Research ArticleChemistry

Ultrafast photoactivation of C─H bonds inside water-soluble nanocages

See allHide authors and affiliations

Science Advances  22 Feb 2019:
Vol. 5, no. 2, eaav4806
DOI: 10.1126/sciadv.aav4806


Light energy absorbed by molecules can be harnessed to activate chemical bonds with extraordinary speed. However, excitation energy redistribution within various molecular degrees of freedom prohibits bond-selective chemistry. Inspired by enzymes, we devised a new photocatalytic scheme that preorganizes and polarizes target chemical bonds inside water-soluble cationic nanocavities to engineer selective functionalization. Specifically, we present a route to photoactivate weakly polarized sp3 C─H bonds in water via host-guest charge transfer and control its reactivity with aerial O2. Electron-rich aromatic hydrocarbons self-organize inside redox complementary supramolecular cavities to form photoactivatable host-guest charge transfer complexes in water. An ultrafast C─H bond cleavage within ~10 to 400 ps is triggered by visible-light excitation, through a cage-assisted and solvent water–assisted proton-coupled electron transfer reaction. The confinement prolongs the lifetime of the carbon-centered radical to enable a facile yet selective reaction with molecular O2 leading to photocatalytic turnover of oxidized products in water.


Photon-mediated activation of chemical bonds has long been regarded as an efficient method to trigger chemical reactions (14). Since the fundamental time scale of photon absorption is in femtoseconds, the ensuing nonequilibrium nuclear dynamics on the excited-state potential energy landscape can lead to facile and efficient chemistry. However, due to the ultrafast rates of photon-energy redistribution within the various internal molecular degrees of an excited molecule, carrying out photoactivated bond-selective chemistry has remained a challenge (5). In recent times, it has been realized that preorganization of reactants preceding light excitation can be a key for controlling photoactivated reactions (68). Enzymes provide a powerful inspiration in this context as they lower the activation barrier of chemical reactions by orienting the substrates appropriately in their active sites for bond-selective polarization and reactivity (9). To mimic the concept of bond polarization by preorganization, chemists have designed synthetic active sites by constructing supramolecular cavities to host chemical reactions that are usually difficult to carry out in free solution (1012). With the design of diverse cavities having tunable shapes, confinement sizes, and electronic properties, remarkable discoveries of new reactions and difficult chemical cascades have been made possible (13, 14). Here, we show that the concept of supramolecular host-guest preorganization can be harnessed to selectively photoactivate benzylic sp3 C─H bonds and enable its controlled reaction with molecular oxygen to achieve photocatalytic turnover of oxidized products.

Activation of weakly polarized C─H bonds has remained a grand synthetic challenge in organic chemistry (1517). Natural enzymes carry out C─H bond activation inside their hydrophobic cavities with precise positioning of the hydrocarbon substrate near the H atom abstraction site, via preorganization. In the enzyme cytochrome P450, the O2 activation site that generates reactive metal-oxo species is capable of abstracting a hydrogen atom from the requisite C─H bond (9). Synthetic transition metal complexes successfully mimic the generation of high-valent metal-oxo species (1821), although they often undergo self-degradation concurrently with the diffusive C─H functionalization reaction. Recent template-based C─H activation methodologies have been able to alleviate the lack of preorganization in the activation step (2224), although the additional steps of attachment and removal of the template increase the cost of functionalization. Alternatively, supramolecular cavities have shown tremendous potential for orienting the substrates in the right reactive configuration and for kinetically stabilizing on-pathway reactive intermediates. Fujita and co-workers (25, 26) had demonstrated that tertiary C─H bonds of adamantane can be activated by ultraviolet (UV) light excitation inside a porous Pd6L412+ nanocage where adamantyl hydroperoxide and its tertiary alcohol were formed as products. Although the cage-induced C─H bond reactivity was promising, a lack of an industrially viable catalytic methodology and limited understanding of the reaction mechanisms made it difficult to access a wide range of hydrocarbon substrates.

To activate a C─H bond via H atom abstraction reaction inside a water-soluble cavity, a host-guest photoactivation scheme, which should be universal in its operation, will be efficient. We recently reported an ultrafast (few picoseconds) light-induced H atom abstraction reaction via proton-coupled electron transfer (PCET) (2729) inside a cationic Pd6L412+ nanocage (30) by preorganizing ionizable N─H or O─H bonds with proximal solvent water molecules (31). The incipient acidity of the photoexcited N─H or O─H polar groups in the radical cation state was exploited to generate long-lived neutral radicals (31). We therefore hypothesize that a mild yet green catalytic method for selective C─H functionalization can be established by combining light activation with substrate preorganization inside a water-soluble supramolecular cavity (Fig. 1). Previously, PCET cascades (32, 33) or hydrogen atom transfer reactions (34, 35) have been used to indirectly activate C─H bonds in photoredox catalysis (36). In addition, Mayer and co-workers (37) have recently shown that C─H bonds can be cleaved or formed in free solution via multisite concerted proton-electron transfer by suitably prepositioning a carboxylate base near to the requisite C─H bond within the substrate. This reaction strategy is promising, although it suffers from a diffusion-limited activation of C─H bonds in the presence of stoichiometric amounts of chemical oxidants or reductants. Consequently, inspired by these ideas, here, we postulate that ArĊH2 radicals (Ar = aryl) can be photogenerated in confinement by triggering the PCET reaction (Fig. 1) directly at the benzylic C─H site of ArCH3 whose acidity can be substantially altered in the radical cation state (pKa can be < −10 for the radical cation) (3841). Thus, the ultrafast photogeneration of a long-lived carbon-centered neutral radical species inside a supramolecular cavity is expected to ensure temporal segregation of the subsequent O2-dependent functionalization steps, thereby enabling an almost enzyme-like execution of organic oxidation reactions in water.

Fig. 1 Strategy for ultrafast C─H bond activation followed by photooxidation in water.

Our bioinspired host-guest CT paradigm for C─H bond oxidation inside a cationic Pd6L412+ nanocage using visible light (Vis), molecular O2, and water. The Pd6L412+ nanocage encapsulates organic substrates in water and provides a C─H photoactivation route through the ultrafast PCET reaction to oxidize the substrate C─H bond under ambient conditions.


