Recovery of Dimethyl Sulfoxide from Aqueous Solutions by Highly

Oct 18, 2012 - Recovery of Dimethyl Sulfoxide from Aqueous Solutions by Highly Selective Adsorption in .... Adsorption of HMF from Water/DMSO Solution...
2 downloads 0 Views 2MB Size
Article pubs.acs.org/Langmuir

Recovery of Dimethyl Sulfoxide from Aqueous Solutions by Highly Selective Adsorption in Hydrophobic Metal−Organic Frameworks Anjaiah Nalaparaju and Jianwen Jiang* Department of Chemical and Biomolecular Engineering, National University of Singapore, 117576, Singapore S Supporting Information *

ABSTRACT: Metal−organic frameworks (MOFs) have emerged as a new family of nanoporous materials. While gas separation in MOFs has been extensively investigated, liquid separation is scarcely examined and lacks a microscopic understanding. A molecular simulation study is reported here for the recovery of dimethyl sulfoxide (DMSO) from aqueous solutions in three hydrophobic MOFs, namely, Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. Type I adsorption isotherms are observed for DMSO, while H2O exhibits type V adsorption isotherms with hysteresis. The saturation capacities of both DMSO and H2O decrease following the order of Zn4O(bdc)(bpz)2 > Zn(bdc)(ted)0.5 > ZIF-71, in accordance with the variation of free volume and porosity in the three MOFs. As attributed to hydrophobic frameworks, the three MOFs are highly selective toward DMSO adsorption from DMSO/H2O mixtures. The highest selectivity is predicted up to 1700 in ZIF-71. This simulation study provides molecular insight into the separation mechanism of DMSO/H2O liquid mixtures and suggests that hydrophobic MOFs are superior candidates for DMSO recovery.

1. INTRODUCTION Adsorption in porous materials is a technically feasible and economically effective approach for the recovery/separation of organics from aqueous solutions. It has the advantage of offering less contamination and heating for target compounds.1 A wide range of porous materials, such as activated carbons, resins, zeolites, and their derivatives, have been investigated as adsorbents to recover organics.2 The adsorption of C1−C5 alcohols from dilute aqueous solutions was studied in a organophilic silicalite.3 Recovery of heat-sensitive carboxylic acids in polymer sorbents was reported.4 The effects of surface chemistry and pore structure of activated carbon on the adsorption of polar and nonpolar solvents were examined.5 Generally, an ideal adsorbent should possess large surface area and pore volume and high adsorption capacity. There has been considerable interest to develop new porous materials for highperformance recovery of organics from aqueous solutions. In the past decade, metal−organic frameworks (MOFs) have emerged as a new family of porous materials.6 MOFs can be synthesized from various inorganic clusters and organic linkers and, thus, possess a wide range of surface areas and pore sizes. More fascinatingly, the judicious selection of building blocks allows for the structure and functionality to be tailored in a rational manner. Consequently, MOFs are considered as versatile materials for storage, separation, catalysis, and biomedical applications.7,8 Most experimental and simulation studies for MOFs have been primarily focused on gas storage and separation, particularly the storage of H2 and CH4 for friendly energy carriers and the separation of CO2-containing gas mixtures for CO2 capture.9−12 A handful of experimental studies have also been reported on the exploitation of MOFs © 2012 American Chemical Society

toward liquid separation. For example, the adsorption capacity and rate of benzene were measured in MIL-101 from a 1000 ppm solution.13 On the basis of the size difference, the selective adsorption of water over methanol was observed in a rationally tuned microporous MOF.14 Biobutanol purification from a fermentation medium containing C1−C5 alcohols, acetone, and water was demonstrated by adsorption in ZIF-8.15 A total of 29 adsorbents comprising of MOFs, zeolites, activated carbon, and silica gel were examined for the separation of liquid mixtures.16 In contrast, simulation studies for liquid separation in MOFs are scarce.17 This is primarily because the force field used in simulation may not yield activities consistent with those experimentally measured. To the best of our knowledge, only two simulation studies have been reported on liquid separation in MOFs, one for water desalination18 and the other for biofuel purification.19 In the latter, the separation of biofuel mimicked by water/ethanol mixtures was simulated in two MOFs, namely, hydrophilic Na-rho-ZMOF and hydrophobic Zn4O(bdc)(bpz)2. The predicted permselectivities in the two MOFs are largely determined by adsorption rather than diffusion. In addition, hydrophobic Zn4O(bdc)(bpz)2 is superior to hydrophilic Na-rho-ZMOF in terms of separation efficiency. The aim of this study is to investigate the recovery of dimethyl sulfoxide (DMSO) from aqueous solutions using MOFs. DMSO is miscible with most organic compounds, as attributed to its aprotic polar and amphiphilic nature. In addition, DMSO is essentially odorless, thermally stable, and Received: August 23, 2012 Revised: October 11, 2012 Published: October 18, 2012 15305

