Selective and Reversible Solvent Uptake in Tetra‐4‐(4‐pyridyl)phenylmethane‐based Supramolecular Organic Frameworks

Abstract The dynamic behavior of supramolecular organic frameworks (SOFs) based on the rigid tetra‐4‐(4‐pyridyl)phenylmethane (TPPM) organic tecton has been elucidated through 3D electron diffraction, X‐ray powder diffraction and differential scanning calorimetry (DSC) analysis. The SOF undergoes a reversible single‐crystal‐to‐single‐crystal transformation when exposed to vapours of selected organic solvents, moving from a closed structure with isolated small voids to an expanded structure with solvated channels along the b axis. The observed selectivity is dictated by the fitting of the guest in the expanded SOF, following the degree of packing coefficient. The effect of solvent uptake on TPPM solid‐state fluorescence was investigated, evidencing a significant variation in the emission profile only in the presence of chloroform.

It is therefore expected that, according to Kitaigorodskii studies on organic crystals, the molecules assemble in the solid state to minimise free space. [46] However, the relative weakness of the supramolecular interactions likely involved in stabilizing the crystal structure of TPPM enhances the probability of obtaining a flexible network [47] susceptible of external, stimulimediated structural changes. [31] In this work, we present the dynamic behaviour of TPPM which, in its crystalline form, can reversibly switch from a nonporous empty to a filled solvated phase and vice-versa, when exposed to vapours of organic solvents and heat, respectively. Remarkably, this single-crystal-to-single-crystal transformation [48][49][50][51] is selectively triggered only by a group of solvents, namely chloroform, dichloroethylene, trichloroethylene, benzene and toluene. In the specific case of chloroform uptake, a significant variation of the fluorescence emission profile was observed.

Results and Discussion
To study the solid-state behaviour of TPPM, crystals of the tecton were grown from the slow evaporation of a chloroform solution and analysed through X-ray diffraction methods. TPPM crystallizes as a 1 : 1 CHCl 3 solvate, TPPM · CHCl 3 , in the space group C2/c. The asymmetric unit consists of half a molecule of TPPM and half a molecule of the guest, disordered over two equivalent positions by a rotation around the CÀ H bond ( Figure 1a). The TPPM units assemble to give a supramolecular three-dimensional network held together by several CÀ H···N interactions. In particular, each TPPM forms symmetry-related C12BÀ H12B···N1A, C3À H3A···N1B and C10À H10A···N1B contacts ( Figure S2). This assembly contains channels parallel to the b   3 . The solvent molecules filling the channels have been drawn in space filling mode. c) View of the two symmetry-related orientations of ethanol in the crystal structure of TPPM · EtOH. d) View of benzene sandwiched between two TPPM ligands in TPPM · C 6 H 6 . e) View of acetonitrile inside the channels in TPPM · 0.5CH 3 CN. axis direction filled by the disordered solvent molecules (Figure 1b). Inside the pores, chloroform forms two CÀ H···C aromatic interaction involving the atoms C5A and C8A of TPPM (Figure S3).
Interestingly, the same supramolecular network also forms when TPPM is crystallized from different solvents, as reported in the literature for DMF [41] (CSD [52] refcode IVETEJ) and DMSO [42] (CSD refcode: ZISXAC). In both isomorphous crystals, the solvent molecules occupy the channels in a 1 : 1 ratio. This prompted us to further study the robustness and the reproducibility of the SOF, by soaking single crystals of the chloroform solvate (for 2 h at room temperature) in different media such as ethanol, benzene and acetonitrile, potentially capable of forming hydrogen bonds or π···π stacking interactions with TPPM. In all three cases, the crystals retained their transparency and habit; X-ray diffraction analysis confirmed the substitution of chloroform by the bulk solvents, and the formation of the three solvates TPPM · EtOH, TPPM · C 6 H 6 and TPPM · 0.5CH 3 CN. All solvates crystallize in the space group C2/c with comparable unit cell dimensions; TPPM · EtOH and TPPM · C 6 H 6 possess the same supramolecular network described above for TPPM · CHCl 3 , TPPM · DMF and TPPM · DMSO ( Figure S4 and Table S1). The asymmetric unit of TPPM · EtOH consists of half a molecule of TPPM and half a molecule of the guest, which, due to symmetry, occupies the channels formed by the ligand with two different, symmetry-related orientations (see Figure 1c). Analogously, the asymmetric unit of TPPM · C 6 H 6 consists of half a molecule of TPPM and half a molecule of the guest. Figure 1d shows benzene sandwiched between two symmetry-related TPPM ligands, highlighting the shape complementarity of the guest with the channels formed by the host. In the case of TPPM · 0.5CH 3 CN, the asymmetric unit comprises half a molecule of TPPM and one fourth of a molecule of acetonitrile. Interestingly, TPPM is disordered over two positions, one of which is not compatible with the presence of the solvent; this means that in the crystal, both a solvate ( Figure 1e) and an unsolvated ( Figure S9b) phase are present. The N1S atom of acetonitrile forms a CÀ H···N interaction with a pyridine moiety ( Figure S8). The framework of the solvated phase of TPPM · 0.5CH 3 CN still contains channels along the b axis direction, but it is consolidated by a different set of interactions, comprising also CÀ H···π·interactions ( Figure S9a). The framework of the unsolvated phase, which is quite compact and contains unconnected voids summing up to 6 % of the unit cell volume, is also consolidated by CÀ H···N and CÀ H···π interactions ( Figure S9b).
