Effects of Molecular Sieving and Electrostatic Enhancement in the

Sep 10, 2010 - Stijn Van der Perre , Tom Van Assche , Belgin Bozbiyik , Jeroen Lannoeye , Dirk. ... David Peralta , Gérald Chaplais , Angélique Simo...
0 downloads 0 Views 1MB Size
Download PDF
pubs.acs.org/Langmuir © 2010 American Chemical Society

Effects of Molecular Sieving and Electrostatic Enhancement in the Adsorption of Organic Compounds on the Zeolitic Imidazolate Framework ZIF-8 Matthew T. Luebbers, Tianjiao Wu, Lingjuan Shen, and Richard I. Masel* Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana, Illinois 61801 Received June 25, 2010. Revised Manuscript Received August 24, 2010 In this work, the adsorption behavior of a range of organic vapors and gases on the zeolitic imidazolate framework, ZIF-8, is investigated using an inverse gas chromatography (IGC) methodology at the zero-coverage limit and elevated temperatures. The measured thermodynamic values and surface energies for the adsorption of n-alkanes on ZIF-8 are found to be reduced from those previously reported for IRMOF-1. This reduction is most likely an effect of the predominately organic accessible surface of ZIF-8 and the resulting weaker interactions in comparison to IRMOF-1. The pore aperture size of ZIF-8, which is significantly reduced from that of IRMOF-1, is seen to introduce molecular sieving effects for branched alkanes, aromatics, and heavily halogenated compounds. Deformation polarizabilities of the adsorbates were used to calculate the specific adsorption free energy, and it is determined that the specific effects account for around 1-5 kJ/mol, or between 10% and 70% of the total free energy of adsorption for the sorbates studied (at 250 °C). The importance of electrostatic forces was seen in the significantly enhanced adsorption of propylene and ethylene in comparison to their respective alkanes and in the direct correlation shown between the specific components of the free energy of adsorption and the adsorbate’s dipole moment.

Introduction The separation of propane and propylene mixtures (as well as other paraffin/olefin mixtures) is a very difficult and expensive commercial process due to the small differences in relative volatility and the large fractional distillation apparatus required.1 Because of this difficulty, several alternative processes have been explored2 including adsorption-based separation utlizing zeolites and similar traditional microporous materials.3,4 Similarly, work is being done to effectively sieve linear alkanes from larger branched alkanes over zeolite and silicalite adsorbents for the boosting of octane ratings for better fuels.5-8 The efficiency of these processes is tied directly to the abilities of the adsorbent, and due to limitations with current materials novel adsorbents with tailored properties are needed. One exciting class of microporous materials which represents a great opportunity for the improvement of a vast array of adsorption processes is metal-organic frameworks (MOFs, also referred to as porous coordination polymers). MOFs are a class of crystalline materials that often possess unprecedented high porosity, substantial thermal stability (often over 400 °C), and the allure of potentially being designable. Over roughly the past decade, there has been a significant amount of research regarding the synthesis and applications of MOFs including demonstrated *To whom correspondence should be addressed. E-mail: r-masel@ illinois.edu. (1) Lamia, N.; Jorge, M.; Granato, M. A.; Paz, F. A. A.; Chevreau, H.; Rodrigues, A. E. Chem. Eng. Sci. 2009, 64(14), 3246. (2) Bryan, P. F. Sep. Purif. Rev. 2004, 33(2), 157. (3) Da Silva, F. A.; Rodrigues, A. E. AIChE J. 2001, 47(2), 341. (4) Yang, R. T.; Kikkinides, E. S. AIChE J. 1995, 41(3), 509. (5) Denayer, J. F. M.; Ocakoglu, R. A.; Thybaut, J.; Marin, G.; Jacobs, P.; Martens, J.; Baron, G. V. J. Phys. Chem. B 2006, 110(17), 8551. (6) Devriese, L. I.; Cools, L.; Aerts, A.; Martens, J. A.; Baron, G. V.; Denayer, J. F. M. Adv. Funct. Mater. 2007, 17(18), 3911. (7) Huddersman, K.; Klimczyk, M. AIChE J. 1996, 42(2), 405. (8) Ocakoglu, R. A.; Denayer, J. F. M.; Marin, G. B.; Martens, J. A.; Baron, G. V. J. Phys. Chem. B 2003, 107(1), 398.

Langmuir 2010, 26(19), 15625–15633

separations of alkane isomers.9-11 There are tens of thousands of MOF structures cataloged in the Cambridge Structural Database;12-15 however, issues regarding the mechanical and chemical stability of many MOF structures currently prevent applications of these materials to many processes. A great number of MOF structures are not stable after the removal of the solvent (or other guest molecules), and many other MOFs (most notably IRMOF-1) decompose quickly to a nonporous phase when exposed to even very small amounts of moisture.16-18 One recently developed class of MOFs that shows great potential to overcome the issues of stability is zeolitic imidazolate frameworks (ZIFs). ZIFs are formed by the crystallization of a transition metal species bound to the nitrogens of an imidazole compound and are so named because of their structural similarities to many zeolites (including characteristic angles of 145°, the preferred Si-O-Si angle seen in many zeolites).19 There are (9) Barcia, P. S.; Zapata, F.; Silva, J. A. C.; Rodrigues, A. E.; Chen, B. L. J. Phys. Chem. B 2007, 111(22), 6101. (10) Chen, B. L.; Liang, C. D.; Yang, J.; Contreras, D. S.; Clancy, Y. L.; Lobkovsky, E. B.; Yaghi, O. M.; Dai, S. Angew. Chem., Int. Ed. 2006, 45(9), 1390. (11) Pan, L.; Olson, D. H.; Ciemnolonski, L. R.; Heddy, R.; Li, J. Angew. Chem., Int. Ed. 2006, 45(4), 616. (12) Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Wachter, J.; O’Keeffe, M.; Yaghi, O. M. Science 2002, 295(5554), 469. (13) Yaghi, O. M.; O’Keeffe, M.; Ockwig, N. W.; Chae, H. K.; Eddaoudi, M.; Kim, J. Nature 2003, 423(6941), 705. (14) Ferey, G. Chem. Soc. Rev. 2008, 37(1), 191. (15) Rowsell, J. L. C.; Yaghi, O. M. Microporous Mesoporous Mater. 2004, 73 (1-2), 3. (16) Hafizovic, J.; Bjorgen, M.; Olsbye, U.; Dietzel, P. D. C.; Bordiga, S.; Prestipino, C.; Lamberti, C.; Lillerud, K. P. J. Am. Chem. Soc. 2007, 129(12), 3612. (17) Hausdorf, S.; Baitalow, F.; Seidel, J.; Mertens, F. J. Phys. Chem. A 2007, 111(20), 4259. (18) Kaye, S. S.; Dailly, A.; Yaghi, O. M.; Long, J. R. J. Am. Chem. Soc. 2007, 129(46), 14176. (19) Park, K. S.; Ni, Z.; Cote, A. P.; Choi, J. Y.; Huang, R. D.; Uribe-Romo, F. J.; Chae, H. K.; O’Keeffe, M.; Yaghi, O. M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103(27), 10186.

