Article pubs.acs.org/Langmuir
Prediction of Molecular Separation of Polar−Apolar Mixtures on Heterogeneous Metal−Organic Frameworks: HKUST‑1 Tom R. C. Van Assche, Tim Duerinck, Stijn Van der Perre, Gino V. Baron, and Joeri F. M. Denayer* Department of Chemical Engineering, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussel, Belgium S Supporting Information *
ABSTRACT: Due to the combination of metal ions and organic linkers and the presence of different types of cages and channels, metal−organic frameworks often possess a large structural and chemical heterogeneity, complicating their adsorption behavior, especially for polar−apolar adsorbate mixtures. By allocating isotherms to individual subunits in the structure, the ideal adsorbed solution theory (IAST) can be adjusted to cope with this heterogeneity. The binary adsorption of methanol and n-hexane on HKUST-1 is analyzed using this segregated IAST (SIAST) approach and offers a significant improvement over the standard IAST model predictions. It identifies the various HKUST-1 cages to have a pronounced polar or apolar adsorptive behavior.
■
INTRODUCTION Many of the most advanced porous solids available today have a complex structure, with heterogeneous chemical and structural properties. Metal−organic frameworks (MOFs) are a telling example; these hybrid structures, built from metal cations and organic linkers, often consist of cages with different sizes, shapes, and chemical properties (e.g., HKUST-1; Figure 1), as different functionalities can be structured throughout the material. As a result, such structures typically show complex and unexpected adsorption behavior, the prediction of which is very challenging. In particular, understanding or even predicting multicomponent adsorption of mixtures containing both polar and apolar molecules is of great relevance, for example in relation to the adsorptive removal of volatile organic compounds1 or gas separation in the presence of humidity.2 Also for heterogeneous catalysis, in which the reaction performance is governed by the distribution of the different reacting species in the catalyst pores, accurate models for multicomponent adsorption equilibrium are required. A common approach to predict the adsorption of mixtures is the use of explicit models as the extended Langmuir isotherm. Such models are easy to implement but are only thermodynamically consistent under strict assumptions and have limited predictive capability. Molecular modeling is a much more advanced and powerful method to study adsorption mechanisms of pure compounds, yet it requires a huge computational effort to obtain multicomponent equilibria for molecular mixtures. Besides, it requires a suitable atomistic description of the adsorbent and strongly relies on the applied force fields. Especially, the adsorption of polar adsorbates is far more challenging and could require alteration of the model’s partial charges to fit experimental data.3 In contrast, a faster calculation method and predictive tool for multicomponent adsorption is the ideal adsorbed solution theory (IAST), put forward by Myers and Prausnitz,4 and derivations thereof such as the real adsorbed solution theory (RAST).5 The elegance of IAST lies © 2014 American Chemical Society
in its thermodynamic consistency for ideal mixtures and its ability to yield predictions based solely on pure component isotherms. IAST-based models have already been developed for binary adsorption on flexible MOFs.6−8 Nevertheless, IAST is not suited for heterogeneous materials, since a key assumption of IAST is the composition of the adsorbed phase to be uniform throughout the structure.9 Although energetic heterogeneity can be implemented into IAST by accounting for a series of pure component isotherms with sites of varying energy and a corresponding energy distribution function,10 the model outcome for this heterogeneous IAST (HIAST) method is very sensitive to the (arbitrary) choice of the energy parameters. The present work focuses on HKUST-1,11 one of the most studied MOFs, which is commercially available and has been studied for hydrogen storage,12 gas sweetening,13 carbon capture,14 xylene separation,15 and alkane−alkene16 separation. HKUST-1 is assembled from copper(II) cations and benzene1,3,5-tricarboxylate (BTC) linker molecules, which form the copper paddle-wheel building blocks, containing free metal sites, shown in Figure 1. The microporous HKUST-1 structure can be subdivided into three distinct types of cages,15,17−22 all adding to the structure’s heterogeneity, shown in Figure 1. Small octahedral S1 pockets, formed by four BTC linkers, have an internal diameter of approximately 5 Å.20 Second, larger L3 cages containing all of the free metal sites are located between the S1 pockets. The S1 pockets are only accessible via windows connecting to the L3 cages. The L3 cages have an internal diameter of approximately 9 Å.20 Finally, large L2 cages are organized in alternating fashion with the L3 cages, indicated in Figure 1. Their internal diameter is near 9 Å.20 The main difference between the L2 and L3 cages resides in the absence of free metal sites in the L2 cages. Received: May 24, 2014 Published: June 10, 2014 7878
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
■
Article
EXPERIMENTAL SECTION
HKUST-1 was obtained from Sigma-Aldrich under the commercial name Basolite C300 and stored in its activated state. The n-hexane (>96%, Biosolve) and methanol (>99.9%, Sigma-Aldrich) vapor phase isotherms (313 K) were obtained using two different gravimetric techniques. For n-hexane, a gravimetric uptake device (SGA-100H, VTI) using nitrogen (>99.998%, Air Liquide) gas as carrier was used (dynamic method). The methanol vapor phase isotherm measured with the same device was reproduced from previous work.17 The activation and measurement procedure can be found elsewhere.17 Additionally, vapor phase isotherms were measured, without carrier gas, using a pressure-controlled method (IGA, Hidden Isochema) (static method). This method allows one to achieve lower vapor pressures. Activation was performed under vacuum at 453 K for at least 6 h. Isotherm fitting is performed on all available data for nhexane and on data obtained using the static method only for methanol. Argon porosimetry measurements were performed on a Quantasorb Autosorb AS-1 (Quantachrome Instruments) device at 87 K. Vaporphase breakthrough experiments were performed with methanol and n-hexane loaded nitrogen streams, produced with temperaturecontrolled saturators. More details of the experimental setup can be found elsewhere29 and in the Supporting Information. For the breakthrough experiments the powder was pelletized and packed in a 10 cm long stainless steel column with 2.54 mm internal diameter and regenerated before use. More details on the pelletization and regeneration procedures can be found in the Supporting Information. The general IAST procedure can be found in the original paper by Myers and Prausnitz.4 The IAST calculations were performed using Newton−Raphson-based iterations in Microsoft Excel 2007.
■
RESULTS AND DISCUSSION The differentiation between the S1, L2, and L3 cages, outlined in the Introduction, and a basic understanding of the adsorption mechanisms provides a framework for predicting and understanding the adsorption and separation of polar− apolar mixtures. The overall knowledge on the adsorption properties of HKUST-1, as discussed above, is implemented by allocating parts of the experimental isotherms to the three individual cages. Subsequently, the IAST methodology can be applied to the individual cages rather than the HKUST-1 unit cell as a whole. Several variants of this overall IAST improvement have appeared throughout the literature. Only recently, segregated IAST (SIAST) was proposed by Swisher et al.30 as the terminology to denote these models. The SIAST and IAST methods have been applied to MOFs,31,32 albeit typically for apolar adsorbates and relying on computer simulated pure component and/or binary isotherms. For light gases, good results have been obtained using IAST, yet the applicability of IAST has been shown to degenerate for mixtures with disparate size and for differing polarities,31 as is the case for a methanol/n-hexane mixture. As polar adsorbate, methanol was chosen. Its experimental isotherm (Figure 2) displays a two-step adsorption mechanism. It must be noted here that the binary breakthrough experiments (vide infra) are performed using nitrogen as a carrier gas. Therefore, isotherms were also measured using the dynamic method, in which nitrogen is also used as carrier gas. Nevertheless, this dynamic method only allows measurements starting from relatively high vapor pressures, while the static method allows appreciation of the isotherm at a greater vapor pressure range, as shown in Figure 2. Both methods yield similar results in their overlapping vapor pressure range, with a slightly higher uptake for methanol using the dynamic method.
Figure 1. (Top) HKUST-1 structure with cages S1, L2, and L3. (Bottom) Front view of the upper panel showing the paddle-wheel units (purple), free metal sites pointing downward in the L3 cages of the lower plane (light green), and free metal sites pointing upward into the top plane L3 cages (yellow). Lower plane indicated by green (L3 cages) and blue (L2 cages) spheres; the top plane is not shown.
