Traffic Lights for Catalysis: Stimuli-Responsive Molecular and Extended Catalytic Systems

2.1.1 Changes of catalyst geometric parameters as a function of light

Some of the first approaches for the regulation of catalytic performance were realized through functionalization of molecular organocatalysts with photoresponsive moieties (Schemes 1 and 2) which allowed for tailoring variables such as reaction selectivity, yield, or rate. One of the mechanisms for such modulation of catalytic performance relies on alteration of either geometric parameters or physicochemical properties of the photoswitch-catalyst pair. For instance, photoswitches for which isomerization is accompanied with large structural transformations (e.g., azobenzene, hydrazone, or rotaxane derivatives), have been utilized to selectively block or expose catalytically active sites using an excitation wavelength as a “knob” for catalytic activity regulation.^{1, 50, 51, 67-69} This particular approach relies on rational design of a photoresponsive unit which undergoes, for instance, E-to-Z isomerization (e.g., azobenzene or hydrazone derivatives).^{1, 50, 51, 69} In this case, one photoisomer blocks the catalytically active site (Scheme 1) while the other provides open access for the substrate to reach this site. Thus, the photoswitchable moiety behaves similar to a gate, allowing or preventing catalyst-substrate interactions. Many of these studies employ azobenzene-based catalysts since the cis-to-trans isomerization could modulate catalytic processes by regulating the spatial separation of substrates and catalysts. For example, incorporation of azobenzene within crown ethers has been shown to yield photoswitchable phase transfer catalysts (PTCs).⁵⁰

Overall, crown ether derivatives have been used as PTCs in many organic reactions because of their ability to solubilize ionic species in organic solvents through coordination of the ions within the crown ether structure.^70-72 For example, chelation of the cationic species of a salt by crown ethers stabilizes the cation and enhances the solubility of the salt. Consequently, the anionic species can exhibit increased basicity or nucleophilicity since it is no longer stabilized by the counterion. As a result, reaction yield can be altered by addition of PTCs.⁶⁹ For instance, azobenzene-functionalized binaphthyl crown ether (ABCE) was utilized to promote alkylation of glycine Schiff bases by the reaction with potassium hydroxide in toluene (Figure 1 and Table 1, #2).⁵⁰ Irradiation of the ABCE solution by 365 nm caused photoisomerization of trans-azobenzene to cis-azobenzene, which effectively opened the aperture of the crown ether as shown in Figure 1. In the “open aperture” state, the crown-ether-based PTC was able to facilitate the base-catalyzed alkylation of glycine, resulting in high yield because potassium cations were efficiently solubilized through their coordination within the aperture of the crown ether. In contrast, switching from cis- to trans-azobenzene reduced the aperture size of the crown ether, resulting in a drastic decrease of reaction yield.⁵⁰ In addition, cis-to-trans photoisomerization resulted in a significant decrease in enantioselectivity, indicating that the reversible change in crown ether geometry could be used to control enantiomeric ratios of the products. Although more detailed studies are necessary to uncover mechanistic details, especially those pertaining to tunable enantioselectivity, this work clearly demonstrates that integration of the azobenzene-based photoswitches offers noninvasive control of both reaction yield and the enantiomeric ratio of the product mixture.⁵⁰

Table 1. Influence of stimuli-responsive moieties on catalytic outcomes.

A conceptually different approach to switchable catalysis based on coupling of azobenzene photoswitches with crown ethers has been demonstrated using two 15-crown-5 ether moieties connected by a central azobenzene unit (Table 1, #10).⁶⁹ In contrast to the previously described study, which utilized azobenzene photoisomerization to change the size of the crown ether aperture, an alternative approach relied on azobenzene trans-to-cis transformations to modulate the distance between the two crown ether groups.⁶⁹ In this case, the cis-azobenzene isomer brought the two crown ether groups in proximity to promote strong binding of cations by both crown ethers simultaneously. This approach was employed to probe base-catalyzed ring-opening polymerization of l-lactides in which the acetate anions were activated by binding of the potassium cations within the cis-azobenzene-containing crown ether catalysts. The authors suggest that electrostatic interactions between the potassium and acetate ions when the potassium cation was not chelated by the crown ether (i.e., in the presence of cis-azobenzene) could cause poor chain end activation. As a result, irradiation with 320-nm light, leading to photoisomerization from trans- to cis-azobenzene, coincided with a nine-fold increase of l-lactide polymerization rate.⁶⁹

Azobenzene-based photoswitches have also been employed for the design of biphasic organocatalysts which promote solution-phase reactions and can be precipitated from solution upon photoisomerization to facilitate separation of catalysts from the product mixture.¹ For instance, Han and co-workers developed a series of azobenzene-based organocatalysts bearing catalytically active tertiary amine groups along with amide groups to facilitate hydrogen bonding interactions. Notably, the E-isomers of these catalysts exhibited planar structures with significant π-π stacking interactions, while photoisomerization to the Z-isomers caused a loss of crystalline packing resulting in a phase transition. That is, E-to-Z photoisomerization coincided with a phase transition from crystalline solids to liquids. As a result, this concept was applied for facile recycling of homogeneous catalysts. For example, a reaction mixture consisting of the soluble Z-enriched catalyst (formed upon irradiation with a 340-nm wavelength) led to higher reaction rates than those of the insoluble E-enriched catalyst by an order of magnitude for Michael addition reactions. Moreover, a 2.4-fold increase in product yield was detected for the Z-isomer. After reaction completion, 87 % of the catalyst could be recovered by irradiation at 430 nm to induce Z-to-E photoisomerization. Upon photoisomerization, the catalyst precipitated from the product mixture and could be recovered using filtration and re-used in at least three subsequent cycles.¹ Thus, integration of azobenzene photoswitches in organocatalysts is an avenue not only for controlling reaction rates and yields but also achieving homogeneous catalyst recycling.