Photoinduced host-guest charge transfer creates radical cations and long-lived neutral radicals inside the Pd6L412+ nanocage

To validate our biomimetic host-guest paradigm for metal site–independent C─H bond photoactivation, we chose three alkyl-aromatic substrates: 9-methylanthracene (9-MA; 1), 1-methylnaphthalene (1-MeNap; 2), and toluene (Tol; 3), which have varying degrees of aromaticity and distinct redox potentials. Incarcerating these electron-rich aromatic guest molecules inside a water-soluble Pd6L412+ nanocage (figs. S2 to S4) (30) resulted in the generation of color in the solution emanating from new charge transfer (CT) bands in the visible part of the spectrum (Fig. 2A, inset). The CT absorption band for 1 ⊂ Cage is broad with a peak maximum at 475 nm (Fig. 2A). However, 1-MeNap and Tol that have smaller ring sizes compared to 9-MA show blueshifted CT transitions for their respective incarcerated complexes, i.e., 2 ⊂ Cage and 3 ⊂ Cage, relative to that of 1 ⊂ Cage. These emergent CT features were absent in the absorption spectra of both free guest molecules dissolved in organic solvents and the aqueous solution of the empty Pd6L412+ cage (figs. S5 and S8). Incarceration of the guest molecules was confirmed by the observation of upfield 1H NMR (nuclear magnetic resonance) spectral shifts of guest protons (figs. S6, S9, and S11) due to hydrophobicity of the cavity and aromatic ring currents of the triazine ligands in the walls. We could quantify from the NMR peak area integration that two of 9-MA molecules, three of 1-MeNap, and four to five of Tol molecules have occupied the cage cavity. For 9-MA, the large size of the anthracene ring (long axis is ~9.3 Å) allows only two guests per cavity since the cage diameter is roughly around ~ 20 Å (fig. S6). The steady-state spectroscopy on the host-guest complexes therefore predicts that radical cationic states of guest molecules can be photogenerated in confinement.

Fig. 2 Spectroscopic detection of charge transfer intermediates and ultrafast generation of neutral benzylic radicals.

(A) Absorption spectra of the host-guest complexes with 9-MA (1), 1-MeNap (2), and Tol (3) inside the Pd6L412+ nanocage. The inset shows the visible color of the aqueous complexes. (B) Transient absorption spectra of guest cation radicals inside the cavity at ~1 ps after photoexcitation to host-guest CT. (C) Comparative decay kinetics of the radical cation state of 2 ⊂ Cage in D2O versus H2O, tracked by plotting the excited-state population decay with time, derived from pump-probe transient absorption data on 2 ⊂ Cage, excited by a 400-nm pump. (D) Comparative lifetimes of all the three alkyl aromatic hydrocarbons in their radical cation states and corresponding KIEs for the benzylic PT reactions from radical cations to the solvent to form neutral radicals. (E) Scheme for PCET reaction for 2 ⊂ Cage and 3 ⊂ Cage, leading to the generation of the long-lived (>2.5 ns) neutral benzylic radicals. PT time scales for 2 ⊂ Cage and 3 ⊂ Cage are 354 ± 25 ps and 46 ± 4 ps respectively. OD, optical density.

To track the photogenerated radical cation and neutral radical states of the guest molecules inside the cavity, we carried out femtosecond transient absorption measurements. Figure 2B shows the radical cation signatures for the incarcerated guests, 1, 2, and 3, after an ultrafast electron transfer reaction (<100 fs) to the nanocage, which forms a corresponding radical anion (25, 26), all recorded at 1 ps subsequent to photoexcitation. For both 9-MA and 1-MeNap, we observed sharp features in the 675- to 750-nm region, while for Tol, a sharp peak at 450 nm was observed. Notably, the radical cation features for all the three molecules were slightly redshifted by ~10 to 20 nm from their respective free-solution absorption maxima, providing evidence for confinement of the radical cations (42, 43). The cage-incarcerated radical cationic state of the aromatic substrates was observed to decay in a picosecond time scale to form the corresponding neutral benzyl radicals after donating a proton to the solvent water molecule since it is an exergonic process (41). The neutral radical states of these alkyl aromatics are typically characterized by their substantially low molar absorption cross section (44). Exponential fits of the radical cation decay kinetics for the signature spectra observed at 714 nm for 2 ⊂ Cage (Fig. 2C) revealed a time constant of 354 ± 25 ps for the generation of the long-lived neutral radical signal. Although the cavity-incarcerated Tol radical cation of 3 ⊂ Cage complex also decays with a monoexponential rate, the time scale is an order of magnitude faster for the deprotonation step. We observe a radical cation decay time scale of 46 ± 4 ps for the formation of neutral benzyl radical (PhĊH2) for 3 ⊂ Cage subsequent to photoexcitation of the weak CT band at 400 nm (fig. S32). In contrast, the incarcerated 9-MA radical cation gets deprotonated with two distinct time constants of 14 ± 0.8 ps and 47 ± 9 ps, respectively (figs. S28 and S29). The large variation in guest deprotonation rates, from tens to hundreds of picoseconds, indicates that the cavity preorganizes the sp3 C─H bond differently in the case of Tol, 9-MA, and 1-MeNap for proton transfer (PT) to solvent water. The PT reaction from a benzylic ─CH3 becomes facile because of substrate polarization in the photoexcited host-guest CT state. We therefore assign this as a stepwise ET-PT or bidirectional PCET, one of the classes within the generalized PCET reactions (27, 28, 45).

The disparity in PCET reaction time scales motivates us to probe the differential positioning of the reactive methyl group with respect to the proton-accepting water molecule for the three alkyl-aromatic substrates. To understand the buried nature of the methyl group, its dynamic mobility inside the cavity, and the corresponding heterogeneity of the guest population, we compared the 1H NMR linewidth of sp3 methyl protons in all the three alkyl-aromatic substrates. We have performed temperature-dependent 1H NMR measurements on the host-guest complexes to identify conformational heterogeneity of the guest inside the cage. At room temperature (fig. S7), we find that, in the case of 9-MA ⊂ Cage, the 1H NMR linewidth is ~10 times broader than that for 1-MeNap (6.5 Hz; fig. S10) and Tol (7 Hz; fig. S11) inside the cage and two distinguishable conformers of 9-MA coexist. We observe two distinct peaks [δ 0.44 and δ 0.34 parts per million (ppm)] buried under the broad envelope, which is prominently revealed by lowering the temperature to 5°C . We could therefore expect two distinct deprotonation time scales representative of two major populations with altered solvent exposure. However, the extremely large upfield shift (δfree = 2.78 ppm to δcaged = −0.13 ppm) of methyl protons for 1-MeNap suggests that the C─H is highly shielded through its proximity to triazine walls and possibly positioned away from the interfacial water molecules. Temperature-dependent 1H NMR on the 2 ⊂ Cage complex shows a single peak throughout the temperature range (25° to 5°C) with gradual broadening at low temperatures, dictating that one kind of conformation predominantly exists in the ground state (fig. S10).

Solvent-assisted sequential PCET reactions for confined alkyl-aromatic substrates

Convincing proof for water-assisted substrate deprotonation of the radical cation leading to neutral alkyl-aromatic radical formation was obtained by measuring the H/D kinetic isotope effect (KIE) of the radical cation decay rates in H2O and D2O. We find that the substrate deprotonation time scales are altered by a D2O/H2O solvent exchange. For instance, the C─H deprotonation in D2O for nanocage-confined 1-MeNap (Fig. 2C) takes longer (456 ± 35 ps) than in H2O (354 ± 25 ps), indicating that the secondary KIE value of 1.29 arises from a closely interacting solvent water molecule. In the case of nanocage-confined Tol, we find a slightly higher KIE value of 1.44 than in 1-MeNap. The KIE again confirms the formation of a neutral benzyl radical state. The decay of the cage-confined 9-MA radical cation state shows biphasic deprotonation kinetics such that both of the steps [fast (14 ps) and slow (47 ps)] show secondary KIE values of 1.43 and 2.32, respectively (figs. S28 and S29). We summarize the secondary KIE values for all the three substrates in Fig. 2D to highlight the distinguished role of the water solvent in the primary step of −CH3 deprotonation.