Published on Web 09/10/2010

DOI: 10.1021/la102582g

15625

Article

several members of this class of MOFs, and many of them have been reported for potential roles in interesting applications such as large, selective uptake of carbon dioxide.20-22 One particularly interesting member of the ZIF family and the one investigated here is ZIF-8 (2-methylimidazole zinc salt). ZIF-8 has been demonstrated in many applications including the kinetic separation of propane/propylene,23 the separation of hydrogen,24 and the oxidation of CO in a ZIF-8 impregnated with gold.25 ZIF-8 can be described as having a formula of Zn(mIM)2 in which mIM represents 2-methylimidazolate and having a sodalite-like topology with internal cavities 1.16 nm in diameter connected by windowlike pore apertures 0.34 nm across.26,27 In comparison to many other MOFs, ZIF-8 has shown amazing thermal and chemical stability due to its hydrophobic nature and strong bonds. It has been demonstrated as being stable to 550 °C as well as for extended periods of time in boiling benzene, methanol, water, and aqueous NaOH.19 This type of stability truly opens up many applications to MOFs that are not able to be realized by the vast majority of these materials. Our previous work addressed the adsorption of volatile organic compounds (VOCs) in the water-sensitive MOF, IRMOF-1, for multiple samples with different surface areas and structures in an attempt to probe the effects of varied degrees of degradation upon the adsorption behavior.28 It is the purpose of this paper to expand upon that previous work and to address the adsorption of a substantially more stable and commercially available sample of ZIF-8. By utilizing a pulse-injection inverse gas chromatography (IGC) technique, the adsorption of alkanes, alkenes, as well as compounds with various other properties is systematically studied on ZIF-8. The equilibrium constants and thermodynamics of adsorption at the low coverage limit are explored, and correlations are made between the molecular properties of the adsorbates and the observed adsorption behavior. Molecular properties commonly used to analyze and explain adsorption behavior (i.e., vapor pressure29 and polarizability30,31) are employed in an attempt to discriminate between the specific and dispersive components of adsorption. The effects of hydrogen bonding upon adsorption behavior are also explored. An electrostatic enhancement in the adsorption of alkenes over the corresponding alkane is observed as is the molecular sieving of branched alkanes from linear alkanes. The specific component of the free energy of adsorption is calculated for some compounds and found to correlate well to the dipole moment of the adsorbate. (20) Banerjee, R.; Furukawa, H.; Britt, D.; Knobler, C.; O’Keeffe, M.; Yaghi, O. M. J. Am. Chem. Soc. 2009, 131(11), 3875. (21) Wang, B.; Cote, A. P.; Furukawa, H.; O’Keeffe, M.; Yaghi, O. M. Nature 2008, 453(7192), 207. (22) Willans, C. E.; French, S.; Barbour, L. J.; Gertenbach, J. A.; Junk, P. C.; Lloyd, G. O.; Steed, J. W. Dalton Trans. 2009, 33, 6480. (23) Li, K. H.; Olson, D. H.; Seidel, J.; Emge, T. J.; Gong, H. W.; Zeng, H. P.; Li, J. J. Am. Chem. Soc. 2009, 131(30), 10368. (24) Bux, H.; Liang, F. Y.; Li, Y. S.; Cravillon, J.; Wiebcke, M.; Caro, J. J. Am. Chem. Soc. 2009, 131(44), 16000. (25) Jiang, H. L.; Liu, B.; Akita, T.; Haruta, M.; Sakurai, H.; Xu, Q. J. Am. Chem. Soc. 2009, 131(32), 11302. (26) Pantatosaki, E.; Pazzona, F. G.; Megariotis, G.; Papadopoulos, G. K. J. Phys. Chem. B 2010, 114(7), 2493. (27) Perez-Pellitero, J.; Amrouche, H.; Siperstein, Flor R.; Pirngruber, G.; Nieto-Draghi, C.; Chaplais, G.; Simon-Masseron, A.; Bazer-Bachi, D.; Peralta, D.; Bats, N. Chem.;Eur. J. 2010, 16(5), 1560. (28) Luebbers, M. T.; Wu, T.; Shen, L.; Masel, R. I. Langmuir 2010, 26(13), 11319. (29) Saintflour, C.; Papirer, E. Ind. Eng. Chem. Prod. Res. Dev. 1982, 21(4), 666. (30) Donnet, J. B.; Park, S. J.; Balard, H. Chromatographia 1991, 31(9-10), 434. (31) Menzel, R.; Lee, A.; Bismarck, A.; Shaffer, M. S. P. Langmuir 2009, 25(14), 8340.

15626 DOI: 10.1021/la102582g

Luebbers et al.

