2.1. Crystalline assemblies with double hierarchical architecture DFA-TPY
The preparation of DFA-TPY (DFA-TPY: Double Fractal Architecture-Tripyridine) is starting from precursor TPY (Fig. 2). The choice of solvent, concentration, and control of the evaporation rate at different stages is fundamental for the obtention of fractal superstructures. Initially, crystalline assemblies of FA-TPY (FA-TPY: Fractal Architecture-Tripyridine) with millimetre size and fractal-like morphology were obtained from a solution in toluene of TPY precursor (> 1.2mg/mL) and low evaporation rate of toluene (> 3 days). Then, FA-TPY was immersed in a solution of TPY in toluene (> 0.8mg/mL) and the solvent was left to evaporate at a fast rate (~ 20 hours) to obtain the final assembly, DFA-TPY. The relatively fast nucleation of the newly generated smaller crystals on the surface of the main FA-TPY crystalline assembly yields a dense array of crystals. Scanning electron microscopy (SEM) confirmed that the secondary generation of crystals, growing at the surface of the larger FA-TPY assemblies, have average diameters of less than 1 micrometre (Fig. 2). Thus, the final crystalline assembly DFA-TPY has multiscale morphology and double hierarchical architecture. During the crystallization process and after thermal activation, the crystalline integrity of DFA-TPY is maintained, as confirmed by powder X-ray diffraction (PXRD) (Fig. S4) and matches with the phase of a recently reported microporous HOF [21]. The thermal stability of DFA-TPY building block was also studied by thermogravimetric analysis (TGA) under N2 (Fig. S5). An initial and gradual 5% weight loss below 300 oC was attributed to the loss of toluene within the packing. A second weight loss of > 70% between 400 oC and 500 oC was attributed to the thermal decomposition of TPY. This confirms that the DFA-TPY precursor is thermally stable below 400 oC.
2.2. Crystalline assemblies with hierarchical porosity-architecture, P2-TPY@P2-TPY
Initially, P2-TPY (P2-TPY: micro-macro Porosity-tripyridine) hollow crystals were also prepared from precursor TPY but at low concentration (< 0.6 mg/mL) and low evaporation rate of toluene (> 4 days). SEM images of the crystals revealed macropores within the main hollow interior (Fig. 3b). The crystal structure of P2-TPY was solved by SCXRD analysis (Table S1). In the crystal packing, there are six TPY molecules and nine toluene molecules in the asymmetric unit. The packing is represented by the TPY molecules arranged in pairs and all six unique molecules form an infinite π∙∙∙π stacked column parallel to b (Fig. 3c). The first molecule is adjacent to a symmetry equivalent of the sixth molecule. The closest contacts between atoms of central rings are significantly shorter than usual π∙∙∙π stacking distances, being in the range 3.26–3.35 Å. The spread of the centroid-centroid distances and pattern of closest contacts supports the monoclinic space group selection due to the pattern of variation. If the TPY molecules are regarded as arranged in layers in the a/c plane, then the toluene molecules are in the space between every other layer in the spaces between stacks of molecules. There is no additional void volume or disorder. With the toluene molecules removed, there are three unique ‘voids’, so six in total in the unit cell. Each equates to about 5% of the cell volume. The void volumes for each individual void are: 538, 545, and 530 Å3. Overall ca. 30% of the cell volume is occupied by toluene molecules, thus defining the presence of micropores where the longest diagonal distance within each pore is close to 1.6 nm.
Next, crystals of P2-TPY with micrometre size, having initial hierarchical micro and macroporosity, were immersed in a solution of TPY precursor in toluene (0.6mg/mL) and the solvent was left to evaporate at fast rate (~ 20 minutes). The relatively fast nucleation of the newly generated smaller crystals of P2-TPY on the surface of the main hollow P2-TPY crystals yields a dense array of nanocrystals with average diameters of 100–200 nm, covering most of the surface of the main hollow crystal, including the inner main hollow structure, thus forming structures with double hierarchy of architecture with porosity. SEM confirmed that the secondary generation of P2-TPY nanocrystals growing at the surface of larger crystals have rod morphology (Fig. 3b) forming a multiscale morphology that resembles a nanoforest. This strategy is also versatile in terms of scale. For example, by modifying the size of the seed crystals of P2-TPY, from micrometre to millimetre size, a secondary generation of micrometre size crystals can be grown, which can be easily visualized by optical microscopy (Fig. S6). During the crystallization process for the formation of P2-TPY@P2-TPY with hierarchical porosity and hierarchical architecture, the crystalline integrity of P2-TPY is maintained, as confirmed by powder X-ray diffraction (PXRD) (Fig. S7). However, during the activation of P2-TPY@P2-TPY at 110°C, single-crystal to single-crystal transformation occurred to a new stable crystalline microporous phase, matching with the DFA-TPY phase (Fig. 3d). The thermal stability of P2-TPY@P2-TPY was studied by TGA, confirming similar stability to DFA-TPY (Fig. S8).
