Photoswitchable Catalysis by a Self-Assembled Molecular Cage

A heteroleptic [Pd2L2L’2]4+ coordination cage containing a photoswitchable azobenzene-derived ligand catalyzes the Michael addition reaction between methyl vinyl ketone and benzoyl nitromethane within its cavity. The corresponding homoleptic cages are catalytically inactive. The heteroleptic cage can be reversibly disassembled and reassembled using 530 and 405 nm light, respectively, allowing catalysis within the cage to be switched OFF and ON at will.

O ver the past three decades, supramolecular cages have evolved from simple hosts to systems with impressive catalytic functions. 1Inspired by how enzymes accelerate reactions, catalytic cages can leverage their activity in multiple ways.These include methods that rely on reducing the entropy of activation, for example, using binding to limit conformational freedom 2 or by encapsulating more than one substrate to increase effective concentration. 3Cages can also use electrostatic forces to drive catalysis, for example leading to either enhanced basicity 4 or acidity 5 of the substrate compared to the nonbound species, and/or stabilizing any subsequent higher energy species. 6Other methods of accelerating reactions include binding substrates in higher energy conformational states, 7 or using the local high concentration of ions around the cage portals. 8When more than one of these mechanisms is used simultaneously, very high activity can be observed. 9olecular cages have also been used to control regioselectivity, although these reactions are not typically catalytic. 10 key feature of enzyme catalysis is the regulated activity.This area of cage catalysis remains significantly underdeveloped and invariably relies on either the endo or exo binding of a guest that is not a substrate. 11One way to mediate cage catalysis would be using light irradiation.Light-responsive molecular cages have been prepared using photoswitchable ligands and metal ions. 12These include those based on diarylethene photoswitches that form cages with different geometries and cavity sizes, allowing selective guest binding. 13iazocine-based cages are unique as they can be switched from a thermodynamically stable Z-isomer to the E-isomer using UV light, forming metastable cages that encapsulate guests. 14A related example uses two diazocine ligands that allow light to selectively disassemble one cage and assemble another. 15e have recently reported an azobenzene-based molecular cage that reversibly responds to visible light to change its composition from a [Pd 2 L 4 ] 4+ lantern-like structure to a [PdL 2 ] 2+ monomeric product. 16However, guest molecules do not readily bind within the cavity of the lantern-like cage, 16 instead preferentially binding on the exterior.Molecular cages can also act as photosensitizers 17 and have been used to drive photochemical reactions away from equilibrium. 18Photoswitchable catalysis 19 has progressed significantly since the earliest reports, 20 with examples of enantioselective catalysis, 21 polymerization 22 and cooperative catalysis. 23In these examples, the catalysis is switched by changing the electronic properties of a donor atom, 24 blocking an active site with steric bulk, 20 bringing together cooperative organocatalytic groups, 21a or forming a more reactive functional group. 25While there are reports of switchable catalysis within macrocycles, 26 on surfaces of nanoparticles, 27 and using rotaxanes, 28 to the best of our knowledge, there are no reports of photoswitchable catalysis using discrete self-assembled species.Herein we report light-regulated catalysis using a heteroleptic [Pd 2 L 2 L′ 2 ] 4+ cage system (Figure 1).
To prepare the BAr F salt, E-1 was reacted with [Pd(Py*) 4 ] (BAr F ) 2 (where Py* = 3-chloropyridine, see Supporting Information S4) 32 in CD 3 CN to give [Pd 2 (E-1) 4 ](BAr F ) 4 , which was characterized by NMR and ESI-MS data (see Supporting Information S6.4,S6.5).Disappointingly, [Pd 2 (E-1) 4 ](BAr F ) 4 showed little evidence of guest binding, nor was it able to catalyze the representative Michael addition reaction of methyl vinyl ketone and benzoyl nitromethane in the presence of 18-crown-6, similar to the behavior of a related cage 16 (Supporting Information S5.2, S5.3, S18.4.3).There are several possible reasons for this lack of host-guest chemistry and reactivity.It could be that the relative size and shape of the cavity of [Pd 2 (E-1) 4 ](BAr F ) 4 are unsuitable for encapsulating the substrates.A molecular model of the homoleptic cage also reveals another potential problem; [Pd 2 (E-1) 4 ] 4+ likely has a pronounced twisted conformation (Figure 2b), a consequence of the nonparallel coordination vectors of E-1, similar to a related cage. 16It has previously been shown that guest binding inside non-twisted, D 4h symmetric Pd 2 L 4 cages is facilitated by the formation of hydrogen bonds with the two sets of four polarized orthopyridyl protons that point directly into the cavity. 31These interactions also drive Michael addition catalysis by stabilizing the deprotonated nucleophile and co-binding the electrophile to reduce the entropy of activation. 