Article pubs.acs.org/JPCA
Reversible Ligand Exchange in a Metal−Organic Framework (MOF): Toward MOF-Based Dynamic Combinatorial Chemical Systems Adam F. Gross, Elena Sherman, Sky L. Mahoney, and John J. Vajo* HRL Laboratories, LLC, 3011 Malibu Canyon Road, Malibu, California 90265, United States S Supporting Information *
ABSTRACT: Reversible benzene dicarboxylate/2-bromobenzene dicarboxylate ligand exchange has been studied in the cubic metal−organic framework MOF-5. Significant exchange (up to ∼50%), with continuous compositional variation, was observed using exsitu 1H NMR following treatment over ∼6 h at ∼85 °C in 10−40 mM ligand solutions. Exchange occurred without significant structural degradation as characterized by X-ray diffraction, nitrogen adsorption, and scanning electron microscopy. Solid-state 13C NMR was used to show that exchanged ligands were incorporated into the framework lattice and not simply adsorbed within the pores. Exchange was found to be sensitive to the small free energy changes caused by the ligand concentration in the exchanging solution indicating that exchange is energetically nearly degenerate. This demonstration of reversible, nearly isoenergetic exchange indicates that mixed ligand MOFs could be developed as dynamic combinatorial chemical systems.
■
INTRODUCTION Metal−organic frameworks (MOFs), consisting of organic linker ligands and metal or oxo−metal cluster ion vertices, have been synthesized from an enormous variety of ligands and metals. Although numerous individual MOFs are known, most often, each MOF is composed of a single ligand/metal pair.1 However, for selected sets of ligands with a common linking chemistry and similar dimensions, randomly mixed multiple linker MOFs have been formed.2−5 A well-studied example containing Zn4O vertices and benzene dicarboxylate (BDC)based linkers is Zn4O(X-BDC)3 (MOF-5, when X = H), which has been synthesized with up to eight different simultaneously incorporated BDC-based linkers.3 These linkers all contain the same dicarboxylate linking chemistry but have various functional groups (X) attached to the BDC phenyl rings including X = NH2, Br, Cl, NO2, and CH3. MOFs with multiple metal vertices have also been synthesized using chemically similar metal cations.6 For example, lanthanide-based MOFs containing mixtures of up to three cations (Yb3+, Er3+, and Nd3+) have been reported.6a These mixed multiple ligand/vertices MOFs render the number of formally chemically distinct MOF materials virtually unlimited because even micrometer scale MOF crystals contain ∼1020 linker and vertices making nearly continuous composition variation possible. This variation is important because the properties of MOFs such as porosity, gaseous and fluid phase adsorptivity, and catalytic and optical activity depend not only on the ligands and metal cations7−9 but also on their concentrations and possibly on their spatial organization within a MOF crystal. For example, mixed linker MOF-5 materials were shown to have improved hydrogen storage capacities and CO/CO2 adsorption selectivities as compared to single ligand frameworks.3 Moreover, the effects were not simple linear combinations of the properties of their © 2013 American Chemical Society
single-linker counterparts. Although clearly enhanced properties were observed, these studies examined only a few possible variations selected by the starting compositions. The existence of mixed composition MOFs and the effects of composition on a MOF’s properties suggests the intriguing possibility that if exchange of ligands and/or metal cations were reversible and occurred with small changes in free energy, multiple ligand/ metal MOFs could form a MOF-based dynamic combinatorial chemical system with variation in both composition and spatial organization. Dynamic combinatorial chemical systems utilize a library of components to dynamically form supramolecular structures that reduce free energy in response to environmental pressures, such as a molecular template.10,11 This approach could lead to MOFs with compositions and internal organization that do not need