Metal–Organic Frameworks with Internal Urea-Functionalized

Oct 4, 2017 - Introduction of a urea R–NH–CO–NH–R group as a seven-membered diazepine ring at the center of 4,4′-biphenyl-dicarboxylic acid ...
0 downloads 0 Views 7MB Size
Research Article www.acsami.org

Cite This: ACS Appl. Mater. Interfaces 2017, 9, 37419-37434

Metal−Organic Frameworks with Internal Urea-Functionalized Dicarboxylate Linkers for SO2 and NH3 Adsorption Sebastian Glomb, Dennis Woschko, Gamall Makhloufi, and Christoph Janiak* Institut für Anorganische Chemie und Strukturchemie, Heinrich-Heine-Universität Düsseldorf, Universitätsstraße 1, Düsseldorf 40225, Germany S Supporting Information *

ABSTRACT: Introduction of a urea R−NH−CO−NH−R group as a sevenmembered diazepine ring at the center of 4,4′-biphenyl-dicarboxylic acid leads to a urea-functionalized dicarboxylate linker (L12−) from which four zinc metal−organic frameworks (MOFs) could be obtained, having a {Zn4(μ4O)(O2C−)6} SBU and IRMOF-9 topology (compound [Zn4(μ4-O)(L1)3], 1, from dimethylformamide, DMF) or a {Zn2(O2C−)4} paddle-wheel SBU in a 2D-network (compound [Zn2(L1)2(DEF)2·2.5DEF], 2, from diethylformamide, DEF). Pillaring of the 2D-network of 2 with 4,4′-bipyridine (bipy) or 1,2-bis(4-pyridyl)ethane (bpe) gives 3D frameworks with rhombohedrally distorted pcu-a topologies ([Zn2(L1)2(bipy)], 3 and [Zn2(L1)2(bpe)], 4, respectively). The 3D-frameworks 1, 3, and 4 are 2-fold interpenetrated with ∼50% solvent-accessible volume, albeit of apparently dynamic porous character, such that N2 adsorption at 77 K does not occur, while H2 at 77 K (up to ∼1 wt %) and CO2 at 293 K (∼5 wt %) are adsorbed with large hystereses in these flexible MOFs. The ureafunctionalized MOF 3 exhibits an uptake of 10.9 mmol g−1 (41 wt %) of SO2 at 293 K, 1 bar, which appears to be the highest value observed so far. Compounds 3 and 4 adsorb 14.3 mmol g−1 (20 wt %) and 17.8 mmol g−1 (23 wt %) NH3, respectively, which is at the top of the reported values. These high uptake values are traced to the urea functionality and its hydrogen-bonding interactions to the adsorbents. The gas uptake capacities follow the specific porosity of the frameworks, in combination with pore aperture size and affinity constants from fits of the adsorption isotherms. KEYWORDS: MOFs, metal−organic frameworks, urea-functionalized linker, sulfur dioxide sorption, ammonia sorption, interpenetration, hydrogen bonding, gas sorption



INTRODUCTION Metal−organic frameworks (MOFs) are potentially porous, crystalline, and extended organic−inorganic hybrid compounds built from metal nodes and bridging organic linkers.1 MOFs have attracted widespread research interest because of their high surface area, tailorable porosity, and tunable composition. They promise, among others, applications in small molecule adsorption, storage and separation,2−7 catalysis,8,9 sensing,10,11 drug delivery,12 and heat transformation.13 Flexible or soft porous networks belong to the so-called third-generation of porous coordination polymers (PCPs). They have gained much attention because of their interesting flexible and dynamic behavior.14−16 Third generation PCPs show a reversible dynamic response dependent on external stimuli, such as changes in pressure17 or temperature18 or even in the presence/absence of specific guest molecules. As a result of these properties, potential applications for flexible MOFs are in the selective gas adsorption/separation or chemical sensing.11 Consequently, it is of significant importance to judiciously functionalize the pore surface of MOFs19 with selective functional groups for enhanced gas selectivity.20 Of interest would be the separation of harmful gases, such as SO2, NH3, and H2S, which are, for example, NH···OSO···HNNH···NH3, >CO··· HN(H)H···OCNH··· OSO···HN< (cf., abstract scheme). NH3 has a maximal loading of about 4 mol(NH3) per mol(L1) in the strong binding site, which goes along with its possible assembly of several hydrogen-bonded NH3 molecules around the urea group of the L1 linker (cf., abstract scheme).

