Redox Triggers Guest Release and Uptake Across a Series of Azopyridine‐Based Metal–Organic Capsules

Precise control over guest release and recapture using external stimuli is a valuable goal, potentially enabling new modes of chemical purification. Including redox moieties within the ligand cores of molecular capsules to trigger the release and uptake of guests has proved effective, but this technique is limited to certain capsules and guests. Herein, the construction of a series of novel metal–organic capsules from ditopic, tritopic, and tetratopic ligands is demonstrated, all of which contain redox‐active azo groups coordinated to FeII centers. Compared to their iminopyridine‐based analogs, this new class of azopyridine‐based capsules possesses larger cavities, capable of encapsulating more voluminous guests. Upon reduction of the capsules, their guests are released and may then be re‐encapsulated when the capsules are regenerated by oxidation. Since the redox centers are on the ligand arms, they are modular and can be attached to a variety of ligand cores to afford varying and predictable architectures. This method thus shows promise as a generalized approach for designing redox‐controlled guest release and uptake systems.


Introduction
Metal-organic capsules, [1] which possess well-defined inner cavities to bind guest molecules, have shown utility in applications that include chemical purification, [2] drug delivery, [3] molecular sensing, [4] and catalysis. [5]For these uses, it is necessary to control the process of guest release and uptake.To this end, different stimuli-responsive functional groups [6] have been incorporated into the backbones of cage ligands, allowing guest uptake and release to be coupled to physical stimuli such as light, [7] or chemical modifications, [8] such as chemical redox [9] or protonation. [10]edox stimuli have been shown to regulate guest binding affinity in metal-organic capsules, through the inclusion of redoxactive motifs on the ligands or subcomponents incorporated into the host capsules. [11]Sallé and co-authors reported the DOI: 10.1002/adma.202302580first example of redox-controlled guest release in metal-organic capsules that incorporated a redox-active extended tetrathiafulvalene (exTTF) motif. [12]Upon oxidation, the bent butterfly-shaped ligands transformed into flat aromatic panels, resulting in cage disassembly and guest release.We also reported a naphthalenediimide-based metal-organic capsule, which utilized electron transfer to control the uptake and release of both neutral and charged guests. [13]hese previous systems relied upon embedding redox-active motifs in the ligand cores, which necessarily dictated capsule size and shape, along with guest-binding properties. [14]A more general method of redox-controlled guest release, which disentangles the capsule core structure from redox-responsiveness, would enable the design of capsules tailored for a wider range of possible guests.In this work, we introduce and develop the strategy of incorporating redox-responsive diazopyridine motifs into the peripheries of metal-organic capsules, providing a general strategy for the design of capsule systems capable of reversible redox-controlled guest release and uptake.
Instead of focusing on the ligand cores, we decided to modify the ligand arms, [15] replacing the dynamic pyridylimine bonds used in the subcomponent self-assembly of metal-organic capsules (Figure 1a) [16] with redox-active azo groups (Figure 1b).The close chemical similarity between these two functionalities led us to anticipate the production of structures analogous to their imine-based congeners.In addition to forming metalcoordination bonds, azopyridine groups can reversibly accept electrons or donate electrons, resulting in complexes that can exist in multiple stable oxidation states. [17]hen redox-active azopyridine moieties are installed on the periphery of a capsule, they do not impose restrictions on the design of the ligand core, enabling greater freedom to construct a capsule with a desired set of host-guest properties.Here, we report three new multitopic azopyridine ligands: ditopic A, tritopic B, and tetratopic C (Figure 1c).Each ligand produced with iron(II) a metal-organic capsule with a distinct geometry: edgelinked Fe II 4 A 6 tetrahedron 1, face-capped Fe II 4 B 4 tetrahedron 2, and Fe II 8 C 6 pseudo-cube 3, respectively.Coordination of the ligands to the metal centers lowers the energy of the ligand * orbitals, allowing for easier reduction of the capsules, as compared to free ligands. [18]All three capsules released their guests upon reduction and re-encapsulated them following re-oxidation.Our incorporation of redox-active azopyridine groups into coordination cages thus opens up new possibilities for electrically controlled guest release and uptake, with possible applications in the field of chemical purification.