For apolar adsorbates, such as Ar,23,24 N2,25 CO2,3 CH4,3 and small alkanes,26 the S1 pockets have been identified as the preferential adsorption sites at low pressure. At higher pressure, adsorption also occurs in the larger L2 and L3 cages. Disregarding the free metal sites in the L3 cages, the L2 and L3 cages are reasonably similar in size and chemistry. Therefore, often no distinction is made between these cages regarding the adsorption of apolar adsorbates.24,26,27 For polar adsorbates, however, the distinction between L2 and L3 cages must be made. Molecules such as methanol and water17,28 preferentially adsorb near the free metal sites in the L3 cages. In recent work, we have shown that the L2 and S1 cages show a more apolar behavior compared to the L3 cages.17 The approach presented here aims at gaining a deeper insight in the adsorptive separation of polar−apolar mixtures on HKUST-1 and the prediction of such separations. The IAST approach is refined by explicitly accounting for the adsorption properties of the individual subunits in the structure of crystalline HKUST-1 to obtain a much more reliable description of the multicomponent adsorption equilibrium. A qualitative and even quantitative prediction of multicomponent adsorption for polar−apolar vapor mixtures is achieved. 7879
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
Figure 2. Experimental isotherm of methanol and n-hexane on HKUST-1 (313 K) measured using a gravimetric static uptake and gravimetric dynamic uptake method. Langmuir−Sips−Sips fit for nhexane and Sips−Sips fit for methanol are represented by the blue and red curves, respectively.
Figure 3. (A) Experimental isotherms of n-hexane and methanol on HKUST-1 at 313 K, combined with fitting of methanol (red) and nhexane (blue). Allocated isotherms of n-hexane (blue) and methanol (red) at 313 K for the (B) L3 cages, (C) L2 cages, and (D) S1 cages of HKUST-1. Results are expressed per mass unit of the total HKUST-1 structure.
The methanol experimental isotherm was fitted using a double-Sips model (Figure 2). Its first isotherm step can be explained mainly by initial adsorption near the free metal sites and filling of the L3 cages. During the second isotherm step, the L2 cages start filling, eventually producing the characteristic two-step methanol isotherm. Simulations have demonstrated that methanol shows a type I adsorption isotherm in the S1 cages.17 Guttiérrez-Sevillano et al. have simulated that a small amount of the methanol molecules that are adsorbed at low vapor pressure can be located in or near the S1 cages, yet mainly in the L3 cages.20 The L2 cages only start adsorbing methanol at higher total loadings.20 The adsorption affinity (assessed by ΔG or the Henry constant20) of the S1 cages lies in between that of the more favorable L3 cages and the least favorable sites in the L2 cages. We have qualitatively accounted for the preferential adsorption of methanol in the L3 cages by allocating the initial isotherm step (first Sips branch of the double-Sips model) solely to these L3 cages. The second Sips branch of the double-Sips model, representing the second isotherm step shown in Figure 2, is accordingly allocated to the combined behavior of the L2 and S1 cages. On the basis of the relative volume of both cages, 75% of the latter Sips isotherm is assigned to the L2 cages and the remaining 25% to the S1 pockets. The detailed allocation procedure can be found in the Supporting Information, and the allocated isotherms are shown in Figure 3B−D. In this work, n-hexane is selected as apolar adsorbate. Its very favorable isotherm is shown in Figure 2, showing measurements of a dynamic and static method. The mole-based capacity of methanol is higher than for n-hexane due to the larger molar volume of the latter. Volume-based capacities (see Supporting Information), however, are more similar. The nhexane isotherm displays a small step at low vapor pressures, followed by the occurrence of a second larger step near 10 Pa nhexane vapor pressure. This isotherm shape is similar to those reported for C4 and C5 alkanes, where a step in the isotherm was observed at low vapor pressures as soon as one molecule is adsorbed per S1 cage.26 The first step in the isotherm is accordingly attributed to adsorption in the S1 cages and fitted using a Langmuir equation with a saturation capacity of one molecule per S1 cage (Figure 3D). The large remaining step is attributed to the adsorption in the larger cages L2 and L3. To
obtain an accurate description of this step, important for applying IAST-based methods, a double-Sips isotherm was used. This double-Sips part, still representing the bulk of nhexane adsorption, can be divided into two equal shares over the L2 and L3 cages, considering their similar size and shape. As a result, new double-Sips isotherm equations for n-hexane in both the L2 as L3 cages are obtained. The detailed allocation procedure can be found in the Supporting Information and the allocated isotherms of n-hexane are shown in Figure 3B−D. Density functional analysis of the HKUST-1 argon isotherm does confirm a bimodal pore size distribution (see the Supporting Information), where only the distinction between the smaller S1 pockets and the larger (L2 and L3) cages can be made. This suggests a very similar adsorption behavior of the L2 and L3 cages for the nonpolar argon molecules, a qualitative adsorption mechanism we have applied for n-hexane above. We must, however, mention that some apolar adsorbates can display interaction with the free metal sites of HKUST-1. Examples include H2, N2, and CH4.22,33−36 A clear distinction between argon and methane can be made,22 where for methane an additional isotherm step can be attributed to the occupation of both the S1 cages and the free metal sites.22,36 Accordingly, a difference in methane adsorption in the L2 and L3 cages could be expected, contrary to the suggested similarity we have applied here for n-hexane. Previous work on larger alkanes than methane, i.e., propane, reveals an isotherm shape similar to that obtained on n-hexane (Figure 2).16 Rubes et al. have calculated the adsorption enthalpies of propane adsorbed at the free metal sites (−24.3 kJ/mol) to be significantly smaller compared to adsorption in the S1 cages and windows (−43.3 to −31.0 kJ/ mol), while the experimental enthalpies vary from −36 to −49 kJ/mol.16 For larger alkanes, such as n-hexane, the effect of the free metal sites might become small as, for example, the van der Waals forces increase. To understand the complete adsorption mechanism, further study is required, and it is clear that for smaller apolar adsorbates, such as methane, the isotherm allocation procedure should increasingly acknowledge the presence of the free metal sites. Predictions of the adsorption of n-hexane/methanol mixtures were performed using (1) IAST in a standard manner (based 7880
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
Figure 4. Experimental results (○), IAST prediction (gray), and SIAST prediction (green) for equimolar methanol/n-hexane separation on HKUST-1 at 313 K. (A) Selectivity and (B) close up of part A, with linear scale. Binary isotherms of (C) methanol and (D) n-hexane for an equimolar mixture.
Figure 5. Predicted binary isotherms of equimolar n-hexane/methanol mixture at 313 K for the (A) L3 cages, (B) L2 cages, and (C) S1 cages of HKUST-1, based on SIAST. Results are expressed per mass unit of the total HKUST-1 structure. (D) Illustrative overview of the SIAST predicted binary adsorption mechanism with increasing total vapor pressure from left to right, using the following color scheme: methanol, red; n-hexane, blue.