Figure 19. Solvent-accessible surface (dark brown) using a probe radius of 1.2 Å for the cavities in (a) 1, (b) 3, and (c) 4 (the three images are not drawn to scale) from a “void calculation” with the program Mercury (Mercury CSD 3.9, Program for Crystal Structure Visualization, Exploration and Analysis from the Cambridge Crystallographic Data Center, Copyright CCDC 2001−2016, http://www.ccdc. cam.ac.uk/mercury/). Note that the pore system in 3 is onedimensional, but in 1 and 4 it is two-dimensional. See section S11 with Figures S29−S31 in the Supporting Information for additional images also with the contact surface and dimensions of the pore aperture cross sections.



CONCLUSIONS The rigid ditopic 4,4′-biphenyl-dicarboxylate linker with a urea group built in as a diazepine unit (L12−) gives rise to metal− 37429

DOI: 10.1021/acsami.7b10884 ACS Appl. Mater. Interfaces 2017, 9, 37419−37434

Research Article

ACS Applied Materials & Interfaces

Table 6. Affinity Constants with Maximal Loading from (Dual-Site) Langmuir Fits of Adsorption Isotherms for Compounds 1, 3, and 4a CO2b

gas compound 1 3 4

affinity const/mbar

−1

SO2c

max load/mol mol

0.00062 0.00077 0.00194

0.9 1.3 1.2

−1d

e

affinity const /mbar

−1

0.00209 0.754 0.207

NH3c −1d,e

max load/mol mol 0.3 0.5

e

affinity const /mbar 0.115 0.062

−1

max load/mol mol−1d,e 3.7 4.2

a See section 10 in the Supporting Information for fits of adsorption isotherms using the software DynaSim V 1.1.0.5 by Quantachrome GmbH 2015. DynaSim is a tool for interpretation and evaluation of experimental data from dynamic experiment. bLangmuir model, except for 1 where dual-site Langmuir−Sips was employed. cDual-site Langmuir model, except for 1 where only a fit with the Henry model was successful. For SO2, the fit was only possible up to 300 Torr. dLoading in mol(gas)/mol(L1 linker). eAffinity constant and corresponding maximal loading for stronger binding site 1.

Because of this, only part of the reflected radiation reaches the detector, so that low relative intensities are measured at 2θ < 7°. In all diffractograms, the most intense reflection was normalized to 1. For the comparison of the experimental diffractograms of solventcontaining samples with the simulated diffractograms, one needs to be aware that the latter were derived from crystal data where a considerable amount of solvent-derived electron density in the voids had been removed by the SQUEEZE option in PLATON.35 Even if the crystal solvent in the voids is disordered, its electron density still contributes to diffraction. Hence, the experimental diffractograms on the as-synthesized samples differ in their intensities and can have additional peaks (see section 8 in the Supporting Information). Nitrogen physisorption for BET surface determination, and carbon dioxide, hydrogen, and sulfur dioxide isotherms (0−1 bar) were measured with a Quantachrome Autosorb iQ MP at 77 K for nitrogen and hydrogen and at 273 K (ice/water bath) and 293 K (active thermostatting) for carbon dioxide and sulfur dioxide, respectively. The Autosorb iQ MP is equipped with oil-free vacuum pumps (ultimate vacuum NH···OSO···HNCO···HN(H)H···OC