Experimental Section The sample of ZIF-8 was used as purchased from SigmaAldrich (Basolite Z1200). As recommended from the manufacturer, the material was first activated at 100 °C under vacuum conditions for a period of 4 h before any measurements or characterizations were performed. The surface area of the material is provided by the manufacturer as being between 1300 and 1800 m2/g with a bulk density of 0.35 g/cm3 and particle size of around 4.9 μm. The surface area of the sample was measured using standard nitrogen adsorption methodology at liquid nitrogen temperature (-196 °C) using a Quantachrome Nova 2200e gas sorption analyzer and ultrahigh purity (UHP) nitrogen. The sample was first degassed under vacuum at 130 °C overnight. A nitrogen isotherm was measured at 21 different relative pressures ranging from 2.9  10-5 to 0.988, and standard Langmuir and BrunauerEmmett-Teller (BET) theories were applied to calculate the specific surface area. For the IGC studies, dry MOF powder was packed into an approximately 15 cm length of 0.53 mm ID guard column (Restek, 10045). The first step was to insert a plug made of IP (intermediate polarity) deactivated borosilicate glass wool (Restek, 20789) about 1 cm long into the column. The mass of the capillary and glass wool plug were measured and recorded. After fixing a sufficient plug into one end of the capillary, the column was connected to a vacuum pump and a vacuum of around 10 psi was applied using a metering valve for control. The open end of the column was used to suction around 3-4 cm (∼2 mg) of dry MOF powder into the column with occasional mechanical vibrations used to assist the uniform packing. After reweighing the column (now with MOF), a second glass wool plug was inserted into the open end of the column. Excess guard column was trimmed from the packed bed to reduce dead volume and peak broadening leaving around 1 cm open on each end. The final weight of the shortened column was recorded. The guard column packed with MOF was then connected to two lengths (7 and 30 cm) of 0.25 mm ID guard column (Restek, 10049) using butt connector fittings (Supelco, 23804) with M-2B butt connector ferrules (Supelco, 22455-U). The column assembly was then connected into the split/splitless inlet of an Agilent 6890 gas chromatograph using the shorter length of connecting guard column (oriented in such a way that the direction of flow is the same as that of the suction during the packing process) with the longer length of connecting column left open during the duration of activation. The gas chromatograph was set up using UHP helium as a carrier gas which was first passed through a 7 μm particle filter and then a universal trap (Agilent, RMSH 2). A split/splitless fast focus inlet linear with 2.3 mm ID (Supelco, 2879505-U) was used in the gas chromatograph inlet. For activation, the inlet of the gas chromatograph was pressurized to 20 psi and the oven temperature is slowly ramped (1 °C/ min) to 275 °C where the temperature was held for 10 h before being slowly returned to room temperature. This step was done to remove any residual guest molecules in the MOF and to increase the available porosity for IGC measurements. The packed column was then disassembled and reweighed to account for any mass changes due to removal of residual solvent in the MOF. The packed column was then reassembled and the effluent end of the column was connected to a flow meter (Omega, FMA-A2300) and the flow rates were recorded for all combinations of temperature and pressure investigated (up to 270 °C). The flow meter readings were converted to helium by multiplication by a conversion factor. The measured flow rate values which were used in subsequent calculations are available in the Supporting Information as Table SI-1. After measuring and recording the flow rates for each set of temperature and pressure, the outlet of the column was connected to a flame ionization detector (FID) and pulse injections were performed. The gas chromatograph’s auto injector (Agilent, 7683B) Langmuir 2010, 26(19), 15625–15633

Luebbers et al.

Article

and a 10 μL gastight syringe (Agilent, 5181-3354) were used to inject 0.2-5.0 μL of headspace vapor onto the column. Dead volume was measured by injecting methane at 200-270 °C at several inlet pressures and measuring the holdup time assuming no retention. For studying the adsorption of the organic compounds, every measured value of a corrected specific retention volume (i.e., for each combination of adsorbate, temperature, and pressure) involved measuring the retention for three different sizes of injection. This was done to ensure that the peak position was independent of the concentration, thus verifying that the measurement corresponds to an ideal, infinite dilution behavior.

Results and Discussion In order to verify that the sample of ZIF-8 under investigation is representative of the high surface area form of the material, a nitrogen adsorption isotherm was measured at 77K (-196 °C). This isotherm (which is provided in the Supporting Information as Figure SI-1) shows the expected Type I shape under IUPAC classifications, which is typical for microporous materials, and a saturation volume of 383 cm3 STP/g (where STP represents the conversion of the adsorbed nitrogen to the corresponding volume at standard temperature and pressure). This isotherm agrees very well with the data presented in literature19 regarding both the relative pressure required for saturation and the volume adsorbed at saturation. Using the standard techniques to calculate specific surface area from nitrogen adsorption data, it was determined that the surface area was 1504 m2/g by the three-point BET method (the Langmuir model utilizing the whole isotherm gives a value of 1729 m2/g). Based upon the measured nitrogen saturation volume of 383 cm3(STP)/g and a liquid nitrogen density of 0.8081 g/cm3, the pore volume was calculated as 0.592 cm3/g. This pore volume agrees very well with the experimental (0.59 cm3/g) and calculated (0.52 cm3/g) values reported recently.27 These calculated values fit well within the manufacturer-described properties and help to verify that the sample investigated here is of high quality. These results indicate the increased stability of the material which is obtainable in a pure form much easier than the watersensitive IRMOF-1 which we have previously investigated.28 Chromatographic Behavior. The initial study performed was to study the effect of flow rate upon the corrected retention volumes calculated (see Supporting Information for the equations used) using the measured retention time, column temperature, James-Martin “compressibility correction factor”, and the measured flow rate. The specific corrected retention volumes discussed throughout represent the volume of carrier gas (converted to STP) needed to elute through the ZIF-8 packed bed. A plot demonstrating the effect of flow rate upon the retention of representative adsorbates is shown in the Supporting Information in Figure SI-3. In this sample plot, it was seen that the corrected specific retention volume for propylene at 50 °C and n-hexane at 230 °C are virtually independent of flow rate. This indicates that the column is operating in a mode where mass transfer concerns do not impact elution time, and it implies that the values obtained for the corrected retention volumes (and all subsequently calculated values) are indicative of equilibrium behavior. Data for n-octane show a greater variation with flow rate, but the trend is opposite of what would be expected if representing the effects of slow diffusion (i.e., the retention volume measured here is greater at higher flow rates) and is most likely a result of uncertainty in peak maximum positions due to the nonideal column performance. For subsequent trials, the column was operated at the highest inlet pressure studied to allow for maximum column efficiency as shown later in this paper. Langmuir 2010, 26(19), 15625–15633

Figure 1. Example chromatograms for compounds adsorbing on ZIF-8 with behavior suitable for thermodynamic analysis. (a) Effect of increased n-alkane chain length. (b) Retention of some compounds with hydrogen-bond basicity. (c) Effect of temperature upon the elution of THF.

