Criegee intermediate inside fullerene cage: Evidence for size-dependent reactivity

Fullerenes are promising candidates for catalysis. The inner spherical radii of the fullerenes C₆₀ and C₇₀ are 3.7 and 4.6 Å, which are large enough to encapsulate atoms and molecules, and provide potential environments for endohedral chemistry.^1,2 In recent years, the arc-discharge methods and the emergence of a unique synthetic approach, molecular surgery,^3,4 have allowed the synthesis of fullerene complexes that can encapsulate a wide variety of species, including metal ions,^5,6 noble gases,^7,8 molecular hydrogen,^9,10 water molecules,^11,12 hydrogen fluoride (HF),¹³ and nitrogen atoms.¹⁴ These complexes have attracted a lot of attention and have been the focal point of numerous experimental and theoretical studies. The synthetic feasibility of encaging small molecules inside the sub-nanospace of fullerenes raises a mechanistically appealing fundamental question: how does the nanoscale confinement influence the properties and reactivity of the enclosed species?

The available limited experimental and theoretical data suggest that the cavity inside fullerenes provides a unique environment for chemical reactivity.^10,15–24 The confined molecules display the quantization of their coupled translational and rotational degrees of freedom¹⁵ and exhibit phenomena such as nuclear spin isomerism¹⁶ and ortho-para conversion.^10,17 The nuclear spin conversion of the caged water molecules in H₂O@C₆₀ impacts the dielectric constant of the material.¹⁸ The rotational and vibrational constants of the encapsulated HF molecules are red-shifted relative to free HF.¹³ The ¹³C resonance for Xe@C₆₀ is shifted downfield by 0.95 ppm.^19 129Xe NMR showed a line 179.2 ppm downfield from xenon gas. The free-energy barrier at 298 K for the hydrogen exchange reaction 3H₂ → 3H₂ is reduced from 88.8 kcal/mol for the free reaction to 36.2 kcal/mol for the reaction within C₆₀, reflecting at least 3 orders of magnitude rate enhancement due to the encagement.¹⁷ The presence of noble gas atoms or molecular hydrogen influences the equilibrium constant for a Diels-Alder (DA) reaction between 9,10-dimethylanthracene and C₆₀ or C₇₀.²⁰ 

The insertion of metal ions into the fullerene cage is also found to greatly promote the DA cycloadditions to C₆₀. Matsuo experimentally studied the DA cycloaddition of 1,3-cyclopentadiene and 1,3-cyclohexadiene to Li⁺@C₆₀.^21,22 These reactions were found to be 1000- and 2700-fold faster than the corresponding additions to the 6–6 bond of empty C₆₀, respectively.^21,22 The size of a fullerene cage also plays an important role in modifying chemical behavior inside a given cage. For example, the shape of the inner metal carbide (M₂C₂) cluster in endohedral fullerenes is found to be influenced by the size of a fullerene cage.²³ Dorn and co-workers suggested that decreasing the size of the fullerene would compress M₂C₂ clusters from a nearly linear, stretched geometry to a constrained butterfly structure.²³ Deng and Popov also predicted similar cage-dependent behavior of M₂C2 units.²⁴ Though the properties and intramolecular chemistry of the encapsulated single atoms or molecules have been studied previously, the bimolecular reactivity inside the fullerene cavity is yet to be explored.

We report the results of quantum chemical calculations and Born-Oppenheimer molecular dynamics (BOMD) simulations on the reaction between Criegee intermediates and water inside fullerene cages of different sizes. Criegee intermediates are carbonyl oxides that are key branching points in the ozone-olefin cycloadditions²⁵ and are important because of their role in the nighttime chemistry of the OH radical in the troposphere. Ever since Criegee intermediates have directly been detected in the gas-phase,^26,27 interest in their chemistry has grown exponentially.²⁸ Criegee intermediates are not only atmospherically important but are also relevant in general synthetic organic reactions en route to value-added chemicals.^29,30 In the results reported here, we show that the hydration of a Criegee intermediate in different sized fullerene cages occurs differently and provides the compelling evidence of a size-dependent reactivity within the sub-nanospace of a fullerene. These results not only offer deeper mechanistic insights into the effect of fullerene-cage confinement on the Criegee chemistry but should also open up new avenues of experimental and theoretical studies for studying other bimolecular processes inside fullerene cages.