Synthesis of Azopyridine-based Capsules and their Host-Guest Chemistry
Reported procedures [19] were used to prepare 2-nitrosopyridine (Figure 1b), which was then treated with di-, tri-, and tetratopic anilines to yield the azopyridine-based ligands A, B, and C (Section S1.2, Supporting Information).Capsules 1 and 2 were obtained by treating ligands A and B, respectively, with iron(II) bis(trifluoromethanesulfonyl)imide (triflimide, Tf 2 N − ) in acetonitrile at 50 °C overnight.The formation of capsule 3 from ligand C and iron(II) triflimide required a higher temperature, with the reaction taking place under microwave heating at 120 °C for 3 h after pre-dissolving the reactants at 50 °C. 1H NMR spectroscopy and mass spectrometry confirmed the formation of all three capsules in solution.Unlike free azopyridine groups, isomerization is effectively inhibited in these cages, as the transconfiguration is "locked" after coordinating to metal centers, as observed in other systems. [17,18]he 1 H NMR spectrum of capsule 1 showed a single set of peaks (Figure S7, Supporting Information), indicating the selective formation of the T-symmetric diastereomer in solution.This outcome differed from the mixture of T, C 3 , and S 4 symmetric diastereomers produced in the case of the congener of 1 containing ditopic iminopyridine ligands, 1′. [20]Capsule 1 encapsulated anions such as PF 6 − and TfO − , and single crystals of TfO − ⊂1 suitable for X-ray structure determination were obtained by slow diffusion of diethyl ether into an acetonitrile solution of TfO − ⊂1.This X-ray structure also revealed a T-symmetric cage framework, consistent with our NMR observations in solution (Figure 2a; Figure S106, Supporting Information).Notably, although triflate was too large to be encapsulated within pyridylimine-based capsule 1′, encapsulation of triflate within the azopyridine-based capsule 1 was observed both in solution and in the solid state.
To understand the differences in host-guest chemistry between the azopyridine-based capsules and their imine-based analogs, the X-ray structure of PF 6 − ⊂1′, was considered alongside that of TfO − ⊂1 (Figure 3) as they both exhibited T-symmetric frameworks; the empty imine capsule 1′ only crystalized as the S 4 -symmetric diastereomer. [20]Encapsulated guests were removed to calculate the host cavity volumes of both capsules using Molovol. [21]The cavity volume of TfO − ⊂1 was 117.7 Å 3 , more than twice as large as the cavity of PF 6 − ⊂1′ (52.1 Å 3 ) calculated us-ing identical parameters, even though the metal-to-ligand bond lengths and metal-metal distances are very similar in both capsules.The average metal-to-ligand bond lengths in TfO − ⊂1 and PF 6 − ⊂1′ are 1.97 and 1.96 Å, respectively, which is consistent with the low-spin Fe II observed by NMR in both cases.The average Fe … Fe distance for TfO − ⊂1 was 12.87 Å, as compared with 12.70 Å in PF 6 − ⊂1′, indicating that both capsules are similar in size despite their considerable difference in cavity volumes.
We attribute this difference in cavity volumes to the greater flexibility of the azopyridine-based capsule, which can expand its cavity to encapsulate triflate.Although the substitution of a CH group by a nitrogen atom led to only subtle changes in bond lengths, the absence of a hydrogen on the azo bonds provided additional conformational freedom to the ligands.
As shown in Figure 3a, this extra freedom could be gauged by measuring the angle  at the metal-bound nitrogen atom.A smaller  value pushes the phenyl rings at the capsule edges outward, giving rise to a larger cavity.For azopyridine host 1, the NNC angle  is 111 ± 1°, which deviates substantially from the ideal sp 2 hybridization angle of 120 o .However, in  iminopyridine-based counterpart 1′, the average  value is 115 ± 4°, closer to the ideal 120 o .We infer that the presence of the imine protons in 1′ prevents the outward bending of the terminal phenyl rings, thereby limiting the flexibility of the imine cage to adapt to larger guests by reducing .In azopyridine host 1, lacking imine protons, the terminal phenyl rings experience less hindrance, thus giving rise to greater flexibility to adopt a larger cavity volume.