For the IGC study, the raw data obtained for the elution of the VOC from the packed column is in the form of the FID signal versus time. Chromatograms for the detected elution of the series of nonpolar n-alkane probes are represented by slightly asymmetric peaks with positions that are independent of the injected amount. This behavior is demonstrated in Figure 1a for selected n-alkanes. This independence of peak position upon concentration verifies that the study is performed in the linear (Henry’s) section of the adsorption isotherm and that infinite dilution is a valid assumption.32 It can be observed by comparing the chromatograms reported here to those in the work that we have previously reported28 for similarly packed MOF columns that the chromatograms for the ZIF-8 packed bed are much broader and more asymmetric in (32) Conder, J. R.; Young, C.L. Physicochemical Measurements by Gas Chromatography; Wiley: New York, 1979.

DOI: 10.1021/la102582g

15627

Article

many cases. This decreased column performance is most likely the result of two factors: decreased flow rate and decreased mass transfer (and kinetics of desorption) in the ZIF-8 column. The decreased flow rate is primarily due to the size of the particles supplied by the vendor. The reported particle size is 4.9 μm which is roughly half of the typical size of MOF particles used in the previous work.28 The increased pressure drop through the column makes it difficult to achieve higher flow rates, and as shown in data provided in Figure SI-2 in the Supporting Information the number of theoretical plates is rapidly increasing with flow rate for the range of flows studied here. In other words, the pressure drop is too large to operate the column under optimal conditions. The slight asymmetry in the chromatograms of many studied compounds also indicates the presence of decreased mass transfer and slowed desorption kinetics which are most likely results of the much smaller pore sizes in the ZIF-8 material in comparison to those of IRMOF-1.19 The concentration-independence of peak positions and symmetric peak shapes are observed for a wide range of studied adsorbates. Most notably, compounds with substantial hydrogen bond basicity (e.g., ketones, diethyl ether, ethyl acetate) fit into this category unlike when they were studied on the IRMOF-1 samples.28 This is most likely a result in the significant chemical and structural differences between ZIF-8 and IRMOF-1 and the absence of any Brønsted or Lewis acid sites in the ZIF structure.19,23 As an example of the chromatograms for these compounds, those obtained for ethyl acetate, MEK, and acetone eluting through the column of ZIF-8 are plotted in Figure 1b. Additionally, chromatograms for THF are shown in Figure 1c at various temperatures, further demonstrating the improved behavior of these types of adsorbates in comparison to adsorption on IRMOF-1.28 It can be seen that the peak positions are independent of the amount injected and that the peaks are not severely asymmetric. Ideal chromatograms were not obtained for all compounds however. As shown in Figure 2a, injections of ethylene consistently resulted in the elution of two distinct peaks. It appears that this observation is not the result of a contaminant in the gas cylinder or delivery system, but a real phenomenon of adsorption on ZIF-8. The reason for this behavior is not clear at this time. For the calculations reported in this work, the retention time for the first (quickest eluting) peak is used. Another complication is demonstrated in the chromatogram supplied in Figure 2b which demonstrates the results obtained for toluene at a range of temperatures. It can be seen that the elution of toluene occurs very quickly with minimal retention and broad tailing. The position of the peak maxima does not change with temperature as would be expected, although the amount of tailing can be seen to decrease somewhat at higher temperatures. This behavior is representative of all aromatic compounds studied as well as cyclopentane and cyclohexane. The data for these compounds demonstrates the size exclusion of these larger cyclic compounds due to the restricted size of the pore aperture for ZIF8. The tailing effects observed are most likely an artifact of impeded mass transfer through the packed bed and strong interactions with the ZIF crystals’ exposed surfaces. In addition, branched alkanes (except for isobutane) are also excluded from entering the pores of the ZIF material as demonstrated in Figure 2c for 2-methylbutane (isopentane). In this plot, 2-methylbutane can be seen to elute very quickly with practically no retention as opposed to n-pentane which shows measurable retention and symmetrical peaks. Similar results are obtained for 2-methylheptane (isooctane) as well as the halogenated compounds: carbon tetrachloride, 1,1,1-trichloroethane, and tetra15628 DOI: 10.1021/la102582g

Luebbers et al.

Figure 2. Representative chromatograms on ZIF-8 for compounds with nonideal behavior. (a) Examples of the two peaks detected for ethylene (retention data for the first peak was used in calculations). (b) Quick elution and broad tailing peaks for toluene that are representative of aromatics studied. (c) Apparent molecular sieving between isomers resulting in rapid elution of isopentane (no significant retention) compared to n-pentane.

chloroethylene. The kinetic diameters of the branched alkanes and those compounds with several bulky chlorine atoms exceed the small pore aperture size of ZIF-8 (0.34 nm). Although the pore apertures of ZIF-8 are not circular and have some flexibility to accommodate guest molecules,27 the cutoff for an adsorbent to allow entry to monobranched alkanes is at or above 0.5 nm, significantly larger than the openings in the ZIF-8 structure.33 It is worth noting at this point that the structure of ZIF-8 must be considerably more flexible than previously believed in order to accommodate many of the guests studied here. For example, the authors were in fact initially surprised that the n-alkane species were able to enter into the pores of ZIF-8 (33) Eder, F.; Lercher, J. A. Zeolites 1997, 18(1), 75.

Langmuir 2010, 26(19), 15625–15633

Luebbers et al.

