Two-Step Synthesis of Heptacyclo[6.6.0.02,6 .03,13 .04,11 .05,9 .010,14 ] tetradecane from Norbornadiene: Mechanism of the Cage Assembly and Post-synthetic Functionalization.

A selective and scalable two-step approach to the dimerization of norbornadiene (NBD) into its thermodynamically most stable dimer, heptacyclo[6.6.0.02,6 .03,13 .04,11 .05,9 .010,14 ] tetradecane, (HCTD) is reported. Calculations indicate that the reaction starts with the Rh-catalyzed stepwise homo Diels-Alder cyclisation of NBD into its exo-cis-endo dimer. Treatment of this compound with acid promotes its evolution to HCTD via a [1,2]-sigmatropic rearrangement. The assemblies of 7,12-disubstituted cages from 7-(alkyl/aryl) NBDs, as well as the selective post-synthetic C-H functionalization of the core HCTD scaffold at position C1, or positions C1 and C4 are described.

Introduction

The apparently simple structures of cubane, adamantane, dodecahedrane, and other highly symmetric cage hydrocarbons have fascinated chemists since decades and inspired numerous routes, mainly based on rearrangements, to achieve their synthesis in an efficient manner. Probably because of its availability, the most broadly studied cage hydrocarbon from these already mentioned is adamantane 1. In fact, this architecture has found application in areas as diverse as material science, and medicine.[ , ] Adamantane derived amines are approved drugs that display potent antiviral activity. For example, amantadine 2 is used to treat dyskinesia associated with parkinsonism and influenza caused by type A influenzavirus, tromantadine 3 has been identified as an active agent to treat herpes simplex virus infections, and bromantane 4 is an atypical psychostimulant and anxiolytic drug (Scheme 1). Adamantyl substituents have also been strategically used to modulate the lipophilicity of already known pharmaceuticals.

Scheme 1

Comparison of the structures of adamantane, diamantane, and HCTD.

The chemistry of the second member of the diamandoids series, diamantane 5, is well studied too. Although the reduced symmetry of this molecule (D 3) when compared with adamatane makes its selective functionalization a priori more challenging, convenient approaches have been developed over the years, which allow the hydroxylation, halogenation, amination, alkylation, or arylation of this compound, in some cases even on multi‐gram scale. The rigid structure of functionalized diamantanes has also found application as linker for the synthesis of polymers or in the design of mechanochemical devices, such as the molecular gyroscope 6, among others.

Given the impressive attention attracted by lower diamondoids, it is surprising that the chemistry of HCTD 7, the D 2‐symmetric isomer of diamantane in which all carbon atoms are embedded in fused 5‐membered rings, is comparatively unexplored. The reason probably lies in the fact that a practical synthesis able to selectively deliver this highly compact scaffold has remained a challenge for decades. The available routes invariably produced complex isomeric mixtures of norbornadiene (NBD) dimers, are stoichiometric in metal or simply afforded poor yields. Only recently has a practical protocol for its synthesis based on Ru‐catalysis been reported. Although still not studied in detail, cumulative evidence suggests that the cage closure starts with the formation of an endo‐endo NBD dimer A, which evolves through a sequence of consecutive olefin insertion steps into B.[ , ] Final C−H oxidative addition followed by C−C reductive coupling are probably the last steps of the cycle, delivering 7 and regenerating the Ru active species (Scheme 1). This synthetic route has finally paved the way to use the rigid HCTD cage as building block. On the other hand, the efficient post‐synthetic functionalization of the naked HCTD skeleton still remains unsolved. Only a handful of methods are reported, and most of them are characterized by exiguous yields or cleavage of the HCTD cage.