The dihedral angle  between the N-Fe-N chelate planes and phenylene rings (Figure 3c) was also inferred to impact the cavity volume.In host 1′, the average value of  is 75°, [20] within the reported range of 69 ± 9°for analogous non-strained capsules. [22]The average  angle in 1 is 46°, however, outside the 69 ± 9°range.In adopting a narrower  angle, the phenylene rings defining the edges of 1 protrude less into the cage cavity (Figure 3c), thus leaving more space for guest encapsulation than in the case of host 1′.The greater flexibility that the azopyridine lends to capsule 1 thus translates into adaptability to bind larger guests.
The 1 H NMR spectrum of capsule 2 displayed two sets of signals with one set dominating, unlike its pyridylimine congener 2′, [23] which showed only one set.Two corresponding sets of peaks for encapsulated triflimide were also observed in the 19 F NMR spectrum (Figure S27, Supporting Information).These results were consistent with the presence of two pairs of diastereomers in solution, with distinct combinations of the handiness (Δ/Λ) of the metal corners and orientations (clockwise (C)/anticlockwise (A)) of the ligand panels. [24]The 1 H NMR spectrum indicated that both possess T point symmetry, consistent with the presence of both (C 4 Δ 4 /A 4 Λ 4 )-2 and (A 4 Δ 4 /C 4 Λ 4 )-2.
Capsule 2 showed similar guest encapsulation properties to its congener 2′.Both diastereomers of 2 were observed to encapsulate adamantane and 1-fluoroadamantane, as indicated by the presence of two sets of peaks for the host-guest complexes in 1 H NMR spectra (Figure 4).Encapsulation of neutral guests in the cationic capsule may be entropically favorable, as discussed in detail in Section S2.3 (Supporting Information).However, the binding affinities of adamantane and 1-fluoroadamantane were weaker for azopyridine-based capsule 2 than for 2′, as following the addition of 5 equivalents of guests, free capsule 2 was still observed (Figure 4).
Single-crystal X-ray diffraction revealed the solid-state structure of adamantane⊂2 (Figure 2b).The asymmetric unit consists of both enantiomers of the (C 4 Δ 4 /A 4 Λ 4 )-adamantane⊂2 diastereomer, and one free B ligand (Figure S110, Supporting Information).The T symmetry exhibited by the X-ray structure of 2 was consistent with its solution-state 1 H NMR spectrum.
We compared the X-ray structures of adamantane⊂2 with its imine congener 2′.The host cavity volume of 2 in adamantane⊂2 is 216.6 Å 3 , while the inner cavity volume of 2′ is 192.8Å 3 , roughly 10% smaller than azopyridine-based 2. The dihedral angles  between N-Fe-N chelate planes and phenylene rings of both capsules are similar (2: 71 o , 2′: 78 o ), with both in the 69 ± 9 o "normal" range. [22]A similar trend in  angles at the imine N atom (Figure 3a) was found between 2 and 2′ as with 1 and 1′.In 2, the average  was 111 o , deviating more from the ideal 120 o than did its imine analog 2′ ( = 117 o ).This reduction in  is inferred to contribute to the 10% increase in cavity volume in the case of 2 over 2′.
Encouraged by the observation that azopyridine-based capsule 1 encapsulated larger guests than its imine congener 1′, we also investigated the capability of 2 to bind larger guests.For instance, diamantane has a Van der Waals volume of 196.4 Å 3 , too large for the inner cavity of 2′(192.8Å), as we verified by 1 H NMR (Figure S100, Supporting Information).Upon the addition of diamantane into a solution of 2, however, new signals for host 2 and encapsulated diamantane were observed (Figure 4).Unlike adamantane and 1-fluoroadamantane, which bound to both diastereomers of 2, as reflected in the observation of two sets of host-guest complex 1 H NMR peaks, the larger diamantane was only accommodated by the diastereomer of 2 with the larger cavity, since only one set of 1 H NMR peaks was observed for diaman-tane⊂2.This effect results from the difference in cavity volumes between distinct diastereomeric cage configurations, [24] where only one is large enough to accommodate diamantane.The encapsulation of diamantane within 2 but not 2′ also highlights the flexibility of such azopyridine-based hosts, which are capable of reconfiguration to accommodate larger guests.