To further dissect the proton removal reaction at the substrate end, we studied the C─H versus C─D cleavage reaction rates in 9-MA (figs. S30 and S31). The observation of clear primary isotope effects in proton abstraction steps show that the C─H bond cleaves with the assistance of a water molecule or water cluster and that the rate-limiting step involves both the radical cation and solvent H2O/D2O. Primary KIE values for both steps with deuterated d12–9-MA show a lower value of 1.37 and 1.6, respectively (fig. S31), than the secondary KIE values obtained for the D2O exchange experiment (1.43 and 2.32; figs. S28 and S29). Such a distinction in KIE values suggest that the bidirectional PCET reaction is solvent-assisted, presumably through a water cluster around the sp3 C─H bond. We hypothesize that the variation in KIE values for the three alkyl-aromatic substrates results from a combination of differential excited-state proton acidities, substrate packing inside the cavity, and differential solvent (H2O) accessibility to the reactive ─CH3 functionality.

We next mapped the exposure of the ─CH3 group in our confined alkyl-aromatic substrates to the water cluster independently by measuring the solvation time scale (4648) of the excited CT state. All the three host-guest complexes explored here show a specific coupled CT state (31, 48) with an optical transition in the near-infrared region between 850 and 1100 nm, obtained from transient absorption (fig. S33). Previous studies have established the dynamical blueshift observed for this coupled CT feature as a representative of the solvent rearrangement time scale in the photogenerated CT state (48). We find that, in 9-MA ⊂ Cage, solvent reorganization takes place in a biphasic manner with two time scales of ~900 fs and 4.2 ps (fig. S34), which we assign as local and bulk solvent relaxation time scales of the CT state, respectively. In contrast, for 1-MeNap, we find solvent relaxation time scales of 1.3 and ~34 ps, respectively. The 10-fold slower time scale for bulk water rearrangement in 1-MeNap compared to that in 9-MA hints at the buried nature of the methyl protons with respect to the bulk water. Therefore, the sluggish time scale (~34 ps) for the bulk rearrangement of the water shell around the photoexcited 2 ⊂ Cage complex finally leads to the slowest bidirectional PCET reaction.

We deduced the lifetime of the radical cation and neutral radical population in 2 ⊂ Cage and 3 ⊂ Cage by analyzing the time-resolved absorption data with a two-species sequential model, as shown in Fig. 2E. The picosecond time scales (14 to 350 ps) for deprotonation reactions reported here are slower than the solvent water reorganization time of 4 to 34 ps around the cationic nanocage. Although we find that the solvent reorganization time scale is critical to the radical cation deprotonation rates, we believe that other factors including molecular dynamics inside the cavity also contribute to the efficiency of the neutral radical generation rates. The directly measured rate of this cage-confined bidirectional PCET reaction at the carbon-site is at least 103 times faster than that previously reported for C─H bond activation reactions (40, 45, 4951). We believe that preorganization of the C─H bond with respect to the proton-accepting water cluster through cage incarceration leads to this ultrafast bond breaking rates.

9-MA undergoes facile photooxidation inside nanocage

To test the reactivity of the photogenerated and cage-stabilized alkyl-aromatic neutral radicals, we carried out photooxidation of all three guests, 1, 2, and 3, with molecular oxygen by forming stoichiometric host-guest complexes (figs. S6, S9, and S11) with the Pd6L412+ nanocage, as summarized in Fig. 3, A to C. Resonance excitation to the CT state with green light centered at 530 nm leads to complete oxidation of 9-MA within 5.5 hours under ambient conditions (figs. S35 and S36A). After 1.5 hours of reaction, the photoproducts primarily include quinones and hydroxylated compounds, i.e., 9,10-anthraquinone (40%), hydroxy-quinone (30%), 9-anthracenealcohol (9-AnCH2OH; 15%), and 9-methyl-10-hydroxyanthracene (15%) (figs. S37 and S38). At a higher O2 pressure of ~1.5 atm, the product distribution substantially changes with ~10-fold faster reaction rates (figs. S36B and S39). The reaction was made photocatalytic by loading 2 mole percent (mol %) of the cage in the aqueous phase with excess solid powder of 9-MA (fig. S41). The product distribution obtained for the catalytic reaction shows that the flux favors the hydroxylated products at the expense of the anthraquinone product. Since anthraquinone formation requires multi-e/H+ removal steps, substrate turnover steps during catalysis limit the over-oxidation of the anthracene framework by shortening the residence time of the pathway intermediates or products inside the cavity.

Fig. 3 Visible light–triggered photoreactions inside the Pd6L412+ cage and the corresponding photocatalytic cycle.

(A) Photooxidation of 9-MA (1) inside the cage with aerial O2 for 1.5 hours generates four photoproducts dominated by the 9,10-anthraquinone. The nanocage-mediated photooxidation of substrates mimics the action of natural oxidase enzymes, such as cytochrome P450. (B) Cage-incarcerated 1-MeNap (2) leads to two oxidized products under aerobic photooxidation for 7.5 hours. (C) Tol (3) gives rise to a single photooxidized product, benzaldehyde, from the photoreaction under high O2 pressure (~1.5 atm) for 8.5 hours. (D) Gas chromatography–mass spectrometry (GC-MS) chromatograms showing photocatalytic conversion of Tol to benzaldehyde in a time-lapsed manner in the presence of a catalytic (5 mol %) amount of nanocage. (E) Higher alkyl analog of Tol, cumene, 4 (3° benzylic site), forms three photooxidized products upon 400-nm light-emitting diode (LED) illumination for 33 hours in the presence of 5 mol % cage catalyst. Isolation of cumene hydroperoxide proves the generation of benzylic hydroperoxides from the initial carbon-centered radicals at the benzylic site after reaction with molecular O2. (F) Catalytic cycle for Tol photooxidation starts with the activation half where, after uptake of the substrate, PCET-induced selective sp3 C─H photoactivation takes place in water. In the oxidation part of the catalytic cycle, benzaldehyde formation from neutral benzyl radical goes via benzyl hydroperoxide (this 1° hydroperoxide could not be isolated because of its unstable nature) generation and subsequent heterolytic cleavage of the O─O bond. Then, excess Tol molecules, present outside the cage, displace the benzaldehyde product from the cavity and replenish the Tol ⊂ Cage complex for the next catalytic cycle.

To confirm the influence of residence times on the product distribution of 9-MA photooxidation, we separately carried out the photoreaction on the cage-loaded 9-AnCH2OH, which is an intermediate in the 9,10-anthraquinone formation pathway. In this case, we observe sole formation of the anthraquinone product (fig. S42B), demonstrating that the residence time of the primary alcohol formed in the cavity during the oxidation of 9-MA directly influences reaction selectivity. Our proposed reaction sequence was also confirmed through 1H NMR experiments, where we found that the loss of the 9-AnCH2OH product correlated with the increase of the quinone formation (fig. S42A). The formation of an anthraquinone photoproduct from both 9-MA and 9-AnCH2OH is not entirely surprising since the middle ring in the anthracene moiety has a diene character, thereby enabling the electrophilic addition of O2 at the 9,10 position. The identity of the chemical products that we obtain from the nanocage-mediated photoreaction (for plausible mechanism, see scheme S2) matches exactly with the enzyme-catalyzed oxidation products of 9-MA and 9-AnCH2OH (52). In essence, we find that our methodology of photoinduced PCET activation and subsequent oxidation for 9-MA inside the cage mimics the enzyme-mediated reaction conceptually without the direct involvement of a conventional metal-based catalyst.