Herein, we describe a conceptually different approach to the synthesis of 7 and its derivatives based on an initial asynchronous homo‐Diels–Alder dimerization of NBD, followed by the acid‐catalyzed isomerization of the intermediate open dimer 8 into 7. Calculations at the PBE0‐D3BJ(PCM)/def2‐TZVPP//TPSS‐D3BJ/def2‐SVP level of theory support this mechanism. Moreover, studies on the post‐synthetic functionalization of the core architecture have identified bromination, but specially nitroxylation, as the most efficient entries to 1‐substituted HCTDs. The utility of these reactions is illustrated by the regioselective preparation of twelve HCTD derivatives in synthetically relevant yields, which include hydroxy‐, amino‐ and aryl‐ substituents. Preliminary examples of 1,4‐difunctionalized cages are also reported.

Results and Discussion

Initial Work Hypothesis

The difficulties associated with the synthesis of 7 come from the necessity to effectively control the very first C−C bond formation event between the two NBD moieties; a priori it needs to proceed with endo‐endo diasteroselectivity to afford a direct precursor of 7. Hence, the metal used as template must hold the two η4‐NBDs ligands that are going to be connected facing each other with their endo‐sides. The dire consequence of this unique situation is that the usual tool to gain control over stereoselectivity, the use of costumed auxiliary ligands, is quite limited here. Specifically, when using Rh catalysts, the employment of chelating ligands quenches any reactivity towards dimerization, while the imposition of significant steric requirements leads to the wrong hapticity in the NBD ligands. As a consequence, undesired mixtures of dimers, preferentially formed via less sterically demanding exo‐coupling pathways, are obtained.

At this stage we envisioned that if compound 8, the exo‐cis‐endo dimer of NBD, could be prepared selectively, it might be used as a precursor for 7. Under conditions still to be optimized, treatment of 8 with Lewis or Brønsted acid should promote the [1,2]‐sigmatropic rearrangement of 8 into 9; while this last compound is already known to evolve, albeit in poor yield, to the thermodynamic sink 7 by treatment with HI (Scheme 2 a). Hence, with these isomerizations effectively achieved, the synthesis of 7 gets reduced to the selective preparation of 8 via homo‐Diels–Alder reaction (Scheme 2 b).[ , ]

Scheme 2

Working hypothesis and synthesis of precatalyst 11.

Based on our previous studies employing cationic ancillary ligands, we expect RhI complex 11 bearing the chelating phosphine 10 to be a suitable precatalyst for this [2+2+2] cycloaddition. The dicationic nature of 10 makes the Rh atom electron deficient and this facilitates, as shown below, the expansion of the coordination sphere of Rh to a fifth ligand. That is fundamental to assemble the two NBDs with the right hapticity around the metal. Moreover, the steric hindrance created in one side of the metal by the dicationic ligand is expected to favor the now desired exo‐[2+2+2] cyclization. Precatalyst 11 is prepared by reaction of 10 with [Rh(nbd)Cl]2 in acetone (see the Scheme 2 c and the Supporting Information), but as it will be mentioned later, it does not need to be isolated. Identical results were observed when the dimerization experiments were carried out with 10 and [RhCl(cod)]2 mixed in situ.

Catalysis

The catalytic performance of 11 (2 mol %) on the desired dimerization was initially tested in 1,2‐dichloroethane at 70 °C. Under these conditions the desired reaction is not efficient (only 7 % conversion); interestingly however, 8 is the major component from the obtained mixture of products (Table 1, entry 1). We were glad to see that addition of NaB(C6F5)4 to the reaction mixture under otherwise identical conditions was enough to improve the overall conversion up to 89 %. This additive substantially increases the solubility of 11 in 1,2‐dichloroethane by anion exchange and probably helps to abstract the chloride ligand from the coordination sphere of Rh. However, considerable amounts of trimer 12 were also formed together with minor amounts of 9 and other dimers (Table 1, entry 2). Having 9 in the reaction mixture is not problematic because this compound is in the route towards 7; on the other hand, the trimerization of NBD needs to be minimized. Further optimization of the reaction conditions indicated that increasing the working temperature to 90 °C was sufficient to achieve complete substrate conversion in 16 h., while working under more diluted conditions ([NBD]i=0.2 M) proved to be essential to nearly suppress the formation of undesired 12. Under the optimized conditions the desired dimer 8 is obtained as a mixture with 9 in a circa 9:1 ratio, respectively, and quantitative yield (Table 1, entry 4). Identical results were observed when these experiments were carried out using a 2:1 mixture of 10 and [RhCl(cod)]2 or [RhCl(nbd)]2 instead of preformed 11. No reaction was observed when ligand 10 was replaced by 1,2‐bis(diphenylphosphino)benzene (dppbz) under otherwise identical conditions.