Capsule 3 was prepared at [C] = 1.2 mm, as at higher concentrations (3.6 mm, Figure S40, Supporting Information) we observed the exclusive formation of an Fe II 16 C 12 extended tetrahedron (Figure S40, Supporting Information), in a similar fashion to a Zn II 16 L 12 capsule reported previously. [25]The X-ray crystal structure of pseudo-cube 3 revealed T h symmetry, with four Δ Fe II centers and four Λ centers (Figure 2c), which is in line with our 1 H NMR observations (Figure S30, Supporting Information), as only one set of ligand signals was observed.Mass spectrometry also confirmed the composition of capsule 3 (Figure S39, Supporting Information).This structure and the larger Fe II 16 C 12 one are homologs of the products formed by the iminopyridine analog of ligand C with Zn II , [25] again highlighting the similar behavior of the iminopyridine and azopyridine ligand series.The 1 H NMR spectrum of 3 displayed a major set of peaks consistent with the X-ray structure of 3 (Figure S30, Supporting Information), alongside a minor set of peaks corresponding to the Fe II 16 C 12 product.Comparison of the  and  values, as defined in Figure 3, were again instructive when the X-ray structures of 3 and 3′ were compared.Both 3 and 3′ displayed dihedral angles  (3: 71 o , 3′: 68 o ) within the 69 ± 9 o "normal" range. [22]Azopyridine capsule 3 possesses a cavity volume of 1083.4Å 3 , about 11% larger than the 974 Å 3 cavity of iminopyridine-based 3′, again following the trend we observed for capsules 1/1′ and 2/2′.This volume increase also tracks with the smaller  value of 3′, which we again attribute to the lower steric demands of the azopyridine moieties.Capsule 3 was observed to encapsulate three corannulene molecules by 1 H NMR spectroscopy and mass spectrometry (Figures S101 and  S109, Supporting Information).

Electrochemical Properties of Azopyridine-Based Capsules
The electrochemical properties of redox-active ligands A-C and capsules 1-3 were studied by cyclic voltammetry in acetonitrile with a tetrabutylammonium triflimide electrolyte (Figure 5; Figures S116-S118, Supporting Information).Both ligands and capsules were reduced, and the reduction was reversible over three cycles.Compared with the capsules, the ligands were more difficult to reduce, with the first pseudo-reversible reduction waves appearing at −1.5 V versus Fc/Fc + .
The lower-lying * orbital of the coordinated azopyridine led to a more accessible reduction of the capsules, as expected.The first reduction waves of capsules 1-3 occurred at ca. −0.5 V, similar to a mononuclear Fe II azopyridine complex. [17]Capsule 2 was reduced at −0.555 V, the most negative potential observed among the three capsules.We infer that the electron-donating tertiary amine groups of the ligand B may render reduction less thermodynamically favorable for 2. Cyclic voltammetry suggested that capsules 1-3 could be reduced by decamethylferrocene Fe(C 5 Me 5 ) 2 , with reduction potentials in all cases at least 0.035 V more negative than decamethylferrocinium.

Redox-Induced Guest Release and Uptake in Azopyridine-Based Capsules
Reversible redox behaviors in pyridylimine-based capsules are difficult to achieve without installing redox-active motifs in the ligand cores, [13] with the reduction of capsules lacking such moieties leading to irreversible chemical degradation.For example, the redox properties of pyridylimine-based capsule 2′ were investigated with different reducing agents.Mild reductants such as decamethylferrocene (Figure S136, Supporting Information) and cobaltocene were not successful in reducing imine bonds, [12] whereas the stronger reducing agent decamethylcobaltocene led to irreversible precipitation of degradation products following addition to an acetonitrile solution of 2′ (Figure S137, Supporting Information).
The redox properties of coordinated azopyridine groups in first-row transition metal complexes have been investigated fruitfully. [17,18,26]The Goswami group reported a Ni II azopyridine complex that loses one coordinated azopyridine upon reduction, − ⊂1 (bottom) was first treated with Fe(C 5 Me 5 ) 2 , which was dissolved in CDCl 3 to ensure its solubility, resulting in capsule reduction (middle), then then an acetonitrile solution of AgNTf 2 was added to recover the host-guest complex by oxidation (top).The whole process was carried out using one host-guest sample.1,3,5-Methoxybenzene was used as an internal standard, with peaks marked by asterisks.d) Redox of PF 6 − ⊂1 monitored by 19 F NMR spectroscopy.The host-guest complex (bottom) encapsulated PF 6 − , which was released upon reduction (middle), and re-bound following re-oxidation (top).