Similar to azobenzene, hydrazone derivatives, which undergo E-to-Z isomerization around a central imine bond upon exposure to UV irradiation, also exhibit significant structural changes upon isomerization, and therefore, can be utilized to block or expose catalytically active sites.⁵¹ For example, bifunctional cinchona alkaloid-squaramide catalysts functionalized with hydrazone groups were used to control the enantiomeric excess of several conjugate addition reactions through irradiation with light (Figure 2). As shown in Figure 2, irradiation with a 395-nm excitation wavelength resulted in formation of the Z-catalyst, which was deactivated by intramolecular hydrogen bonding. In this case, the intramolecular hydrogen bonding caused the hydrazone group to block the active site and prevent binding of the substrate to the catalyst. In contrast, the formation of the E-catalyst exposed the active site to promote substrate binding and activation toward nucleophilic attack. Remarkably, the reaction rates for these catalysts in the “on” state (E-hydrazone) were 16 times higher than those of the “off” state (Z-hydrazone).⁵¹ This example illustrates that coupling of hydrazone derivatives with organocatalysts could be used for tuning the reaction rate through selective blocking of the active sites, i.e., preventing hydrogen bonding between the substrate and catalyst.⁵¹

Although the majority of studies in the area of stimuli-responsive photocatalysts utilize azobenzene or hydrazone derivatives as described above,^{1, 50, 51, 69} other classes of relatively less explored photoswitches offer other avenues for switchable catalysis.^{52, 73-75} For example, some overcrowded alkenes (i.e., tetrasubstituted alkenes with bulky substituents such as 9-(6-bromo-3,5,8-trimethylthiochroman-4-ylidene)-N,N-dimethyl-9H-fluoren-2-amine) exhibit photoswitchable behavior similar to E-to-Z isomerism.⁷³ A photoresponsive tetrasubstituted alkene core has been used to build molecular motors for controlling the catalytic activity in the Michael reaction between (E)-3-bromo-β-nitrostyrene and 2,4-pentanedione.⁷³ In the described system, formation of the E-photoisomer resulted in 60 % conversion after 18 h, whereas the Z-isomer resulted in only 19 % conversion under the same conditions, showing the potential for crowded alkenes to behave as “on/off” switchable catalysts. The other less explored stimuli-responsive molecules which could be utilized to block and expose active sites, include rotaxane and helicene derivatives.^{67, 76} They provide unique opportunities to expand the library of switchable catalysts because their conformational and configurational changes induced by irradiation can provide the control of chiral space.⁷³ For instance, the position of rotaxane macrocycles around a catalytically active core could be used to regulate the facial binding of substrates as a function of an excitation wavelength.⁶⁷ In a similar direction, helicene derivatives functionalized with catalytically active 4-N-methylaminopyridine groups have been shown to facilitate asymmetric Steglich rearrangements of o-carboxylazlactones.⁷⁶

2.1.2 Changes in electronic properties of catalysts as a function of light

In contrast to the above discussion focusing on catalyst geometrical changes, a conceptually different approach to photoswitchable organocatalysis could be achieved through modulation of the catalyst electronic structure.^{52, 54, 74, 75, 77} In this direction, diarylethene derivatives (Scheme 2) have played a significant role.^{52, 74, 78} Their photoisomerization is based on reversible bond formation leading to alternation between conjugated and non-conjugated forms.^{52, 74, 78} For instance, photocyclization of diarylethene derivatives is accompanied by changes in π-conjugation of the photoswitch backbone and, in some cases, release of substituent groups.⁵² Indeed, Shirinian and co-workers have shown the ability of a series of halogenated diarylethene derivatives to release bromine cations upon photoisomerization (Figure 3 and Table 1, #20). Such photolysis could be used to carry out bromination reactions in good yields as well as to catalyze the double Friedel–Crafts addition of indoles to aldehydes.⁵² In the latter case, catalysis of the Friedel–Crafts reaction occurred through the Lewis acidic bromine cation which could only be formed through photoisomerization of the halogenated diarylethene-based photoswitch. As a result, the catalytic reaction could be initiated using UV irradiation.⁵²

A different approach, based on selective changes in charge delocalization involving diarylethene derivatives, takes advantage of light-induced changes in π-conjugation to alter the activity of a catalytic site.^{74, 77} According to computational analysis, diarylethene derivatives have the potential to modulate the behavior of frustrated Lewis pair (FLP) catalysts through charge delocalization.⁷⁴ It was predicted, for instance, that functionalization of a diarylethene core with amine and boronic ester groups could result in generation of an FLP capable of catalyzing hydrogen recombination reactions.⁷⁴ Without exposure to an appropriate excitation wavelength, the diarylethene core is “open,” so there is minimal electronic communication between the acid and base moieties of the FLP catalyst. In contrast, exposure to UV irradiation causes photoisomerization to the colored (i.e., fully conjugated) isomer, which allows for electronic delocalization throughout the catalyst. This delocalization promotes a resonance form in which a negative charge is formally generated at the Lewis base, and a positive charge is formally located at the Lewis acid. As a result, the theoretically predicted proton affinity of the amine group of the closed diarylethene-based catalyst is significantly higher than that of the open (colorless) form.⁷⁴ Likewise, the theoretical hydride affinity of the Lewis acidic boron atom is increased for the closed isomer.⁷⁴

Rather than changing the electronic structure of FLPs through the photocyclization reaction of diarylethene derivatives, another avenue is to use the photoswitchable molecules which can undergo charge separation upon photoisomerization, leading to modulation of electron density, as well as proton or hydride affinity of an active site.⁷⁵ For example, functionalization of borane-phosphine or borane-amine FLPs with the dihydroazulene-vinylheptafulvene photoswitch has allowed for light-induced changes in H₂ activation abilities.⁷⁵

Tailoring organocatalyst performance could also be accomplished via light-induced changes in its anion binding affinity.⁷⁹ For instance, Venkataramani and co-workers probed tunable ion binding using tripodal 1,3,5-tris-functionalized benzenes bearing triazole-containing azobenzene photoswitches, which could undergo trans-to-cis photoisomerization under UV excitation.⁷⁹ In particular, the binding affinity of triazole functionalities to chloride anions could be reduced due to formation of the cis-azobenzene form upon irradiation with a 365-nm excitation wavelength.⁷⁹ In contrast, the chloride anion binding affinity was increased by photoisomerization of the catalyst to the trans-azobenzene isomer under 405-nm irradiation.⁷⁹ In fact, based on isothermal titration calorimetry experiments performed on the trans- and cis-azobenzene forms of the catalyst, the dissociation constant associated with binding of chloride ions by the triazole functionalities was two times higher for the cis-azobenzene-based catalyst than the trans-azobenzene-based catalyst, which is in line with a reduced binding affinity for the cis-catalyst.⁷⁹ These findings were applied toward probing tritylation reactions (using trityl chloride) in which the photoswitchable catalyst promoted reaction rates by strong binding of the chloride counterion upon exposure to a 405-nm excitation wavelength. As expected based on the determined dissociation constants, the isolated yields of tritylation reactions, carried out in the presence of the trans-catalyst, were approximately twice of those carried out in the presence of the cis-catalyst. As a result, the reaction yields could be reversibly controlled through photoswitch isomerization.⁷⁹