dx.doi.org/10.1021/la3034116 | Langmuir 2012, 28, 15305−15312

Langmuir

Article

Figure 1. Crystal structures of Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. Color code: C, gray; O, red; H, white; and Cl, green. Zn clusters are denoted as polyhedra. The sizes are not on the same scale. occupy the axial sites to extend the 2D layers into a three-dimensional (3D) framework and are disordered along the crystallographic 4-fold axis. The structure belongs to a space group of P4/mmm, with lattice constants of a = 10.929 Å and c = 9.608 Å. The middle panel of Figure 1 shows the crystal structure (3 × 3 × 3 unit cells) on the (100) plane, constructed from experimental crystallographic data and firstprinciples optimization.26 There exist interlacing pores in the framework of Zn(bdc)(ted)0.5. Specifically, the open square pores are along the z axis, with a diameter ranging from 7.3 to 9.2 Å, which are connected by a small window along the x and y axes, with a diameter of 3.6 Å. ZIF-71 has a space group of Pm3̅m and lattice constant of 28.554 Å.27 A unit cell of ZIF-71 possesses a truncated cuboctahedron (α cage) with 48 Zn atoms. Each Zn atom is coordinated with four N atoms of 4,5-dicholoroimidazolate ligands to form four-coordinated molecular building block. The substitution of oxygen in rho-zeolite with imidazolate-based ligands leads to an open framework in ZIF-71, almost twice as large as in rho-zeolite. Thus, the α cages possess a large diameter of 16.8 Å connected by a window of 4.8 Å. The free volume Vfree in each of the three MOFs was evaluated from

has a low level of ecotoxicity. In comparison to common solvents, such as dimethylformamide and tetrahydrofuran, DMSO is more environmentally friendly. Thus, DMSO is an ideal solvent in chemical and biological industries.20 Particularly, DMSO is often used as a co-solvent with H2O to form biphase medium in various practical applications. For example, the chemotherapeutic drug busulfan is dissolved in DMSO/ H2O mixtures to enhance its solubility and stability.21 The formulation of polymers, dyes, and electronics is widely conducted in DMSO/H2O mixtures.22 For recycling purposes, DMSO is required to be recovered from aqueous solutions. Although a three-stage distillation technology has been applied for DMSO recovery,23 it is energy-intensive because the majority of water needs to be evaporated and not economical when aqueous solutions are dilute in DMSO. Therefore, adsorption might be a better technology to recover DMSO from dilute aqueous solutions. Toward this end, DMSO recovery is examined in the current study by simulation in three MOFs, Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. A deep microscopic understanding from simulation is indispensable to improve the separation performance. The three MOFs are hydrophobic and stable in water and organic solvents. The reason that only hydrophobic MOFs are considered is because our recent study has revealed that the hydrophobic MOF outperforms hydrophilic counterparts in biofuel purification.19 In section 2, the simulation models and methods are briefly described. In section 3, the adsorption isotherms of pure DMSO and H2O are first presented, as well as their structural properties in the MOFs, and then the adsorption and separation of DMSO/H2O mixtures from aqueous solutions are discussed. Finally, the concluding remarks are summarized in section 4.

Vfree =

∫V

He exp[− uad (r )/kBT ]dr

(1)

where uHe ad is the interaction between non-adsorbing helium and the framework. Helium was modeled as a Lennard−Jones (LJ) particle, with σHe = 2.58 Å and εHe/kB = 10.22 K.28 The ratio of the free volume over the framework volume gives porosity. The atomic charges and LJ parameters of the three MOFs are listed in the Supporting Information, as adopted from our previous studies.19,26,29 In each MOF, the atomic charges were calculated by the density-functional theory (DFT) on a fragmental cluster. The DFT calculations used the Becke exchange plus Lee−Yang−Parr functional (B3LYP) and were carried out using Gaussian 03.30 The basis set of 6-31G(d) was used for all atoms, except metal atoms, for which the LANL2DZ basis set was used. LANL2DZ is a double-ζ basis set and contains effective pseudo-potentials to represent the potentials of nucleus and core electrons. The atomic charges were fit to the electrostatic potentials using the Merz−Kollman scheme.31,32 The intermolecular interactions of DMSO and H2O were represented by LJ and Coulombic potentials