Table 1

Rh‐catalyzed dimerization of NBD. Optimization of reaction conditions.

Entry	Additive	T [°C]	Ratio 8:9:12:13 [%]	Conv. [%]
1[a]	–	70	75:10:15:0	7
2[a]	NaB(C6F5)4	70	67:2:22:9	89
3[a]	NaB(C6F5)4	90	76:7:12:5	98
4[b]	NaB(C6F5)4	90	86:10:1:3	97

[a] Experiments carried out with an initial NBD concentration of 0.3 M; [b] initial NBD concentration of 0.2 M.

–

In an attempt to gain further knowledge on the nature of the actual catalytic species, the orange oils remaining after extraction of the reaction mixtures with n‐pentane were combined and dissolved in acetonitrile. Slow evaporation of that solution produced a few crystals of two different Rh complexes 14 and 15, whose structures were determined by X‐ray crystallography (Figure 1).

Figure 1

X‐ray structures of 14 and 15. H atoms and SbF6 anions omitted for clarity; ellipsoids set at 50 % probability.

In 14 the first coordination sphere about the metal center can be described as a distorted square pyramidal environment. The base of the pyramid is formed by the chelating phosphine 11 and a η4‐NBD unit, while a loosely coordinated acetonitrile ligand is accommodated at the apical position. The fact that a d8 RhI coordinates a fifth ligand beautifully illustrates the enhanced electrophilicity at that metal center, which is surely induced by the dicationic ancillary ligand. Moreover, the feasibility of a simultaneous binding of ligand 10 and NBD to the Rh atom is demonstrated. Unfortunately, the extreme insolubility of 14 in common organic solvents once crystallized does not permit its spectroscopic characterization. The structure of 15 is very similar; the only difference being that a chloride anion occupies the apical position at Rh instead CH3CN. Both, 14 and 15 are catalytically competent when submitted to the optimized reaction conditions.

At this stage, the interconversion of the obtained 8/9 mixture into 7 was attempted via treatment of the crude dimerization reaction mixtures with acid. A complete series of Lewis and Brønsted acids including BF3, AlCl3, HBF4, HSbF6, and [H(OEt2)2][B(C6F5)4] were tested. Optimal results were achieved when using [H(OEt2)2][B(C6F5)4] (5 mol %) in highly diluted conditions (0.05 M); this minimizes the formation of undesired polymers. Our overall two‐step protocol has been scaled up to one gram of NBD and is not only practical for the synthesis of 7 but also for the preparation of 7,12‐disubstituted cages 16–20, from 7‐(alkyl/aryl) substituted norbornadienes (Scheme 3).

Scheme 3

Substrate scope. Yields are of the isolated products from the corresponding 7‐substituted NBD (two steps). X‐ray structure of 18, ellipsoids set at 50 % probability.

Mechanistic Studies

The mechanism for the dimerization of NBD into 8 was investigated using density functional theory (DFT at the PBE0‐D3BJ(PCM)/def2‐TZVPP//TPSS‐D3BJ/def2‐SVP level. Questions regarding the mechanism at the outset included 1) what is the lowest energy pathway for formation of product 8; and 2) why does ligand 10 lead to the unique formation of product 8 in contrast to methods using monodentate phosphine ligands.[ , ]

Multiple pathways were considered in devising the most favorable pathway for product formation. Traditional transition metal catalyzed NBD dimerization using monodentate phosphine ligands involves bis‐coordination of two NBD units followed by a reaction sequence of oxidative cyclization, carbometalation, and reductive elimination (Mechanism A; Supporting Information, Scheme S1). This pathway is disfavored when bidentate phosphine ligands are used because they prevent a second NBD from coordinating. In view of the isolation of 14 and 15, in which the Rh atom is pentacoordinated, this does not seem to be a problem when using 10 as the ancillary ligand.

