Control over Phase Transformations in a Family of Flexible Double Diamondoid Coordination Networks through Linker Ligand Substitution

In this work, we present the first metal–organic framework (MOF) platform with a self-penetrated double diamondoid (ddi) topology that exhibits switching between closed (nonporous) and open (porous) phases induced by exposure to gases. A crystal engineering strategy, linker ligand substitution, was used to control gas sorption properties for CO2 and C3 gases. Specifically, bimbz (1,4-bis(imidazol-1-yl)benzene) in the coordination network X-ddi-1-Ni ([Ni2(bimbz)2(bdc)2(H2O)]n, H2bdc = 1,4-benzenedicarboxylic acid) was replaced by bimpz (3,6-bis(imidazol-1-yl)pyridazine) in X-ddi-2-Ni ([Ni2(bimpz)2(bdc)2(H2O)]n). In addition, the 1:1 mixed crystal X-ddi-1,2-Ni ([Ni2(bimbz)(bimpz)(bdc)2(H2O)]n) was prepared and studied. All three variants form isostructural closed (β) phases upon activation which each exhibited different reversible properties upon exposure to CO2 at 195 K and C3 gases at 273 K. For CO2, X-ddi-1-Ni revealed incomplete gate-opening, X-ddi-2-Ni exhibited a stepped isotherm with saturation uptake of 3.92 mol·mol–1, and X-ddi-1,2-Ni achieved up to 62% more gas uptake and a distinct isotherm shape vs the parent materials. Single-crystal X-ray diffraction (SCXRD) and in situ powder X-ray diffraction (PXRD) experiments provided insight into the mechanisms of phase transformation and revealed that the β phases are nonporous with unit cell volumes 39.9, 40.8, and 41.0% lower than the corresponding as-synthesized α phases, X-ddi-1-Ni-α, X-ddi-2-Ni-α, and X-ddi-1,2-Ni-α, respectively. The results presented herein represent the first report of reversible switching between closed and open phases in ddi topology coordination networks and further highlight how ligand substitution can profoundly impact the gas sorption properties of switching sorbents.


■ INTRODUCTION
Stimuli-responsive materials 1 create the opportunity for properties that are infeasible in rigid equivalents, such as guest selectivity 2,3 and gas uptake triggered by external stimuli. 4 Such phenomena are typically enabled by structural transformations, in some cases involving extreme structural changes. 5−9 These materials are of interest in multiple areas of application, including gas storage, 10 gas separation, 11 molecular sensing, 12 and proton conduction. 13 Flexible metal−organic frameworks (MOFs) are promising members of the dynamic porous solids family, 14−16 due in part to their amenability to crystal engineering, 17 which allows for systematic investigation of structure-function relationships. A well-explored crystal engineering strategy is to use ligand substitution to create families of isostructural materials. 18 In the case of rigid materials, ligand substitution can retain pore shape and size while changing pore chemistry to determine how this influences uptake and selectivity (Scheme 1, left). 19 −21 In the case of flexible MOFs, which can undergo transformations from phases of reduced porosity (or nonporosity) to highly porous phases through "gate-opening" mechanisms, control over the "switching" or "breathing" is of particular interest. This is because switching MOFs can offer improved working capacities as opposed to traditional sorbents with Langmuirtype isotherm profiles and, therefore, are expected to benefit gas-related technologies if their gate-opening/closing events occur within the right pressure range (Scheme 1, right, shaded area). 22,23 Their technological potential has intensified the quest for switching MOF sorbents; however, insufficient understanding of the switching mechanism limits the ability of crystal engineers to design them from first principles. There exist a relatively small number of known switching MOF platforms 24 with around 60 reported sorbents, most of which are individual MOFs not associated with a platform. Analysis of these platforms suggests that structural transformations can be driven by both intranetwork and internetwork phenomena, including geometric changes of the coordination environment, 5 linker motion, 6,25 and network expansion in twodimensional (2D) frameworks 26,27 or network sliding in threedimensional (3D) frameworks. 28 Since the presence of at least one of these phenomena is likely to be a prerequisite for flexibility, the choice of the metal node and organic linker would be expected to impact structural dynamics, allowing for systematic studies of structure and function.