Nanocage enables selective photooxidation of 1-MeNap and Tol

For 1-MeNap and Tol, the reduced number of fused benzene ring prohibits a direct 1,4 addition of O2, thereby reducing the possibilities of para-quinone formation. Blue-light illumination [460-nm light-emitting diode (LED)] on the host-guest complex of 2 ⊂ Cage for 7.5 hours yields 1-naphthaldehyde in 60% and 7,8-diol in 40% yield (Fig. 3B and figs. S43 and S44). The higher selectivity for C─H photooxidation to yield aldehyde and alcohol without any naphthoquinone formation demonstrates the generality of our host-guest photoactivation paradigm. Photocatalysis was performed with 5 mol % of the cage while maintaining the product ratios. In contrast, the enzymatic oxidation of naphthalenes by naphthalene dioxygenases usually produce ring-degraded products after the formation of aromatic diols, aldehydes, and quinones (53). The emergence of selectivity toward ─CH3 oxidation by our host-guest CT paradigm indicates that specificity toward the primary C─H bond can be amplified by tuning the delocalization of the substrate cation radical. Separately, we observed that 1-naphthaldehyde forms from the photooxidation of 1-naphthylalcohol, thereby confirming that the aldehyde product is exclusive if the intermediate alcohol is formed. Since we do not isolate any intermediate alcohol from 1-MeNap photoreaction, we conjecture that 1-napthaldehyde formation can proceed via a concerted 4e/4H+ pathway (54) after C─H photoactivation. Alternatively, we can also rationalize that alcohol, being more reactive toward oxidation, can have a very small steady-state concentration, making it difficult to identify in the photoproduct mixture.

Tol, due to its single benzene ring, has a more spatially localized π-electron cloud and shows a blueshifted CT band relative to 9-MA or 1-MeNap after incarceration (Fig. 2A). By using the PCET trigger operated at 400-nm LED illumination for 3 ⊂ Cage, the photogenerated long-lived benzyl radical inside the cavity reacted with molecular O2 to give benzaldehyde as a single product. When the photoreaction was carried out under an average stoichiometry of ~4 Tol molecules per cage, we obtained 60% yield for the reaction and 100% selectivity for benzaldehyde in 8.5 hours (Fig. 3C). Extending this remarkable result to catalytic conditions, we found that, when operated with 5 mol % loading of the cage, 93% of Tol molecules converted to benzaldehyde with 100% selectivity in 40 hours, as shown in the gas chromatography–mass spectrometry (GC-MS) trace in Fig. 3D (fig. S46). No other side products such as benzyl alcohol or benzoic acid or ester were observed. To the best of our knowledge, this PCET-mediated photocatalytic route is a novel methodology for selective chemical synthesis of benzaldehyde from toluene under ambient conditions in water with high conversion efficiency. Note that, although both stoichiometric and catalytic conversions for selective benzaldehyde formation have been targeted previously (55, 56), these methods have not been optimized for industrial scalability. In addition, the industrial method of Etard oxidation for benzaldehyde synthesis requires high temperatures and use of caustic solvents (57). Apart from these forays, metal-oxide cages have also been used as photocatalytic containers, although the reactions examined were mostly in the heterogeneous phase with activation at a metal site. In contrast, our method relies on the green chemistry principle, i.e., usage of water as a solvent and light and oxygen as reagents, without direct involvement of a metal center in the bond activation while the entire photocatalysis is carried out at room temperature.

Radical stability is increased by alkyl substitutions on the methyl group

Since the photocatalytic Pd6L412+ nanocage cavity mimics an enzyme active site, the displacement of the photoproduct, benzaldehyde, by fresh Tol is essential for maintaining catalytic turnover (see the photocatalytic cycle in Fig. 3F). Since PCET activation of substrate C─H bonds is much faster than the product formation time scale, the residence time of the benzaldehyde inside the cavity is crucial for determining the catalytic turnover rate. Because of the presence of a polar ─CHO functional group, benzaldehyde has higher free energy of solvation in water than Tol, causing facile displacement of benzaldehyde from the cavity by excess Tol after single turnover. Since there were no intermediates observed during conversion of Tol to benzaldehyde, we hypothesized that the neutral benzyl radical upon reaction with O2 and subsequent electron and proton recombination step forms the organic hydroperoxide. Previous works on benzyl radical oxidation by O2 gas or chemical oxidants (55, 58, 59) has also shown benzaldehyde formation along with other oxidized side products. We found out that, depending on the intrinsic stability of aromatic benzylic hydroperoxide, kinetic partitioning via either heterolytic cleavage I or heterolytic cleavage II of the O─O bond can occur (see the schemes in fig. S49). Although our cage-confined benzyl radical reaction with molecular O2 has similarity to literature precedents, we demonstrate temporal and chemical control on the obtained oxidized products. To elucidate the mechanism for benzaldehyde formation inside the cavity, we used isopropyl-benzene or cumene, 4, as a mechanistic probe (see Fig. 3E and fig. S49). Since the isopropyl group can form a stable tertiary radical after the PCET reaction, the photogenerated organic hydroperoxide (4c, 28%) and the corresponding tertiary alcohol (4b, 36%) are stable as compared to their primary analogs, thereby allowing their observation in the GC-MS traces along with the one carbon-removed product, acetophenone (4a, 35%). These stable hydroperoxides at the tertiary carbon of adamantane were also reported previously by Fujita and co-workers (25). We therefore demonstrate that the lifetime of the organic hydroperoxide plays a critical role in directing the selectivity of products formed by also photocatalytically (3 mol % cage) converting ethylbenzene to acetophenone (93%), with a minor population of secondary alcohol (7%) (fig. S48).

Nanocage-confined alkyl-aromatic substrate reactivity is altered by substitutions on the phenyl group

We rationalized that the photocatalytic rates of benzaldehyde formation from Tol could be tuned by systematically introducing electron-donating and electron-withdrawing substituents on the phenyl ring of Tol. To this end, we compared substrate conversion rates and product yields for three different Tol derivatives, namely, methyl (─CH3), methoxy (─OCH3), and fluoro (─F) derivatives with substitution at the para position (in Fig. 4, A and B). While methyl and methoxy groups are electron-donating substituents, the fluoro group has a strong electron-withdrawing effect. We find that, for methyl substitution in para-xylene [for one-dimensional (1D) and two-dimensional (2D) NMR characterization; figs. S17 and S18], the substrate conversion rate increases by twofold relative to Tol, thereby demonstrating a modest extent of reaction tunability induced by this particular phenyl ring substituent through stabilization of the radical cation state (fig. S50). On the other hand, in the fluoro-derivative case (for 1D and 2D NMR characterization; figs. S23 and S24) the reaction does not proceed beyond 70% in 40 hours, demonstrating that the electron-withdrawing substitution in the ring not only slows down the conversion rate by 1.3 times but also affects the turnover step due to additional noncovalent interactions via the ─F site (fig. S54). Unexpectedly, the electron-donating ─OMe substitution (for 1D and 2D NMR characterization; figs. S21 and S22) also shows slower reaction rates but with high conversion percentages (~94%) and exclusive product selectivity (97%) (fig. S53). Presumably, in the case of para-OMe substitution, both the electronic effects and intermolecular interactions caused by substrate packing in confinement drive product selectivity and catalytic rates.