There are thus two alternative main pathways that can be envisioned; the one that starts with an inner‐sphere attack of the second NBD to Rh through only one double bond, or an outer‐sphere mechanism in which a non‐coordinated NBD attacks an already coordinated NBD. The inner‐sphere attack of a NBD can only take place on the unsaturated 16 electron cationic Rh formed from Cl dissociation; the isolation of 14 demonstrates that this happens easily under the reaction conditions applied. An outer‐sphere attack can take place on either the Rh‐Cl complex 15 or in a tricationic complex resembling 14. All three mechanisms were explored (Figure 2; Supporting Information, Figures S2 and S3), and the lowest energy pathway (4.3 kcal mol−1 lower than the next closest one) is discussed here, which is the inner‐sphere mechanism.

Figure 2

Gibbs free‐energy profile computed at the PBE0‐D3BJ(PCM)/def2‐TZVPP//TPSS‐D3BJ/def2‐SVP level of DFT. The green pathway corresponds to the endo pathway and magenta the exo pathway relating to NBD approach. The “=” between TS3 and Prod symbolizes the presence of a higher‐energy, shallow intermediate.

This mechanism may lead to up to four different NBD dimers depending on the diasteroselectivity of the attack; we only show in Figure 2 the calculated pathways leading to products 8 (Prod) and 9 (Prod). The mechanism initiates through association of the second NBD to the Rh atom using one of its π‐systems to form the corresponding 18 electron complexes in either an exo (INT1) or endo (INT1) configuration, with exo association being favored by 2.3 kcal mol−1. This energetic preference is reflected structurally in a shorter distance between the Rh center and C atoms of the associating π‐system in INT (2.41 Å; 2.50 Å) than in the INT (2.59 Å; 2.68 Å), indicating a stronger interaction and suggesting that endo association is sterically more hindering than the exo. The subsequent oxidative cyclisation, which is calculated to be the rate determining step of the mechanism, takes place with the exo pathway maintaining the energetic preference by 2.5 kcal mol−1 (compare TS1 and TS1). While the initial intermediates after TS1 can be considered 16 electron complexes, the second π‐system of the endo pathway, INT2′, weakly coordinates to the Rh center to form INT2 resulting in a higher stabilization for the endo over exo intermediate (Figure 2). The participation of the second π‐system is geometrically restricted from coordinating in INT2 and thus, the Rh atom in INT2 remains unsaturated. The second C−C bond is next formed through carborhodation leading to INT3. INT3 is substantially lower in energy than INT3 due to the stabilization provided by coordination of the olefin π‐bond, which cannot be compensated by the weaker σ C−H agostic interaction in INT3. The products are then formed through reductive elimination in the final step forming the cyclopropane ring. From this analysis it can be concluded that even though the second, nonreactive π‐system in the endo pathway helps with stabilization through coordination, this assistance only takes place after the rate determining step TS1; as consequence, the formation of 8 over 9 is favored. Moreover, it is clear how trimer 12 may be formed through the association of a third NBD unit to Prod.

The formation of the other two diasteromeric NBD dimers was also considered (Supporting Information, Figure S1); the corresponding TS1 for each are calculated to be higher in energy than the lowest energy exo pathway discussed here.

