Solvent Templating and Structural Dynamics of Fluorinated 2D Cu-Carboxylate MOFs Derived from the Diffusion-Controlled Process

The layered 2D MOFs, owing to their enhanced flexibility and tunability, have recently emerged as a promising alternative to the 3D microporous MOFs in the quest for novel responsive functional materials. However, maintaining the simultaneous control over self-assembly of molecular building blocks as well as ordered stacking of MOF layers poses a significant synthetic challenge. We report on the controlled 2D MOF formation based on a case study of solvent-templated growth of a series of 2D Cu(II)–carboxylate MOFs varying in stacking modes and distances using a diffusion-controlled MOF deposition approach in various solvent mixtures. Moreover, we demonstrate the structural dynamics of the developed 2D MOFs involving both in-plane and out-of-plane movements of the individual 2D layers triggered by solvent exchange, which allowed for selective postsynthetic transformations between the developed 2D MOFs. We also investigated the gas adsorption properties of the developed MOFs, which demonstrates a remarkable crystal size effect on the N2 adsorption capacity using a model 2D MOF system.


■ INTRODUCTION
Metal−organic frameworks (MOFs) composed of organic linkers and inorganic nodes have been attracting the immense attention of researchers due to their diverse and strikingly robust architectures combined with high surface areas and great chemical diversity. As a result, MOFs have found numerous applications in gas storage and separation, catalysis, sensors, energy storage, etc. 1−6 One of the strategies for enhancing the functionality of MOFs involves the use of fluorinated ligands, which greatly influence the material properties such as hydrophobicity, chemical stability, gas adsorption, and separation capabilities, due to specific chemical characteristics of the fluorine-containing moieties. 7−11 Moreover, rational introduction of highly electronegative fluorine atoms might allow for a fine-tuning of the electronic properties of MOFs such as luminescence or conductivity. 12 However, to fully take advantage of the structure-related properties of MOFs, their formation must be adequately controlled in order to achieve hybrid materials of desired properties. 13 One of the ongoing challenges in the rational synthesis of MOFs is related to the MOFs isomerism, i.e., the ability to adopt multiple structural arrangements of the same building blocks leading to a number of materials with different properties. 14,15 An efficient and simple solution of this issue is the use of various template molecules for the controlled MOF growth, which has already been demonstrated as a powerful approach for synthesis and formulation of desired MOFs. 16,17 In particular, the use of inherently present solvent molecules as templates for MOFs self-assembly appears as a very promising, but it is an often underdeveloped strategy. In fact, replacing one solvent for another in a typical solvothermal MOF synthesis procedure usually requires adjustment of other variables (temperature, concentrations, additives, etc.) for efficient MOF crystallization, which hinders the inference on solvent-templating in MOF synthesis. 18,19 Over the last two decades, the continuously developed methods of MOFs synthesis and crystal engineering have led to the discovery of countless of new MOF structures. 20 This immense effort has been mostly driven by the applicationoriented approach utilizing MOF rigid networks and welldefined porosity for practical aspects. However, some applications require substantial structural tunability and responsiveness of the applied porous materials, which can be achieved by rational synthetic design of the MOF networks using flexible or reactive linkers and nodes or by a postsynthetic modification strategy introducing the desired functionality to the prepared frameworks. 21−23 Within the pursuit for responsive MOFs, 2D-layered materials have emerged as a promising alternative to the 3D frameworks due to their enhanced network flexibility and ability to adopt various stacking modes and distances of 2D MOF sheets. 24,25 This structural isomerism is determined by cooperative supramolecular interactions between individual layers as well as potential guest molecules, which can be controlled by the use of templates influencing the molecular and supramolecular MOFs self-assembly processes. 18,19 Interestingly, the stacking geometry of individual layers in 2D MOFs can also be adjusted dynamically by introduction or removal of various guest molecules within the premade MOF structure. As a result, the 2D layers might perform motion either in-plane ("sliding" movement, Figure 1a) or out-of-plane (Figure 1b), which leads to significant structural changes, e.g., opening/closing the 1D pores perpendicular to the 2D sheets 26−28 or changing the stacking distance between the individual layers. 