As previously described, hydrazone derivatives could be used to modulate geometric parameters in a catalytic system, but they have also been used to tune the barrier for proton transfer steps through intramolecular hydrogen bonding in either the E- or Z-photoisomer.⁵³ In some cases, intramolecular hydrogen bonding between the amine proton of the hydrazone core and a substituent acting as a hydrogen bond acceptor can be induced by photoisomerization, resulting in changes in pK_a of the amine proton.⁸⁰ Thus, hydrazone derivatives present an opportunity to modulate pH of a reaction medium without introducing additional acids or bases.⁸¹ Even slight changes in acidity or basicity can be used to promote product formation in pH-sensitive reactions (e.g., transesterification).⁵³ On the example of the transesterification of boronate esters with diols, Kalow and co-workers utilized a hydrazone photoswitch as an internal base catalyst to facilitate the rate-determining proton transfer mechanistic step (Figure 4a).⁵³ In these studies, E-to-Z isomerization of the hydrazone functionality resulted in breaking of intramolecular hydrogen bonding interactions between the hydrazone amide proton and proximal nitrogen atom. As a result, hydrogen bonding between the boronate ester proton and nitrogen was possible, which promoted deprotonation of the boronate ester and subsequent transesterification (Figure 4).

2.1.3 Photoacids and photobases

While several well-studied classes of photochromic molecules have been successfully applied in photoswitchable catalysis as described above, a different direction involves the use of photoacids and photobases as means to achieve light-controlled organic transformations.^{54, 82} For example, these organic photoacids and bases exhibit significant changes in pK_a depending on irradiation conditions. Unlike the previously described systems in which the photoswitch undergoes either a geometric or electronic change that perturbs the active catalyst, photoacids and photobases produce a catalytic amount of an acid or base after exposure to a particular excitation wavelength (Figure 4bb).⁵⁴ For example, several studies concerning quinoline photobasicity have shown an increase in approximately 10 pK_a units for quinoline derivatives exposed to UV light.^{54, 83} In particular, the excited states of the studied quinoline photobases are of sufficient energy to facilitate proton transfer from the reaction media (e.g., water or alcohols), which yields the conjugate base upon irradiation with an appropriate excitation wavelength (Figure 4b). The mechanism for generation of conjugate bases proceeds through abstraction of a proton from a hydrogen-bonded donor.⁵⁴

While the majority of reported photoacids and bases are of the Brønsted type as described above, development of light-activated Lewis acids and bases is of particular interest due to their ability to initiate polymerizations of relatively unreactive monomers, as well as promote Diels-Alder and Friedel-Crafts reactions, among other synthetically useful transformations.⁸² For example, functionalization of a diarylethene-containing photoswitch with a triflate naphthalene moiety allowed for ejection of the triflate anion upon exposure to UV irradiation. As a result, a carbocation was formed at the photoactive core which was stabilized by the extended π-conjugation of the colored (i.e., UV-activated) diarylethene photoisomer. The stabilized carbocation was then evaluated in its ability to promote the bulk polymerization of epoxy monomers and the Mukaiyama-Aldol reaction between benzaldehyde and silyl enolate derivatives.⁸² Presence of an insoluble polymer containing end-terminal groups corresponding to the photoacid after UV excitation showed that the prepared catalyst was active toward the selected cationic polymerization reactions. As a control experiment, the described Mukaiyama-Aldol reactions were unsuccessful without UV irradiation, and addition of previously studied Brønsted photoacids could not initiate product formation.⁸² Thus, this approach demonstrates the successful preparation of a Lewis photoacid with switchable catalytic activity.⁸²

Based on the mentioned studies above, it is clear that the light-induced alteration of geometric parameters of photoswitch-catalyst pairs has the potential to modulate reaction rates,^{1, 51, 69, 79} yield,^{1, 50, 52, 79} and enantiomeric ratios for chiral products.⁷³ Rational design of photoswitchable catalysts in this area has not only resulted in “on/off” catalysts, but also in systems for which the possible outcomes are no longer binary.^{63, 84-86} That is, switchable catalysts could serve to promote different reaction mechanisms with regio- or enantioselectivity depending on the state of the photoswitch.^{73, 87, 88}

Despite significant progress toward such systems, development of fully reversible switchable catalysts capable of promoting variable product formation remains a challenging task. For instance, instantaneous temporal control of organic transformations would allow for precise tuning of product formation; however, many of the reported photoswitchable organocatalysts with large geometric changes possess relatively slow photoisomerization kinetics or utilize a secondary stimulus for the reverse isomerization process (e.g., heat or acid).^89-93 At the same time, photoswitchable organocatalysts relying on changes in the electronic properties of the photoswitch-catalyst pair have received well-deserved attention in recent years,^94-96 but the development of a true “on/off” system, which fully suppresses reaction progress in the “off” state, has yet to be achieved. The latter issue is associated with the fact that most catalysts in the “off” state frequently possess some residual catalytic activity. This residual activity coupled with low photostationary states (PSS) for some systems (i.e., a percentage of the catalyst remains as one photoisomer despite treatment with an appropriate excitation wavelength) minimizes the change in reaction outcome upon photoisomerization.⁵¹ Addressing the mentioned challenges associated with switchable organocatalysis relies on performing deep fundamental photophysical and mechanistic studies. Thus, the combined efforts of the synthetic organic and photophysical chemistry communities are necessary to advance the development of new photoswitch-catalyst pairs with high PSS ratios and near-complete reversal of catalytic activity upon isomerization of the stimuli-responsive molecules.