Post‐synthetic Functionalization of the HCTD Cage

The C−H functionalization of adamantane and related structures is well developed; hence, we decided to make use of these already existing protocols to attempt the derivatization of 7. Initially, the bromination of the HCTD cage was targeted. Direct treatment with Br2 under a variety of conditions resulted to be completely inefficient and 7 was re‐isolated from the reaction mixtures untouched. Hence, we turned our attention to the functionalization of 7 employing Schreiner's protocol using [Br3C]. radicals generated under phase transfer catalytic conditions. To our delight, bromine derivative 21 bearing the Br‐substituent in position 1 could be obtained using this method, albeit in moderate yield. Importantly, although 7 contains two non‐equivalent tertiary C−H bonds, the 6‐Br derivate was not detected under these conditions (Scheme 4).

Scheme 4

Preparation of 1‐substituted HCTD. Reagents and conditions: a) CBr4 (1 equiv), [BnNEt3]Cl (7 mol %), 50 % NaOH (aq.)/CH2Cl2, 70 °C for 18 h; b) HNO3/CH2Cl2, r.t.; c) H2SO4 and then H2O; d) H2SO4/C6H6, 50 °C; e) H2SO4/CH3CN, 50 °C; f) H2SO4/NCCH2CN, 50 °C; g) H2SO4/ClCH2CN, 50 °C. X‐ray structures of 23 and 27, ellipsoids set at 50 % probability.

A more general approach however for the preparation of functionalized HCTDs is nitroxylation, which we carried out in neat HNO3 under identical conditions to those employed to substitute diamantanes. Once obtained, the nitroxyl derivative 22 can be hydrolyzed to produce alcohol 23 in much better yield than previously reported. Additionally, treatment of 22 with H2SO4 generates the transient carbocation intermediate that can be trapped either with an arene following a Friedel–Crafts alkylation mechanism, such as in the case of 24, or with nitriles in a typical Ritter‐type reaction. Amides 25–27 are obtained in good to excellent yields following this method; the structures of 25 and 27 being confirmed by X‐ray diffraction (Scheme 4 and the Supporting Information). Functionalization of 7 at position 1 destroys the symmetry of the original cage and as a consequence; compounds 21–27 are chiral even though they are obtained as racemates.

Following the methodology already described, HCTD analogues to adamantane derived clinically approved drugs have been prepared as well. For example, hydrolysis of 27 under acidic conditions delivers amine 28, which can be considered an amantadine analogue; while reaction of the same compound with deprotonated 2‐(dimethylamino)ethanol affords 30, the HCTD version of Tromantadine 3. Finally, re‐subjection of 22 to nitroxylation conditions affords the dinitroxyl derivative 31. The relative 1,4‐disposition of the two ‐ONO2 units was initially suspected by NMR spectroscopy, which indicated the formation of a C2‐symmetric product, and later confirmed by X‐ray crystallography. Hydrolysis of 31 cleanly delivers diol 32. The isolation of these two compounds represents a first milestone for the preparation of di‐functionalized HCTD cages (Scheme 5).

Scheme 5

Preparation of HCTD analogues to adamantane‐derived drugs and dinitroxylation of the cage. Reagents and conditions: a) thiourea (1.2 equiv), 10 h. and then aq. NaOH solution; b) MeNCS (1.0 equiv), EtOH, r.t. 12 h; c) Me2NCH2CH2OH (1.5 equiv), NaH (1.5 equiv), THF, reflux, 3 h.; d) HNO3, 16 h. r.t. 16 h; e) H2SO4 and then H2O. X‐ray structure of 31 and 32, ellipsoids set at 50 % probability and solvent molecules omitted for clarity.

Conclusion

We have presented an efficient two‐steps protocol for the synthesis of HCTD 7. For the first one a Rh‐catalyzed homo Diels–Alder dimerization is utilized. The dicationic ancillary ligand that the Rh center bears makes possible this reaction by both, activation of the π‐system, and limiting formation of undesired products of trans‐dimerization, as supported by DFT. Subsequently, the initial dimers obtained are readily transformed into 7 via acidic treatment. Making use of the chemistry already developed for the C−H functionalization of adamantanes, a series of HCTD derivatives have been prepared in synthetically useful yields. Ongoing research in our group is focused on the development of protocols allowing the asymmetric mono‐ and di‐functionalization of 7.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.