Mixed crystals are defined by IUPAC as a type of crystal containing a second constituent, which fits into and is distributed in the lattice of the host crystal. 29 Mixed crystals are known to exist for multiple classes of compounds, including MOFs, for which the term MTV MOFs has been coined. 30 Non-stoichiometric substitution of multiple ligands or metals into a MOF represents one of the crystal engineering approaches used to fine-tune the properties of multicomponent systems. For example, ligand-based mixed crystals have been utilized to tune gas and water sorption profiles of both rigid 31,32 and flexible 4,33,34 MOFs. In addition, rare-earth-based mixed crystals have been extensively studied for their luminescence properties, 35,36 while transition metal mixed crystals have been studied for their gas sorption performance. 37 The properties of the mixed crystals often 38 fall between the properties of the parent materials. An alternate strategy to exploit isostructural MOFs is the use of epitaxial crystal growth 39,40 to create multi-phase "MOF-on-MOF" materials.
In this work, we report a platform of MOFs with double diamondoid (ddi) topology comprised of 8-cmetal nodes that exhibit switching properties. The single-crystal and powder Xray analysis of the parent member of this platform, [Ni 2 (bimbz) 2 (bdc) 2 (H 2 O)·6DMF] n (bimbz = 1,4-bis-(imidazol-1-yl)benzene, H 2 bdc = 1,4-benzendicarboxylic acid, DMF = N,N-dimethylformamide; Scheme 2), herein referred to as X-ddi-1-Ni-α, were reported in 2008 (CSD Refcode BONMAT). 65 Interestingly, the authors reported reversible framework transformations upon solvent exchange based on powder X-ray diffraction (PXRD) measurements. Herein, we elaborate on this study by extending the X-ddi platform to Xddi-2-Ni, in which bimbz was replaced by bimpz (3,6bis(imidazol-1-yl)pyridazine, Scheme 2) and a 1:1 mixed crystal variant, X-ddi-1,2-Ni, thereby allowing us to investigate the effect of ligand substitution on sorption properties. ■ EXPERIMENTAL SECTION Materials and Methods. The linkers bimbz and bimpz were synthesized by modified reported procedures (see Supporting Information for detailed procedures). 66 Other reagents and solvents were commercially available and were used without further purification. Thermogravimetric analysis (TGA) was performed using a TA Instruments Q50 analyzer at a rate of 10.00°C/min from 25 to 550°C under nitrogen gas flow. Differential scanning calorimetry (DSC) analysis was performed on a TA Instruments Q2000 system at a rate of 5°C/min from 25 to 250°C under nitrogen gas flow. Fourier transform infrared (FTIR) spectroscopy was performed on a PerkinElmer Spectrum 100 FTIR spectrometer with ATR and a Spotlight 200 FTIR microscope attachment. Raman spectra were collected on a Horiba LabRAM 1A spectrometer, Scheme 1. Control Over Pore Chemistry Via Crystal Engineering Strategies Can Be Used to Fine-Tune Both Rigid (Left) and Switching (Right) Sorbents for Their Sorbent−Sorbate Affinity, Which in Turn Impacts Uptakes (U) or Gate-Opening Pressures (P go ), Respectively Scheme 2. Chemical Structure of the Three Ligands Used in the Synthesis of the X-ddi Platform Studied Herein: H 2 bdc, bimbz, and bimpz equipped with an Olympus BX40 confocal microscope. Elemental analysis was performed using an Exeter Analytical CE 440 elemental analyzer. Scanning electron microscopy (SEM) measurements were carried out on a Hitachi SU-70 instrument, using a 3 kV acceleration voltage and a working distance of 15 mm. 1 H NMR spectroscopy was performed using a JEOL ECX400 spectrometer operating at 400 MHz. Powder X-ray diffraction (PXRD) diffractograms were recorded on a PANalytical Empyrean diffractometer operated at 40 kV and 40 mA, using Cu Kα radiation (λ = 1.5406 Å). Variable-temperature PXRD (VT-PXRD) studies were performed on a PANalytical X'Pert diffractometer operated at 40 kV and 40 Ma, using Cu Kα radiation (λ = 1.5406 Å). Data were collected from 4−40°(2θ).