2.2.1 Changes in geometric parameters of catalysts as a function of light

One area in which modulation of an organometallic complex's coordination geometry has been successfully used to alter catalytic activity is the noninvasive control of the products of metathesis and polymerization reactions.⁹⁹ Changes in the catalyst's coordination geometry are reported to affect not only the reaction rate, but also the stereo- and regioselectivity, as well as polymer tacticity, directly affecting bulk material properties.⁹⁹ For instance, a 1^st generation Grubbs catalyst modified with an azobenzene-containing cyclic(alkyl)(amino)carbene ligand for olefin metathesis exhibited switchable catalytic activity based on azobenzene blocking or exposing the active site (Figure 5 and Table 1, #3).⁵⁵ In particular, the formation of trans-azobenzene exposed the active site and promoted the ring-closing metathesis reaction with >99 % conversion of diethyl diallyl malonate when exposed to visible light (Figure 5).⁵⁵ In contrast, irradiation with UV light caused photoisomerization to the cis-azobenzene ligand which blocked the active site, and as a result, less than 1 % conversion was detected when the reaction was performed under UV excitation. Remarkably, this system represents near complete control of catalytic activity because the “off” form of the catalyst (e.g., cis-azobenzene) possesses almost no activity toward product formation.⁵⁵ Control experiments involving the ruthenium-based catalyst without the azobenzene functionality confirmed that the catalytic activity was not altered by the irradiation conditions alone, but rather could be attributed to photoisomerization of the ligand. In order to determine whether the observed changes in catalytic activity originated from the steric hindrance imposed by the cis-azobenzene photoswitch or changes in the electronic properties of the carbene moiety, a model rhodium carbonyl complex bearing the same azobenzene-carbene ligand was probed by Fourier transform infrared (FTIR) spectroscopy. Based on the FTIR stretching frequencies for the carbonyl groups, the authors calculated the Tolman electronic parameter (TEP) for the complex before and after UV irradiation. Since the TEP value was unaffected by azobenzene photoisomerization, it was concluded that changes in steric bulk at the metal center must alter the catalytic activity.⁵⁵ This example provokes further investigation in the area of photochromic organometallic compounds since it clearly demonstrates the precise temporal control over reaction progress that could be achieved through rationally designed photochromic organometallic complexes.

Not only could photoswitchable ligands be utilized to block or expose active sites, but they could also be used to control the distance between two metal centers involved in cooperative catalysis.⁵⁶ In Figure 6, two titanium(salen) complexes bridged by an azobenzene group could facilitate asymmetric sulfoxidation reactions only when the Ti^IV centers were in close proximity to one another.⁵⁶ In particular, the true catalytically active species was a μ-oxo-bridged [{(salen)Ti(μ-O)}₂], which could only be achieved after the trans-to-cis photoisomerization of the azobenzene ligand took place. Thus, irradiation with UV light to form cis-azobenzene promoted asymmetric sulfoxidation of a variety of prochiral sulfide substrates through the control of the spatial arrangement of two Ti^IV catalysts. In contrast, photoisomerization to trans-azobenzene, leading to a larger distance between the catalytic centers, decreased catalytic activity. For instance, up to 98 % conversion with 93 % chemoselectivity and >99 % enantiomeric excess was detected under UV irradiation, while only 68 % conversion, 86 % chemoselectivity, and 78 % enantiomeric excess were obtained for the same reaction conducted in the dark.⁵⁶

Using a similar strategy, Marinetti and co-workers have synthesized a binuclear gold(I) catalyst, Au₂Cl₂L (L=1,2-bis(3-(diphenylphosphino)-2,6-difluorophenyl)diazene), possessing a photochromic azobenzene functionality.¹⁰¹ After structural confirmation of both the cis- and trans-azobenzene-containing complexes by single-crystal X-ray diffraction, their catalytic activity was evaluated on the example of intramolecular hydroamination of N-alkenyl urea.¹⁰¹ The spectroscopic analysis revealed an increased reaction rate for the cis-azobenzene-containing complex as compared to the trans-azobenzene-containing catalyst. For instance, 91 % conversion was detected after 44 minutes for the cis-complex, while only 70 % conversion was detected for the trans-complex under the same reaction conditions. Further, the kinetic studies for the trans-azobenzene catalyst were in line with control experiments carried out with an analogous mononuclear complex. Thus, these two examples illustrate that irradiation with UV light, resulting in formation of cis-azobenzene, enhanced catalytic activity through cooperation between the two metal centers, whereas the presence of trans-azobenzene minimized cooperativity by increasing the distance between the active sites.

Implementation of azobenzene for controlling the spatial arrangement of two metal centers has also been successfully applied in anion abstraction catalysis.¹⁰⁶ An azobenzene core functionalized with tellurium-containing arms was used as a basis for the catalyst development.¹⁰⁶ While the electrophilic tellurium centers exhibited strong binding of halide counterions, binding strength was modulated by photoisomerization of the azobenzene core.¹⁰⁶ In particular, formation of cis-azobenzene through irradiation at 590 nm brought the tellurium centers in close proximity and increased the binding strength for halides. In contrast, the binding affinity was reduced for the trans-azobenzene-based catalyst. This strategy has been applied for tailoring the catalytic activity of Friedel-Crafts alkylation reactions occurring between benzhydryl chloride and 1,3,5-trimethoxybenzene.¹⁰⁶ Indeed, reaction yield was improved from 9 % to 58 % under continuous irradiation with a 590-nm excitation wavelength, which promotes formation of the cis-azobenzene-based catalyst and led to halide abstraction from the substrate. In general, this work represents a relatively novel direction in switchable catalysis built upon synergy of photochromic moieties with chalcogen bonding motifs.

In addition to selective blocking of active sites or alternation of the spatial separation of catalytic metal centers, switchable catalysis can also be achieved through changes in the coordination environment accompanied with ligand photoisomerization. For example, an oxazolinyl-2-azopyridine ligand coordinated to rare earth metal centers (M=La or Eu) could switch between bi- and tridentate coordination arrangements.¹⁰⁰ This system overcomes a unique challenge in that it relies on coordination of the azobenzene moiety through the nitrogen atom while maintaining efficient photoisomerization around the N=N bond. Coordination of the photoswitchable ligand to the metal center occurred through three nitrogen atoms, one from the azobenzene, one from a pyridine, and one from an oxazoline functionality. In particular, the trans-azobenzene allows for coordination of all three nitrogen atoms to a lanthanum or europium center, while irradiation with 365-nm light promoted formation of cis-azobenzene and limited the coordination of the ligand to only the pyridine and oxazoline nitrogen atoms (i.e., switching from a tridentate to bidentate ligand). Such changes in coordination modes have been associated with changes in the ratio of enantiomers present in the product mixture for intermolecular cyclization of sulfonamide and aldehyde derivatives.¹⁰⁰ Under optimized conditions, the enantiomeric excess of the reaction could be switched from 28 % to 76 % excess of the R-enantiomer based on excitation wavelength (UV or dark, respectively).¹⁰⁰ Thus, the trans-azobenzene ligand facilitated a tridentate binding motif which resulted in higher enantioinduction. Control reactions involving non-photochromic bidentate or tridentate oxazoline ligands revealed similar trends to that of the tridentate ligands, which resulted in higher enantioselectivity than in the case of the bidentate ligands. Further mechanistic studies are necessary to shed light on the basis for the enantioselectivity of each photoisomer, and therefore, they are critical to reveal principles for the design of new switchable catalysts.