Article

Table 1. Values of the Henry Constants (mmol/kg/kPa) for Different Adsorbates on ZIF-8 Packed Column at Listed Temperaturesa Henry constants (mmol/kg/kPa) temperature (°C) adsorbed species

50

70

180

200

230

250

270

ethane 2.91 1.67 ethylene 9.78 5.64 propane 14.55 8.98 propylene 63.16 32.59 n-butane 54.77 28.35 isobutene 30.62 14.59 n-pentane 29.36 19.62 11.54 8.17 5.99 n-hexane 77.38 48.24 26.09 17.73 12.44 n-heptane 194.01 114.32 56.88 36.91 24.69 n-octane 115.52 72.00 45.28 n-nonane 220.12 129.55 80.03 diethylether 22.75 14.82 8.68 6.13 4.51 ethylacetate 61.00 32.03 16.27 10.39 7.13 1,2-dichloroethylene 16.53 11.41 6.95 5.04 3.76 acetone 22.75 11.95 6.31 4.14 2.92 MEK 47.25 25.17 13.03 8.59 6.05 trichloroethylene 41.53 27.16 15.36 10.71 7.69 acetaldehyde 5.52 3.46 2.19 1.56 1.11 bromoform 108.84 63.42 42.33 THF 12.79 7.87 5.31 dichloromethane 9.77 7.05 4.22 3.08 2.28 a The listed values for the Henry constants are calculated from the corrected retention volumes measured on ZIF-8. The full range of corrected retention volumes and Henry constants are provided in the Supporting Information. The cases where the measured retention volume was less than the dead volume are shown in bold due to increased experimental uncertaintiy for these data points.

considering a reported kinetic diameter greater than or equal to 0.43 nm34 for these compounds; however, the chromatographic performance of these compounds does not show any indications of significant diffusion limitations. As was the case in the previous work on IRMOF-1, a variety of alcohols and amines (i.e., methanol, ethanol, isopropanol, n-propanol, tert-butanol, 1-hexanol, 1-heptanol, triethylamine, and propylamine) were also investigated but are not included due to the absence of any detectable elution peaks.28 This complication is attributed to strong interactions between those compounds with significant hydrogen bonding acidity and the framework. Unexpectedly, injections of 1-pentene also did not show a detected elution peak. The impact of various molecular properties upon adsorption behavior will be discussed in more detail, but these compounds offer a chromatographic complication that cannot be overcome. Equilibrium Behavior. The measured Henry constants for the adsorption of the organic compounds on ZIF-8 are provided in Table 1 at selected representative temperatures. The full set of corrected retention volumes and Henry constants are provided in Tables SI-2 and SI-3 in the Supporting Information. An increased retention of n-alkanes with increasing chain length is seen as would be expected from the additive nature of the dispersive interactions between these compounds and the additional methylene groups added to each member of the series. One observation from Table 1 is that the values calculated for the equilibrium constants on ZIF-8 are appreciably reduced at the temperatures studied in comparison to those previously reported for IRMOF-1.28 Some compounds (i.e., ethane, acetaldehyde, dichloromethane, and acetone) have very small retention volumes which are often slightly less than the corrected specific dead volume (0.018 L/g) (34) Breck, D. W. Zeolite Molecular Sieves: Structure, Chemistry, and Use; Wiley-Interscience: New York, 1974.

Langmuir 2010, 26(19), 15625–15633

which was calculated using methane pulse injections. There is greater uncertainty for these data points, and for this reason they are shown in bold in Table 1. It can also be seen in Table 1 that the measured Henry constants are 3-4 times higher for ethylene than on ethane and 3.5-5.5 times higher for propylene than on propane. As mentioned previously, injections of 1-pentene do not show a detectable elution peak. It is a possibility that once inside the framework the longer 1-pentene chain is susceptible to reactions or rearrangement which prevents it from diffusing out. The behavior of 1-pentene aside, these observations indicate an enhancement in the adsorbent-adsorbate interactions due to the presence of the π-electrons of the alkenes. This is an opposite trend to that seen in previous work on IRMOF-1.28 As discussed later in this work, the dipole moment of the adsorbate is found to directly correlate to the strength of the interaction with the framework. This enhancement in adsorption seen for alkenes is therefore most likely an artifact related to the polarizability of these compounds through the distortion of the π-electron cloud in an external electric field. The dipoles induced in these molecules by an electric field present within the ZIF-8 framework lead to electrostatic effects and significantly enhance the adsorbate/adsorbent interactions. Another observation of note from Table 1 is that the measured Henry constants for n-butane are on the order of 2 times higher than those measured for isobutane under these conditions. As previously discussed, the narrow pore apertures of ZIF-8 appear to act as a molecular sieve preventing the larger, branched alkanes from diffusing into the material. Isobutane is the only branched alkane that gave measurable retention data although it is reduced from its linear isomers. The chromatograms measured for isobutane are not asymmetric and severely tailing as were those for the other branched alkanes studied (i.e., 2-methylbutane and 2-methylheptane); however, the retention of isobutane on ZIF-8 is in fact relatively small. This observation implies that the molecular sieve effect is greater for the branched isomers of the larger alkanes. Thermodynamics of Adsorption. The temperature dependence of the retention volumes can be explored to derive values for the differential enthalpy of adsorption, ΔH, via the van’t Hoff relationship as discussed in the Supporting Information. Representative van’t Hoff plots for adsorption on ZIF-8 are shown in Figure 3 for (a) a series of n-alkanes, (b) the lighter gaseous compounds, and (c) the hydrogen bond base compounds. It can be seen that an exponential relationship exists between the measured corrected retention volumes and the inverse absolute temperature. Linearity similar to that seen in Figure 3 is observed for all compounds studied. To calculate enthalpies from IGC data, the range and intervals used for the van’t Hoff calculations need to be sufficiently small to maintain linearity of the isosteres and minimize any additional uncertainties in the calculated slopes. There is, in general, greater linearity in the isosteres measured on ZIF-8 than was seen in the previously reported work with IRMOF-1,28 implying a less temperature-dependent adsorption process for the ZIF material. In the previous work, it was found that too wide a temperature range could alter the calculate enthalpies by more than 10 kJ/ mol;28 however, in the work reported here, similar temperature ranges with the same compounds are found to only alter the enthalpies by 1-2 kJ/mol. The values used for the enthalpy determination are given in the Supporting Information (Table SI-4) along with the data points used and the quality of fit. The calculated values for the enthalpy, ΔH, free energy, ΔG, and entropy, ΔS, of adsorption are all listed in Table 2 at the stated temperatures along with the enthalpy of vaporization at DOI: 10.1021/la102582g