In the first step, the gas-phase reaction of the simplest Criegee intermediate, CH₂OO with H₂O inside fullerenes C₆₀, C₇₀, C₈₄, and C₁₂₀ was examined at the density functional theory (DFT) level. The geometries of all the stationary points on the potential energy surfaces were fully optimized using the M06-2X³¹ density functional theory method and cc-pVDZ and cc-pVTZ³² basis sets. All fullerene calculations were done at the M06-2X/cc-pVTZ level except for the C₁₂₀ ones that were done at the M06-2X/cc-pVDZ level. These gas-phase calculations were performed using Gaussian 09³³ software and at a standard state of 298.15 K and 1 atm. The Cartesian coordinates of key optimized species are provided in the supplementary material document.

In the second step, the BOMD simulations were performed on the basis of DFT methods as implemented in the CP2K³⁴ code. In the BOMD simulations, the fullerenes C₇₀, C₈₄, and C₁₂₀ and one CH₂OO molecule were considered. The dimensions of the simulation box are x = 35 Å, y = 35 Å, and z = 35 Å, which are large enough to avoid interactions between adjacent periodic images of fullerene cages. Prior to the BOMD simulations, the system was fully relaxed using the DFT method. For optimizing the system at the DFT level, the exchange and correlation interactions of electrons are treated with the Becke-Lee-Yang-Parr (BLYP)^35,36 functional, and Grimme’s dispersion correction method³⁷ is applied to account for the weak dispersion interaction. A double-ζ Gaussian basis set combined with an auxiliary basis set³⁸ and the Goedecker-Teter-Hutter (GTH) norm-conserved pseudopotentials^39,40 is adopted to treat the valence electrons and the core electrons, respectively. An energy cutoff of 280 Ry was set for the plane wave basis set and 40 Ry for the Gaussian basis set. All the BOMD simulations were performed in the constant volume and temperature (NVT) ensemble, with the Nose-Hoover chain method for controlling the temperature of the system. The integration step was set as 1 fs, which had been proven to achieve sufficient energy conservation for the water system. All the BOMD simulations were performed at a constant temperature of 300 K.