When the standard IAST model is used on the undivided isotherms, a large overprediction of the selectivity of methanol over n-hexane is obtained (Figure 4A,B). The IAST selectivity rapidly rises with increasing total vapor pressure and exceeds a value of 70 near a total vapor pressure of 10 kPa. These results are explained by a packing effect: IAST favors molecules with a higher saturation loading on a mole basis when reaching higher total vapor pressures. SIAST predicts a far lower selectivity for methanol over n-hexane at higher total vapor pressures. At low vapor pressures, SIAST as well as IAST, predict the preferential adsorption of n-hexane followed by a selectivity inverse near 100 Pa total vapor pressure. Then, the SIAST methanol over nhexane selectivity remains nearly constant at a value of 3.2−3.8
on the complete isotherms in Figure 3A) and (2) by applying IAST to the three cage types separately. The latter SIAST approach makes use of the hypothetical isotherms for the individual S1, L2, and L3 cages shown in Figure 3B−D to cope with the structural and chemical heterogeneity of HKUST-1. This methodology is thermodynamically correct for ideal mixtures if the three cage types show independent adsorption behavior, which is a significant advantage over many other multicomponent approaches that are thermodynamically inconsistent. Binary adsorption data were obtained from breakthrough experiments using a column packed with HKUST-1 pellets. 7881
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
for total vapor pressures of 0.5−10 kPa (Figure 4A,B), corresponding remarkably well to the experimental selectivity, which varies around an average value of 3.7. Figure 4C,D shows the amounts adsorbed with equimolar mixtures as a function of pressure, as obtained experimentally and from model predictions. SIAST clearly outperforms IAST at predicting the experimental methanol and especially the n-hexane loading. The SIAST approach accounts mathematically for the dual character of HKUST-1 and allows one to maintain a significant n-hexane loading up to large total vapor pressures, in good agreement with the experimental data, while the methanol loading is slightly overpredicted. Contrarily, standard IAST significantly overestimates methanol adsorption and completely underestimates n-hexane adsorption. The binary isotherms for n-hexane and methanol (Figure 5A−C) for the individual cages clearly reveal the adsorption mechanism, as predicted by SIAST. The L3 cages containing free metal sites show preferential adsorption of methanol over n-hexane, in accordance with their polar character. Additionally, entropic effects play a significant role, favoring the adsorption of the smaller methanol molecules. In contrast, the L2 cages preferentially adsorb n-hexane. Only at very high total vapor pressures, the selectivity will eventually be inversed as n-hexane molecules in the L2 cages are replaced by smaller methanol molecules, as this is entropically favorable. For comparison, similar behavior is predicted for an apolar MOF (see the Supporting Information). For the S1 pockets, an almost exclusive filling with n-hexane molecules is predicted, over a large pressure range. In general, the ability of each cage type to maintain its pronounced, in some cases almost exclusive, adsorption of either methanol or n-hexane results in a more constant selectivity over a broad vapor pressure range, as compared with IAST predictions. This binary adsorption mechanism as predicted by SIAST is shown graphically in Figure 5D. The SIAST method significantly improved the IAST prediction and almost exactly predicts methanol/n-hexane selectivity. The slight overprediction of experimental methanol adsorption could be located in the L3 cages as the other cages display only limited methanol adsorption (Figure 5). Further improvements here could involve acknowledging additional adsorption sites (free metal sites, cage windows).
errors can occur in predicting mixture adsorption when only relying on pure component isotherms and the standard IAST methodology. From the present analysis for the methanol/nhexane case, it is clear that standard IAST predictions should be used with caution, especially when screening or eliminating adsorbents for an application, purely based on pure component data. For computational data, a SIAST approach on key adsorption sites might significantly improve the predictions. This work also contributes to understanding the individual contributions of the structural subunits to guide engineering of novel MOFs, where, for example, specific cages are blocked to improve selectivity.20 Similar adsorptive behavior to HKUST-1 can also be expected with other MOFs comprising free-metal sites containing paddle-wheel building blocks,37 especially the tbo isorecticular series.38 For HKUST-1 and analogues, we anticipate the selectivity for polar−apolar mixtures to remain low in general, due to large local heterogeneity. However, the structured adsorption of polar molecules in one type of cages and apolar adsorbates in adjacent cages is very interesting from the catalytic point of view, where chemical reactions between polar and apolar reactants, i.e., hydration of alkenes39 or alkylation of aromatics,40 can only proceed efficiently when both species are brought in close contact in the catalyst pores. Similarly structured MOFs are also interesting to capture both polar and apolar pollutants from air simultaneously.
CONCLUSIONS The heterogeneity of HKUST-1 has a pronounced influence on the adsorption of mixtures containing polar and apolar compounds. Especially, the free metal sites located in the L3 cages have an affinity for polar adsorbates. In comparison to other polar adsorbents such as zeolites with associated cations, the polar sites in the HKUST-1 MOF are anchored rather than mobile. The concentration of these sites in a single type of cage can result in varying local adsorption properties. To cope with this structural heterogeneity, isotherms were qualitatively allocated to three distinct adsorption sites in the HKUST-1 structure. Accordingly, IAST could be applied to the individual sites to predict the overall adsorption behavior. This segregated IAST approach yields quantitatively good predictions of vapor phase methanol/n-hexane mixture adsorption on HKUST-1, based on a qualitative description of the adsorption mechanisms of polar and apolar adsorbates. Further distinction of adsorption sites, such as the S1 windows and the free metals sites, and more accurate isotherm allocation could further improve the predictions. This work also emphasizes that large
ACKNOWLEDGMENTS This research is funded by a Ph.D. grant (T.R.C.V.A.) from the Agency for Innovation by Science and Technology (IWT) Flanders. J.F.M.D. acknowledges FWO-Vlaanderen for financial support.
■
ASSOCIATED CONTENT
S Supporting Information *
Materials and methods, argon porosimetry, isotherm allocation, and IAST on apolar MOF. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
■ ■
REFERENCES
(1) Huang, Z.-H.; Kang, F.; Zheng, Y.-P.; Yang, J.-B.; Liang, K.-M. Adsorption of Trace Polar Methyl-Ethyl-Ketone and Non-Polar Benzene Vapors on Viscose Rayon-Based Activated Carbon Fibers. Carbon 2002, 8, 1363−1367. (2) Li, G.; Xiao, P.; Webley, P.; Zhang, J.; Singh, R.; Marshall, M. Capture of CO2 from High Humidity Flue Gas by Vacuum Swing Adsorption with Zeolite 13X. Adsorption 2008, 14, 415−422. (3) Castillo, J. M.; Vlugt, T. J. H.; Calero, S. Understanding Water Adsorption in Cu-BTC Metal−Organic Frameworks. J. Phys. Chem. C 2008, 112, 15934−15939. (4) Myers, A. L.; Prausnitz, J. M. Thermodynamics of Mixed-Gas Adsorption. AIChE J. 1965, 11, 121−127. (5) Erto, A.; Lancia, A.; Musmarra, D. A Real Adsorbed Solution Theory Model for Competitive Multicomponent Liquid Adsorption
7882