29,30 The structural dynamics of 2D MOFs enables their promising gas adsorption properties related to the "gate opening" mechanism, which is advantageous compared to the typical gas storage in rigid 3D structures due to the high working capacity. 31,32 Moreover, 2D MOF materials are used as precursors for ultrathin MOF nanosheets, which find many applications due to their exceptionally high surface/volume atom ratios and large surface areas. 24, 33−36 One of the prevalent synthons in 2D MOFs are M 2 (μ-O 2 CR) 4 , paddlewheel units adopting square coordination geometry, which form flat or wavy MOF layers upon selfassembly with various bicarboxylates. 37 The Cu(II)-based 2D MOFs are the most extensively studied systems in that group, which is related to their relatively high stability and straightforward solvothermal synthetic approach. 37 Known structures of Cu(II)-based 2D MOFs involve vast arrays of bicarboxylate linkers, including the aforementioned fluorinated ligand systems. For example, in the presence of Cu(II) precursors and the 4,4′-(hexafluoroisopropylidene)bisbenzoate (hfipbb) linkers, 2D MOFs are formed, which could be described using the formula [Cu(hfipbb)·Solv] (where Solv = various solvent molecules coordinated to the Cu centers). The literature data includes four examples of Cu(hfipbb) structures, all prepared in solvothermal conditions, which differ in internal structure of the networks, interpenetration, as well as layer stacking distance and geometry, which results from varying preparation conditions ( Figure 2). 38−41 Similar variations have also been observed for other 2D MOF systems, which indicates that their supramolecular structure can be influenced by various subtle effects like solvent type, temperature, pH, additives, etc. 14,15 Stemming from our interest in development of efficient preparation of hybrid inorganic−organic materials, 30,43−48 herein we report on the solvent templating effect on the selfassembly of Cu 2 (μ-O 2 CR) 4 paddlewheel units and 4,4′-(hexafluoroisopropylidene)bisbenzoate (hfipbb) linkers yielding a family of 2D MOFs (collectively identified as a Cu(hfipbb) family). In contrast to the previous reports, 38 −41 we applied the diffusion-controlled MOF synthesis methodology, involving the MOF formation in a very low precursor concentration regime. 30 Using various solvent mixtures, we isolated and structurally characterized five members of the Cu(hfipbb) family, which differ in both stacking geometry (I-x vs I-xy) and distance between the 2D layers ( Figure S1). We also demonstrate the structural dynamics of the developed MOFs involving both in-plane and out-of-plane movements of the 2D layers, triggered by exchange or removal of the guest solvent molecules ( Figure 3). Remarkably, the revealed transitions between various members of the Cu(hfipbb) family are selectively allowed or disallowed based on the type of the substrate and guest molecules.

■ EXPERIMENTAL SECTION
Materials and Methods. All chemicals were purchased from commercial sources and used without further purification. Potassium 4,4′-(hexa-fluoroisopropylidene)bisbenzoate solution was prepared by mixing appropriate amounts of KOH and 4,4′-(hexa-fluoroisopropylidene)bisbenzoic acid in water until all of the solids were dissolved and the pH of the solution was close to neutral. PXRD data were collected on the Empyrean diffractometer (PANalytical) employed with Ni-filtered Cu Kα radiation (40 kV, 40 mA) using Bragg− Brentano geometry and a Si zero-background holder. VTPXRD data were collected using the above instrument equipped with the Anton Paar TTK-450 sample chamber, which was evacuated (10 −2 mbar) during the measurements. IR spectra were recorded using the FTIR Bruker-Tensor II system. Gas sorption studies were undertaken using a Micromeritics Instrument Corporation (Norcross, Georgia, USA) ASAP 2020 system. Approximately 100−150 mg of the corresponding solid product was transferred to a preweighed sample tube and evacuated under vacuum at 100°C on the gas adsorption apparatus until the outgas rate was <5 μmHg. All gases used were of 99.999% purity. Helium was used for the free space determination after sorption analysis. Adsorption isotherms of N 2 and H 2 were measured at 77 K in a liquid nitrogen bath. Adsorption isotherms of CO 2 were measured at 273 and 298 K in methanol bath controlled by circulating refrigerated medium. Elemental analyses were carried out with an UNICUBE Elementar Analyzer (GmbH). SEM imaging was performed using FEI Nova NanoSEM 450 system equipped with a field emission electron gun operating at 2 kV. The optical microscope images were collected using the Olympus BX53 microscope.