to be designed a priori but rather are self- or evolutionarily optimized for adsorption, catalytic or optical activity, or molecular recognition or sensing by interaction with properly chosen environments.11 Reversible formation of mixed linker MOFs based on Zr(IV)/BDC-X (UiO-66-X), Al/BDC-X (MIL-53), and In/ BDC-X (MIL-68) were reported recently.12,13 Exchange was observed between solid UiO-66-NH2 and solid UiO-66-Br, or between solid MIL type MOFs, when crystals were placed in various liquid phases. This suggests that exchange might be indirectly occurring by a dissolution/recrystallization mechanism. Nevertheless, the exchange of BDC-NH2 into UiO-66-Br and the reverse, BDC-Br into UiO-66-NH2, was observed to occur in water and other solvents over ∼5 days.12 The extent of exchange was significant (10−75%). Exchange by BDC-NH2 Received: January 29, 2013 Revised: April 10, 2013 Published: April 15, 2013 3771
dx.doi.org/10.1021/jp401039k | J. Phys. Chem. A 2013, 117, 3771−3776
The Journal of Physical Chemistry A
Article
occurred to a greater extent (∼2×) reflecting a small thermodynamic bias. Although some bias between different ligands is expected, the result that the bias was small is important because for exchange in a dynamic combinatorial system to be influenced by the environment, the combinatorial members should be nearly thermodynamically degenerate.10 Exchange of dissolved ligands with solid zeolite imidazolate frameworks (ZIFs) was shown to be reversible for a subset of ligands.13,14 A Cd-based ZIF with an ethylimidazolate linker (CdIF-4) could be exchanged with 2-methylimidazolate (forming SALEM-1) or 2-nitroimidazolate (forming CdIF-9), but the reverse exchange was only possible for SALEM-1.14 In another report, 4-bromo-1H-imidazole replaced up to 30% of the ligands in ZIF-71; however, the reversibility of this exchange was not measured.13 Direct reversible exchange of metal vertices has been reported.13,15,16 For example, in a [Cd4O]6+-based MOF, the reversible substitution of Pb2+ for Cd2+ was reported to occur with smoothly varying intermediate mixed cation concentrations again with only a slight thermodynamic (or perhaps entirely kinetic) bias.15a Most relevant to MOF-5 used in this work, the exchange of up to 25% of Zn atoms with Ni was recently reported, which required soaking Zn-based MOF-5 in a Ni(NO3)2·6H2O in DMF solution for one year at room temperature.17 In related areas, reversible exchange of ligands was reported in metal−organic polyhedra, which are soluble finite molecular complexes as opposed to extended crystals.18 In addition, the irreversible exchange of 4,4′-bipyridine for a longer bipyridal ligand in 2D and 3D MOFs was observed.19 New chemical functionalities may also be introduced after synthesis as a method to modify MOF chemistry, although this is similarly an irreversible process.20 Heterogeneous MOF structures with core−shell architectures4 as well as MOF/nanoparticle composites21 have also been synthesized. In this work, we demonstrate the direct reversible ligand exchange of Br-BDC for BDC and vice versa in MOF-5 and show that exchange occurs with small changes in free energy. This establishes the dynamical character of MOFs required for MOF-based dynamic combinatorial chemical systems.
■
Table 1. Samples, Compositions, and Exchange Conditions sample designation 1 2 3 4 5 6 7 8
sample description synthesized BDC-based MOF-5 Br-BDC-exchanged 1 BDC-exchanged 2 Br-BDC-exchanged 1 Br-BDC-exchanged 1 synthesized Br-BDC-based MOF BDC-exchanged 6 directly synthesized BDC/ Br-BDC mixed linker MOF-5
BDC/BrBDC mol % ratio
exchange conditions time (h)/ T (°C)
100/0
−
84.8/15.2 92.5/7.5 79.0/21.0 47.4/52.6 0/100
6/85 6/85 6/90 24/85 −
40.7/59.3 53.9/46.1
6/85 −
Figure 1. Powder XRD patterns for as-synthesized and exchanged samples. (1) As-synthesized BDC MOF-5, (2) Br-BDC-exchanged 1, (6) as-synthesized Br-BDC-based MOF-5, (7) BDC-exchanged 6, and (8) as-synthesized BDC/Br-BDC mixed linker MOF-5. All of the patterns reflect the symmetry of MOF-5, indicating that exchange occurs without significant structural rearrangement.