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
onto Granular Activated Carbon. Microporous Mesoporous Mater. 2012, 154, 45−50. (6) Coudert, F. X.; Mellot-Draznieks, C.; Fuchs, A. H.; Boutin, A. Prediction of Breathing and Gate-Opening Transitions Upon Binary Mixture Adsorption in Metal−Organic Frameworks. J. Am. Chem. Soc. 2009, 131, 11329−11331. (7) Ortiz, A. U.; Springuel-Huet, M. A.; Coudert, F. X.; Fuchs, A. H.; Boutin, A. Predicting Mixture Coadsorption in Soft Porous Crystals: Experimental and Theoretical Study of CO2/CH4 in MIL-53(Al). Langmuir 2012, 28, 494−498. (8) Coudert, F. X. The Osmotic Framework Adsorbed Solution Theory: Predicting Mixture Coadsorption in Flexible Nanoporous Materials. Phys. Chem. Chem. Phys. 2010, 12, 10904−10913. (9) Krishna, R.; van Baten, J. M. Segregation Effects in Adsorption of CO2-Containing Mixtures and Their Consequences for Separation Selectivities in Cage-Type Zeolites. Sep. Purif. Technol. 2008, 61, 414− 423. (10) Valenzuela, D. P.; Myers, A. L.; Talu, O.; Zwiebel, I. Adsorption of Gas Mixtures: Effect of Energetic Heterogenity. AIChE J. 1988, 34, 397−402. (11) Chui, S. S. Y.; Lo, S. M. F.; Charmant, J. P. H.; Orpen, A. G.; Williams, I. D. A Chemically Functionalizable Nanoporous Material [Cu3(TMA)2(H2O)3]n. Science 1999, 283, 1148−1150. (12) Liu, Y.; Brown, C.M.; Neumann, D.A.; Peterson, V.K.; Kepert, C.J. Inelastic Neutron Scattering of H2 Adsorbed in HKUST-1. J. Alloys Compd. 2007, 446-447, 385−388. (13) Chowdhury, P.; Mekala, S.; Dreisbach, F.; Gumma, S. Adsorption of CO, CO2 and CH4 on Cu-BTC and MIL-101 Metal−Organic Frameworks: Effect of Open Metal Sites and Adsorbate Polarity. Microporous Mesoporous Mater. 2012, 152, 246− 252. (14) Millward, A. R.; Yaghi, O. M. Metal−Organic Frameworks with Exceptionally High Capacity for Storage of Carbon Dioxide at Room Temperature. J. Am. Chem. Soc. 2005, 127, 17998−17999. (15) Peralta, D.; Barthelet, K.; Pérez-Pellitero, J.; Chizallet, C.; Chaplais, G.; Simon-Masseron, A.; Pirngruber, G. D. Adsorption and Separation of Xylene Isomers: CPO-27-Ni vs HKUST-1 vs NaY. J. Phys. Chem. C 2012, 116, 21844−21855. (16) Rubes, M.; Wiersum, A. D.; Llewellyn, P. L.; Grajciar, L.; Bludský, O.; Nachtigall, P. Adsorption of Propane and Propylene on CuBTC Metal−Organic Framework: Combined Theoretical and Experimental Investigation. J. Phys. Chem. C 2013, 117, 11159−11167. (17) Van Assche, T. R. C.; Duerinck, T.; Gutiérrez-Sevillano, J. J.; Calero, S.; Baron, G. V.; Denayer, J. F. M. High Adsorption Capacities and Two-Step Adsorption of Polar Adsorbates on Copper−Benzene1,3,5-tricarboxylate Metal−Organic Framework. J. Phys. Chem. C 2013, 117, 18100−18111. (18) Yang, L.; Naruke, H.; Yamase, T. A Novel Organic/Inorganic Hybrid Nanoporous Material Incorporating Keggin-type Polyoxometalates. Inorg. Chem. Commun. 2003, 6, 1020−1024. (19) Finsy, V.; De Bruyne, S.; Alaerts, L.; De Vos, D.; Jacobs, P. A.; Baron, G. V.; Denayer, J. F. M. Shape Selective Adsorption of Linear and Branched Alkanes in the Cu3(BTC)2 Metal−Organic Framework. Stud. Surf. Sci. Catal. 2007, 170, 2048−2053. (20) Gutiérrez-Sevillano, J. J.; Vicent-Luna, J. M.; Dubbeldam, D.; Calero, S. Molecular Mechanisms for Adsorption in Cu−BTC Metal Organic Framework. J. Phys. Chem. C 2013, 117, 11357−11366. (21) Calero, S.; Gutiérrez-Sevillano, J. J.; García-Pérez, E. Effect of the Molecular Interactions on the Separation of Nonpolar Mixtures Using Cu−BTC Metal−Organic Framework. Microporous Mesoporous Mater. 2013, 165, 79−83. (22) Getzschmann, J.; Senkovska, I.; Wallacher, D.; Tovar, M.; Fairen-Jimenez, D.; Düren, T.; van Baten, J. M.; Krishna, R.; Kaskel, S. Methane Storage Mechanism in the Metal−Organic Framework Cu3(btc)2: An in Situ Neutron Diffraction Study. Microporous Mesoporous Mater. 2010, 136, 50−58. (23) Vishnyakov, A.; Ravikovitch, P. I.; Neimark, A. V.; Bülow, M.; Wang, Q. M. Nanopore Structure and Sorption Properties of Cu− BTC Metal−Organic Framework. Nano Lett. 2003, 3, 713−718.
(24) Krungleviciute, V.; Lask, K.; Heroux, L.; Migone, A.D.; Lee, J.Y.; Li, J.; Skoulidas, A. Argon Adsorption on Cu3(benzene-1,3,5tricarboxylate)2(H2O)3 Metal−Organic Framework. Langmuir 2007, 23, 3106−3109. (25) Krawiec, P.; Kramer, M.; Sabo, M.; Kunschke, R.; Fröde, H.; Kaskel, S. Improved Hydrogen Storage in the Metal−Organic Framework Cu3(BTC)2. Adv. Eng. Mater. 2006, 8, 293−296. (26) Chmelik, C.; Kärger, J.; Wiebcke, M.; Caro, J.; van Baten, J. M.; Krishna, R. Adsorption and Diffusion of Alkanes in CuBTC Crystals Investigated Using Infra-Red Microscopy and Molecular Simulations. Microporous Mesoporous Mater. 2009, 117, 22−32. (27) Hamon, L.; Jolimaître, E.; Pirngruber, G. D. CO2 and CH4 Separation by Adsorption Using Cu−BTC Metal−Organic Framework. Ind. Eng. Chem. Res. 2010, 49, 7497−7503. (28) Küsgens, P.; Rose, M.; Senkovska, I.; Fröde, H.; Henschel, A.; Siegle, S.; Kaskel, S. Characterization of Metal−Organic Frameworks by Water Adsorption. Microporous Mesoporous Mater. 2009, 120, 325− 330. (29) Van Assche, T. R. C.; Denayer, J. F. M. Fabrication and Separation Performance Evaluation of a Metal−Organic Framework Based Microseparator Device. Chem. Eng. Sci. 2013, 95, 65−72. (30) Swisher, J. A.; Lin, L. C.; Kim, J.; Smit, B. Evaluating Mixture Adsorption Models Using Molecular Simulation. AIChE J. 2013, 59, 3054−3064. (31) Cessford, N. F.; Seaton, N. A.; Düren, T. Evaluation of Ideal Adsorbed Solution Theory as a Tool for the Design of Metal−Organic Framework Materials. Ind. Eng. Chem. Res. 2012, 51, 4911−4921. (32) Van Heest, T.; Teich-McGoldrick, S. L.; Greathouse, J. A.; Allendorf, M. D.; Sholl, D. S. Identification of Metal−Organic Framework Materials for Adsorption Separation of Rare Gases: Applicability of Ideal Adsorbed Solution Theory (IAST) and Effects of Inaccessible Framework Regions. J. Phys. Chem. C 2012, 116, 13183− 13195. (33) Peterson, V. K.; Liu, Y.; Brown, C. M.; Kepert, C. J. Neutron Powder Diffraction Study of D2 Sorption in Cu3(1,3,5-benzenetricarboxylate)2. J. Am. Chem. Soc. 2006, 128, 15578−15579. (34) Bordiga, S.; Regli, L.; Bonino, F.; Groppo, E.; Lamberti, C.; Xiao, B.; Wheatley, P. S.; Morris, R. E.; Zecchina, A. Adsorption Properties of HKUST-1 toward Hydrogen and Other Small Molecules Monitored by IR. Phys. Chem. Chem. Phys. 2007, 9, 2676−2685. (35) Prestipino, C.; Regli, R.; Vitillo, J. G.; Bonino, F.; Damin, A.; Lamberti, C.; Zecchina, A.; Solari, P. L.; Kongshaug, K. O.; Bordiga, S. Local Structure of Framework Cu(II) in HKUST-1 Metallorganic Framework: Spectroscopic Characterization upon Activation and Interaction with Adsorbates. Chem. Mater. 2006, 18, 1337−1346. (36) Chen, L.; Grajciar, L.; Nachtigall, P.; Düren, T. Accurate Prediction of Methane Adsorption in a Metal−Organic Framework with Unsaturated Metal Sites by Direct Implementation of an Ab Initio Derived Potential Energy Surface in GCMC Simulation. J. Phys. Chem. C 2011, 115, 23074−23080. (37) Pham, T.; Forrest, K. A.; Nugent, P.; Belmabkhout, Y.; Luebke, R.; Eddaoudi, M.; Zaworotko, M. J.; Space, B. Understanding Hydrogen Sorption in a Metal−Organic Framework with OpenMetal Sites and Amide Functional Groups. J. Phys. Chem. C 2013, 117, 9340−9354. (38) Eubank, J. F.; Mouttaki, M.; Cairns, A.J.; Belmabkhout, Y.; Wojtas, L.; Luebke, R.; Alkordi, M.; Eddaoudi, M. The Quest for Modular Nanocages: tbo-MOF as an Archetype for Mutual Substitution, Functionalization, and Expansion of Quadrangular Pillar Building Blocks. J. Am. Chem. Soc. 2011, 133, 14204−14207. (39) Kallo, D.; Mihalyi, R.M. Mechanism of l-Butene Hydration over Acidic Zeolite and Ion-Exchange Resin Catalysts. Appl. Catal., A 1995, 1, 45−56. (40) Dejaegere, E. A.; Thybault, J. W.; Marin, G. B.; Baron, G. V.; Denayer, J. F. M. Modeling of Toluene Acetylation with Acetic Anhydride on H-USY Zeolite. Ind. Eng. Chem. Res. 2011, 50, 11822− 11832.