2.2.2 Changes in electronic properties of catalysts as a function of light

Rather than geometric control of reaction rate or yield, a different approach to switchable catalysis can be achieved through modulation of the electronic properties of the metal center for catalytically active organometallic complexes.^{57-60, 103} For example, change in a metal center's oxidation state could significantly alter the catalytic activity. Subtle shifts in ligand field strength, as well as LMCT, MLCT, or PET could be utilized to affect the metal oxidation states.^{78, 109-111} Thus, in this section, we will outline the ways in which photoswitchable ligands can be used to incite changes in the electronic properties of corresponding organometallic complexes and, as a result, catalytic activity through the mentioned processes.

One approach taken by Bogliotti and co-workers relies on “photo-uncaging,” or the light-induced dissociation of ligands from metal complexes.⁵⁷ In the mentioned studies, an azobenzene-based ligand caused dissociation of a phosphine group from a ruthenium complex based on changes in steric hindrance due to azobenzene photoisomerization (Figure 7). In particular, exposure to visible light promoted the trans-form of the azobenzene ligand and resulted in ejection of the phosphine ligand. While the ruthenium complex by itself was not catalytically active, the presence of free phosphine released by the ruthenium complex promoted reduction of a palladium(II) pre-catalyst to an active palladium(0) metal center. Consequently, the palladium(0) complex exhibited catalytic activity toward cross-coupling (Sonogashira) reactions between aryl iodides and trimethylsilyl acetylene. For instance, the reaction yield could be improved from 10 % to 92 % by exposure of the reaction mixture to visible light irradiation, leading to release the phosphine group. Thus, reaction rate could be tuned as a function of irradiation time since formation of the active catalyst was directly associated with the light-induced dissociation of the phosphine group.⁵⁷ Although it is a significant step toward noninvasive control of critical synthetic transformations, this approach toward switchable catalysis is irreversible. For instance, the reduction of palladium(II) by the free phosphine cannot be reversed even upon photoisomerization to reform the cis-azobenzene ligand. Despite its irreversibility, “photo-uncaging” offers the advantage of being applicable to a wide variety of catalysts without requiring individualized photoswitches (i.e., novel photoswitchable ligands for each organometallic catalyst).

Tunable metal oxidation states are an attractive direction for switchable catalysis, but, in many cases, it is sufficient to alter the electron density of the metal center without utilizing complete oxidation or reduction to modulate catalytic activity. For example, the rate of industrially-relevant hydrosilylation reactions involving a platinum-based Karstedt's catalyst are sensitive to the electron density at the platinum metal center.⁵⁸ In particular, coordination of electron-deficient ligands can render the platinum metal center inactive toward hydrosilylation of alkenes. While high catalytic activity is generally desirable, the high activity of platinum catalysts (even in small amounts) toward hydrosilylation can cause premature curing of polymers on the industrial scale.^{58, 112} Thus, selective inhibition of these catalysts through binding of electron-deficient alkenes is one attractive approach to modulate catalytic behavior.⁵⁸ In this direction, reversible binding of electron-deficient ligands could alter the electron density of the catalyst and allow for tunable activity.⁵⁸ Branda and co-workers demonstrated this possibility through careful design of a diarylethene-based ligand functionalized with electron-withdrawing cyano groups (Figure 8).⁵⁸ In the open form of the diarylethene photoswitch, the cyano groups are located at an electron-deficient double bond which exhibited strong binding to the platinum center of Karstedt's catalyst and minimized catalytic activity. In contrast, irradiation with UV light caused photoisomerization to the closed form of diarylethene, which possesses a conjugated π-system that eliminates the electron-deficient alkene. As a result, the closed diarylethene photoswitch did not bind to the catalyst, preserving catalytic activity toward hydrosilylation reactions.⁵⁸ Based on ¹H NMR spectroscopic analysis of the reaction mixture, the authors concluded that the prepared diarylethene-based ligand was an effective inhibitor for the Karstedt's catalyst under visible light. However, the system suffered from a poor photostationary state under UV irradiation. That is, exposure to the appropriate excitation wavelength to induce photoisomerization to the closed form (i.e., without the electron deficient alkene) was not quantitative, resulting in a mixture of photoisomers and noticeable inhibition of the catalyst caused by the residual open photoisomer (Figure 8). Thus, this system achieves the desired inhibition of the catalyst but requires further optimization to obtain complete photoisomerization to the uninhibited form under UV excitation.

A different approach to utilizing the change in electronic structure associated with diarylethene photoisomerization was achieved by Wolf and co-workers by modifying a copper complex with a diarylethene-based ligand (Figure 9).⁵⁹ In this example, extension of conjugation throughout the ligand backbone upon isomerization from the open to closed diarylethene photoisomers resulted in decreased electron density at the copper metal center. As shown in Figure 9, the reduced electron density at the metal center resulted in decreased yields for hydroboration reactions conducted under UV irradiation in comparison with those conducted under visible light.⁵⁹ Thus, on the example of both hydrosilylation and hydroboration reactions, diarylethene-based ligands are a promising avenue for control of electron density of metal centers and the corresponding reactivity.

In a different example by Chen and co-workers, azobenzene-containing palladium and nickel catalysts were used to control the bulk material properties produced by ethylene polymerization and copolymerization reactions.¹⁰³ Preparation of azobenzene-functionalized palladium complexes for ethylene polymerization revealed a 12-fold increase in chain transfer rate under UV irradiation to promote trans-to-cis isomerization of azobenzene.¹⁰³ The authors attribute this increase in chain transfer rate to an increased electrophilicity of the palladium metal center as a result of azobenzene photoisomerization perturbing the electronic structure of the ligand backbone.¹⁰³ Further, the relative distance between the azobenzene functionalities and the metal center suggest that the changes in reactivity were not due to steric factors. Within the same series of studies, sandwich-type azobenzene-functionalized naphthalenyl amine palladium and nickel complexes were prepared which could affect both the electronic and steric properties of the metal center upon azobenzene photoisomerization. In this case, photoisomerization of the azobenzene moiety directly impacted the steric bulk at the metal center, which coincided with decrease of the catalytic activity under UV irradiation. This sharp difference in behavior between the sandwich complexes and those with azobenzene groups at the outside of the coordination sphere can be attributed to competing electronic and steric factors. These results are in line with commonly used α-diimine nickel catalysts, which exhibit increased polyethylene molecular weight and decreased branching density with increasing ligand steric bulk.¹⁰³ Thus, these studies show how design of photochromic ligands can alter the mechanism through which switchable catalysis is achieved (i.e., steric or electronic factors). For instance, changes in the electronic and steric properties of the transition metal complexes associated with azobenzene photoisomerization resulted in tunable polymer density, tensile strength, and molecular weight.¹⁰³ With this in mind, incorporation of photoswitches in polymerization catalysts is an avenue for creating a library of materials from the same starting reagents simply by varying irradiation parameters. Along these lines, Chen and co-workers have also probed ring-opening polymerization by tuning the electronic structure of photoresponsive salicylaldimine Zn^II complexes based on azobenzene photoisomerization.⁹⁷ In each of the described studies by the Chen group, substantial differences in catalytic activity could be achieved through reversible photoisomerization of a ligand.^{97, 103}

