MOFs are organic-inorganic hybrid crystalline materials with well-defined pores of molecular dimensions.1-6 Compared with 3D MOF crystals, two-dimensional (2D) MOF nanosheets have higher flexibility and surface area, better processability, and more easily accessible active sites, which are particularly advantageous for applications requiring continuous defect-free pore structures, such as molecular sieving, separation, sensing and catalysis.7-12 Although we share a library with more than 60, 000 kinds of MOFs, only very limited number of 2D MOF nanosheets have been prepared via top-down exfoliation12-15or bottom-up synthesis.16,17 In most cases, however, the obtained MOF nanosheets have uneven thicknesses, limited lateral sizes, many defects and low production yields. It remains a great challenge to prepare uniform MOF nanosheets with a well-defined molecular thickness at large scale.
It is conceivable that layer-structured MOFs can be used to produce 2D MOF sheets by mechanical or chemical exfoliation, which is the most common and effective way for the production of 2D materials.13,18-20 The success of this strategy relies on the strong in-plane linkages coupled with weak interlaminar (out-of-plane) interaction, which ensures that the cleavage of the parent layered crystal occurs along the desired direction. Direction-specific interactions can be introduced into MOFs by rationally designed synthesis.
From the physics point of view, in the liquid exfoliation process of the layered 3D crystals, the 2D layers in the lattice are subjected to shear force (Fs) and interlaminar interaction ( ) (Fig. 1A). In the moment Fs > , the interlayer slippages and layer separation occur, which results in the formation of nanosheet. The collision momentum component parallel to the layer plus the viscous force equal to Fs, which tends to separate the layers (Fig. S1, S2; Equation S1), and the component vertical to the layers tends to tear the layers into smaller pieces. Fs can be greatly enhanced by sonication along the direction of the power source (detailed information in SI, around 1E-10 - 1E-8 N for 1 nm2 real force area). Fi equals to the summation of pairwise interactions between adjacent layers (Equation S2), which depends on the strength and number of the pairwise interaction in unit area (detailed information in SI). Its component parallel to the layer plane competes against Fs to avoid exfoliation,21 and the value was calculated as 6E-9 N for graphene in 1 nm2 contact area.
Obviously, to make the layer exfoliation possible, we need to decrease the value and increase the Fs value. Specifically, three key factors should be considered. At first, the bulk crystals should have layered structure and weak interlaminar interaction so as to achieve the key criterion Fs > . In the second, Fs value is positively correlated with the external driving force (vs) and real force area l∙h∙ (1+Δ∙w). As such, thicker layer (h) and porous structure (Δ) can effectively enhance Fs. At last, the layer should be mechanically strong so as to avoid damage caused by the vertical component of collision momentum, which can be mitigated with porous structure.
Even though only weak intermolecular interactions (Van der Waals force, π-π stacking, CH-π interaction etc.) exist between the layers, 7,10,15,22-25 the significant number of densely arranged bonding sites give rise to sufficient interaction force to lock the periodical position of the 2D layers in the crystal lattice,26 which presents the biggest hurdle for the production of monolayer nanosheets and its commercial applications.27-29 Weakening the interlaminar interaction in layered MOFs is the ultimate solution and requires a rational design to substantially reduce the bonding energy and the number of bonding sites in unit area.
Paradoxically, sufficiently weak interlayer binding may cause relative sliding between layers, leading to disordered stacking and thus the loss of structural periodicity in one-dimension (Fig. 1B); Consequently, the resulting materials do not show diffraction peaks in conventional in-house powder X-ray diffraction (XRD) characterization, and thus may be considered amorphous. Unlike the recently reported amorphous MOFs (aMOFs), an emerging function material family with great application potential,30-33 are amorphized MOFs by introduction of disorder into the parent crystalline frameworks through heating, pressure (both hydrostatic and nonhydrostatic), and ball-milling. They are amorphous in three-dimensions instead of in one-dimension and lack any long-range periodic order.