First, we examined the calculated structure and energetics of the C₆₀ C₇₀, C₈₄, and C₁₂₀ encapsulated CH₂OO (CH₂OO@C₆₀, CH₂OO@C₇₀, CH₂OO@C₈₄, and CH₂OO@C₁₂₀). The equilibrium geometries of the encapsulated CH₂OO are shown in Fig. 1. The comparative analysis of CH₂OO with and without a fullerene cage (Table S1 of the supplementary material) reveals that the encapsulation does not induce any major structural change in the Criegee intermediate. The calculated interaction energy for CH₂OO@C₆₀ is +15.8 kcal/mol (Fig. 1) suggests that the encaging of CH₂OO inside C₆₀ is energetically not favorable. On the other hand, the calculated interaction energies for CH₂OO@C₇₀, CH₂OO@C₈₄, and CH₂OO@C₁₂₀ are −27.6, −24.8, and −21.0 kcal/mol, respectively. These interaction energies are at least 36.8 kcal/mol more favorable than that for CH₂OO@C₆₀, indicating that the encapsulation in a larger fullerene cage (>C₆₀) stabilizes CH₂OO, and the laboratory synthesis of the endohedral complexes of C₇₀, C₈₄, and C₁₂₀ with Criegee intermediates might be achievable. We also examined the effect of C₇₀ confinement on the larger C₂ Criegee intermediates. Interestingly, the binding energy of anti-CH₃CHOO@C₇₀ is +25.3 kcal/mol (Fig. S1 of the supplementary material), suggesting that the encapsulation of anti-CH₃CHOO inside the C₇₀ cage is strongly disfavored. On the other hand, the encapsulation energy of syn-CH₃CHOO is only 6.1 kcal/mol above the thermoneutral zero (Fig. S1 of the supplementary material), suggesting that selecting a larger fullerene cage will make the endohedral chemistry of syn-CH₃CHOO feasible. This stark difference in the binding preference of two C₂ conformers toward fullerene C₇₀ is due to the presence of an intramolecular hydrogen bond between the terminal Criegee oxygen and the —C—H fragment of the syn-methyl group in syn-CH₃CHOO@C₇₀ that avoids steric clashes between Criegee intermediates and the walls of the fullerene cage. It is important to mention here that the presence of an intramolecular hydrogen bonding in syn-CH₃CHOO makes it more stable in the gas-phase than anti-CH₃CHOO. Specifically, Anglada et al. predicted syn-CH₃CHOO to be 3.49 kcal/mol more stable than anti-CH₃CHOO in the gas-phase at the CCSD(T)/aug-cc-pVTZ//B3LYP/6-311 + G(2df,2p) level of theory.⁴¹ Though this intramolecular hydrogen bond has traditionally been used to explain the slower gas-phase reactivity of syn-CH₃CHOO relative to anti-CH₃CHOO,^41,42 the results here suggest that this interaction could be utilized to guide the endohedral chemistry of larger Criegee intermediates inside the fullerene cages.

Considering that the C₇₀ or larger fullerene cage encapsulation of the simplest Criegee intermediate CH₂OO is favorable, we next examined the plausible gas-phase reaction between CH₂OO and H₂O inside C₇₀, C₈₄, and C₁₂₀ cages. Despite our attempts to optimize the (CH₂OO⋯H₂O)@C₇₀ and (CH₂OO⋯H₂O)@C₈₄ complexes using various starting geometries, the global minimum for the caged CH₂OO⋯H₂O complex could not be characterized. Instead, all the starting geometries converged to the final product of the reaction, α-hydroxy methyl hydroperoxide (HMHP, (HO)CH₂(OOH)). The formation of HMHP from CH₂OO and H₂O inside C₇₀ and C₈₄ cages has reaction energies of −61.3 and −82.6 kcal/mol, respectively. That was quite surprising, yet an interesting outcome considering the fact that the CH₂OO⋯H₂O → (HO)CH₂(OOH) reaction in the gas-phase involves an appreciable barrier of ∼6.9 kcal/mol at the M06-2X/cc-pVTZ level of theory and has a reaction energy of −55.9 kcal/mol (Fig. 2). It is important to mention here that though the previously calculated data for the gas-phase CH₂OO⋯H₂O → (HO)CH₂(OOH) reaction is available at relatively higher levels of theory, we herein have calculated the gas-phase reaction profile at the M06-2X/cc-pVTZ level of theory to allow comparison between the gas-phase and caged reactions at the same theoretical footing. For example, Anglada et al. have studied the same reaction at the CCSD(T)/aug-cc-pVTZ//B3LYP/6-311 + G(2df,2p) level of theory.⁴¹ Though the M06-2X/cc-pVTZ calculated energies of individual stationary points for the gas-phase CH₂OO⋯H₂O reaction are significantly different from those previously calculated at the CCSD(T)/aug-cc-pVTZ//B3LYP/6-311 + G(2df,2p) theoretical level, the M06-2X/cc-pVTZ calculated barrier of 6.9 for the gas-phase CH₂OO⋯H₂O reaction is in reasonable agreement with the CCSD(T)/aug-cc-pVTZ//B3LYP/6-311 + G(2df,2p) calculated value of 7.5 kcal/mol. On the other hand, the (CH₂OO⋯H₂O)@C₁₂₀ complex could be characterized at the M06-2X/cc-pVDZ level (Fig. 3), suggesting that the facile formation of HMHP inside C₇₀ and C₈₄ cages is due to the confinement effects inside fullerene cages. Note that we have used a slightly smaller basis set of cc-pVDZ for the C₁₂₀ system due to its relatively larger size as compared to C₆₀, C₇₀, and C₈₄ systems.