7883
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Prediction of Molecular Separation of Polar−Apolar Mixtures on Heterogeneous Metal−Organic Frameworks: HKUST‑1 Tom R. C. Van Assche, Tim Duerinck, Stijn Van der Perre, Gino V. Baron, and Joeri F. M. Denayer* Department of Chemical Engineering, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussel, Belgium S Supporting Information *
ABSTRACT: Due to the combination of metal ions and organic linkers and the presence of different types of cages and channels, metal−organic frameworks often possess a large structural and chemical heterogeneity, complicating their adsorption behavior, especially for polar−apolar adsorbate mixtures. By allocating isotherms to individual subunits in the structure, the ideal adsorbed solution theory (IAST) can be adjusted to cope with this heterogeneity. The binary adsorption of methanol and n-hexane on HKUST-1 is analyzed using this segregated IAST (SIAST) approach and offers a significant improvement over the standard IAST model predictions. It identifies the various HKUST-1 cages to have a pronounced polar or apolar adsorptive behavior.
■
INTRODUCTION Many of the most advanced porous solids available today have a complex structure, with heterogeneous chemical and structural properties. Metal−organic frameworks (MOFs) are a telling example; these hybrid structures, built from metal cations and organic linkers, often consist of cages with different sizes, shapes, and chemical properties (e.g., HKUST-1; Figure 1), as different functionalities can be structured throughout the material. As a result, such structures typically show complex and unexpected adsorption behavior, the prediction of which is very challenging. In particular, understanding or even predicting multicomponent adsorption of mixtures containing both polar and apolar molecules is of great relevance, for example in relation to the adsorptive removal of volatile organic compounds1 or gas separation in the presence of humidity.2 Also for heterogeneous catalysis, in which the reaction performance is governed by the distribution of the different reacting species in the catalyst pores, accurate models for multicomponent adsorption equilibrium are required. A common approach to predict the adsorption of mixtures is the use of explicit models as the extended Langmuir isotherm. Such models are easy to implement but are only thermodynamically consistent under strict assumptions and have limited predictive capability. Molecular modeling is a much more advanced and powerful method to study adsorption mechanisms of pure compounds, yet it requires a huge computational effort to obtain multicomponent equilibria for molecular mixtures. Besides, it requires a suitable atomistic description of the adsorbent and strongly relies on the applied force fields. Especially, the adsorption of polar adsorbates is far more challenging and could require alteration of the model’s partial charges to fit experimental data.3 In contrast, a faster calculation method and predictive tool for multicomponent adsorption is the ideal adsorbed solution theory (IAST), put forward by Myers and Prausnitz,4 and derivations thereof such as the real adsorbed solution theory (RAST).5 The elegance of IAST lies © 2014 American Chemical Society
in its thermodynamic consistency for ideal mixtures and its ability to yield predictions based solely on pure component isotherms. IAST-based models have already been developed for binary adsorption on flexible MOFs.6−8 Nevertheless, IAST is not suited for heterogeneous materials, since a key assumption of IAST is the composition of the adsorbed phase to be uniform throughout the structure.9 Although energetic heterogeneity can be implemented into IAST by accounting for a series of pure component isotherms with sites of varying energy and a corresponding energy distribution function,10 the model outcome for this heterogeneous IAST (HIAST) method is very sensitive to the (arbitrary) choice of the energy parameters. The present work focuses on HKUST-1,11 one of the most studied MOFs, which is commercially available and has been studied for hydrogen storage,12 gas sweetening,13 carbon capture,14 xylene separation,15 and alkane−alkene16 separation. HKUST-1 is assembled from copper(II) cations and benzene1,3,5-tricarboxylate (BTC) linker molecules, which form the copper paddle-wheel building blocks, containing free metal sites, shown in Figure 1. The microporous HKUST-1 structure can be subdivided into three distinct types of cages,15,17−22 all adding to the structure’s heterogeneity, shown in Figure 1. Small octahedral S1 pockets, formed by four BTC linkers, have an internal diameter of approximately 5 Å.20 Second, larger L3 cages containing all of the free metal sites are located between the S1 pockets. The S1 pockets are only accessible via windows connecting to the L3 cages. The L3 cages have an internal diameter of approximately 9 Å.20 Finally, large L2 cages are organized in alternating fashion with the L3 cages, indicated in Figure 1. Their internal diameter is near 9 Å.20 The main difference between the L2 and L3 cages resides in the absence of free metal sites in the L2 cages. Received: May 24, 2014 Published: June 10, 2014 7878
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
■
Article
EXPERIMENTAL SECTION
HKUST-1 was obtained from Sigma-Aldrich under the commercial name Basolite C300 and stored in its activated state. The n-hexane (>96%, Biosolve) and methanol (>99.9%, Sigma-Aldrich) vapor phase isotherms (313 K) were obtained using two different gravimetric techniques. For n-hexane, a gravimetric uptake device (SGA-100H, VTI) using nitrogen (>99.998%, Air Liquide) gas as carrier was used (dynamic method). The methanol vapor phase isotherm measured with the same device was reproduced from previous work.17 The activation and measurement procedure can be found elsewhere.17 Additionally, vapor phase isotherms were measured, without carrier gas, using a pressure-controlled method (IGA, Hidden Isochema) (static method). This method allows one to achieve lower vapor pressures. Activation was performed under vacuum at 453 K for at least 6 h. Isotherm fitting is performed on all available data for nhexane and on data obtained using the static method only for methanol. Argon porosimetry measurements were performed on a Quantasorb Autosorb AS-1 (Quantachrome Instruments) device at 87 K. Vaporphase breakthrough experiments were performed with methanol and n-hexane loaded nitrogen streams, produced with temperaturecontrolled saturators. More details of the experimental setup can be found elsewhere29 and in the Supporting Information. For the breakthrough experiments the powder was pelletized and packed in a 10 cm long stainless steel column with 2.54 mm internal diameter and regenerated before use. More details on the pelletization and regeneration procedures can be found in the Supporting Information. The general IAST procedure can be found in the original paper by Myers and Prausnitz.4 The IAST calculations were performed using Newton−Raphson-based iterations in Microsoft Excel 2007.
■
RESULTS AND DISCUSSION The differentiation between the S1, L2, and L3 cages, outlined in the Introduction, and a basic understanding of the adsorption mechanisms provides a framework for predicting and understanding the adsorption and separation of polar− apolar mixtures. The overall knowledge on the adsorption properties of HKUST-1, as discussed above, is implemented by allocating parts of the experimental isotherms to the three individual cages. Subsequently, the IAST methodology can be applied to the individual cages rather than the HKUST-1 unit cell as a whole. Several variants of this overall IAST improvement have appeared throughout the literature. Only recently, segregated IAST (SIAST) was proposed by Swisher et al.30 as the terminology to denote these models. The SIAST and IAST methods have been applied to MOFs,31,32 albeit typically for apolar adsorbates and relying on computer simulated pure component and/or binary isotherms. For light gases, good results have been obtained using IAST, yet the applicability of IAST has been shown to degenerate for mixtures with disparate size and for differing polarities,31 as is the case for a methanol/n-hexane mixture. As polar adsorbate, methanol was chosen. Its experimental isotherm (Figure 2) displays a two-step adsorption mechanism. It must be noted here that the binary breakthrough experiments (vide infra) are performed using nitrogen as a carrier gas. Therefore, isotherms were also measured using the dynamic method, in which nitrogen is also used as carrier gas. Nevertheless, this dynamic method only allows measurements starting from relatively high vapor pressures, while the static method allows appreciation of the isotherm at a greater vapor pressure range, as shown in Figure 2. Both methods yield similar results in their overlapping vapor pressure range, with a slightly higher uptake for methanol using the dynamic method.
Figure 1. (Top) HKUST-1 structure with cages S1, L2, and L3. (Bottom) Front view of the upper panel showing the paddle-wheel units (purple), free metal sites pointing downward in the L3 cages of the lower plane (light green), and free metal sites pointing upward into the top plane L3 cages (yellow). Lower plane indicated by green (L3 cages) and blue (L2 cages) spheres; the top plane is not shown.