15629

Article

Figure 3. Representative van’t Hoff plots for the adsorbates measured on the ZIF-8 packed column.. The sorbates shown in plot (a) are for the n-alkanes labeled on the plot. The adsorbates in plot (b) are ethane ([), ethylene (9), propane (2), propylene (), isobutane (b), and n-butane (]). The adsorbates in plot (c) are diethyl ether ([), ethyl acetate (9), acetone (2), MEK (), and 1,2dichloroethylene (]).

25 °C.35 The equations used in the calculation of these values are discussed in the Supporting Information. The temperature listed is the mean value of the data points used in the linear fit used for calculating ΔH. The bottom of Table 2 gives the slope of a line plotted for the thermodynamic values versus n-alkane chain length for the pentane to nonane n-alkane series. Values for ΔG were calculated at every temperature for each adsorbate, and a full table of these values is included in the Supporting Information (Table SI-6). In Table 2, the enthalpies of adsorption measured for the n-alkanes at 250 °C on ZIF-8 are slightly greater (5-7 kJ/mol) (35) Lide, D. R. CRC Handbook of Chemistry and Physics, 90th ed.; CRC Press: Boca Raton, FL, 2009-2010.

15630 DOI: 10.1021/la102582g

Luebbers et al.

than the enthalpies of vaporization listed for 25 °C, indicating the enhanced interaction between the ZIF surface and the nonpolar compounds. The lighter gaseous compounds, which are measured at 60 °C, showed greater enhancement over the heat of vaporization. This is most likely an effect of the temperature probed as well as the enhanced access to stronger binding sites for these smaller analytes. The enhancement for ethylene and propylene over their paraffin counterparts which was previously discussed for equilibrium properties can also be seen in the calculated adsorption enthalpy values. The previously discussed π-electron enhancement of the adsorption process appears to be significant as the heat of adsorption for propylene is found to be nearly 10 kJ/mol higher than that of propane. The other polar analytes measured have varying degrees of enhanced adsorption enthalpies in comparison to the heats of vaporization with the most strongly enhanced being THF (around 13 kJ/mol enhancenment). The calculated values for the free energies and enthalpies of adsorption of all the compounds studied on ZIF-8 are significantly reduced from those calculated for IRMOF-1. The largest decreases in the values for enthalpy of adsorption on ZIF-8 when compared to IRMOF-1 are found in the hydrogen-bond bases which showed very large specific heats of adsorption on IRMOF1, a result of the absence of strong acid sites in the structure of ZIF-8.19,23 In general, the decreased energetics of the ZIF-8 surface can be attributed to the location of the adsorption sites in comparison to other MOFs such as IRMOF-1. It has been reported that the most energetically favorable sites for adsorption in ZIF-8 are close to the imidazole rings and that the Zn atoms are far from the accessible surface seen by a sorbate.27 In comparison, the most energetic sites in IRMOF-1 have been found to be around the metal-oxide clusters.27 The weaker interactions seen in ZIF-8 can therefore be attributed to the predominately organic nature of the accessible adsorption sites. The effects of chain length upon the calculated values of enthalpy, entropy, and free energy are plotted for a series of n-alkanes in Figure 4. The enthalpy and free energy associated with the adsorption of an additional methylene, ΔHCH2 and ΔGCH2, respectively, can be obtained from the slope of a line fit to the data in Figure 4, and these values are summarized at the bottom of Table 2 along with the entropy value. The linearity of the thermodynamic values for n-alkanes is due to the additive nature of dispersive interactions and is commonly shown in experimental and computational research. The slope of the linear relationships for ΔH and ΔG calculated for ZIF-8 are significantly reduced from the values calculated for IRMOF-1 which were previously reported (slopes of ΔH around 9 kJ/mol and ΔG around 4 kJ/mol),28 again indicating a less energetic surface for the ZIF-8 sample. The value for ΔGCH2 calculated and shown in Table 2 for the series of n-alkanes is commonly used through the Dorris and Gray method36-38 to estimate the dispersive component of the surface energy, γsD (see the Supporting Information for the equations used), which is listed in Table 3. The value of γsD is an important term for the evaluation of the acitivity of a solid surface and is related to the London interactions and the heterogeneity and defects of a surface. The values for γsD (between 45 and 55 mJ/m2) calculated for the ZIF-8 sample here are relatively low compared to those of other microporous materials such as (36) Dorris, G. M.; Gray, D. G. J. Colloid Interface Sci. 1980, 77(2), 353. (37) Yla-Maihaniemi, P. P.; Heng, J. Y. Y.; Thielmann, F.; Williams, D. R. Langmuir 2008, 24(17), 9551. (38) Almazan-Almazan, M. C.; Perez-Mendoza, M.; Domingo-Garcia, M.; Fernandez-Morales, I.; del Rey-Bueno, F.; Garcia-Rodriguez, A.; Lopez-Garzon, F. J. Carbon 2007, 45(9), 1777.

Langmuir 2010, 26(19), 15625–15633

Luebbers et al.