18f] This phenomenon suggested that the installation of azopyridine corners at capsule vertices might potentially lead to the reduction-driven capsule disassembly and subsequent release of encapsulated guests.
Chemical reductions were performed for each capsule 1-3, with different guests encapsulated (PF 6 − ⊂1, 1-fluoroadamantane⊂2 and corannulene⊂3) in order to investigate the generality of this redox-controlled guest-release strategy in azopyridine capsules.Decamethylferrocene was selected as the reducing agent based on our cyclic voltammetry results, and CDCl 3 was chosen as the solvent to ensure the solubility of all components.No evidence was observed for oxidation of the cage vertices to Fe III , consistent with the stability of Fe II under similar redox conditions employed for a previously reported Fe II -iminopyridine-based complex. [13]Following reduction, the system was then oxidized with AgNTf 2 to recover the capsule.Reduction and oxidation were monitored by 1 H and 19 F NMR spectroscopy (Figure 6).
To calibrate the amount of reducing agent to be added to the host solution, we first carried out titrations into an acetonitrile solution of PF 6 − ⊂1, monitored by 1 H NMR spectroscopy (Figure S119, Supporting Information).The signals broadened significantly upon the addition of Fe(C 5 Me 5 ) 2 , and within the diamagnetic region, only one peak remained, which shifted progressively upon addition.After the addition of 6 equivalents of Fe(C 5 Me 5 ) 2 , the only signal remaining no longer shifted, which we inferred to indicate the completion of the reduction.The same amount of reducing agent was also added to empty capsule 1, resulting also in only one signal remaining in the diamagnetic region assigned to the host.This signal for reduced capsule 1 appeared at the same chemical shift as for PF 6 − ⊂1, consistent with the conclusion that after reduction, the PF 6 − guest no longer bound to the host (Figure S126, Supporting Information).A 1 H DOSY experiment was performed on the reduced product of PF 6 − ⊂1, which indicated possible fragmentation after reduction (Figure S113, Supporting Information).However, the characterization of the reduced species remained challenging, as multiple products may exist simultaneously after the reduction of a capsule with multiple ligands and metal centers.Mass spectrometry did not enable meaningful analysis of the reduced products, which could reflect their instability under the ionization conditions employed or their reaction with atmospheric oxygen (Figure S129, Supporting Information).The broadness of the NMR signals also prevented detailed analysis of the reduced products.18f] The 19 F NMR spectrum also showed the disappearance of the signals corresponding to encapsulated PF 6 − .Results from both 1 H and 19 F NMR spectroscopy were thus consistent with the release of encapsulated PF 6 − upon reduction.When AgNTf 2 (6 equiv.) was added to the reduced product, capsule 1 H NMR signals reappeared, and signals corresponding to encapsulated PF 6 -were observed again in 19 F NMR.These observations are consistent with the reversibility of capsule 1 reduction and guest re-encapsulation, with a recovery rate of ca.90% based upon NMR integration versus the internal standard 1,3,5trimethoxybenzene.Further cycles of in situ chemical redox occurred reversibly, with 81% capsule 1 recovery after five cycles (Figure 6c; Figure S124, Supporting Information).
Similar redox-controlled guest release and uptake were observed in capsule 2. Among the guests that were encapsulated by 2, 1-fluoroadamantane was chosen to investigate reversibility because the process could be followed using 19 F NMR spectroscopy.Fe(C 5 Me 5 ) 2 (6 equiv) was added to the 1-fluoroadamantane⊂2 solutions, as previously described.The capsule signals then disappeared, and a new set of signals appeared.According to 1 H DOSY NMR (Figure S132, Supporting Information), the solvodynamic radius of the species present in solution was 8.38 Å.Since the average Fe … Fe distance in 2 was 15.39 Å, we infer the host to have disassembled, with the release of the originally encapsulated guest.This inference was further supported by 19 F NMR spectroscopy, with the intensity of the peak corresponding to encapsulated 1-fluoroadamantane diminishing significantly upon reduction (Figure S131, Supporting Information).Upon oxidation with AgNTf 2 (6 equiv), capsule 2 was recovered, resulting in guest re-uptake.By-products were also observed in the 1 H NMR spectrum, attributed to irreversible processes that potentially involve the redox-noninnocent N-CH 3 triazine groups.These processes decreased the capsule recovery rate to 75% (Figure S133, Supporting Information).