For apolar adsorbates, such as Ar,23,24 N2,25 CO2,3 CH4,3 and small alkanes,26 the S1 pockets have been identified as the preferential adsorption sites at low pressure. At higher pressure, adsorption also occurs in the larger L2 and L3 cages. Disregarding the free metal sites in the L3 cages, the L2 and L3 cages are reasonably similar in size and chemistry. Therefore, often no distinction is made between these cages regarding the adsorption of apolar adsorbates.24,26,27 For polar adsorbates, however, the distinction between L2 and L3 cages must be made. Molecules such as methanol and water17,28 preferentially adsorb near the free metal sites in the L3 cages. In recent work, we have shown that the L2 and S1 cages show a more apolar behavior compared to the L3 cages.17 The approach presented here aims at gaining a deeper insight in the adsorptive separation of polar−apolar mixtures on HKUST-1 and the prediction of such separations. The IAST approach is refined by explicitly accounting for the adsorption properties of the individual subunits in the structure of crystalline HKUST-1 to obtain a much more reliable description of the multicomponent adsorption equilibrium. A qualitative and even quantitative prediction of multicomponent adsorption for polar−apolar vapor mixtures is achieved. 7879
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
Figure 2. Experimental isotherm of methanol and n-hexane on HKUST-1 (313 K) measured using a gravimetric static uptake and gravimetric dynamic uptake method. Langmuir−Sips−Sips fit for nhexane and Sips−Sips fit for methanol are represented by the blue and red curves, respectively.
Figure 3. (A) Experimental isotherms of n-hexane and methanol on HKUST-1 at 313 K, combined with fitting of methanol (red) and nhexane (blue). Allocated isotherms of n-hexane (blue) and methanol (red) at 313 K for the (B) L3 cages, (C) L2 cages, and (D) S1 cages of HKUST-1. Results are expressed per mass unit of the total HKUST-1 structure.
The methanol experimental isotherm was fitted using a double-Sips model (Figure 2). Its first isotherm step can be explained mainly by initial adsorption near the free metal sites and filling of the L3 cages. During the second isotherm step, the L2 cages start filling, eventually producing the characteristic two-step methanol isotherm. Simulations have demonstrated that methanol shows a type I adsorption isotherm in the S1 cages.17 Guttiérrez-Sevillano et al. have simulated that a small amount of the methanol molecules that are adsorbed at low vapor pressure can be located in or near the S1 cages, yet mainly in the L3 cages.20 The L2 cages only start adsorbing methanol at higher total loadings.20 The adsorption affinity (assessed by ΔG or the Henry constant20) of the S1 cages lies in between that of the more favorable L3 cages and the least favorable sites in the L2 cages. We have qualitatively accounted for the preferential adsorption of methanol in the L3 cages by allocating the initial isotherm step (first Sips branch of the double-Sips model) solely to these L3 cages. The second Sips branch of the double-Sips model, representing the second isotherm step shown in Figure 2, is accordingly allocated to the combined behavior of the L2 and S1 cages. On the basis of the relative volume of both cages, 75% of the latter Sips isotherm is assigned to the L2 cages and the remaining 25% to the S1 pockets. The detailed allocation procedure can be found in the Supporting Information, and the allocated isotherms are shown in Figure 3B−D. In this work, n-hexane is selected as apolar adsorbate. Its very favorable isotherm is shown in Figure 2, showing measurements of a dynamic and static method. The mole-based capacity of methanol is higher than for n-hexane due to the larger molar volume of the latter. Volume-based capacities (see Supporting Information), however, are more similar. The nhexane isotherm displays a small step at low vapor pressures, followed by the occurrence of a second larger step near 10 Pa nhexane vapor pressure. This isotherm shape is similar to those reported for C4 and C5 alkanes, where a step in the isotherm was observed at low vapor pressures as soon as one molecule is adsorbed per S1 cage.26 The first step in the isotherm is accordingly attributed to adsorption in the S1 cages and fitted using a Langmuir equation with a saturation capacity of one molecule per S1 cage (Figure 3D). The large remaining step is attributed to the adsorption in the larger cages L2 and L3. To
obtain an accurate description of this step, important for applying IAST-based methods, a double-Sips isotherm was used. This double-Sips part, still representing the bulk of nhexane adsorption, can be divided into two equal shares over the L2 and L3 cages, considering their similar size and shape. As a result, new double-Sips isotherm equations for n-hexane in both the L2 as L3 cages are obtained. The detailed allocation procedure can be found in the Supporting Information and the allocated isotherms of n-hexane are shown in Figure 3B−D. Density functional analysis of the HKUST-1 argon isotherm does confirm a bimodal pore size distribution (see the Supporting Information), where only the distinction between the smaller S1 pockets and the larger (L2 and L3) cages can be made. This suggests a very similar adsorption behavior of the L2 and L3 cages for the nonpolar argon molecules, a qualitative adsorption mechanism we have applied for n-hexane above. We must, however, mention that some apolar adsorbates can display interaction with the free metal sites of HKUST-1. Examples include H2, N2, and CH4.22,33−36 A clear distinction between argon and methane can be made,22 where for methane an additional isotherm step can be attributed to the occupation of both the S1 cages and the free metal sites.22,36 Accordingly, a difference in methane adsorption in the L2 and L3 cages could be expected, contrary to the suggested similarity we have applied here for n-hexane. Previous work on larger alkanes than methane, i.e., propane, reveals an isotherm shape similar to that obtained on n-hexane (Figure 2).16 Rubes et al. have calculated the adsorption enthalpies of propane adsorbed at the free metal sites (−24.3 kJ/mol) to be significantly smaller compared to adsorption in the S1 cages and windows (−43.3 to −31.0 kJ/ mol), while the experimental enthalpies vary from −36 to −49 kJ/mol.16 For larger alkanes, such as n-hexane, the effect of the free metal sites might become small as, for example, the van der Waals forces increase. To understand the complete adsorption mechanism, further study is required, and it is clear that for smaller apolar adsorbates, such as methane, the isotherm allocation procedure should increasingly acknowledge the presence of the free metal sites. Predictions of the adsorption of n-hexane/methanol mixtures were performed using (1) IAST in a standard manner (based 7880
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
Figure 4. Experimental results (○), IAST prediction (gray), and SIAST prediction (green) for equimolar methanol/n-hexane separation on HKUST-1 at 313 K. (A) Selectivity and (B) close up of part A, with linear scale. Binary isotherms of (C) methanol and (D) n-hexane for an equimolar mixture.
Figure 5. Predicted binary isotherms of equimolar n-hexane/methanol mixture at 313 K for the (A) L3 cages, (B) L2 cages, and (C) S1 cages of HKUST-1, based on SIAST. Results are expressed per mass unit of the total HKUST-1 structure. (D) Illustrative overview of the SIAST predicted binary adsorption mechanism with increasing total vapor pressure from left to right, using the following color scheme: methanol, red; n-hexane, blue.