While the previous examples focused on a photoswitchable ligand altering the catalytic activity of a metal center, it is feasible to design systems which work in the alternative fashion, i.e., modulation of the catalytic activity of a photoswitchable molecule through coordination to a metal center. For instance, reversible coordination of a catalytically active ligand to a metal cation could perturb, for example, the Lewis basicity of the ligand and alter its chemical properties. In these systems, the metal center does not directly catalyze any organic transformations, but rather modulates the activity of an organic counterpart. One such example utilized an azopyridine photoswitchable ligand in combination with a nickel(II) porphyrin complex to achieve switchable catalytic activity toward nitroaldol reactions (Figure 10).⁶⁰ Dimethylaminopyridine (DMAP) is known to act as a basic catalyst for nitroaldol reactions and, therefore, changes in the basicity of the nitrogen lone pair in the pyridine group could affect catalytic activity. Functionalization of DMAP with an azobenzene photoresponsive derivative followed by complexation with Ni^II cations allowed for reversible coordination of DMAP to the Lewis-acidic metal center, resulting in tunable basicity of the DMAP catalyst.⁶⁰ In particular, irradiation with 530-nm light caused formation of cis-azobenzene and promoted coordination of DMAP to the metal center. Consequently, the catalytic activity was reduced by a factor of 2.2.⁶⁰ In contrast, irradiation with 435-nm light caused photoisomerization to trans-azobenzene, which promoted de-coordination of DMAP from the metal center and restored the catalytic activity to that of the free catalyst (i.e., uncoordinated DMAP).⁶⁰

In general, the area of photoswitchable organometallic catalysts is currently dominated by azobenzene-based ligands which can modulate the geometry around the metal center.^{55-57, 60, 97, 101, 103, 106} While these efforts have certainly opened the door for many exciting studies, the scope of reactions catalyzed by the prepared complexes could be expanded through in-depth mechanistic studies of the stimuli-responsive catalyst behavior and application of a broader variety of photoswitch motifs. Furthermore, switchable enantiodivergent catalysis (i.e., catalysts which can produce either enantiomer of a product based on photoisomerization of a ligand) could also be an attractive direction that requires rational design of photoresponsive ligands with specific geometric arrangements around a metal center to promote stereoselective synthesis.⁹⁹

2.3.1 Self-assembly of catalyst systems as a function of light

One of the strategies for controlling the catalytic activity of extended structures is to incorporate photoswitches for which isomerization is accompanied by significant changes in polarity (e.g., spiropyran, Scheme 2) to affect self-assembly processes. For example, assembly of micelles or nanotubes, which often depends on organization of amphiphilic molecules, could be controlled through switching between hydrophilic and hydrophobic forms of a photoresponsive molecule (e.g., switching from merocyanine to spiropyran).^{61, 127} Systems for which catalysis occurs only through the assembled supramolecular structure can, therefore, be modulated through isomerization of the photochromic fragment.¹²⁴ For example, Tan and co-workers achieved catalysis of asymmetric sulfoxidation reactions in aqueous media through the light-activated self-assembly of spiropyran-containing metallomicelles (Figure 11).⁶¹ In these studies, a series of spiropyran-based chiral diblock salen Ti^IV copolymers were used to form metallomicelles capable of catalyzing the asymmetric sulfoxidation of aryl alkyl sulfides.⁶¹ In contrast to the majority of studies that have utilized spiropyran derivatives in organic solvents, the mentioned work was conducted in an aqueous environment, which preferentially stabilized the charge-separated merocyanine form of this photoresponsive molecule even in the dark. Thus, exposure to visible light induced isomerization from the hydrophilic merocyanine isomer to the hydrophobic spiropyran form. As a result, the metallomicelles underwent self-assembly to the catalytically-active organized structure in the dark, followed by disassembly and precipitation from aqueous solution under exposure to visible light.⁶¹ The catalytic activity of this assembled metallomicelle was first evaluated through probing the asymmetric oxidation of phenyl methyl sulfide in the dark. After one hour, >97 % yield with 99 % enantioselectivity was detected.⁶¹ Further studies using the same metallomicelles showed promising yields for a wider variety of aryl alkyl sulfide substrates, including p-methoxyphenyl methyl sulfide, o-methoxyphenyl methyl sulfide, phenyl n-butyl sulfoxide, and phenyl n-hexyl sulfoxide. The studied spiropyran-containing metallomicelles not only allowed for control of reaction yields for several sulfoxidation reactions, but they also demonstrated a pathway for facile recycling of catalysts. In particular, irradiation at 550 nm caused photoisomerization to the hydrophobic form of the metallomicelle, which precipitated from the reaction mixture and could be recycled through filtration.