2.3. Crystalline assemblies with hierarchical architectures-composition, P2-TPY@Hybrid-1
Crystals of P2-TPY@Hybrid-1 were obtained from dipping P2-TPY crystals in a solution of 3COOH (1.3mg/mL) in pure ethanol and left to evaporate at slow evaporation rate (Fig. 4). Initial examination of the millimetre size P2-TPY crystals by optical microscopy after 2 days of crystallization indicated the presence of new micrometre size crystals growing on the surface of P2-TPY, including its hollow interior. The resulting surface of P2-TPY@Hybrid-1 with the new crystals were initially structurally characterized by Energy-dispersive X-ray (EDX) analysis for C, N and O elements (Fig. S9), indicating a homogeneous distribution of C, O and N. This result suggests that initially, a new crystalline phase containing similar amounts of N and O, from both ligand precursors, is growing on the surface of the P2-TPY crystals. PXRD also corroborated this hypothesis, confirming the presence of 2 different crystalline phases, the phase that corresponds to P2-TPY and a new co-crystalline phase in Hybrid-1 where both 3COOH and TPY could be present (Fig. S10 and S11). To explain the presence of TPY molecules in the crystals of Hybrid-1, P2-TPY crystals were submerged in an ethanol solution for 3 weeks and UV-Vis analysis of the resulting solution was done after filtration. UV-Vis analysis confirmed the presence of a residual amount of TPY in solution, which indicates a partial solubilization of P2-TPY at the surface of the crystal in Ethanol (Fig. S12). Therefore, we conclude that TPY at the surface of the crystal is transferred to solution during the crystallization, and combined with 3COOH also in solution, co-crystallizing as a new crystal phase on P2-TPY. A similar mechanism of dissolution-recrystallisation was recently observed in MOFs [22]. To study the crystal packing of the newly formed crystals Hybrid-1 and understand the interactions between both precursors, we attempted to crystallize 3COOH and TPY simultaneously by combining a solution of 3COOH in ethanol and TPY in toluene which was left for several days with slow evaporations. Clear long crystals from the mixture were obtained and characterized by SCXRD (Table S2). The crystals of Hybrid-1 revealed an asymmetric unit consisting of seven pairs of 3COOH molecules forming hydrogen bonds to TPY in a 1:1 ratio. H-bonding interactions are observed between all three arms of TPY and 3COOH through the N pyridyl acceptor and OH carboxyl donor atoms (2.49 Å average) (Fig. 4c). Further stabilizing interactions consist of π-π stacking interactions between the central triazine ring, middle phenyl rings and outer pyridyl rings with a step of around 3.34 Å between the planes. The arrangement consists of seven stacked molecules, either a stacked set of CCT-TCTC (C:3COOH and T:TPY) molecules (with a central 60° rotation marked with a hash) or stack of CTCT-TTC in a herringbone arrangement with each group rotated 60° to each other (Fig. S2). The planarity of the stacked and paired 3COOH and TPY molecules is accommodated with an increasing twist angle, ending up to 36° between the middle C6H4 rings and the outer pyridyl rings for the TPY when moving up along the stacking axes, ending at 37° for the 3COOH phenyl rings and the outer carboxyl group. This interlaced, stacking arrangement of the 3COOH and TPY pairs accommodates a single open channel, along the c axis (Fig. 4c, S2b).