2. MODELS AND METHODS Figure 1 illustrates the crystal structures of Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. Zn4O(bdc)(bpz)2 belongs to a space group of P42/mcm with lattice constants of a = 11.5228 Å and c = 25.7692 Å.24 The framework is assembled by tetrahedral Zn4O with mixed ligands 1,4-benezenedicarboxylate (bdc) and 3,3′,5,5′-tetramethyl-4,4′-bipyrazolate (bpz). Each Zn4O is edge-bridged by four pyrazolate and two carboxylate groups, resulting in an octahedral Zn4O(O2C−)2(NN−)4 building unit. All of the Zn4O clusters are connected by bpz linkers on the (001) plane and further pillared along the (001) plane to form a porous network. Zn4O(bdc)(bpz)2 contains cube-like cavities connected by a window of 4.9 × 6.8 Å2 along the (100, 010) plane and another window of 5.7 × 5.7 Å2 along the (001) plane. The phenyl rings of bdc ligands and the methyl groups of bpz ligands contribute to the hydrophobic nature of the cavities. Zn(bdc)(ted)0.5 possesses a paddle-wheel structure, in which metal oxides Zn2(COO)4 are bridged by bdc linkers to form a twodimensional (2D) square-grid net [Zn2(1,4-bdc)2].25 The ted pillars

⎡⎛ ⎞12 ⎛ ⎞6 ⎤ qiqj σij σij U (rij) = ∑ 4εij⎢⎢⎜⎜ ⎟⎟ − ⎜⎜ ⎟⎟ ⎥⎥ + r ⎝ rij ⎠ ⎦ 4πε0rij i,j ⎣⎝ ij ⎠

(2)

where rij is the distance between atom i and j, εij and σij are the well depth and collision diameter, respectively, qi is the atomic charge, and ε0 = 8.8542 × 10−12 C2 N−1 m−2 is the permittivity of the vacuum. The Lorentz−Berthelot combining rules were used to estimate the cross LJ parameters. Specifically, DMSO was modeled by the P1 potential with methyl groups treated as united atoms.33 The bond lengths S−C and S−O are 1.80 and 1.53 Å, respectively, and the bond angles ∠OSC and ∠CSC are 106.75° and 97.4°, respectively. Among the available force fields for DMSO, the P1 potential best describes the representation of charge separation on S−O atoms. In addition, the P1 potential yields 15306

dx.doi.org/10.1021/la3034116 | Langmuir 2012, 28, 15305−15312

Langmuir

Article

thermodynamic, structural, and transport properties consistent with experimental data. H2O was mimicked by the three-point transferable interaction potential (TIP3P) model.34 The O−H bond length is 0.9572 Å, and the ∠HOH angle is 104.52°. The TIP3P model gives reasonably good interaction energy compared to the experimental value.35 Table 1 lists the LJ potential parameters and atomic charges of DMSO and H2O.

treatment, the simulation was accelerated by 2 orders of magnitude. A spherical cutoff equal to half of the box length was used to evaluate the LJ interactions, and long-range corrections were added beyond the cutoff. For the Coulombic interactions, Ewald summation with a tinfoil boundary condition was used, where the real/reciprocal space partition parameter and the cutoff for reciprocal lattice vectors were 0.2 Å−1 and 8, respectively, to ensure convergence. The number of trial moves in a typical simulation was 2 × 107, in which the first 107 moves were used for equilibration and the subsequent 107 moves were used for ensemble averages. Different types of trial moves were randomly attempted in the GCMC simulation, namely, displacement, rotation, and partial regrowth at a neighboring position, complete regrowth at a new position, swap between reservoirs, including creation and deletion with equal probability, and the exchange of molecular identity (for mixtures).40

Table 1. LJ Potential Parameters and Atomic Charges of DMSO and H2O adsorbate DMSO H2O

site

σ (Å)

ε/kB (K)

q (e)

CH3 S O OW HW

3.80 3.40 2.80 3.151

147.94 119.96 35.99 76.47

0.0 0.54 −0.54 −0.834 +0.417

3. RESULTS AND DISCUSSION 3.1. Pure DMSO and H2O. Figure 2a shows the adsorption isotherms of pure DMSO in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71 as a function of reduced pressure (P/Psat) on a linear scale. Because of the similarity of the hydrophobic nature of the three MOFs, the adsorption behavior of DMSO has a resemblance. The adsorption isotherms can be classified as type I, implying strong adsorbate−adsorbent interactions. At infinite dilution, the isosteric heats Qost in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71 are 61.3, 51.5, and 63.5 kJ/mol, respectively. The saturation capacity is 10 mmol/g in Zn4O(bdc)(bpz)2 and drops to 8.0 mmol/g in Zn(bdc)(ted)0.5 and 4.4 mmol/g in ZIF-71. Such a hierarchy is consistent with the decreasing order of free volume and porosity in the three MOFs. Specifically, the free volumes in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71 are approximately 0.87, 0.79, and 0.44 cm3/g, respectively, and the porosities are 69.4, 64.9 and 50.4%, respectively. To clearly understand the adsorption of DMSO at low pressures, Figure 2b shows the isotherms on a semi-log scale of P/Psat. A sharp rise of adsorption in Zn(bdc)(ted)0.5 and Zn4O(bdc)(bpz)2 is observed at P/Psat = 8 × 10−4 and 1.5 × 10−3, respectively. In ZIF-71, however, the adsorption increases gradually and approaches saturation after P/Psat = 10−1. Figure 3 illustrates the center-of-mass distributions of DMSO in the three MOFs at 5 × 10−5 kPa (P/Psat = 6.3 × 10−4) and 5 × 10−4 kPa (P/Psat = 6.3 × 10−3). At 5 × 10−5 kPa, DMSO in Zn4O(bdc)(bpz)2 is adsorbed proximal to the Zn4O cluster corner, as well as the phenyl ring and methyl group. In Zn(bdc)(ted)0.5, DMSO is located in the open square pore. In ZIF-71, DMSO is localized only in the small six-memebered ring, which possesses strong overlap of the surface potential, as