Cage-like molecules have been used as idea building blocks to construct porous 3D layered structures with weak interlaminar interaction.34-36 Bicyclocalix[2]arene[2]triazines tri-carboxylicacid (BCTA), a cage-like D3h symmetric molecule (Fig. 1C),37 can self-assemble into 3D layered structure with big pores. An interesting feature of the obtained layered structure is that, there are only very weak and scattered interaction pairs between the layers (Fig. S2). The enclosed calixarene cavities and rigid structure of BCTA brings molecular recognition properties and inherent micropores,38-40 and reduces the real contact area between layers. Taking advantage of this feature, we fabricate a layered MOF (denoted as IPM-1) using BCTA to achieve weak interlaminar interaction. We demonstrate that IPM-1 comprises highly crystalline monolayers stacked in a disordered manner, differing from both traditional “crystalline” and “amorphous” phase. Furthermore, the weak interlaminar interaction enables facile exfoliation of IPM-1 into uniform, molecularly thin 2D MOF sheets with a rather high yield by simple treatment. Our study not only provides an effective strategy for designing precursor materials of truly 2D MOFs, but also highlights an important intermediate state between traditional “crystalline” and “amorphous”, revealing that seemingly unsuccessful products may be highly useful.
IPM-1 was prepared from BCTA and MnCl2 in N,N-dimethylformamide (DMF) (Fig. S3-S6). IPM-1 easily loses 3D crystallinity in the ambient environment and its crystal structure can only be solved by low-temperature single crystal X-ray diffraction performed at -80 oC. As the single crystal structure shown in Fig. 2A, it clearly indicates that IPM-1 has a separated layered structure with a layer thickness of 1.15 nm, which is nearly double the height of BCTA (Fig. S2, S7A). In the single layer, this network is very rigid with high mechanical strength topologically, extending in the ac plane with three types of regular pores (the pore sizes are 0.8nm, 1.3nm, and 1.5nm, Fig. 2B). In the crystal lattice, there are two types of [Mn3(O2C)6] clusters, i.e., [Mn3(O2C)6]‧4H2O and [Mn3(O2C)6]‧2H2O‧2DMF, arranged into two alternatively-arranged rows in the layer plane, as shown in Fig. S7B. These two types of [Mn3(O2C)6] clusters are connected by trigeminal BCTA ligands to form a 2D network with 3,6-connected net nodes (Schläfli symbol {4^3}2{4^6;6^6;8^3}, with very high structural stability, Fig. 2C).
Along the [0, -1, 1] axis, one single IPM-1 layer looks like a square-wave, similar to black-phosphorus monolayer in some extent. The net-like layers stacked in a ABAB arrangement along the crystallographic b axis via very weak interlayer O–H···O hydrogen bonding [dO-H…O = 2.618 and 2.404 Å] between these two different [Mn3(O2C)6] clusters, whereby these hydrogen bonding are the sole interaction between the layers (Fig. 2A, S7C). This results in a 3D neutral supramolecular framework with three ellipsoid channels as illustrated in Fig. 2D, S8. There are two types of regular rectangle channels that exist between the square-wave layers for a half-phase difference (the sizes are 1.3*0.5 nm and 1.3*0.2 nm respectively, Fig. S9). All these channels vertical or parallel to the layer surface have significantly reduced the number of the weak hydrogen-bonding in unit area between adjoining layers. Meanwhile, the 3D crystal is stabilized by an interlocking structure, which is an interpenetrating network formed by the coordinated DMF molecules on [Mn3(O2C)6] clusters (Fig. 1E). In brief, the extremely low interlaminar interaction discovered in IPM-1 crystal makes it a perfectly candidate for our experiment.
Scanning electron microscopy (SEM) images have confirmed that the apparent structure of IPM-1 is highly consistent with the single crystal data, very distinct and smooth layered structure was observed(Fig. 3A, S10). Tapping-mode atomic force microscopy (AFM) and transition electron microscopy (TEM) images confirmed that IPM-1 is readily exfoliated into monolayer nanosheets in the gram-scale via simple ultrasonic exfoliation (Fig S11-S13). These monolayer nanosheets have unprecedent evenness and homogeneity. IPM-1 nanosheets can be prepared by routine ultrasonication treatment in different solvents (Fig S14). The as-prepared colloidal suspension remained stable at room temperature for several months, with significant Tyndall effect (Fig. 3A inset).