The channel was analysed using Solvent Channel Mapping [23] to reveal infinite pores running along the c-axis. There are four channels, two with larger diameters and two smaller diameters, with the larger pore mapped to accommodate a probe molecule of 2.6 Å radius. Overall ca. 9.8% of the cell volume is occupied by solvent molecules using a probe radius of 1.2 Å, thus defining the presence of pores with maximum dimensions of 1.3×1.1 nm. FTIR analysis of P2-TPY@Hybrid-1 crystals (Fig. S13 and S14) confirmed the presence of COOH group for the band centred at 1686 cm-1, characteristic of the stretching band of C = O groups. Compared to the FTIR spectra for pure 3COOH crystallised in Ethanol, and previously reported [24], there was a clear shifted for the band of the carbonyl group of Δδ ~ 10cm-1, indicating a change in the chemical environment for the carbonyl group, probably due to the presence of CHN….COOH hydrogen bond interactions in the crystal and the interface. Additional shifts are also visible in the bands attributed to P2-TPY in the regions of 3033 − 3021 cm-1and 2653 − 2528 cm-1 for 3COOH. To understand the driving mechanism of growth of Hybrid-1 on P2-TPY, we compared the unit cells of both crystals, Hybrid-1 and P2-TPY, showing very different cell parameters, (a = 34.4 Å, b = 25.1 Å, c = 55.7 Å, b = 90.1° vs a = 17.4 Å, b = 40.0 Å, c = 17.4 Å, b = 119.9°), indicating lattice mismatch. We then calculated the contribution of each interaction: H-bonding and π–π stacking, for the stacked dimers of the co-crystal Hybrid-1 (Fig. S15-16). The DFT calculations were performed using version 5.0.3 of the ORCA program [25]. The range-separated hybrid functional, wB97X-V, was used to approximate the exchange–correlation functional. The polarized triple-zeta basis set, def2-TZVP [26], as well as the general auxiliary basis, Def2/J [27], were used in the calculations. The values in energy for both contributions, H-bonding and π–π stacking, accounts for 12 kcal/mol and 27 kcal/mol, respectively. The obtained energy value for the H-bond interaction on the co-crystals of Hybrid-1 for COOH….N is higher than the H-bond value for N….N in P2-TPY (12 kcal/mol vs 5.4 kcal/mol, respectively [21]), which would suggest that the driven mechanism of growth of Hybrid-1 on the surface of P2-TPY would be the stronger H-bond interactions originated at the interface between TPY and 3COOH crystals. The thermal stability of P2-TPY@Hybrid-1 building block was also studied by thermogravimetric analysis (TGA) under N2 (Fig. S17). An initial 13% weight loss below 150 oC was attributed to the loss of solvent trapped within the packing. A second weight loss of > 60% below 500 oC was attributed to the thermal decomposition of both precursors. The porosity of P2-TPY@Hybrid-1 was also studied by gas adsorption isotherms after sample activation (Fig. S18). The CO2 adsorption isotherm at 223 K of P2-TPY@Hybrid-1 shows an uptake capacity of 0.35 mmol/g, smaller than that of the original P2-TPY [21] and it could be attributed to the partial collapse of the framework during activation. This partial amorphization after activation of P2-TPY@Hybrid-1 was also corroborated by PXRD analysis, clearly showing a change in the phase for Hybrid-1 (Fig. S11).
2.4. Crystalline assemblies with hierarchical architectures-composition, HOF-3COOH@P2-TPY
First, crystals of HOF-3COOH were obtained from slow crystallization (~ 2 weeks) of a Methanol solution of 3COOH (1.4mg/mL) precursor at room temperature (Fig. 5). SCXRD was used to determine the crystal packing of HOF-3COOH (Table S3). Packing in the unit cell is accommodated by both ring π-π stacking and the interlocking together of the 3COOH molecules, dominated by hydrogen bonding involving all three carboxyl COOH groups (Fig. 5c). The rotatable bond of the carboxylic acid allows for two possible hydrogen bonding orientations in the crystal lattice seen here, generating an average bond length for OH carboxyl donor atoms. The 3COOH molecules interact through π∙∙∙π stacking of the central benzene ring as a column consisting of sets four molecules plus three molecules, packed as a herringbone arrangement resulting in twenty-eight (4x7) independent molecules in the ASU. The total solvent accessible volume is 29404 Å3 made from four equivalent ‘voids’ of 7351 Å3 or 8.7% each. The overall void volume is 14750 Å3 or 17% of the unit cell volume (84656 Å3), when calculated using solvent molecule of diameter of 1.4 Å. There is a continuous single channel running along the ac axes [23] (Fig. S3) with the narrowest region accommodating a probe molecule radius of 2 Å3 and other two dimensions of about 1.4x1.6 nm. Within the channels there are no fixed ordering of solvent based on the residual electron density maps. The thermal stability of HOF-3COOH building block measured by the TGA curve under N2 (Fig. S19) experienced an initial 5% weight loss below 80oC, attributed to the loss of solvent within the pores. A second weight loss of > 20% between 200oC and 600oC was attributed to any remaining solvent trapped and the thermal decomposition of the molecular precursors. The CO2 