Ruthenium-Catalyzed [2 + 2] versus Homo Diels-Alder [2 + 2 + 2] Cycloadditions of Norbornadiene and Disubstituted Alkynes: A DFT Study.

The ruthenium-catalyzed [2 + 2] and homo Diels-Alder [2 + 2 + 2] cycloadditions of norbornadiene with disubstituted alkynes are investigated using density functional theory (DFT). These DFT calculations provide a mechanistic explanation for observed reactivity trends with different functional groups. Alkynyl phosphonates and norbornadiene form the [2 + 2 + 2] cycloadduct, while other functionalized alkynes afford the respective [2 + 2] cycloadduct, in excellent agreement with experimental results. The computational studies on the potential energy profiles of the cycloadditions show that the rate-determining step for the [2 + 2] cycloaddition is the final reductive elimination step, but the overall rate for the [2 + 2 + 2] cycloaddition is controlled by the initial oxidative cyclization. Two distinct mechanistic pathways for the [2 + 2 + 2] cycloaddition, cationic and neutral, are characterized and reveal that Cp*RuCl(COD) energetically prefers the cationic pathway.

Introduction

Fused bicyclo[2.2.1]alkenes constitute a diverse class of chemical compounds and are powerful synthetic intermediates for the formation of complex, highly substituted ring systems.[1a−1e] Among these, bicyclo[2.2.1]hepta-2,5-diene, colloquially known as norbornadiene 1 (NBD), has been used as a key intermediate in the total synthesis of natural products such as prostaglandin endoperoxides PGH2 and PGG2,[2]cis-Trikentrin B,[3] and β-santalol.[4] Photochemical isomerization between NBD derivatives and quadricyclanes is of interest as a potential photoswitchable molecule that can undergo closed cycles of solar light harvesting and energy storage and release.[5a−5f] Because of the rigid bicyclic structure of NBD, there is significant ring strain energy in this molecule.[6] In addition, the presence of two isolated double bonds homoconjugated through space[7] along with two distinct faces (exo and endo) allows for unique reactivity not available in conventional alkenes.[8] NBD’s inherent ring strain and dual-faced nature can be exploited for the facile establishment of multiple stereocenters in a single reaction by way of face-selective olefin chemistry (Scheme , 2)[9a−9g] or ring-opening reactions (Scheme , 9 and 7).[10a−10e] Of particular interest are transition-metal-catalyzed cycloadditions, as they are among the most powerful and frequently used methods for the construction of ring systems.[11] Recent developments in transition-metal-catalyzed [2 + 1],[12] [2 + 2] (Scheme , 8),[13a−13d] [2 + 2 + 1] (Scheme , 5),[14a−14e] [2 + 3] (Scheme , 4),[15a−15e] [2 + 2 + 2 + 2] (Scheme , 3),[16a−16c] and [2 + 2 + 6] (Scheme , 6)[17] reactions have provided efficient methods for the production of 3–8 membered rings.

Scheme 1

Previously Reported Chemistry of Norbornadiene

The development of ruthenium-catalyzed transformations has burgeoned over the past decade.[18a−18e] Among the various ruthenium complexes, Cp*RuCl(COD) has been found to be the catalyst of choice in many reactions such as [2 + 2 + 2] cycloadditions,[19a−19c] hydrogenations,[20a−20c] cross-benzannulations,[21a−21e] and Alder-ene reactions.[22a,22b] Our group, along with others, have been largely involved in the construction of cyclobutene adducts via ruthenium-catalyzed [2 + 2] cycloadditions.[23a−23u] In 2011, our group described the ruthenium-catalyzed homo Diels–Alder (HDA) [2 + 2 + 2] cycloadditions of alkynyl phosphonates (10, X = phosphonate) with NBD 1 to afford phosphonate-substituted deltacyclenes 12 (Scheme , right).[24] In agreement with previously reported ruthenium-catalyzed cycloadditions of alkynes with bicyclic alkenes, a quick screen of substituted acetylenes (10, X ≠ phosphonate) resulted in the exclusive formation of [2 + 2] cycloadducts 11 (Scheme , left), suggesting that HDA [2 + 2 + 2] cycloaddition is unique to the electronics of phosphonate-substituted alkynes when Cp*RuCl(COD) is used as the precatalyst (Scheme ).