In DMSO/H2O liquid mixtures, the fugacity of component i was estimated by

⎛ V (P − Pisat) ⎞ fi = PisatϕisatXiγi exp⎜ i ⎟ RT ⎠ ⎝

(3)

sat where Psat i is the saturation pressure, ϕi is the fugacity coefficient, Xi is mole fraction in liquid feed solution, γi is the activity coefficient, V̅ i is the partial molar volume, and P and T are operating pressure and temperature, respectively. At the operating conditions considered in this study (P = 1 bar and T = 298 K), ϕsat i and the Poynting factor (the exponential term) are approximately equal to unity. The saturation pressures of DMSO and H2O at 298 K are 0.079 and 3.14 kPa, respectively.36 The activity coefficients were estimated by the nonrandom two liquid (NRTL) model with parameters from the literature,37 and good agreement was found between the estimated and experimental results. Differently, we should note that Sandler and co-workers recently implemented expanded ensemble simulation to determine adsorbate fugacities in liquid mixtures.38 Adsorption of DMSO/H2O liquid mixtures in the three MOFs was simulated by the grand canonical Monte Carlo (GCMC) method. In addition, adsorption of pure DMSO and H2O was also examined. Because the chemical potentials of an adsorbate in adsorbed and bulk phases are identical at thermodynamic equilibrium, the GCMC method allows one to directly relate the fugacity in both phases and has been widely used to simulate adsorption. The simulation box contained 9 (3 × 3 × 1) unit cells of Zn4O(bdc)(bpz)2, 27 (3 × 3 × 3) unit cells of Zn(bdc)(ted)0.5, or 1 unit cell of ZIF-71. Thus, the box sizes of the three MOFs were approximately the same. The periodic boundary conditions were exerted in three dimensions. The framework atoms were assumed to be rigid, and the unit cell was divided into fine grids, with the potential energies pretabulated and subsequently used by interpolation during simulation.39 In such a

Figure 2. Adsorption isotherms of pure DMSO in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71 on a (a) linear and (b) semi-log scale. 15307

dx.doi.org/10.1021/la3034116 | Langmuir 2012, 28, 15305−15312

Langmuir

Article

are the favorable adsorption sites in Zn4O(bdc)(bpz)2; the C2 atom of the methyl group is relatively less favorable. Similarly, the favorable sites in ZIF-71 are the Zn atom of the ZnN4 cluster and the C1 atom of imidazolate. Figure 5 shows the adsorption and desorption isotherms of H2O in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. It

Figure 3. Center-of-mass distributions of DMSO at 5 × 10−5 kPa (P/ Psat = 6.3 × 10−4, top) and 5 × 10−4 kPa (P/Psat = 6.3 × 10−3, bottom) in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71.

reflected in the large Qost. At 5 × 10−4 kPa, DMSO starts to fill the pore in Zn4O(bdc)(bpz)2, form a layer in Zn(bdc)(ted)0.5, and enter marginally into the α cage in ZIF-71. In this context, the sharp rise of adsorption in Zn4O(bdc)(bpz)2 and Zn(bdc)(ted)0.5 in Figure 2b can be attributed to pore filling and layer formation, respectively. However, the adsorption in ZIF-71 increases gradually because of the weak potential in the large α cage. The detailed structural information of DMSO in the three MOFs is provided by the analysis of radial distribution functions g(r), which were calculated from gij(r ) =

Figure 5. Isotherms of pure H2O in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. The filled and open symbols are adsorption and desorption, respectively.