Synthesis of [Cu(hfipbb)·H 2 O] I-x (1). Solutions of copper(II) nitrate trihydrate (241.6 mg, 1.0 mmol, in 2 mL of H 2 O) and potassium 4,4′-(hexa-fluoroisopropylidene)bisbenzoate (1 mmol, in 2 mL of H 2 O) were prepared. The solutions were placed in a special glass reactor that prevented mixing of the substrates ( Figure S2). Next, the reactor was gently filled with water to create a diffusion path  Figure 2. Overview of the 2D MOFs formed by Cu 2+ and hfipbb linkers including data from the literature (solvothermal synthesis, red arrows) and this work (diffusion-controlled synthesis, blue arrows). Geometry of 2D layers stacking was named in accordance to the literature data. 42 3-mepy = 3-methylpyridine.
Inorganic Chemistry pubs.acs.org/IC Article between substrates. The diffusion controlled process was conducted for 4 weeks at room temperature which resulted in formation of turquoise flake-shaped single crystals of 1. The product was collected and washed with water and THF and dried in vacuo. PXRD analysis confirmed the phase purity of the product. Yield: 370 mg (78%  51. Synthesis of the microcrystalline powder of 2 was performed in the following way: solutions of copper(II) nitrate trihydrate (241.6 mg, 1.0 mmol, in 2 mL of H 2 O/MeOH mixture (1:1 by volume)) and potassium 4,4′-(hexa-fluoroisopropylidene)bisbenzoate (1 mmol, in 2 mL of H 2 O/MeOH mixture (1:1 by volume)) were mixed at RT, which resulted in an immediate formation of a turquoise precipitate. The reaction mixture was centrifuged, and the liquid phase was removed. The resulting solid was washed with THF and dried in vacuo. PXRD analysis confirmed the phase purity of the product. Yield: 460 mg (95% X-ray Structure Determination. The crystals of all complexes were selected under Paratone-N oil, mounted on the nylon loops, and positioned in the cold stream on the diffractometer. The X-ray data for materials 1, 2, 3, 4, and 5 were collected at 100(2) K on a SuperNova Agilent diffractometer using Mo Kα radiation (λ = 0.71073 Å). 49 The structures were solved by direct methods and refined using SHELXL-2018/3. 50 All non-hydrogen atoms were refined with anisotropic displacement parameters. Hydrogen atoms attached to carbon atoms were added to the structure model at geometrically idealized coordinates and refined as riding atoms. Crystal

■ RESULTS AND DISCUSSION
Synthesis and Structures of the Cu 2D MOFs. The supramolecular architectures of 2D MOFs can be affected by various subtle effects as mentioned above. In this regard, we wondered how the diffusion-controlled MOF synthesis methodology will direct the crystal structures of 2D MOFs. To this aim we selected 2D Cu-MOFs based on a paddlewheel Cu 2 (OOR) 4 cluster and a fluorinated bicarboxylate linker as a model system. In control experiments, we used Cu(NO 3 ) 2 and K 2 hfipbb salts as precursors in the diffusion-controlled synthesis in water and various water/alcohol (MeOH, EtOH, iPrOH, and tBuOH) mixtures. The reaction conducted in water yielded [Cu(hfipbb)·H 2 O] I-x (1; the lower index indicates the 2D layers stacking type 42 ) (Figure 3), which has been previously reported as the result of solvothermal synthesis in water at 160°C. 38 (Figure 3), all isolated in a form of turquoise flake-shaped single crystals. The resulting materials were characterized by elemental analysis, PXRD and single-crystal X-ray diffraction (SCXRD), FTIR spectroscopy (for details see Figures S3−S7), and a gas adsorption study involving N 2 , H 2 , and CO 2 as adsorbates.