Gas Sorption Measurements. For gas sorption experiments, high-purity gases were used as received from BOC Gases Ireland: CO 2 (99.995%), N 2 (99.9995%%), C 3 H 4 (97.0%), C 3 H 6 (99.5%), C 3 H 8 (99.95%). A Micromeritics 3Flex surface area and pore size analyzer 3500 was used for collecting the low-pressure sorption isotherms for CO 2 , N 2 , and C3 gases. The temperature of 77 K was maintained using liquid nitrogen, and the temperature of 195 K was maintained using a dry ice−acetone mixture. Bath temperatures of 273 and 298 K were controlled with a Julabo ME (v.2) recirculating control system containing a mixture of ethylene glycol and water. A Hiden Isochema XEMIS-001 gravimetric sorption analyzer was used for collecting the high-pressure sorption isotherms for CO 2 .
In Situ Powder X-Ray Diffraction Measurements. In situ PXRD measurements were carried out on a Rigaku Smartlab instrument with Cu Kα radiation (λ = 1.5406 Å) connected to a BELSORP-18PLUS volumetric sorption instrument, attached to a cryostat system. The samples were activated under a high vacuum at 85°C for 12 h, and the weight of the evacuated samples was determined under an inert atmosphere. CO 2 sorption was carried out at 195 K, and PXRD data were simultaneously measured at each equilibrium point from 4−40°(2θ).
X-ddi-1-Ni-α ( Figure 1A) crystallized in the orthorhombic space group Fdd2. The asymmetric unit (ASU) is comprised of one Ni 2+ ion, one bdc 2− linker, one bimbz linker, and half a coordinated water molecule. The molecular building block (MBB) consists of two equivalent Ni 2+ centers connected by a μ 2 -OH 2 bridge and two bidentate (η 2 ) carboxylate groups, while the coordination sphere for each Ni 2+ is completed by one monodentate (η 1 ) carboxylate group and two nitrogen atoms from two bimbz linkers, resulting in pseudo-octahedral geometry. Therefore, the resulting MBB has eight points of extension ( Figure 1B) and self-assembles into a 3D framework with ddi topology ( Figure 1C, see Section S13), which is a type of self-penetrated dia-c net. 67 Although this MBB has yet to be encountered in switching MOFs, a CSD survey (version 2022.2.0) revealed that it is found in 446 structures containing any type of metal, out of which 105 are coordination networks. Upon inspecting the results on the TopCryst database 68 for this topology and analyzing them based on cluster representation, one other valence-bonded structure with ddi topology was found (CSD refcode BONMEX). 65 Additionally, an inspection of the results of the CSD survey for this node yielded one more structure with ddi topology (CSD refcode IQIMEC), which brings, to the best of our knowledge, the total number of coordination networks with this topology to three. The MBBs are connected by alternating η 2 −η 1 coordination modes of bdc 2− and bimbz linkers in anticonfigurations ( Figure 1D) that self-assemble into a noninterpenetrated framework with alternating chains of bdc 2and bimbz or bimpz linkers ( Figure 1E).
Despite being comprised of similar components to X-ddi-1-Ni-α, X-ddi-2-Ni-α ( Figure 1A) crystallized in the monoclinic space group Cc (Tables S4−S7). The crystallographic c axis is reduced by ca. 44%, resulting in a unit cell volume that is equivalent after being normalized for the reduced crystallographic symmetry (13125.4 Å 3 for 1 vs 12823.2 Å 3 for 2, for Z = 8). The change in the space group in comparison to the In the case of X-ddi-2-Ni-α, the ASU contains two bimpz linkers with nitrogen atoms oriented in opposite directions, replacing the C 2v axis in X-ddi-1-Ni-α with a mirror plane (C s ) ( Figure  2, left and Figure S8). As a result, the pore environment in Xddi-2-Ni-α is less symmetrical than in X-ddi-1-Ni-α, as shown by the guest-accessible space (Figure 2, right). The relationship between the pyridazine ring rotation and the space group of the crystal was further studied by a variable-temperature SCXRD experiment on the same single crystal (Figures S9− S11, see Section S2 for experimental details). Even though the crystal solved as monoclinic at low temperature (100 K), at higher temperature (298 K) it solved as orthorhombic, indicating that the rotation of the pyridazine ring is hindered at low temperature, therefore creating a less symmetric environment than at high temperature. The space group determination was further supported by solving X-ddi-1-Ni-α and X-ddi-1,2-Ni-α as monoclinic (100 K), for which PLATON detected missed symmetry ( Figures S12 and S13). The rectangular channels along the crystallographic b and a axes for X-ddi-1-Ni-α and X-ddi-2-Ni-α, respectively, account for guest-accessible space of ca. 47.3 and 45.2% (Figure 2), respectively (for a probe radius of 1.2 Å). Thermogravimetric analysis (TGA) indicated six DMF molecules per Ni 2 unit for all compounds (Figures S14−S16).