When the standard IAST model is used on the undivided isotherms, a large overprediction of the selectivity of methanol over n-hexane is obtained (Figure 4A,B). The IAST selectivity rapidly rises with increasing total vapor pressure and exceeds a value of 70 near a total vapor pressure of 10 kPa. These results are explained by a packing effect: IAST favors molecules with a higher saturation loading on a mole basis when reaching higher total vapor pressures. SIAST predicts a far lower selectivity for methanol over n-hexane at higher total vapor pressures. At low vapor pressures, SIAST as well as IAST, predict the preferential adsorption of n-hexane followed by a selectivity inverse near 100 Pa total vapor pressure. Then, the SIAST methanol over nhexane selectivity remains nearly constant at a value of 3.2−3.8
on the complete isotherms in Figure 3A) and (2) by applying IAST to the three cage types separately. The latter SIAST approach makes use of the hypothetical isotherms for the individual S1, L2, and L3 cages shown in Figure 3B−D to cope with the structural and chemical heterogeneity of HKUST-1. This methodology is thermodynamically correct for ideal mixtures if the three cage types show independent adsorption behavior, which is a significant advantage over many other multicomponent approaches that are thermodynamically inconsistent. Binary adsorption data were obtained from breakthrough experiments using a column packed with HKUST-1 pellets. 7881
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
for total vapor pressures of 0.5−10 kPa (Figure 4A,B), corresponding remarkably well to the experimental selectivity, which varies around an average value of 3.7. Figure 4C,D shows the amounts adsorbed with equimolar mixtures as a function of pressure, as obtained experimentally and from model predictions. SIAST clearly outperforms IAST at predicting the experimental methanol and especially the n-hexane loading. The SIAST approach accounts mathematically for the dual character of HKUST-1 and allows one to maintain a significant n-hexane loading up to large total vapor pressures, in good agreement with the experimental data, while the methanol loading is slightly overpredicted. Contrarily, standard IAST significantly overestimates methanol adsorption and completely underestimates n-hexane adsorption. The binary isotherms for n-hexane and methanol (Figure 5A−C) for the individual cages clearly reveal the adsorption mechanism, as predicted by SIAST. The L3 cages containing free metal sites show preferential adsorption of methanol over n-hexane, in accordance with their polar character. Additionally, entropic effects play a significant role, favoring the adsorption of the smaller methanol molecules. In contrast, the L2 cages preferentially adsorb n-hexane. Only at very high total vapor pressures, the selectivity will eventually be inversed as n-hexane molecules in the L2 cages are replaced by smaller methanol molecules, as this is entropically favorable. For comparison, similar behavior is predicted for an apolar MOF (see the Supporting Information). For the S1 pockets, an almost exclusive filling with n-hexane molecules is predicted, over a large pressure range. In general, the ability of each cage type to maintain its pronounced, in some cases almost exclusive, adsorption of either methanol or n-hexane results in a more constant selectivity over a broad vapor pressure range, as compared with IAST predictions. This binary adsorption mechanism as predicted by SIAST is shown graphically in Figure 5D. The SIAST method significantly improved the IAST prediction and almost exactly predicts methanol/n-hexane selectivity. The slight overprediction of experimental methanol adsorption could be located in the L3 cages as the other cages display only limited methanol adsorption (Figure 5). Further improvements here could involve acknowledging additional adsorption sites (free metal sites, cage windows).
errors can occur in predicting mixture adsorption when only relying on pure component isotherms and the standard IAST methodology. From the present analysis for the methanol/nhexane case, it is clear that standard IAST predictions should be used with caution, especially when screening or eliminating adsorbents for an application, purely based on pure component data. For computational data, a SIAST approach on key adsorption sites might significantly improve the predictions. This work also contributes to understanding the individual contributions of the structural subunits to guide engineering of novel MOFs, where, for example, specific cages are blocked to improve selectivity.20 Similar adsorptive behavior to HKUST-1 can also be expected with other MOFs comprising free-metal sites containing paddle-wheel building blocks,37 especially the tbo isorecticular series.38 For HKUST-1 and analogues, we anticipate the selectivity for polar−apolar mixtures to remain low in general, due to large local heterogeneity. However, the structured adsorption of polar molecules in one type of cages and apolar adsorbates in adjacent cages is very interesting from the catalytic point of view, where chemical reactions between polar and apolar reactants, i.e., hydration of alkenes39 or alkylation of aromatics,40 can only proceed efficiently when both species are brought in close contact in the catalyst pores. Similarly structured MOFs are also interesting to capture both polar and apolar pollutants from air simultaneously.
CONCLUSIONS The heterogeneity of HKUST-1 has a pronounced influence on the adsorption of mixtures containing polar and apolar compounds. Especially, the free metal sites located in the L3 cages have an affinity for polar adsorbates. In comparison to other polar adsorbents such as zeolites with associated cations, the polar sites in the HKUST-1 MOF are anchored rather than mobile. The concentration of these sites in a single type of cage can result in varying local adsorption properties. To cope with this structural heterogeneity, isotherms were qualitatively allocated to three distinct adsorption sites in the HKUST-1 structure. Accordingly, IAST could be applied to the individual sites to predict the overall adsorption behavior. This segregated IAST approach yields quantitatively good predictions of vapor phase methanol/n-hexane mixture adsorption on HKUST-1, based on a qualitative description of the adsorption mechanisms of polar and apolar adsorbates. Further distinction of adsorption sites, such as the S1 windows and the free metals sites, and more accurate isotherm allocation could further improve the predictions. This work also emphasizes that large
ACKNOWLEDGMENTS This research is funded by a Ph.D. grant (T.R.C.V.A.) from the Agency for Innovation by Science and Technology (IWT) Flanders. J.F.M.D. acknowledges FWO-Vlaanderen for financial support.
■
ASSOCIATED CONTENT
S Supporting Information *
Materials and methods, argon porosimetry, isotherm allocation, and IAST on apolar MOF. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
■ ■
REFERENCES
(1) Huang, Z.-H.; Kang, F.; Zheng, Y.-P.; Yang, J.-B.; Liang, K.-M. Adsorption of Trace Polar Methyl-Ethyl-Ketone and Non-Polar Benzene Vapors on Viscose Rayon-Based Activated Carbon Fibers. Carbon 2002, 8, 1363−1367. (2) Li, G.; Xiao, P.; Webley, P.; Zhang, J.; Singh, R.; Marshall, M. Capture of CO2 from High Humidity Flue Gas by Vacuum Swing Adsorption with Zeolite 13X. Adsorption 2008, 14, 415−422. (3) Castillo, J. M.; Vlugt, T. J. H.; Calero, S. Understanding Water Adsorption in Cu-BTC Metal−Organic Frameworks. J. Phys. Chem. C 2008, 112, 15934−15939. (4) Myers, A. L.; Prausnitz, J. M. Thermodynamics of Mixed-Gas Adsorption. AIChE J. 1965, 11, 121−127. (5) Erto, A.; Lancia, A.; Musmarra, D. A Real Adsorbed Solution Theory Model for Competitive Multicomponent Liquid Adsorption
7882
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