SCXRD analysis of the developed 2D MOFs revealed a common distinctive structural feature, namely, the presence of topologically equivalent 2D MOF layers based on [Cu 2 (OOR) 4 ] paddlewheel-type nodes connected by V-shaped hfipbb linkers. All of the observed 2D layers were composed of two independent interpenetrated networks and included 1D open channels parallel to the layer surface ( Figure 4). However, the materials prepared using various solvents significantly differed in the supramolecular arrangement of the 2D layers exhibiting two stacking types (I-x and I-xy) as well as various stacking distances, which indicates a strong solvent templating effect on the supramolecular structure of the studied materials. Remarkably, these various stacking arrangements were not related to the presence of any covalent or hydrogen bonding between the individual MOF sheets but rather to the weak interactions influenced by the presence of various solvent guest molecules either in the 1D channels or coordinated to the axial positions of the [Cu 2 (OOR) 4 ] paddlewheels. Notably, PXRD analysis of the studied 2D MOFs (1−5) revealed strong preferred orientation of crystals along the direction of 2D layers ( Figure S8). This observation clearly indicates that the flake shape of the crystals is a result of significantly faster crystal growth in that direction. Albeit this is not unexpected for 2D-structured materials, in the case of the Cu(hfipbb) family, it might kinetically limit processes involving transportation of molecules through the 1D channels, which are parallel to the MOF layers ( Figure 4).
As evidenced by the SCXRD study, the crystal structure of 1 is composed of [Cu 2 (OOR) 4 ]·2H 2 O clusters connected by dianionic hfipbb linkers ( Figure S1a), which form a 2D MOF with the interlayer distance of 11.02 Å. The axial positions of the [Cu 2 (OOR) 4 ] clusters are occupied by water molecules (Figure 5a), which can be removed by heating the material in vacuum at 100°C for 3 h, resulting in a color change to sapphire-like. Detailed investigations of structural changes upon thermal treatment of 1 were conducted using the in situ variable temperature X-ray powder diffraction in vacuum (VTPXRD, Figure S9), which revealed a phase transition at ca. 125°C, similar to that previously reported for analogous Cr(II)-based material. 30 The crystal structure of 2 differs from  Inorganic Chemistry pubs.acs.org/IC Article that observed for 1 in both stacking distance (12.80 Å) and stacking type (I-xy) of the MOF layers, as revealed by the SCXRD analysis ( Figure S1b). This structural difference is related to the MeOH molecules coordinated to the axial positions of the [Cu 2 (OOR) 4 ] units in 2 (Figure 5b), which influences the supramolecular arrangement of the MOF layers. Thermal treatment of 2 also led to desolvated sapphire-colored material, which was observed as a phase transition at ca. 75°C in the VTXPRD study ( Figure S10). Surprisingly, the product of 2 desolvation exhibited a strikingly similar PXRD pattern to 1, which suggests that the phase transition involves a change in the 2D layer stacking geometry. Moreover, unlike any other 2D Cu-MOFs described here, 2 can be easily prepared by direct precipitation at room temperature by mixing of the solutions of Cu(NO 3 ) 2 and K 2 hfipbb in H 2 O/MeOH, yielding a fine blue powder of the desired 2D MOF material (2-P) containing microcrystals ca. 10 μm in size, as confirmed using PXRD analysis ( Figure S11) and SEM imaging ( Figure S12). According to the SCXRD analysis, the crystal