Fig. 4 Tuning the substrate and cavity electronics for photocatalytic oxidation of Tol.

(A) Product distribution for Tol and each of the Tol derivatives at 5 mol % catalyst loading of the Pd6L412+ nanocage and under high O2 pressure (~1.5 atm) with 400-nm LED illumination. All the reactants showed benzaldehyde as the major product with >94% yields. Minor products of alcohol and acid are also observed for certain derivatives. (B) Comparison of photocatalytic reaction time (gray bars) and percentage conversion (orange bars) for Tol, para-methoxy Tol, para-Fluoro Tol, para-xylene, meta-xylene, and ortho-xylene inside the Encage. (C) Reaction rates for conversion of Tol to benzaldehyde using two different Pd6L412+ nanocages, Encage and Bipycage, are plotted.

To understand the dependence of sp3 C─H bond oxidation on the position of the CH3 substituent in the phenyl ring, the reaction rates of positional regioisomers, meta-xylene (for 1D and 2D NMR characterization; figs. S19 and S20) and ortho-xylene (for 1D and 2D NMR characterization; figs. S15 and S16), were compared to that obtained for para-xylene. For meta-xylene, we observe slower rates of conversion relative to para-xylene and pure Tol oxidation (fig. S52), while ortho-methyl substitution led to conversion rates that were slightly faster than the parent Tol but slower than para-xylene rates (fig. S51). The decrease in rates from para- to ortho- to meta- presumably enunciates strong influence of regioisomeric ring substitution effects on C─H bond oxidation. However, the packing of the substrates are also slightly different for the three xylene derivatives (figs. S16, S18, S20, and S25). The chemical shifts of the ─CH3 protons in 1H NMR spectra for the three xylene derivatives correlate well to their catalytic rates of oxidation (fig. S55). The observed chemical shifts from the NMR data indicate faster oxidation reaction rates when the methyl protons are toward the open pore/interfacial water (closer to pyridine Hβ/Hα) rather than orienting to the core triazine ring (C3N3) of the cavity ligand walls. Although two equivalent methyl groups are available for the photooxidation reaction in the xylene derivatives, we find only single ─CH3 oxidation, reinforcing high regioselectivity for the exposed C─H bond. The anomalous reaction rates for para-OMe and para-Fluoro substituents can be explained by the differential packing of these substrates in the cage cavity, as highlighted by the clear changes observed in the chemical shifts and cross-correlation peak intensities in both 1D 1H NMR and 2D 1H-1H ROESY (rotating frame nuclear Overhauser effect spectroscopy) NMR spectral data. We find lower integrated peak volumes for cross-peaks arising between ─CH3 protons of the guest and the triazine protons (Hβ and Hα on the pyridine) of the cage in para-OMe and para-Fluoro substituent compared to all the xylene-incarcerated complexes (table S2). In Fig. 5C, the data indicate greater distances between substrate CH3 protons and cage pyridine-Hα/pyridine-Hβ for para-OMe and para-Fluoro substituted Tol relative to xylenes inside the cavity. When ─CH3 protons are at a distance less or equal to 5 Å to the pyridine-Hβ/Hα, the oxidation reaction proceeds at a faster rate. Para-xylene, ortho-xylene, and meta-xylene show decreases in the catalytic rates as the guest-to-wall distances increase. Beyond 5.5 Å, the catalytic rates appear to be independent of the guest-to-wall distances, as indicated by the similar catalytic rates of para-OMeTol and para-FluoroTol oxidation. Tol ⊂ Cage shows a different rate of catalysis, although the distance of CH3 protons to triazine Hβ is similar to that of ortho-xylene ⊂ Cage. In addition, for the para-OMe derivative, the signal intensities for the OCH3-triazine Hβ/Hα 1H-1H-correlation peaks were larger than that for the corresponding CH3-triazine Hβ/Hα correlation peaks, suggesting that the reactive C─H bond is pointing away from the cage triazine walls (fig. S22). The structural information derived from 1D and 2D NMR indicates altered packing of the molecules upon changing the substituents on the ring. The data in this subsection illustrate that, in addition to the ring electronic effects, which dictate reaction rates, the packing of the molecules inside the cage cavity also plays a critical role in dictating the product formation time scales.

Fig. 5 Molecular packing of Tol inside the nanocage and correlation with photocatalytic rates.

(A) 2D ROESY (rotating frame nuclear Overhauser effect spectroscopy) spectrum of Tol ⊂ Cage complex in D2O in 500 MHz. Cross-correlation peaks are marked by i (Hβ, Tol-CH3), ii (Hα, Tol-CH3), iii (Hβ, Tol-aromatic), and iv (Hα, Tol-aromatic). Distances between protons are calculated by considering the cross peak (Hα, en-CH2) as a reference peak and distance between them (r = 5.35 ± 0.36 Ǻ) as a reference distance. (B) Optimized geometry of single Tol incarcerated inside Pd6L412+ cage. (C) Correlation of methyl-to-triazine distances (1H-1H ROESY NMR distances) with the photocatalytic oxidation rates of the para-substituted Tol derivatives.

The nanocage as a prototypical photoenzyme: Tunable cage electronics and cavity size as evolutionary strategies for controlling reaction kinetics

By combining the enzyme-like substrate encapsulation behavior of the nanocage with light activation, we have introduced a modular photocatalytic method, robust and general enough for a diverse range of alkyl-aromatic substrate oxidation reactions. The cationic Pd6L4 nanocages can be thought to evolve like an enzyme to modulate the catalytic rate of Tol oxidation. Figure 4C compares the reaction time course of photocatalytic Tol oxidation inside two different Pd6L412+ cavities constructed by using ethylenediamine (en) and 2,2′-bipyridyl (bipy) moieties as external cis-chelating ligands to hold the Pd2+ site. The change in the Pd2+ ligands alters both the electronic structure and the cavity size of the cage (30, 60). In case of regular “en”-capped nanocage (Encage), we see product formation in ~40 hours with 93% conversion, while for the Bipycage, the conversion is slower (in ~56 hours with ~90% conversion) although with equal product selectivity (fig. S56). Note here that the reaction proceeds via a lag phase, which is independent of the nanocage used and possibly reflects the initial encapsulation dynamics of the guest inside the host necessary to trigger the catalytic reaction. For another variant of the nanocage, where the external ligand of Pd2+ is TMEDA (N,N,N′,N′-tetramethylethylenediamine), we observe a prominent selectivity change in the Tol photocatalytic oxidation reaction (fig. S57). After 26 hours of photocatalysis, while 53% of Tol is converted into products, ~20% of the product mixture comprises benzyl alcohol, ortho-cresol and para-cresol, and the rest is benzaldehyde. These “mutations” of the cavity therefore ensure that the nanocages studied here can be evolved by a change of Pd ligands, thereby making these supramolecular structures excellent mimics of natural enzymes. We thus view the Pd6L412+ nanocage as a prototypical “photoenzyme” (61), which can oxidize organic substrates in water with the help of molecular oxygen and can be evolved in an enzyme-like fashion.