adsorption isotherm at 223 K of HOF-3COOH shows an uptake capacity of 3.48 mmol/g (Fig. S20), which can be considered excellent for a HOF [28]. Next, HOF-3COOH crystals were immersed in a warm solution of TPY in toluene (0.6 mg/mL) and the solvent was left to evaporate at slow rate (~ 18 hours), yielding HOF-on-HOF crystals of HOF-3COOH@P2-TPY. SEM also confirmed the morphology of P2-TPY as nanometric rod crystals (Fig. 5) growing on the surface of HOF-3COOH, thus forming core-shell crystals with combined hierarchical architectures and composition. EDX analysis of HOF-3COOH@P2-TPY for C, N and O elements showed the presence of C and N at the shell and only C and O at the solid core of the crystal, which was expected (Fig. S21). PXRD also confirmed the presence of the P2-TPY and the maintained integrity of HOF-3COOH, (Fig. S22-S23). FTIR analysis of HOF-3COOH@P2-TPY crystals confirmed the presence of characteristic peaks of P2-TPY and HOF-3COOH (Fig. S24). Additionally, the peaks attributed to P2-TPY in the regions of 1566 cm-1-1510 cm-1 and 2960–3021 cm-1 are shifted compared to pure P2-TPY, indicating a new environment and compatible with the presence of CHN….COOH H-bonds within the surface. The thermal stability of HOF-3COOH@P2-TPY building blocks was also studied by TGA under N2. The TGA curve (Fig. S25) experienced an initial 8% weight loss below 100oC, attributed to the loss of solvents within the pores. A second weight loss of > 50% between 200oC and 450oC was attributed to any remaining solvent trapped within the pores and the thermal decomposition of the molecular precursors. The porosity of HOF-3COOH@P2-TPY was also studied by gas adsorption isotherms after sample activation (Fig. S26). The CO2 adsorption isotherm at 223 K of HOF-3COOH@P2-TPY shows an uptake capacity of 2.3 mmol/g, which corresponds to a moderate adsorption. To our knowledge, this is the first stable HOF-on-HOF crystal. The dissimilar parameters for the unit cells in both crystals (a = 54.6 Å, b = 31.2 Å, c = 54.5 Å, b = 114.6° vs a = 17.4 Å, b = 40.0 Å, c = 17.4 Å, b = 119.9°) suggest lattice mismatch. Like in Hybrid-1 and P2-TPY@Hybrid-1, stronger H-bonding interactions between the COOH and pyridine groups at the interface could be the driving mechanism for the growth of P2-TPY on the surface of HOF-3COOH.
2.5. Superhydrophobicity, hierarchy, and contaminant trapping
Hierarchy in HOFs could be used as an elegant route for the introduction of multifunctionality in crystalline materials for specific applications. For example, the hierarchical architecture of some of these materials could be used to introduce special wettability and trapping properties, (Fig. 6). Previous studies have demonstrated that hierarchical materials can alter the surface’s wettability [29]. However, the wettability of HOFs has been hardly explored, with only a few reports published, showing the amphiphilic nature of building blocks due to the presence of aromatic hydrophobic blocks and hydrophilic functional groups [30–32]. To increase the hydrophobic nature of HOFs, previous reports were based on designing fully fluorinated pores [30]. We propose that HOFs with multiscale architectures could be good candidates as hydrophobic materials. Some of these hierarchical materials display low wettability in presence of water, forming liquid marbles and floating on water (Fig. S27). Indeed, when we measured the contact angle for P2-TPY@P2-TPY crystals, angles of > 150 ° were obtained, which can be considered as superhydrophobic (Fig. 6). To our knowledge, this is the first reported HOF with superhydrophobic properties. The high hydrophobicity was also demonstrated for DFA-TPY (Fig. 6) and P2-TPY@Hybrid-1 (Fig. S27). The hydrophobicity and hierarchical fractal nature of DFA-TPY could be advantageous in applications for challenging separations such as selective oil-water adsorption and microplastics trapping, as they are 2 of the most relevant environmental contaminants. Oil in water is a well-known and persistent contaminant [33] and, similarly, microplastics can be found in the environment and even the human body, with the toxicity that entails [34]. Traditional methods for the removal of microplastics in wastewater plants are not effective for micrometre size plastics [35] and new methods are required [36, 37]. As a proof of concept, we tested DFA-TPY for the removal of both oil and solid contaminants from water. When a mixture of water and hexane or petroleum ether (Fig. 6) dyed with oil red were prepared, DFA-TPY was able to completely remove 99% oil in minutes, proving the efficacy of hydrophobicity, with no visual evidence of remaining oil. For microplastics, a water solution containing PET microplastics were filtered through DFA-TPY, with 98% of the total amount of microplastics removed. The mechanism of microplastic trapping is expected to be favoured by hydrophobic interactions and the hierarchical nature of the fractal DFA-TPY (Fig. 6). Beyond its efficiency on removing contaminants and due to its crystalline nature, DFA-TPY can be easily regenerated after end of its life to new crystals by recrystallisation.