Improved Acid Resistance of a Metal–Organic Cage Enables Cargo Release and Exchange between Hosts

Abstract The use of di(2‐pyridyl)ketone in subcomponent self‐assembly is introduced. When combined with a flexible triamine and zinc bis(trifluoromethanesulfonyl)imide, this ketone formed a new Zn4L4 tetrahedron 1 bearing twelve uncoordinated pyridyl units around its metal‐ion vertices. The acid stability of 1 was found to be greater than that of the analogous tetrahedron 2 built from 2‐formylpyridine. Intriguingly, the peripheral presence of additional pyridine rings in 1 resulted in distinct guest binding behavior from that of 2, affecting guest scope as well as binding affinities. The different stabilities and guest affinities of capsules 1 and 2 enabled the design of systems whereby different cargoes could be moved between cages using acid and base as chemical stimuli.


Synthesis and characterization 2.1 Preparation of tetrahedron 1
Scheme S1. Subcomponent self-assembly of 1.

X-ray crystallography
Data were with collected either using a Bruker D8 VENTURE equipped with high-brilliance IμS Cu-Kα radiation (1.54178 Å), with ω and ψ scans at 180(2) K or at Beamline I19 of Diamond Light Source employing silicon double crystal monochromated synchrotron radiation (0.6889 Å) with ω scans at 100(2) K. [2] Data integration and reduction were undertaken with Xia2 [3] or CrysalisPro. [4] Subsequent computations were carried out using the WinGX-32 graphical user interface. [5] Empirical absorption corrections were applied to the data using the AIMLESS tool [6] in the CCP4 suite. [7] Structures were solved by direct methods using SHELXT-2013 [8] then refined and extended with SHELXL-2013. [9] In general, non-hydrogen atoms with occupancies greater than 0.5 were refined anisotropically.
Carbon-bound hydrogen atoms were included in idealised positions and refined using a riding model. solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data. These measures and the use of a high intensity laboratory source allowed data to be collected to ca. 0.86 Å resolution. The asymmetric unit was found to contain one complete Zn4L4 assembly and associated counterions and solvent molecules. Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other and thermal S16 parameter restraints (SIMU, RIGU) were applied to all atoms except for zinc.
Two of the non-coordinated pyridine rings were modelled as disordered over two or three locations with bond length and thermal parameter restraints applied in order to obtain a reasonable refinement.
Several of the anions within the structure also show evidence of disorder. Three tetrafluoroborate anions were modelled as disordered over up to five locations. Several solvent molecules were also modelled as disordered over multiple locations and/or with partial occupancy. Substantial bond length and thermal parameter restraints were applied to facilitate a reasonable refinement of the disordered anions and solvent molecules and most low occupancy disordered moieties were modelled with isotropic thermal parameters.
The SQUEEZE [10] function of PLATON [11] was employed to remove the contribution of the electron density associated with a small amount of highly disordered solvent, which gave a potential solvent an acetonitrile solution of the complex. The crystals immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data.
The asymmetric unit was found to contain one complete Zn4L4 assembly and associated counterions and solvent molecules. Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other and thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for zinc.

S17
The anions within the structure show evidence of substantial disorder. Seven triflate anions were modelled as disordered over two locations and multiple anions were modelled with partial occupancy.
Several solvent molecules were also modelled as disordered over multiple locations and/or with partial occupancy. Substantial bond length and thermal parameter restraints were applied to facilitate a reasonable refinement of the disordered triflate anions and solvent molecules and most low occupancy disordered groups were modelled with isotropic thermal parameters.
Further reflecting the solvent loss and poor diffraction properties there is a significant amount of void volume in the lattice containing smeared electron density from disordered solvent. Consequently the SQUEEZE [10] function of PLATON [11] was employed to remove the contribution of the electron density associated with this highly disordered solvent, which gave a potential solvent accessible void of 1233.7 Å 3 per unit cell (a total of approximately 308 electrons). Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diethyl ether only those solvent molecules that could be modelled with discrete atom positions are included in the formula. to collect data. Despite these measures and the use of a high intensity laboratory source few reflections at greater than 1.05 Å resolution were observed. Nevertheless, the quality of the data is far more than sufficient to establish the connectivity of the structure. The asymmetric unit was found to contain one complete Zn4L4 assembly and associated counterions and solvent molecules. Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other and thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for zinc and rhenium.