■
RESULTS AND DISCUSSION Structural and Compositional Characterization of Exchange. Figure 1 shows XRD patterns of the as-synthesized MOFs (1, 6, and 8) and the singly exchanged samples (2 and 7). All of the patterns index to the known XRD powder pattern for MOF-5,3,23 indicating crystallographic structural retention during exchange. The ligand compositions (mole percent BDC/mole percent Br-BDC), analyzed using 1H NMR of digested samples (Supporting Information), are given in Table 1. The analyses indicate that 15.2% of the BDC ligands in BDC MOF-5 (1) were exchanged with Br-BDC at 85 °C (sample 2). For Br-BDC MOF-5 (6) exchanged with BDC (yielding 7), replacement of 40.7% of the Br-BDC ligand occurred. Higher levels of exchange were achieved by increasing the temperature or the reaction time, samples 4 and 5, respectively. For sample 4, with 21% exchange, no degradation was seen by XRD, although after treatment for 24 h at 85 °C and exchange of 52.6% of the ligands (sample 5), additional small diffraction peaks were observed (Supporting Information Figure S1) The surface areas and pore volumes versus Br-BDC content for the exchanged and directly synthesized samples are shown in Figure 2. For the directly synthesized samples (1, 6, and 8) both the surface area and pore volume decrease linearly with
EXPERIMENTAL METHODS
Exchange was conducted by mixing ∼500 μm crystals of activated MOFs with 40 mM solutions of Br-BDC or BDC in diethylformamide (DEF) that contained 488 mM water.22 To initiate exchange, the suspensions were heated for 0.5−6 or 24 h at 85 or 90 °C. Crystals were observed throughout the exchange process. After exchange, crystals underwent six solvent exchanges to remove free ligands and were activated prior to analysis. No yields were calculated because some crystals were lost during washing in preparation for postexchange analysis. After activation, the samples were characterized using X-ray diffraction (XRD), nitrogen adsorption for surface area and pore volume, solid-state 1H and 13C MAS NMR, 1H solution NMR following digestion, scanning electron microscopy (SEM), and energy dispersive X-ray analysis (EDX). Sample designations, descriptions, and exchange conditions are given in Table 1. Further details concerning the synthesis, exchange, and characterization are given in the Supporting Information. 3772
dx.doi.org/10.1021/jp401039k | J. Phys. Chem. A 2013, 117, 3771−3776
The Journal of Physical Chemistry A
Article
Figure 3. 13C MAS NMR spectra of as-synthesized and exchanged MOF-5. As-synthesized samples: 1 (BDC MOF-5), 6 (Br-BDC MOF5), and 8 (mixed linker MOF-5). Exchanged sample: 7 (BDCexchanged 6). The similarity of the spectra for 7 and 8 suggests that BDC is fully incorporated into the framework during exchange.