Scheme 2

[2 + 2] and HDA [2 + 2 + 2] Cycloadditions of Disubstituted Alkynes 10 and NBD 1 Using Cp*RuCl(COD) as the Catalyst

Here, we pursue an in-depth investigation of the reaction mechanism by performing density functional calculations to map out the ruthenium-catalyzed [2 + 2] and HDA [2 + 2 + 2] cycloadditions of norbornadiene and alkynyl phosphonates. The results provide theoretical support on the chemoselectivity of the experimentally observed Ru-catalyzed cycloadducts.

Computational Details

All computations in this study were carried out with the Gaussian 16 C.01 suite of programs.[25] Geometry optimizations of all of the intermediates and transition states were carried out at the Becke three-parameter hybrid functional[26] combined with the Lee, Yang, and Parr (LYP) correlation functional,[27] B3LYP, with the double-ζ basis set def2SVP[28] and Grimme’s dispersion with Becke–Johnson damping (GD3BJ).[29] The optimization was carried out without imposing any symmetry constraints. Solvent effects (solvent = N,N-dimethylformamide or DMF) were taken into account using the polarized continuum model (PCM) of Tomasi and co-workers[30a,30b] and were involved in all geometry optimization and frequency calculations. Although N-methyl-2-pyrrolidone (NMP) was experimentally determined to be the optimal solvent for the reaction, DMF was chosen to be the solvent of choice in the calculations. NMP is not implemented in Gaussian and the physical parameters for NMP are not published in Cramer and Truhlar’s solvent database.[31a,31b] Given the similarity in structure and dielectric constant (DMF ε = 37.2; NMP ε = 33), we considered this a good approximation. Harmonic vibrational frequencies were computed to verify the nature of the stationary points. Low imaginary frequencies were obtained for selected intermediates and transition states (see the Supporting Information for explanation and analysis). To ensure that the transition states link the products to the expected reactants, the normal modes corresponding to the imaginary frequencies were animated. To obtain the free energies (temperature of 433.15 K, pressure of 1 atm), the zero-point energies, thermal motion, and entropy corrections were added to the total electronic energy. Optimized structures are illustrated using CYLview.[32]

The relative Gibbs free energy for each intermediate and transition state was calculated by taking the difference of the sums of the Gibbs free energy for each reaction coordinate from the reference point (norbornadiene 1, alkyne 10, and the Cp*Ru(COD)Cl precatalyst). In terms of each pathway, each [2 + 2] reaction coordinate was composed of ΔG433.15(COD + IN/TS), while each [2 + 2 + 2] reaction coordinate was composed of ΔG433.15 (COD + IN/TS + Cl–).