Crystals of TPPM · CHCl 3 were also exposed both to water vapours and soaked in distilled water for seven days to assess the stability of the solvate in humid and wet conditions. After these treatments, the X-ray diffraction analysis of the crystals revealed the presence of the unaltered chloroform solvate.
The host-guest interactions for the different solvates were compared by means of a Hirshfeld analysis, performed with CrystalExplorer17. [53] The Hirshfeld surface of TPPM is quite similar for TPPM · EtOH and TPPM · C 6 H 6 , indicating that the presence of the guest does not influence much the supramolecular interactions consolidating the framework of the host ( Figure S10). The Hirshfeld surface of TPPM · CHCl 3 shows an additional bright red spot relative to the CÀ H···C aromatic interactions shown in Figure S3, while the red spots on the Hirshfeld surface of TPPM · 0.5CH 3 CN derive from the CÀ H···π interactions ( Figure S9a) which are missing in the other solvates. For all the forms, the general trend is a predominance of dispersion forces and of H···H, C···H/H···C and N···H/H···N contacts (see fingerprint plots in Figure S11 and Table S2).
This peculiar behaviour prompted us to investigate the possibility to isolate an empty crystal structure of TPPM alone. To this purpose, a batch of single crystals of TPPM · CHCl 3 was heated at 100°C for 2 h. Part of the batch was grinded and analysed through X-ray powder diffraction; the comparison of the diffractograms of the chloroform solvate and the activated form clearly showed the formation of a new phase showing high micro-crystallinity, probably due to the lack of disordered solvent inside the channels. The remaining single crystals show an increased mosaicity ( Figure S12) which hampers the use of single crystal X-ray diffraction and calls for 3D electron diffraction (3D ED), [54][55][56] which can be successfully employed with nanometric samples possessing small crystalline domains. Although TPPM crystals are sensitive to the electron beam, thanks to a special low dose set up in which the electron dose is reduced below 0.05 el s À 1 Å À 2 and the diffraction patterns are collected with a Timepix single electron detector, [57] a full 3D ED experiment covering a 120°of reciprocal space can be performed without the amorphization of the crystals or a significant loss in resolution of the diffraction signal. The 3D ED data have been collected in precession electron diffraction mode (PEDT) and allowed an ab-initio structure determination of TPPM using direct methods. Remarkably, the data quality was suitable for the dynamical structure refinement of the TPPM structure; dynamical refinement takes into account, in the modelling of diffracted intensity, the effect of dynamical scattering through a Bloch formalism. [58] The structure is refined together with the thickness of the sample and both agreement factors and quality in the structure determination approach single crystal X-ray precision. [59] In the case of TPPM, dynamical refinement led to an improvement of the structural model and allowed to localize most of the hydrogen atoms ( Figure S16). This is the first time that a SOF structure has been refined at a high precision taking into account dynamical scattering. The activated TPPM, which is indeed the empty phase, crystallizes in the monoclinic space group C2/c like the parent solvates. The biggest changes in the cell parameters involve the a axis, which goes from 27.9295(5) to 31.40(6) Å, the β angle, that likewise increases from 120.1390(10) to 133.1(1)°and the volume, which slightly decreases from 3628.2(1) Å 3 to 3596(12) Å 3 (see Table S1). In the lattice, each TPPM forms symmetry-related C11BÀ H11B···Cg1 and C10AÀ H10 A···Cg2 contacts with four distinct adjacent molecules ( Figure S17; Cg1 and Cg2 are the centroids of rings C8AÀ C12A/N1A and C8BÀ C12B/N1B, respectively). These interactions are lacking in the solvated forms and are responsible for the more compact structure of the framework. The same central reference molecule is also surrounded by another set of four ligands through weak, CÀ H···N interactions involving each of the pyridine rings, acting as H-bond acceptors towards the CÀ H groups of the phenyl moieties. Also in this case, the pattern is rather different from that present in the solvated forms, where each TPPM unit behaves both as Hbond donor and acceptor. The different interactions formed in the empty and solvated phase are also evidenced by the Hirshfeld surface analysis shown in Figure S10.