Preorganization of Tol assists polarization of methyl C─H bonds inside a nanocage

One of the notable features of the artificial photooxidase system presented here is that the sp3 C─H bond activation occurs without any direct interaction with a metal center. However, the cage itself is made up of six Pd2+ metal ions in octahedral symmetry where each ion is coordinatively saturated within a symmetric square-planar geometry. We estimated the spatial proximity of the ─CH3 group in Tol to the Pd2+ sites in the cavity through detailed 1H-1H correlation 2D NMR experiments. As shown in Fig. 5A, the ROESY experiment provides direct evidence for spatial proximity (~5.1 Å; table S1) of the Tol CH3 protons to the β protons of the triazine pyridine rings that form the cavity walls. In addition, the aromatic protons of Tol also exhibit spatial coupling, albeit with lower strength (r ~ 5.7 to 6.7 Å), indicating that the benzene ring is closer to the central ring of the triazine wall. These spatial constraints allow us to estimate the average separation of the Tol CH3 protons from the Pd site to be >6.7 Å (see model in Fig. 5B, fig. S13, and table S1). NMR data thus provide evidence that there is no direct interaction of the sp3 C─H center with Pd2+. On the other hand, the electrostatic field generated by the six Pd2+ ions may still be strong enough to possibly prepolarize the alkyl-aromatic substrates. These strong fields can potentially influence both the π-electron density of aromatic ring and the σ density on the C─H bonds. We used Raman spectroscopy to directly probe any ground-state prepolarization effects on the substrate arising from nano-confinement in the presence of strong electric fields (E-fields). Ground-state Raman spectroscopy (λex = 532 nm) enumerated guest-specific C─H bond stretching frequencies for both the methyl sp3 C─H bond and the sp2 C─H bond in the benzene ring. In Fig. 6A, the incarcerated Tol molecules (3 ⊂ Cage) show two distinct C─H stretches at 2952 cm−1 (phenyl C─H) and 2849 cm−1 (methyl sp3 C─H), which are remarkably redshifted by 105 and 73 cm−1 respectively, relative to their values observed for bulk-free Tol. The large redshift in the C─H bond stretching frequency cannot be explained by [C─H•••O] H-bonding as these effects are limited to modest shifts of 15 to 20 cm−1 (62). However, solvent-induced E-fields can produce large Stark shifts [with a tuning rate of ~0.4 cm−1/(MV/cm)] in vibrational frequencies for the polarizable C─O bond (63). We therefore believe that the Raman data hint at an overall ground-state electrostatic polarization of the C─H bonds inside the charged cavity. For the unpolarized C─H bond, the Stark tuning rate of the cavity E-fields is expected to be much larger than that for the C─O oscillator.

Fig. 6 Ground-state polarization of the preorganized sp3 C─H bond.

(A) Steady-state Raman traces for free Tol (top, blue) and caged Tol (bottom, black) along with the fit (red line) after an excitation at 532 nm. Phenylic and methyl C─H stretches are both redshifted because of cage confinement. (B) Ground-state Raman frequency shifts (zoomed in) of methyl C─H stretches (three modes are color-coded) in the presence of E↑↑ (top, E-field parallel to the C3 axis of the Tol molecule) and E↑↓ (bottom, E-field opposite to the C3 axis of the Tol molecule) full spectra are presented in figs. S58 and S59. (C) Highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of Tol are plotted in the absence (first column) and presence of E↑↑ (at 10 and 25 V/Å) (middle pair of columns) and E↑↓ (at 5 and 25 V/Å) (last pair of columns) to show the polarization effect on the frontier orbitals due to an external E-field. a.u., arbitrary units.

E-fields inside the cavity lead to bond polarization

To capture the effect of C─H bond polarization due to static E-fields, we computed the electronic structure of a free Tol molecule under field strengths, varying from 0 to 25 V/Å and along two distinct directions. For an E-field orientation parallel (↑↑) to the phenyl-methyl bond along the ─CH3 symmetry axis, we find (Fig. 6B) that the methyl C─H stretching mode frequencies (fig. S58) undergo marked redshifts relative to their values in the absence of an E-field. The frequencies are found to increase monotonically with the field strength. The E-field–dependent shift is not mode sensitive at lower E-fields (<10 V/Å) but becomes mode sensitive at higher E-fields. For the ↑↑ E-field direction, a field strength of 10 V/Å reproduces the experimentally observed redshift of 73 cm−1 for the sp3 C─H stretches (Fig. 6B and fig. S58). We find (fig. S58C) that the character of the modes is retained (i.e., modes remain purely methyl modes) over the entire range of E-field strengths spanning from 0 to 25 V/Å. Our calculations further show that the C─H bond frequency shifts are highly sensitive to the direction of the E-field. The application of an antiparallel (↑↓) E-field to the phenyl-methyl bond, along the ─CH3 symmetry axis, leads to a blueshift in the methyl C─H stretch mode frequencies (relative to zero field values) initially at low field strengths (<5 to 6 V/Å), but it is reversed to a redshift at higher field strengths (Fig. 6B). Further analysis (fig. S59) reveals that higher ↑↓ E-field strengths (>5 to 6 V/Å) result in admixture of methyl and phenyl C─H stretch modes and produce mode-selective frequency shifts. Notably, for the ↑↓ E-field direction, higher field strengths >15 to 25 V/Å are required to produce the experimentally observed redshift (~73 cm−1). For both ↑↑ and ↑↓ E-field directions, field strengths that reproduce the experimental C─H stretch redshifts do not alter the molecular geometry (fig. S60) but rather polarize the molecular orbitals (Fig. 6C) in a direction-sensitive manner. Using a point charge description of the bare cationic nanocage, we found an E-field strength of ~5 to 6 V/Å inside the cavity at a distance of ~7 Å from the Pd2+ sites. Together, our experimental and computational investigations indicate a substantial prepolarization of the ground-state methyl C─H bonds induced due to the E-field within the cavity.

Our model of bond prepolarization inside the nanocage resembles enzyme-mediated catalysis, which is also driven by confined E-fields generated by active site residues (64). Our results above suggest that changes in the substrate ring aromaticity, which alter the delocalization of molecular orbitals, may further tune the prepolarization of the methyl C─H bonds by E-fields. To examine the effect of the aromatic ring structure on the E-field–induced shifts in methyl C─H stretch frequencies, we compared ↑↑ E-field–induced (5 to 15 V/Å) shifts on mode frequencies of 9-MA and Tol (fig. S61). We find that the change in substrate aromaticity distinctly alters the C─H stretch mode frequency shifts due to the applied E-field. While the ↑↑ E-field–induced shifts in methyl C─H stretches for Tol are not mode sensitive (fig. S61, D and E), the ↑↑ E-field–induced shifts for 9-MA show pronounced mode sensitivity. For instance, all three Tol sp3 C─H stretch modes show experimentally relevant redshifts (~73 cm−1) at E-field strengths of ~10 V/Å. In contrast, three C─H stretches of methyl in 9-MA show distinct E-field–dependent shifts (fig. S61B). While two of the modes (blue and green) show a redshift (~73 cm−1) at E-field strengths <10 V/Å, the third mode (red) shows the similar shifts at values >10 V/Å. The different aromaticity of 9-MA and Tol leads to different extents of E-field–induced MO polarizations (fig. S61F and Fig. 6C), which tune the response of the methyl C─H stretch frequencies and therefore the ground-state prepolarization of C─H bonds by the cavity E-fields. It has not escaped our attention that, apart from E-fields, the polarization induced due to emergence of the new host-guest CT states should also affect the C─H bond stretching frequency.