With the high concentration nanosheets suspension, AFM images show plenty of rigid and high flatness nanosheets (~ 0.1- 0.2 mg/ml, Figure 3B, S12), with a lateral size range from several to more than ten micrometres. A large-size wafer was produced on the substrate with a thickness of 1-2 nm from a 1-5 μg/ml nanosheets suspension (Fig. S13). As demonstrated by the AFM analyses on forty-three different sites (Fig. 3C, S13), more than 90% of them had a thickness of 1.1±0.2nm, which confirms that the nanosheets are presented as monolayer with high homogeneity. Fig. 3D shows an enlarged area of the dispersed nanosheet. The height profile reveals that the nanosheet is extremely flat, with an even step height of 1.1 nm that is consistent with the thickness of the monolayer in crystal structure. The infrared spectrum of the nanosheets prepared in a large-batch is in good agreement with that of the freshly-prepared IPM-1 crystal (Fig. S15), stronger and more distinct absorbance bands were observed for nanosheets in the fingerprint 1300-700 cm-1 region. Thus, it can be concluded that the chemical structure of IPM-1 remains stable during the exfoliation process. The rigid and smooth appearance of IPM-1 nanosheets identified a high mechanical strength, as ultrathin nanosheets easily curled under such conditions.
The crystallinity of IPM-1 nanosheets was directly evidenced by low-dose high-resolution TEM (HR-TEM).41,42 Lattice fringes are clearly observed in the HR-TEM image with the information transfer up to ~4 Å (Fig. 3E, S16). The fast Fourier transform (FFT) pattern of the HR-TEM image (Fig. 3G) matches reasonably well with the simulated electron diffraction pattern of IPM-1 along the [001] direction (Fig. 3F, based on the CIF of IPM-1 monolayer). The three labelled spots correspond to the 110 (1.96 nm), 010 (2.09 nm), and 100 (2.02 nm) reflections, respectively. These experimental results indicate the preservation of high crystallinity in the 2D nanosheet after the exfoliation process and the retention of the monolayer structure observed in crystal.
The bulk IPM-1 crystals appear highly “unstable” according to XRD. The powder XRD pattern of freshly prepared IPM-1 sample can be fitted to the simulated one (Fig. 4A), but IPM-1 is considered as an “amorphous” material according to XRD after various regular treatments such as heating, drying or solvent soaking (Fig. 4B). However, it should be noted that there is no change in the crystal appearance (Fig. S17). Generally speaking, it is commonly believed that there is no long-range order in an amorphous material. Very recently, evidences show that the zeolitic imidazolate framework (ZIF) glass does not even have short-range periodicity.43 However, the experimental results already confirmed the stability and 2D crystallinity of IPM-1. It is obviously a deviation from the ordinary which cannot be explained using the current X-ray crystallography.
In the regular treatments, the coordinated DMF in the IPM-1 crystal lattice dissociated, leading to the degradation of the interlocking structure. The calculated for IPM-1 is around 2E-10 N per nm2 contact area based on the single crystal structure. The thermal motion of the solvent molecules in IPM-1 channels gives rise to a weak Fs due to fluctuation, which can reach the same scale as in a small region (see SI), leading to random interlayer slippage. The crystal lattice lost the atomic periodicity in one dimension in turn. However, this slippage still is localized, as a significant number of hydrogen bonding between the layers keep the IPM-1 “crystal” stable. Under sonication exfoliation condition, the 1.15 nm layer thickness and porous structure promoted the Fs to 1E-9 - 1E-8 N per nm2 scale. The large-pore structure can effectively reduce the damage to the nanosheet. Hence, IPM-1 has successfully been exfoliated into the observed monolayer nanosheets in gram scale with large lateral size and high homogeneity.