The (CH₂OO⋯H₂O)@C₁₂₀ complex is 18.8 kcal/mol more stable relative to CH₂OO@C₁₂₀ and H₂O. The transition state lies 4.4 kcal/mol above the (CH₂OO⋯H₂O)@C₁₂₀ complex. The formation of HMHP@C₁₂₀ has a reaction energy of −63.0 kcal/mol relative to CH₂OO@C₁₂₀ and H₂O and has a reaction energy of −84.0 kcal/mol relative to separated CH₂OO, H₂O, and C₁₂₀, respectively. The HMHP@C₁₂₀ complex is 24.3 kcal/mol more stable than the separated C₁₂₀ and HMHP. It is interesting to compare the results of the Criegee hydration in the gas-phase with those inside the C₁₂₀ cage. The CH₂OO⋯H₂O complex in the gas-phase is 13.6 kcal/mol more stable than separated CH₂OO and H₂O. Inside the C₁₂₀ cage, the same complex is 39.8 kcal/mol more favorable than the starting precursors, which translates into a significant stabilization of 26.2 kcal/mol because of the nanoconfinement. The M06-2X/cc-pVDZ calculated barrier for the Criegee hydration in the gas-phase is 5.3 kcal/mol, which is 0.9 kcal/mol larger than that inside the C₁₂₀ cage. Although the barrier lowering due to the fullerene-cage confinement is not dramatic, the encapsulation energy of CH₂OO@C₁₂₀ or (CH₂OO⋯H₂O)@C₁₂₀ complex is significantly favorable, which should make this reaction occur rapidly under fullerene confinement. It is also important to point out that the product of Criegee hydration, HMHP, is better stabilized inside a larger cage, i.e., the HMHP@C₇₀ complex is only 5.4 kcal/mol more stable than free HMHP, whereas the HMHP@C₈₄ and the HMHP@C₁₂₀ complexes are 26.7 and 24.3 kcal/mol, respectively, more stable than free HMHP.

In order to deeply understand the Criegee reactivity inside a fullerene cage (>C₆₀), we performed BOMD simulations at 300 K. The BOMD simulations are a powerful computational approach that offers useful insights into the dynamic behavior of a molecular system at a given temperature and allows the study of a chemical reaction as a dynamic time-dependent process. Specifically, the BOMD simulations were performed to better understand the dynamic behavior of Criegee intermediates inside C₇₀, C₈₄, and C₁₂₀ cages. The simulation results show that CH₂OO inside these fullerene cages remain dynamically stable for the 20 ps duration (Movies S1–S3 of the supplementary material). Compared to the gas-phase, the key C—O and O—O bond distances of CH₂OO inside the C₇₀, C₈₄, and C₁₂₀ cages undergo minor elongations of 0.05-0.10 Å during the simulated time (Fig. 4). No major deformation in the structure of the fullerene cage occurs during simulations, suggesting that the fullerene cage provides a stable framework for reactivity.