Article

Table 2. Calculated Thermodynamic Values for Adsorption on the ZIF-8 Sample along with the Heat of Vaporization at 25° C for Ref 35a ΔHvap (25 °C) (kJ/mol)

T (°C)

ΔHads (kJ/mol)

ΔGads (kJ/mol)

ΔSads (J/(mol 3 K))

ethane ethylene propane propylene

5.16 NA 14.79 14.24

60 60 60 60

23.85 25.32 20.46 30.02

0.54 3.74 4.95 8.76

69.97 64.77 46.54 63.79

n-butane isobutane n-pentane n-hexane n-heptane n-octane n-nonane diethylether ethyl acetate 1,2-dichloroethylene acetone MEK trichloroethylene acetaldehyde bromoform THF dichloromethane

21.02 19.23 26.43 31.56 36.57 41.49 46.55 27.1 35.6 NA 30.99 34.79 34.54 25.47 46.06 31.99 28.82

60 60 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250

29.75 28.64 32.74 37.59 42.93 48.54 52.98 32.62 42.43 30.58 39.13 39.08 34.40 34.03 48.77 45.45 30.50

8.38 6.14 8.38 11.75 14.93 17.84 20.39 7.13 9.43 6.28 5.43 8.60 9.56 1.17 17.29 8.22 4.14

64.15 67.54 46.57 49.41 53.51 58.69 62.28 48.73 63.09 46.44 64.43 58.28 47.48 62.82 60.19 71.17 50.39

slope of thermo values for n-alkanes at 250 °C 5.14 3.01 4.07 The slopes of the thermodynamic values with respect to the n-alkane carbon number are listed at the bottom. Heats of adsorption are calculated from the van’t Hoff equation. Theory and equations used are provided in the Supporting Information. a

Figure 4. Effects of n-alkane chain length upon the thermodynamics of adsorption on ZIF-8. Enthalpy (ΔH) and free energy (ΔG) are given in kJ/mol, and entropy (ΔS) is in J/mol/K. Table 3. Values of the Dispersive Component of Surface Energy, γSD (mJ/m2), at Each Temperature As Calculated for ZIF-8 through the Method of Dorris and Gray36,37 T (°C)

γCH2 (mJ/m2)

ΔGCH2 (kJ/mol)

γsD (mJ/m2)

alkanes used

180 190 200 210 220 230 240 250 260 270

42.15 41.57 40.99 40.41 39.83 39.25 38.67 38.09 37.51 36.93

3.56 3.50 3.46 3.41 3.38 3.09 3.05 3.01 2.97 2.92

57.46 56.55 56.11 55.18 54.88 46.55 46.01 45.64 45.02 44.33

c5-c7 c5-c7 c5-c7 c5-c7 c5-c7 c5-c9 c5-c9 c5-c9 c5-c9 c5-c9

activated carbons and zeolites which often have values of γsD reported in the range of 200-500 mJ/m2. The temperature effect upon values of γsD shown here is typical, and the values reported here are lower than those reported for IRMOF-1.28 The reduced values for γsD demonstrate the relatively inactive surface of the Langmuir 2010, 26(19), 15625–15633

ZIF-8 as compared to IRMOF-1 resulting from significantly altered chemistries of the pores and the primarily organic nature of the ZIF-8 binding sites.27 Specific Components of Adsorption. There are several methodologies used in an attempt to separate the dispersive interactions from the specific to allow for greater understanding of observed adsorption phenomena. One of the methods commonly employed is to plot the calculated free energy of adsorption (or the retention volume) versus the vapor pressure of the analyte29 using a line formed by the nonpolar n-alkane series to represent the dispersive component of adsorption. By looking at the offset of the measured values for other probe molecules from the dispersive n-alkane line, the specific component of free energy can be determined. This method has been applied successfully for some simple materials, but as can be seen in Figure 5a it does not apply well to the ZIF-8 sample studied here. In this figure, the value of the vapor pressure is calculated at the given temperature using the LeeKesler method39 and the values of the critical parameters.35,40 Values used for this calculation are given in the Supporting Information in Table SI-5. The problem seen in Figure 5a is that all of the points for polar probes lie below the nonpolar alkane line and the resulting values for the specific component of ΔG are positive, a value which makes no physical sense. From the observation that most of the data points lie below this line, it can be stated that the forces governing the analyte-analyte interactions (determining the vapor pressure) are very different from those governing the analyte-adsorbent interactions. This is not surprising considering that the measurements reported here are in the Henry’s region of adsorption at which the analyteanalyte interactions are negligible and only the interactions between the analyte and the strongest active site are probed. In addition, these plots indicate that there does not appear to be any correlation relating the intermolecular forces governing the (39) Lee, B. I.; Kesler, M. G. AIChE J. 1975, 21(3), 510. (40) Perry, R. H. Perry’s Chemical Engineer’s Handbook, 7th ed.; McGraw-Hill: 1997.

DOI: 10.1021/la102582g

15631

Article

Luebbers et al.

vapor-liquid equilibrium with those controlling the low coverage adsorption behavior. The failure of such models in identifying the specific component of adsorbate-adsorbent interactions is not unprecedented, and has led to the development of the usage of alternative molecular properties. One of the more successful properties used in place of vapor pressure is the deformation polarizability as shown in Figure 5b.30,31 The deformation polarizability is a term directly derived from the fundamental London equation, and its

Figure 5. Correlation of the measured free energies of adsorption at 250 °C on the ZIF-8 packed column to molecular properties of the adsorbates is shown. In (a) the fit to the natural logarithm of vapor pressure is shown. In (b) the fit to deformation polarizability is shown. Solid markers represent the series of n-alkanes (with the nonpolar component fit linearly), and the empty markers represent other measured analytes. Vapor pressures are calculated at 250 °C by the Lee-Kesler method.

calculation is discussed in the Supporting Information. The application of this term to such a system is in effect probing the relationship of measured retention to the molecule’s fundamental electronic dispersive ability. To calculate the deformation polarizability, molecular polarizabilities can be obtained from literature35 and the characteristic electronic frequency can be calculated as discussed in the Supporting Information. The values of both polarizability and frequency used in Figure 5b are also given in the Supporting Information in Table SI-5. From Figure 5b, it can be seen that the deformation polarizability does a much better job of giving negative values for specific component of the free energy in adsorption. In fact, there are no data points which lie below the n-alkane “nonpolar” line. The success of plotting the calculated free energy of adsorption versus deformation polarizability allows for the calculation of the specific component of the free energy of adsorption, ΔGsp, by measuring the offset above the nonpolar, n-alkane line. The values of the total, dispersive, and specific components of the free energy of adsorption are all given in Table 4 as well as the percentage of the interaction related to specific forces. For all of the compounds other than n-alkanes, the interaction with ZIF-8 shows greater than 10% of the interaction being related to specific forces. Acetaldehyde has a particularly large specific component of the free energy of adsorption (greater than 200% of its total), but some care should be taken in this particular data point as the retention (as previously mentioned) is particularly low for this compound and at this temperature. However, the following

Figure 6. Linear relationship between the specific component of the free energy of adsorption at 250 °C for ZIF-8 and the dipole moment of the adsorbate along with the discussed outlier.