should be noted that H2O adsorption simulated in Zn(bdc)(ted)0.5 is consistent with experimental measurement, as reported in our previous study.26 All of the isotherms are of S-shaped type V, which signifies the adsorption of weakly interacting adsorbate in the microporous adsorbent. The isosteric heats of H2O adsorption at infinite dilution are 27.3, 18.3, and 15.8 kJ/mol in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71, respectively. As expected, these values are substantially smaller than DMSO adsorption. At pressures below saturation, vanishingly small adsorption of H2O is observed, indicating that the three MOFs are hydrophobic and no preferential affinity for H2O. It is energetically unfavorable for H2O to break hydrogen bonds in the bulk phase and then adsorb in a hydrophobic environment. Only at sufficiently high pressures can adsorption be distinctly observed. At P/Psat = 2.6, 4.4, and 10, H2O exhibits a sharp rise of adsorption and approaches saturation in Zn(bdc)(ted)0.5, Zn4O(bdc)(bpz)2, and ZIF-71, respectively. In comparison to Zn4O(bdc)(bpz)2 and Zn(bdc)(ted)0.5, the onset pressure in ZIF-71 is higher because of the large α cages. The occurrence of a sharp rise in adsorption is due to capillary condensation through nucleation of H2O and subsequent formation of liquid-like H2O. In each MOF, a hysteresis loop is observed in adsorption and

ΔNij(r , r + Δr )V 4πr 2ΔrNN i j

(4)

where r is the distance between species i and j, ΔNij(r, r + Δr) is the number of species j around i within a shell from r to r + Δr, V is the system volume, and Ni and Nj are the numbers of species i and j, respectively. Figure 4 shows the g(r) of DMSO at 5 × 10−5 kPa, where the loadings in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71 are 0.116, 0.0931, and 1.48 molecules per unit cell, respectively. Pronounced peaks are observed in Zn4O(bdc)(bpz) 2 and ZIF-71 but not in Zn(bdc)(ted)0.5. This suggests that the favorable adsorption sites in Zn4O(bdc)(bpz)2 and ZIF-71 are at specific locations; however, they are relatively more homogeneous in Zn(bdc)(ted)0.5. Such a phenomenon is in accordance with the centerof-mass distributions of DMSO in Figure 3. Specifically, the Zn atom of the Zn4O cluster and the C1 atom of the phenyl ring

Figure 4. Radial distribution functions of DMSO at 5 × 10−5 kPa in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. 15308

dx.doi.org/10.1021/la3034116 | Langmuir 2012, 28, 15305−15312

Langmuir

Article

Figure 6. Adsorption isotherms of DMSO/H2O mixtures versus molar concentration cDMSO (top x axis) and mole fraction XDMSO (bottom x axis) in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. The regime of low XDMSO is in the inset.

Figure 7. Distributions of DMSO (green) and H2O (pink) for a DMSO/H2O mixture (XDMSO = 0.05) in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71.

desorption. As attributed to how easy H2O can fill and drain the pores, hysteresis suggests the existence of metastable states and local minima in the grand free energy of the system.41 A similar behavior was observed for H2O in hydrophobic zeolites42,43 and alkanes in carbon nanotubes.44,45 This type of hysteresis is the signature of the phase transition in ordered open pores. Because of the large α cages in ZIF-71, condensed water in ZIF71 exists in a continuous liquid-like state. Thus, the extent of hysteresis is greater than in Zn4O(bdc)(bpz)2 and Zn(bdc)(ted)0.5. In addition, in comparison to DMSO adsorption in Figure 2, H2O has larger saturation capacities in all of the three MOFs. This is because H2O is smaller than DMSO, and thus, a greater number of H2O molecules are adsorbed at saturation. As observed for DMSO, however, the saturation capacity of H2O also decreases in the order of Zn4O(bdc)(bpz)2 > Zn(bdc)(ted)0.5 > ZIF-71, following the decreasing order of free volume and porosity in the three MOFs. 3.2. DMSO/H2O Liquid Mixtures. Figure 6 shows the adsorption isotherms of DMSO/H2O mixtures in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71 versus XDMSO, the mole composition of DMSO in aqueous solutions. The largest XDMSO under study is 5% (approximately 20% by weight), which is a common composition of DMSO aqueous solutions industrially encountered. Apparently, DMSO is much more preferentially adsorbed over H2O, as attributed to the selective adsorption in hydrophobic MOFs. With increasing XDMSO, the adsorption amounts of DMSO and H2O in Zn4O(bdc)(bpz)2 and Zn(bdc)(ted)0.5 increase and approach saturation. In ZIF-71, DMSO adsorption monotonically increases and H2O adsorption is nearly constant. This is because DMSO adsorption in ZIF-71 is low within the range of XDMSO considered. Specifically, the amount of DMSO adsorption in ZIF-71 is

only 0.3 mmol/g at XDMSO = 0.05, substantially lower than 4.4 mmol/g for pure DMSO at saturation. In the regime of low XDMSO ( Zn(bdc)(ted)0.5 > ZIF-71. In DMSO/H2O liquid mixtures, the three hydrophobic MOFs are highly selective toward DMSO adsorption. The substantially high selectivity predicted suggests that DMSO can be efficiently recovered in the three MOFs. Nevertheless, Zn4O(bdc)(bpz)2 and Zn(bdc)(ted)0.5 outperform ZIF-71 in DMSO recovery if considering both adsorption capacity and selectivity. Currently, simulation studies for liquid separation in MOFs are rare. A deep microscopic understanding from a molecular level is thus indispensable for the development of new MOFs for DMSO recovery and other practically important liquid separations.