33In the case of [Pd 2 (E-1) 4 ] 4+ , we attribute the lack of host-guest chemistry and catalysis to the twisted, propeller-like conformation of the Pd(pyridyl) 4 units.This twisting perturbs the convergent cavity-directed hydrogenbond donor atoms, hindering the formation of favorable electrostatic interactions that infer catalytic activity.Looking at an alternative cage design and to address the twisted conformation that could hinder catalysis, we targeted a heteroleptic system combining rigid ligand 2 with E-1.This combination of ligands was selected for their shape complementarity. 35On its own, ligand 2 is reported 34 to react with palladium(II) ions to form a [Pd 3 (2) 6 ] 6+ double-walled triangle in acetonitrile and a [Pd 4 (2) 8 ] 8+ double-walled square in DMSO and never a [Pd 2 (2) 4 ] 4+ dimer, which is also our observation (Figure 2aiv, Supporting Information S8 and S9). 34hen one equivalent of each of ligand E-1, ligand 2, and [Pd(MeCN) 4 ](BF 4 ) 2 are combined in DMSO-d 6 a single new species is formed within 10 min at room temperature (Figure 2aiii, Supporting Information S10).An identical result is obtained if the homoleptic cages [Pd 2 (E-1) 4 ](BF 4 ) 4 and [Pd 4 (2) 8 ](BF 4 ) 8 are combined in DMSO-d 6 and heated with a heat gun for 5 min (see Supporting Information S10.1), indicating that the heteroleptic cage is the thermodynamic product.Multinuclear NMR experiments (Supporting Information S10.2) confirm that the heteroleptic cage is symmetrical with a single environment for each of the ligands, E-1 and 2. Interligand ROESY interactions indicate that both ligands are coordinated to the same metal ion (Supporting Information S10.2).Diffusion NMR experiments in DMSO-d 6 (see Supporting Information S13) gave hydrodynamic radii in line with expectations: [Pd 2 (E-1) 2 (2) 2 ] 4+ (9.1 nm), smaller than that of the homoleptic [Pd 2 (E-1) 4 ] 4+ (10.5 nm) and the macrocyclic [Pd 4 (2) 8 ] 8+ (9.9 nm).ESI-MS confirmed the composition as [Pd 2 (1) 2 (2) 2 ] 4+ with a series of cations with isotope patterns corresponding to sequential loss of BF 4 anions (Supporting Information S10.3).
Finally, a single crystal suitable for X-ray diffraction unambiguously confirmed the heteroleptic species as cis-[Pd 2 (E-1) 2 (2) 2 ](BF 4 ) 4 (Figure 2c, CCDC: 2343886, Supporting Information S10.4), in line with expectations from shape complementarity prediction. 35The cage has a Pd•••Pd separation of 9.92 Å and has a BF 4 anion inside the cavity, confirming its ability to bind guests.The pyridyl units are not significantly twisted (angles between the trans pyridyl rings range from 16 to 22°), and therefore the ortho-pyridyl hydrogen-bond donors project into the cavity for optimal guest binding and catalysis.
Turning to the equivalent BAr F salt, we then tested whether the heteroleptic cage [Pd 2 (E-1) 2 (2) 2 ](BAr F ) 4 could be formed and, in particular, whether it could be generated from the rearrangement of the two homoleptic structures.Homoleptic structures [Pd 2 (E-1) 4 ](BAr F ) 4 and [Pd 3 (2) 6 ](BAr F ) 6 / [Pd 4 (2) 8 ](BAr F ) 8 can be assembled with [Pd(Py*) 4 ](BAr F ) 2 in CD 3 CN, and were characterized by NMR (Supporting Information S6.4,S9.4) and ESI-MS data (Supporting Information S6.5, S9.5).When a 1:1 mixture of these two homoleptic cages was combined in CD 3 CN and heated at 50 °C for 30 min, the heteroleptic cage was formed quantitatively (Supporting Information S10.5).This indicates that swapping from BF 4 to BAr F counteranions does not lead to problems with kinetic trapping.However, the solvent that is optimal for catalysis�dichloromethane�is poorly coordinating and therefore does not promote the rapid ligand exchange required for cage switching.We found that a solvent mixture of 11:1 CD 2 Cl 2 /CD 3 CN, was a good compromise to maximize the host-guest chemistry while providing some coordinating properties to facilitate cage rearrangement (see Supporting Information S14, S15).Using these mixed solvent conditions, we investigated the binding of methyl vinyl ketone and benzoyl nitromethane within [Pd 2 (E-1) 2 (2) 2 ](BAr F ) 4 (Supporting Information S17) as well as the ability of [Pd 2 (E-1) 2 (2) 2 ]-(BAr F ) 4 to catalyze the Michael addition reaction between these two substrates (Supporting Information S18). 36When methyl vinyl ketone is added to a sample of the heteroleptic cage, only minor shifts were observed in the 1 H NMR peaks of the cage, with no substantial changes when 18-crown-6 is also added (see Supporting Information 17.1).By contrast when benzoyl nitromethane is added a new set of deshielded signals are observed that correspond to the cage with bound deprotonated benzoyl nitromethane.These signals increase in intensity once 18-crown-6 is added (see Supporting Information 17.2).The signals that show the biggest difference in chemical shift compared to the "empty cage" correspond to the protons adjacent to the pyridyl nitrogen of the nonswitchable ligand 2, and of the inwardly directed ortho-pyridyl CH of the photoswitchable ligand 1.These chemical shift differences are in line with those observed from simple Pd 2 L 4 cages 1e,5,31 and are consistent with the binding of the substrates inside the cage through CH•••O hydrogen bonds (Figure 4a). 33We have also found that triflate binds tightly inside the cage, similarly evidenced by the appearance of a second set of cage signals with deshielded inward-facing proton resonances (Supporting Information S17.4).