structures of 3 and 4 are essentially similar to that observed for 2 but differ in the type of molecule coordinated to the Cu centers (Figures 5b and S1c,d) as well as in the interlayer stacking distances (12.78 and 13.52 Å for 3 and 4, respectively). The VTPXRD study revealed that 3 undergoes a slight phase transition at ca. 150°C , which does not affect the layer stacking distance ( Figure  S13), whereas 4 did not exhibit any changes up to 200°C ( Figure S14). This observation suggests limited mobility of iPrOH molecules, as compared to smaller EtOH molecules, in the relatively narrow channels of the investigated MOF. Finally, the SCXRD study of 5 revealed that its supramolecular structure consists of I-xy stacked 2D layers with the interlayer stacking distance of 14.81 ( Figure S1e). In contrary to the products 2, 3, and 4 prepared using MeOH, EtOH, and iPrOH, the Cu coordination sites in 5 were saturated by water molecules, while tBuOH molecules were only present inside pores, interacting by hydrogen bonding (Figure 5c). Remarkably, the presence of tBuOH molecules templated the MOF growth into formation of a I-xy stacked structure as opposed to the I-x stacking in the product formed in pure H 2 O as solvent. However, as revealed by the PXRD study, 5 was unstable outside of the mother liquor and spontaneously decomposed into a more closely packed structure with interlayer distance of 13.71 Å after ca. 48 h of storage at RT. The same phase transition of 5 was observed in the VTPXRD study in vacuum already at 50°C ( Figure S15).
Structural Dynamics Study of the Developed 2D MOFs. As demonstrated above, the supramolecular arrangement of the 2D MOF layers in the Cu(hfipbb) family is heavily dependent on the type of ligands incorporated in the [Cu 2 (OOR) 4 ] units. Taking that into consideration, we wondered if both the stacking distance and the stacking mode of the studied 2D MOFs could be postsynthetically modified by a phase transition triggered by solvent exchange (Figure 6). Therefore, we have designed a series of experiments in which 1, 2, 3, and 4 (5 was excluded due to low stability) were soaked in various solvents (H 2 O, MeOH, EtOH, or iPrOH) for 6 h, followed by PXRD analysis in order to find whether the studied members of the Cu(hfipbb) family can be interchangeably transformed into one another. Remarkably, the results demonstrated significant selectivity of the studied postsynthetic processes depending on the type of substrate and solvent used ( Table 1, Figures S16−19). As seen from the results of this study, the transformation from 1 to 2 and from 2 to 1 could be easily conducted by application of the appropriate solvent (MeOH and H 2 O, respectively). In a similar fashion, both 3 and 4 could be converted into 1 by soaking with H 2 O, but the reverse processes from 1 to 3 and from 1 to 4 were not effective. Furthermore, both soaking 2 with EtOH and soaking 3 with MeOH yielded mixed phases. Finally, 4 proved to be unreactive toward MeOH and EtOH, and any attempts to obtain 4 by postsynthetic solvent exchange were unsuccessful. Noteworthy, as mentioned above, heating of 4 in vacuum did not lead to any structural change, while soaking 4 in water easily yielded 1, which suggests that water molecules facilitate mobility of iPrOH through the narrow 1D channels in the material. The demonstrated selective transformations involving both in-plane and out-of-plane movements of MOF layers clearly indicate that the key factors influencing the network dynamics are related to the type of guest solvent molecules.