The net result of this single-crystal-to-single-crystal phase transition is that of a compact framework in which the original channels become non-connected small voids located among the TPPM units; these voids sum up to a volume of ca. 145 Å 3 (4 % of the total unit cell volume) compared to the volume of ca. 610 Å 3 (17 % of the total unit cell volume) occupied by the channels filled with solvent in the solvated form TPPM · S ( Figure 2).
Interestingly, this framework is quite similar to the unsolvated form present in the TPPM · CH 3 CN crystals, which can be seen as a sort of intermediate phase in which, however, the unit cell is that of the solvated form.
The desolvation process was also monitored in detail through X-ray powder diffraction analysis. To this purpose, the microcrystalline powder of TPPM · CHCl 3 was used as TPPM · S phase reference and measured by a temperature resolved in situ X-ray powder diffraction analysis. The TPPM · S phase, during the heating treatment, undergoes a single-crystal-to-singlecrystal phase transition leading to the empty TPPM phase. From the temperature-resolved diffractograms it is evident that crystallinity is retained throughout the whole process (see Figures 3 and S18). The phase transition does not present a gradual change of lattice parameters, but only a reflection intensity change of the two involved species. LeBail refinement was performed for each collected diffractogram, leading to a temperature-dependent correlation between powder profile parameters and the phase fraction (Figures 3 and S19). Interestingly, the dynamic behaviour shown by TPPM is reversible. Indeed, when the empty crystalline TPPM is exposed to the vapours of a series of organic solvents, it switches back to the solvated phase TPPM · S, as evidenced by PXRD analysis (Figures 3 and S20). The vapour uptake is fast (completed after ca. 2 minutes or less) and selective, as the phase transformation occurs with chloroform, dichloroethane, trichloroethylene, benzene, toluene but not with acetonitrile, ethanol, tetrachloroethylene or acetone. Cyclohexane also triggers the phase transformation, but its uptake time is much longer (ca. 20 minutes). Dichloromethane can be also partially absorbed by empty TPPM yielding, however, an unstable phase at room conditions. The affinity towards a group of solvent vapours shown by the SOF was also investigated by 1 H NMR spectroscopy through competition experiments. Chloroform and ethanol were selected as representative of the two classes of vapours uptaken and not uptaken by the empty form. The exposure of the ethanol solvate for different intervals of time (2 and 20 minutes) caused the progressive substitution of ethanol inside the pores, as shown in the spectra reported in Figure 4 by the disappearance of its signals and the appearance of the chloroform ones. The reverse uptake did not occur when the crystalline TPPM · CHCl 3 solvate was exposed to EtOH vapours ( Figure S35).
The framework stability of the TPPM · S phases was investigated through DSC analysis. The characterizations were performed in an open system to minimize possible counterpressure effects. The collected thermograms present an endothermic phase transition, related to the release of guest molecules embedded in the TPPM framework (see Figures 5  and S21). It is worth noticing that the TPPM · S phases directly obtained by vapor absorption present higher desorption enthalpies (ΔH Des ) than both TPPM · EtOH and TPPM · 0.5CH 3 CN ( Figures S22 and S23) suggesting that thermodynamic stability can play a role in the selectivity of the absorption process. The release of solvent molecules during the phase transition was also confirmed by TGA ( Figure S24) which, together with 1 H NMR spectroscopy, was also used to determine the amount of absorbed solvent molecules, reported as solvate stoichiometry ( Figures S25, S33 and S34). The dynamical behaviour of TPPM was studied performing the DSC analysis in a closed system, in which the solvent desorption is followed by the absorption process during the cooling path ( Figure 5). This experiment was repeated several times to confirm the reversibility of the process; indeed, after each cycle the desorption (ΔH Des ) and absorption (ΔH Abs ) enthalpies did not show any significant change. This suggests that also the amount of absorbed and released guest after every repetition remains constant.