Upon activation of X-ddi-2-Ni, additional phases with intermediate pore volumes (less open than X-ddi-2-Ni-α and more open than X-ddi-2-Ni-β) were observed. In order to gain access to an intermediate pore volume, as-synthesized X-ddi-2-Ni-α was exchanged with methanol, which has a smaller diameter than DMF ( Figure S27). The solvent exchange resulted in a less open phase, X-ddi-2-Ni-γ ( Figure 2B) with guest-accessible space of ca. 43.6%, and a transformation to the space group Fdd2, through rotation of the pyridazine ring ( Figure S8). Leaving the sample to dry in air for ca. 10 min caused further narrowing of the pore size to X-ddi-2-Ni-δ ( Figure 2B), guest-accessible space being ca. 16.6%. The bimpz linker in this structure was found to be disordered over two positions ( Figure S28, see Section S2 for refinement details). PXRD measurements validated bulk-phase purity and confirmed the relationships between the phases of X-ddi-2-Ni ( Figures S29−S32). The bulk phase of X-ddi-2-Ni-δ was found to spontaneously lose guest MeOH molecules in the air,  Figure S33). SCXRD analysis revealed that the flexibility in X-ddi-1-Ni and X-ddi-2-Ni can be rationalized in three ways. First, the structural transformations from α to β in X-ddi-1-Ni and α to β, γ, and δ in X-ddi-2-Ni could be enabled by the flexible nature of the bimbz and bimpz linkers, respectively ( Figure  3A). Rotation and bending motions of the imidazole rings with respect to the central rings allowed for a stepwise decrease of the distance between two MBBs from 13.403 and 13.296 Å to 13.203 and 13.044 Å from α to β in X-ddi-1-Ni and X-ddi-2-Ni, respectively ( Figure S34). Second, the seemingly rigid bdc 2linker could induce hinge-like motions through its carboxylate carbon atoms (Figure 3B), facilitating a variety of coordination angles that enable pore expansion or contraction ( Figure S35). Therefore, the shortest Ni to Ni distance between two MBBs contracted from 10.712 and 10.818 Å to 10.421 and 10.541 Å from α to β and the η 1 coordination angle ranged from 20.33 and 13.60 to 40.84 and 43.61°from α to β in X-ddi-1-Ni and X-ddi-2-Ni, respectively. Third, the dinuclear Ni unit might afford access to diverse coordination geometries for the metal centers ( Figures 3C and S36), thereby characterizing it as a "soft" MBB. The orientation of O and N atoms from monodentate bdc 2− and bimbz/bimpz linkers can change to accommodate structural changes. A collective effect of the factors mentioned above resulted in flexibility triggered by common organic solvents, heat, or gas molecules ( Figure S37). In turn, this flexibility enabled changes in the distances and angles between the MBBs that comprise the pore opening ( Figures S38−S40), which can be summarized by the differences in unit cell parameters upon structural transformations, as shown in Figure 3D.