Results and Discussion

Discussion of the Mechanism

Previous studies have suggested that the neutral [Cp*RuCl] moiety 16 is likely to be the active catalytic species in the [2 + 2] cycloaddition.[33a,33b] Dissociation of one of the double bonds of the cyclooctadiene (COD) ligand from the Cp*Ru(COD)Cl precatalyst followed by the ligand association with the alkyne would provide complex 15 (Scheme ). Upon dissociation of the COD ligand to form the coordinatively unsaturated species 16, either a second alkyne or NBD 1 could complex with 16 (Scheme ). It has been noted that the number of equivalents of NBD and alkyne had a considerable effect on the efficacy of the reaction. Excess of the alkene component improved the yield of the desired cycloadduct, while the use of excess alkyne decreased the yield dramatically.[33a,33b] Because Ru is known to form stronger π-complexes with alkynes than with alkenes,[34a,34b] the formation of the coordinatively saturated complex 17 is hypothesized to inhibit the active catalytic cycle. An excess of alkene results in the complexation to the exo-face of NBD, affording complex IN, starting the [2 + 2] catalytic cycle. Complexation of 16 with the endo-face of NBD was not investigated as an alternative complex, as the intermediate would not lead to the experimentally observed exo-cycloadducts. The full catalytic cycle for the [2 + 2] cycloaddition is constituted by three elementary steps (Scheme ): coordination of the alkyne to the exo-face of NBD to afford species IN; oxidative cyclization of the Ru–alkene–alkyne π-complex to provide metallacyclopentene IN via TS; and reductive elimination of metallacyclopentene IN to form the final [2 + 2] cycloadduct 17. The HDA [2 + 2 + 2] pathway is thought to begin in a similar manner proposed by Trost and co-workers in their 1993 report on CpRu(COD)Cl-catalyzed bis-homo-Diels–Alder cycloadditions of COD with alkynes.[35] The active cationic species is formed through the ionization of the chloride ligand under the condition of a polar solvent.[36a,36b] This coordinatively unsaturated species can complex to the endo-face on NBD to form species IN; oxidative cyclization of the Ru–alkene–alkyne π-complex will provide metallacyclopentene IN; migratory insertion of IN forms the second carbon–carbon bond; and reductive elimination of intermediate IN forms the last carbon–carbon bond, furnishing the HDA [2 + 2 + 2] cycloadduct 18. According to our calculations, the reductive elimination transition states TS have greater relative energy differences than the oxidative cyclization transition states TS in all cases. This indicates that TS2 is the rate-determining step for the [2 + 2] pathway (Scheme ). Conversely, the initial oxidative cyclization TS possesses a relative energy difference that is greater than both the migratory insertion TS and the reductive elimination TS, suggesting TS is the rate-determining step for the HDA [2 + 2 + 2] pathway (Scheme ). To determine the preferred pathway, the energy difference between the two competing transition states TS and TS was compared. Because of the potential unsymmetrical nature of the disubstituted alkynes, two configurations of the alkyne, A and B, were considered throughout both pathways—these further two competing pathways were denoted by including either a or b in the numbering systems (ie INa vs INb).

Scheme 3

Activation of the Cp*RuCl(COD) Precatalyst

Scheme 4

Proposed Mechanisms for the Ru-Catalyzed [2 + 2] and HDA [2 + 2 + 2] Cycloadditions of Norbornadiene 1 and Disubstituted Alkynes 10 for Path A (Top) and Path B (Bottom)

Effects of Phosphonate Substitution

We began studying the effects of phosphonate substitution on the Ru-catalyzed [2 + 2] and HDA [2 + 2 + 2] cycloadditions of NBD and alkynes. As reported by Kettles in 2011, both alkyl phosphonate moieties examined (Table , entries 1 and 2) were completely selective for the HDA adduct.[24] Because the theoretical results discussed above addressed the reductive elimination TS and oxidative cyclization TS as the rate-determining steps for [2 + 2] and HDA [2 + 2 + 2] cycloadditions, respectively, only these transition states will be discussed in detail (see the Supporting Information for the ΔG433.15‡ of all transition states). For the two alkynyl phosphonates, the relative Gibbs free energies at 433.15 K, ΔG433.15‡, for TS and TS with respect to the isolated reactants (NBD 1, alkyne 10, and Cp*RuCl(COD)) were calculated (see the Supporting Information). In both cases of alkyl substitution (Table , entries 1 and 2), the reductive elimination TSb was favored over TSa. This is most likely due to the increased negative charge placed on the carbon adjacent to the phosphonate moiety, an effect known to accelerate reductive elimination. In terms of HDA [2 + 2 + 2], both alkynyl phosphonates preferred oxidative cyclization TSb over TSa by 2–4 kcal/mol. Once again, it was noticed that adjacent electron-withdrawing functional groups, through either inductive or mesomeric effects, lowered the activation barrier for reductive elimination. To determine the preferred pathway, the differences in relative Gibbs free energies at 433.15 K, ΔΔG433.15‡, with respect to their preceding intermediates were calculated. The energy differences between the four possible transition states were computed to determine the lowest energy rate-determining transition state, qualitatively revealing whether the [2 + 2] or HDA [2 + 2 + 2] pathway is preferred, as well as the order of carbon–carbon bonds formed. Among the four transition states, TSb was the lowest in free energy for both dimethylphosphonate 10b and diethylphosphonate 10a (Table , entries 1 and 2), indicating the HDA [2 + 2 + 2] pathway is more energetically favorable. The theoretically predicted activation energy differences are in good agreement with the experimentally observed products,[24] as the lowest energy rate-determining step leads to the expected product. The potential energy profiles for the model reaction investigated, and the corresponding geometries for the most favored path, for the predicted Ru-catalyzed [2 + 2] and HDA [2 + 2 + 2] cycloadditions between NBD 1 and alkynyl phosphonate 10a are shown in Figure (see the Supporting Information for energy profiles for alkyne 10b).