S18
There is a significant amount of thermal motion in the extremities of the molecule, notably around the uncoordinated pyridyl groups; a number of restraints were required for the realistic modelling of these groups. Even with these restraints the thermal parameters of some pyridyl rings are larger than ideal resulting in high carbon, hydrogen and nitrogen Uiso min/max ratios. Modelling these groups as disordered over multiple discrete positions did not improve the overall model.
The anions within the structure show evidence of substantial disorder. All seven triflimide anions were modelled as disordered over two or three locations and multiple anions were modelled with partial occupancy. Some minor occupancy locations for the disordered anions could not be located in the electron density map and were therefore not modelled. Substantial bond length and thermal parameter restraints were applied to facilitate a reasonable refinement of the disordered anions most low occupancy disordered sites were modelled with isotropic thermal parameters.
Further reflecting the solvent loss and poor diffraction properties there is a significant amount of void volume in the lattice containing smeared electron density from disordered solvent. Consequently the SQUEEZE [10] function of PLATON [11] was employed to remove the contribution of the electron density associated with this highly disordered solvent, which gave a potential solvent accessible void of 3370 Å 3 per unit cell (a total of approximately 864 electrons). Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diethyl ether only those solvent molecules that could be modelled with discrete atom positions are included in the formula.
Specific refinement details: The crystals of [PF6⊂1]·5.25PF6·1.75NTf2·7.5MeCN·0.5Et2O were grown by diffusion of diethyl ether into an acetonitrile solution of [Zn4L4]·8NTf2 containing excess Bu4NPF6. The crystals immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream S19 was required to collect data. Despite these measures and the use of a high intensity laboratory source few reflections at greater than 1.1 Å resolution were observed. Nevertheless, the quality of the data is far more than sufficient to establish the connectivity of the structure. The asymmetric unit was found to contain two complete Zn4L4 assemblies and associated counterions and solvent molecules. Bond lengths and angles within pairs of chemically identical organic ligands were restrained to be similar to each other and thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for zinc.
One coordinated pyridine ring was modelled as disordered over two locations with bond length and thermal parameter restraints applied in order to obtain a reasonable refinement. In addition there is a significant amount of thermal motion in the extremities of the molecule, notably around the uncoordinated pyridyl rings; a number of restraints were required for the realistic modelling of these groups. Even with these restraints the thermal parameters of some pyridyl rings are larger than ideal resulting in high carbon, hydrogen and nitrogen Uiso min/max ratios. Modelling these groups as disordered over multiple discrete positions did not improve the overall model.
Several of the anions within the structure show evidence of disorder. Three hexafluorophosphate anions were modelled as disordered over two or three locations and multiple anions were modelled with partial occupancy. Some minor occupancy locations for the disordered anions could not be located in the electron density map and were therefore not modelled. Several solvent molecules were also modelled as disordered over multiple locations and/or with partial occupancy. Substantial bond length and thermal parameter restraints were applied to facilitate a reasonable refinement of the disordered hexafluorophosphate anions and solvent molecules and most low occupancy disordered groups were modelled with isotropic thermal parameters.
Further reflecting the solvent loss and poor diffraction properties there is a significant amount of void volume in the lattice containing smeared electron density from disordered solvent. Consequently the SQUEEZE 7 function of PLATON 8 was employed to remove the contribution of the electron density associated with this highly disordered solvent, which gave a potential solvent accessible void of 1948 Å 3 per unit cell (a total of approximately 606 electrons). Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diethyl ether only those solvent molecules that could be modelled with discrete atom positions are included in the formula.
Specific refinement details: The crystals of [SbF6⊂1]·6.2SbF6·0.8NTf2 were grown by diffusion of diethyl ether into an acetonitrile solution of 1·8NTf2 containing excess KSbF6. The crystals immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data.
Despite these measures and the use of synchrotron few reflections at greater than 1.02 Å resolution were observed. Nevertheless, the quality of the data is far more than sufficient to establish the connectivity of the structure. The asymmetric unit was found to contain one half of a Zn4L4 assembly and associated counterions. Bond lengths and angles within the two chemically identical organic ligands were restrained to be similar to each other and thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for zinc and antimony.