When both substrates are present, this clear indication of substrate binding is accompanied by substrate consumption and the generation of the Michael addition product, as monitored using 1 H NMR spectroscopy (Figure 4, Supporting Information S18).An 11% catalyst loading of heteroleptic species [Pd 2 (E-1) 2 (2) 2 ](BAr F ) 4 converted 24% of the benzoyl nitromethane to the Michael addition product in 10 h.
Having shown that we can reversibly switch the heteroleptic cage with light and that it is also an active catalyst, it was time for photoswitchable catalysis!The [Pd 2 (E-1) 2 (2) 2 ](BAr F ) 4 cage was successfully disassembled (530 nm light for 10 min) and reassembled (405 nm light for 5 min) in 11:1 CD 2 Cl 2 /CD 3 CN   2) 2 ](BAr F ) 4 with a 530 nm LED before the substrates are added (Figure 5b, Supporting Information S19.3).The reaction can then be activated at will by irradiating the sample with a 405 nm LED.The long thermal half-life of the photoswitch ensures that the cage remains in the state it is programmed after the irradiation is stopped.The responsiveness of the system to visible light demonstrates that using a molecular photoswitch to control self-assembly can lead to excellent control of the chemical reactivity.We also show that the system can be subjected to at least five cycles of photoswitching without any effect on catalytic performance (Supporting Information S19.4).
(see Supporting Information S19), showing similar behavior to the BF 4 salt in DMSO-d 6 (Figure 3). 31Benzoyl nitromethane, methyl vinyl ketone, and 18-crown-6 were added to the sample, and the reaction was monitored using 1 H NMR spectroscopy (Figure 5a), showing the cage was catalytically active.Next, the sample was irradiated with a 530 nm LED for 10 min, which resulted in a 10-fold decrease in the rate of product formation as the cage was disassembled.The reaction was then reactivated by irradiating with a 405 nm LED for 5 min.Following this reactivation, the rate of product formation was almost identical to that prior to 530 nm irradiation, showing that the photoswitching is completely reversible.
In conclusion, we have shown the first example of photoswitchable catalysis within a self-assembled molecular cage.The mechanism of catalysis relies on electrostatic interactions within the cavity, which is possible only in the heteroleptic cage with a cavity preorganized for guest binding.The catalysis can be switched ON and OFF with visible light (405 nm and 530 nm, respectively) and is entirely reversible.Controlling the catalytic activity of self-assembled cavities with nondestructive visible light is a new method for directing chemical reactions.Combining photoswitchable ligands with facile ligand-exchange reactions allows a system to be driven toward assemblies composed of different components, with programmable stoichiometries, shapes, affinities, and now catalytic functions.We anticipate future examples could include different selfassembled cages each capable of catalyzing different reactions, allowing more complex multistep reactions to be performed simply by using visible light.

Figure 5 .
Figure 5. Photoswitchable catalysis by heteroleptic cage [Pd 2 (E-1) 2 (2) 2 ](BAr F ) 4 in CD 2 Cl 2 :CD 3 CN 11:1 monitored by 1 H NMR spectroscopy.a) ON/OFF/ON cycle and b) OFF/ON/OFF cycle.Reaction conditions: benzoyl nitromethane (17 mM), methyl vinyl ketone (33 mM), and 18-crown-6 (12 mM), [Pd] = 2.0 mM.Irradiation by 405 nm (5 min) and 530 nm (10 min) outside of the NMR instrument, with colored bars representing the time between NMR measurements; see Supporting Information S19 for details.The system can also be kept dormant by first irradiating [Pd 2 (E-1) 2 (2) 2 ](BAr F ) 4 with a 530 nm LED before the substrates are added (Figure5b, Supporting Information S19.3).The reaction can then be activated at will by irradiating the sample with a 405 nm LED.The long thermal half-life of the photoswitch ensures that the cage remains in the state it is programmed after the irradiation is stopped.The responsiveness of the system to visible light demonstrates that using a molecular photoswitch to control self-assembly can lead to excellent control of the chemical reactivity.We also show that the system can be subjected to at least five cycles of photoswitching without any effect on catalytic performance (Supporting Information S19.4).