The calculated radial distribution functions (RDFs) for the encapsulated CH₂OO provide useful information about its spatial distributions inside C₇₀, C₈₄, and C₁₂₀ cages. Figure 5 displays the RDFs(R) for CH₂OO@C₇₀, CH₂OO@C₈₄, and CH₂OO@C₁₂₀ from the BOMD simulations; R here is the distance from the center of mass of CH₂OO to the carbons of a fullerene cage. The results show that the CH₂OO RDFs inside C₇₀ and C₈₄ cages have relatively narrower distributions and the peaks for these RDFs are centered around 4.0 Å, which is close to the interior radius of these cages, whereas the CH₂OO RDF inside the C₁₂₀ cage has a relatively broad distribution with a peak around 5.0 Å. Overall, the most probable distribution of Criegee intermediates inside a fullerene cage resides at a distance of 4.0–5.0 Å from the nearest carbons of the fullerene cage. We next calculated the mean square displacements (MSDs) of CH₂OO inside these cages at 300 K. The calculated MSDs are shown in Fig. 5. The results suggest that the Criegee intermediates inside C₇₀ or C₈₄ is less mobile, whereas it exhibits relatively higher mobility inside the C₁₂₀ cage, which is consistent with the size of a fullerene cage.

We also examined the dynamics of HMHP inside C₇₀, C₈₄, and C₁₂₀ fullerene cages. The simulation results show that HMHP inside these cages remain dynamically stable for the 20 ps duration (Movies S4–S6 of the supplementary material). Figure 6 displays the RDFs(R) for HMHP@C₇₀, HMHP@C₈₄, and HMHP@C₁₂₀ from the BOMD simulations. The dynamics behavior of HMHP inside these cages is similar to that of Criegee intermediates, i.e., the RDFs for HMHP inside C₇₀ and C₈₄ cages have relatively narrower distributions and the peaks for these RDFs are centered around 3.7 and 4.3 Å, respectively, whereas the HMHP RDF inside the C₁₂₀ cage has a broader distribution with a peak around 5.0 Å. We also calculated the MSDs of HMHP inside these cages at 300 K. The calculated MSDs are shown in Fig. 6. Just like Criegee intermediates, HMHP inside the C₇₀ or C₈₄ cage is less mobile, whereas it exhibits relatively higher mobility inside the C₁₂₀ cage, which is consistent with the size of a fullerene cage. Interestingly, HMHP inside C₇₀ and C₈₄ cages is relatively less mobile than CH₂OO. But inside the larger cage of C₁₂₀, HMHP is relatively more mobile than CH₂OO. This is because HMHP is larger than CH₂OO, which makes it less mobile inside smaller cages of C₇₀ and C₈₄. The mobility of HMHP inside C₁₂₀ is not adversely impacted because C₁₂₀ is larger in size than C₇₀ or C₈₄. Overall, these new results may open up new avenues of experimental and theoretical research for trapping and harnessing the chemical reactivity inside fullerene cages.

In conclusion, our density functional theory calculations show that the hydration of CH₂OO within a fullerene cage (>C₆₀) occurs favorably, suggesting the role of space confinement in chemical reactivity. The Born-Oppenheimer molecular dynamics simulations indicate that CH₂OO and the product of its hydration, α-hydroxy methyl hydroperoxide, remain dynamically stable for the 20 ps time scale at 300 K and their self-diffusion coefficients inside fullerene cages correlate with the size of a cage. These results reveal that the confinement of the fullerene cage provides a unique environment for executing bimolecular chemistries. The conclusions from this study should serve as useful and new leads for studying other bimolecular chemical processes inside fullerene cages.

See supplementary material for the calculated equilibrium geometries and interaction energies of anti-CH₃CHOO and syn-CH₃CHOO encapsulated C₇₀ fullerene, calculated structural parameters of the uncaged and caged Criegee intermediates, Cartesian coordinates of key species optimized at M06-2X/cc-pVTZ and M06-2X/cc-pVDZ levels of theory, and trajectories of the BOMD simulations for the simplest Criegee intermediate as well as for α-hydroxy methyl hydroperoxide inside C₇₀, C₈₄, and C₁₂₀, respectively.

This work is supported by the National Science Foundation (Grant No. CHE-1500217) and by the University of Nebraska Holland Computing Center.