The cationic nanocage presented here is a prototypical photooxidase that mimics natural enzymes by encapsulating organic substrates in water and photocatalytically generates oxidized products in the presence of molecular oxygen. Using the nanocage, we have developed a modular general approach for redox-activated reactions, which can be diversified by a suitable and rational choice of guest and its complementary host. In addition, our work enunciates the critical role of E-fields in driving reactions under nano-confinement. Tailoring the E-field features should enable selective bond polarization in complex polyatomic molecules trapped inside nano-hosts. The success of our prototypical photoenzyme is further highlighted by our ability to tune cavity electronics along with molecular packing to control both catalytic reactivity and selectivity. We anticipate that our paradigm will lead to the development of new prototypical photoenzymes for redox-activated C─C or C─N bond coupling reactions and that, through these green methodologies, the environmental impact of chemical industries can be alleviated.



9-MA (98%), 1-MeNap, ethylbenzene (anhydrous, 99.8%), isopropylbenzene (cumene), para-substituted Tol derivatives, and 2,4,6-tri(4-pyridyl)-1,3,5-triazine were purchased from Sigma-Aldrich and TCI Chemicals Pvt. Ltd., respectively. We used high-performance liquid chromatography–grade (HPLC) Tol (99.7% purity) for all the photoreactions. Solvents and other reagents were purchased from SD Fine Chemicals and Sigma-Aldrich. Deuterated water (D2O; 99.9 atom %D) was procured from Sigma-Aldrich. The Pd6L4 cage was synthesized following the reported protocol and discussed in supporting information. All the incarceration strategies are discussed in the Supplementary Materials.

Resonance Raman spectroscopy

For resonance Raman measurements, we used an excitation wavelength of 532 nm, which was generated from a frequency-doubled diode-pumped solid-state (DPSS) yttrium-aluminum-garnet–Nd (Nd:YAG) laser (WITec) and coupled to the Alpha 300R confocal Raman microscope [WITec GmbH, Ulm (Germany)]. A 100-μm optical fiber was used to collect the backscattered light where a lens-based ultrahigh throughput spectrometer (UHTS300) with 1800 grooves/mm grating was used and coupled to a back-illuminated charge-coupled device camera (CCD, 1024 × 128 pixels, Peltier-cooled to −65°C) for detection. The spectral resolution of the spectrograph was ∼2 cm−1. The laser was focused into the solution flowing through a flow cuvette (path length is 2 mm) using a 10× objective. The power of the laser source (532 nm) was 3 to 4 mW at the sample stage.

Transient absorption measurements

All pump-probe measurements were executed using an ultrafast transient absorption spectrometer, located in the ultrafast biophysics and photomaterials laboratory of the Department of Chemical Sciences, Tata Institute of Fundamental Research, India (65). An oscillator output with a center wavelength of 810 nm, a bandwidth of ∼100 nm, and a repetition rate of 80 MHz was amplified 106 times using a commercial regenerative amplifier. The output of the amplifier was 30 fs/4 mJ per pulse, with a repetition rate of 1 kHz and a bandwidth of ∼65 nm. We generated a 400-nm pump by frequency doubling a portion of the 810-nm output from the amplifier from a beta barium borate (BBO) crystal. The amplifier output was directed to an optical parametric amplifier (Coherent OPerA Solo Ultrafast Optical Parametric Amplifier system) for generation of the 490-nm pump pulse. For measurements, a pump beam was attenuated to ∼200 to 400 nJ per pulse. The white-light broadband probe continuum (440 to 1400 nm) was generated by focusing a portion of the amplified 810-nm output on a 2-mm-thick sapphire and then directing it to a multichannel detector. The pump and probe pulses were focused and overlapped spatially and temporally within the sample cuvette. For measurements, we used a flow cuvette of 2-mm path length. Kinetic fitting of the raw data traces were performed using a mathematical program for deconvoluting the time constants from the recorded instrument response function (IRF) in IGOR Pro 5 software. Global and target analyses of the transient absorption data were carried out with the help of the Surface Xplorer and Glotaran software (66).

Photoreaction on host-guest complexes

We used commercial cheap LEDs (photon flux, ~80 to 100 mW) for all the photoreactions. Then, a green LED (λ = 530 nm) for 9-MA ⊂ Cage and a blue LED (λ = 460 nm) for 1-MeNap ⊂ Cage were illuminated for 1.5 and 7.5 hours, respectively, in a side illumination fashion (distance of the LED from the vial was kept at ~5 cm). For Tol ⊂ Cage, we used a round bottom flask connected with a water condenser to avoid any loss of Tol because of its volatile nature, and an O2 balloon (pressure is ~1.5 atm) was attached at the top of the condenser. Then, we illuminated with a blue LED (λ = 400 nm) for 8 to 40 hours depending on whether it is a stoichiometric photoreaction or a photocatalytic reaction. After completion of the reaction, the aqueous solution was extracted with an organic solvent, dichloromethane (CH2Cl2). The extracted photoproducts were either purified by column chromatography or analyzed by GC-MS, specifically for Tol and its derivatives.

Molecular modeling of Tol inside the Pd6L412+ cage to estimate distances between substrate methyl C─H bonds and cage Pd2+ centers

The initial coordinates of the Pd6L412+ (Pd, ethylenediamine-palladium) cationic cage were obtained from the Cambridge Crystallographic Data Centre (CCDC number 277006) and optimized using density functional theory (DFT) at a B3LYP/Lanl2dz/6-31G* level of theory. Then, a single Tol molecule was placed close to one of the four triazine moieties of the optimized cationic cage using the distance constraints (see table S1) obtained from 2D ROESY NMR experiments (Fig. 5A). The Tol-cage complex was then minimized in water with the polarizable continuum model (PCM) (67), keeping the cage structure fixed and using the same level of theory mentioned above. For the resultant optimized Tol-cage complex, average distances between all hydrogen atom pairs of Tol and cage and between Tol H and Pd atoms were calculated and compared with NMR ROESY experimental data (see section S1.3).