Rather than self-assembly occurring due to light-induced changes in polarity as described above, a different approach is to utilize changes in geometric configurations to promote self-assembly of catalytically-active systems. Functionalization of a peptide chain with an azobenzene moiety, for instance, has been used to control the self-assembly of peptide fibrils through promotion of π-π stacking interactions for the trans-azobenzene-containing peptide chain.⁶² In this system, the trans-isomer of the azobenzene functionality allowed for noncovalent interactions, such as the mentioned π-π stacking in combination with other hydrophobic interactions. As a result, the trans-azobenzene-based peptide chains underwent self-assembly into an organized peptide fibril (Figure 12). In contrast, exposure to UV irradiation resulted in formation of the cis-azobenzene photoisomer. The significant geometric rearrangement upon photoisomerization broke the noncovalent interactions between the assembled peptide subunits and resulted in destruction of the organized fibril system. As shown in Figure 12, the assembled fibril could promote hydrolysis of p-nitrophenyl acetate as an artificial hydrolase enzyme mimic, but the disassembled state was unable to do so. The mechanism for the hydrolysis reaction promoted by the assembled peptide fibril relies on catalytically active histidine residues which facilitate deprotonation of water molecules to initiate hydrolysis.⁶² He and co-workers attributed the detected higher reaction rate for the trans-azobenzene-containing system to cooperation between the active histidine residues located in proximity to each other in the self-assembled structure.⁶² That is, the aligned histidine residues could cooperatively lower the pK_a of bound water molecules to promote deprotonation. As a result, the assembled hydrolase mimic exhibited a catalytic activity of 95.87 μM min⁻¹, while irradiation with UV light to cause disassembly reduced the catalytic activity to 65.56 μM min⁻¹.⁶² Thus, control of the catalytic activity and of the self-assembly process was possible through irradiation of the system to cause azobenzene isomerization. Further studies from Bandyopadhyay and co-workers have also illustrated that integration of azobenzene in enzyme mimics allows for precise tuning of the reaction outcomes.¹²⁹ Rather than light-activated self-assembly of protein structures, azobenzene photoisomerization could also cause relatively small geometric changes in the distance between amino acid residues in a racemase mimic, which modulated the kinetics of the racemization reaction converting L-alanine to D-alanine.¹²⁹ Similarly, modulation of the size of the catalytic pocket for peptide-based catalysts through azobenzene photoisomerization has been implemented to control the site-selectivity in biomimetic acetylation of sugars.⁶⁸ Advances in biomimetic catalysts are not limited to azobenzene-containing systems, and the binding affinity of substrates and peptides has also been successfully regulated using diarylethene¹³⁰ and spiropyran¹²⁷ derivatives.

2.3.2 Tailoring membrane permittivity as a function of light

In addition to modulation of geometric parameters and self-assembly processes, incorporation of light-responsive molecules in micelles or vesicles has also allowed for tailorable membrane permittivity which could be altered to regulate flow of reactants through a membrane (Figure 13).^{63, 131} In this direction, Bruns and co-workers utilized photoresponsive donor-acceptor Stenhouse adducts (DASAs) to construct nanoreactors with membranes possessing switchable permeability.⁶³ In this case, the DASAs could be converted between nonpolar triene-enol and polar cyclopentanone forms upon irradiation with visible light.⁶³ As illustrated in Figure 13, embedding DASAs in self-assembled vesicles prepared from amphiphilic block copolymers provided an opportunity to switch between a nonpolar and polar membrane surrounding a horseradish peroxidase enzyme. Upon irradiation with visible light, the prepared nanoreactors exhibited increased permeability, allowing for diffusion of reactants into the vesicle which housed the enzyme.⁶³ Specifically, diffusion of pyrogallol to horseradish peroxidase in the presence of hydrogen peroxide resulted in catalytic formation of purpurogallin. Notably, the reaction was only effective under constant irradiation with visible light, indicating that diffusion of reactants toward a catalytic site could be controlled via irradiation with an appropriate excitation wavelength. Moreover, design of two nanoreactors responsive to different excitation wavelengths (e.g., green or red light) allowed for control of cascade catalysis in which the product formed by one reaction was allowed to diffuse into a second nanoreactor with high spatiotemporal control.⁶³ Notably, the concept of photoswitchable nanoreactors has also been successfully demonstrated using spiropyran as the photochromic moiety, and the resulting system was capable of regulating pathways for incompatible tandem catalysis.¹²⁷ For example, multi-step synthesis requiring more than one catalyst which could interact unfavorably (i.e., catalyst poisoning or increased formation of byproducts), could be achieved in a one-pot procedure if the catalytically active species were separated by micelles.¹²⁷ Thus, construction of photochromic micelles, which behave as photoresponsive nanoreactors, represents a pathway to synergistically catalyze multiple organic transformations in an efficient manner.

2.3.3 Nanoparticle catalytic activity as a function of light

A different direction which has gained significant attention in recent years is coupling inorganic NPs with organic photochromic ligands because the catalytic behavior of the resulting NPs could be tuned using light.^{64, 65, 122-126, 132, 133} The catalytic behavior and stability of many NPs (e.g., gold NPs) are affected by several factors, including the nature and conformation of organic ligands.^{64, 65, 122-126, 132, 133} For example, a passivating layer of organic ligands is necessary to prevent NP aggregation and precipitation, but the presence of organic ligands also usually reduces the accessible surface area and catalytic activity.^{122, 125} Thus, photochromic ligands, which could reversibly change their configuration on the surface of the NP, is one of the avenues for regulating the activity and recyclability of NP-based catalysts.^{64, 121-123, 126, 133} For instance, Knecht and co-workers demonstrated that switching of azobenzene-containing peptide ligands bound to the surface of gold NPs could alter the rate of the catalytic reduction of 4-nitrophenol to 4-aminophenol.¹²² These studies probed the possibility of alternation of binding affinity of material-binding peptides as a function of azobenzene photoisomerization. While the peptide ligands exhibit strong binding to the NP surface through noncovalent interactions, these interactions can be weakened by small changes (e.g., point mutations affecting residue sequence) which alter the conformation of the peptide with respect to the NP surface.¹²² With this in mind, integration of an azobenzene moiety within the peptide sequence could allow for light-induced “mutations” that alter the peptide ligand binding strength. In fact, the presence of trans-azobenzene coincided with an increased rate (10.5±0.2×10⁻² s⁻¹) for the reduction of 4-nitrophenol to 4-aminophenol, which could be attributed to weaker binding and dissociation of some ligands to expose a larger portion of the catalytically active NP surface. In contrast, irradiation with 365-nm light to yield cis-azobenzene was accompanied with stronger binding of the peptide ligands and reduced reaction rate (9.4±0.7×10⁻² s⁻¹).¹²² A different example coupled a proline organocatalyst with gold NPs by covalently bonding the proline functionality to an aliphatic tail which could be coordinated to the metal surface by a thiol end group.⁶⁴ An azobenzene spacer was used within the aliphatic chain, allowing for modulation of the distance between the proline catalyst and NP surface (Figure 14). For instance, irradiation with 365-nm light caused photoisomerization to form cis-azobenzene, which brought the catalyst close to the NP surface and suppressed catalytic activity by preventing close interactions between catalyst and substrate through steric hindrance.⁶⁴ These latter examples⁶⁴ illustrate how incorporation of photoresponsive ligands on the surface of inorganic NPs is a pathway to control catalytic activity of either the metal NP or an organic catalyst depending on the system design.