(5)

where Yi and Xi are the compositions of species i in the adsorbed phase and feed solution, respectively. Figure 8a shows the adsorption selectivity of DMSO over H2O as a function of XDMSO in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71. With increasing XDMSO, the selectivity in each MOF initially increases and then decreases within the range of XDMSO under study. The initial increase is attributed to the enhanced interaction of DMSO with multiple adsorption sites in the framework. However, the interaction strength drops when most adsorption sites are occupied, and furthermore, smaller H2O molecules are more likely to fill into the framework. Consequently, a decrease in selectivity is seen at high XDMSO. In particular, a sharp increase of selectivity in Zn4O(bdc)(bpz)2 is observed at XDMSO = 0.02, which is due to the sharp rise of DMSO adsorption at this XDMSO (see Figure 6). In general, the three MOFs appear to be highly selective toward DMSO adsorption because of their hydrophobic frameworks. Particularly, ZIF-71 is the most hydrophobic and, thus, has the highest selectivity. Approximately, this can be quantified by the difference in Qost between DMSO and H2O. Specifically, Qost(DMSO) − Qost(H2O) is 34.0, 33.2, and 47.7 kJ/mol in Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71, respectively. Therefore, ZIF-71 has the largest difference in Qost. Figure 8b shows the mole fraction YDMSO in the adsorbed phase as a function of XDMSO. Within the range of XDMSO from 0.02 to 0.05, YDMSO is about 0.85, 0.92, and 0.98 in Zn(bdc)(ted)0.5, Zn4O(bdc)(bpz)2, and ZIF-71, respectively. This suggests that DMSO in dilute aqueous solutions can be efficiently recovered by adsorption in MOFs. While the adsorbed phase in ZIF-71 is composed of almost 100% DMSO, the recovery capacity of DMSO in ZIF-71 is significantly smaller than in Zn4O(bdc)(bpz)2 and Zn(bdc)(ted)0.5. From a practical point of view, Zn4O(bdc)(bpz)2 and Zn(bdc)(ted)0.5 might be better candidates among the three MOFs for DMSO recovery.



ASSOCIATED CONTENT

S Supporting Information *

Atomic charges and LJ potential parameters of framework atoms in Zn4O(bdc)(bpz)2, ZIF-71, and Zn(bdc)(ted)0.5. This material is available free of charge via the Internet at http:// pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors are grateful to Prof. Stanley I. Sandler for helpful suggestion and the National University of Singapore (R-279000-297-112) and the National Research Foundation of Singapore (R-279-000-261-281) for financial support.



REFERENCES

(1) Vane, L. M. Separation technologies for the recovery and dehydration of alcohols from fermentation broths. Biofuels, Bioprod. Biorefin. 2008, 2, 553−588. (2) Deng, S. Sorbent technology. Encyclopedia of Chemical Processing; Taylor and Francis Group: New York, 2006. (3) Milestone, N. B.; Bibby, D. M. Concentration of alcohols by adsorption on silicalite. J. Chem. Technol. Biotechnol. 1981, 31, 732− 736. (4) Garcia, A. A.; King, C. J. The use of basic polymer sorbents for the recovery of acetic acid from dilute aqueous solution. Ind. Eng. Chem. Res. 1989, 28, 204−212. (5) Li, L.; Quinlivan, P. A.; Knappe, D. R. U. Effects of activated carbon surface chemistry and pore structure on the adsorption of

4. CONCLUSION A molecular simulation study has been reported for DMSO recovery from aqueous solutions by adsorption in MOFs. The MOFs examined include Zn4O(bdc)(bpz)2, Zn(bdc)(ted)0.5, and ZIF-71 with hydrophobic frameworks. DMSO exhibits strong adsorption in the three MOFs with type I isotherms, whereas H2O adsorption is weak and characterized as type V. In Zn4O(bdc)(bpz)2, DMSO is adsorbed close to the metal cluster corner, phenyl ring, and methyl group. In Zn(bdc)(ted)0.5 and ZIF-71, DMSO is located inside the open square 15310