Estimating the effect of E-fields on the ground-state alkyl C─H Raman stretch frequencies for Tol and 9-MA

We calculated Raman frequencies for Tol and 9-MA molecules in the presence of E-fields of varying strengths and directions to estimate E-field–induced prepolarization of methyl C─H bonds inside the charged Pd cage. The E-field strengths were varied from 0.05 to 26 V/Å, and the fields were applied along the methyl group symmetry axis, either parallel (↑↑) or antiparallel (↑↓) to the methyl-phenyl bond. For 9-MA, we carried out calculations only with the ↑↑ E-field. First, structures were optimized using the DFT method with the 6-31G* basis set on all atoms, and the B3LYP hybrid exchange-correlational functional and vibrational Raman frequencies were calculated at the same level of theory in the presence of E-fields. All calculations were performed in the Gaussian 09 program (68). Further details of the calculations and analysis of calculated Raman modes as a function of E-field strength are provided in section S1.3.


Supplementary material for this article is available at

Section S1. Materials and methods

Fig. S1. Calibration distance for 2D ROESY NMR.

Fig. S2. Octahedral nanocage.

Fig. S3. 1H NMR spectrum of a synthesized Pd6L412+ nanocage in D2O.

Fig. S4. 13C NMR spectrum of a synthesized Pd6L412+ nanocage.

Fig. S5. Visible charge transfer complex of 9-MA.

Fig. S6. Host-guest complexation and NMR upfield shifts.

Fig. S7. Temperature dependence of population heterogeneity of caged 9-MA.

Fig. S8. Charge transfer complex with 1-MeNap.

Fig. S9. Host-guest complexation of 1-MeNap.

Fig. S10. Temperature-dependent 1H NMR of methyl protons.

Fig. S11. Complexation of Tol and nanocage.

Fig. S12. 13C NMR spectrum of Tol ⊂ Cage.

Fig. S13. Optimized structure of Tol inside a nanocage.

Fig. S14. Substituent-dependent steady-state absorption of host-guest CT complexes of Tol derivatives.

Fig. S15. Formation of ortho-xylene ⊂ Cage complex.

Fig. S16. 2D ROESY spectrum of ortho-xylene ⊂ Cage complex in D2O in 800 MHz.

Fig. S17. Formation of para-xylene ⊂ Cage complex.

Fig. S18. ROESY spectrum of para-xylene ⊂ Cage complex in D2O (800 MHz).

Fig. S19. Formation of meta-xylene ⊂ Cage complex.

Fig. S20. 1H-1H 2D ROESY NMR spectrum of meta-xylene ⊂ Cage complex in D2O in 800 MHz.

Fig. S21. Complexation of para-OMeTol with nanocage.

Fig. S22. 2D 1H-1H ROESY NMR spectrum of para-OMeTol ⊂ Cage complex in D2O.

Fig. S23. Complexation of para-FluoroTol with nanocage.

Fig. S24. 2D 1H-1H ROESY NMR spectrum of para-FluoroTol ⊂ Cage complex in D2O (800 MHz).

Fig. S25. Electronics- and regioisomerism-dependent molecular packing of Tol derivatives inside the nanocage.

Fig. S26. 1H NMR spectra of Tol ⊂ Bipycage (top) in D2O and its comparison with Tol ⊂ Encage in D2O.

Fig. S27. 2D 1H-1H ROESY NMR spectrum of Tol ⊂ Bipycage complex in D2O (800 MHz).

Fig. S28. Transient absorption of radical cation of 9-MA ⊂ Cage and corresponding decay.

Fig. S29. PCET reaction for 9-MA ⊂ Cage complex.

Fig. S30. Substrate isotope dependence on C─H vs. C─D bond cleavage and primary KIE.

Fig. S31. C─H versus C─D bond cleavage and primary KIE.

Fig. S32. Tol radical cation decay inside the nanocage.

Fig. S33. Excited-state absorption of coupled host-guest CT state.

Fig. S34. Solvation relaxation of the excited host-guest CT state.

Fig. S35. Photoreaction on encapsulated 9-MA, illuminated by cheap green LED (530 nm).

Fig. S36. Rate of photoreaction is dependent on O2 concentration.

Fig. S37. Photooxidized product identification.

Fig. S38. Both benzylic and ring oxidation take place for 9-MA.

Fig. S39. Photoreaction under high O2 pressure for 30 min.

Fig. S40. UV light–mediated photoreaction under ambient O2.

Fig. S41. Photocatalysis on caged 9-MA.

Fig. S42. Control photoreactions for mechanism elucidation.

Fig. S43. Selective photooxidation products of 1-MeNap.

Fig. S44. Spectral characteristics of the photoproducts.

Fig. S45. Mass spectrum of photooxidation product, benzaldehyde.

Fig. S46. Concomitant decay of the substrate and rise of the oxidized product with a lag phase.

Fig. S47. Benzyl alcohol to benzaldehyde formation.

Fig. S48. Photooxidation of secondary benzylic radicals.

Fig. S49. Photoreactivity of cage incarcerated tertiary benzylic radicals under O2.

Fig. S50. Photooxidation of para-xylene inside the nanocage.

Fig. S51. Ortho-xylene oxidation by light.

Fig. S52. Photocatalytic oxidation of meta-xylene.

Fig. S53. Oxidation of para-OMeTol with blue light.

Fig. S54. Electron-withdrawing substituent effect on photocatalytic oxidation of para-FluoroTol.

Fig. S55. Photocatalytic oxidation rates are strongly dependent on molecular packing of the substrates inside the cage.

Fig. S56. Cage electronics perturb the photocatalytic reaction rates.

Fig. S57. Cage behavior like a prototypical photoenzyme: Tuning photoproduct selectivity by changing the cage electronics.

Fig. S58. Parallel E-field and C─H Raman stretches.

Fig. S59. Antiparallel E-field effects on C─H Raman stretches.

Fig. S60. E-field effects on C─H bond lengths.

Fig. S61. Effects of aromaticity on E-field induced C─H bond polarization.

Scheme S1. Making of the nanocage.

Scheme S2. Plausible mechanism for photooxidation of 9-MA inside cage.

Scheme S3. Oxidation of 1-MeNap by light.

Table S1. Comparison between 2D NMR and molecular modeling–obtained relative distances.

Table S2. Geometric separation of Tol methyl protons from the triazine walls for different derivatives.

References (6972)

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We acknowledge A. Datta and her students for help with the electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) mass spectroscopic measurements and the TIFR National NMR facility, M. V. Joshi, and D. Jadhav for the help with the NMR measurements. We also acknowledge V. Polshettiwar and M. Dhiman for the help in the GC-MS measurements. We acknowledge P. Roy for help in a few transient absorption experiments, suggestions, and fruitful discussions. We thank S. Ghosal for help with 2D NMR experiments and valuable inputs in the manuscript. In addition, we value suggestions from R. Gera, A. Jha, K. Vijayalakshmi, and S. Paul. We also acknowledge the suggestions from S. Karmakar and A. Ambastha while editing the manuscript. Funding: J.D. and R.V. thank the Tata Institute of Fundamental Research and the Department of Atomic Energy (DAE), India for financial support. Author contributions: J.D. conceived the research, while A.D. and J.D. planned the experiments. A.D. performed the experiments and analyzed the experimental data with contributions from J.D. The electronic structure calculations were performed by I.M. and R.V. The manuscript was written by J.D. and A.D. with substantial contributions from I.M. and R.V. All the authors discussed the scientific content in detail and agreed with the final form of the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article