The dynamic release of corannulene from capsule 3 was also demonstrated upon reduction using Fe(C 5 Me 5 ) 2 (Figure S134, Supporting Information).Capsule 1 H NMR signals disappeared upon reduction, as with the other two capsules.In the case of 3, however, we did not observe complete guest release, which might be due to aromatic stacking between the partially disassembled capsule and guests.The amount of capsule 3 recovered after one cycle was calculated to be 87% based on the internal standard (Figure S135, Supporting Information)

Conclusion
The preparation of capsules 1-3 thus establishes the utility of the azopyridine motif in the construction of capsules isomorphous to their iminopyridine counterparts, but with slightly larger cavity volumes.Future azopyridine capsules may thus be designed to encapsulate a predictable range of different guests, based upon the guest-binding properties of their iminopyridine congeners.As all three azopyridine-based capsules displayed reversible redox-triggered guest release and uptake for their respective types of guests, suggesting that our approach could be extended to other capsules with different ligand cores, to achieve guest release and encapsulation as required.A detailed investigation into the chemical processes that follow electron transfer during the redox process is currently ongoing.The use of electricity to effect redox-driven guest release could enable the design of new electrically fueled chemical purification processes, whereby a guest is taken up selectively from a mixture, and released in pure form, as is currently under investigation as well.
[CCDC 2247767-2247769 contain the supplementary crystallographic data for this paper.These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif ].

Figure 1 .
Figure 1.a) Subcomponent self-assembly process of Fe II metal-organic capsules, where imine bonds are formed during coordination to the metal center.b) Synthetic scheme for the class of azopyridine-based Fe II metal-organic capsules introduced in this work.c) Synthesis of the new azopyridine-based metal-organic capsules 1, 2, and 3.

Figure 3 .
Figure 3. Structural comparison between azopyridine capsule 1 and its iminopyridine congener 1′: a) narrower  angles at nitrogen lead the phenylene edges to bow outward, bringing about b) an expansion of the cavity volume; c) narrower dihedral angles  between N-Fe-N chelate planes and terminal phenyl rings (colored blue) lead the phenylene rings in 1 to protrude less into the central cavity of 1 than for 1′, likewise impacting the cavity volumes.

Figure 4 .
Figure 4. a) The crystal structure of adamantane⊂2, with the guest removed and showing the central void space highlighted in mesh.b) 1 H NMR spectra of, from bottom to top, guest-free capsule 2, adamantane⊂2, 1-fluoroadamantane⊂2, and diamantane⊂2.Blue stripes in the 1 H NMR spectra highlight signals corresponding to free capsule 2. c) The two diastereomers of capsule 2, which originate from the two different combinations of the handedness of the metal corners and the orientations of the triazine panels.The guests, adamantane and 1-fluoroadamantane, were encapsulated in both diastereomers, but diamantane was only bound by one diastereomer.

Figure 6 .
Figure 6.a) Cartoon illustrations of guest release and uptake via redox in the three azopyridine-based metal-organic capsules 1-3.b) Percentage of the metal-organic capsule 1 remaining after five cycles of in situ redox.c) Redox of PF 6 − ⊂1 monitored by 1 H NMR spectroscopy.The host-guest complex PF 6− ⊂1 (bottom) was first treated with Fe(C 5 Me 5 ) 2 , which was dissolved in CDCl 3 to ensure its solubility, resulting in capsule reduction (middle), then then an acetonitrile solution of AgNTf 2 was added to recover the host-guest complex by oxidation (top).The whole process was carried out using one host-guest sample.1,3,5-Methoxybenzene was used as an internal standard, with peaks marked by asterisks.d) Redox of PF 6 − ⊂1 monitored by19 F NMR spectroscopy.The host-guest complex (bottom) encapsulated PF 6 − , which was released upon reduction (middle), and re-bound following re-oxidation (top).