Langmuir
Article
onto Granular Activated Carbon. Microporous Mesoporous Mater. 2012, 154, 45−50. (6) Coudert, F. X.; Mellot-Draznieks, C.; Fuchs, A. H.; Boutin, A. Prediction of Breathing and Gate-Opening Transitions Upon Binary Mixture Adsorption in Metal−Organic Frameworks. J. Am. Chem. Soc. 2009, 131, 11329−11331. (7) Ortiz, A. U.; Springuel-Huet, M. A.; Coudert, F. X.; Fuchs, A. H.; Boutin, A. Predicting Mixture Coadsorption in Soft Porous Crystals: Experimental and Theoretical Study of CO2/CH4 in MIL-53(Al). Langmuir 2012, 28, 494−498. (8) Coudert, F. X. The Osmotic Framework Adsorbed Solution Theory: Predicting Mixture Coadsorption in Flexible Nanoporous Materials. Phys. Chem. Chem. Phys. 2010, 12, 10904−10913. (9) Krishna, R.; van Baten, J. M. Segregation Effects in Adsorption of CO2-Containing Mixtures and Their Consequences for Separation Selectivities in Cage-Type Zeolites. Sep. Purif. Technol. 2008, 61, 414− 423. (10) Valenzuela, D. P.; Myers, A. L.; Talu, O.; Zwiebel, I. Adsorption of Gas Mixtures: Effect of Energetic Heterogenity. AIChE J. 1988, 34, 397−402. (11) Chui, S. S. Y.; Lo, S. M. F.; Charmant, J. P. H.; Orpen, A. G.; Williams, I. D. A Chemically Functionalizable Nanoporous Material [Cu3(TMA)2(H2O)3]n. Science 1999, 283, 1148−1150. (12) Liu, Y.; Brown, C.M.; Neumann, D.A.; Peterson, V.K.; Kepert, C.J. Inelastic Neutron Scattering of H2 Adsorbed in HKUST-1. J. Alloys Compd. 2007, 446-447, 385−388. (13) Chowdhury, P.; Mekala, S.; Dreisbach, F.; Gumma, S. Adsorption of CO, CO2 and CH4 on Cu-BTC and MIL-101 Metal−Organic Frameworks: Effect of Open Metal Sites and Adsorbate Polarity. Microporous Mesoporous Mater. 2012, 152, 246− 252. (14) Millward, A. R.; Yaghi, O. M. Metal−Organic Frameworks with Exceptionally High Capacity for Storage of Carbon Dioxide at Room Temperature. J. Am. Chem. Soc. 2005, 127, 17998−17999. (15) Peralta, D.; Barthelet, K.; Pérez-Pellitero, J.; Chizallet, C.; Chaplais, G.; Simon-Masseron, A.; Pirngruber, G. D. Adsorption and Separation of Xylene Isomers: CPO-27-Ni vs HKUST-1 vs NaY. J. Phys. Chem. C 2012, 116, 21844−21855. (16) Rubes, M.; Wiersum, A. D.; Llewellyn, P. L.; Grajciar, L.; Bludský, O.; Nachtigall, P. Adsorption of Propane and Propylene on CuBTC Metal−Organic Framework: Combined Theoretical and Experimental Investigation. J. Phys. Chem. C 2013, 117, 11159−11167. (17) Van Assche, T. R. C.; Duerinck, T.; Gutiérrez-Sevillano, J. J.; Calero, S.; Baron, G. V.; Denayer, J. F. M. High Adsorption Capacities and Two-Step Adsorption of Polar Adsorbates on Copper−Benzene1,3,5-tricarboxylate Metal−Organic Framework. J. Phys. Chem. C 2013, 117, 18100−18111. (18) Yang, L.; Naruke, H.; Yamase, T. A Novel Organic/Inorganic Hybrid Nanoporous Material Incorporating Keggin-type Polyoxometalates. Inorg. Chem. Commun. 2003, 6, 1020−1024. (19) Finsy, V.; De Bruyne, S.; Alaerts, L.; De Vos, D.; Jacobs, P. A.; Baron, G. V.; Denayer, J. F. M. Shape Selective Adsorption of Linear and Branched Alkanes in the Cu3(BTC)2 Metal−Organic Framework. Stud. Surf. Sci. Catal. 2007, 170, 2048−2053. (20) Gutiérrez-Sevillano, J. J.; Vicent-Luna, J. M.; Dubbeldam, D.; Calero, S. Molecular Mechanisms for Adsorption in Cu−BTC Metal Organic Framework. J. Phys. Chem. C 2013, 117, 11357−11366. (21) Calero, S.; Gutiérrez-Sevillano, J. J.; García-Pérez, E. Effect of the Molecular Interactions on the Separation of Nonpolar Mixtures Using Cu−BTC Metal−Organic Framework. Microporous Mesoporous Mater. 2013, 165, 79−83. (22) Getzschmann, J.; Senkovska, I.; Wallacher, D.; Tovar, M.; Fairen-Jimenez, D.; Düren, T.; van Baten, J. M.; Krishna, R.; Kaskel, S. Methane Storage Mechanism in the Metal−Organic Framework Cu3(btc)2: An in Situ Neutron Diffraction Study. Microporous Mesoporous Mater. 2010, 136, 50−58. (23) Vishnyakov, A.; Ravikovitch, P. I.; Neimark, A. V.; Bülow, M.; Wang, Q. M. Nanopore Structure and Sorption Properties of Cu− BTC Metal−Organic Framework. Nano Lett. 2003, 3, 713−718.
(24) Krungleviciute, V.; Lask, K.; Heroux, L.; Migone, A.D.; Lee, J.Y.; Li, J.; Skoulidas, A. Argon Adsorption on Cu3(benzene-1,3,5tricarboxylate)2(H2O)3 Metal−Organic Framework. Langmuir 2007, 23, 3106−3109. (25) Krawiec, P.; Kramer, M.; Sabo, M.; Kunschke, R.; Fröde, H.; Kaskel, S. Improved Hydrogen Storage in the Metal−Organic Framework Cu3(BTC)2. Adv. Eng. Mater. 2006, 8, 293−296. (26) Chmelik, C.; Kärger, J.; Wiebcke, M.; Caro, J.; van Baten, J. M.; Krishna, R. Adsorption and Diffusion of Alkanes in CuBTC Crystals Investigated Using Infra-Red Microscopy and Molecular Simulations. Microporous Mesoporous Mater. 2009, 117, 22−32. (27) Hamon, L.; Jolimaître, E.; Pirngruber, G. D. CO2 and CH4 Separation by Adsorption Using Cu−BTC Metal−Organic Framework. Ind. Eng. Chem. Res. 2010, 49, 7497−7503. (28) Küsgens, P.; Rose, M.; Senkovska, I.; Fröde, H.; Henschel, A.; Siegle, S.; Kaskel, S. Characterization of Metal−Organic Frameworks by Water Adsorption. Microporous Mesoporous Mater. 2009, 120, 325− 330. (29) Van Assche, T. R. C.; Denayer, J. F. M. Fabrication and Separation Performance Evaluation of a Metal−Organic Framework Based Microseparator Device. Chem. Eng. Sci. 2013, 95, 65−72. (30) Swisher, J. A.; Lin, L. C.; Kim, J.; Smit, B. Evaluating Mixture Adsorption Models Using Molecular Simulation. AIChE J. 2013, 59, 3054−3064. (31) Cessford, N. F.; Seaton, N. A.; Düren, T. Evaluation of Ideal Adsorbed Solution Theory as a Tool for the Design of Metal−Organic Framework Materials. Ind. Eng. Chem. Res. 2012, 51, 4911−4921. (32) Van Heest, T.; Teich-McGoldrick, S. L.; Greathouse, J. A.; Allendorf, M. D.; Sholl, D. S. Identification of Metal−Organic Framework Materials for Adsorption Separation of Rare Gases: Applicability of Ideal Adsorbed Solution Theory (IAST) and Effects of Inaccessible Framework Regions. J. Phys. Chem. C 2012, 116, 13183− 13195. (33) Peterson, V. K.; Liu, Y.; Brown, C. M.; Kepert, C. J. Neutron Powder Diffraction Study of D2 Sorption in Cu3(1,3,5-benzenetricarboxylate)2. J. Am. Chem. Soc. 2006, 128, 15578−15579. (34) Bordiga, S.; Regli, L.; Bonino, F.; Groppo, E.; Lamberti, C.; Xiao, B.; Wheatley, P. S.; Morris, R. E.; Zecchina, A. Adsorption Properties of HKUST-1 toward Hydrogen and Other Small Molecules Monitored by IR. Phys. Chem. Chem. Phys. 2007, 9, 2676−2685. (35) Prestipino, C.; Regli, R.; Vitillo, J. G.; Bonino, F.; Damin, A.; Lamberti, C.; Zecchina, A.; Solari, P. L.; Kongshaug, K. O.; Bordiga, S. Local Structure of Framework Cu(II) in HKUST-1 Metallorganic Framework: Spectroscopic Characterization upon Activation and Interaction with Adsorbates. Chem. Mater. 2006, 18, 1337−1346. (36) Chen, L.; Grajciar, L.; Nachtigall, P.; Düren, T. Accurate Prediction of Methane Adsorption in a Metal−Organic Framework with Unsaturated Metal Sites by Direct Implementation of an Ab Initio Derived Potential Energy Surface in GCMC Simulation. J. Phys. Chem. C 2011, 115, 23074−23080. (37) Pham, T.; Forrest, K. A.; Nugent, P.; Belmabkhout, Y.; Luebke, R.; Eddaoudi, M.; Zaworotko, M. J.; Space, B. Understanding Hydrogen Sorption in a Metal−Organic Framework with OpenMetal Sites and Amide Functional Groups. J. Phys. Chem. C 2013, 117, 9340−9354. (38) Eubank, J. F.; Mouttaki, M.; Cairns, A.J.; Belmabkhout, Y.; Wojtas, L.; Luebke, R.; Alkordi, M.; Eddaoudi, M. The Quest for Modular Nanocages: tbo-MOF as an Archetype for Mutual Substitution, Functionalization, and Expansion of Quadrangular Pillar Building Blocks. J. Am. Chem. Soc. 2011, 133, 14204−14207. (39) Kallo, D.; Mihalyi, R.M. Mechanism of l-Butene Hydration over Acidic Zeolite and Ion-Exchange Resin Catalysts. Appl. Catal., A 1995, 1, 45−56. (40) Dejaegere, E. A.; Thybault, J. W.; Marin, G. B.; Baron, G. V.; Denayer, J. F. M. Modeling of Toluene Acetylation with Acetic Anhydride on H-USY Zeolite. Ind. Eng. Chem. Res. 2011, 50, 11822− 11832.
7883
dx.doi.org/10.1021/la5020253 | Langmuir 2014, 30, 7878−7883