Two non-coordinated pyridine rings were modelled as disordered over two locations with bond length and thermal parameter restraints applied in order to obtain a reasonable refinement. In addition there is a high level of thermal motion throughout the structure, most notably around the other non-coordinated pyridyl rings; a number of restraints were required for the realistic modelling of these groups. Even with these restraints, the thermal parameters of some pyridyl rings are larger than ideal. Modelling further pyridyl rings as disordered over multiple discrete positions did not improve the overall model.
The anions within the structure show evidence of disorder. One anion lattice site was modelled as a disordered mixture of hexafluoroantimonate and triflimide with refined occupancies of 0.6 and 0.4 respectively. Substantial bond length and thermal parameter restraints were applied to facilitate a reasonable refinement of the disordered anion and most low occupancy disordered groups were modelled with isotropic thermal parameters. The encapsulated hexafluoroantimonate, which is located on a special position, shows a high level of thermal motion; attempts were made to model this anion as disordered over multiple discrete positions but this did not improve the overall model.
Further reflecting the solvent loss and poor diffraction properties there is a substantial amount of void volume in the lattice containing smeared electron density from disordered solvent and five anions per Zn4L4 assembly (assigned to hexafluoroantimonate in the formula). Despite many attempts to model this region of disorder as a combination of solvent and anion molecules no reasonable fit could be found, even with the use of restraints or rigid body constraints. Consequently the SQUEEZE [10] function of PLATON [11] was employed to remove the contribution of the electron density associated with the highly disordered solvent and anions, which gave a potential solvent accessible void of 8741 Å 3 per unit cell (a total of approximately 3050 electrons). Since the diffuse solvent molecules could not be assigned conclusively to acetonitrile or diethyl ether they are not included in the formula.
Specific refinement details: The crystals of [SbF6⊂2]·7SbF6·0.5MeCN were grown by diffusion of diethyl ether into an acetonitrile solution of 2·8NTf2 containing excess KSbF6. The crystals immediately lost solvent after removal from the mother liquor and rapid handling prior to flash cooling in the cryostream was required to collect data.
The crystals were very small and were subject to rapid beam damage during data collection using synchrotron radiation. Consequently few reflections at greater than 1.3 Å resolution were observed and the quality of the integration is less than ideal. Due to the beam damage only 95% data completeness could be achieved. Nevertheless, the quality of the data is sufficient to establish the connectivity of the structure. The asymmetric unit was found to contain two complete of a Zn4L4 assemblies and associated counterions and solvent molecules.
In order to obtain a reasonable model for the organic parts of the structure the GRADE program [12] was employed using the GRADE Web Server [13] to generate a full set of bond distance and angle restraints (DFIX, DANG, FLAT). Due to the thermal motion and less than ideal resolution, thermal parameter restraints (SIMU, RIGU) were applied to all atoms except for iron and antimony to facilitate anisotropic refinement. Even with these restraints some thermal parameters remain larger than ideal as a consequence of the high level of thermal motion throughout the structure. Bond length and thermal parameter restraints were also applied to the hexafluoroantimonate anions.
Further reflecting the solvent loss and poor diffraction properties there is a substantial amount of void volume in the lattice containing smeared electron density from disordered solvent and 5.5 anions per Zn4L4 assembly (assigned to hexafluoroantimonate in the formula). Despite many attempts to model this region of disorder as a combination of solvent and anion molecules no reasonable fit could be found, even with the use of restraints or rigid body constraints. Consequently the SQUEEZE [10] function of PLATON [11] was employed to remove the contribution of the electron density associated with the S22 highly disordered solvent and anions, which gave a potential solvent accessible void of 7547 Å 3 per unit cell (a total of approximately 2533 electrons). Since the diffuse solvent molecule(s) could not be assigned conclusively to acetonitrile or diethyl ether only those solvent molecules that could be modelled with discrete atom positions are included in the formula. solvents of crystallization are omitted for clarity. We infer that the steric hindrance between the free pyridine rings and their neighboring phenylene rings restricts the free rotation of these phenylenes, consistent with the phenomenon observed in the 1 H NMR spectrum, which shows four distinct signals for these protons ( Figure S1).