Table 4. Specific Component of the Adsorption Free Energies (kJ/mol) of Various Adsorbates on ZIF-8 along with the Dipole Moment of the Compounda

n-pentane n-hexane n-heptane n-octane n-nonane

ΔGtotal (kJ/mol)

ΔGdispersive (kJ/mol)

ΔGspecific (kJ/mol)

% specific

8.38 11.75 14.93 17.84 20.39

8.52 11.75 14.54 18.13 20.36

-0.14 0.00 0.40 -0.29 0.04

-1.65 -0.03 2.67 -1.64 0.18

dipole moment (D)

diethylether 7.13 6.30 0.83 11.60 1.10 ethylacetate 9.43 6.10 3.32 35.25 1.78 1,2-dichloroethylene 6.28 5.03 1.25 19.87 0.00 acetone 5.43 1.84 3.59 66.17 2.88 MEK 8.60 5.22 3.38 39.33 2.78 trichloroethylene 9.56 8.59 0.97 10.16 0.80 acetaldehyde 1.17 -1.67 2.84 242.79 2.75 bromoform 17.29 11.58 5.70 32.99 4.75 dichloromethane 4.14 2.13 2.01 48.62 1.60 a The dispersive free energy is calculated as the offset above a nonpolar n-alkane reference line in a plot of free energy of adsorption versus deformation polarizability.

15632 DOI: 10.1021/la102582g

Langmuir 2010, 26(19), 15625–15633

Luebbers et al.

Article

analysis verifies that this observed behavior is most likely a real phenomenon due to a significant dipole moment and low dispersive abilities. It is observed that the value of the specific component of free energy correlates very well to the dipole moment as shown in Figure 6. Based on a linear fit of the data presented in Figure 6, the effect of an adsorbate’s dipole moment to the free energy of adsorption is approximately 1.2 kJ mol-1 D-1. The main outlier to this trend is that a significant specific component is calculated for 1,2-dichloroethylene despite no permanent dipole moment. As previously discussed regarding the enhanced adsorption of alkenes, there is an apparent enhancement in the adsorbent-adsorbate interaction due to the presence of π-electrons. The 1.25 kJ/mol specific component of the free energy of adsorption for 1,2dichloroethylene is most likely a result of the electrostatic interactions between an induced dipole in this adsorbate created by an electric field within the framework.

Conclusions The work reported here applies an IGC method utilizing micropacked capillaries to the study of VOC adsorption in a zeolitic imidazolate framework, ZIF-8. Equilibrium and thermodynamic values of adsorption are reported for a range of organic compounds into a commercially available sample of ZIF-8 in an attempt to explore trends in the interactions of adsorbed compound with the framework and to identify the effects of molecular properties upon the observed adsorption behavior. As was previously reported for IRMOF-1,28 it is observed that the corrected retention volumes measured for adsorbates on ZIF-8 are independent of flow rate, suggesting the measurements are representative of equilibrium conditions. This fact was not obvious a priori considering the greatly reduced pore geometries for ZIF-8, which approach the kinetic diameter of the compounds studied. It was observed that aromatic compounds, branched alkanes, and heavily halogenated compounds elute very quickly with little retention and tailing peaks, indicating a molecular sieving effect that prevents the these compounds from entering the narrow apertures of the framework crystal. The uptake of many of the compounds that were studied hint at surprising flexibility of

Langmuir 2010, 26(19), 15625–15633

the ZIF-8 framework as even the n-alkanes have kinetic diameters of 0.43 nm or larger,34 significantly greater than the reported ZIF8 pore aperture size of 0.34 nm.26,27 It can be summarized from this work that for the adsorption of gases and vapors onto ZIF-8 there are several important factors. ZIF-8 possesses lower surface energy than IRMOF-1 and many other microporous materials due to the primarily organic nature of the accessible binding sites.27 As was observed in IRMOF-1, hydrogen bond interactions appear to be a dominating factor as acids (i.e., alcohols and amines) bind too strongly to properly study in this method; however, the adsorption of bases (e.g., ketones) appears to be much less effected in ZIF-8 than in IRMOF-1 due to the absence of strong acid sites in ZIF-8. Electrostatic interactions were observed to be very important, as the specific component of the adsorption free energy was found to correlate very strongly to the adsorbates’ dipole moments. Additionally, it was observed that a significant enhancement in the adsorption of alkenes over their corresponding alkane resulted from the polarizability of the π-electrons and the interactions of the induced dipole with the framework structure. The ability for ZIF-8 to discriminate in the adsorption of both branched versus linear alkanes and paraffins versus olefins along with the material’s exceptional stability indicate great potential for this material in several important industrial applications. Acknowledgment. This work was supported by Cbana Laboratories, Inc. Any opinions, findings, and conclusions or recommendations expressed in this manuscript are those of the authors and do not necessarily reflect the views of Cbana Laboratories, Inc. Supporting Information Available: Equations and theory used for analysis; tabulated values of flow rates through the columns, retention volumes, Henry constants, enthalpy calculation values, free energies, values used for polarizability and vapor pressure calculations, and the calculation data for the dispersive component of the surface energy. This material is available free of charge via the Internet at http:// pubs.acs.org.

DOI: 10.1021/la102582g

15633