Gas Adsorption Studies. Structural analyses of the developed 2D MOFs revealed the presence of 1D channels with relatively narrow openings of 2.99, 3.12, 2.79, 2.97, and 3.16 Å (determined using the Poreblazer software) 51 for 1, 2, 3, 4, and 5, respectively. In order to evaluate the gas adsorption properties of the studied MOFs we initially conducted isothermal N 2 adsorption studies at 77 K involving 1, 2, 3, and 4 (5 was excluded due to low stability) after the thermal activation conducted in vacuum, which revealed low capacities of all of the investigated materials ( Figure S20). This   Figure S21) were significantly greater than those observed for N 2 in case of every tested material, which confirms our view. Further studies involving materials 1−4 revealed significant CO 2 adsorption capacities at 298 K ( Figure  7) and 273 K ( Figure S22). The CO 2 adsorption values of 1, 2, and 3 were similar and varied in the range from 20 to 27 cm 3 /g at 850 mmHg; sample 4 exhibited a lower capacity of ca. 11 cm 3 /g at 850 mmHg, which is in accordance with the observations mentioned above of limited ability to remove iPrOH molecules by thermal and vacuum treatment. The ability of the 1−4 to adsorb CO 2 , with a kinetic diameter (3.30 Å) very similar to the pore openings, indicates that the materials are likely prone to a temperature triggered gateopening effect. 47,52 The above absorption studies were conducted using samples in the form of flake-shaped crystals; thus, we wondered if the material formulation could affect the gas adsorption capabilities of the studied materials. To this aim, we selected material 2, which conveniently could be obtained both in the form of highly anisotropic single crystals (2, prepared by diffusion-controlled deposition) and fine microcrystalline powder (2-P, prepared by direct precipitation, crystal size ca. 10 μm, Figures S12 and S23). Remarkably, the H 2 adsorption properties of 2-P and 2 at 77 K were dramatically different (Figure 8a), which clearly indicates that crystals of a smaller size could easily be penetrated by gas molecules. This observation of material formulation affecting its gas adsorption capabilities is likely due to the kinetic effects related to a limited number of pore entrances in the bigger flake-like crystals as well as limited mobility of the gas molecules in the narrow 1D channels. This limiting effect was observed to be alleviated at 273 and 298 K as 2 and 2-P revealed very similar CO 2 adsorption capacities at these temperatures (Figure 8b), which suggests increased mobility through the 1D channels, probably related to the framework flexibility at higher temperatures. In order to further test the hypothesis of absorption capabilities of 2 being related to the crystal size we investigated H 2 adsorption properties of a ground sample of 2 (marked as 2-G). To this aim, single crystals of 2 were manually ground for 10 min in an agate mortar, soaked in MeOH for 24 h, and dried, resulting in a fine powder of 2-G.
The PXRD pattern of 2-G perfectly matched that of 2-P ( Figure S11), while the estimated crystal sizes were 50−500 μm for 2, 5−40 μm for 2-G, and ca. 10 μm for 2-P ( Figures  S12 and S23). Conducting the H 2 adsorption experiment at 77 K using sample 2-G resulted in an adsorption isotherm very similar to the one observed for 2-P (Figure 8a), which additionally confirms that in the case of the studied 2D MOF family, the crystal size has a significant impact on gas adsorption properties due to the kinetic effects corresponding to the number of pore entrances on the sides of the flakeshaped crystals.

■ CONCLUSIONS
In conclusion, we demonstrated that the diffusion-controlled preparation of 2D MOFs is a very efficient way for controlling their supramolecular structures using solvent-templating effect. Using this methodology we obtained the Cu(hfipbb) family of 2D MOFs which, depending on the solvent used, differed in stacking distance and stacking geometry of the layers. We investigated the gas adsorption properties of the developed 2D MOFs, which revealed relatively low N 2 and moderate CO 2 adsorption capacities. Remarkably, we observed a strong effect of crystal size on the N 2 adsorption capacity using a model 2D MOF system, which is most likely related to the limited number of pore entrances in the bigger flake-shaped crystals.
Moreover, we demonstrated the structural dynamics involving both in-plane and out-of-plane movement of 2D layers in the studied 2D MOF family triggered by the solvent exchange under mild conditions. Remarkably, the selectivity and yields of the investigated transformations depended on the type of introduced and replaced solvent molecules, which indicates the role of product stability as well as solvent permeability in controlling of these transformations. Further developments of this approach could lead to new procedures of  Overall, we demonstrated a comprehensive case study of the family of Cu(II) carboxylate-based 2D MOFs involving the diffusion-based synthetic strategy, allowing for precise analysis of the role of solvent type, as an isolated variable, which might be a valuable tool for more in-depth understanding of MOFs formation in general.