To further investigate the selectivity of the guests, their packing coefficient was calculated as the ratio between the guest volume (V Guest ) and the volume of the guest-binding site (V) present in the TPPM · S phase (see Section 5 in the Supporting Information). This parameter has been proposed by Rebek to predict the best binding accommodation of guests inside molecular capsules in solution, which is reached when  . Solvent competition studies followed by 1 H NMR spectroscopy. The TPPM · EtOH phase was exposed to CHCl 3 vapours for 2 and 20 minutes. The solid was then dissolved in CD 2 Cl 2 with few drops of methanol-d 4 . 1 H NMR spectra were collected and compared to the one of the initial solvate. the coefficient is ca. 55 %. [60] Recently, this method has been shown to work nicely also for the absorption of gases inside the pores of supramolecular hosts. [61] Indeed, it is remarkable that in our case, both CHCl 3 and DCE show a packing coefficient of 55 and 56 %, respectively, and almost all the other guests that are absorbed by TPPM present packing coefficients in the range 60 %-70 %. On the contrary, all the solvents that are not uptaken or form unstable host-guest complexes, show a low coefficient packing, typically in the range 35 %-44 %.
The selective vapour uptake prompted us to investigate whether TPPM solid-state spectroscopic properties are influenced by the nature of the absorbed solvent. With this aim, UV-Vis absorption and emission spectra were recorded on powder samples of TPPM in thin layer form before and after their exposure to solvent vapours ( Figures S37-S43). For all tested samples, the absorption profile ( Figure S44) was found to be in good agreement with the one obtained for TPPM in DCM solution ( Figure S36). On the contrary, in the emission spectra of the solid samples, the rising of a second, more intense band centred at 480 nm was observed (Figure 6a), in addition to the one at 360 nm characteristic of the solubilized TPPM. This behaviour is in line with the formation of excimers in tightly packed crystalline structures of aromatic molecules, as confirmed by the excitation profiles. [62] Interestingly, the ratio between the emission intensity at the two maxima varies among all the series of tested solvates and differs from the TPPM empty form (Table S5). As depicted by the diagram reported in Figure 6b, the emission spectra were collected for three different samples and the significance levels for each mean value of intensity ratio was obtained by one-way ANOVA calculation by Tukey Test. Only in the case of TPPM · CHCl 3 a significance level lower than the threshold value of 0.05 was recorded. These results make TPPM a potential candidate for the uptake and detection of chloroform vapours in air. As volatile organic compound (VOC), the release of chloroform in the atmosphere resulting from a wide range of industrial activities is of high interest and has a severe impact on both human and environmental health. [63] In this respect, the stability of both TPPM and TPPM · S in presence of water is of particular relevance, as humidity is the main interferent in environmental monitoring of VOCs.

Conclusion
In summary, we have elucidated the crystal structure of the solvated and empty forms of a crystalline SOF based on the rigid, tetrahedral organic tecton TPPM. This structural analysis, in which 3D ED has played an essential role, has paved the way for a deeper understanding of the dynamic behaviour of the material by means of X-ray powder diffraction and DSC. This  reversible single-crystal-to-single-crystal phase transition is remarkably triggered by a selective group of organic vapours; the origin of this selectivity has been also elucidated and lies in the difference of packing coefficient. In addition, we have shown that the solid-state fluorescence of TPPM is affected by solvent uptake. The combined features of dynamic behaviour and selectivity can be exploited in the recognition of chloroform vapours through fluorescence measurements. We believe that this paper is another step towards a better comprehension of stimuli-responsive materials held together by weak supramolecular interactions.

Experimental Section
Synthesis, characterization and experimental details are provided as Supporting Information. This includes TPPM and SOFs preparation, detailed SCXRD, PXRD, 3D ED, DSC, TGA, NMR and spectroscopic studies and packing coefficient calculations.