dx.doi.org/10.1021/la3034116 | Langmuir 2012, 28, 15305−15312

Langmuir

Article

experimental and simulation study. J. Phys. Chem. C 2010, 114, 6602−6609. (27) Banerjee, R.; Phan, A.; Wang, B.; Knobler, C.; Furukawa, H.; O’Keeffe, M.; Yaghi, O. M. High-throughput synthesis of zeolitic imidazolate frameworks and application to CO2 capture. Science 2008, 319, 939−943. (28) Hirschfelder, J. O.; Curtiss, C. F.; Bird, R. B. Molecular Theory of Gases and Liquids; John Wiley: New York, 1964. (29) Nalaparaju, A.; Zhao, X. S.; Jiang, J. W. Molecular understanding for the adsorption of water and alcohols in hydrophilic and hydrophobic zeolitic metal−organic frameworks. J. Phys. Chem. C 2010, 114, 11542−11550. (30) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifforde, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 03, Revision D.01; Gaussian, Inc.: Wallingford, CT, 2004. (31) Singh, U. C.; Kollman, P. A. An approach to computing electrostatic charges for molecules. J. Comput. Chem. 1984, 5, 129− 145. (32) Besler, B. H.; Merz, K. M.; Kollman, P. A. Atomic charges derived from semiempirical methods. J. Comput. Chem. 1990, 11, 431− 439. (33) Luzar, A.; Chandler, D. Structure and hydrogen-bond dynamics of water−dimethyl sulfoxide mixtures by computer simulations. J. Chem. Phys. 1993, 98, 8160−8173. (34) Jorgenson, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Simple potential function for water. J. Chem. Phys. 1983, 79, 926. (35) Mark, P.; Nilsson, L. Structure and dynamics of liquid water with different long-range interaction truncation and temperature control methods in molecular dynamics simulations. J. Comput. Chem. 2002, 23, 1211−1219. (36) National Institute of Standards and Technology (NIST). NIST Chemistry Webbook; Linstrom, P. J., Ed., NIST: Gaithersburg, MD, 2010. (37) Peng, Y.; Ping, L. J.; Lu, S. L.; Mao, J. W. Vapor−liquid equilibria for water + acetic acid + N,N-dimethylformamide or dimethyl sulfoxide at 13.33 kPa. Fluid Phase Equilib. 2009, 275, 27−32. (38) Xiong, R. C.; Sandler, S. I.; Vlachos, D. G. Molecular screening of alcohol and polyol adsorption onto MFI-type zeolites. Langmuir 2012, 28, 4491−4499. (39) June, R. L.; Bell, A. T.; Theodorou, D. N. Prediction of low occupancy sorption of alkanes in silicalite. J. Phys. Chem. 1990, 94, 1508−1516. (40) Frenkel, D.; Smit, B. Understanding Molecular Simulation: From Algorithms to Applications; Academic Press: San Diego, CA, 2002. (41) Monson, P. A. Understanding adsorption/desorption hysteresis. Microporous Mesoporous Mater. 2012, 160, 47−66. (42) Fleys, M.; Thompson, R. W. Monte Carlo simulations of water adsorption isotherms in silicalite and dealuminated zeolite Y. J. Chem. Theory Comput. 2005, 1, 453−458. (43) Cailliez, F.; Stirnemann, G.; Boutin, A.; Demachy, I.; Fuchs, A. H. Does water condense in hydrophobic cavities? A molecular simulation study of hydration in heterogeneous nanopores. J. Phys. Chem. C 2008, 112, 10435−10445. (44) Jiang, J. W.; Sandler, S. I.; Smit, B. Capillary phase transitions of n-alkanes in a carbon nanotube. Nano Lett. 2004, 4, 241−244.