Table 1

Relative Free Energies in kcal/mol at 422.15 K in DMF with Respect to the Preceding Intermediate for the Ru-Catalyzed [2 + 2] and HDA [2 + 2 + 2] Cycloadditions of Norbornadiene 1 with Alkynyl Phosphonates 10a,b

Figure 1

Potential energy profile of the Ru-catalyzed [2 + 2] and HDA [2 + 2 + 2] cycloadditions of norbornadiene 1 with alkynyl phosphonate 10a at 433.15 K in DMF. Energies are Gibbs free energies in kcal/mol with respect to separated reactants 1 + 10a + Cp*RuCl(COD).

Investigating Functionalized Acetylenes

To gain further insights into the dependency of the phosphonate moiety on the HDA [2 + 2 + 2] cycloaddition of NDA 1 and alkynes 10, a series of alternative functional groups were examined (Table ). For all of the reactions, the difference in relative Gibbs free energies at 433.15 K, ΔΔG433.15‡, for the reductive elimination TS and oxidative cyclization TS with respect to their preceding intermediates are given in Table . Replacement of the phosphonate moiety with sulfide 10d or bromide 10e flipped the reactivity with the [2 + 2] cycloaddition pathway, having lower activation energy than HDA [2 + 2 + 2] cycloaddition (Table , entries 2 and 3). Again, the reductive elimination TS and oxidative cyclization TS are predicted to be rate-determining steps for [2 + 2] and HDA [2 + 2 + 2] cycloadditions, respectively. The theoretically predicted activation energy differences at 433.15 K, ΔΔG433.15,DMF‡ are in good agreement with the experimentally observed products. The potential energy profiles for the alkynyl sulfide investigated, and the corresponding geometries for the most favored path, for the predicted Ru-catalyzed [2 + 2] and HDA [2 + 2 + 2] cycloadditions between NBD 1 and alkynyl phosphonates 10d are shown in Figure (see the Supporting Information for energy profiles for alkynes 10e). Similar to the findings in Table , it is predicted that pathway B is favored for asymmetric alkynes, as it is energetically more favorable for reductive elimination TSb to occur adjacent to the heteroatom functional group. The relevant molecular diagrams of the rate-determining transition states and adjacent intermediates for Ru-catalyzed [2 + 2] (INb, TSb, and 16) and HDA [2 + 2 + 2] (IN, TSb, and INb) cycloadditions between NBD 1 and alkynes 10a and 10d are shown in Figures and 4. Metal orbitals in IN and TS are similar, except for the notable charge transfer from the alkyne to the NBD ligand during TS (Figure ). The positive orbital overlap, as seen in the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of TS can be attributed to the electron transfer from the alkyne to the alkene ligand with characteristics from the LUMO of IN being noted in the HOMO of IN (Figure ). Like the alkynyl phosphonate (Figure ), the alkynyl sulfide reaction undergoes a similar charge transfer from the alkyne to the NBD ligand via TS (Figure ), however, with a smaller orbital overlap compared to 10a (Figure ). In terms of [2 + 2] molecular diagrams, a charge transfer can be seen taking place in ruthenapentacycle via the reductive elimination TS transition state (Figures and 4). Upon inspection of 10a (Figure ), the frontier molecular orbitals in TS are dominated by antibonding characteristics, which may explain the high-energy transition state. On the other hand, 10d exhibits a positive orbital overlap in TS (Figure ). This difference in the bonding nature may explain the preferential [2 + 2] and [2 + 2 + 2] pathways for 10a and 10d, respectively. The relevant geometries with Mulliken charges for the alkynyl phosphonate and sulfide are given in Figures and 6, respectively.