Combining the areas of switchable catalysis and bioorthogonal chemistry has resulted in a pathway for noninvasive control of catalytic processes even in living cells.⁶⁵ In this direction, Qu and co-workers designed heterogeneous palladium catalysts by grafting of Pd⁰ NPs to a macroporous silica support (Figure 15).⁶⁵ While the prepared NPs were capable of catalyzing Suzuki-Miyaura cross-coupling reactions by themselves, modification with supramolecular azobenzene and β-cyclodextrin functionalities allowed the system to be controlled by light. First, the Pd⁰ NPs were embedded inside the pores of the silica support, and the support was then functionalized with photoresponsive azobenzene groups. As shown in Figure 15, trans-azobenzene promoted interactions with β-cyclodextrin, which served to block the pores of the silica support and prevent catalysis from occurring at the Pd⁰ NPs.⁶⁵ Exposing the system to UV light promoted the trans-to-cis isomerization of azobenzene, which broke the interactions with β-cyclodextrin and exposed the Pd⁰ catalysts inside the pores. As a result, higher catalytic activity was detected under UV irradiation than visible light. To advance these studies further, living cancer cells were incubated in a suspension of the prepared NPs embedded in the silica support. Then, the NP-containing cells were exposed to a fresh solution of non-fluorescent starting materials for a Pd-catalyzed cross-coupling reaction. The product of the cross-coupling reaction was a mitochondria-specific fluorescent probe. Only under UV light did the cross-coupling reaction occur, and the reaction progress could be tracked using fluorescence imaging of the cells. Control experiments utilizing a commercially available mitochondrial dye revealed that the light-gated Suzuki-Miyaura cross-coupling reaction was capable of producing fluorescent probes for cellular imaging in vitro. Thus, these studies foreshadow new pathways for the development of noninvasive imaging, diagnosis, and treatment using photoresponsive heterogeneous catalysis.

2.3.4 Catalytic activity of extended structures as a function of light

While the previously described study utilized macroporous silica to support the catalyst, many studies in the area of heterogeneous catalysis have focused on COFs and MOFs as other suitable platforms due to their high surface area and tailorable porosity, which promotes mass transport of the reactants throughout the catalytic framework. Additionally, these frameworks could be tailored to include either active metals within the metal nodes or organocatalysts as parts of linkers connecting secondary building units.^{66, 113-120} Given the several successful examples of COF- or MOF-based heterogeneous catalysis, integration of photochromic moieties into these systems^{49, 134, 135} is a logical approach to expand the role of COFs and MOFs in the field of switchable catalysis. For example, switchable generation of ¹O₂ for oxidation of various organic substrates⁶⁶ or for photodynamic therapy¹⁰ is one area where COFs and MOFs have been implemented successfully. For instance, a COF constructed from porphyrin- and diarylethene-based linkers could be used to modulate the photocatalytic oxidation of amines through selective ¹O₂ generation (Figure 16).⁶⁶ In this system, the triplet excited state of the porphyrin linker by itself promotes conversion of ³O₂ to ¹O₂; however, incorporation of a diarylethene-based linker creates a competitive energy transfer pathway that can suppress generation of ¹O₂.^{66, 136} In particular, the lowest triplet energy of the open form of the diarylethene-based linker exceeded that of the porphyrin linker, which prevents energy transfer from occurring in the presence of the open diarylethene photoisomer. At the same time, the lowest triplet energy of the porphyrin linker lies above that of ³O₂, allowing for efficient conversion of ³O₂ to ¹O₂. In contrast, photoisomerization of diarylethene from the open to closed form results in a lower energy triplet excited state that is compatible with energy transfer from the porphyrin derivative to the diarylethene linker. As such, switching from the open to closed form of the diarylethene linker upon exposure to blue light caused the ¹O₂ generation pathway to be suppressed by the favored energy transfer pathway. This concept was demonstrated experimentally through the photocatalytic oxidative coupling of amines to yield imines under various irradiation conditions. Indeed, enhanced generation of ¹O₂ associated with the open diarylethene photoisomer resulted in reaction conversions of up to 99 % for oxidative coupling of benzylamine in comparison to 63 % conversion detected for the COF containing closed diarylethene.

Other approaches to synergistic coupling of diarylethene and porphyrin derivatives in COFs have involved oxidation of N-methylpyridinium salts under alternating irradiation conditions.¹¹⁸ In this case, the open form of the diarylethene linker promoted catalytic oxidation of 1-methylquinoline-1-ium iodide to 1-methylquinolin-2(1H)-one by molecular oxygen, and the detected conversion was 99 %, which the authors attribute to both higher specific surface area and changes in electronic structure due to photoisomerization. Irradiation with UV light to induce photoisomerization to the closed form of the diarylethene linker resulted in a reduced conversion of 61 % on average under the same reaction conditions.¹¹⁸ Further studies are necessary to fully uncover the mechanism for the enhanced catalytic activity of the open diarylethene derivative in comparison with the closed photoisomer. One of the hypotheses for the observed behavior could be attributed to the fact that the closed isomer suppresses electron transfer from the porphyrin derivative to the substrates.¹¹⁸

In addition to the described studies focusing on photoresponsive COFs, it is worth mentioning that hydrazone-based COFs have been utilized as efficient photocatalysts for several types of organic transformations, but the role of the hydrazone derivative has mainly been limited to a light-harvesting unit rather than a “knob” for control of catalytic activity.^{105, 116} Thus, future studies involving the photoswitching ability of hydrazone-based COFs in conjunction with photocatalysis will undoubtedly yield valuable insights for the development of new switchable catalytic materials.

Integration of azobenzene-based linkers in MOFs has also been employed to modulate the response of heterogeneous catalysts.¹²⁰ For example, an azobenzene-functionalized MOF was evaluated in its ability to catalyze several Knoevenagel condensation reactions, which are typically base-catalyzed processes.^137-140 The azobenzene photoswitch could then serve a dual role. First, the basicity of the N=N bond of the azobenzene was probed for promotion of the condensation reaction, and E/Z photoisomerization could change the available pore space and control diffusion of the reactants through the catalyst. For instance, the adsorption capacity of the azobenzene-MOF was detected to decrease under UV irradiation and increase when the MOF was stored in the dark, indicating that photoisomerization of the azobenzene group affected the pore volume. These results were supported by the catalytic yields, which were elevated when the reactions were conducted in the dark and suppressed under UV irradiation. For example, the condensation of salicylaldehyde proved to be extremely sensitive to azobenzene photoisomerization as the conversion was determined to be 93 % in the dark and only 6 % under UV excitation. Further, the reaction conversion was dependent on substrate size because it is a limiting factor in diffusion rate.¹²⁰ Thus, photochromic MOFs offer the ability to not only modulate catalytic activity, but also selectivity using a single platform.