Sorption Analysis. Even though X-ddi-1-Ni, X-ddi-2-Ni, and X-ddi-1,2-Ni are isostructural in their closed or β phases, their sorption isotherm profiles are distinct. Low-pressure isotherms measured on the three closed phases at 195 K showed flexibility toward CO 2 (Figure 4), with saturation uptakes (at P/P 0 = 1) of 3.47, 3.92, and 5.63 mol·mol −1 for Xddi-1-Ni-β, X-ddi-2-Ni-β, and X-ddi-1,2-Ni-β, respectively. In the case of X-ddi-1-Ni ( Figure 4A), gas sorption revealed negligible uptake until P/P 0 ≈ 0.79, followed by sharp increase in uptake (i.e., gate-opening). However, corresponding in situ PXRD measurements ( Figure S41) suggested incomplete gateopening, as the characteristic peaks of the closed phase are present even at P/P 0 = 1 (marked with an asterisk in Figure  4A), indicating that the open and closed phases continued to coexist at 1 bar. 69 Conversely, X-ddi-2-Ni ( Figure 4B) displayed steadily increasing uptake with increasing pressure up to P/P 0 ≈ 0.49 (and uptake of 1.24 mol·mol −1 ), before showing sudden gate-opening. The initial uptake before gateopening corresponds to ca. 10 CO 2 molecules per unit cell, which suggests the possibility of gas diffusion into the nonporous β phase 70,71 ( Figure S42) and is consistent with the change in the relative intensities of the peaks of the in situ PXRD patterns recorded up to P/P 0 ≈ 0.49 ( Figure S43). Xddi-1,2-Ni ( Figure 4C) exhibited a single-step isotherm with gate-opening at P/P 0 ≈ 0.65 and uptake of 0.70 mol·mol −1 before opening. The in situ PXRD patterns for X-ddi-1-Ni and X-ddi-1,2-Ni demonstrated that the initial gas uptake before gate-opening is related to sorption phenomena associated with the closed phases. Through gate-opening, both X-ddi-2-Ni and X-ddi-1,2-Ni underwent a structural transformation to a more open phase resembling the δ form discussed in the crystallographic section. Examination of the in situ PXRD patterns at P/P 0 = 1 ( Figure S44) revealed that X-ddi-1,2-Ni expanded further than X-ddi-2-Ni, as the former displayed signals corresponding to characteristic peaks of the γ form. Nevertheless, the mixed crystal achieved an intermediate gateopening value compared to the parent compounds and an improved uptake by 62 and 44% with respect to X-ddi-1-Ni and X-ddi-2-Ni, respectively.
Interestingly, low-pressure gas sorption for C 3 H 4 at 273 K afforded more open phases than those observed for CO 2 at 195 K ( Figure 5). This can be ascribed to the permanent dipole moment of C 3 H 4 and its large van der Waals size, which enables highly exothermic adsorption relative to CO 2 , as we have reported in switching sql topology networks. 44,45 X-ddi-1-Ni-β displayed a single-step type F-IV isotherm 8 with gateopening at 0.50 bar and uptake of 3.32 mol·mol −1 at 1 bar. Xddi-2-Ni-β exhibited a two-step type F-IV 2 (F-IV m ) isotherm under the same conditions: the first step occurred at 0.13 bar with an uptake of 2.00 mol·mol −1 after opening, while the second step appeared at 0.65 bar with an uptake of 6.34 mol· mol −1 at 1 bar. As for the CO 2 experiments, X-ddi-1,2-Ni-β showed a stepped isotherm for C 3 H 4 with gate-opening between that of the parent compounds, at 0.27 bar, and uptake of 3.32 mol·mol −1 at 1 bar. We observed the same trend of gate-opening pressure (P go ) values for the first transformation induced by C 3 H 4 (at 273 K) as seen in CO 2 isotherms (at 195 K): X-ddi-2-Ni < X-ddi-1,2-Ni < X-ddi-1-Ni. Exposure to C 3 H 4 induced all three compounds to transform from the closed (β) phase to a phase of intermediate porosity, with uptakes ranging from 2.08 to 3.83 mol·mol −1 ( Figure 5). However, since X-ddi-2-Ni displayed the lowest P go values out of the three compounds, the second transformation (from the phase of intermediate porosity to a phase of higher porosity) for X-ddi-2-Ni occurred below 1 bar at 273 K. In contrast, the corresponding second transformations for X-ddi-1-Ni and X-ddi-1,2-Ni were not observed below 1 bar at 273 K. Sorption measurements for C 3 H 6 and C 3 H 8 at 273 K revealed no uptake for all three compounds ( Figure S50). SEM images obtained after sorption of the C3 gases ( Figure S51) indicated that the particle size had diminished, while PXRD measurements showed that all three compounds reverted to their respective closed phases after desorption ( Figure S52).