Table 2

Relative Free Energies in kcal/mol at 433.15 K in the DMF solution with Respect to the Preceding Intermediate for the Ru-Catalyzed [2 + 2] and HDA [2 + 2 + 2] Cycloadditions of Norbornadiene 1 with Acetylenes 10a,c,d

Figure 2

Potential energy profile for the Ru-catalyzed [2 + 2] and HDA [2 + 2 + 2] cycloadditions of norbornadiene 1 with alkynyl phosphonate 10d at 433.15 K in DMF. Energies are Gibbs free energies in kcal/mol with respect to separated reactants 1 + 10d + Cp*RuCl(COD).

Figure 3

Frontier orbital isosurface plots (isosurface value = 0.03 ) of the rate-determining transition states and adjacent intermediates for the Ru-catalyzed [2 + 2] (right) and HDA [2 + 2 + 2] (left) cycloadditions of NBD 1 with alkynyl phosphonate 10a at 433.15 K in DMF in the ground states taken from Figure . Surfaces computed at the B3LYP/def2SVP level of theory with eigenvalues converted to kcal/mol.

Figure 4

Frontier orbital isosurface plots (isosurface value = 0.03 ) of the rate-determining transition states and adjacent intermediates for the Ru-catalyzed [2 + 2] (right) and HDA [2 + 2 + 2] (left) cycloadditions of NBD 1 with alkynyl phosphonate 10d at 433.15 K in DMF in the ground states taken from Figure . Surfaces computed at the B3LYP/ def2SVP level of theory with eigenvalues converted to kcal/mol.

Figure 5

Mulliken charges of the rate-determining transition states and adjacent intermediates for the Ru-catalyzed [2 + 2] (top) and HDA [2 + 2 + 2] (bottom) cycloadditions of NBD 1 with alkynyl phosphonate 10a at 433.15 K in DMF.

Figure 6

Mulliken charges of the rate-determining transition states and adjacent intermediates for the Ru-catalyzed [2 + 2] (top) and HDA [2 + 2 + 2] (bottom) cycloadditions of NBD 1 with alkynyl phosphonate 10d at 433.15 K in DMF.

Effects of Aryl Substitution

For greater completeness and reliability, we studied four more reactions: NBD 1 with aryl-substituted diethylphosphonate alkynes 10e–h (Table ). For all four reactions, the differences in relative Gibbs free energies at 433.15 K, ΔΔG433.15‡ are given in Table . In agreement with dimethylphosphonate 10b or diethylphosphonate 10a (Table ), the lowest energy rate-determining step corresponds to the oxidative cyclization TS in the HDA [2 + 2 + 2] cycle with pathway B being a more energetically accessible route to the deltacyclane product 18 (Table ) (the energy profile diagrams for the five alkynyl phosphonates 10e–h investigated can be found in the Supporting Information). The theoretically predicted activation energy differences are in good agreement with the experimentally observed products,[24] as the lowest energy rate-determining step leads to the expected product.

Table 3

Relative Free Energies in kcal/mol at 433.15 K in the DMF solution with Respect to the Preceding Intermediate for the Ru-Catalyzed [2 + 2] and HDA [2 + 2 + 2] Cycloadditions of Norbornadiene 1 with Alkynyl Phosphonates 10a,e–h