organic contaminants from aqueous solution. Carbon 2002, 40, 2085− 2100. (6) Yaghi, O. M.; O’Keefe, M.; Ockwig, N. W.; Chae, H. K.; Eddaoudi, M.; Kim, J. Reticular synthesis and design of new materials. Nature 2003, 423, 705−714. (7) Horcajada, P.; Serre, C.; Maurin, G.; Ramsahye, N. A.; Balas, F.; Vallet-Regi, M.; Sebban, M.; Taulelle, F.; Ferey, G. Flexible porous metal−organic frameworks for a controlled drug delivery. J. Am. Chem. Soc. 2008, 130, 6774−6780. (8) Long, J. R.; Yaghi, O. M. The pervasive chemistry of metal− organic frameworks. Chem. Soc. Rev 2009, 38, 1213−1214. (9) Suh, M. P.; Park, H. J.; Prasad, T. K.; Lim, D. W. Hydrogen storage in metal−organic frameworks. Chem. Rev. 2012, 112, 782−835. (10) Sumida, K.; Rogow, D. L.; Mason, J. A.; McDonald, T. M.; Bloch, E. D.; Herm, Z. R.; Bae, J. H.; Long, J. R. Carbon dioxide capture in metal−organic frameworks. Chem. Rev. 2012, 112, 724− 781. (11) Getman, R. B.; Bae, Y. S.; Wilmer, C. E.; Snurr, R. Q. Review and analysis of molecular simulations of methane, hydrogen and acetylene storage in metal−organic frameworks. Chem. Rev. 2012, 112, 703−723. (12) Jiang, J. W.; Babarao, R.; Hu, Z. Q. Molecular simulations for energy, environmental and pharmaceutical applications of nanoporous materials: From zeolites, metal−organic frameworks to protein crystals. Chem. Soc. Rev. 2011, 40, 3599−3612. (13) Jhung, S. H.; Lee, J. H.; Yoon, J. W.; Serre, C.; Ferey, G.; Chang, J. S. Microwave synthesis of chromium terephthalate MIL-101 and its benzene sorption ability. Adv. Mater. 2007, 19, 121−125. (14) Chen, B. L.; Ji, Y. Y.; Xue, M.; Fronczek, F. R.; Hurtado, E. J.; Mondal, J. U.; Liang, C. D.; Dai, S. Metal−organic framework with rationally tuned micropores for selective adsorption of water over methanol. Inorg. Chem. 2008, 47, 5543−5545. (15) Saint Remi, J. C.; Remy, T.; Van Hunskerken, V.; van de Perre, S.; Duerinck, T.; Maes, M.; De Vos, D.; Gobechiya, E.; Kirschhock, C. E. A.; Baron, G. V.; Denayer, J. F. M. Biobutanol separation with the metal−organic framework ZIF-8. ChemSusChem 2011, 4, 1074−1077. (16) Van Assche, T. R. C.; Remy, T.; Desmet, G.; Baron, G. V.; Denayer, J. F. M. Adsorptive separation of liquid water/acetonitrile mixtures. Sep. Purif. Technol. 2011, 82, 76−86. (17) Jiang, J. W. Recent development of in silico molecular modeling for gas and liquid separations in metal−organic frameworks. Curr. Opin. Chem. Eng. 2012, 1, 138−144. (18) Hu, Z. Q.; Chen, Y. F.; Jiang, J. W. Zeolitic imidazolate framework-8 as a reverse osmosis membrane for water desalination: Insight from molecular simulation. J. Chem. Phys. 2011, 134, 134705. (19) Nalaparaju, A.; Zhao, X. S.; Jiang, J. W. Biofuel purification by pervaporation and vapor permeation in metal−organic frameworks: A computational study. Energy Environ. Sci. 2011, 4, 2107−2116. (20) Yu, Z. W.; Quinn, P. J. Dimethyl-sulfoxide: A review of its applications in cell biology. Biosci. Rep. 1994, 14, 259−281. (21) Bhagwatwar, H. P.; Phadungpojna, S.; Chow, D. S.; Andersson, B. S. Formulation and stability of busulfan for intravenous administration in high-dose chemotherapy. Cancer Chemother. Pharmacol. 1996, 37, 401−408. (22) Park, S. J.; Yoon, T. I.; Bae, J. H.; Seo, H. J.; Park, H. J. Biological treatment of wastewater containing dimethyl sulphoxide from the semi-conductor industry. Process Biochem. 2001, 36, 579− 589. (23) DMSO Recovery; www.arkema.com/dmso. (24) Hou, L.; Lin, Y. Y.; Chen, X. M. Porous metal−organic framework based on μ4-oxo tetrazinc clusters: Sorption and guestdependent luminescent properties. Inorg. Chem. 2008, 47, 1346−1351. (25) Lee, J. Y.; Olson, D. H.; Pan, L.; Emge, T. J.; Li, J. Microporous metal−organic frameworks with high gas sorption and separation capacity. Adv. Funct. Mater. 2007, 17, 1255−1262. (26) Chen, Y. F.; Lee, J. Y.; Babarao, R.; Li, J.; Jiang, J. W. A highly hydrophobic metal organic framework Zn(bdc)(ted)0.5 for adsorption and separation of CH3OH/H2O and CO2/CH4: An integrated 15311

dx.doi.org/10.1021/la3034116 | Langmuir 2012, 28, 15305−15312

Langmuir

Article

(45) Jiang, J. W.; Sandler, S. I. Capillary phase transitions of linear and branched alkanes in carbon nanotubes from molecular simulation. Langmuir 2006, 22, 7391−7399. (46) Omota, L. M.; Iulian, O.; Omota, F.; Ciocirlan, O. Densities and derived properties of water, dioxane and dimethyl sulfoxide binary and ternary systems. Rev. Roum. Chim. 2009, 54, 63−73.

15312

dx.doi.org/10.1021/la3034116 | Langmuir 2012, 28, 15305−15312