IPM-1 is thermally stable up to 440 oC (Fig. S6), and has a BET surface area of 210 m2/g and a CO2 absorption of 41 cm3/g after activation at 150 oC under vacuum (As shown in Fig. 4C, 4D). IPM-1 has a typical type I N2 adsorption-desorption isotherm. The adsorption was apparent in the low-pressure region (P/P0 < 0.05) and the desorption of N2 was reversible with an observable hysteresis. IPM-1 nanosheet has nearly a doubled BET surface area (360 m2/g) and CO2 absorption (58 cm3/g) as compared to bulk IPM-1 (BET and CO2 absorption were performed with 130 mg IPM-1 nanosheets, as shown in inset of Fig. 4E). It is known that heterocalixaromatics can bind to CO2 via its nitrogen-rich cavities.44 Thus, the availability of the fully exposed calixarene cavities in IPM-1 nanosheet leads to this remarkable CO2 absorption observed. More importantly, while bulk IPM-1 show a broad and ambiguous pore size distribution, the pore size distribution for IPM-1 nanosheet is highly distinct (Fig. 4E), which is identical to the theoretical values as shown in Fig. 2B. This could be attributed to the removal of the irregular pores in the bulk IPM-1 as a result of interlayer slippage after complete separation, and no BET hysteresis was observed in the N2 desorption isotherm of the IPM-1 nanosheet (Fig. 4C).
High pressure BET test results show that the blocked channels and calixarene cavities in IPM-1 could be opened at high pressure. At 273 K, when the pressure reached 60 atm (maximum pressure of 200 atm), N2 absorption of IPM-1 increased from 2.8 cm3/g to 7.0 cm3/g (the interaction between N2 molecules and the frameworks is very weak). Meanwhile, the CO2 absorption (performed with the same batch sample) increased from 18 cm3/g to 56 cm3/g at 38 atm (Fig. 4E). Both desorption revealed hysteresis, but they were reversible at reduced pressure, which confirmed the high structural stability of IPM-1 layers. 45,46
It can be concluded from the above experimental results: IPM-1 exhibits a good physical and chemical stability, and it is 1D-amouphous due to the localized interlayer slippage, which presents an intermediate phase between the classic crystalline and amorphous phase. The interlayer slippage leads to the blockage of the channels and micropores instead of structure collapsing. After exfoliation, IPM-1 nanosheets possess high macroscopic homogeneity and restored microporosity. In addition, the stacking fault caused by the interlayer slippage in IPM-1 is variable at the angstrom scale. More interestingly, neither aggregation of nanosheets nor crystalline phase recovery has been detected with the as-prepared nanosheets. This cheap, easy, and scalable preparation of IPM-1 monolayer nanosheets can further render its commercialization highly feasible.
Why IPM-1 has been determined as “amorphous” by in-house XRD characterization? As we know, the molecules are in chaotic thermal motion within the confined space of the crystal lattice, which leads to the loss in the periodicity. If the initial distance between two adjacent molecules is set as ε(0), after time t, the distance can be described as ( , the Lyapunov exponent, Fig. 1A, S18).47,48 For an effective diffraction crystal plane with n atoms, the detection error will be amplified by times. In an ideal close packed crystal, the thermal motion is confined in picometers scale, but the wavelength of the X-ray generally used in XRD is around 1 Å. Therefore, the influence of this kind of thermal motion to XRD observation is limited. In addition, the lattice wave makes the atoms correlated in long range, , λ→0, the error in the diffraction crystal planes can be omitted.
In IPM-1, experimental results indicate that the movement of the atoms along the layer plane can increase to angstrom scale due to the interlayer slippage. Given a 2 Å ε(max), the λIPM-1 can reach 0.1 in the crystal lattice (λ value is decided by ). The corresponding detection error will increase times, resulting in an extremely poor S/N ratio.
In this study, a 1D-amouphous MOF, IPM-1, is designed and prepared. Experimental results verified its in-plane 2D periodic structure and misaligned arrangement in the last dimension, which is similar to the quasicrystal in some extent,49 as they are both partially periodical in atomic arrangement. This intermediate state can be attributed to the extremely low interlaminar interaction and the resulting localized interlayer slippage. The exfoliated IPM-1 nanosheet presents the first mass-producible 2D monolayer MOF nanosheet, and exhibits high uptake capacity and selectivity for CO2 adsorption. In addition, the “amorphous” status observed in IPM-1 may be vital towards the successful preparation of monolayer 2D material.