Cationic versus Neutral Mechanism

Previous reports on the Ru-catalyzed HDA [2 + 2 + 2] cycloadditions of NBD 1 have suggested alternative catalytic cycles to the one presented in this report. In 2004, Tenaglia and co-workers reported the Ru-catalyzed HDA [2 + 2 + 2] cycloaddition of NBD using RuCl2(PPh3)2(NBD) as the precatalyst.[37] The mechanistic rationale for the observed results involved a neutral ruthenium species shuttling between 2+ and 4+ oxidation states. Although different mechanistic pathways may account for the HDA reaction depending on the proposed active catalytic species, we were interested if our precatalyst, Cp*RuCl(COD), could go through a neutral catalytic cycle. The exchange of the COD ligand for NBD would generate the 18-electron complex IN (Scheme ). The oxidative coupling of NBD to Ru allows for the formation of ruthenacyclobutane as a part of the 16-electron tetracyclic framework IN, which can coordinate to the alkyne partner 10 to afford IN. Migratory insertion of the alkyne would form ruthenacyclohexane IN. Reductive elimination of IN accounts for the formation of deltacyclane 18, as well as the regeneration of IN by coordination of NBD 1. As described in the previous catalytic cycle (Scheme ), two configurations of the alkyne, A and B, were considered throughout the pathway.

Scheme 5

Proposed Mechanism for the Neutral Ru-Catalyzed HDA [2 + 2 + 2] Cycloaddition of Norbornadiene 1 and Disubstituted Alkyne 10

To address the possibility of a neutral pathway, we studied our model reaction: the Ru-catalyzed cycloaddition of alkyne 10a and NBD 1. The differences in relative Gibbs free energies at 433.15 K, ΔΔG433.15‡, for the four transition states with respect to their preceding intermediate are given in Table . The oxidative cyclization TS has an energy barrier (11.1–31.9 kcal/mol) greater than alkyne association TS, migratory insertion TS, and reductive elimination TS transition states. This indicates that the oxidative cyclization is rate-determining for the overall neutral pathway. Comparing the energies for the rate-determining steps for cationic (TS) and neutral (TS) mechanisms, we see a strong potential energy difference of 18.6–23.8 kcal/mol (TS–TS), suggesting that the cationic mechanism is the preferred pathway. The potential energy profile for the neutral pathway investigated, and the corresponding geometries for the most favored path, for the predicted Ru-catalyzed HDA [2 + 2 + 2] cycloadditions between NBD 1 and alkynyl phosphonates 10a are shown in Figure .

Table 4

Relative Free Energies in kcal/mol at 433.15 K in the DMF solution with Respect to the Preceding Intermediate for the Ru-Catalyzed HDA [2 + 2 + 2] Cycloadditions of Norbornadiene 1 with Alkynyl Phosphonates 10a via a Neutral Mechanism

Figure 7

Potential energy profile of the PCM solvation model for the Ru-catalyzed HDA [2 + 2 + 2] cycloadditions of norbornadiene 1 with alkynyl phosphonate 10a in DMF via a neutral mechanism at 433.15 K in DMF. Energies are Gibbs free energies in kcal/mol with respect to separated reactants 1 + 10a + Cp*RuCl(COD).

Conclusions

In summary, the mechanism and chemoselectivity of the Ru-catalyzed [2 + 2] and HDA [2 + 2 + 2] cycloadditions of norbornadiene with disubstituted alkynes have been studied by DFT calculations. We performed hybrid DFT calculations to address the unexplored mechanism of the Ru-catalyzed HDA [2 + 2 + 2] of norbornadiene. Two mechanistic pathways were investigated, the cationic and neutral reactions. In both cases, the oxidative cyclization step was found to be a rate-determining step in the whole catalytic cycle. Under the reaction conditions explored, we found the cationic pathway is favored over the neutral pathway by 18.6–23.8 kcal/mol. The results provide theoretical support toward the unique reactivity of alkynyl phosphonates. In good agreement with experimental results, calculations revealed that the HDA [2 + 2 + 2] pathway is more energetically favorable than the [2 + 2] pathway with alkynyl phosphonates. Likewise, replacement of the phosphonate moiety with a bromide or sulfide group flipped the reactivity, with the [2 + 2] pathway being favored. In combination with these experimental results, the theoretical studies presented herein may shift the understanding of Ru-catalyzed cycloadditions and the effects of alkynyl functionality.