Volume calculations
In order to determine the available void space within 1 and 2, VOIDOO calculations [14] based on the crystal structures (with the anionic guests removed) were performed. A virtual probe with a radius of 1.4 Å (set by default, water-sized) was employed, and the standard parameters tabulated below were used. [15] The results are shown in Table S1 and Figure S22. [H]0 represent the concentrations of host-guest complex at equilibrium and total host, respectively) as a function of the total concentration of guest added and fitting the data to the 1:1 binding model. [16] However for some particular guests, peak overlap between those of the host and host-guest complex prevented the accurate integration of peaks for each species and therefore the binding constants were The binding constants of cages 1 and 2 for the investigated neutral and anionic guests are tabulated in Table S2.

Acid resistance studies of cages
The studies of the acid resistance ability of cages were investigated in CD3CN by progressive addition of different amounts of triflimide acid (HNTf2). After each addition of acid to the cage solution, the sample was kept at room temperature for five minutes to reach equilibrium prior to recording the corresponding 1 H NMR spectrum. The amount of the remaining cage was determined by integration of the 1 H NMR spectra with 1,3,5-triethylbenzene as an internal standard. N,N-diisopropylethylamine was used as the base to test the recovery of the cage after its disassembly.

Cargo delivery between cages 1 and 2
The experiments of SbF6 − delivery from cage 2 to cage 1 using acid and base as chemical stimuli were carried out as following: To an NMR tube, a solution (500 μL) of SbF6 − ⊂2 at 3 mM was prepared by adding SbF6 − (1.0 equiv) to 2 (3 mM) in CD3CN using 1,3,5-triethylbenzene as an internal standard. It was found that under these conditions, around 95% of the cage was occupied by the SbF6 − guest with 5% of the cage being empty. Solid cage 1 (1.0 equiv) was then added into the solution to prepare an equimolar mixture of 1 and SbF6 − ⊂2 (3 mM). Different equivalents of acid (HNTf2, up to 9.0 equiv) and base (N,N-diisopropylethylamine, up to 9.0 equiv) were sequentially added and 1 H NMR spectra were recorded to monitor the SbF6 − transfer from cage 2 to cage 1. Control experiments without adding any acid or base into the solution were also performed to investigate the unassisted cargo delivery process from 2 to 1 by recording the 1 H NMR spectra of the mixture after different periods of time. comparison. Typical peaks from these species have been labelled. The solution was found to take 45 hours at room temperature to reach a steady state after mixing 1 and SbF6 − ⊂2, which contained approximately 80% SbF6 − ⊂1 and 20% SbF6 − ⊂2.

Cargo exchange between cages 1 and 2
The experiments of cargo exchange between SbF6 − ⊂2 and 1,3-dioxane⊂1 using acid and base as chemical stimuli were carried out as following: To an NMR tube, a solution (450 μL) of SbF6 − ⊂2 at 3.3 mM was prepared by adding SbF6 − (1.0 equiv) to 2 (3.3 mM) in CD3CN using 1,3,5-triethylbenzene as an internal standard. It was found that under these conditions, around 95% of the cage was occupied by the SbF6 − guest with 5% of the cage being empty. A solution (450 μL) of 1,3-dioxane⊂1 (3.3 mM), which was prepared by mixing the 1,3-dioxane guest (40 equiv) and 1 (3.3 mM) in CD3CN, was then added to the sample to prepare an equimolar mixture of 1,3-dioxane⊂1 and SbF6 − ⊂2 (1.7 mM).
Different equivalents of acid (HNTf2, up to 10 equiv) and base (N,N-diisopropylethylamine, up to 10 equiv) were sequentially added and 1 H NMR spectra were recorded to monitor the cargo exchange between the two cages. Control experiments without adding any acid or base into the solution were also performed to investigate the unassisted cargo exchange process between cages 1 and 2, by recording the 1 H NMR spectra of the mixture after different periods of time.