High-pressure gas sorption studies for CO 2 at 273 K afforded similar isotherm profiles to the low-pressure isotherms collected at 195 K. In the case of X-ddi-1-Ni, no gate-opening was observed ( Figure S53) since this event occurred at high P/ P 0 values at 195 K. Instead, an isotherm profile resembling that before gate-opening was obtained, with an uptake of 1.06 mol· mol −1 at 35 bar. X-ddi-2-Ni showed appreciable uptake (2.13 mol·mol −1 ) up to 19 bar, followed by a step and an increase in uptake to reach the saturation value of 4.39 mol·mol −1 at 35 bar ( Figure S54). X-ddi-1,2-Ni displayed incomplete gateopening ( Figure S55). Repeated cycling diminished the uptake of X-ddi-1-Ni, whereas, in the cases of X-ddi-2-Ni and X-ddi-1,2-Ni, the step moved to lower-pressure values. PXRD measurements after both high-and low-pressure CO 2 gas sorption ( Figure S56) showed that the samples remained crystalline after repeated cycling and validated that they had reverted to their respective closed phases.
The transformations of the three variants of the X-ddi platform from their as-synthesized open phases (α) to the respective closed phases (β) were found to be among the most extreme structural transformations in 3D switching MOFs reported thus far (Table S9) as measured by unit cell volume reduction. Specifically, when compared to the open phases, reductions of 39.9, 40.8, and 41.0% for X-ddi-1-Ni, X-ddi-2-Ni, and X-ddi-1,2-Ni, respectively, were observed, ranking the X-ddi platform 6th after the DUT-8(Ni), Co(bdp), DUT-98, MIL-53, and X-dia-1-Ni platforms. Notably, the unit cell volume change in these platforms is associated with physical sample expansion or contraction, which needs to be considered when shaping flexible MOFs for industrial applications 72 but can be inherently advantageous for sensing devices. 73 In terms of the effect of ligand substitution on gate-opening pressure (P go ), a review of CO 2 sorption studies conducted at 195 K on 3D switching MOF platforms (Table S10) revealed that the effect of ligand substitution on P go ranged from 0.02 to 0.70 bar, our results lying in the middle of this range. Halogen functionalization on JUK-8 48 achieved simultaneous shifting of P go and improvement of CO 2 uptake at 1 bar by 46.5%; however, this was accompanied by a large increase in hysteresis (up to 5100% increase compared to the parent material). A series of mixed crystals based on carboxylate ligands with flexible pendant groups 51 demonstrated a variety of P go values based on the linker ratio, with no significant effect on uptake and increase in hysteresis. Change in the conformational freedom of the ligand in the X-pcu-n-Zn platform 28 showed that reduction in hysteresis is possible, but the effect of ligand substitution on gas uptake was insignificant. Hence, to the best of our knowledge, the X-ddi platform is the first example of a switching sorbent platform where ligand substitution induced both a shift in P go and an increase in uptake for CO 2 (at 195 K and 1 bar) while reducing hysteresis compared to the parent material.

■ CONCLUSIONS
We report the structural transformations of two new stimuliresponsive MOFs with ddi topology, X-ddi-1-Ni and X-ddi-2-Ni, as well as the mixed crystal variant X-ddi-1,2-Ni. Through this platform, we introduce a new molecular building block that can induce dynamic behavior triggered by gases (CO 2 , C 3 H 4 ) and liquids (DMF, methanol). The three variants were evaluated for their sorption properties, and their differences in response to guest molecules were attributed to differences in pore chemistry. SCXRD and variable pressure in situ PXRD provided insights into the flexibility modes of the ddi platform. Additionally, the continuous breathing of X-ddi-2-Ni was validated with multiple crystal structures with different pore