BDC MOF-5 (6, yielding 7). The BDC-based MOF-5 (1) spectrum has peaks at 129, 136, and 175 ppm due to the carbon atoms in the phenyl ring, in the phenyl ring attached to the carboxylic acid, and in the carboxylic acid groups, respectively. This data matches the known 13C MAS NMR peak positions for BDC MOF-5.24 In addition to those peaks, both the directly synthesized mixed linker MOF-5 (8) and the BDC-exchanged Br-BDC MOF-5 (7) have peaks at 131.3, 126.8, and 122.9 ppm. These peak positions are very close to those reported for a mixed BDC/Br-BDC MOF-5.3 The peak at 126.8 ppm is from the carbon in the phenyl ring attached to the Br, and the peaks at 122.9 and 131.3 ppm are from other carbons in the phenyl ring whose resonance was shifted by Br.3 The Br-BDCbased MOF-5 (6) is similar with small additional peaks due to residual solvent. The growth of the strong resonance at 129 ppm upon exchange of BDC into 6 (yielding 7) and the close match between the spectra of 7 and 8 indicate that the exchanged and directly synthesize samples are chemically very similar. Finally, the −COO carbon peaks in all samples occur at 175 ppm, which is shifted relative to the free ligand peaks at 173 ppm,3 confirming the absence of either the exchanging or displaced ligands in acid form. Kinetic and Mechanistic Details. The kinetics of the exchange reaction were investigated by heating samples of BDC MOF-5 (1) for 0.5−6 h at 85 °C in 10 or 40 mM Br-BDC in DEF containing 488 mM water, recovering the samples, digesting the samples in acid, and then determining the extent of exchange with solution 1H NMR. The results are shown in Figure 4. The exchange does not reach equilibrium within 6 h, although it is initially rapid, with half of the total observed exchange occurring in ∼0.5 h. The kinetics are not well described by a parabolic growth curve but are approximately logarithmic.25 This suggests that exchange is not occurring uniformly throughout individual crystals or from the solution/ crystal interface by simple diffusion. Rather, the logarithmic growth curve indicates that the initially rapid exchange leads to the formation of a surface layer barrier.25 This barrier may be related to steric effects caused by the Br-BDC ligand. As BrBDC substitutes into the BDC MOF-5 framework, the pore size is reduced due to the bulky Br-BDC molecules and additional molecules experience a larger steric barrier for diffusing through the framework. Thus, the local Br content
Figure 2. Specific surface areas and pore volumes versus Br-BDC content of as-synthesized and exchanged samples. Top: Surface area determined by N2 adsorption BET analysis. Sample designations refer to Table 1. Red solid curve through the starting samples 1 and 6 and the exchanged samples 2, 3, 4, 5, and 7 (filled circles) is a guide to the eye. Dashed line shows surface area dependence for as-synthesized samples 1, 6, and 8 (filled square). Solid line shows surface area dependence from a series of mixed linker MOFs reported in ref 2. Bottom: Total pore volume for same sample set determined by N2 adsorption using a single-point pore volume. Arrows indicate the exchange paths used in both graphs. Triangles (top and bottom) show surface area and pore volume, respectively, for an as-synthesized mixed linker sample from ref 3. The pore volumes for the exchanged samples fall under the linear dependences seen from direct syntheses indicating some filling or degradation of the pores.
Br-BDC content as expected for the heavier Br-BDC ligand if each ligand, BDC or Br-BDC, contributed a given surface area or pore volume without any nonlinear interactions. As shown in Figure 2, the trend for surface area is similar to that reported for a series of directly synthesized mixed BDC/Br-BDC MOFs.2 The surface areas and pores volumes of the exchanged samples remain high suggesting that the ligands are indeed exchanging rather than simply filling the pores. However, the surface areas and pore volumes fall below a linear trend indicating some degradation or clogging of the pores during exchange. The deviation from the linear dependence is up to ∼30% at ∼50% exchange. This is not surprising because the greatest amount of exchange from initially either 100% or 0% Br-BDC probably results in the greatest degradation. The exchanged sample data do agree, however, with the results from ref 3 of the direct synthesis of mixed linker MOF-5. Molecular Characterization Using NMR. Solid-state 1H NMR of the exchanged samples (Supporting Information Figure S2) showed no indication of the carboxylic acid protons from the free ligands reinforcing that exchange, as opposed to physisorption of ligands on the framework, had occurred. Further comparison with as-synthesized samples was not possible due to the large widths of the solid-state 1H NMR resonances. However, 13C MAS NMR spectra, shown in Figure 3, reveal differences between pure and exchanged mixed linker samples clearly indicating the incorporation of BDC into Br3773
