with the extract obtained by Soxhlet extraction for 72 h employing ethanol confirm that indeed both MTBS and trimesic acid are removed from the pores (Figure S4, Supporting Information). Moreover, it is observed that the specific pore volume of sample D increases significantly to about 0.76 cm3/g after the treatment by Soxhlet extraction. Thus, the observed creeping weight loss is tentatively ascribed to the decomposition of MTBS and trimesic acid occluded in the pores. The TG experiments reveal that the total amount of water adsorbed differs for the five samples employed in the present study (Table 2). However, the water loading is not fully in line with the specific pore volume determined by nitrogen adsorption. Here, we have to conclude that the hydrophilicity of the samples is also influenced by the synthesis route. The comparison of the theoretical composition of Cu3(BTC)2 with the compositions determined for our samples shows that the material which has been synthesized in refluxing ethanol (EtOH) under atmospheric pressure correlates well with the theoretical composition (Table 2). With a molar ratio of nC/nCu ) 6.2, the material has a chemical composition very close to the proposed stoichiometry of Cu3(BTC)2. The hydrothermally synthesized Cu3(BTC)2 samples possess nC/nCu ratios which are too low. Thus, an excess of copper is found which could be due to the formation of Cu2O particles, which has been reported in previous studies.14 However, no reflections due to Cu2O (or CuO) are observed in our XRD powder patterns. In contrast, the presence of extra-framework Cu2+ species has been observed by electron spin resonance (ESR) spectroscopy.33 The chemical compositions of electrochemically synthesized samples also diverge noticeably from the theoretical composition. Both samples show copper contents which are too low. On the other hand, the hydrogen and carbon contents are too high, leading to molar carbon to copper ratios of ∼9 (Table 2). The results of the chemical analysis are in line with the nitrogen adsorption data and the TGA experiment results. The capacity for nitrogen of the electrochemically synthesized samples D and E is reduced compared to the N2 adsorption capacity measured for the samples synthesized hydrothermally (samples A and B) and the sample synthesized in refluxing EtOH under atmospheric pressure (sample C). From the TGA experiments, we have learned that the electrochemically synthesized materials (samples D and E) show an additional weight loss which was tentatively ascribed to organic material which is not part of the MOF framework. Thus, an organic compound might block the pores, which results in lower nitrogen capacity for the electrochemically synthesized samples. Therefore, it is anticipated that, during the fast electrochemical synthesis of Cu3(BTC)2, trimesic acid and MTBS coprecipitate with the MOF and are captured within the structure and/or are blocking the pore entrances. Another possibility is that the framework has defects in its structure due to copper which is not coordinated as proposed for the paddle-wheel unit. These defects could lead to a partial collapse of the framework. However, the XRD patterns do not show a lower crystallinity of the electrochemically synthesized samples. 3.2. Adsorption of Isobutane and Isobutene. The highresolution isobutene and isobutane adsorption isotherms at 303 K are compared in Figures 7 and 8, respectively. For isobutene adsorption, Langmuir-type isotherms (type I according to the IUPAC classification) are observed for all samples. Thus, we assume a strong interaction of isobutene with the surrounding (33) Po¨ppl, A.; Kunz, S.; Himsl, D.; Hartmann, M. J. Phys. Chem. C 2008, 112, 2678.

Hartmann et al.

Figure 7. Adsorption of isobutene at 303 K on different Cu3(BTC)2 samples.

Figure 8. Adsorption of isobutane at 303 K on different Cu3(BTC)2 materials.

framework (possibly the copper sites and the aromatic rings via π-π interactions) which is responsible for the efficient adsorption at low pressure. A close inspection of the adsorption isotherms at low pressure reveals a step at about 250 Pa which is indicative of change of the underlying adsorption interactions. A similar (more pronounced) “kink” has been observed by Zhu et al. for the adsorption of isobutane in silicalite-1 and attributed to the large difference in adsorption entropy between the molecular location in the intersections and the channels.34 Again, the adsorption capacity is different for the samples prepared by different routes (Figures 7 and 8). The observed adsorption isotherms are not in line with the specific pore volumes determined by nitrogen adsorption, and thus, it has to be assumed that the pore volume accessible to nitrogen is not necessarily also fully accessible to larger molecules such as isobutene and isobutane. The shape of the isobutane adsorption isotherms observed for samples A, B, and C is markedly different from the shape of the isobutane isotherms recorded for the two electrochemically synthesized samples D and E (Figure 8), which also show type I isotherms as have been detected for isobutene adsorption. In contrast, for samples A, B, and C, isotherms are measured which are S-shaped (type V according to the IUPAC classification). Type III and type V isotherms are characteristic of weak gas-solid interactions. Type III isotherms are observed for nonporous or macroporous solids, while type V isotherms are found for mesoporous or microporous solids as in the present case. The weakness of the adsorbent-adsorbate interactions will cause the uptake at low relative pressures to be small; however, once some (34) Zhu, W.; Kapteijn, F.; Moulijn, J. A. Chem. Phys. Phys. Chem. 2000, 2, 1989.

Separation of Isobutene and Isobutane on Cu3(BTC)2

Langmuir, Vol. 24, No. 16, 2008 8639

Figure 9. Comparison of isobutene and isobutane adsorption isotherms at 303 K on Cu3(BTC)2 (sample C).

molecules are adsorbed, the adsorbate-adsorbate interactions will promote the adsorption of further molecules so that the isotherm will become convex to the pressure axis.35 Type V isotherms are rarely observed and often involve the adsorption of water. Compared to those of ethane21 and propane,24 the isobutane adsorption isotherm is much steeper and almost horizontal at higher pressures. A similar behavior has been observed for the adsorption of these gases on zeolite 13X and has been ascribed to the fact that the adsorption temperature is below the critical temperature of isobutane (401.8 K), while the adsorption temperature is above the critical temperatures of ethane and propane.36 This means that the much less volatile isobutane is more likely to condense on the surface, since it has a much larger relative saturation value than the other adsorbates. On the other hand, isobutane has a much larger van der Waals volume of 0.048 m3/kmol than ethane and propane (0.020-0.027 m3/kmol). Thus, it is expected that isobutane fills the pores at low pressure and then gives a horizontal isotherm, which is indeed observed in the present study. Compared to samples A, B, and C, the accessible pore volume is much lower in the electrochemically synthesized samples D and E. Thus, the appearance of a type I isotherm for the latter sample might be related to the reduced accessible pore volume which allows adsorbate-adsorbate interactions to dominate at significantly lower pressure. Since we have assumed that pore blocking by trimesic acid and/or MTBS might occur in these samples, we cannot rule out that interaction of isobutane with the occluded organics might be responsible for the observed shape of the adsorption isotherm. In Figure 9, the isobutane and isobutene adsorption isotherms of sample C recorded at 303 K are compared. While the isotherms are strikingly different in the low pressure region (