dx.doi.org/10.1021/jp401039k | J. Phys. Chem. A 2013, 117, 3771−3776
The Journal of Physical Chemistry A
Article
acidic (due to the Zn salt) than the exchange solutions used in this work. Because MOF dissolution is generally favored by acidic conditions and likely did not occur in the (more acidic) core/shell structure study, dissolution is even less likely in the exchange experiments in this work. Finally, no crystals were observed on the container walls during exchange, which would be a sign of crystallization from solution that is observed during MOF-5 synthesis. Due to the thermodynamic stability of MOF5 in the exchange conditions, the absence of crystal nucleation on the container walls, and previous work at higher temperatures and longer times demonstrating heterostructure growth rather than dissolution, exchange is the likely the dominant process in these experiments. Dynamical Character of Ligand Exchange. The applicability of these exchange reactions for use in a dynamic combinatorial system requires reversibility. To demonstrate reversible ligand exchange, the partially Br-BDC-exchanged sample 2 was further treated in a solution of 40 mM BDC in DEF with 488 mM water (yielding 3). Approximately 50% of the initially incorporated Br-BDC was removed, lowering the Br-BDC ligand fraction from 15.2% to 7.5% (Table 1). The MOF-5 XRD structure was maintained (Supporting Information Figure S6) while the surface area and pore volume were similar to singly exchanged samples (Figure 2). Exchange beginning from both 0% and 100% Br-BDC and reversible exchange in a single sample depending on the concentrations of the ligands in the surrounding solution indicates that there is not a large thermodynamic driving force for any particular ligand composition, i.e., that both BDC and Br-BDC ligands are energetically similar. From the experiments in Table 1, we estimate the difference in free energy for the exchange of Br-BDC into MOF-5 to be in the range of −3.3 to 1.4 kJ/mol (details in the Supporting Information). This range is on the order of ±kT (∼3 kJ/mol) at the reaction temperature of 85 °C. Such a small free energy difference demonstrates that, at least for the BDC/Br-BDC ligand pair, MOFs are a dynamic system that can be reversibly chemically modified by the surrounding environment. The reversibility of ligand exchange is an important step toward using MOFs in dynamic combinatorial chemical systems.10,11 In these systems, supramolecular assemblies form from a library of components and evolve over time in response to thermodynamic driving forces (i.e., pressures) in their environment. These thermodynamic pressures can be any material, such as component concentrations or templates, or energetic variables, such as temperature, that influence the system free energy and stabilize products formed from a subset of the components. Thus, dynamic combinatorial systems have the potential to create materials with environmentally responsive properties.11 MOFs that undergo reversible ligand or metal exchange could act as extended assemblies for dynamic combinatorial libraries consisting of multiple ligands and metals in solution. The metals and ligands could provide catalytic, optical, or gas sorption properties that are optimized by the correct thermodynamic pressures. For example, if a templating material such as CO228 were introduced into a MOF solution with multiple exchanging ligands, it is possible that adsorption of the CO2 on the MOF and its interaction with the different ligands would lead to a mixed ligand MOF with ligand concentrations and possibly even atomic scale spatial organization optimized for CO2 adsorption. Considerable further work is needed to explore this possibility.
Figure 4. Extent of exchange versus time. Beginning with 1 (BDC MOF-5), exchange was conducted in 10 or 40 mM (as shown) BrBDC with 488 mM H2O in DEF at 85 °C. Solid curves are fits versus time (t) to a logarithmic dependence given by k1 ln(k2t + k3) with fitting parameters ki. Dashed curve shows the best fit for 40 mM BrBDC with a parabolic dependence given by kt1/2. Both exchanges are best described by logarithmic kinetics.
may vary within individual crystals leading to a Br-rich surface layer. An attempt was made to map this distribution. However, no gradients in the Br concentration were found using SEMbased EDX, which has a spatial resolution of a few micrometers.26 The exchange reaction was also examined using SEM on BrBDC based MOF-5 (6) before and after exchange with BDC. As-synthesized, 6 was composed of 100−500 μm sized crystals with clean surfaces. After exchange, the crystal size of 7 was
